conservation of plant genetic resources.

Forest Ecology and Management 333 (2014) 76–87

Contents lists available at ScienceDirect

Forest Ecology and Management

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / f o r e c o

The role of forest genetic resources in responding to biotic and abiotic factors in the context of anthropogenic climate change

http://dx.doi.org/10.1016/j.foreco.2014.04.006 0378-1127/� 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

⇑ Corresponding author. Tel.: +1 2502982363. E-mail addresses: [email protected] (R.I. Alfaro), [email protected]

(B. Fady), [email protected] (G.G. Vendramin), [email protected] (I.K. Dawson), [email protected] (R.A. Fleming), [email protected] (C. Sáenz-Romero), [email protected] (R.A. Lindig-Cisneros), tmurdock@uvic. ca (T. Murdock), [email protected] (B. Vinceti), [email protected] (C.M. Navarro), [email protected] (T. Skrøppa), [email protected] (G. Baldinelli), [email protected] (Y.A. El-Kassaby), [email protected] (J. Loo).

René I. Alfaro a,⇑, Bruno Fady b, Giovanni Giuseppe Vendramin c, Ian K. Dawson d, Richard A. Fleming a, Cuauhtémoc Sáenz-Romero e, Roberto A. Lindig-Cisneros f, Trevor Murdock g, Barbara Vinceti h, Carlos Manuel Navarro i, Tore Skrøppa j, Giulia Baldinelli k, Yousry A. El-Kassaby l, Judy Loo h

a Canadian Forest Service, Natural Resources Canada, Canada b INRA, UR629, Ecologie des Forêts Méditerranéennes, Avignon, France c National Research Council, Institute of Biosciences and Bioresources, Firenze, Italy d The World Agroforestry Centre, Headquarters, PO Box 30677, Nairobi, Kenya e Instituto de Investigaciones Agropecuarias y Forestales, Universidad Michoacana de San Nicolás de Hidalgo, Morelia, Michoacán, Mexico f Centro de Investigaciones en Ecosistemas, Universidad Nacional Autónoma de México (CIECO-UNAM), Morelia, Michoacán, Mexico g Pacific Climate Impacts Consortium, University of Victoria, Victoria, British Columbia, Canada h Bioversity International, Via dei Tre Denari, 472/a 00057 Maccarese (Fiumicino), Rome, Italy i Azuero Earth Project, Apartado 0749-00015, Pedasí, Panamá j Norwegian Forest and Landscape Institute, Box 115, 1431 Ås, Norway k Dept. of Development Studies, School of Oriental and African Studies, SOAS, University of London, UK l Department of Forest and Conservation Sciences, Faculty of Forestry, The University of British Columbia, Vancouver, BC V6T 1Z4, Canada

a r t i c l e i n f o

Article history: Available online 6 May 2014

Keywords: Tree genetic variation Climate change Adaptation Natural disturbances

a b s t r a c t

The current distribution of forest genetic resources on Earth is the result of a combination of natural pro- cesses and human actions. Over time, tree populations have become adapted to their habitats including the local ecological disturbances they face. As the planet enters a phase of human-induced climate change of unprecedented speed and magnitude, however, previously locally-adapted populations are rendered less suitable for new conditions, and ‘natural’ biotic and abiotic disturbances are taken outside their historic distribution, frequency and intensity ranges. Tree populations rely on phenotypic plasticity to survive in extant locations, on genetic adaptation to modify their local phenotypic optimum or on migration to new suitable environmental conditions. The rate of required change, however, may outpace the ability to respond, and tree species and populations may become locally extinct after specific, but as yet unknown and unquantified, tipping points are reached. Here, we review the importance of forest genetic resources as a source of evolutionary potential for adaptation to changes in climate and other eco- logical factors. We particularly consider climate-related responses in the context of linkages to distur- bances such as pests, diseases and fire, and associated feedback loops. The importance of management strategies to conserve evolutionary potential is emphasised and recommendations for policy-makers are provided.

� 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

1. Introduction

Forests cover approximately 30% of the world’s total land mass (FAO, 2010) and are an integral part of life on earth, providing a range of services at local, national and global levels. Projected changes in climate, both gradual and extreme events, pose a serious threat to forestry (IPCC, 2011). As such, international organizations are currently engaged in actions to address the interconnected chal- lenges of deforestation, forests degradation and desertification in a changing environment. Not only does climate change pose a threat to forests themselves, but also to the millions of people who depend

R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87 77

on them directly for their livelihoods (Dawson et al., 2014, this special issue), and to the billions who are supported by forests through the provision of environmental services that are vital to humanity (UNEC, 2009; FAO, 2010).

Global climate change projections depend on future rates of greenhouse gas emissions, but expected temperature increases range from 1.1 �C to 2.9 �C by 2090–2099 (compared to 1980– 1999) for a low (B1) emissions scenario, 1.7 �C to 4.4 �C for a med- ium (A1B) scenario and 2.0 �C to 5.4 �C for a high (A2) scenario (Solomon et al., 2007). Even a change at the lower end of this range is significant for forests and trees. Considerable changes in precip- itation are also projected, with locations that are currently dry receiving generally less precipitation and locations that are cur- rently relatively wet receiving more (Solomon et al., 2007). Evi- dence for negative effects of climate change on forests globally is mounting (Allen et al., 2010). In North America, for example, whitebark pine (Pinus albicaulis Engelm.) is dying due to a combi- nation of drought-induced stress, mountain pine beetle attack (Dendroctonus ponderosae Hopkins) and blister rust (Cronartium ribicola A. Dietr.) that is attributed to climate change (Campbell and Antos, 2000; Smith et al., 2008; Zeglen, 2002). Other negative effects attributed to climate change include: the massive die-off (on 12,000 km2) of Pinus edulis (Engelm.) in the southwestern USA (Breshears et al., 2005); the sudden decline of Populus tremu- loides (Michx.) in the USA’s Rocky Mountains (Rehfeldt et al., 2009); the decline in Cedrus atlantica ([Endl.] Manetti ex Carrière) in the Middle Atlas mountains of Morocco (Mátyás, 2010); the decline of Fagus sylvatica L. in southwest Hungary (Mátyás et al., 2010); and the replacement of F. sylvatica by more drought-toler- ant Quercus ilex L. in Catalonia, northeast Spain (Peñuelas et al., 2007).

Although in this paper our focus is the challenges in responding to anthropogenic climate change, it should be noted that human- included environmental alteration also carries some potential ben- efits for forest production in particular regions, where net produc- tivity may be raised due to increases in CO2 levels and temperature (in contemporary cold regions), if drought stress does not become limiting. For crops, modelling shows that drought often becomes constraining despite elevated CO2 levels acting as a ‘fertilizer’ (Parry et al., 2004). In cold climates, it is not unusual for natural tree populations to be located under sub-optimal conditions, with the discrepancy between the inhabited and the optimal climate increasing with the severity of climate (Rehfeldt et al., 2004). In such locations, an increase in temperature, coupled with at least stable precipitation, may result in increased wood yields in the short- to medium-term. Projected examples of such increases include: Pinus banksiana in the North American Great Lakes region (Mátyás and Yeatman, 1992; Mátyás, 1994); Pinus contorta, Pinus sylvestris and Larix sibirica in Siberia (Rehfeldt et al., 1999, 2001, 2004); Picea glauca in southern Quebec (Beaulieu and Rainville, 2005); and Pseudostsuga menziesii in western North America (Leites et al., 2012a,b). In the longer-term, however, declines are expected as adaptive and plastic capacities to respond to change are exhausted (Mátyás et al., 2010).

Here, we address the role that forest genetic resources (FGR, the genetic variation in trees of present or potential benefit to humans; FAO, 1989) can play in responding to anthropogenic climate change. The present distribution of FGR globally is the result of nat- ural geological, ecological and genetic processes, which, over thou- sands of years, and along with the influence of man, have resulted in adaptation to local environments (Alberto et al., 2013). Included in this is adaptation to local disturbances, such as fire, insects and diseases. We review the pressures on FGR imposed directly by changing climate, as well as the indirect impacts on forests induced by changes in the biotic (e.g., insect and disease) and abiotic (e.g., fire, flood) disturbances that affect them. In particular, we consider

climate-related responses in the context of linkages to distur- bances and associated feedback loops, an issue not widely addressed in previous reviews on climate change and tree genetic resources. We conclude by discussing the feasibility of various management options to utilize the genetic variation in trees to respond to climate change and present options for policy-makers.

2. The impacts of climate change on FGR

Impacts are experienced through several demographic and genetic processes (Kremer et al., 2012; Savolainen et al., 2011). Some are directional and gradual, such as trends in increasing tem- perature and reducing rainfall, while others involve abrupt change, including drought, flood, fire and sudden pest invasions (in this paper we refer to these as catastrophic events; Scheffer et al., 2001; Scheffer and Carpenter, 2003). If environmental change is directional and continuous, fast-maturing trees in particular may have the potential to adapt genetically (Hamrick, 2004). At the receding edge of species distributions in particular, however, the magnitude and speed of projected anthropogenic climate change is likely to surpass adaptive capacity in many cases, resulting in local extirpations (Davis and Shaw, 2001). As climate changes, spe- cies and genotypes within species that are mal-adapted may be replaced by fitter ones that are already present at a site or by geno- types migrating from elsewhere. At the ecosystem level, the result will be a change in the relative abundance of species and genotypes in the landscape. Such changes may be unpredictable, with signif- icant changes in net ecosystem productivity possible (Thornley and Cannell, 1996; Wang et al., 2012). Extirpation of ecologically important keystone species will have critical impacts on coexisting organisms and their adaptation.

Climate change may also result in high variability in tempera- ture and precipitation, with an increase in incidence of extreme events, such as flooding, late frosts and intensive summer droughts, amongst other events (IPCC, 2011) (Table 1). In some areas, such as the Mediterranean and the Neo-tropics, an increase in seasonality is also expected (Alcamo et al., 2007; Meir and Woodward, 2010). Under such conditions, natural selection may not result in efficient adaptation because selection pressures are multi-directional, involving traits that may be inversely correlated at the gene level (Jump and Peñuelas, 2005). The standing genetic variation in populations may then not be large enough to create the rare new genotypic combinations that are required. Ecosys- tems affected by abrupt change may sustain rapid and widespread transformation as ecological tipping points are exceeded (Lenton, 2011). Given the pivotal role of trees in ecosystem function, abrupt climate change impacts on them may thus have profound conse- quences for forests as a whole (Whitham et al., 2006). Irreversible loss of ecosystem integrity and function may follow, with replace- ment by new non-endemic ecosystems (Gunderson and Holling, 2002; Mooney et al., 2009).

3. Responses of tree populations to environmental change

3.1. Adaptation and ‘standing’ genetic variation

Tree populations rely on three interplaying mechanisms to respond to environmental change: adaptation, migration; and phe- notypic plasticity (Davis and Shaw, 2001; Jump and Peñuelas, 2005). Genetic adaptations that make a population more suited for survival are achieved through gene frequency changes across generations (Koski et al., 1997). Many tree species have high genetic variability in adaptive traits and can therefore grow under a wide range of conditions (Gutschick and BassiriRad, 2003). Indeed, phenotypic traits of adaptive importance, such as drought

Table 1 Forest genetic resources under pressure: climatic drivers of change.

Direct effects of changing climate These include high tree mortality through extreme climatic events, particularly drought in combination with widespread regeneration failure (IPCC, 2011). Malhi et al. (2009), for example, examined the evidence for anthropogenic climate change leading to future large-scale ‘‘dieback’’ in Amazonian rainforest. Analysis suggested that dry-season water stress is likely to increase in eastern Amazonia over the 21st century, with the region tending toward a climate more appropriate to seasonal forests. Due to their deep roots, trees are able to persist under extreme weather events such as droughts for longer periods than many non-woody taxa can, but this persistence should not be over-estimated. For example, in an experiment in the Amazon in which rainfall was restricted to mimic savannah conditions, Nepstad et al. (2007) demonstrated that there was a lag of only three years before the increased mortality of mature trees due to limited water availability

Effects of changing climate on organisms associated with trees

In particular, changes in the biology of insect pests and diseases may make ecosystems more susceptible to tree mortality (Alfaro et al., 2010). Because of improved environmental conditions for the pest and reduced tree resistance due to increased stress, pests may react to climate change with range expansions and/or increases in attack severity (Raffa et al., 2013). Since many pests have short generation times, large populations and strong dispersal abilities, they may adapt to environmental change more quickly than host trees (Harrington et al., 2001)

Changes in abiotic disturbance regimes These include changes in fire regimes, flooding, landslides and/or hurricanes. Fire and climate are closely linked and are also associated with changes in land use (Piñol et al., 1998; Pausas, 2004). Coupled climate and fire-risk models (Moriondo et al., 2006) suggest not only an increase in the frequency of fires but also in fire size and length of the fire- risk season, with some areas subject to risk that were not before. Malhi et al. (2009) considered how tipping points may be reached in Amazonian rainforest by a combination of increased dryness and an increased incidence of fire events

Invasion by organisms foreign to local ecosystems

This includes the increased risk of establishment by invasive species which accidentally arrive into ports of entry, through globalized commerce. By making new niches available, climate change will facilitate the survival of mammals, insects, diseases and/or weeds foreign to endemic ecosystems. These include invasive exotic trees introduced for production and amenity purposes that are more precocious, have higher seed dispersal distances, are more fecund and/ or are more adaptable than existing species, or that are better suited to new environmental conditions (Peterson et al., 2008)

78 R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87

tolerance, cold-hardiness, resistance to pests and diseases, and flowering and fruiting period, have been shown to vary across eco- logical and geographic gradients to an extent that may be as important as the differences observed amongst species (Alberto et al., 2013; Petit and Hampe, 2006). The result is local adaptation along these gradients (Alberto et al., 2013; Savolainen et al., 2007). Navarro et al. (2002, 2005), for example, found that Cedrela odorata L. populations sampled from areas with long dry periods were more adapted to drought than those collected from wet areas. In relation to pests, Alfaro et al. (2013) indicated that populations of Sitka spruce (Picea sitchensis [Bong.] Carr.) with resistance to Pis- sodes strobi Peck were more common in areas with intense pest pressure than in areas where the pest was absent. The process of adaptation to climate change is influenced by migration and genetic drift, with fitness trait values shifting over generations to track environmental change and to ensure the survival of tree populations, with the emergence of endemic populations and spe- ciation (Futuyma, 2010; Kremer et al., 2012; Savolainen et al., 2011).

Although a large amount of genetic diversity per se does not guarantee adaptation and adaptability (Gomulkiewicz and Houle, 2009), the high within-population genetic diversity observed in many forest tree species (but see Vendramin et al., 2008 for a coun- ter example) can support an optimistic view that climate change challenges may be met by standing genetic variation in many cases (Hoffmann and Sgro, 2011). Many forest trees, for example, have high genetic diversity in important adaptive traits, such as tallness, longevity and defense mechanisms (Petit and Hampe, 2006). Trees also often have high fecundity (El-Kassaby et al., 1989), which cre- ates a large gene pool to select from. The speed of adaptive response within populations also depends on the size of the popu- lation; the heritability of fitness-related traits; interconnectedness; and the intensity, direction and duration of the selection pressure.

Field trials have been central to demonstrating the extent and distribution of genetic diversity in fitness-related traits in tree spe- cies (Kremer et al., 2002; Savolainen et al., 2007). Experiments have been conducted mostly on boreal and temperate species and a few commercially important tropical trees (Aitken et al., 2008; Alberto et al., 2013). Recently, however, there has been a move to include a wider range of less commercial species in the tropics (Ræbild et al., 2011). New studies on indigenous African

fruit trees, for example, have specifically considered traits impor- tant in the context of climate change adaptation (see www.safrui- t.org). The information being obtained on the effects of different treatments on root development, seedling vigour and other impor- tant adaptive characteristics will inform the strategies by which planting material of these fruit trees is supplied to African small- holders (Sanou et al., 2007). In addition to common garden trials, recent molecular-level studies have demonstrated allelic shifts in genes related to drought and heat tolerance amongst tree popula- tions, variables that are relevant for local adaptation (Grivet et al., 2011).

Evidence from field experiments suggests that a balance between divergent selection across contrasting ecological sites and reproductive contact has maintained enough genetic diversity to support adaptation to changing environments in the past (Kremer et al., 2010). Certainly, it has been demonstrated that maintaining high genetic diversity within and amongst tree popu- lations can increase ecosystem resilience (Whitham et al., 2006; Thorsen and Kjær, 2007), especially when trees are keystone spe- cies (Barbour et al., 2009). Intra-specific diversity can promote both resilience to pest attack and the productivity of individual species; economic modelling has, for example, shown that in some cases more optimal production under climate change will be attained in plantations by ‘‘composite provenancing’’ from within a species’ range (Bosselmann et al., 2008; Hubert and Cottrell, 2007).

The fast pace of anthropogenic climate change and the compar- atively long generation interval of many trees, however, mean that there may be insufficient time for natural selection to give rise to genotypes within populations that are adapted to new environ- ments (Jump et al., 2006). When environmental conditions change at a rate beyond the point where they cause demographic declines, the adaptive challenges faced by populations are markedly differ- ent from those experienced during demographic expansions (Gomulkiewicz and Holt, 1995). In a race between decline and evo- lutionary change, if genetic change is too slow population extinc- tion will be the result. Only when the pace and extent of environmental change is moderate, when a population is initially large, and when evolutionary potential is high, is a population likely to be rescued through adaptation (Gomulkiewicz and Holt, 1995; Gomulkiewicz and Houle, 2009).

R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87 79

3.2. Migration via pollen and seed movement

Pollen- and seed-mediated gene flow can facilitate adaptation to new environmental conditions by replenishing population genetic variation (Bridle et al., 2010; Le Corre and Kremer, 2003; Polechova et al., 2009), and by reducing the effects of genetic drift in small stands (Alleaume-Benharira et al., 2006; Lopez et al., 2009). Under climate change, the asymmetric gene flow from large central populations to small peripheral ones (Kirkpatrick and Barton, 1997; Lenormand, 2002) should prove beneficial for popu- lations at the leading edge of migration fronts, but possibly mal- adaptive for populations at the rear edge (Hampe and Petit, 2005). Pollen is known on occasions to travel very long distances, particularly in wind-dispersed broadleaves and conifers (Liepelt et al., 2002), but also sometimes for animal-pollinated species (Jha and Dick, 2010; Kramer et al., 2008; Oddou-Muratorio et al., 2005; Ward et al., 2005). Paleoecological reconstructions of the recolonisation of temperate zones during the Holocene have also suggested that seeds are capable of travelling long distances rap- idly (Brewer et al., 2002; Nathan et al., 2002), in the range of sev- eral hundreds of meters per year. Landscape genetic approaches, macrofossil evidence and theoretical studies, however, indicate that cryptic refugia may have been overlooked, considerably reducing migration estimates (McLachlan et al., 2005; Roques et al., 2010; Willis and van Andel, 2004). In addition, modern esti- mates of contemporary seed dispersal, although pointing to the existence of long distance dispersal events, generally indicate that median migration rates are in the range of a few tens of meters per year (Amm et al., 2012; Clark et al., 1998; Sagnard et al., 2007; Willson, 1993).

Whereas such modest migration rates are enough to keep pace in mountain and tropical conifer biomes, migration rates of over 1 km per year may be needed, even under quite modest scenarios of temperature change, in tropical and boreal broadleaf biomes (Loarie et al., 2009). In addition, rates of natural migration are reduced by forest degradation and fragmentation, which therefore increase vulnerability to climate change (Kellomäki et al., 2001; Malcolm et al., 2002). Trees in agricultural land or planted in cor- ridors can enhance pollen-mediated gene flow between forest patches (Ward et al., 2005), allowing more effective responses to change (Bhagwat et al., 2008; Thuiller et al., 2008). Mediterranean and other mountainous regions, where strongly contrasted topog- raphy on a meso-or micro-geographic scale prevail, may prove to be amongst the few biomes where climate change velocity will not outpace migration rates (Loarie et al., 2009), provided that land use change and man-made habitat fragmentation does not limit natural migration processes.

Abundant seed production is needed for efficient migration (and local adaptation, see Section 3.1). Predicting how climate change modifies tree fecundity remains a formidable challenge, however, because flowering phenology and seed production are regulated by complex endogenous (e.g., hormonal) and exogenous (e.g. climate) factors that are not completely understood yet. Selås et al. (2002), for example, indicated that spruce seed production in Norway is subject to a negative autocorrelation that lags by 1 year, i.e., good seed years (mast years) are preceded by low seed years, a phenomenon common to many trees. These authors found that seed production during mast years was directly related to higher temperatures in the previous spring and summer, late spring frost and summer precipitation of the last 2 years. On the other hand, more recently, Kelly et al. (2013), analysing extensive data sets from five plant families, found that a warm spring or summer in the previous year had a low predictive ability for seed production. Kelly et al. (2013) developed a model for the prediction of seed production that was based on temperature differentials over sev- eral seasons. They concluded that mast seeding will be unaffected

by gradual increases in mean temperature, because this will have little effect on the temperature differential over multiple years. Instead, yearly climatic variability may determine the amount of seed produced. This model was recently found to be an accurate predictor of acorn production in valley oak, Quercus lobata Jeps (Pearse et al., 2014).

Increased mortality under climate change reduces tree density (especially at the receding edge), which will also affect the quan- tity (and genetic quality) of seed crops (Restoux et al., 2008). Changes in climate may also result in asynchronies between flower development and pollinator availability which, for trees that depend on animal vectors, may reduce the seed crop (Dawson et al., 2011), at least until new mutualistic relations are established between trees and pollinators (see Section 4.1). Many tropical tree species that are pollinated by insects, birds, or bats may be affected (Hegland et al., 2009).

3.3. The role of phenotypic plasticity

Phenotypic plasticity is defined as the capacity of a particular genotype to express different phenotypes under different environ- mental conditions (de Jong, 2005; Pigliucci and Murren, 2003). The concept is often extended to populations and species, with ‘plastic’ trees those with flexible morphology and physiology that grow at least reasonably well under a range of different environmental stresses without genetic change (Gienapp et al., 2008). A degree of phenotypic plasticity is found in most trees (Piersma and Drent, 2003; Rehfeldt et al., 2001; Valladares et al., 2005), but var- ies substantially amongst and within species (Aitken et al., 2008; Bouvarel, 1960; Skrøppa et al., 2010). Even in species with very lit- tle genetic diversity, such as Pinus pinea L. (Vendramin et al., 2008), strong phenotypic plasticity is expressed for growth-related traits, which may have helped the species colonise new environments (Mutke et al., 2010).

At least in the short term, high plasticity is likely to favour tree survival under changing environmental conditions, although trade-offs between traits can be expected. As processes related to phenotypic plasticity may oppose those related to genetic adapta- tion, however, in the longer term, survival may not be favoured (Aitken et al., 2008). Since phenotypic plasticity has a heritable basis and may be selected for under changing environments (Nicotra et al., 2010), complex interactions between traits are pos- sible, depending on the magnitude and structure of change (Chevin et al., 2010). Selecting populations and genotypes that demonstrate good levels of phenotypic plasticity (based on multi-locational field trials and environmental data) may be an appropriate man- agement response to climate change for plantation forestry and agroforestry, especially for regions where greater variation in weather conditions is anticipated. Multi-site field trials sometimes reveal that trees have more plastic responses than would be expected based on their existing geographic distributions (e.g., Pinus radiata D. Don., Gautam et al., 2003).

Epigenetic phenomena (modification of DNA expression but not the nucleotide sequence, e.g., through DNA methylation, histone modification and mRNA regulation) may affect phenotypic plastic- ity and adaptive potential (Hedhly et al., 2008). Epigenetic effects caused by environmental stresses can be maintained across several generations and vary across populations and individuals (Bossdorf et al., 2008; Yakovlev et al., 2010). Since epigenetic modifications can be reversed, they can be considered as relatively ‘‘plastic’’, pro- viding for a rapid response to change while avoiding the need for additional genetic diversification (Lira-Medeiros et al., 2010). According to Aitken et al. (2008), the epigenome may provide a temporary buffer against climatic variability, providing time for the genome to ‘‘catch up’’ with change.

80 R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87

Epigenetic effects have been demonstrated in the phenology of bud set in Picea abies (L.) Karst. Progenies of this species whose embryos develop in warm environments are less cold hardy than those that develop at lower temperatures (Skrøppa and Johnsen, 2000; Johnsen et al., 2005, 2009). Similar effects have been observed in: progeny from Picea glauca and in P. glauca � P. engel- mannii (Parry ex Engelman.) (Webber et al., 2005); in Pinus sylves- tris L. (Dormling and Johnsen, 1992); and in Larix spp. (Greenwood and Hutchison, 1996). Epigenetic phenomena have also been hypothesised to explain the phenotypic plasticity of the genetically depauperate Pinus pinea (see earlier in this section, Vendramin et al., 2008). There is, however, a general lack of information on epigenetic effects in angiosperm trees (Rohde and Junttila, 2008).

Fig. 1. Forest transformation by natural disturbances. For fire-adapted tree species in British Columbia, Canada, elimination of the old forest canopy by stand-replacing fire triggers massive forest regeneration (top two panes). The bottom photograph illustrates massive mortality of lodgepole pine by mountain pine beetle in British Columbia. The diagram shows the removal of the mature lodgepole pine canopy by mountain pine beetle (left) and forest transformation to a different species not affected by beetle (right). Canopy mortality by disturbances creates enormous economic losses, but at the same time provides conditions for forest regeneration. This provides new opportunities for natural selection to operate, resulting in a new generation of trees better adapted to new climatic conditions, and which eventually will replace old canopies born under the climate of over one hundred years earlier. Photographs: Canadian Forest Service.

4. Responses of tree populations to catastrophic biotic and abiotic disturbances

Tree populations have developed mechanisms to respond to naturally occurring disturbances within their range. North Ameri- can conifers, for example, have adapted to outbreaks of the defoli- ating insect spruce budworm (Choristoneura fumiferana Clem.) that have recurred at periodic intervals (�every 35 years) at least since the middle of the Holocene, 6000 years ago (Simard et al., 2011). Climate change may however cause range expansions in herbivo- rous insects (Murdock et al., 2013) and in diseases, causing increased mortality in non-adapted populations. This is illustrated by whitebark pine, where a warming climate has increased the access of stands to native bark beetles that are now able to reach higher elevations, resulting in high mortality due to low defenses in trees that have had little previous contact with this beetle (Raffa et al., 2013). Recent modelling supports the view that large areas of current whitebark pine habitat are likely to become cli- matically unsuitable over the coming decades (McLane and Aitken, 2012). Increasingly, warm winters and earlier springs, which cause greater drying of soils and forest fuels, are also pre- dicted to increase the number of large wildfires and the total area burned in temperate and some tropical forests (Malhi et al., 2009).

Tree populations respond to abrupt, non-linear environmental changes through the mechanisms already outlined: natural selec- tion favours genotypes with increased tolerance or resistance to disturbances, and phenotypic plasticity plays a role. It is well known, for example, that populations of Pinus contorta Dougl. ex Loud. and P. banksiana Lamb. from parts of North America more prone to natural fires have a higher proportion of serotinous cones than those from elsewhere. Serotinous cones remain tightly closed until a hot fire has destroyed standing trees, then releasing seed to initiate rapid post-fire regeneration. There is also evidence that in the Mediterranean ecosystem, fire selects tree species and individ- uals with a particular combination of functional traits including serotiny, thick bark and high water use efficiency (Fady, 2012; Budde et al., 2014). Populations of many Mediterranean plants per- sist after fire due to their capacity to form a resistant seed bank (Lamont et al., 1991; Keeley and Fotheringham, 2000). Although many tree species that grow in semi-arid regions have developed mechanisms that confer a degree of resistance to periodic fires, this may not be the case in more humid forests, where increased fire frequency due to climate change may eliminate fire-sensitive spe- cies (Verdu and Pausas, 2007). Regions that newly experience reg- ular wildfires may evolve in close association with fire as the main driver, with rapid species and genotype transitions from fire-sensi- tive to fire-resistant (i.e., a rapid change in micro-evolutionary pat- tern may occur).

Large stand-replacing fires or widespread insect and disease outbreaks, although often resulting in large economic losses, do eliminate forests that were adapted to old climatic conditions

and provide a ‘fresh start’ with new regeneration opportunities (Fig. 1). Such successional forests will eventually adapt to new cli- mate through natural selection, particularly at the seedling stage. Selective shifts in traits related to fire resistance may, however, have negative effects on economically important associated traits. For example, Schwilk and Ackerly (2001) indicated that trees that embrace fire as a species survival strategy are more likely to favour traits such as short height, flammable foliage and no self-pruning.

4.1. Co-evolution and biotic disturbances

‘Co-evolution’ describes a situation where two (or more) species reciprocally affect each other’s evolution (Janzen, 1980; Pimentel, 1961), such as the classic case of host-pathogen interaction, where changes in R-gene resistance in the host lead to corresponding changes in v-gene virulence in the pathogen, triggering further rounds of change in one and then the other (Person, 1966). In trees, such gene-for-gene relationships have, for example, been found in a number of North American white pines in their interaction with blister rust (Kinloch, 2003; Kinloch and Dupper, 2002). Further important examples of co-evolution in trees include interactions with herbivores and pollinators. In the former case, a number of constitutive and induced defence systems, both mechanical defences (e.g., resin canals, sclereid cells and thorns) as well as chemical defences (e.g., the production of toxic phenols and

R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87 81

terpenoids), have evolved in response to herbivory (Alfaro et al., 2002; Cooper and Owen-Smith, 1986; Franceschi et al., 2005). Insects and pathogens have developed mechanisms to de-activate these defences and even utilize them for their own benefit; for example, some insects use tree terpenes as precursors for their communication pheromones (Erbilgin et al., 2014) or incorporate them into their own defence systems (Higginson et al., 2012).

The relationships between trees and associated herbivores, par- asites and pollinators are strongly influenced by environmental factors. It is well known, for example, that drought stress reduces the ability of conifers to defend against bark beetles due to changes in plant defences (Ayres and Lombardero, 2000; Safranyik and Carroll, 2006). Climate change-mediated insect epidemics are already observed in Canada, where the mountain pine beetle has had severe economic consequences for forestry (Konkin and Hopkins, 2009; Fig. 1). In the Canadian province of British Colum- bia, an outbreak of mountain pine beetle, which began in the early part of the last decade and is only now (2014) abating, attacked more than 13 million hectares of Pinus contorta forests. The cause of this sustained outbreak is believed to have been a long series of unusually warm winters (Safranyik and Carroll, 2006). As with fire, however, large scale mortality does provide an opportunity for wide-scale regeneration (Axelson et al., 2010) and hence more rapid adaptation to changing climate.

Overall, pest-resistant tree genotypes occur more frequently in areas where climate is most favourable to the insect and the lowest resistance levels are found where the insect is absent (Alfaro et al., 2008). As global environmental changes influence the distribution of the insect, an associated adaptive response by the tree will be required.

The mutualistic relationship between trees and insect or verte- brate pollinators is of considerable interest in the context of cli- mate change. The current view of ecologists recognizes that plant–pollinator relationships are not always a strict one-on-one co-evolutionary process; instead, there are many plant pollinator systems where diverse pollinator assemblages can lead to the maintenance of pollination services, plant reproduction and persis- tence, and relationships change over time and space (Burkle and Alarcón, 2011 and references therein). Under climate change, trees may be able to rely on new pollinators that shift their attention to them. According to Burkle and Alarcón (2011), the inherent plastic- ity of plant–pollinator interactions suggests that many species should be able to persist by responding to environmental changes quickly, even though their mutualistic partners may be different.

4.2. Responses to alien invasive species

Under climate change, FGR are likely to be increasingly threa- tened by alien invasive species i.e., more competitive trees, fungal and other diseases and herbivores that do not occur naturally in their local ecosystems, and to which they lack adequate defenses. The acceleration of global trade has increased the likelihood of cross-continental introductions of alien species, which may become more widely established in new ecosystem niches created by global warming (Koskela et al., 2009; Koskela et al., 2014, this special issue; Peterson et al., 2008). When forest ecosystems are already disturbed by other anthropogenic activities, they may have little resistance to invasive species, especially when climate change is also considered, with extreme results possible (Moore, 2005). There are, for example, numerous cases of exotic trees invading forest ecosystems (Richardson, 1998). Lack of resistance to alien invaders, especially in temperate forests, is more severe when the number of endemic species found in them is reduced (Petit et al., 2004; Simberloff et al., 2002). The consequences of exotic pest invasions may be a catastrophic elimination of FGR, such as the cases of chestnut blight and white pine blister rust (Kinloch,

2003). At a provenance level, exotic introductions may result in hybridisation and out-breeding depression in local tree popula- tions already stressed by climate change, but, more positively, hybridisation may also introduce the new genetic variation required by trees to adapt to novel environments (Hoffmann and Sgro, 2011).

5. FGR-based strategies to respond to climate change

Isbell et al. (2011) stated that ‘‘many species are needed to maintain multiple functions at multiple times and places in a changing world’’. From a forest management perspective, adapting to climate change requires the adoption of the ‘‘precautionary prin- ciple’’ and maintaining options in the form of inter- and intra-spe- cific diversity (a form of insurance policy) (UNESCO, 2005). This should increase the resilience of natural and planted forests under environmental variability, especially if the component parts of sys- tems and their interactions respond differently to disturbances (Fleming et al., 2011; Kindt et al., 2006; Steffan-Dewenter et al., 2007). As climate change progresses, poorly-performing trees will be naturally replaced by alternatives that are better suited to new conditions, altering the relative abundance of different species and genotypes in landscapes (Jump and Peñuelas, 2005). As resil- ience rests on the maintenance of genetic, species and ecosystem diversity, management strategies should support diversification at all three levels (Millar et al., 2007; Jump et al., 2008).

Although humans impacts on forests over time have often involved (genetic) resource depletion (e.g., in the Mediterranean, Fady et al., 2008), silvicultural interventions can provide opportu- nities to manage forests better under climate change. Several of the management interventions required to support natural and planted production forests in the context of climate change can be considered as good practice under ‘business as usual’ scenarios (Guariguata et al., 2012). Forest managers sometimes question, however, whether interventions specifically formulated to respond to climate change are economically justified, as tropical foresters are likely to consider commercial agriculture and unplanned log- ging more important production threats (Guariguata et al., 2012). Interviews of foresters in Europe indicate that they are sometimes similarly ambivalent in implementing specific management responses to climate change, partly reflecting uncertainties in cli- mate impacts and appropriate responses (Milad et al., 2013).

As part of the toolkit that foresters can use to adapt forests to climate change, the distribution of FGR and their silviculture can be modified in space and time (Sagnard et al., 2011; Lefèvre et al., 2013). To date, few countries have however taken practical steps to reduce the risk of FGR loss due to climate change. Relevant steps are usually only indirectly incorporated into action plans for forest management under climate change. In France, for example, FGR are not explicitly mentioned in the national adaptation strat- egy (ONERC, 2007). They are, however, part of the action plan for forests, one of the sectors included in the national strategy for bio- diversity, where recommendations for their conservation and sus- tainable use are explicitly mentioned (MAP, 2006).

5.1. Assisted migration

Assisted migration involves human movement of tree seed and seedlings from current locations to sites modelled to experience analogous environmental conditions in the future (Guariguata et al., 2008; McLachlan et al., 2007). Such movements may be lat- itudinal, longitudinal or altitudinal, and are designed to reduce extinction risks for those species not able to naturally migrate quickly enough, and to maintain forest productivity (Heller and Zavaleta, 2009; Marris, 2009; Millar et al., 2007). Assisted

82 R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87

migration may be undertaken over long distances, or just beyond the current range limit of particular genotypes and populations, or within the existing range (Winder et al., 2011). A gradual form of assisted migration could consist of reforestation of harvested sites with seed from adjacent locations likely to be better adapted to the planting site under future climate (e.g., in the Northern hemisphere, using seed from sources to the south; in mountainous regions using seed from lower elevations).

Aubin et al. (2011) and Winder et al. (2011) reviewed the pros and cons of the assisted migration approach. One problem is that the selection between different global climate models (GCMs) and the methods for downscaling to detailed geographic levels are still areas of active research and thereby introduce uncertainty in modelling, especially for marginal environments (Fowler et al., 2007). Clearly, areas of high probability for a given future environ- ment, based on ensemble forecasting across GCM (and across the various statistical models available for determining species distri- butions) should be priorities for action (Rehfeldt et al., 2012).

Much uncertainty is also due to the unknown future trajectories of greenhouse gas emissions in the longer term, as these will depend on technological developments that increase or decrease emissions (IPCC, 2011). For more immediate future scenarios, how- ever, the variation amongst models is small; for México, for the decade centered on the year 2030, for example, it is only about ±0.2 �C of mean annual temperature (Sáenz-Romero et al., 2010). Another difficulty in modelling is that the current distributions of tree species, which form the basic input data for determining likely future distributions, are often not well known (McLachlan et al., 2007; Rehfeldt and Jaquish, 2010), especially in the tropics, where sometimes complex topographies and high biodiversity paradoxi- cally make accurate predictions even more urgent. In the light of uncertainties in modelling, the United Kingdom’s Forestry Commission (2011) considers risk minimisation as the best approach, by maintaining existing genetic variation, promoting migration, encouraging natural regeneration and supporting prov- enance mixing in plantations (Hubert and Cottrell, 2007).

As already noted (see Section 4.2), interventions that involve moving tree species into entirely new areas is hotly debated because of potential disturbances to indigenous flora and fauna. There are also numerous commercial forestry examples where the introduction of ill-adapted genetic resources has resulted in massive production failures. For example, 30,000 ha of Pinus pinas- ter Aiton plantations were destroyed in the Landes region of France during the winter of 1984 to 1985 following the introduction of non-frost-resistant material from the Iberian Peninsula (Timbal et al., 2005). Careful thought to all environmental factors should therefore be given before climate-related assisted migrations are undertaken. In mountain regions, upwards associated transloca- tions may not be an option if populations are already at or near the summit (translocation must then be to different mountains), or if edaphic conditions are unsuitable (Lauer, 1973). Certainly, the establishment of viable populations at extremely high altitudes would be very challenging (Sáenz-Romero et al., 2010, 2012).

Another challenge to assisted migration that is specific to long- living perennials is that, where climate is changing quickly, large differences in conditions may be observed over an individual tree’s lifespan. To find species or genotypes well adapted to conditions at establishment and at productive maturity (e.g., for some species, perhaps a century later) may therefore be difficult. In order to achieve a proper balance, the interval to production/maturity needs to be considered, and multiple stepped translocations over time may be required (Soto-Correa et al., 2012). In addition, changes to pest outbreak risk could simultaneously occur as a result of climate change, and this should be factored into assisted migration decisions (Murdock et al., 2013).

Another useful approach is to conduct assisted migration on assemblages of species with positive interactions that reduce cli- mate risks. For example, a ‘‘first-stage’’ species may be planted as a nurse crop to provide protection from temperature extremes for a second tree. Such an approach has been applied to Abies reli- giosa (Kunth) Schltdl. et Cham., using the leguminous shrub Lupi- nus elegans Kunth as a nurse plant for seedlings (Blanco-García et al., 2011). Within species, assisted gene flow, where genes are exchanged between populations by moving individuals or gametes, has also the potential to control and reduce mal-adapta- tion (Aitken and Whitlock, 2013).

5.2. Selection and breeding

Climate change-related traits including plasticity and adapta- tion to increased drought need to be incorporated more actively into breeding programs (IUFRO, 2006). Many existing provenance trials were established before the need to respond to large scale anthropogenic environmental change was considered an impor- tant research issue and the traits measured have therefore often not been the most important ones from this perspective. Neverthe- less, information from old trials can be reinterpreted in the context of climate threats (Aitken et al., 2008; Alberto et al., 2013). New tri- als established to assess explicit responses to climate change are being established in a number of countries (see, e.g., http://treeb- reedex.eu/).

Traits needed to respond to different climatic conditions not often considered previously in breeding include:

� Pest and disease resistance: As noted above (Section 4), climate- change-mediated increases in pest and disease attack are a cru- cial issue in commercial forestry. To date, one of the most extensive programmes to develop trees with resistance to insect pests in temperate regions is in British Columbia (Alfaro et al., 2013; King and Alfaro, 2009). Using a conventional breeding approach, Picea sitchensis genotypes with resistance to the white pine weevil were screened and deployed in reforesta- tion programmes (Alfaro et al., 2013; Moreira et al., 2012). Such traits may be controlled by only a few loci as a result of gene- for-gene co-evolution (sensu Thompson and Burdon, 1992), as already described (Section 4.1), making breeding easier. � Drought resistance: For many tree species, such as in the Medi-

terranean and parts of the tropics, altered moisture regimes will be of greater concern than temperature changes (Santos-del- Blanco et al., 2013). Drought stress induces a range of physio- logical and biochemical responses in plants and an assortment of genes with diverse functions are induced or repressed in organ-specific changes (Kreps et al., 2002; Shinozaki and Yamaguchi-Shinozaki, 2007), which may make breeding more difficult. Perdiguero et al. (2013), for example, using microarray analysis, detected that up to 113 genes were significantly induced by drought in two Mediterranean pine species. Spe- cies-dependent features shape the transcriptome response; for example, almost none of the 27 genes reliably responsive to water stress in Arabidopsis thaliana (L.) Heynh., differentially regulated under drought in poplar and pine (Polle et al., 2006). Candidate genes for drought tolerance include those involved in the synthesis of abscisic acid, transcriptional regula- tors of drought-inducible pathways, and late embryogenesis abundant proteins; shifts at such loci have been linked to global warming (Hoffmann and Willi, 2008). � Fire resistance/tolerance: Since fire incidence and severity will

increase in many regions under climate change, breeding for features such as serotiny, thicker bark and higher water use effi- ciency may all be required (e.g., Jump et al., 2008).

R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87 83

� Cyclone resistance/salt tolerance: Rising sea levels and an increase in the frequency of storms have the potential to wreak heavy damage on coastal forests, with low elevation islands at particular risk. Differential abilities to withstand storms and salinity are found more commonly amongst, rather than within, species, but the possibility of intra-specific selection should be further explored. Increasing storm frequency in the Pacific due to climate change has led to efforts to identify cyclone-resistant species such as Endospermum medullosum L.S.Sm. for large-scale planting. In Vanuatu, for example, the establishment of 20,000 ha of plantations of this species is planned over the next 20 years. � Phenotypic plasticity: Important but generally poorly under-

stood, the plasticity of particular tree species and populations is vital for responding to climate change, and can be studied in common garden tests (Rehfeldt et al., 1999, 2002; Vitasse et al., 2010). Plasticity across environments can be quantified and response functions for particular populations generated, which describe the change in a trait as a function of the transfer distance or the change in an environmental factor (Rehfeldt et al., 1999, 2002). Populations vary in their response functions: in Pinus contorta, for example, some populations have a high growth rate over a much wider range of climatic conditions than others do (Wang et al., 2006).

At a strategic level, the feasibility of classical breeding approaches as a response to climate change needs to be consid- ered. Yanchuk and Allard (2009) reviewed 260 activities for pest and disease breeding in trees, and found relatively few examples where resistant or tolerant material had been developed and deployed operationally. They concluded that future programs to tackle increased pest and disease incidence caused by rapid cli- mate change were likely to have limited success if they relied on conventional breeding approaches (but see the case in this section above on P. sitchensis and white pine weevil). The long life cycle, large size, and (generally) poorly characterised genetics of trees all make breeding responses to climate change more costly and slower than for annual species. Indeed, in the neo-tropics, Guariguata et al. (2008) were unable to identify any changes to industrial tree breeding approaches that were aimed specifically to this end.

A breeding response to climate change requires agile and accu- rate methods that can deliver the needed genetic improvements but with substantially reduced time and resources. More than ever, breeding programs need to target several traits simultaneously, while conserving large genetic bases for unpredictable adaptation needs (Eriksson et al., 1993). The recent development of Next Gen- eration Sequencing and Genotyping by Sequencing approaches offers an unlimited number of genetic markers, creating opportuni- ties for new developments. These include pedigree reconstruction, so the breeding phase of tree improvement can be by-passed (e.g., ‘‘Breeding Without Breeding’’; El-Kassaby and Lstiburek, 2009), with additional simplifications in testing (El-Kassaby et al., 2011); the use of pedigree-free models that can deliver genetic assessments with unprecedented precision, with the added advan- tage of applicability to unstructured natural populations (El- Kassaby et al., 2012; Klápšte et al., 2013; Korecký et al., 2013); and selection methods that utilize information from the entire gen- ome (Meuwissen et al., 2001). Additionally, new methods for bul- king-up and delivering the improvements of breeding are needed for commercially important species, as traditional methods (e.g., seed orchards) are slow. Renewed efforts are needed for improving and simplifying vegetative propagation methods, starting from the conventional production of rooted-cuttings through to somatic embryogenesis.

6. Conclusion

Forest resilience and ecosystem stability are required to ensure the future flow of ecosystem services over space and time in the support of world societies (FAO, 2010). These depend on maintain- ing genetic diversity, functional species diversity and ecosystem diversity (beta diversity) across forest landscapes and over time. Only adapted and adaptable genetic material will, for example, efficiently mitigate global carbon emissions. From a forest manage- ment perspective, adapting to climate change (and mitigating its effects) requires the adoption of the ‘‘precautionary principle’’ and maintaining options including intra-specific diversity (UNESCO, 2005). Tree species generally contain high genetic diver- sity in many of the traits and genes analysed, which supports this principle (Jump et al., 2008), but the potential of trees to respond to climate change should not be over-estimated (Nepstad et al., 2007).

In determining human responses to climate change for the for- estry sector, there needs to be good supporting evidence if the active engagement of forest managers is to be obtained to support management interventions that proceed beyond good ‘business as usual’ practice (Guariguata et al., 2012; Milad et al., 2013). This evi- dence includes reliable science-based estimates of risks and the benefits of management for the mitigation of climate change impacts. Responses based on assisted migration need to include the consideration of all environmental factors, as the consequences of only partial consideration (response to a single or a few variables only) may be catastrophic (cf. Timbal et al., 2005), with such mea- sures then losing credibility with forest managers. For assisted migration, modelling should consider potential damage by biotic and abiotic disturbances; for example, potential increases in pest and fire risk as a result of stress in the new area (Murdock et al., 2013).

Assisted migration responses to climate change that are based on greater dependency on the trans-national exchange of forest genetic resources require an appropriate policy and legislative environment to support transfer, including by the harmonisation of phytosanitary requirements, as noted by Koskela et al. (2009). At a national level, policies defining seed zones will need to be modified to allow the assisted migration of genetic material within nations. Countries developing national forestry action plans should also be encouraged to specifically include genetic level responses to climate change in their plans, which has sometimes, but not always, been the case to date (Hubert and Cottrell, 2007).

Designing proper responses to climate change requires a greater understanding of the extent of phenotypic plasticity in trees for important traits, the adaptive significance of plasticity, the differ- ences in phenotypic plasticity amongst different genetic levels (genotypes, families, populations, etc.), and the trade-offs between plastic and adaptive responses (Aitken et al., 2008). Also required is further research on epigenetic effects, especially in angiosperm trees (Rohde and Junttila, 2008). Plastic and adaptive responses can be studied in multi-locational common garden experiments that specifically consider climate-related traits in measurement and design (Rehfeldt et al., 2002; Vitasse et al., 2010). For ani- mal-pollinated species in particular, research is also needed on the effects of climate change on tree reproductive capacity, such as how elevated temperatures may affect mutualisms with pollin- ators, and how the changed availability of mutualistic partners influences the persistence of interacting species (Hegland et al., 2009).

As in previous climate change episodes, forest genetic resources will recombine to produce new variants, which through natural or assisted selection will produce the genotypes required to continue providing the ecosystem services that societies need from forests.

84 R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87

But, as climate change progresses it will be important to monitor the adaptation of trees, stands and ecosystems, and to intervene with efforts to support adaptation where needed.

Acknowledgements

For compiling and incorporating suggestions and changes from the contributors, checking citations and references, and general English copyediting of this manuscript, the authors would like to thank biologist and editor Maggie Paquet, Port Alberni, British Columbia, Canada.

References

Aitken, S.N., Whitlock, M.C., 2013. Assisted gene flow to facilitate local adaptation to climate change. Annu. Rev. Ecol. Evol. Syst. 44, 367–388.

Aitken, S.N., Yeaman, S., Holliday, J., Wang, T., Curtis-McLane, S., 2008. Adaptation, migration or extirpation: climate change outcomes for tree populations. Evol. Appl. 1, 95–111.

Alberto, F.J., Aitken, S.N., Alia, R., Gonzalez-Martinez, S.C., Hänninen, H., Kremer, A., Lefèvre, F., Lenormand, T., Yeaman, S., Whetten, R., Savolainen, O., 2013. Potential for evolutionary responses to climate change – evidence from tree populations. Glob. Change Biol. 19, 1645–1661.

Alcamo, J.M., Floerke, M., Maerker, 2007. Future long-term changes in global water resources driven by socio-economic and climatic changes. Hydrol. Sci. J. 52, 247–275.

Alfaro, R.I., Borden, J.H., King, J.N., Tomlin, E.S., McIntosh, R.L., Bohlmann, J., 2002. Mechanisms of resistance in conifers against shoot infesting insects. In: Wagner, M.R., Clancy, K.M., Lieutier, F., Paine, T.D. (Eds.), Mechanisms and Deployment of Resistance in Trees to Insects. Kluwer Academic Press, Dordnecht, the Netherlands, pp. 101–126.

Alfaro, R.I., Axelson, J.N., Hawkes, B., 2008. Mountain pine beetle increases the complexity of fire-origin lodgepole pine stands in British Columbia, Canada. J. Ecosyst. Manage. 9, 1–5.

Alfaro, R.I., Hantula, J., Carroll, A., Battisti, A., Fleming, R., Woods, A., Hennon, P.E., Lanfranco, D., Ramos, M., Müller, M., Lilja, A., Francis, D., 2010. Forest Health in a changing environment. In: Mery, G., Katila, P., Galloway, G., Alfaro, R.I., Kanninen, M., Lobovikov, M., Varjo, J. (Eds.), Forests and Society. Responding to Global Drivers of Change. International Union of Forest Research Organizations (IUFRO), Vienna, Austria, pp. 113–134.

Alfaro, R.I., King, J.N., vanAkker, L., 2013. Delivering Sitka spruce with resistance against white pine weevil in British Columbia, Canada. Forest. Chron. 89, 235– 245.

Alleaume-Benharira, M., Pen, I.R., Ronce, O., 2006. Geographical patterns of adaptation within a species’ range: interactions between drift and gene flow. J. Evol. Biol. 19, 203–215.

Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier, M., Kizberger, T., Rigling, A., Breshears, D.D., Hogg, E.H., Gonzalez, P., Fensham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.H., Allard, G., Running, S.W., Semerci, A., Cobb, N., 2010. A global overview of drought- and heat-induced tree mortality reveals emerging climate change risks for forests. For. Ecol. Manage. 259, 660–684.

Amm, A., Pichot, C., Dreyfus, P., Davi, H., Fady, B., 2012. Improving the estimation of landscape scale seed dispersal by integrating seedling recruitment. Ann. For. Sci..

Aubin, I., Garbe, C.M., Colombo, S., Drever, C.R., McKenney, D.W., Messier, C., Pedlar, J., Saner, M.A., 2011. Why we disagree about assisted migration: ethical implications of a key debate regarding the future of Canada’s forests. Forest. Chron. 87, 755–765.

Axelson, J.N., Alfaro, R.I., Hawkes, B.C., 2010. Changes in stand structure in uneven- aged lodgepole pine stands impacted by mountain pine beetle epidemics and fires in central British Columbia. Forest. Chron. 86, 87–99.

Ayres, M.P., Lombardero, M.J., 2000. Assessing the consequences of global change for forest disturbance from herbivores and pathogens. Sci. Total Environ. 262, 263–286.

Barbour, R.C., O’Reilly-Wapstra, J.M., DeLittle, D.W., Jordan, G.J., Steane, D.A., Humphreys, J.R., Bailey, J.K., Whitham, T.G., Potts, B.M., 2009. A geographic mosaic of genetic variation within a foundation tree species and its community- level consequences. Ecology 90, 1762–1772.

Beaulieu, J., Rainville, A., 2005. Adaptation to climate change: genetic variation is both a short- and a long-term solution. Forest. Chron. 81, 704–709.

Bhagwat, S.A., Willis, K.J., Birks, H.J.B., Whittaker, R.J., 2008. Agroforestry: a refuge for tropical biodiversity? Trends Ecol. Evol. 23, 261–267.

Blanco-García, A., Sáenz-Romero, C., Martorell, C., Alvarado-Sosa, P., Lindig- Cisneros, R., 2011. Nurse plant and mulching effects on three conifer species in a Mexican temperate forest. Ecol. Eng. 37, 994–998.

Bossdorf, O., Richards, C.L., Pigliucci, M., 2008. Epigenetics for ecologists. Ecology 11, 106–115.

Bosselmann, A.S., Jacobsen, J.B., Kjær, E.D., Thorsen, B., 2008. Climate Change, Uncertainty and the Economic Value of Genetic Diversity: A Pilot Study on Methodologies. Landscape, Hørsholm, Denmark.

Bouvarel, P., 1960. Note sur la résistance au froid de quelques provenances de pin maritime. Rev. Forest. Fran. 7, 495.

Breshears, D.D., Cobb, N.S., Rich, P.M., Price, K.P., Allen, C.D., Balice, R.G., 2005. Regional vegetation die-off in response to global-change-type drought. Proc. Natl. Acad. Sci. 102, 15144–15148.

Brewer, S., Cheddadi, R., Beaulieu, J.L., deReille, M., 2002. The spread of deciduous Quercus throughout Europe since the last glacial period. For. Ecol. Manage. 156, 27–48.

Bridle, J.R., Polechova, J., Kawata, M., Butlin, R.K., 2010. Why is adaptation prevented at ecological margins? New insights from individual-based simulations. Ecol. Lett. 13, 484–494.

Budde, K.B., Heuertz, M., Hernández-Serrano, A., Pausas, J.G., Vendramin, G.G., Verdú, M., González-Martínez, S.C., 2014. In situ genetic association for serotiny, a fire-related trait, in Mediterranean maritime pine (Pinus pinaster Aiton). New Phytol. 201, 230–241.

Burkle, L.A., Alarcón, R., 2011. The future of plant–pollinator diversity, Understanding interaction networks across time, space, and global change. American J. Botany 98, 528–538.

Campbell, E.M., Antos, J.A., 2000. Distribution and severity of White pine blister rust and mountain pine beetle on whitebark pine in British Columbia. Can. J. For. Res. 30, 1051–1059.

Chevin, L.-M., Lande, R., Mace, G.M., 2010. Adaptation, plasticity, and extinction in a changing environment: towards a predictive theory. PLoS Biol. 8 (4), e1000357. http://dx.doi.org/10.1371/journal.pbio.1000357.

Clark, J.S., Macklin, E., Wood, L., 1998. Stages and spatial scales of recruitment limitation in southern Appalachian forests. Ecol. Monogr. 68, 213–235.

Cooper, S.M., Owen-Smith, N., 1986. Effects of plant spinescence on large mammalian herbivores. Oecologia 68, 446–455.

Davis, M.B., Shaw, R.G., 2001. Range shifts and adaptive responses to quaternary climate change. Science 292, 673–679.

Dawson, I.K., Vinceti, B., Weber, J.C., Neufeldt, H., Russell, J., Lengkeek, A.G., Kalinganire, A., Kindt, R., Lillesø, J.-P.B., Roshetko, J., Jamnadass, R., 2011. Climate change and tree genetic resource management: maintaining and enhancing the productivity and value of smallholder tropical agroforestry landscapes. A review. Agroforest. Syst. 81, 67–78.

Dawson, I.K., Leakey, R., Clement, C.R., Weber, J.C., Cornelius, J.P., Roshetko, J.M., Vinceti, B., Kalinganire, A., Masters, E., Jamnadass, R., 2014. The management of tree genetic resources and the livelihoods of rural communities in the tropics: non-timber forest products, smallholder agroforestry practices and tree commodity crops. For. Ecol. Manage. 333, 9–21.

de Jong, G., 2005. Evolution of phenotypic plasticity: patterns of plasticity and the emergence of ecotypes. New Phytol. 166, 101–117.

Dormling, I., Johnsen, Ø., 1992. Effects of the parental environment on full-sib families of Pinus sylvestris. Can. J. For. Res. 22, 88–100.

El-Kassaby, Y.A., Lstiburek, M., 2009. Breeding without breeding. Genet. Res. 91, 111–120.

El-Kassaby, Y.A., Fashler, A.M.K., Crown, M., 1989. Variation in fruitfulness in a Douglas-fir seed orchard and its effect on crop-management decisions. Silvae Genet. 38, 3–4.

El-Kassaby, Y.A., Cappa, E.P., Liewlaksaneeyanawin, C., Klápšt, J., Lstiburek, M., 2011. Breeding without breeding: is a complete pedigree necessary for efficient breeding? PLoS ONE 6, e25737. http://dx.doi.org/10.1371/ journal.pone.0025737.

El-Kassaby, Y.A., Klápšt, J., Guy, R.D., 2012. Breeding without Breeding: selection using the genomic best linear unbiased predictor method (GBLUP). New Forest. 43, 631–637.

Erbilgin, N., Ma, C., Whitehouse, C., Shan, B., Najar, A., Evenden, M., 2014. Chemical similarity between historical and novel host plants promotes range and host expansion of the mountain pine beetle in a naïve host ecosystem. New Phytol. 201, 940–950.

Eriksson, G., Namkoong, G., Roberts, J.H., 1993. Dynamic gene conservation for uncertain futures. For. Ecol. Manage. 62, 15–37.

Fady, B., 2012. Biogeography of neutral genes and recent evolutionary history of pines in the Mediterranean Basin. Ann. Forest Sci. 69, 421–428.

Fady, B., Lefèvre, F., Vendramin, G.G., Ambert, A., Régnier, C., Bariteau, M., 2008. Genetic consequences of past climate and human impact on eastern Mediterranean Cedrus libani forests. Implications for their conservation. Conserv. Genet. 9, 85–95.

FAO, 1989. Plant Genetic Resources: Their Conservation In situ for Human Use. Food and Agriculture Organisation of the United Nations, Rome, Italy.

FAO, 2010. Global Forest Resources Assessment 2010. FAO Forestry Paper No. 163. Food and Agriculture Organisation of the United Nations, Rome, Italy.

Fleming, R., Kanowski, P., Brown, N., Jenik, J., Kahumbu, P., Plesnik, J., Devisscher, T., 2011. Emerging Perspectives on Forest Biodiversity. In: Emerging Issues in our Global Environment. Division of Early Warning and Assessment. United Nations Environment Programme, Nairobi, Kenya, pp. 46–59.

Forestry Commission, 2011. Forests and Climate Change. UK Forestry Standard Guidelines. Forestry Commission, Edinburgh, UK.

Fowler, H.J., Blenkinsop, S., Tebaldi, C., 2007. Review: linking climate change modelling to impacts studies: recent advances in downscaling techniques for hydrological modelling. Int. J. Climatol. 27, 1547–1578.

Franceschi, V.R., Krokene, P., Christiansen, E., Krekling, T., 2005. Anatomical and chemical defenses of conifer bark against bark beetles and other pests. New Phytol. 167, 353–376.

Futuyma, D.J., 2010. Evolutionary constraint and ecological consequences. Evolution 64, 1865–1884.

R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87 85

Gautam, M.K., Mead, D.J., Clinton, P.W., Chang, S.X., 2003. Biomass and morphology of Pinus radiata coarse root components in a sub-humid temperate silvopastoral system. For. Ecol. Manage. 177, 387–397.

Gienapp, P., Teplitsky, C., Alho, J.S., Mills, J.A., Merila, J., 2008. Climate change and evolution: disentangling environmental and genetic responses. Mol. Ecol. 17, 167–178.

Gomulkiewicz, R., Holt, R.D., 1995. When does evolution by natural selection prevent extinction? Evolution 49, 201–207.

Gomulkiewicz, R., Houle, D., 2009. Demographic and genetic constraints on evolution. Am. Nat. 174, 218–229.

Greenwood, M.S., Hutchison, K.W., 1996. Genetic after-effects of increased temperature in Larix. In: Hom, J., Birdsey, R., O’Brian, K. (Eds.), Proceedings of the 1995 Meeting of the Northern Global Change Program. USDA Forest Service report, Radnor, pp. 56–62.

Grivet, D., Sebastiani, F., Alia, R., Bataillon, T., Torre, S., 2011. Molecular footprints of local adaptation in two mediterranean conifers. Mol. Biol. Evol. 28, 101–116.

Guariguata, M.R., Cornelius, J.P., Locatelli, B., Forner, C., Sanchez-Azofeifa, G.A., 2008. Mitigation needs adaptation: tropical forestry and climate change. Mitig. Adapt. Strat. Glob. Change 13, 793–808.

Guariguata, M.R., Locatelli, B., Haupt, F., 2012. Adapting tropical production forests to global climate change: risk perceptions and actions. Int. Forest. Rev. 14, 27– 38.

Gunderson, L.H., Holling, C.S., 2002. Panarchy: understanding transformations in human and natural systems. Island Press, Washington, DC, USA.

Gutschick, V.P., BassiriRad, H., 2003. Extreme events as shaping physiology, ecology, and evolution of plants: toward a unified definition and evaluation of their consequences. New Phytol. 160, 21–42.

Hampe, A., Petit, R.J., 2005. Conserving biodiversity under climate change: the rear edge matters. Ecol. Lett. 8, 461–467.

Hamrick, J.L., 2004. Response of forest trees to global environmental changes. For. Ecol. Manage. 197, 323–335.

Harrington, R., Fleming, R.A., Woiwod, I.P., 2001. Climate change impacts on insect management and conservation in temperate regions: can they be predicted? Agric. Forest Entomol. 3, 233–240.

Hedhly, A., Hormaza, J.I., Herrero, M., 2008. Global warming and sexual plant reproduction. Trends Plant Sci. 14, 30–36.

Hegland, S.J., Nielsen, A., Lazaro, A., Bjerknes, A.-L., Totland, O., 2009. How does climate warming affect plant-pollinator interactions? Ecol. Lett. 12, 184–195.

Heller, N.E., Zavaleta, E.S., 2009. Biodiversity management in the face of climate change: a review of 22 years of recommendations. Biol. Conserv. 142, 14–32.

Higginson, A.D., Speed, M.P., Ruxton, G.D., 2012. Effects of anti-predator defence through toxin sequestration on use of alternative food microhabitats by small herbivores. J. Theor. Biol. 300, 368–375.

Hoffmann, A.A., Sgro, C.M., 2011. Climate change and evolutionary adaptation. Nature 470, 479–485.

Hoffmann, A.A., Willi, Y., 2008. Detecting genetic responses to environmental change. Nat. Rev. Genet. 9, 421–432.

Hubert, J., Cottrell, J., 2007. The role of forest genetic resources in helping British forests respond to climate change. Information Note. Forestry Commission, Edinburgh, UK.

IPCC. 2011. Summary for Policymakers, in: Intergovernmental Panel on Climate Change Special Report on Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation Field, C. B., Barros, V., Stocker, T.F., Qin, D., Dokken, D., Ebi, K.L., Mastrandrea, M. D., Mach, K. J., Plattner, G.-K., Allen, S. K., Tignor, M., Midgle, P. M. (Eds).

Isbell, F., Calcagno, V., Hector, A., Connally, J., Harpole, W.S., Reich, P.B., Scherer- Lorenzen, M., Schmid, B., Tilman, D., van Ruijven, J., Weigelt, A., Wilsey, B.J., Zavaleta, E.S., Loreau, M., 2011. High plant diversity is needed to maintain ecosystem services. Nature 477, 199–202.

IUFRO. 2006. Proceedings of the IUFRO Division 2 Joint Conference: Low Input Breeding and Conservation of Forest Genetic Resources: Antalya, Turkey, 9–13 October 2006. Edited by F. Isik. <http://www-genfys.slu.se/staff/dagl/Meetings/ Antalya06/Turkey06Proceedings.pdf>.

Janzen, D.H., 1980. When is it co-evolution? Evolution 34, 611–612. Jha, S., Dick, C.W., 2010. Native bees mediate long-distance pollen dispersal in a

shade coffee landscape mosaic. Proc. Natl. Acad. Sci. USA 107, 3760–13764. Johnsen, O., Fossdal, C.G., Nagy, N., Molmann, J., Daehlen, O.G., Skrøppa, T., 2005.

Climatic adaptation in Picea abies progenies is affected by the temperature during zygotic embryogenesis and seed maturation. Plant, Cell Environ. 28, 1090–1102.

Johnsen, Ø., Kvaalen, H., Yakovlev, I.A., Dæhlen, O.G., Fossdal, C.G., Skrøppa, T., 2009. An epigenetic memory from time of embryo development affects climatic adaptation in Norway spruce. In: Gusta, L.V., Wisniewski, M.E., Tanino, K.K. (Eds.), Plant Cold Hardiness. From the Laboratory to the Field. CABI, Wallingford, pp. 99–107.

Jump, A.S., Peñuelas, J., 2005. Running to stand still: adaptation and the response of plants to rapid climate change. Ecol. Lett. 8, 1010–1020.

Jump, A.S., Hunt, J.M., Martinez-Izquierdo, J.A., Peñuelas, J., 2006. Natural selection and climate change: temperature-linked spatial and temporal trends in gene frequency in Fagus sylvatica. Mol. Ecol. 15, 3469–3480.

Jump, A.S., Marchant, R., Peñuelas, J., 2008. Environmental change and the option value of genetic diversity. Trends Plant Sci. 14, 51–58.

Keeley, J.E., Fotheringham, C.J., 2000. Role of fire in regeneration from seeds. In: Fenner, M. (Ed.), Seeds: the Ecology of Regeneration in Plant Communities. CAB International, Wallingford, pp. 311–330.

Kellomäki, S., Rouvinen, I., Peltola, H., Strandman, H., Steinbrecher, R., 2001. Impact of global warming on the tree species composition of boreal forests in Finland and effects on emission of isoprenoids. Glob. Change Biol. 7, 531–544.

Kelly, D., Geldenhuis, A., James, A., Holland, E.P., Plank, M.J., Brockie, R.E., Cowan, P.E., Harper, G.A., Lee, W.G., Maitland, M.J., Mark, A.F., Mills, J.A., Wilson, P.R., Byrom, A.E., 2013. Of mast and mean: differential-temperature cue makes mast seeding insensitive to climate change. Ecol. Lett. 16, 90–98.

Kindt, R., Damme, P., Simons, J., Beeckma, H., 2006. Planning tree species diversification in Kenya based on differences in tree species composition between farms. I. Analysis of tree uses. Agroforest. Syst. 67, 215–228.

King, J., Alfaro, R.I., 2009. Developing Sitka spruce populations for resistance to the white pine weevil. BC Ministry of For. Range, For. Sci. Prog. Tech. Rep. 050. Victoria, BC, Canada.

Kinloch Jr., B.B., 2003. White pine blister rust in north america: past and prognosis. Phytopathology 93, 1044–1047. http://dx.doi.org/10.1094/ PHYTO.2003.93.8.1044.

Kinloch Jr., B.B., Dupper, G.E., 2002. Genetic specificity in the white pine–blister rust pathosystem. Phytopathology 92, 278–280.

Kirkpatrick, M., Barton, N.H., 1997. Evolution of a species’ range. Am. Nat. 150, 1–23. Klápště, J., Lstibůrek, M., El-Kassaby, Y.A., 2013. Estimates of genetic parameters and

breeding values from western larch open-pollinated families using marker- based relationship. Tree Genet. Genomes. 10, 241–249.

Konkin, D., Hopkins, K., 2009. Learning to deal with climate change and catastrophic forest disturbances. Unasylva 60, 17–23.

Korecký, J., Klápště, J., Lstibůrek, M., Kobliha, J., Nelson, D.C., El-Kassaby, Y.A., 2013. Comparison of genetic parameters from marker-based relationship, sibship, and combined models in Scots pine multi-site open-pollinated testsTree. Genet. Genomes. 9, 1227–1235.

Koskela, J., Vinceti, B., Dvorak, W., Bush, D., Dawson, I., Loo, J., Kjaer, E.D., Navarro, C., Padolina, C., Jamnadass, R., Graudal, L., Ramamonjisoa, L., 2009. The use and exchange of forest genetic resources for food and agriculture. Background Study Paper No. 44. The Commission on Genetic Resources for Food and Agriculture, Food and Agriculture Organization of the United Nations, Rome, Italy.

Koskela, J., Vinceti, B., Dvorak, W., Bush, D., Dawson, I.K., Loo, J., Kjaer, E.D., Navarro, C., Padolina, C., Bordács, S., Jamnadass, R., Graudal, L., Ramamonjisoa, L., 2014. Utilization and transfer of forest genetic resources: A global review. Fores. Ecol. Manage. 333, 22–34.

Koski, V., Skrøppa, T., Paule, L., Wolf, H., Turok, J., 1997. Technical guidelines for genetic conservation of Norway spruce (Picea abies (L.) Karst). International Plant Genetic Resources Institute, Rome, Italy. 42 p.

Kramer, A.T., Ison, J.L., Ashley, M.V., Howe, J.F., 2008. The paradox of forest fragmentation genetics. Conserv. Biol. 22, 878–885.

Kremer, A., Kleinschmit, J., Cottrell, J., Cundall, E.P., Deans, J.D., Ducousso, A., Konig, A.O., Lowe, A.J., Munro, R.C., Petit, R.J., Stephan, B.R., 2002. Is there a correlation between chloroplastic and nuclear divergence, or what are the roles of history and selection on genetic diversity in European oaks? For. Ecol. Manage. 156, 75– 87.

Kremer, A., Le Corre, V., Petit, R.J., Ducousso, A., 2010. Historical and contemporary dynamics of adaptive differentiation in European oaks. In: DeWoody, A., Bickham, J., Michler, C., Nichols, K., Rhodes, G., Woeste, K. (Eds.), Molecular Approaches in Natural Resource Conservation. Cambridge University Press.

Kremer, A., Ronce, O., Robledo-Arnuncio, J.J., Guillaume, F., Bohrer, G., Nathan, R., Bridle, J.R., Gomulkiewicz, R., Klein, E.K., Ritland, K., Kuparinen, A., Gerber, S., Schueler, S., 2012. Long-distance gene flow and adaptation of forest trees to rapid climate change. Ecol. Lett. 15, 378–392.

Kreps, J.A., Wu, Y.J., Chang, H.S., Zhu, T., Wang, X., Harper, J.F., 2002. Transcriptome changes for Arabidopsis in response to salt, osmotic, and cold stress. Plant Physiol. 130, 2129–2141.

Lamont, B.B., Le Maitre, D.C., Cowling, R.M., Enright, N.J., 1991. Canopy seed storage in woody plants. Bot. Rev. 57, 277–317.

Lauer, W., 1973. The altitudinal belts of the vegetation in the central Mexican highlands and their climatic conditions. Arct. Alp. Res. 5, 99–113.

Le Corre, V., Kremer, A., 2003. Genetic variability at neutral markers, quantitative trait loci and trait in a subdivided population under selection. Genetics 164, 1205–1219.

Lefèvre, F., Boivin, T., Bontemps, A., Courbet, F., Davi, H., Durand-Gillmann, M., Fady, B., Gauzere, J., Gidoin, C., Karam, M.-J., Lalagüe, H., Oddou-Muratorio, S., Pichot, C., 2013. Considering evolutionary processes in adaptive forestry. Ann. Forest Sci.. http://dx.doi.org/10.1007/s13595-013-0272-1.

Leites, L.P., Rehfeldt, G.E., Robinson, A.P., Crookston, N.L., Jaquish, B.C., 2012a. Possibilities and limitations of using historic provenance tests to infer forest species growth responses to climate change. Nat. Resour. Model. 25, 409–433.

Leites, L.P., Robinson, A.P., Rehfeldt, G.E., Marshall, J.D., 2012b. Height-growth response to climate changes differs among populations of Douglas-fir: a novel analysis of historic data. Ecol. Appl. 22, 199–206.

Lenormand, T., 2002. Gene flow and the limits to natural selection. Trends Ecol. Evol. 17, 183–189.

Lenton, T.M., 2011. Early warning of climate tipping points. Nat. Climate Change 1, 201–209.

Liepelt, S., Bialozyt, R., Ziegenhagen, B., 2002. Wind-dispersed pollen mediates postglacial gene flow among refugia. PNAS 99, 14590–14594.

Lira-Medeiros, C.F., Parisod, C., Fernandes, R.A., Mata, C.S., Cardoso, M.A., Ferreira, P.C., 2010. Epigenetic variation in mangrove plants occurring in contrasting natural environment. PLoS ONE 5, e10326.

Loarie, S.R., Duffy, P.H., Hamilton, H., Asner, G.P., Field, C.B., Ackerly, D.D., 2009. The velocity of climate change. Nature 462, 1052–1055.

86 R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87

Lopez, S., Rousset, F., Shaw, F.H., Shaw, R.G., Ronce, O., 2009. Joint effects of inbreeding and local adaptation on the evolution of genetic load after fragmentation. Conserv. Biol. 23, 1618–1627.

Malcolm, J.R., Markham, A., Neilson, R.P., Garaci, M., 2002. Estimated migration rates under scenarios of global climate change. J. Biogeograp. 29, 835–849.

Malhi, Y., Aragao, L.E.O.C., Galbraith, D., Huntingford, C., Fisher, R., Zelazowski, P., Sitch, S., McSweeney, C., Meir, P., 2009. Exploring the likelihood and mechanism of a climate-change-induced dieback of the Amazon rainforest. Proc. Natl. Acad. Sci. USA 106, 20610–20615.

MAP (Ministère de l’Agriculture et de la Pêche). 2006. Stratégie nationale pour la biodiversité. Plan d’action forêt. Paris, France. <http://agriculture.gouv.fr/IMG/ pdf/biodiversit%C3%A9_foret_051006.pdf>.

Marris, E., 2009. Planting the forest of the future. Nature 459, 906–908. Mátyás, C., 1994. Modeling climate change effects with provenance test data. Tree

Physiol. 14, 797–804. Mátyás, C., 2010. Forecasts needed for retreating forests. Nature 464, 1271. Mátyás, C., Yeatman, C.W., 1992. Effect of geographical transfer on growth and

survival of jack pine (Pinus banksiana Lamb.) populations. Silvae Genet. 43, 370– 376.

Mátyás, C., Berki, I., Czúcz, B., Gálos, B., Móricz, N., Rasztovits, E., 2010. Future of beech in Southern Europe from the perspective of evolutionary ecology. Acta Silv. Lign. Hung. 6, 91–110.

McLachlan, J.S., Clark, J.S., Manos, P.S., 2005. Molecular indicators of tree migration capacity under rapid climate change. Ecology 86, 2088–2098.

McLachlan, J., Hellmann, J.J., Schwartz, M.W., 2007. A framework for debate of assisted migration in an era of climate change. Conserv. Biol. 21, 297–302.

McLane, L., Aitken, S.N., 2012. Whitebark pine (Pinus albicaulis) assisted migration potential: testing establishment north of the species range. Ecol. Appl. 22, 142– 153.

Meir, P., Woodward, F.I., 2010. Amazonian rain forests and drought: response and vulnerability. New Phytol. 187, 553–557.

Meuwissen, T.H.E., Hayes, B.J., Goddard, M.E., 2001. Prediction of total genetic value using genome wide dense marker maps. Genetics 157, 1819–1829.

Milad, M., Schaich, H., Konold, W., 2013. How is adaptation to climate change reflected in current practice of forest management and conservation? A case study from Germany. Biodivers. Conserv. 22, 181–1202.

Millar, C.I., Stephenson, N.L., Stephens, S.L., 2007. Climate change and forests of the future: managing in the face of uncertainty. Ecol. Appl. 17, 2145–2151.

Mooney, H., Larigauderie, A., Cesario, M., Elmquist, T., Hoegh-Guldberg, O., Lavorel, S., Mace, G.M., Palmer, M., Scholes, R., Yahara, T., 2009. Biodiversity, climate change and ecosystem services. Curr. Opin. Environ. Sustain. 1, 46–54.

Moore, B.A., 2005. Alien invasive species: impacts on forests and forestry. A review. Forest Health and Biosecurity Working Paper FBS/8E. Forestry Department, FAO, Rome.

Moreira, X., Alfaro, R.I., King, J.N., 2012. Constitutive defenses and damage in Sitka spruce progeny obtained from crosses between white pine weevil resistant and susceptible parents. Forestry 85, 87–97.

Moriondo, M., Good, P., Durao, R., Bindi, M., Giannakopoulos, C., Corte-Real, J., 2006. Potential impact of climate change on fire risk in the Mediterranean area. Clim. Res. 31, 85–95.

Murdock, T.Q., Taylor, S.W., Flower, A., Mehlenbacher, A., Montenegro, A., Zwiers, F.W., Alfaro, R.I., Spittlehouse, D.L., 2013. Pest outbreak distribution and forest management impacts in a changing climate in British Columbia. J. Environ. Sci. Policy 26, 75–89. http://dx.doi.org/10.1016/j.envsci.2012.07.026.

Mutke, S., Gordo, J., Chambel, M.R., Prada, M.A., Álvarez, D., Iglesias, S., Gil, L., 2010. Phenotypic plasticity is stronger than adaptative differentiation among Mediterranean stone pine provenances. Forest Syst. 19, 354–366.

Nathan, R., Katul, G.G., Horn, H.S., Suvi, M.T., Oren, R., Avissar, O., Pacala, S.W., Levin, S.A., 2002. Mechanisms of long-distance dispersal of seeds by wind. Nature 418, 409–413.

Navarro, C., Ward, S., Hernández, M., 2002. The tree Cedrela odorata (Meliaceae): a morphologically subdivided species in Costa Rica. Revista. Biol. Trop. 50, 21–29.

Navarro, C., Cavers, S., Pappinen, A., Tigerstedt, P., Lowe, A., Merila, J., 2005. Contrasting quantitative traits and neutral genetic markers for genetic resource assessment of Mesoamerican Cedrela odorata. Silvae Genet. 54, 281–292.

Nepstad, D.C., Tohver, I.M., Ray, D., Moutinho, P., Cardinot, G., 2007. Mortality of large trees and lianas following experimental drought in an Amazon forest. Ecology 88, 2259–2269.

Nicotra, A.B., Atkin, O.K., Bonser, S.P., Davidson, A.M., Finnegan, E.J., Mathesius, U., Poot, P., Purugganan, M.D., Richards, C.L., Valladares, F., van Kleunen, M., 2010. Plant phenotypic plasticity in a changing climate. Trends Plant Sci. 15, 684–692.

Oddou-Muratorio, S., Klein, E.K., Austerlitz, F., 2005. Pollen flow in the wildservice tree, Sorbus torminalis (L.) Crantz. II. Pollen dispersal and heterogeneity in mating success inferred from parent-offspring analysis. Mol. Ecol. 14, 4441– 4452.

ONERC (Observatoire National sur les effets du changement climatique). 2007. Stratégie nationale d’adaptation au changement climatique. La documentation française, Paris, France. <http://www.developpement-durable.gouv.fr/IMG/pdf/ Strategie_Nationale_2-17_Mo-2-2.pdf>.

Parry, M.L., Rosenzweig, C., Iglesis, A., Livermore, M., Fisher, G., 2004. Effects of climate change on global food production under SRES emissions and socio- economic scenarios. Global Environ. Change 14, 53–67.

Pausas, J.G., 2004. Changes in fire and climate in the eastern Iberian Peninsula (Mediterranean basin). Climat. Change 63, 337–350.

Pearse, I.S., Koenig, W.D., Knops, J.M.H., 2014. Cues versus proximate drivers: testing the mechanism behind masting behavior. Oikos 123, 179–184.

Peñuelas, J., Oyaga, R., Boada, M., Jump, A.S., 2007. Migration, invasion and decline: changes in recruitment and forest structure in a warming-linked shift of European beech forest in Catalonia (NE Spain). Ecography 30, 830–838.

Perdiguero, P., Barbero, M.D., Cervera, M.T., Collada, C., Soto, A., 2013. Molecular response to water stress in two contrasting Mediterranean pines (Pinus pinaster and Pinus pinea). Plant Physiol. Biochem. 67, 199–208.

Person, C., 1966. Genetic polymorphism in parasitic systems. Nature 212, 266–267. http://dx.doi.org/10.1038/212266a0.

Peterson, A.T., Stewart, A., Mohamed, K.I., Araujo, M.B., 2008. Shifting global invasive potential of European plants with climate change. PLoS ONE 3, e2441.

Petit, R.J., Hampe, A., 2006. Some evolutionary consequences of being a tree. Annu. Rev. Ecol. Evol. Syst. 37, 187–214.

Petit, R.J., Bialozyt, R., Garnier-Géré, P., Hampe, A., 2004. Ecology and genetics of tree invasions: from recent introductions to Quaternary migrations. For. Ecol. Manage. 197, 117–137.

Piersma, T., Drent, J., 2003. Phenotypic flexibility and the evolution of organismal design. Trends Ecol. Evol. 18, 228–233.

Pigliucci, M., Murren, C.J., 2003. Genetic assimilation and a possible evolutionary paradox: can macroevolution sometimes be so fast as to pass us by? Evolution 57, 1455–1464.

Pimentel, D., 1961. Animal population regulation by the genetic feed-back mechanism. Am. Nat. 95, 65–79.

Piñol, J., Terradas, J., Lloret, F., 1998. Climate warming, wildfire hazard, and wildfire occurrence in coastal eastern Spain. Climat. Change 38, 345–357.

Polechova, J., Barton, N., Marion, G., 2009. Species’ range: adaptation in space and time. Am. Nat. 174, 186–204.

Polle, A., Altman, A., Jiang, X.N., 2006. Towards genetic engineering for drought tolerance in trees. In: Tree Transgenesis: Recent Developments. Springer- Verlag, Berlin, pp. 75–297.

Ræbild, A., Larsen, A.S., Jensen, J.S., Ouedraogo, M., Groote, S., Damme, P., Bayala, J., 2011. Advances in domestication of indigenous fruit trees in the West African Sahel. New Forest. 41, 297–315.

Raffa, K.F., Powell, E.N., Townsend, P.A., 2013. Temperature-driven range expansion of an irruptive insect heightened by weakly coevolved plant defenses. Proc. Natl. Acad. Sci. USA 110, 2193–2198.

Rehfeldt, G.E., Jaquish, B.C., 2010. Ecological impacts and management strategies for western larch in the face of climate-change. Mitig. Adapt. Strat. Glob. Change 15, 283–306.

Rehfeldt, G.E., Ying, C.C., Spittlehouse, D.L., Hamilton, D.A., 1999. Genetic responses to climate in Pinus contorta: niche breadth, climate change, and reforestation. Ecol. Monogr. 69, 375–407.

Rehfeldt, G.E., Wykoff, W.R., Ying, C.C., 2001. Physiologic plasticity, evolution, and impacts of a changing climate on Pinus contorta. Climat. Change 50, 355–376.

Rehfeldt, G.E., Tchebakova, N.M., Parfenova, Y.I., Wykoff, W.R., Kuzmina, N.A., Milyutin, L.I., 2002. Intraspecific responses to climate in Pinus sylvestris. Glob. Change Biol. 8, 912–929.

Rehfeldt, G.E., Tchebakova, N.M., Parfenova, E.I., 2004. Genetic responses to climate and climate-change in conifers of the temperate and boreal forests. Recent Res. Dev. Genet. Breed. 1, 113–130.

Rehfeldt, G.E., Ferguson, D.E., Crookston, N.L., 2009. Aspen, climate and sudden decline in western USA. For. Ecol. Manage. 258, 2353–2364.

Rehfeldt, G.E., Crookston, N.L., Sáenz-Romero, C., Campbell, E., 2012. North American vegetation model for land use planning in a changing climate: a statistical solution to large classification problems. Ecol. Appl. 22, 119–141.

Restoux, G.E., Silva, D., Sagnard, F., Torre, F., Klein, E., Fady, B., 2008. Life at the margin: the mating system of Mediterranean conifers. Web Ecol. 8, 94–102.

Richardson, D.M., 1998. Forestry trees as invasive aliens. Conserv. Biol. 12, 18–26. Rohde, A., Junttila, O., 2008. Remembrances of an embryo: long-term effects on

phenology traits in spruce. New Phytol. 177, 2–5. Roques, L., Hamel, F., Fayard, J., Fady, B., Klein, E.K., 2010. Recolonisation by

diffusion can generate increasing rates of spread. Theor. Popul. Biol. 77, 205– 212.

Sáenz-Romero, C., Rehfeldt, G.E., Crookston, N.L., Pierre, D., St-Amant, R., Beaulieu, J., Richardson, B., 2010. Spline models of contemporary, 2030, 2060 and 2090 climates for Mexico and their use in understanding climate-plant impacts on vegetation. Climat. Change 102, 595–623.

Sáenz-Romero, C., Rehfeldt, G.E., Duval, P., Lindig-Cisneros, R., 2012. Abies religiosa habitat prediction in climatic change scenarios and implications for monarch butterfly conservation in Mexico. For. Ecol. Manage. 275, 98–106.

Safranyik, L., Carroll, A., 2006. The biology and epidemiology of the mountain pine beetle in lodgepole pine forests. In: Safranik, L., Wilson, B. (Eds.), The Mountain Pine Beetle. A Synthesis of Biology, Management and Impacts on Lodgepole Pine. Natural Resources Canada, Canadian Forest Service, Pacific Forestry Centre, Victoria, BC. Canada, pp. 3–66.

Sagnard, F., Pichot, C., Dreyfus, P., Jordano, P., Fady, B., 2007. Modelling seed dispersal to predict seedling recruitment: recolonization dynamics in a plantation forest. Ecol. Model. 203, 464–474.

Sagnard, F., Oddou-Muratorio, S., Pichot, C., Vendramin, G.G., Fady, B., 2011. Effect of seed dispersal, adult tree and seedling density on the spatial genetic structure of regeneration at fine temporal and spatial scales. Tree Genet. Genom. 7, 37–48.

Sanou, H., Korbo, A., Tougani, A., Rabiou, A., Kambou, S., Ouedraogo, M., Diallo, B.O., Parkouda, C., Ræbild, A., Jensen, J.S., 2007. Protocol for establishment of trials with baobab and tamarind within the SAFRUIT project. Working Paper No. 21- 2007. Forest and Landscape Denmark, Hørsholm, Denmark.

Santos-del-Blanco, L., Bonser, S.P., Valladares, F., Chambel, M.R., Climent, J.M., 2013. Plasticity in reproduction and growth among 52 range-wide populations of a

R.I. Alfaro et al. / Forest Ecology and Management 333 (2014) 76–87 87

Mediterranean conifer: adaptive responses to environmental stress. J. Evol. Biol. 26, 1912–1924.

Savolainen, O., Pyhajarvi, T., Knurr, T., 2007. Gene flow and local adaptation in trees. Annu. Rev. Ecol. Evol. Syst. 38, 595–619.

Savolainen, O., Kujala, S.T., Sokol, C., Pyhäjärvi, T., Avia, K., Knürr, T., Kärkkäinen, K., Hicks, S., 2011. Adaptive potential of northernmost tree populations to climate change, with emphasis on scots pine (Pinus sylvestris L.). J. Hered. 102, 526–536. http://dx.doi.org/10.1093/jhered/esr056.

Scheffer, M., Carpenter, S.R., 2003. Catastrophic regime shifts in ecosystems: linking theory to observation. Trends Ecol. Evol. 18, 648–656.

Scheffer, M., Carpenter, S., Foley, J., Folke, C., Walker, B., 2001. Stochastic events can trigger large state shifts in ecosystems with reduced resilience. Nature 413, 591–596.

Schwilk, D.W., Ackerly, D.D., 2001. Flammability and serotiny as strategies: correlated evolution in pines. Oikos 94, 326–336.

Selås, V., Piovesan, G., Adams, J.M., Bernabei, M., 2002. Climatic factors controlling reproduction and growth of Norway spruce in southern Norway. Can. J. For. Res. 32, 217–225.

Shinozaki, K., Yamaguchi-Shinozaki, K., 2007. Gene networks involved in drought stress response and tolerance. J. Exp. Bot. 58, 221–227.

Simard, S., Morin, H., Krause, C., 2011. Long-term spruce budworm outbreak dynamics reconstructed from subfossil trees. J. Quat. Sci. 26, 734–738.

Simberloff, D., Relva, M.A., Nuñez, M.A., 2002. Gringos en el bosque: introduced tree invasion in a native Nothofagus/Austrocedrus forest. Biol. Invasions 4, 35–53.

Skrøppa, T., Johnsen, Ø., 2000. Patterns of adaptive genetic variation in forest tree species; the reproductive environment as an evolutionary force in Picea abies. In: Mátyás, C. (Ed.), Forest Genetics and Sustainability. Kluwer Academic Publications, Dordrecht, The Netherlands, pp. 49–58.

Skrøppa, T., Tollefsrud, M., Sperisen, C., Johnsen, Ø., 2010. Rapid change in adaptive performance from one generation to the next in Picea abies—Central European trees in a Nordic environment. Tree Genet. Genom. 6, 93–99.

Smith, C.M., Wilson, B., Rasheed, S., Walker, R.C., Carolin, T., Shepherd, B., 2008. Whitebark pine and white pine blister rust in the Rocky Mountains of Canada and northern Montana. Can. J. Forest Res. 38, 982–995.

Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L., 2007. In: IPCC 2007: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK.

Soto-Correa, J.C., Sáenz-Romero, C., Lindig-Cisneros, R., Sánchez-Vargas, N.M., Cruz- de-León, J., 2012. Genetic variation between Lupinus elegans Kunth provenances, altitudinal seed zoning and assisted migration. Agrociencia 46 (6), 593–608.

Steffan-Dewenter, I., Kessler, M., Barkmann, J., Bos, M.M., Buchori, D., Erasmi, S., Faust, H., Gerold, G., Glenk, K., Gradstein, S.R., Guhardja, E., Harteveld, M., Hertel, D., Höhn, P., Kappas, M., Köhler, S., Leuschner, C., Maertens, M., Marggraf, R., Migge-Kleian, S., Mogea, J., Pitopang, R., Schaefer, M., Schwarze, S., Sporn, S.G., Steingrebe, A., Tjitrosoedirdjo, S.S., Tjitrosoemito, S., Twele, A., Weber, R., Woltmann, L., Zeller, M., Tscharntke, T., 2007. Tradeoffs between income, biodiversity, and ecosystem functioning during tropical rainforest conversion and agroforestry intensification. Proc. Natl. Acad. Sci. USA 104, 4973–4978.

Thompson, J.N., Burdon, J.J., 1992. Gene-for-gene co-evolution between plants and parasites. Nature 360, 121–125.

Thornley, J.H.M., Cannell, M.G.R., 1996. Temperate forest responses to carbon dioxide, temperature and nitrogen: a model analysis. Plant, Cell Environ. 19, 1331–1348.

Thorsen, B.J., Kjær, E.D., 2007. Forest genetic diversity and climate change: economic considerations. In: Koskela, J., Buck, A., Teissier du Cros, E. (Eds.), Climate Change and Forest Genetic Diversity: Implications for Sustainable Forest Management in Europe. Bioversity International, Rome, Italy, pp. 69–84.

Thuiller, W., Albert, C., Araujo, M.B., Berry, P.M., Cabeza, M., Guisan, A., Hickler, T., Midgley, G.F., Paterson, J., Schurr, F.M., Sykes, M.T., Zimmermann, N.E., 2008. Predicting global change impacts on plant species’ distributions: future challenges. Perspect. Plant Ecol., Evol. Syst. 9, 137–152.

Timbal, J., Bonneau, M., Landmann, G., Trouvilliez, J., Bouhot-Delduc, L., 2005. European non boreal conifer forests. In: Andersson, F.A. (Ed.), Ecosystems of the World (6): Coniferous Forests. Elsevier, Amsterdam, The Netherlands, pp. 131– 162.

UNEC (United Nations Economic Council). 2009. Reversing the loss of forest cover, preventing forest degradation in all types of forests and combating desertification including in low forest cover countries. Report of the Secretary-General, United Nations Forum on Forests, Eighth session, New York, 20 April-1 May, 2009. E/CN.18/2009/5.

UNESCO. 2005. The Precautionary Principle. World Commission on the Ethics of Scientific Knowledge and Technology (COMEST); <http://unesdoc.unesco.org/ images/0013/001395/139578e.pdf> (accessed 28. 03.12).

Valladares, F., Arrieta, S., Aranda, I., Lorenzo, D., Sánchez-Gómez, D., Tena, D., Suárez, F., Pardos, J.A., 2005. Shade tolerance, photoinhibition sensitivity and phenotypic plasticity of Ilex aquifolium in continental Mediterranean sites. Tree Physiol. 25, 1041–1052.

Vendramin, G.G., Fady, B., González-Martínez, S.C., Hu, F.S., Scotti, I., Sebastiani, F., Soto, A., Petit, R.J., 2008. Genetically depauperate but widespread: the case of an emblematic Mediterranean pine. Evolution 62, 680–688.

Verdu, M., Pausas, J.G., 2007. Fire drives phylogenetic clustering in Mediterranean Basin woody plant communities. J. Ecol. 95, 1316–1323.

Vitasse, Y., Bresson, C.C., Kremer, A., Michalet, R., Delzon, S., 2010. Quantifying phenological plasticity to temperature in two temperate tree species. Funct. Ecol. 24, 1211–1218.

Wang, T., Hamann, A., Yanchuk, A., O’Neill, G.A., Aitken, S.N., 2006. Use of response functions in selecting lodgepole pine populations for future climate. Glob. Change Biol. 12, 2404–2416.

Wang, W., Peng, C., Kneeshaw, D.D., Larocque, G., Song, X., Zhou, X., 2012. Quantifying the effects of climate change and harvesting on carbon dynamics of boreal aspen and jack pine forests using the TRIPLEX-Management model. For. Ecol. Manage. 281, 152–162.

Ward, M., Dick, C.W., Lowe, A.J., 2005. To self, or not to self. a review of outcrossing and pollen-mediated gene flow in neotropical trees. Heredity 95, 246–254.

Webber, J., Ott, P., Owens, J., Binder, W., 2005. Elevated temperature during reproductive development affects cone traits and progeny performance in Picea glauca–engelmannii complex. Tree Physiol. 25, 1219–1227.

Whitham, T.G., Bailey, J.K., Schweitzer, J.A., Shuster, S.M., Bangert, R.K., LeRoy, C.J., Lonsdorf, E.V., Allan, G.J., DiFazio, S.P., Potts, B.M., Fischer, D.G., Gehring, C.A., Lindroth, R.L., Marks, J.C., Hart, S.C., Wimp, G.M., Wooley, S.C., 2006. A framework for community and ecosystem genetics: from genes to ecosystems. Nat. Rev. Genet. 7, 510–523.

Willis, K.J., van Andel, T.H., 2004. Trees or no trees? The environments of central and eastern Europe during the Last Glaciation. Quatern. Sci. Rev. 23, 2369–2387.

Willson, M.F., 1993. Dispersal mode, seed shadows, and colonization patterns. Vegetatio 107 (108), 261–280.

Winder, R., Nelson, E.A., Beardmore, T., 2011. Ecological implications for assisted migration in Canadian forests. Forest. Chron. 87, 731–744.

Yakovlev, I.A., Fossdal, C.G., Johnsen, Ø., 2010. MicroRNAs, the epigenetic memory and climatic adaptation in Norway spruce. New Phytol. 187, 1154–1169.

Yanchuk, A., Allard, G., 2009. Tree improvement programmes for forest health – can they keep pace with climate changes? Unsylva 60, 50–56.

Zeglen, S., 2002. Whitebark pine and white pine blister rust in British Columbia, Canada. Can. J. For. Res. 32, 1265–1274.

Global conservation priorities for crop wild relatives Nora P. Castañeda-Álvarez1,2†*, Colin K. Khoury1,3†, Harold A. Achicanoy1, Vivian Bernau1, Hannes Dempewolf4, Ruth J. Eastwood5, Luigi Guarino4, Ruth H. Harker5, Andy Jarvis1,6, Nigel Maxted2, Jonas V. Müller5, Julian Ramirez-Villegas1,6,7, Chrystian C. Sosa1, Paul C. Struik3, Holly Vincent2 and Jane Toll4

The wild relatives of domesticated crops possess genetic diversity useful for developing more productive, nutritious and resilient crop varieties. However, their conservation status and availability for utilization are a concern, and have not been quantified globally. Here, we model the global distribution of 1,076 taxa related to 81 crops, using occurrence information collected from biodiversity, herbarium and gene bank databases. We compare the potential geographic and ecological diversity encompassed in these distributions with that currently accessible in gene banks, as a means to estimate the comprehensiveness of the conservation of genetic diversity. Our results indicate that the diversity of crop wild relatives is poorly represented in gene banks. For 313 (29.1% of total) taxa associated with 63 crops, no germplasm accessions exist, and a further 257 (23.9%) are represented by fewer than ten accessions. Over 70% of taxa are identified as high priority for further collecting in order to improve their representation in gene banks, and over 95% are insufficiently represented in regard to the full range of geographic and ecological variation in their native distributions. The most critical collecting gaps occur in the Mediterranean and the Near East, western and southern Europe, Southeast and East Asia, and South America. We conclude that a systematic effort is needed to improve the conservation and availability of crop wild relatives for use in plant breeding.

T he challenges to global food security are complex and compounding. Our growing population and changing dietary expectations are projected to increase demand on

food systems for at least the next four decades1–5, outpacing forecasted crop yield gains6. Limitations in land, water and other natural resource inputs, competition for arable soils with non- food crops and other land uses, soil degradation, climate change and the need to minimize harmful impacts on ecosystem services and biodiversity further constrain production potential3,4,7,8. Although gains in food availability may partially be obtained through dietary change and food waste reduction1,3, increases in the productivity, resilience and sustainability of current agricultural systems are clearly necessary5. Key to this sustainable intensification is the use of novel genetic diversity in plant breeding to produce crop varieties containing traits such as drought and heat tolerance, increased pest and disease resistance, and input use efficiency9–11.

As sources of new genetic diversity, crop wild relatives—the wild cousins of cultivated plant species—have been used for many decades for plant breeding, contributing a wide range of beneficial agronomic and nutritional traits12–17. Their utilization is expected to increase as a result of ongoing improvements in information on species and their diversity and advances in breeding tools16,18. However, this expectation is based on the assumption that crop wild relatives will be readily available for research and plant breed- ing, which requires their conservation as germplasm accessions in gene banks as well as functioning mechanisms to enable access to

this diversity10,11. Preliminary assessments of the comprehensive- ness of conservation of wild relatives in gene banks have suggested substantial gaps19,20, and wild populations of a range of species are threatened by the conversion of natural habitats to agriculture, urbanization, invasive species, mining, climate change and/or pol- lution21–23. A concerted effort devoted to improving the conserva- tion and availability of crop wild relatives for crop improvement is thus timely both for biodiversity conservation and for food security objectives24, as the window of opportunity to resolve these deficiencies will not remain open indefinitely20,22.

We conducted a detailed analysis of the extent of representation of the wild relatives of 81 crops in gene banks equipped to provide access to these genetic resources to the global research and breeding community. The crops include major and minor cereals, root and tuber crops, oilcrops, vegetables, fruits, forages and spices, chosen on the basis of their importance to food security, income generation and sustainable agricultural production (Supplementary Table 1). We first modelled the geographic distributions of a total of 1,076 unique crop wild relative taxa from 76 genera and 24 plant families (Supplementary Table 2). We then compared the potential geo- graphic and ecological diversity encompassed in these distributions to that which is currently accessible in gene banks25. To aid conser- vation strategies, we categorized taxa with a final priority score (FPS) for further collecting from the natural habitats of crop wild relatives to increase representation in gene banks, on a scale from zero to ten. The FPS was created by averaging each taxon’s assessed current

1International Center for Tropical Agriculture (CIAT), Km 17, Recta Cali-Palmira, Cali 763537, Colombia. 2School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. 3Centre for Crop Systems Analysis, Wageningen University, Droevendaalsesteeg 1, 6708 PB Wageningen, The Netherlands. 4Global Crop Diversity Trust, Platz der Vereinten Nationen 7, 53115 Bonn, Germany. 5Royal Botanic Gardens, Kew, Conservation Science, Millennium Seed Bank, Wakehurst Place, Ardingly RH17 6TN, UK. 6CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), Km 17, Recta Cali-Palmira, Cali 763537, Colombia. 7Institute for Climate and Atmospheric Science, School of Earth and Environment, University of Leeds, LS2 9JT, UK. †These authors contributed equally to this work. *e-mail: [email protected]

ARTICLES PUBLISHED: 21 MARCH 2016 | ARTICLE NUMBER: 16022 | DOI: 10.1038/NPLANTS.2016.22

NATURE PLANTS | www.nature.com/natureplants 1

© 2016 Macmillan Publishers Limited. All rights reserved

representation in gene banks in regard to overall number of acces- sions, geographic diversity and ecological diversity. High priority for further collecting was assigned for taxa where FPS ≥ 7 (that is, very little or no current representation in gene banks); medium priority where 5 ≤ FPS < 7; low priority where 2.5 ≤ FPS < 5; and sufficiently represented for taxa with FPS < 2.5. Finally, we identified geographic hotspots where considerable richness of high-priority wild relative taxa is concentrated. Such sites represent particularly valuable targets, both for efficient collecting for ex situ conservation in gene banks and for in situ conservation in protected areas.

Results The distributions of crop wild relatives were modelled to occur on all continents except Antarctica, and throughout most of the tropics, subtropics and temperate regions, except the most arid areas and polar zones (Fig. 1). The greatest richness of taxa was modelled in the Mediterranean, Near East and southern Europe, South America, Southeast and East Asia, and Mesoamerica, with up to 84 taxa overlapping in a single 25 km2 grid cell. These richness hotspots largely align with traditionally recognized centres of crop diversity26, although the analysis also identified a number of less well-recognized areas, for example central and western Europe, the eastern USA, southeastern Africa and northern Australia, which also contain considerable richness. Hotspots in tropical and subtropical areas also largely aligned with zones recorded as posses- sing high richness of endemic flora and fauna, and experiencing exceptional degrees of loss of habitat27. Temperate regions identified under the same criteria, for example the California and Cape Floristic Provinces, southwestern Australia, central Chile and New Zealand, had considerably less overlap with areas rich in crop wild relatives.

Wild relative taxa as a class of plant genetic resources were found to be critically under-represented in gene banks. For 313 (29.1% of total) taxa associated with 63 crops, no germplasm accessions exist at all, and a further 257 taxa are represented by fewer than ten acces- sions. A total of 765 (71.1%) taxa were ranked as high priority for further collecting from their natural habitats, 148 (13.8%) as medium priority, 118 (11.0%) as low priority and only 45 (4.2%) as currently sufficiently represented in gene banks (Supplementary Table 2). The mean FPS across all species (7.9 ± 2.5 (mean ± s.d.)) fitted well within the high priority category range (Fig. 2). Lack of geographic and ecological representation in gene banks contributed

significantly to most of the high FPS values, whereas less extreme gaps were generally evident in the total numbers of accessions con- served (Supplementary Fig. 1).

An analysis of wild relatives grouped by their associated crop (that is, by crop gene pool) revealed that 72% of the crop gene pools had been assigned to high priority for further collecting (as an average of FPS scores across associated wild relative taxa), and thus require urgent conservation action (Fig. 2). These included the gene pools of commodity crops of critical importance to global food supplies and/or agricultural production, for example sugarcane (9.2 ± 1.6), sugar beet (8.1 ± 1.6) and maize (6.9 ± 2.1), as well as important food security staples such as banana and plantain (9.4 ± 0.8), cassava (9.0 ± 1.6), sorghum (8.8 ± 1.0), yams (8.5 ± 2.9), cowpea (8.4 ± 1.7), sweet potato (8.4 ± 1.7), pigeon pea (8.4 ± 1.1), millets (8.4 ± 2.7) and groundnut (7.6 ± 1.8) (Fig. 3 and Supplementary Table 1). High priority was also assigned to the gene pools of numer- ous crops important for smallholder income generation in the tropics (for example, cacao and papaya) and minor crops increasing in popu- larity because of their nutritional qualities (quinoa), as well as various other important fruits (for example, grape, apple, watermelon, orange and mango), oilcrops (rapeseed) and forages (alfalfa) possessing con- siderable numbers of wild related taxa. Although all gene pools con- tained taxa with considerable conservation concerns, the wild relatives of fruits, forages, sugar crops, starchy roots and vegetables were those assessed as least well represented in gene banks (Supplementary Fig. 2). Average FPS values across all wild relatives per crop type were 8.8 ± 1.8 for fruits, 8.7 ± 1.7 for forages, 8.6 ± 1.6 for sugar crops, 8.2 ± 2.3 for starchy roots, 8.1 ± 2.4 for vegetables, 7.2 ± 2.6 for pulses, 7.1 ± 2.3 for oilcrops, 7.1 ± 1.9 for spices and 6.4 ± 3.1 for cereals.

None of the 81 assessed crop gene pools demonstrated an average FPS across its wild relatives that would permit its categorization as sufficiently well represented in gene banks (Fig. 2). The wild rela- tives of six crops were assessed as fairly well represented, that is low current priority for further collecting for the gene pools of wheat (3.7 ± 2.4), grass pea (3.7 ± 2.0), chickpea (4.2 ± 2.6) and tomato (4.5 ± 1.9). Wheat and tomato, along with medium-priority crop gene pools such as sunflower (6.3 ± 2.2), rice (6.6 ± 2.5) and potato (6.7 ± 2.6), have a long history of use of wild relatives in crop improvement9,13 and benefit from relatively extensive germ- plasm collections. Other crop gene pools determined as low priority (grass pea and chickpea) have few wild relatives, and these generally present restricted distributions that have been fairly well sampled.

Number of overlapping crop wild relative taxa

1–1 2

13 –2

4 25

–3 6

37 –4

8 49

–6 0

61 –7

2 73

–8 4

Figure 1 | Crop wild relative taxon richness map. The map displays overlapping potential distribution models for assessed crop wild relatives. Dark red indicates greater overlap of potential distributions of taxa, that is, where greater numbers of crop wild relative taxa occur in the same geographic area.

ARTICLES NATURE PLANTS DOI: 10.1038/NPLANTS.2016.22

NATURE PLANTS | www.nature.com/natureplants2

© 2016 Macmillan Publishers Limited. All rights reserved

However, specific taxa were assessed as under-represented in gene banks even within these low-priority gene pools. For example, five taxa related to wheat were assessed as medium or high priority, one taxon related to grass pea as medium priority, three taxa related to chickpea as medium priority and six taxa related to tomato as medium or high priority (Supplementary Table 2).

Proposed hotspots for further collecting for high-priority crop wild relatives were identified across the world’s tropical, subtropical and temperate regions, with the most critical gaps identified in the

Mediterranean, Near East, and southern and western Europe; Southeast and East Asia; and South America (Fig. 4). Up to 43 wild relative taxa (main map in Fig. 4) associated with up to 23 crops (inset map in Fig. 4) may potentially be collected within a single 25 km2 grid cell.

Discussion Our results demonstrate that crop wild relatives are currently under-represented and a systematic effort to improve their

Wheat Grass pea Chickpea

Tomato Soybean

Barley Lentil Bean

Sunflower Onion

Rice Cabbage

Vetch Potato

Lima bean Lettuce

Cottonseed Chili pepper

Oat Maize

Mung bean Zucchini

Pea Pumpkin

Rye Adzuki bean Watermelon

Groundnut Foxtail millet

Urd bean Safflower

Turnip Apple

Cucumber Grape

Black mustard Sugar beet

Pear Almond

Strawberry White Guinea yam

Finger millet Garlic

Pigeon pea Sweet potato

Grapefruit Cowpea

Peach Quinoa

Lagos yam Rapeseed

Alfalfa Mustard

Orange Pearl millet Water yam

Sorghum Cassava

Plum Cherry

Broom millet Bambara Spinach

Sugarcane Carrot Lemon

Eggplant Apricot

Banana and plantain Leek

Asparagus Mango Cacao

Breadfruit Papaya

Faba bean Melon

Pineapple Yautia

A ss

oc ia

te d

cr op

s

Priority score (FPS) for collecting and conserving crop wild relatives

0 2 4 6 8 10

Figure 2 | Collecting and conservation priorities for crop wild relatives by associated crop. Black circles represent the FPS for further collecting for wild relative taxa, with larger grey circles representing the average FPS across taxa per crop gene pool. The blue straight vertical line represents the mean FPS across all crop wild relative taxa within all crop gene pools.

NATURE PLANTS DOI: 10.1038/NPLANTS.2016.22 ARTICLES

NATURE PLANTS | www.nature.com/natureplants 3

© 2016 Macmillan Publishers Limited. All rights reserved

comprehensiveness in gene banks is critically needed. These find- ings are remarkable given the extensive efforts particularly in the past half century by international, regional and national initiatives to conserve the broad diversity of important agricultural crops11,20. Achieving the comprehensive conservation of crop genetic resources ex situ is constrained by technical as well as political and funding challenges in recent decades11, and is most poignant for wild taxa, which are less well researched than crop species and often more difficult to conserve and to utilize11,20,24. Addressing con- servation gaps globally for crop wild relatives, a goal that is specifi- cally targeted in recent major international agreements (the United Nations’ Sustainable Development Goals and the Strategic Plan for Biodiversity28) will require substantial investment and extensive international collaboration. The high spatial resolution of these results is already informing such initiatives24 and can be useful to the development of further efforts.

Here we outline priorities for collecting wild relatives on the basis of their current representation in gene banks (Fig. 2 and Supplementary Table 2), and also provide an assessment of the rela- tive importance to global food supplies and production systems worldwide of their associated crops (Fig. 3 and Supplementary Fig. 2), as well as additional information regarding the contribution of crops to food security and sustainable agriculture (Supplementary Table 1). We recommend filling gaps in ex situ conservation first for the wild relatives of crops significant to these criteria, for

example rice, maize, sugarcane, cassava, potato, bananas and plan- tains, sorghum, millets, sweet potato, yams, groundnut, cowpea and pigeon pea.

To further refine these priorities, additional information and filters are needed. These include incorporating knowledge of threats to populations due to habitat modification, climate change and other impacts. Preliminary field surveys and threat analyses for under-represented taxa are therefore urgently needed. We note that extensive expert evaluations of the results generally confirmed the robustness of our species distribution models and conservation prioritizations but also clearly emphasized the need to address urgent threats to the survival of many crop wild relative populations (Supplementary Fig. 3). Realistic strategies for field collecting and subsequent ex situ conservation resulting in an increased availability of germplasm for plant breeding also require negotiating policy gov- erning germplasm collecting and exchange29,30, assessing field work risks (for example, war and civil strife in regions with high levels of diversity of wild relatives), coordinating timing of field work to maximize the collection of viable seeds and other propagules, prior- itizing target crop gene pools based on the interest of the breeding community in utilizing wild germplasm, and determining the relative difficulty of maintenance of targeted wild germplasm in gene banks. Although the seeds of most wild relatives can be main- tained under standard conditions for long-term conservation ex situ, some wild relatives produce recalcitrant seeds or do not

Barley

Bean

Chickpea

Grass pea Lentil

Quinoa

Safflower

Soybean

Tomato

Vetch

Yautia

RiceWheat

Groundnut

Maize

Sugarcane

Cassava

Potato

Apple

Papaya

Sorghum

Pigeon pea

Sweet potato

Mango

Banana and plantain

Faba bean

Sugar beet

0.02

0.04

0.06

0.08 0.10

0.20

0.40

0.60

0.80 1.00

2.00

4.00

6.00

8.00 10.00

0 1 2 3 4 5 6 7 8 9 10 Mean priority score (FPS) for collecting and conserving crop wild relatives

M ea

n im

po rt

an ce

s co

re o

f a ss

oc ia

te d

cr op

s (l

og 10

) Number of crop wild relative taxa per crop gene pool

10

30

60

90

130

Crop type

Cereals

Forages

Fruits

Oil crops

Pulses

Spices

Starchy roots

Sugar crops

Vegetables

Figure 3 | Collecting priorities for crop wild relatives and the importance of associated crops. The priority scale displays the average FPS across wild relatives per crop. The mean importance class of associated crops displays the significance of crops averaged across global food supplies and agricultural production metrics (see Supplementary Methods). For both axes, the scale is zero to ten, with ten representing the highest priority for further collecting/most important crop. The size of crop gene pool circles denotes the number of wild relative taxa per crop, ranging from 1 (faba bean) to 135 (cassava).

ARTICLES NATURE PLANTS DOI: 10.1038/NPLANTS.2016.22

NATURE PLANTS | www.nature.com/natureplants4

© 2016 Macmillan Publishers Limited. All rights reserved

produce seeds at all. Such wild relatives may require more expensive approaches (for example, in vitro or cryopreservation), and particu- larly for such taxa alternative conservation strategies such as the establishment of in situ conservation reserves may be more effective.

Despite an extensive effort to compile occurrence records from more than 400 different data sources, the wild relatives of a number of important agricultural crops (namely coffee, tea and avocado) were not assessed because of the lack of sufficient accessi- ble data. We also note that a number of agricultural crops are not currently known to possess closely related wild relatives, including taro (Colocasia esculenta), coconut (Cocos nucifera) and date palm (Phoenix dactylifera). Improvements in the generation and accessi- bility of taxonomic, relatedness and geographic information on wild relatives19,31 may permit conservation assessments for some of these gene pools in the future.

The combination of the sampling, geographic and ecological representativeness scores used to determine the extent of conserva- tion of the wild relatives of important agricultural crops in gene banks represents an efficient methodology for prioritizing taxa across crop gene pools given wide variations in the potential diver- sity encompassed in each taxon and the general absence of molecu- lar data for such species. The sampling representativeness score permitted an indication of the total number of germplasm acces- sions estimated as sufficient to represent a taxon, relative to the known extent of the taxon and utilizing all gene bank and reference data regardless of whether geographical coordinates are available. Geographic and ecological variation metrics were used as proxy for genetic diversity and potential functional adaptation to diverse environments, based on the assumption that the genetic compo- sition of plant species varies across geographic range and is associ- ated with adaptation to different ecological conditions32. The increasing power and decreasing costs of direct measures of diversity in genomes may make significant future refinements of priorities achievable10. However, further collecting is still needed for a very large number of wild relatives in order to assemble sufficient samples to perform such genetic assessments and to help resolve taxonomic and gene pool assignment uncertainties33.

Methods Methods used for gathering data, modelling, analyses and the associated references are available in the Supplementary Information.

Interactive maps displaying occurrence data coordinates, potential distribution models, further collecting priority maps and collecting priority categories for the crop wild relatives analysed are available at http://www.cwrdiversity.org/ distribution-map/. Occurrence data used for this analysis are available at http:// www.cwrdiversity.org/checklist/cwr-occurrences.php. Further information on expert evaluations of the gap analysis are available at http://www.cwrdiversity.org/ expert-evaluation/.

Received 04 September 2015; accepted 05 February 2016; published online 21 March 2016

References 1. Tilman, D. & Clark, M. Global diets link environmental sustainability and

human health. Nature 515, 518–522 (2014). 2. Khoury, C. K. et al. Increasing homogeneity in global food supplies and the

implications for food security. Proc. Natl Acad. Sci. USA 111, 4001–4006 (2014). 3. Foley, J. A. et al. Solutions for a cultivated planet. Nature 478, 337–342 (2011). 4. Godfray, H. C. J. et al. Food security: the challenge of feeding 9 billion people.

Science 327, 812–818 (2010). 5. Tilman, D., Balzer, C., Hill, J. & Befort, B. L. Global food demand and the

sustainable intensification of agriculture. Proc. Natl Acad. Sci. USA 108, 20260–20264 (2011).

6. Ray, D. K., Mueller, N. D., West, P. C. & Foley, J. A. Yield trends are insufficient to double global crop production by 2050. PLoS One 8, e66428 (2013).

7. Cordell, D., Drangert, J.-O. & White, S. The story of phosphorus: global food security and food for thought. Global Environ. Chang. 19, 292–305 (2009).

8. Asseng, S. et al. Rising temperatures reduce global wheat production. Nature Clim. Change 5, 143–147 (2015).

9. Zhu, Y. et al. Genetic diversity and disease control in rice. Nature 406, 718–722 (2000).

10. McCouch, S. et al. Feeding the future. Nature 499, 23–24 (2013). 11. Esquinas-Alcázar, J. Protecting crop genetic diversity for food security: political,

ethical and technical challenges. Nature Rev. Genet. 6, 946–953 (2005). 12. Hajjar, R. & Hodgkin, T. The use of wild relatives in crop improvement: a survey

of developments over the last 20 years. Euphytica 156, 1–13 (2007). 13. Xiao, J., Grandillo, S., Ahn, S. N., McCouch, S. R. & Tanksley, S. D. Genes from

wild rice improve yield. Nature 384, 356–358 (1996). 14. Gur, A. & Zamir, D. Unused natural variation can lift yield barriers in plant

breeding. PLoS Biol. 2, e245 (2004). 15. McCouch, S. R. et al. Through the genetic bottleneck: O. rufipogon as a source of

trait-enhancing alleles for O. sativa. Euphytica 154, 317–339 (2007). 16. Tanksley, S. D. & McCouch, S. R. Seed banks and molecular maps: unlocking

genetic potential from the wild. Science 277, 1063–1066 (1997). 17. Guarino, L. & Lobell, D. B. A walk on the wild side. Nature Clim. Change 1,

374–375 (2011). 18. McCouch, S. R., McNally, K. L., Wang, W. & Hamilton, R. S. Genomics of gene

banks: a case study in rice. Am. J. Bot. 99, 407–423 (2012). 19. Vincent, H. et al. A prioritized crop wild relative inventory to help underpin

global food security. Biol. Conserv. 167, 265–275 (2013).

Overlapping high-priority taxa for collecting Overlapping

crop gene pools 1–6

1–6 7–

12 13

–18 19

–2 4

7– 12

13 –18

19 –2

3 25 –3

0 31

–3 6

37 –4

3

Hotspots for collecting taxa associated with various crops

Figure 4 | Proposed hotspots for further collecting activities for high-priority crop wild relatives. The map displays geographic regions where high-priority crop wild relative taxa are expected to occur and have not yet been collected and conserved in gene banks. The inset map shows gaps for under-represented taxa by crop gene pool. Dark red indicates greater overlap of potential distributions of under-represented taxa, where greater numbers of under-represented crop wild relative taxa occur in the same geographic area. For the inset map, greater numbers indicate greater overlap of taxa associated with various crops.

NATURE PLANTS DOI: 10.1038/NPLANTS.2016.22 ARTICLES

NATURE PLANTS | www.nature.com/natureplants 5

© 2016 Macmillan Publishers Limited. All rights reserved

20. Food and Agriculture Organization of the United Nations (FAO) The Second Report on the State of the World‘s Plant Genetic Resources for Food and Agriculture (Commission on Genetic Resources for Food and Agriculture, FAO, 2010).

21. Jarvis, A., Lane, A. & Hijmans, R. J. The effect of climate change on crop wild relatives. Agric. Ecosyst. Environ. 126, 13–23 (2008).

22. Wilkes, G. Urgent notice to all maize researchers: disappearance and extinction of the last wild Teosinte population is more than half completed. A modest proposal for Teosinte evolution and conservation in situ: the Balsas, Guerrero, Mexico. Maydica 52, 49–58 (2007).

23. Brummitt, N. A. et al. Green plants in the red: a baseline global assessment for the IUCN sampled Red List Index for plants. PLoS One 10, e0135152 (2015).

24. Dempewolf, H. et al. Adapting agriculture to climate change: a global initiative to collect, conserve, and use crop wild relatives. Agroecol. Sustain. Food Syst. 38, 369–377 (2013).

25. Ramírez-Villegas, J., Khoury, C., Jarvis, A., Debouck, D. G. & Guarino, L. A gap analysis methodology for collecting crop genepools: a case study with Phaseolus beans. PLoS One 5, e13497 (2010).

26. Vavilov, N. I. Centers of origin of cultivated plants. Bull. Appl. Bot. Plant Breed. 16, (1926).

27. Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B. & Kent, J. Biodiversity hotspots for conservation priorities. Nature 403, 853–858 (2000).

28. Secretariat of the Convention on Biological Diversity (CBD). Decision X/2. The Strategic Plan for Biodiversity 2011-2020 and the Aichi Biodiversity Targets (2010).

29. Food and Agriculture Organization of the United Nations (FAO). International Treaty on Plant Genetic Resources for Food and Agriculture (2009).

30. Secretariat of the Convention on Biological Diversity (CBD). Nagoya Protocol on Access to Genetic Resources and the Fair and Equitable Sharing of Benefits Arising from their Utilization to the Convention on Biological Diversity (2011).

31. Wiersema, J. H., León, B. & Garvey, E. J. in ISHS Acta Horticulturae 948 I International Symposium Wild Relatives of Subtropical Temperate Fruit and Nut Crops (eds Aradhya, M. K. & Kluepfel, D. A.) 285–288 (ISHS, 2012).

32. Eckert, C. G., Samis, K. E. & Lougheed, S. C. Genetic variation across species’ geographical ranges: the central-marginal hypothesis and beyond. Mol. Ecol. 17, 1170–1188 (2008).

33. Goodwin, Z. A., Harris, D. J., Filer, D., Wood, J. R. I. & Scotland, R. W. Widespread mistaken identity in tropical plant collections. Curr. Biol. 25, R1066–R1067 (2015).

Acknowledgements We thank J. Wiersema and B. León for major contributions to taxonomic concepts; the herbaria, gene banks, researchers and other sources that contributed occurrence data to the analysis (Supplementary Table 3); the expert evaluators of gap analysis results (Supplementary Table 4); S. Calderón, I. Vanegas, H. Tobón, D. Arango, H. Dorado and E. Guevara for data inputs and processing; and S. Prager for comments. This work was undertaken as part of the project ‘Adapting Agriculture to Climate Change: Collecting, Protecting and Preparing Crop Wild Relatives’, which is supported by the Government of Norway. The project is managed by the Global Crop Diversity Trust and the Millennium Seed Bank of the Royal Botanic Gardens, Kew, and implemented in partnership with national and international gene banks and plant breeding institutes around the world. For further information, visit the project website: http://www.cwrdiversity.org/. Funding was also provided by the CGIAR Research Program on Climate Change, Agriculture, and Food Security, Cali, Colombia.

Author contributions N.P.C.-A., C.K.K., H.D., R.J.E., L.G., A.J., N.M., J.M., J.R-V. and J.T. conceived and designed the study. N.P.C.-A., C.K.K., H.D., R.J.E., R.H.H., A.J., N.M., J.R-V., C.C.S. and H.V. acquired and contributed data. N.P.C.-A., C.K.K., H.A.A., V.B. and C.C.S. processed the data, performed the analyses and analysed the results. N.P.C.-A., C.K.K., H.D., R.J.E., L.G., A.J., N.M. and J.M. interpreted the results and wrote the manuscript. N.P.C.-A., C.K.K., V.B., H.D., R.J.E., L.G., A.J., N.M., J.M., J.R-V. and P.C.S. edited the manuscript.

Additional information Supplementary information is available online. Reprints and permissions information is available online at www.nature.com/reprints. Correspondence and requests for materials should be addressed to N.P.C.-A.

Competing interests The authors declare no competing financial interests.

ARTICLES NATURE PLANTS DOI: 10.1038/NPLANTS.2016.22

NATURE PLANTS | www.nature.com/natureplants6

© 2016 Macmillan Publishers Limited. All rights reserved

R E V I E W

Quality of core collections for effective utilisation of genetic resources review, discussion and interpretation

T. L. Odong • J. Jansen • F. A. van Eeuwijk •

T. J. L. van Hintum

Received: 8 February 2012 / Accepted: 16 August 2012 / Published online: 15 September 2012

� The Author(s) 2012. This article is published with open access at Springerlink.com

Abstract Definition of clear criteria for evaluation of the

quality of core collections is a prerequisite for selecting

high-quality cores. However, a critical examination of the

different methods used in literature, for evaluating the

quality of core collections, shows that there are no clear

guidelines on the choices of quality evaluation criteria and

as a result, inappropriate analyses are sometimes made

leading to false conclusions being drawn regarding the

quality of core collections and the methods to select such

core collections. The choice of criteria for evaluating core

collections appears to be based mainly on the fact that

those criteria have been used in earlier publications rather

than on the actual objectives of the core collection. In this

study, we provide insight into different criteria used for

evaluating core collections. We also discussed different

types of core collections and related each type of core

collection to their respective evaluation criteria. Two new

criteria based on genetic distance are introduced. The

consequences of the different evaluation criteria are illus-

trated using simulated and experimental data. We strongly

recommend the use of the distance-based criteria since they

not only allow the simultaneous evaluation of all variables

describing the accessions, but they also provide intuitive

and interpretable criteria, as compared with the univariate

criteria generally used for the evaluation of core collec-

tions. Our findings will provide genebank curators and

researchers with possibilities to make informed choices

when creating, comparing and using core collections.

Introduction

Ex-situ germplasm collections have increased enormously

in number and size over the last three to four decades as a

result of global efforts to conserve plant genetic resources

for food and agriculture. The large sizes of many of these

collections, either individually or collectively for a given

species complicate the characterisation, evaluation, util-

isation and maintenance of the conserved germplasm. The

approach of forming core collections was introduced to

increase the efficiency of characterisation and utilisation of

collections stored in the genebanks, while preserving as

much as possible the genetic diversity of the entire col-

lection (Frankel 1984; Brown 1989). Frankel (1984)

defined a core collection as a limited set of accessions

representing, with minimum repetitiveness, the genetic

diversity of a crop species and its wild relatives. From the

original definition, several operational definitions have

been coined (see Brown 1995 and van Hintum et al. 2000).

Core collections have many roles to play in the man-

agement and use of genetic resources. Genebank curators

have the responsibility for conservation, regeneration,

safety duplication, documentation, evaluation and charac-

terisation as well as facilitating access to the genetic

resources in their collections. These activities often require

them to make choices or to set priorities among accessions

because of limited resources (Brown 1995). Because a core

Communicated by R. Varshney.

Electronic supplementary material The online version of this article (doi:10.1007/s00122-012-1971-y) contains supplementary material, which is available to authorized users.

T. L. Odong (&) � J. Jansen � F. A. van Eeuwijk Biometris, Wageningen University and Research Centre,

P.O. Box 100, 6700 AC Wageningen, The Netherlands

e-mail: [email protected]

T. J. L. van Hintum

Centre for Genetic Resources, The Netherlands (CGN),

Wageningen, The Netherlands

123

Theor Appl Genet (2013) 126:289–305

DOI 10.1007/s00122-012-1971-y

collection is smaller in size compared to the whole col-

lection, it enables some operations of the genebank, such as

evaluation (of the selected accessions), to be handled more

efficiently and effectively. The reduced size of a core

collection is a key to its manageability and, in many cases

the representation of the total collection’s diversity enables

the core to function as a reference set of accessions for the

whole collection (Brown and Spillane 1999).

Since the inception of the idea of core collections over

two decades ago, a body of literature on the theory and

practice of core collections has accumulated. Very many

approaches for selecting core collections have been pro-

posed and used (e.g. M-Strat (Gouesnard et al. 2001),

Genetic distance sampling (Jansen and van Hintum 2007),

PowerCore (Kim et al. 2007) and CoreHunter (Thachuk

et al. 2009)). In comparing the options for assembling a

core collection, one of the challenges is to decide on the

evaluation criteria for the quality of the result. Various

criteria for determining the quality of a core collection

have been suggested in the literature, yet very little atten-

tion has been given to the analysis of these quality criteria.

In fact every researcher appears to have his/her own criteria

for the evaluation of core collections.

Thus there is a need to clearly define criteria for the

evaluation of the quality of core collections and to relate

the different types of core collections to those criteria. For

example, a core collection formed for the purpose of cap-

turing accessions with rare or extreme values of the desired

trait(s) (e.g. high resistance to pest or high yield) should be

evaluated differently from one formed with the intention of

representing the (pattern of) genetic diversity in the col-

lection. By the pattern of genetic diversity we refer to the

genetic differences among all the accessions which have

been accumulated as a result of natural processes, species’

characteristics and historical events.

The debate whether to have a single or several core

collections for a given genebank collection is an old but

still an interesting one (see Mackay 1995). The initial idea

behind a core collection favoured the creation of a fixed

core collection, possibly modified in time to accommodate

new knowledge and new diversity (Brown 1989). How-

ever, there is evidence from the literature to suggest that

genebanks are creating different core collections to repre-

sent specific sections of their germplasm collections, e.g.

Chilean bean core collections (Paredes et al. 2010) and

Iberian peninsula common bean core collections (Rodino

et al. 2003) and to cater for specific projects. As pointed out

by Mackay (1995), to support better the use of available

germplasm, sets of diverse accessions need to be estab-

lished with different selection criteria in mind. This idea is

best captured by a computer programme ‘‘core selector’’

developed at Centre for Genetic Resource, The Netherlands

(CGN), where a user is allowed to select online a

maximum diversity subset of accessions that meets his/her

selection criteria (van Hintum 1999). These selections

could be considered core collections since they are repre-

sentative of the genetic variation of a larger group of

germplasm accessions. This concept of objective driven

core collection deviates from the original idea of the core

collection. Brown (1995) recommended that such objective

driven subsets of accessions should have name tags that

indicate their purposes rather than call them core (e.g. acid

tolerant set). Irrespective of the name, it is clear from lit-

erature that these objective driven diverse selections are

quite popular. Recent developments in computer science,

molecular biology and biochemistry suggest that the gen-

eration, storage and processing of data from germplasm

will cease to be a limiting factor when creating diverse

selections.

It should be noted that we are in no way suggesting that

the fixed core collection no longer has merits; the compi-

lation of information on representative samples of a given

collection still adds value to all accessions. The mini-core

collections and reference sets (Odong et al. 2011b;

Upadhyaya et al. 2009) as initiated by the Generation

Challenge Program of the CGIAR are good examples of

core collections serving that purpose. It is important to note

that irrespective of the type of core collections, appropriate

optimisation and evaluation criteria should be used in

creating and evaluating these selections.

In this paper, we will (1) discuss the different types of

core collections and proposed criteria appropriate for

quality evaluation of each type of core collection; (2)

discuss the different criteria used in the literature for

evaluating the quality of core collections and relate each

criterion to the different types of core collections; (3) use

real data sets (molecular marker data) to illustrate the

performance of the proposed quality evaluation criteria

with respect to the different types (and purposes) of core

collections. The outcome of our study will allow

researchers and curators to make informed choices from a

set of alternative approaches.

What is a good core collection?

One of the key goals of defining a core collection is the

efficient utilisation of available genetic resources and this

is best achieved by having clear objectives in mind when

selecting entries for the core (Mackay 1995). The answer to

the question ‘‘what is a good core collection’’ therefore

depends on the objectives for making the core. This can be

‘‘conserving as much variation (phenotypic or genotypic)

as possible in as few as possible accessions’’ or ‘‘optimis-

ing the chance of finding a new allele’’. A second question

is how to measure quality of the core collection, and this

will depend on the type of data available for evaluation.

290 Theor Appl Genet (2013) 126:289–305

123

According to Brown (1989), a good core collection

should have no redundant entries (an entry is an accession

included in the core), represent the whole collection with

regards to species, subspecies and geographical regions and

should be small enough to be easily managed. It was

suggested by Marita et al. (2000) that core collections can

be created with two general purposes in mind (1) maxi-

mising the total genetic diversity in a core (as sometimes

favoured by taxonomists, and geneticists as well as gene-

bank curators) and (2) maximising the representativeness

of the genetic diversity of the whole collection in a core

collection (as sometimes favoured by plant breeders).

Accordingly, maximising the representativeness of genetic

diversity implies also the inclusion of broadly adapted and

heterotic materials containing ‘generalist’ alleles in a core

collection. Earlier, Galwey (1995) stated the above two

purposes of core collections in a slightly different way as:

(1) maximising the representativeness of the full range of

variation present in the whole collection; (2) maximising

the representativeness of the pattern of variation present in

the whole collection.

There is also an aspect of balance between representing

total diversity and the usefulness of the core to the intended

user (Brown 1995). This can be illustrated with some

examples. If a breeder searches for accessions with a spe-

cific trait of interest (e.g. acid tolerance), it is likely that the

best core collection should contain relatively more material

from the primary genepool (see Harlan and de Wet 1971

for description of different types of genepools) as com-

pared to the secondary or tertiary genepool, irrespective of

the amount of diversity within the primary genepool since

there will be a strong preference for material in an adapted

genetic background. Although the chances of getting a rare

allele might be higher in the secondary or tertiary genepool

compared to primary genepool, it is probably cheaper to

evaluate more accessions from the primary genepool than

to use material from the tertiary or secondary genepool in

the breeding program. If a core collection is created in the

search for new resistances, the part of the whole collection

originating for example from area(s) that in the past had

shown to contain resistances should obviously be over-

represented. This implies that the user is often not pri-

marily interested in maximising diversity per se (which

could result in core collections with mainly wild and exotic

material), but rather in optimising the chance of finding

accessions that he/she is looking for as material which is

relatively easy to use in for instance a breeding pro-

gramme. To achieve this, the selection of a core collection

often starts with stratifying accessions into homogeneous

groups (a group can be collection of accessions with sim-

ilar characteristics, e.g. phenotype, genotypes or region of

origin), followed by an arbitrary determination of the

number of accessions to be selected from each group, the

so-called allocation. When a core collection is being

formed for a specific user, the stratification and allocation

processes can be used to ensure that accessions from

(a) particular group(s) (e.g. primary gene pool, modern

varieties or Ethiopian landraces) are given more priority

than justified by the genetic variation contained in that

group. Since each user or curator most often define their

own methods for stratification (dividing accessions into

groups) and for determining the number of accessions to

select from each group, it is difficult to setup uniform

criteria for evaluating those objective driven core

collections.

From the literature, it is not clear how to relate the

purpose of the core collections with the various quality

evaluation criteria, and only very few authors have

attempted this (e.g. Thachuk et al. 2009). Based on the

purposes of core collections as suggested by Galwey

(1995) and Marita et al. (2000), we have identified three

broad types of core collection which will be discussed in

the next section. We relate each of the three types of core

collections with their respective evaluation criteria.

Types of core collections

Based on the purposes for which they are formed core

collections can generally be classified into three types or

categories (i.e. core collections representing (1) individual

accessions; (2) extremes; and (3) distribution of accessions

in the whole collection). In defining the types of core

collections, the term ‘‘accessions’’ refers to elements that

constitute the whole collection and ‘‘entries’’ are elements

of the core collection. Since the core collection is a

selection from the whole collection, all entries are acces-

sions, but only few accessions are entries.

Type 1

A core collection representing the individual accessions of

the whole collection (CC–I). In this case each entry in the

core collection represents one (itself) or more accessions

that jointly make up the whole collection. Each accession

in the whole collection is represented by an entry in the

core which is most similar to it.

This type of core collection (CC–I) aims at a uniform

representation of the original genetic space, with equal

weights across this space and is the most intuitive way of

looking at core collection (see Fig. 1). A core collection of

type CC–I is especially suitable, for situations requiring an

overview of the genetic diversity of the accessions of the

whole collection. Core collections formed for the purposes

of maximising the representativeness of genetic diversity

as suggested by Marita et al. (2000) can be placed in type

Theor Appl Genet (2013) 126:289–305 291

123

CC–I. By ensuring that all accessions in the entire collec-

tion are maximally represented, core collections of type

CC–I provide the best option for obtaining a single ‘‘multi-

purpose’’ or generalist core collections compared to any

type of core collection.

Type 2

A core collection representing the extremes of the whole

collection (CC–X). Implication the diversity of the traits of

the entries of the core collection is maximised.

A core collection of type CC–X is geared towards rep-

resenting the ranges of phenotypes, genotypes or alleles of

the whole collection. A good core collection of type CC–X

has entries that are as different as possible from each other.

A core collection representing the total genetic diversity, as

suggested by Marita et al. (2000), can be considered as a

core collection of type CC–X.

Type 3

A core collection representing the distribution of accessions

of the whole collection (CC–D). In this case, when creating a

core collection one ensures that the proportion of accessions

in that core collection reflects the numerical contributions of

the different regions or categories to the whole collection.

For example, if the majority of the accessions come from a

given geographical region, then the core collection should

adequately reflect the importance of that region.

Implication the distributions of all relevant traits over

the entries of the core are similar (in terms of mean, var-

iance, quartiles, frequencies) to those of the whole

collection.

In our opinion, this core collection of type CC–D is only

of interest if the aim is to give an overview of the com-

position of the whole collection using only a part of the

collection. This type of core collection will be obtained by

maximising the representativeness of the pattern of varia-

tion of traits in the whole collection, as suggested by

Galwey (1995).

Based on the criteria used for the evaluation of core

collections in literature, it appears that either most of the

core collections are intended to be of type CC–D or they

were evaluated with inappropriate criteria (Diwan et al.

1994), sesame core collection: China (Xiurong et. al 2000),

Iberia Peninsula common bean (Rodino et al. 2003),

groundnut (Upadhyaya 2003), peanut (Valencia) (Dwivedi

et al. 2008), USDA soybean core (Oliveira et al. 2010).

This could be an indication of the desire of researchers to

have a single ‘‘multi-purpose’’ core collection from which

one could extract materials for different purposes. It should

be noted that by insisting on selecting a core collection that

reproduces the distribution of traits in the whole collection

one ignores the issue of redundancies and over represen-

tation. As we stated earlier, for researchers aiming at a

‘‘multi-purpose’’ core collection the CC–I type of core

collection would be the much better option compared to the

CC–D type.

Fig. 1 a Multimodal trait distribution of for whole

collection; b distribution of the same trait for a collection of

type CC–I; c distribution of the same trait for a core collection

of type CC–X; d distribution of the same distribution for a

collection of type CC–D

292 Theor Appl Genet (2013) 126:289–305

123

The different types of core collections have been illus-

trated graphically using a multimodal univariate distribu-

tion for the whole collection (Fig. 1).

Quality criteria for evaluating core collections

The process of evaluating a core collection usually

involves a comparison with the whole collection from

which it has been obtained, or a comparison with alterna-

tive core collections (core collected created using different

methods). This requires clear and objective criteria for

assessing the quality of the different types of core

collections.

Irrespective of the type of core collection and the quality

criterion used, the evaluation of quality should be based if

possible on data (traits or characteristics) that were not

used in the selection of the core (van Hintum et al. 2000).

This might sound like an obvious statement, but it is very

often neglected (e.g. Tai and Miller 2000; Wang et al.

2007). For example, one has a dataset of 1,000 accessions

each genotyped with 50 markers, and the objective is to

create a core collection of 20 entries with maximal allelic

richness. If it would concern only the current 50 markers,

this would be a simple optimisation problem. However, the

question is, ‘‘what if the core collection should also be

‘allelic rich’ for all loci that were not genotyped?’’ One

option would be to use half the markers for creating core

collections, and the other half for evaluating the quality of

the resulting core collection(s) (for a good examples see

Mckhann et al. (2004); Ronfort et al. (2006); Balfourier

et al. (2007)). Once the best strategy has been determined

this strategy could then be used on the entire set of markers

to create the final core collection. Since often molecular

data will be used to select a core that is also supposed to

optimise the representation of phenotypic diversity, rele-

vant phenotypic traits should be used for the validation as

well.

In this article, we place emphasis on evaluation criteria

that are based on genetic distances between accessions. The

main advantage of using genetic distance for evaluation of

core collections is that unlike the other criteria used in

literature which handle one variable at a time, all the

variables are used simultaneously. It is also easier and

more intuitive to link distances to the concept of genetic

diversity.

Evaluation of type CC–I

A good criterion for the evaluation of a CC–I core should

be able indicate how well each accession in the whole

collection is represented in the core collection. This

involves establishing the relationships between each

accession in the whole collection with the entries of the

core collection. The relationship between accessions and

entries is best represented by genetic distances between.

For CC–I, we proposed a criteria based on distances

between each accession in the whole collection and the

nearest entry in the core collection (A–NE) (see Fig. 2a, b).

Average distance between each accession and the nearest

entry (A–NE) (Odong et al. 2011b)

In this case, the distance between each accession and the

nearest entry in the core is calculated and averaged over all

the accessions. For the selected accessions (entries) these

Fig. 2 a Eight accessions (1, 2, …, 8) in a 2D space with all pairwise distances (the distance between accession i and j is indicated as Di-j). b The three selected entries (highlighted accessions) based on the

A–NE criterion, minimising the average distance between each

accession and it nearest neighbouring entry (D1-2 ? D2-2 ? D3-3 ? D4-2 ? D5-6 ? D6-6 ? D7-6 ? D8-6)/8

Theor Appl Genet (2013) 126:289–305 293

123

distances are taken as zero (they are closest to themselves).

For example, the value A–NE for Fig. 2 is given as:

A�NE

¼ ðD1�2 þ D2�2 þ D3�3 þ D4�2 þ D5�6 þ D6�6 þ D7�6 þ D8�6Þ

8

where Di�j (i, j = 1, 2, …, n; n is the number of accessions in the whole collection).

For core collections of type CC–I, the value of A–NE

should be as small as possible; the maximum representa-

tion (A–NE = 0) is obtained when each accession is rep-

resented by itself or by an identical duplicate accession in

the core. In core collections that optimise the values of

A–NE (CC–I type of core), the accessions selected as

entries tend to be those at the centres of clusters (i.e.

groups) rather accessions on the outer layer of the clusters.

Evaluation of type CC–X

A good criterion for a core collection of type CC–X (rep-

resenting the extreme values) should be able to quantify

differences between entries of the core collection as well as

being able to measure the inclusion or exclusion of

accessions with extreme values of the relevant traits in the

core. The most intuitive criteria for determining differences

between entries in the core collection are those criteria

based on pairwise distances. The exclusion or inclusion of

accessions with extremes values in the core can be assessed

using frequencies of traits or alleles captured (see Thachuk

et al. 2009). Below we propose a new criterion based on

distances between an entry and the nearest neighbouring

entry (E–NE) and compare it with criteria based on average

pairwise distances between all entries.

Average distance between each entry and the nearest

neighbouring entry (E–NE)

According to this criterion (E–NE), a good core collection

is one that maximises the average distance between each

entry and the nearest neighbouring entry in the core col-

lection. For this criterion, each entry should be as different

as possible from each other. This avoids selecting a few

clusters of similar accessions at the extreme ends of the

distribution, that might occur if one chooses a set of entries

that maximises the average of all pairwise distances

between the entries in the core (E–E) (see Fig. 4). When

calculating E–NE only a subset of pairwise distances

between the entries are used. Using example in Fig. 3, if

accessions 1, 3 and 7 are selected as entries in the core

collection, and if we assume that; (i) entry 1 is the nearest

neighbouring entry to both 3 and 7 (D3�1\D3�7 and D7�1\D7�3); and (ii) entry 3 is the nearest neighbour to entry 1 (D1�3\D1�7) then E–NE is given as:

E�NE ¼ ðD1�3 þ D3�1 þ D7�1Þ

3

where Di�j (i, j = 1, 2, …, n; n is the number of accessions in the whole collection)

Average genetic distances between entries (E–E)

Maximising the average genetic distance between entries

of a core collection has been suggested as a desired quality

criterion for evaluating core collections intended for plant

breeders (Franco 2006, Thachuk et al. 2009). Using

example in Fig. 3, E–E are given as:

E�E ¼ ðD1�3 þ D1�7 þ D3�7Þ

3

Fig. 3 a Eight accessions (1, 2, …, 8) in a 2D space with all pairwise distances

(the distance between accession

i and j is indicated as Di-j). b The three selected entries (highlighted accessions) based

on the E–NE criterion

maximising distances between

each entry and the nearest

neighbouring

(D1-3 ? D3-1 ? D7-1)/3

294 Theor Appl Genet (2013) 126:289–305

123

where Di�j (i, j = 1, 2, …, n; n is the number of accessions in the whole collection).

Figure 4 provides a simple numeric and graphical

comparisons of the three distance-based criteria discussed

above. Although both E–E and E–NE are suitable for the

CC–X type of core, as illustrated in Fig. 4c core collections

with a high average distance between the entries (E–E) can

still have a high level of redundancies. It is clear from

Fig. 4 that despite having the highest E–E (0.573 vs. 0.491

and 0.467) the core collection in Fig. 4c, some entries in

Fig. 4c are too close to each other to be included in a core

collection as reflected by a low value of E–NE. Figure 4a

indicates that minimisation of A–NE leads to the selection

of accessions from the centres of clusters compared to E–E

and E–NE which select accession at the periphery of

clusters.

Evaluation of type CC–D

Ideal criteria for evaluating a core collection of type CC–D

should be able to compare many distributional aspects

simultaneously: centre (mean, mode), spread (variance,

range), shape (symmetry, skewness, number of modes) and

unusual features (gaps, presence of outliers) of all data

simultaneously. For continuous data, we propose the use of

quantile–quantile plots (Gnanadesikan and Wilks 1968)

which provide a visual comparison for two data sets using

several distributional aspects of the data simultaneously.

We also recommend the use of Kullback–Leibler distance

(Kullback and Leibler 1951) which measures the distance

between probability distributions and can be used to

compare the difference in probability distribution between

the core collection and the whole collection. A brief

description of Kullback–Leibler distance is presented in

Electronic Supplementary Material (Appendix 1).

QQ plot

Compared to simple comparison of means or variances the

QQ plot gives a much better overall visual view of how the

distribution of a given trait differs between the core col-

lection and the whole collection. A QQ plot is a graphical

method for comparing two probability distributions by

plotting corresponding quantiles against each other. If the

two distributions are similar, the points in the QQ plot will

lie approximately on a straight line. A QQ plot is generally

a more powerful approach for comparing distributions than

the common technique of comparing histograms of the two

Fig. 4 Examples of core collections, showing the effect

of optimisation of different

criteria on the positioning of

entries (red stars) within the distribution of accessions

(circle) for each core collection, the value of all three evaluation

criteria are given: a the average distance between each accession

and the nearest entry (A–NE) is

minimised (E–E = 0.467;

E–NE = 0.180;

A–NE = 0.038) b the average distance between an entry and

the nearest other entry (E–NE)

is maximised (E–E = 0.491;

E–NE = 0.241;

A–NE = 0.056) c the average distance between entries (E–E)

is maximised (E–E = 0.573;

E–NE = 0.118; A–

NE = 0.094). Thus, for E–E

and E–NE, the larger the value

the higher the quality of the core

collection, the opposite is true

for A–NE

Theor Appl Genet (2013) 126:289–305 295

123

samples, but requires more skills for correct interpretation.

A more quantitative approach for comparing the distribu-

tion of the traits in the whole collection and the core would

be to calculate the Kullback–Leibler distance between the

core collection and the whole collection. Figure 5 shows

QQ plots for the three core collections types shown in

Fig. 1. We have also used the information from QQ plot to

calculate the Kullback–Leibler distance between the dif-

ferent core collections in Fig. 1 and the whole collection.

Common methods used for evaluating core collections

in the literature

Below we give an overview of the various criteria for

evaluating core collections used in the literature and relate

them to the three types of core collection. Given that the

type of data determines how diversity in the whole col-

lection or the core collection should be quantified, we will

also try to relate the evaluation criteria to the different

types of data (see Table 1 for brief descriptions of different

types of data that are being used for selecting and evalu-

ating the quality of core collections). It should be noted that

when evaluating the quality of core collections, most

authors apply several evaluation criteria despite the fact

that those criteria are only suitable for specific aspects of

core collections. The most common criteria used for

evaluating core collections include criteria based on sum-

mary statistics, the Shannon diversity index, class/category

coverage and Chi-square tests of association (see Table 2

below for summary).

Summary statistics

Criteria based on mean, variance and other summary sta-

tistics such as coefficient of variation, range, inter-quartile

range have been used mainly to evaluate the quality of core

collections based on continuous traits (Hu et al. 2000; Tai

and Miller 2000). It involves statistical tests of differences

between means, variances and other summary statistics of

the core and the whole collection. Based on the results of

statistical tests (mainly t tests and F tests) performed on

each trait separately, several evaluation criteria (mean

difference percentage, variance difference percentage,

coincidence rate of change and variable rate of coefficient

of variation, sign test) have been suggested (see Table 3).

Criteria based on means and variances are probably

Fig. 5 QQ plots for different types of core collections shown in Fig. 1. From both the QQ plots and Kullback distance, it is clear that

the distribution of whole collection is best represented by type 1 (CC–

D) core. The Kullback–Leibler distance (Kullback Dist) was

calculated based on values generated by the QQ plot. Random

sampling core collection is only based on 1 data set. The minimum

value of Kullback–Leibler distance is zero (for a core collection with

identical distribution to that of the whole collection)

296 Theor Appl Genet (2013) 126:289–305

123

suitable for the evaluation of a core collection of type

CC–D and will perform very poorly with core collections

of types CC–I and CC–X.

Some authors have questioned the use of differences

between means and variances of core and whole collection

as criteria for evaluating the quality of core collections

(e.g. Kim et al. 2007). There is also a conceptual problem

when comparing a core collection and a whole collection.

Statistically a core collection is a sample from the whole

collection (i.e. a population). Thus the question is not

Table 1 Brief description of of data types used for creating and evaluating core collections

Several types of information can be used for selecting core collections. The most common type of data are (i) passport data (ii) agronomic data

and (iii) molecular marker data

Passport data

Passport data are data about the identity and origin of an accession, including its taxonomic classification, with connected knowledge about

domestication, distribution, breeding history, cropping pattern and utilisation. Example of passport data include the country of origin, the crop

type (e.g. winter or summer wheat), and pedigree

Agronomic data

Agronomic data can be continuous, discrete or categorical. Examples of continuous variables include grain yield, plant height, leaf area, etc.

Discrete variables deal with counts such as the number of fruits or the number of seeds in a pod. Categorical variables may be defined as binary

(presence or absence of a given characteristic), nominal (colour or shape of an organ) or ordinal (a visual scale arranged to represent intensity,

colour or size) (Crossa and Franco 2004). Agronomic traits are usually controlled by multiple genes and typically by environmental factors

Molecular data

Data from molecular or biochemical marker systems can be treated as either continuous (allele frequency) or categorical (presence or absence of

band or allele). Examples of popular molecular data types include those generated by single nucleotide polymorphism (SNP), amplified

fragment polymorphism (AFLP), random amplified polymorphic DNA (RAPD), and simple sequence repeats (SSR)

Table 2 Summary of common methods (criteria) used for evaluation of the quality of core collections in literature

Criteria Type of

variables

General comments

Summary statistics Continuous Compare the mean, variance, etc. of the core with that of whole collection

Comparison is done for one variable at a time and later combined

Mainly suitable for CC–D type of core collections

Principal component

analysis

Continuous Plot of the coordinates of the entries on the main principal components (exploratory) to show spatial

distribution of entries and accessions

Compare two core collections using sum of squares of the their scores along the major PCs (Suitable for

CC–X core type)

Shannon diversity

Index (SH) b

Categorical The highest value is obtained when all the categories in the whole collection are represented in equal

proportion (penalizes redundancy at the category level)

The value of SH of a given core collection should be compared with the maximum possible value (log

(n), where n is the number of classes in the whole collection)

Most authors apply this criterion inappropriately by comparing SH value of the core collection with that

of the whole collection

Suitable for CC–I core type

Class coverage b

Categorical The highest value (1 or 100 %) is obtained when all the categories in the whole collection represented in

the core

Unlike SH it does not correct for redundancy in the core collection

Suitable for CC–I core type

Chi-square goodness-

of-fit b

Categorical This criterion has been used to test for the deviation of the frequency distributions of important

categorical traits between core collection and the whole collection

A good core collection is one in which the frequency distribution of the categories of the core is not

statistically different from that of the whole collection

Suitable for CC–D core type

a For all criteria except Principal component, the criterion is calculated for each variable at a time and later combined

b Can be applied to ccontinuous variables by first putting values into specific number of classes (determining the number classes is challenging)

Theor Appl Genet (2013) 126:289–305 297

123

whether these two samples are different, but could this

sample (core collection) have come from this population

distribution (i.e. the whole collection)? So we should be

dealing with a one-sample test and not a two-sample test. It

is thus clear that the use of QQ plot (Gnanadesikan and

Wilks 1968) and probability distribution based methods

such as the Kullback–Leibler distance (Kullback and Lei-

bler 1951) would be the best option for evaluation of CC–D

types of core collections.

Apart from the criteria described in Table 3, the phe-

notypic correlation coefficient of different traits has also

been used as a criterion for evaluating the quality of core

collections (Reddy et al. 2005; Mahajan et al. 2007). The

pairwise phenotypic correlation coefficients between dif-

ferent traits are calculated separately for the core collection

and whole collection and the values are then compared in

order to determine whether the associations between traits

have between conserved well enough in the core collection.

Principal component analysis

Another exploratory criterion for evaluating core collec-

tions involves the inspection of the spatial distribution of

the entries in plots of principal components (Bisht et al.

1998; Kang et al. 2006, Mahajan et al. 2007). Based on the

method suggested by Noirot et al. (1996), it is possible to

compare two core collections or relate the core collection

with the whole collection based on the sum of squares of

the scores of the entries on the major principal components:

the greater the value, the more diverse the core collection.

This criterion would be suitable for evaluation of core

collections of type CC–X. However, it should be noted that

a core with a higher value for this criterion can still have a

high level of redundancy resulting from the inclusion of

two or more similar accessions from the extreme ends of

the distribution.

Shannon diversity Index (SH)

This criterion is suitable for evaluating core collections

using categorical data; it has been used extensively in the

literature. For a given trait, the Shannon diversity index

(Shannon 1948) is calculated as follows:

SH ¼� Xn

i¼1 pi logðpiÞ

where pi is the frequency of the category i and n is the total

number of categories. The SH penalizes redundancy at the

category level and its maximum value (log(n)) is obtained

when all classes are represented in equal proportions

(i.e. p1 ¼ p2 ¼ �� � ¼ pn ¼ 1=n). Therefore, in terms of SH, the best core collection should be the one with the maxi-

mum attainable value which makes SH a suitable criterion

for core collections of type CC–I. Note that the whole

collection will never attain the maximum possible value of

SH because of redundancy associated with it. A core

Table 3 Common criteria for evaluating the quality of core collections based on summary statistics

Criteria Description

Mean difference percentage (MD)

(Hu et al. 2000 a )

MD ¼ St n

� � � 100 where St is the number of traits with a significant difference between the means of

the whole collection and the core collection; n is the total number of traits. The lower (\20 %) the value of MD the more representative the core collection

Variance difference percentage (VD)

(Hu et al. 2000) VD ¼ St

n

� � � 100 where St is the number of traits with a significant difference between the variances

of the whole collection and the core collection; n is the total number of traits. The larger ([80 %) the value of VD, the more diverse the core collection

Coincidence rate of range (CR)

(Diwan et al. 1995) CR ¼ 1

n

Pn

i¼1

RCðiÞ RWðiÞ � 100

where RCðiÞ and RWðiÞ represent the ranges of the ith trait in the core collection and the whole

collection, respectively; n is the total number of traits

Variable rate of coefficient of

variation (VR) (Hu et al. 2000) VR ¼ 1

n

Pn

i¼1

CVCðiÞ CVWðiÞ

� 100, where CVCðiÞ and CVWðiÞ represent the coefficients of variation of the ith trait

in the core collection and the whole collection, respectively; n is the total number of traits

The Sign (? versus -) test

(Basigalup et al. 1995,

Tai and Miller 2000)

X2 ¼ N1 � N2ð Þ2= N1 þ N2ð Þ. where N1 is the number of variables for which the mean or variance of the core collection is greater than the mean or variance of the whole collection (number of ? signs);

N2 is the number of variables for which the mean or variance of the core collection is less than the mean or variance of the whole collection (number of - signs). The values of X2 should be compared with a Chi-square distribution with 1 degree of freedom

a For a core collection to be representative of the whole collection, the value of MD should not be more than 20 % and the value of CR should be

greater than 80 % (Hu et al. 2000)

298 Theor Appl Genet (2013) 126:289–305

123

collection should be expected to have higher SH values as

compared to the whole collection. Similarity of these SH

values is not an indication of a good core collection, con-

trary to what is often concluded in the literature (e.g. Bisht

et al. 1998; Upadhyaya 2003; Mahalakshmi et al. 2007;

Dwivedi et al. 2008; Upadhyaya et al. 2009).

To apply SH or other measures of diversity to continu-

ous agronomic data, the data should first be converted into

categorical data by putting them into a specific number of

classes.

Class coverage (Coverage)

This criterion reports the percentage or proportion of the

categories in the whole collection that have been retained

in a core collection (Kim et al. 2007). It is defined by:

Coverage ¼ 1

K

XK

k¼1

ACore AWcol

! � 100

where ACore is the sets of categories in the core collection

and AWcol is the sets of classes found in the whole collec-

tion and K is the number of traits. According to this cri-

terion, a good core collection should retain all categories of

a given variable in the whole collection. For the case of

molecular marker data, the categories represent the number

of distinct alleles (akin to allelic richness) in the whole

collection. Class coverage is also a suitable quality crite-

rion for core collections formed with the purpose of ade-

quately representing the accessions in the whole collection

(type CC–I). When this criterion is applied to molecular

markers it will be suitable for core collections aimed at

capturing accessions with rare alleles (type CC–X).

It should be noted that unlike SH, coverage does not

take into consideration the differences in frequency of the

categories represented in the core collection so a core

collection with high coverage can still have a high level of

redundancy. Just like with SH, deciding on the number of

categories (intervals for continuous data) is a major chal-

lenge when calculating coverage.

Chi-square goodness-of-fit

This criterion has been used to test for the deviation of the

frequency distributions of important categorical traits

between core collection and the whole collection (Tai and

Miller 2000; Grenier et al. 2000; Zeuli and Qualset 1993).

Chi-square goodness-of-fit can also be used for continuous

agronomic data converted into categorical data. The Chi-

square values can be computed as:

v2 ¼ Xk

i¼1

ðCFreqi � WCFreqiÞ 2

ðWCFreqiÞ

where CFreqi is the relative frequency of accession from

category i (i ¼ 1; 2; . . .; k) in the core collection and WCFreqi is the relative frequency of accessions from cat-

egory i in the whole collection. The number of degrees of

freedom being the number categories (classes) minus one.

This test (Chi-square) is only suitable when the interest is

in representing the distribution of traits of accessions in the

whole collection (type CC–D).

From the literature, it clear that criteria based on sum-

mary statistics and SH are the most frequently used (see

Table 4). Since most core collections in literature are

evaluated using similar evaluation criteria, one would be

tempted to believe that all those cores were obtained with

the same objective(s) in mind. We highly doubt whether all

those core collections were indeed made with the same

objective(s) in mind.

Illustration using real data sets

Description of the datasets

We used two published data sets, i.e. coconut and common

bean (Odong et al. 2011a, b) to demonstrate the importance

of choosing the right criteria for each type of core collec-

tion (see below description of the data for details). In this

section we also demonstrate that a core collection which

optimises a given criterion in one dataset may not do well

when evaluated using another dataset.

Coconut (Cocos nucifera)

The coconut data consist of 1,014 accessions of coconut

accessions genotyped with 30 SSR markers. The acces-

sions were collected from different regions of the world:

West Africa (32), North America (52), South Asia (62),

Latin America (72), Central America and the Caribbean

(109), East Africa (124), South East Asia (183) and the

Pacific Islands (380). Coconut is a diploid, mainly out-

crossing species. Most of the accessions in this set were

indicated as tall; 43 dwarf accessions were present mainly

from South East Asia. Dwarf coconuts have a high degree

of self-fertilization. Because of its usefulness, coconut has

been extensively distributed around the world. For this

study, the coconut data were selected because it contained

larger numbers of accessions of each of the diverse origins

(a typical albeit virtual genebank germplasm collection).

Common bean (Phaseolus vulgaris)

The common bean data set consisted of 603 accessions

with 296 being described as Andean and 307 as

Theor Appl Genet (2013) 126:289–305 299

123

Mesoamerican types, genotyped with 36 SSR markers.

These accessions originated from 24 different countries

with most of them coming from Peru (184), Mexico (178),

Guatemala (61), Ecuador (35), Colombia (29) and Brazil

(22) and the remaining 18 countries contributed 94 acces-

sions. The common bean is a self-pollinating diploid spe-

cies. Twenty-nine of the 36 SSR markers used in study

belong to known linkage groups.

Comparing the performance of the evaluation of criteria

(A–NE, E–NE and E–E) when applied to different

types of core collections

The aim of this subsection is to demonstrate the importance

of using appropriate quality evaluation criteria when

evaluating the different types of core collection. We cre-

ated core collections of different sizes (5, 10, 15, …, 100) by optimising (minimising or maximising) each of the

three evaluation criteria (A–NE, E–NE and E–E). That is,

for each quality evaluation criterion we created 20 core

collections of different sizes and each core collection was

evaluated using the other two evaluation criteria which

were not used for creating it. For example core collections

created by maximising the value of E–NE are evaluated

using A–NE and E–E criteria. Core collections which

minimises A–NE were created using Genetic Distance

Optimisation method (Odong et al. 2011b) and those which

maximise E–E were created using Corehunter (Thachuk

et al. 2009). We wrote an R programme (available on

request from the authors) for creating core collections

which maximises E–NE. For comparison purposes random

sampling was also used to create core collection of the

same sample sizes (5, 10, 15, …, 100). For both coconut and common bean data, Fig. 6 shows

that in terms of A–NE (representing accessions in the

whole collection), core collections formed by maximisation

of E–NE or E–E perform even poorer than random sam-

pling. On the other hand, the performance of core collec-

tions formed by minimising A–NE performed poorly when

evaluated using E–NE or E–E criteria (see Figs. 7, 8). This

shows that when selecting a core collection, it is essential

to define the objectives clearly and the objectives should be

Table 4 Examples of Core collections from literature showing data and criteria used for their evaluating them

Paper (Core) Data use for

selection

Data use for

evaluation

Criteria use for evaluation

Soybean core collection (Oliveira et al. 2010) P, A, M A, M Summary statistics, Chi-square,

correlations

Sorghum mini-core (Upadhyaya et al. 2009) P, A, M P, A, M Summary statistics, Chi-square, SH,

correlation

Mini-core Japanese rice landraces (Ebana

et al. 2008)

Markers Markers, A Percentage of alleles retained, summary

statistics

Peanut (Valencia) (Dwivedi et al. 2008) P, A, M P, A, M Summary statistics, Chi-square, SH,

correlation

A worldwide bread wheat (Balfourier et al.

2007)

P, Markers P, Markers a

Alleles captured, countries of origins

represented

Pearl millet (Bhattacharjee 2007) P, A, M P, A, M Summary statistics, Chi-square, SH,

correlation

World sesame (Mahajan et al. 2007) P, A, M A, M Summary statistics, correlations, SH, PCA

West African yam Dioscorea spp.

(Mahalakshmi et al. 2007)

P, A, M A Summary statistics, correlation,

Chi-square, SH

USDA rice (Yan et al. 2007) P A, M a

Summary statistics, correlation

Korean Sesame core (Kang et al. 2006) P, A, M A, M Summary statistics, Chi-square PCA

Pigeon pea (Reddy et al. 2005) P, A, M P, A, M Summary statistics, Chi-square, SH,

correlation

Iberia Peninsula common beans (Rodino

et al. 2003)

P A, M Summary statistics, Chi-square

Groundnuts (Upadhyaya 2003) P, M M Summary statistics, Chi-square, SH,

correlation

Sesame -China (Xiurong et al. 2000) P, A, M A, M Summary statistics

Indian Mung Beans (Bisht et al. 1998) A, M M a

Summary statistics, PC, SH

Perennial Medicago (Basigalup et al. 1995) P, A, M A, M Summary statistics

Annual Medicago (Diwan et al. 1994) P, A, M P, A, M a

Summary statistics

A Agronomic data, M Morphological data, P Passport data, PCA Principle component analysis, SH Shannon Diversity Index a

Part or all the data used for the evaluation was different from the one used for forming the core collection

300 Theor Appl Genet (2013) 126:289–305

123

the basis for choosing the evaluation criteria. It is clear

from these two examples that if one evaluates a core col-

lection of type CC–X with inappropriate criteria (e.g. A–

NE instead of E–NE or E–E) there is a high likelihood of

drawing a wrong conclusion. The poor performance of core

collections formed by maximising A–NE (CC–I type of

core collection) when evaluated using E–NE and E–E

indicates the challenges of constructing a single ‘‘multi-

purpose’’ core collection from which one could extract

material of interest. However, the poor performance of the

core obtained by minimising A–NE when accessed using

E–NE or E–E does not mean that such core collections do

not have accessions with extreme characteristics. In CC–I

type of core collections accessions with extreme characters

are those one that represent themselves (i.e. CC–I core put

emphasis on both accessions with common and rare traits).

Fig. 6 Plot of Average distance between each accessions and its

nearest entry in the core (A–NE)

against different sizes of

collections formed by

optimising (minimising or

maximising) different criteria

(E–E, E–NE, A–NE and

Random sampling) using

Coconut (a) and Common beans (b)

Fig. 7 Plot of average distances between the entries in the core collection (E–E) (a) and average distance between an entry and the nearest neighbouring entry (E–NE) (b) against the size of core

collection for cores formed by optimising different criteria (E–E,

E–NE, A–NE and random sampling) for Coconut data (1,014

accessions)

Theor Appl Genet (2013) 126:289–305 301

123

We have shown in Figs. 7 and 8 that for both crops a

core collection that maximises E–NE also performs

(maximises) very well with respect to E–E but the reverse

is not always true (i.e. maximising E–E can result in a

much lower value of E–NE since similar accessions at the

extreme ends of the distributions can be included in the

core). In general, for both coconut and common beans, data

sets comparison based on E–E is less responsive to changes

within the core collection introduced by either changing the

number of entries (5–100) or changes in the optimisation

methods used for forming the core collection. For example,

for both crops (Figs. 7, 8), the changes in E–E between a

core with a size of 5 and a size of 100 ranges between 1.5

and 12 % compared to the changes in E–NE, which lies

between 18 and 54 %. The little response of E–E to

changes within the core collection is due to the fact that as

the core collection size increases, the average distance

between entries (E–E) tends towards the overall mean of

distances between accessions in the whole collection (the

E–E line of random sampling; Figs. 7a, 8a). This is a clear

indication that although both E–NE and E–E can be used

for evaluating the quality CC–X type of core collection,

E–NE appears to be more reliable.

Use of different data sets for evaluating core collections

In this subsection we show that a core collection obtained

by optimising a given criterion using one set of variables

(data set) may not be optimal for another set of variables.

The evaluation of a core collection with the same data set

that was used to create it ignores this simple but very

important point. This is quite important, especially in the

case of molecular markers data where the key assumption

is that by maximising diversity in a given set of marker

loci, the diversity of genes of interest will also be maxi-

mised. In this study we randomly divided the two datasets

into two equal datasets in terms of the number of molecular

markers (18 and 15 markers each for bean and coconut

datasets, respectively). For each crop, one half of the data

was used to form the core collection (training dataset) and

the other half used for evaluation of the resulting core

collection (evaluation set). For the evaluation set we first

determined the maximum (for E–NE) and minimum (A–

NE) possible value of the evaluation criteria. We referred

to this maximum or minimum possible value attainable

from evaluation set as Target (E–NE or A–NE), while

Actual (E–ENE or A–NE) values are obtained when core

collections that were created using the core/training set and

evaluated using the evaluation set. The core collections

formed were of the same samples sizes as those formed in

the previous subsection (5, 10, 15, …, 100). Randomly generated core collections of the same sizes were also

evaluated.

It is clear from Fig. 9 that major differences may occur

between the unknown value we intend to optimise (Target)

and the actual value obtained when the core is formed

using training set and evaluated using another set of data

(Actual). Although the core collections obtained by opti-

mising both E–NE and A–NE performed better than ran-

dom sampling in capturing unknown diversity, the

differences are quite small (5–15 % for E–NE and 1–5 %

for A–NE). A similar result was also observed with the

coconut data (see Electronic Supplementary Material:

Appendix 2). Ronfort et al. (2006) found very little gain in

Fig. 8 Plot of average distances between the entries in the core

collection (E–E) (a) and average distance between an entry and

the nearest neighbouring entry

(E–NE) (b) against the size of core collection for cores formed

by optimising different criteria

(E–E, E–NE, A–NE and random

sampling) for Common bean

(515 accessions) data

302 Theor Appl Genet (2013) 126:289–305

123

the total number of alleles captured using the H and M

strategy (Schoen and Brown, 1995) over random sampling

when evaluation was done using a different set of data.

The H strategy seeks to maximise the total number of

alleles in the core collection by sampling accessions from

groups in proportion to their within-group genetic diver-

sity. On the other hand, the M strategy examines all

possible core collections and singles out those that max-

imise the number of observed alleles at the marker loci.

Their (Ronfort et al. 2006) major explanation was that the

set of inbred lines used in the study had no redundancy,

leaving little room for optimisation to improve the results

over and above random sampling. The relatively small

gain in our case is probably due to limited size (number

of markers) and questionable quality of the data. For a

data set with limited structure, we expect little gain by

minimising A–NE compared to random sampling and this

could explain the small difference observed in the com-

mon bean data, i.e. splitting the common bean data into

two sets weakened the group structure of the data and

thus resulting in very little gain.

For both crops the correlation between distance

matrices formed by the two halves (core and evaluation)

of the data was not very high, i.e. bean (0.79) and coconut

(0.63). It is therefore not very surprising that when core

collections are created using one half of the data set they

perform rather poorly when evaluated with the other half

of the data set.

Conclusions and recommendations

A critical examination of the different methods for the

evaluating the quality of core collections used in the lit-

erature shows that the choices of criteria for evaluating

core collections are sometimes made arbitrarily, resulting

in false conclusions regarding the quality of core collec-

tions and the methods to select them. The criterion of

choice for evaluating the quality of core collections should

be determined by the objectives or type of the core col-

lection. If the core collection is made to represent the

accessions in the collection (CC–I), the evaluation criterion

should reflect that, and a criterion such as the A–NE we

proposed in this paper should be used. If the core is to

represent the range of genotypes and/or phenotypes in the

collection (CC–X), a criterion such as the E–NE should be

used. In addition, we stress that whenever possible or

appropriate, the evaluation of core collections should be

based on data that have not been used for the selection of

the accessions for the core collection. When the core col-

lection is intended for a specific user, the quality will have

to be determined in terms of fitness-for-use such as the ease

with which certain groups of material can be used or the

likelihood of finding traits of interest.

In summary, we introduced two genetic distance-based

criteria (A–NE and E–NE) for evaluating the quality of core

collections. We strongly recommend distance-based criteria

mainly for two reasons: (a) they combine information from

Fig. 9 Plot of average distance between an entry and the nearest neighbouring entry (E–NE) (a) and average distance between each accessions and its nearest entry in the core (A–NE) (b) against the size of core collection for bean data set. The bean data set was split

into two halves with one half used to form collection and the other

half used for evaluation of the core. Target (E–NE and A–NE) values

are the maximum (E–ENE) or minimum (A–NE) possible values for

each criterion for the half of the data used for evaluation (evaluation

set), while actual (E–ENE and A–NE) values are obtained from a core

collections that were using one half (training set) and evaluated using

the quality evaluation half of the data (evaluation set)

Theor Appl Genet (2013) 126:289–305 303

123

all traits simultaneously, instead of using one trait at a time

as most of the evaluation criteria used in literature do;

(b) they are intuitive, easy to interpret and relate to the

concept of representation of genetic diversity. These two

newly proposed distance-based criteria are suitable for

evaluating the two important types of core collections (CC–

I and CC–X). These evaluation criteria can also be used as

optimisation criteria when creating the core collections.

Acknowledgments This work was supported by the Generation Challenge Programme under GCP subprogram 5—Capacity Building

and Enabling Delivery. We would like to thank Carmen de Vicente,

leader subprogram 5 for availing us with the funds. We are also

grateful to people who participated in generating the data used in this

study especially Patricia Lebrun-Turquay (PI—coconut), Matthew

Blair (PI—Common bean). Finally, we would like to thank two

anonymous reviewers for their constructive suggestions for further

improvements.

Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, dis-

tribution, and reproduction in any medium, provided the original

author(s) and the source are credited.

References

Balfourier F, Roussel V, Strelchenko P, Exbrayat-Vinson F, Sourdille

P, Boutet G, Koenig J, Ravel C, Mitrofanova O, Beckert M,

Charmet G (2007) A worldwide bread wheat core collection

arrayed in a 384-well plate. Theor Appl Genet 114(7):1265–1275

Basigalup DH, Barnes DK, Stucker RE (1995) Development of a core

collection for perennial Medicago plant introductions. Crop Sci 35:1163–1168

Bhattacharjee R (2007) Establishment of a pearl millet (Pennisetum glaucum (L.) R. Br.) core collection based on geographical distribution and quantitative traits. Euphytica 155:35–45

Bisht IS, Mahajan RK, Patel DP (1998) The use of characterisation

data to establish the Indian mungbean core collection and

assessment of genetic diversity. Genet Resour Crop Evol 45(2):

127–133

Brown AHD (1989) Core collections—a practical approach to

genetic-resources management. Genome 31(2):818–824

Brown AHD (1995) The core collection at the crossroads. In:

Hodgkin T, Brown AHD, van Hintum TJL, Morales EAV (eds)

Core Collections of Plant Genetic Resources. John Wiley and

Sons, Chichester, pp 3–19

Brown AHD, Spillane C (1999) Implementing core collections—

principles, procedures, progress, problems and promise. In:

Johnson RC, Hodgkin T (eds) Core Collections for Today and

Tomorrow. Crop Science Society of America, Madison, pp 1–10

Crossa J, Franco J (2004) Statistical methods for classifying

genotypes. Euphytica 137:19–37

Diwan N, Gary Bauchan GR, McIntosh MS (1994) A core collection

for the United States Annula Medicago Germplasm collection. Crop Sci 34:279–285

Diwan N, McIntosh MS, Bauchan GR (1995) Methods of developing

a core collection of annual Medicago species. Theor Appl Genet 90:755–761

Dwivedi SL, Puppala N, Upadhyaya HD, Manivannan N, Singh S

(2008) Developing a core collection of peanut specific to

Valencia market type. Crop Sci 48:625–632

Ebana K, Kojima Y, Fukuoka S, Nagamine T, Kawase M (2008)

Development of mini core collection of Japanese rice landrace.

Breeding Sci 58:281–291

Franco J (2006) Sampling strategies for conserving maize diversity

when forming core subsets using genetic markers. Crop Sci

46(2):854–864

Frankel OH (1984) Genetic perspectives of germplasm conservation.

In: Arber WK et al (eds) Genetic manipulation: impact on man

and society. Cambridge University Press, Cambridge, England,

pp 161–170

Galwey NW (1995) Verifying and validating the representativeness of

a core collection. In: Hodgkin T, Brown AHD, van Hintum TJL,

Morales EAV (eds) Core Collections of Plant Genetic

Resources. John Wiley and Sons, Chichester, pp 187–198

Gnanadesikan R, Wilk MB (1968) Probability plotting methods for

the analysis of data. Biometrika 55(1):1–17

Gouesnard B, Bataillon TM et al (2001) MSTRAT: an algorithm for

building germ plasm core collections by maximizing allelic or

phenotypic richness. J Hered 92(1):93–94

Grenier C, Hamon P, Bramel-Cox PJ (2000) Assessment of genetic

diversity in three subsets constituted from the ICRISAT sorghum

collection using random vs non-random sampling procedures A.

Using morpho-agronomical and passport data. Theor Appl Genet

101(1–2):190–196

Harlan JR, de Wet JMT (1971) Toward a rational classification of

cultivated Plants. Taxon 20:509–517

Hu JJ, Zhu J, Xu HHM (2000) Methods of constructing core

collections by stepwise clustering with three sampling strategies

based on the genotypic values of crops. Theor Appl Genet

101(1–2):264–268

Jansen J, van Hintum TJL (2007) Genetic distance sampling: a novel

sampling method for obtaining core collections using genetic

distances with an application to cultivated lettuce. Theor Appl

Genet 114(3):421–428

Kang CW, Kim SY, Lee SW, Mathur PN, Hodgkin T, Zhou MD, Lee

RJ (2006) Selection of a core collection of Korean sesame

germplasm by a stepwise clustering method. Breeding Sci

56(1):85–91

Kim KW, Chung HK et al (2007) PowerCore: a program applying the

advanced M strategy with a heuristic search for establishing core

sets. Bioinformatics 23(16):2155–2162

Kullback S, Leibler RA (1951) On information and sufficiency. Ann

Math Stat 22(1):79–86

Mackay MC (1995) One core or many? In: Hodgkin T, Brown AHD,

van Hintum TJL, Morales EAV (eds) Core Collections of Plant

Genetic Resources. John Wiley and Sons, Chichester, pp 199–209

Mahajan RK, Bisht IS, Dhillon BS (2007) Establishment of a core

collection of world sesame (Sesamum indicum L.) germplasm accessions. Sabrao J Breed Genet 39:53–64

Mahalakshmi V, Ng Q, Atalobhor J, Ogunsola D, Lawson M, Ortiz R

(2007) Development of a West African yam Dioscorea spp. core collection. Genet Resour Crop Evol 54:1817–1825

Marita JM, Rodriguez JM et al (2000) Development of an algorithm

identifying maximally diverse core collections. Genet Resour

Crop Evol 47(5):515–526

McKhann HI, Camilleri C et al (2004) Nested core collections

maximizing genetic diversity in Arabidopsis thaliana. Plant J 38(1):193–202

Noirot M, Hamon S, Anthony F (1996) The principal component

scoring: a new method of constituting a core collection using

quantitative data. Genet Res Crop Evol 43:1–6

Odong TL, van Heerwaarden J, Jansen J, van Hintum TJL, van

Eeuwijk FA (2011a) Determination of genetic structure of

germplasm collections: are traditional hierarchical clustering

methods appropriate for molecular marker data. Theor Appl

Genet 123(2):195–205. doi:10.1007/s00122-011-1576-x

304 Theor Appl Genet (2013) 126:289–305

123

Odong TL, van Heerwaarden J, Jansen J, van Hintum TJL, van

Eeuwijk FA (2011b) Statistical techniques for defining reference

sets of accessions and microsatellite markers. Crop Sci

51(6):2401–2411

Oliveira MF, Nelson RL, Geraldi IO, Cruz CD, de Toledo JFF (2010)

Establishing a soybean germplasm core collection. Field Crops

Res 119:277–289

Paredes MC, Becerra VV, Tay JU, Blair MW, Gabriel BB (2010)

Selection of a representative core collection from the chilean

common bean germplasm. Chilean J Agri Res 70(1):3–15

Reddy LJ, Upadhyaya HD, Gowda CLL, Singh S (2005) Develop-

ment of core collection in pigeon pea [Cajanus cajan (L.) Millspaugh] using geographic and qualitative morphological

descriptors. Genet Resour Crop Evol 52:1049–1056

Rodino AP, Santalla M, Ron AMD, Singh SP (2003) A core

collection of common bean from the Iberian peninsula. Euphy-

tica 131:165–175

Ronfort J, Bataillon T, Santoni S, Delalande M, David J, Prosperi JM

(2006) Microsatellite diversity and broad scale geographic

structure in a model legume: building a set of nested core

collections for studying naturally occurring variation in Medi- cago truncatula. BMC Plant Biol 6:28. doi:10.1186/1471- 2229-6-28

Schoen DJ, Brown AHD (1995) Maximising genetic diversity in core

collections of wild relatives of crop species. In: Hodgkin T,

Brown AHD, van Hintum TJL, Morales EAV (eds) Core

Collections of Plant Genetic Resources. John Wiley and Sons,

Chichester, pp 55–76

Shannon CE (1948) A mathematical theory of communication. Bell

Syst Tech J 27:379–423

Tai P, Miller JD (2000) A core collection for Saccharum spontaneum L. from the world collection of sugarcane. Crop Sci 41(3):

879–885

Thachuk C, Crossa J, Franco J, Dreisigacker S, Warburton M,

Davenport GF (2009) Core Hunter: an algorithm for sampling

genetic resources based on multiple genetic measures. BMC

Bioinformatics 10:243

Upadhyaya HD (2003) Development of a groundnut core collection

using taxonomical, geographical and morphological descriptors.

Genet Resour Crop Evol 50:139–148

Upadhyaya HD, Pundir RPS, Dwivedi SL, Gowda CLL, Reddy VG,

Singh S (2009) Developing a mini core collection of sorghum for

diversified utilization of germplasm. Crop Sci 49:1769–1780

van Hintum TJL (1999) The core selector, a system to generate

representative selections of germplasm collections. Plant Genet

Resour Newslett 118:64–67

van Hintum TJL, Brown AHD, Spillane C, Hodgkin T (2000) Core

collections of plant genetic resources. IPGRI Technical Bullletin

No. 3. International Plant genetic Resources Institute, Rome

Wang JC, Hu J et al (2007) A strategy on constructing core

collections by least distance stepwise sampling. Theor Appl

Genet 115(1):1–8

Wright S (1978) Evolution and the Genetics of Populations: A treatise

in four volumes Volume IV. University of Chicago Press,

Chicago

Xiurong Z, Yingzhong Z, Yong C, Xiangyun F, Qingyuan G, Mingde

Z, Hodgkin T (2000) Establishment of sesame germplasm core

collection in China. Genet Resour Crop Evol 47:273–279

Yan W, Rutger JN, Bryant RJ, Bockelman HE, Fjellstrom RG, Chen

MH, Tai TH, McClung AM (2007) Development and evaluation

of a core subset of the USDA rice germplasm collection. Crop

Sci 47:869–878

Zeuli PLS, Qualset CO (1993) Evaluation of 5 strategies for obtaining

a core subset from a large genetic resource collection of Durum

wheat. Theor Appl Genet 87(3):295–304

Theor Appl Genet (2013) 126:289–305 305

123

RESEARCH ARTICLE

Into the vault of the Vavilov wheats: old diversity for new alleles

Adnan Riaz . Adrian Hathorn . Eric Dinglasan . Laura Ziems .

Cecile Richard . Dharmendra Singh . Olga Mitrofanova .

Olga Afanasenko . Elizabeth Aitken . Ian Godwin .

Lee Hickey

Received: 25 October 2015 / Accepted: 8 February 2016

� Springer Science+Business Media Dordrecht 2016

Abstract Intensive selection in wheat (Triticum

aestivum L.) breeding programs over the past

100 years has led to a genetic bottleneck in modern

bread wheat. Novel allelic variation is needed to break

the yield plateau, particularly in the face of climate

change and rapidly evolving pests and pathogens.

Landraces preserved in seed banks likely harbour

valuable sources of untapped genetic diversity

because they were cultivated for thousands of years

under diverse eco-geographical conditions prior to

modern breeding. We performed the first genetic

characterisation of bread wheat accessions sourced

from the N. I. Vavilov Institute of Plant Genetic

Resources (VIR) in St Petersburg, Russia. A panel

comprising 295 accessions, including landraces,

breeding lines and cultivars was subject to single seed

descent (SSD) and genotyped using the genotyping-

by-sequencing Diversity Arrays Technology platform

(DArT-seq); returning a total of 34,311 polymorphic

markers (14,228 mapped and 20,083 unmapped).

Cluster analysis identified two distinct groups; one

comprising mostly breeding lines and cultivars, and

the other comprising landraces. Diversity was bench-

marked in comparison to a set of standards, which

revealed a high degree of genetic similarity among

breeding material from Australia and the International

Maize and Wheat Improvement Center (CIMMYT).

Further, 11,025 markers (1888 mapped and 9137

unmapped) were polymorphic in the diversity panel

only, thus representing allelic diversity potentially not

present in Australian or CIMMYT germplasm. Open-

access to DArT-seq markers and seed for SSD lines

will empower researchers, pre-breeders and breeders

to rediscover genetic diversity in the VIR collection

and accelerate utilisation of novel alleles to improve

wheat.

Keywords Genetic diversity � Genotyping-by- sequencing � Landraces � Seed bank � Triticum � Wheat breeding

Electronic supplementary material The online version of this article (doi:10.1007/s10722-016-0380-5) contains supple- mentary material, which is available to authorized users.

A. Riaz � A. Hathorn � E. Dinglasan � L. Ziems � C. Richard � D. Singh � L. Hickey (&) Queensland Alliance for Agriculture and Food Innovation,

The University of Queensland, St Lucia, QLD 4072,

Australia

e-mail: [email protected]

O. Mitrofanova

N. I. Vavilov Institute of Plant Genetic Resources,

Saint Petersburg, Russia

O. Afanasenko

Department of Plant Resistance to Diseases, All Russian

Research Institute for Plant Protection, Pushkin,

Russia 196608

E. Aitken � I. Godwin School of Agriculture and Food Sciences, The University

of Queensland, St Lucia, QLD 4072, Australia

123

Genet Resour Crop Evol

DOI 10.1007/s10722-016-0380-5

Introduction

Bread wheat is a staple food crop that was domesti-

cated about 10,000 years ago in the Fertile Crescent of

Western Asia (Shewry 2009; Ray et al. 2013). Wheat

has a complex hexaploid genome (2n = 6x = 42)

contributed by three different progenitors. The first

hybridisation occurred between Triticum urartu

Thum. (genome AA) and Aegilops speltoides Tausch

(genome BB) and resulted in tetraploid Emmer

(Triticum turgidum L.) (genome AABB). The second

hybridisation event occurred between Emmer and

Aegilops tauschii Coss. (genome DD) resulting in

hexaploid bread wheat (AABBDD) (Salse et al. 2008).

During wheat domestication, a limited number of

hybridisation events occurred between progenitor

species, subsequently leading to a relatively narrow

genetic base in hexaploid wheat compared to its wild

relatives (Brenchley et al. 2012). Moreover, trait-

specific selection in wheat breeding programs has

further reduced levels of genetic diversity (Doebley

et al. 2006). Consequently, the rate of genetic gain for

yield and some desirable traits within the modern

wheat germplasm pool is approaching a plateau

(Grassini et al. 2013).

Landraces have arisen through a combination of

natural and artificial selection performed by farmers in

a specific environment, thus are highly adapted to

local conditions (Reif et al. 2005). However, landraces

were developed under a lower selection pressure in

comparison to modern cultivars, therefore collectively

display a broader genetic base (Cavanagh et al. 2013).

Immense genetic diversity for landrace collections has

been reported for many crops, including; wheat, maize

(Zea mays L.), sorghum (Sorghum bicolor L.), barley

(Hordeum vulgare L.), rice (Oryza sativa L.) and oats

(Avena sativa L.) (Cavanagh et al. 2013; Ignjatović-

Micić et al. 2013; Pineda-Hidalgo et al. 2013; Mace

et al. 2013; Pasam et al. 2014; Jilal et al. 2008; Huang

et al. 2010; Montilla-Bascón et al. 2013). Recent

studies have examined landrace collections and iden-

tified novel alleles for tolerance to abiotic and biotic

stress (McIntosh et al. 1998; Bansal et al. 2013; Jaradat

2013; Sthapit et al. 2014; Lopes et al. 2015; Maccaferri

et al. 2015), which could be used by wheat breeding

programs to improve yield stability.

Historical germplasm, such as landraces or old

cultivars, represent a potentially valuable source of

genetic variation (Motley 2006; Jones et al. 2008;

Bhullar et al. 2009; Cavanagh et al. 2013). However,

such material is rarely used in breeding programs,

which typically target elite 9 elite crosses to improve

the likelihood of developing higher yielding cultivars

(Baenziger and DePauw 2009). Fortunately, a propor-

tion of historical wheat germplasm has been main-

tained by gene banks. Approximately 850,000 viable

wheat samples are stored in 229 independent collec-

tions worldwide (Mitrofanova 2012). Recent studies

exploring the genetic diversity for landraces collected

in 1930s by renowned botanist A. E. Watkins have

identified new sources of disease resistance, for

instance leaf rust resistance genes Lr52 (Bansal et al.

2011) and Lr67 (Hiebert et al. 2010) and stripe rust

resistance gene Yr47 (Bansal et al. 2011). The

prominent Russian botanist and geneticist N. I. Vav-

ilov, best known for his theory relating to ‘‘the centres

of origin of cultivated plants’’ (Vavilov 1926), devoted

his life to the improvement of cereal crops. During the

early nineteenth century, Vavilov and his colleagues

travelled around the world collecting seeds, including

many wheat landraces, which were subsequently

stored in a seed bank in Leningrad, now known as

the N. I. Vavilov Institute of Plant Genetic Resources

(VIR) in St Petersburg, Russia. Vavilov’s collections

represent a ‘snap shot’ of landraces cultivated around

the world prior to modern breeding. Currently the VIR

seed bank consists of germplasm derived from almost

100 countries throughout Europe, Asia, Africa, Amer-

ica and Australia, where about 19 % of the accessions

are from various parts of Russia (Mitrofanova 2012).

The development of low-cost high-throughput

DNA marker systems, such as genotyping-by-se-

quencing (GBS) (Elshire et al. 2011; Poland and Rife

2012) or the 90 K single nucleotide polymorphisms

(SNP) platform (Wang et al. 2014), offers a cost-

effective way to explore the genetic diversity con-

tained in landrace collections. Previous studies have

examined VIR accessions for agronomic and disease

traits, such as plant height, resistance to leaf rust, dark

brown leaf spot-blotch and septoria glume blotch

(Tyryshkin and Tyryshkina 2003; Mitrofanova 2012);

however, genetic analysis has been mostly limited to

DNA markers specific for known genes. A whole-

genome approach, such as GBS or SNP markers,

would enable genome-wide association studies

(GWAS) to identify marker-trait associations and

discover novel alleles for desirable traits (Lopes et al.

2015; Sukumaran et al. 2015). This way, useful

Genet Resour Crop Evol

123

genetic variation can be efficiently introgressed into

modern wheat germplasm using marker-assisted

breeding strategies.

In this study, we assemble 295 bread wheat

accessions originally sourced from VIR. We geneti-

cally characterise this diversity panel using the GBS

Diversity Arrays Technology platform (i.e. DArT-seq)

and benchmark levels of genetic diversity using a set

of standards comprising modern cultivars and elite

breeding lines from Australia and the International

Maize and Wheat Improvement Center (CIMMYT).

We anticipate that open-access to this global diversity

panel, including DArT-seq marker profiles and seed

for single seed descent (SSD) lines, will enable GWAS

aiming to identify novel alleles for important target

traits and accelerate the use of genetically diverse

material from VIR in modern wheat breeding

programs.

Materials and methods

Plant materials

Two hundred and ninety-five bread wheat accessions

originally sourced from VIR were assembled to form a

globally diverse panel. The 295 accessions were

imported to Australia by the Australian Grains

Genebank in Horsham, Victoria, Australia (Electronic

Supplementary Material 1). The panel comprised 136

landraces, 32 cultivars, 10 breeding lines and 118

accessions with unknown classification in terms of

their cultivation status. The accessions were collected

from different geographical regions of the world

between 1922 and 1990. This panel also contains 56

accessions originally collected by A.E. Watkins,

which were donated and registered at VIR in 1934

and 1936. Of the 295 accessions, 206 have known

origin information, originating from 28 countries,

spanning 5 different continents of the world, includ-

ing; North America (4), South America (2), Africa (6),

Europe (69), and Asia (125) (Fig. 1). Although the

exact geographical origin of the remaining 89 acces-

sions was unknown—they were collected at the time

of the former Union of Soviet Socialist Republics

(USSR).

A set of standards comprising 20 cultivars and elite

breeding lines from CIMMYT and Australian wheat

breeding programs was assembled (Table 1), which

was used to benchmark the genetic diversity in the

panel of accessions from VIR.

Line purification

A single plant for each of the 295 VIR accessions and

20 standards was grown for line purification in a

temperature controlled glasshouse at The University

of Queensland, St Lucia, QLD, Australia. Seed bank

collections (e.g. landraces) often contain mixtures of

different genotypes (Newton et al. 2010), thus a single

random plant is not likely representative of the

diversity contained within each accession. Although,

this strategy aimed to maximise the number of

accessions sampled from, rather than sampling the

diversity within accessions. A generation of SSD was

used to develop genetically stable lines for subsequent

genotypic and phenotypic analyses. Seeds were

imbibed in trays filled with potting media comprising

composted pine bark fines and placed at 4 �C for 8 weeks to satisfy vernalisation requirements. Plants

were transplanted into 140 mm (1.4 L) ANOVAPot �

pots and grown under constant (24 h) light to accel-

erate plant development (Hickey et al. 2009; 2012).

During the growth cycle, notes were recorded for each

accession, including leaf hairiness, presence of awns,

and seed shape following harvest. The progeny from

each single plant was bulked and formed the pure seed

source for all future experiments.

Field evaluation

Pure seed for each accession in the diversity panel,

along with the set of standards, were sown in a field

nursery located at The University of Queensland

Research Station, Gatton, QLD, Australia. Un-repli-

cated hill plots were sown where each plot contained

six seeds. At 113 days after sowing (DAS) the

accessions were evaluated for growth habit and plant

height was recorded for genotypes exhibiting spring

growth habit.

Genotyping

Young leaf tissue was sampled from the single plant

selections and genomic DNA was extracted using the

CTAB (hexadecyl tri-methyl ammonium bromide)

method following the protocol recommended by

DArT (www.diversityarrays.com). A total of 315

Genet Resour Crop Evol

123

SSD lines (i.e. 295 VIR accessions and 20 standards)

were genotyped using the DArT-seq wheat PstI

microarray platform developed by DArT, Canberra,

Australia, as described by Li et al. (2015).

Analysis of genetic diversity and population

structure

Clustering of individuals was performed using the

partitioning around medoids (PAM) algorithm, the

most common implementation of the k-medoids

algorithm (Reynolds et al. 1992). The PAM algorithm

attempts to partition a population into k clusters based

on the levels of dissimilarity between individuals. The

optimal number of clusters is determined using a

graphical display called a silhouette plot. The algo-

rithm then finds a representative individual (called a

medoid) for each of the k clusters such that the average

dissimilarity of that medoid to all other members of its

cluster is minimised.

Clustering was initially performed for the 295

accessions forming the diversity panel using 34,311

dominant markers (SilicoDArTs). The ‘Jaccards dis-

tance’ (Jaccard 1908) between all 295 individuals was

then calculated using R stats package (R Core Team

2014). Using the resulting 295 9 295 dissimilarity

matrix, the optimal number of clusters (i.e. k = 2) was

estimated using the fpc package (Hennig 2014). This

estimate was then verified by running the PAM

algorithm for a range of cluster sizes (k = 1…5) and visually assessing the silhouette plots for each value of

k. The cluster package was used to run the PAM

algorithm and to generate the final biplots. Monomor-

phic markers were excluded from the analysis

although there were no restrictions placed on rare

alleles (i.e. alleles occurring at low frequency) as they

were considered important in determining genetic

diversity. The procedure was then repeated with the

standards included in the population to investigate the

diversity within the context of elite breeding lines and

cultivars from Australia and CIMMYT. Accessions

were classified according to their cultivation status

(i.e. cultivar, breeding line, landraces and unknown)

and geographic origin (i.e. continent) to explore trends

in genetic diversity based on the clustering analysis.

The 20 standards were also used to benchmark

genetic diversity by identifying ‘novel’ markers that

were only polymorphic in the diversity panel (i.e.

monomorphic in the standards). Markers were posi-

tioned based on the wheat DArT-seq consensus map

and displayed on chromosomes using Map Chart soft-

ware Version 2.2 (Voorrips 2002). To visualise the

distribution of novel markers at the genome level,

markers that were polymorphic only in the diversity

panel were coloured red and markers that were

polymorphic in the standards were coloured black.

Fig. 1 The geographical distribution of accessions with known origin in the diversity panel (206 out of 295)

Genet Resour Crop Evol

123

Results

Phenotypic diversity

During line purification in the glasshouse, the diversity

panel was evaluated for morphological characters. Of

the 286 accessions evaluated, 12.5 % displayed the leaf

hairiness trait. The majority of accessions (98.3 %)

displayed oblong seed shape, whereas only 1.7 %

displayed ovate (round) seed shape. The diversity panel

was also evaluated for the presence of awns, where

25.2 % were awnless, 4.9 % were apically awnleted,

2.4 % were awnleted and 65.08 % were awned. A

sample of phenotypic variation in awn morphology is

displayed in Fig. 2. Based on the classification defined

by Dorofeev et al. (1979), the original accessions from

which the pure lines were sampled, represented 5

species and 30 botanical varieties (Electronic Supple-

mentary Material 1). It should be noted that some SSD

lines developed in this study did not match the botanical

variety assigned for the original accession.For example,

original VIR accessions AUS38778 and AUS39503

both contain a mixture of var. graecum (Koern.) Mansf.

and var. pseudomeridionale (Flaksb.) Mansf.; however

according to our morphology results, the derived SSD

lines (i.e. WLA-017 and WLA-039, respectively)

belong to var. graecum.

In the field, the majority of accessions in the

diversity panel displayed a spring growth habit (i.e.

80.1 %), while 17.2 % displayed a significantly

delayed time to anthesis, indicating a winter growth

habit. A small number of accessions (3.7 %) failed to

germinate, thus were not included in phenotypic

analysis. A total of 237 spring type accessions were

observed, which included 61 accessions with unknown

origin (not presented in Fig. 3). The remaining 176

spring type accessions originated from 27 countries,

mostly from Russia (40), India (35) and Pakistan (32).

Among the 51 winter type accessions, 25 were of

unknown origin, while 26 originated from nine

Table 1 Pedigree information for the 20 standards from Australia and the International Maize and Wheat Improvement Center (CIMMYT)

Genotype Status Pedigree

Australia

Drysdale Cultivar HARTOG*3/QUARRION

EGA Gregory Cultivar PELSART/2*BATAVIA

EGA Wylie Cultivar QT2327/COOK//QT2804

Gladius Cultivar RAC-875/KRICHAUFF//EXCALIBUR/KUKRI/3/RAC-875/KRICHAUFF/4/

RAC-875//EXCALIBUR/KUKRI

Halberd Cultivar SCIMITAR/KENYA-C-6042//BOBIN/3/INSIGNIA-49

Mace Cultivar WYALKATCHEM/STYLET//WYALKATCHEM

QT14783 Breeding line KENNEDY*2/QT8766

RIL114 Breeding line UQ01484/RSY10//H45

Scout Cultivar SUNSTATE/QH71-6//YITPI

Suntop Cultivar SUNCO/2*PASTOR//SUN436E

Westonia Cultivar SPICA/TIMGALEN//TOSCA/3/CRANBROOK/BOBWHITE*2/JACUP

Yipti Cultivar C-8-MMC-8-HMM/FRAME

CIMMYT

Seri M82 Breeding line KAVKAZ/(SIB)BUHO//KALYANSONA/BLUEBIRD

SB062 Breeding line SERI M82/BABAX

ZWB10–37 Breeding line TACUPETO F2001/BRAMBLING//KIRITATI

ZWB11–11 Breeding line ATTILA*2/PBW65*2/5/KAUZ//ALTAR 84/AOS/3/MILAN/KAUZ/4/HUITES

ZWB11–105 Breeding line PFAU/SERI.1B//AMAD/3/WAXWING/4/BABAX/LR42//BABAX*2/3/KURUKU

ZWW10–50 Breeding line ONIX/4/MILAN/KAUZ//PRINIA/3/BAV92

ZWW10–128 Breeding line ESDA/KKTS

ZWW11–36 Breeding line EGA BONNIE ROCK/4/MILAN/KAUZ//PRINIA/3/BAV92

Genet Resour Crop Evol

123

countries, with the largest samples from Russia (11),

Ukraine (4) and Armenia (3) (Fig. 3).

Plant height was measured 113 DAS, at which time

most of the spring type accessions displayed growth

stagesrangingGS65toGS71(i.e.midfloweringtograin

filling). In contrast, winter type accessions were depict-

ing delayed growth, ranging GS21–GS29 (i.e. early to

late tillering). Considering this variation in maturity, the

plant height data for winter type accessions was

excluded from the analysis of the population distribu-

tion. The average height for the 237 spring type

accessions was 103.8 cm, ranging 55–165 cm. The

Australian and CIMMYT standards displayed an aver-

age height of 81.8 and 89.4 cm, respectively (Fig. 4).

Genetic diversity

Genotyping of the diversity panel and standards using

the DArT-seq GBS platform, returned a total of 56,306

SilicoDArTs, of which 34,311 were polymorphic. Of

the polymorphic markers, 14,228 were positioned on

the current DArT-seq consensus map, while 20,083

were unmapped and their chromosomal position was

unknown. Among the 14,228 mapped polymorphic

markers, 1888 were found polymorphic only in the

diversity panel, thus were considered ‘‘novel’’ in

comparison to the Australian and CIMMYT genotypes

(Fig. 5). Among the unmapped polymorphic markers

9137 were novel to the diversity panel. The Sili-

coDArTs provided good coverage across the cen-

tromeric and pericentromeric regions of the seven

homologous groups of chromosomes. A large portion

of the novel polymorphic markers were mapped to the

A and B genomes (32 and 43 %, respectively)

compared to the D genome (25 %). Also, marker

density was higher for the A and B genome chromo-

somes (2.11 and 3.14 markers per cM, respectively)

compared to the D genome chromosomes (1.7 markers

Fig. 2 A sample of phenotypic variation for awns in the diversity panel, where 1) T. aestivum var. aureum (Link) Mansf.;

2) T. aestivum var. pseudomeridionale (Flaksb.) Mansf.; 3) T.

aestivum var. ferrugineum (Alef.) Mansf.; 4) T. aestivum var.

heraticum (Vav. et Kob.) Mansf.; 5) T. spelta L.; 6) erythros-

permum (Koern.) Mansf.; 7) T. sphaerococcum Perc.

Genet Resour Crop Evol

123

per cM). The highest densities of novel polymorphic

markers were observed on chromosomes 2A, 2B, 3B,

6B and 7B and with considerably lower densities on

chromosomes 1A, 1D, 4D and 5D (Fig. 5). Further,

analysis of the DArT SNP data revealed low levels of

heterozygosity in SSD lines forming the diversity

panel, ranging from 0.7 to 1.8 % per chromosome

(Fig. 6).

Population structure

The silhouette method revealed the optimum number

of clusters (k = 2) for the diversity panel. The PAM

cluster analysis for two groups resulted in 171

accessions in cluster 1 and 124 accessions in cluster

2 (Fig. 7). The 42 reported cultivars and breeding

lines within the diversity panel were split across the

two clusters, with 34 accessions (81 %) in cluster 1.

The 136 reported landraces were also divided across

the two groups, with 90 accessions (66 % of lan-

draces) appearing in cluster 2. The population struc-

ture of the diversity panel was re-evaluated by adding

standards to the PAM cluster analysis (Fig. 8a, b). All

of the Australian and CIMMYT standards were

genetically similar and were positioned very close to

one another and were all grouped within cluster 1

(Fig. 8a, b). Cluster 2 mostly comprised landraces

(Fig. 8a).

Genetic diversity corresponding to geographic

origin

Most of the accessions from Europe, all South Amer-

ican accessions and those with unknown origin, were

grouped together in cluster 1, along with the standards

(Fig. 9). Most of the accessions from Asia and all

Fig. 3 The geographical distribution of diversity panel acces- sions with known origin (i.e. 202 out of 295) displaying spring

(blue) and winter (red) growth habits, along with standards from

Australia (brown) and the International Maize and Wheat

Improvement Center (pink). (Color figure online)

10 20 30 40 50 60 70 80 90 10 0

11 0

12 0

13 0

14 0

15 0

16 0

17 0

0

20

40

60

80

Plant height (cm)

N um

be r

of a

cc es

si on

s CIMMYT standards

Australian standards

Mean diversity panel

Fig. 4 Distribution of plant height for accessions in the diversity panel. Population mean indicated by the dotted line

(103.8 cm). The mean plant height for standards from Australia

and the International Maize and Wheat Improvement Center

(CIMMYT) are displayed by arrows (i.e. 81.8 and 89.4 cm,

respectively)

Genet Resour Crop Evol

123

accessions from Africa were also grouped in cluster 1

(Fig. 9). The North American accessions did not show

a clear pattern and were equally distributed across both

clusters (Fig. 9). The genetically diverse landraces in

cluster 2 were largely from Asia, mainly sourced from

India and Pakistan. Accessions with unknown origin

were found genetically similar to accessions from

Europe, most of which were sourced from Russia.

Fig. 5 Distribution of polymorphic markers based on the wheat DArT-seq consensus map. Black bands on chromosomes

indicate markers that were polymorphic in both the diversity

panel and set of standards, while red bands indicate novel

markers which are monomorphic in the set of standards and

polymorphic in the diversity panel. (Color figure online)

Genet Resour Crop Evol

123

Discussion

Through this study we have gained an insight of the

genetic diversity preserved in the wheat collection at

VIR in St Petersburg, Russia. A high degree of novel

alleles were observed in comparison to a set of

standards from Australia and CIMMYT. This diverse

collection includes accessions from 28 countries,

collected over a period spanning 70 years, presenting

a potentially valuable open-access genetic resource for

enriching diversity in modern breeding programs. We

anticipate this will accelerate discovery of novel

alleles for tolerance to abiotic and biotic stresses—

needed to improve wheat productivity with the onset

of climate change and anticipated new pests and

diseases.

Diversity in the panel

The diversity panel was genotyped with 56,306

SilicoDArTs, of which 14,228 polymorphic markers

had a chromosomal position, based on the current

DArT-seq consensus map.Of these, 1888 were deemed

novel to the diversity panel as they were monomorphic

in the standards from Australia and CIMMYT. These

novel markers were distributed across all 21 chromo-

somes, but in particular clusters of novel markers were

located on chromosomes 2A, 2B, 3B, 6B and 7B. It is

important to note that a large number of unmapped

polymorphic markers (i.e. 9137) were also novel in the

diversity panel. While cluster analysis used 34,311

markers (mapped and unmapped), the chromosomal

location of novel markers could only be investigated

Fig. 6 Box plot displaying the proportion of heterozygous SNP markers per chromosome in the diversity panel

Fig. 7 Biplot displaying results from cluster analysis of the 295 accessions in the diversity panel using the partitioning around

medoids (PAM) algorithm. Members of cluster 1 denoted by

circles and members of cluster 2 denoted by triangles. Colour

coding of accessions is based on the following classifications:

cultivars or breeding lines (red), landraces (blue) and ‘unknown’

(green). (Color figure online)

Genet Resour Crop Evol

123

using the subset of mapped markers. However, a

genetic map is not required to identify marker-trait

associations in wheat (Arief et al. 2014), thus the entire

set of polymorphic markers can be used in future

GWAS studies.

Higher marker densities were observed in the A and

B genome chromosomes (2.11 and 3.14 markers per

cM respectively) compared to the D genome chromo-

somes, which may be due to lower rates of recombi-

nation (Akbari et al. 2006; Allen et al. 2011; Cavanagh

et al. 2013; Nielsen et al. 2014). Wang et al. (2014)

used the 90 K SNP chip to genotype 726 wheat

accessions including landraces and found a similar

trend, where only 15 % of the reported markers were

in the D genome. Voss-Fels et al. (2015) also found

large non-polymorphic chromosomal sections in the D

genome, especially on 4D and 7D ([30 cM). The typically low genetic variation in the D genome of

modern wheats means that breeding efforts essentially

act to manipulate diversity largely in the A and B

genomes (White et al. 2008; Jia et al. 2013; Henry and

Nevo 2014; Voss-Fels et al. 2015). Accessions from

this diversity panel could be used to increase genetic

diversity particularly for the D genome in modern

germplasm.

In the future, we anticipate the development of an

improved DArT-seq consensus map and positioning of

unmapped polymorphic markers in this study. This

could improve marker density for the D genome

Fig. 8 a Biplot displaying results from cluster analysis of the 295 accessions in the diversity panel, plus the 20 standards from

Australia and the International Maize and Wheat Improvement

Center (CIMMYT), using the partitioning around medoids

(PAM) algorithm. Members of cluster 1 denoted by circles and

members of cluster 2 denoted by triangles. Colour coding of

accessions is based on the following classifications: diversity

panel accessions (green), Australian standards (blue) nd

CIMMYT standards (red). b Enlarged view of the 20 standards. (Color figure online)

Fig. 9 Biplot displaying results from cluster analysis of the 295 accessions in the diversity panel using the partitioning around

medoids (PAM) algorithm. Members of cluster 1 denoted by

circles and members of cluster 2 denoted by triangles.

Accessions were colour coded according to geographic origin:

Asia (black), Europe (purple), Africa (light blue), North

America (dark blue), South America (red) and Unclassified

(dark green). (Color figure online)

Genet Resour Crop Evol

123

chromosomes, in particular chromosomes 1D, 4D and

5D. Nevertheless, the mapped marker coverage using

the current wheat consensus map is adequate for

effective GWAS aiming to explore this genetic

resource for target traits. The large number of novel

markers highlights the high degree of diversity and

historical recombination among accessions. This cou-

pled with the use of high density SilicoDArTs will

enable precise positioning of QTL in future GWAS

studies.

It was clear that landrace accessions were genetically

more diverse than breeding lines and cultivars, which

tend to group together in the cluster analyses. The group

of most distinct landraces were those from India and

Pakistan, which grouped in the upper section of cluster 2

(Figs. 7, 9). Landraces from India and Pakistan thus

represent a great source of genetic variation for wheat

improvement. Although there was no clear trend in the

clustering of accessions based on growth habit (i.e.

spring and winter types) according to cultivation status,

although most winter type accessions originated from

Russia, Ukraine and Armenia. These countries experi-

ence extremely low temperatures during winter and also

relatively cool temperatures during the wheat growing

season (Schierhorn et al. 2014). The study by Cavanagh

et al. (2013) also found a lack of differentiation between

spring and winter wheats using whole-genome profiles.

This suggests that spring and winter wheats were

selected side-by-side in farmers’ fields and breeding

programs. Flowering time in wheat is a complex trait

and many different genetic factors can lead to early

flowering, thus such differences between spring and

winter genotypes may not be differentiated using a

whole-genome marker scan.

Australian and CIMMYT breeding material have

a narrow genetic base

Wheat breeding efforts for more than 100 years in

Australia have increased farm yield from 0.5 t/ha to

approximately 2 t/ha (Fischer 2009; Fischer et al.

2014). While a large improvement in yield was

achieved via the transition to semi-dwarf varieties

during the Green Revolution, the rate of gain for farm

yield has slowed to just 1 % per year (Fischer 2009;

Fischer et al. 2014). However, this estimate includes

both genetic gains resulting from breeding and

improved management practices. It seems wheat yield

around the world is beginning to plateau (Ray et al.

2013). While breeding strategies must improve to

meet future demands, the intensive selection per-

formed in modern breeding programs has resulted in

bottlenecks in terms of genetic diversity (Cavanagh

et al. 2013), which may restrict future genetic gains.

Since the early 1970s, CIMMYT material has been

extensively used in wheat breeding programs in

Australia. As a result, the majority of Australian

cultivars are either direct CIMMYT lines or contain

CIMMYT lines in their parentage (Brennan and Quade

2006). Of course, the set of 20 standards evaluated in

this study does not capture all diversity in modern

breeding programs around the world; nevertheless it

provides useful insight to gauge the diversity partic-

ularly within the context of wheat pre-breeding and

breeding efforts in Australia. Widespread utilisation of

CIMMYT material globally has led to significant yield

gains, but simultaneously resulted in narrowing the

genetic base of elite breeding material (Cavanagh et al.

2013). This can be problematic in the event of new

pests or pathogens. A recent example is the emergence

of a highly virulent stem rust pathotype Ug99 (Race

TTKSK), first detected in Uganda in 1998, which

rendered 90 % of wheat cultivars susceptible world-

wide (Singh et al. 2011).

The high degree of allelic variation in landraces can

be used to broaden the genetic base of modern wheat

germplasm and improve desirable traits (Smale et al.

2002; Reif et al. 2005; Lopes et al. 2015). Landraces

have contributed many agronomically important traits

in modern cultivars, such as the semi dwarfing gene

Rht8c and photoperiod insensitivity gene Ppd_D1

(formerly known as Ppd1) from the Japanese landrace

‘‘Aka Kamougi’’ (Worland et al. 1998; Ellis et al.

2007). Similarly, disease resistance genes have been

identified, such as leaf rust resistance gene Lr67 from

Pakistani landrace ‘‘PI250413’’ (Dyck and Samborski

1979; Hiebert et al. 2010; Herrera-Foessel et al. 2011).

While this study has genetically characterised 295

diverse wheat accessions from VIR, more accessions

could be genotyped and utilised for breeding, as a total

of 29,209 bread wheat accessions are currently

preserved at VIR (Mitrofanova 2012).

Exploiting the genetic resource

The diversity panel is currently being evaluated for

root architecture traits (seminal root angle and

Genet Resour Crop Evol

123

number) and resistance to key foliar diseases, includ-

ing; leaf rust, stripe rust, stem rust and yellow spot.

The next step is to perform GWAS to identify novel

alleles for these traits. This information could then be

used to profile the environments that contributed novel

alleles. This, in turn, would enable the identification of

similar environments from which germplasm could be

sampled to mine additional diversity from seed banks

using the Focused Identification of Germplasm Strat-

egy (FIGS) approach (Mackay 1995; Mackay and

Street 2004; Bhullar et al. 2009).

This diversity panel is an open-access resource

and available to researchers, pre-breeders and wheat

breeders. A small quantity of pure seed can be

requested from the Australian Grains Genebank

in Horsham, Victoria, Australia (contact:

[email protected]) and will be provided

under a Standard Material Transfer Agreement

(SMTA). The DArT-seq marker data is available upon

request from the corresponding author.

Acknowledgments This research was supported by an Early Career Research Grant and a Ph.D. scholarship from The

University of Queensland, Australia. We acknowledge M.s.

Raeleen Jennings and Dr. Mandy Christopher from the

Department of Agriculture and Fisheries for performing DNA

extractions. We also thank the Australian Grains Genebank

(Horsham, Victoria, Australia) and N. I. Vavilov Institute of

Plant Genetic Resources (St Petersburg, Russia) for providing

seed and passport information for the accessions examined in

this study. We thank Dr. Michael Mackay for providing

feedback on a draft version of this manuscript.

Compliance with ethical standards

Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial rela-

tionships that could be construed as a potential conflict of

interest.

References

Akbari M, Wenzl P, Caig V, Carling J, Xia L, Yang S, Uszynski

G, Mohler V, Lehmensiek A, Kuchel H, Hayden M, Howes

N, Sharp P, Vaughan P, Rathmell B, Huttner E, Kilian A

(2006) Diversity arrays technology (DArT) for high-

throughput profiling of the hexaploid wheat genome. Theor

Appl Genet 113:1409–1420

Allen AM, Barker GLA, Berry ST, Coghill JA, Gwilliam R,

Kirby S, Robinson P, Brenchley RC, D’Amore R,

McKenzie N, Waite D, Hall A, Bevan M, Hall N, Edwards

KJ (2011) Transcript-specific, single-nucleotide

polymorphism discovery and linkage analysis in hexaploid

bread wheat (Triticum aestivum L.). Plant Biotechnol J 9:

1086–1099

Arief VN, DeLacy IH, Dieters MJ, Basford KE (2014) Appli-

cation of marker-trait association profiles in simulating

plant breeding strategies. In: 15th Australasian plant

breeding conference, St Kilda

Baenziger PS, DePauw RM (2009) Wheat breeding: proce-

dures and strategies. In: Carver BF (ed) Wheat: Science

and trade. Wiley-Blackwell Publishing, Ames,

pp 275–308

Bansal UK, Forrest KL, Hayden MJ, Miah H, Singh D, Bariana

HS (2011) Characterisation of a new stripe rust resistance

gene Yr47 and its genetic association with the leaf rust

resistance gene Lr52. Theor Appl Genet 122:1461–1466

Bansal U, Arief V, DeLacy I, Bariana H (2013) Exploring wheat

landraces for rust resistance using a single marker scan.

Euphytica 194:219–233

Bhullar NK, Street K, Mackay M, Yahiaoui N, Keller B (2009)

Unlocking wheat genetic resources for the molecular

identification of previously undescribed functional alleles

at the Pm3 resistance locus. Proc Natl Acad Sci USA

106:9519–9524

Brenchley R, Spannagl M, Pfeifer M, Barker GLA, D’Amore R,

Allen AM, McKenzie N, Kramer M, Kerhornou A, Bolser

D, Kay S, Waite D, Trick M, Bancroft I, Gu Y, Huo N, Luo

M-C, Sehgal S, Gill B, Kianian S, Anderson O, Kersey P,

Dvorak J, McCombie WR, Hall A, Mayer KFX, Edwards

KJ, Bevan MW, Hall N (2012) Analysis of the bread wheat

genome using whole-genome shotgun sequencing. Nature

491:705–710

Brennan JP, Quade KJ (2006) Evolving usage of materials from

CIMMYT in developing Australian wheat varieties. Aust J

Agric Res 57:947–952

Cavanagh CR, Chao S, Wang S, Huang BE, Stephen S, Kiani S,

Forrest K, Saintenac C, Brown-Guedira GL, Akhunova A,

See D, Bai G, Pumphrey M, Tomar L, Wong D, Kong S,

Reynolds M, da Silva ML, Bockelman H, Talbert L,

Anderson JA, Dreisigacker S, Baenziger S, Carter A,

Korzun V, Morrell PL, Dubcovsky J, Morell MK, Sorrells

ME, Hayden MJ, Akhunov E (2013) Genome-wide com-

parative diversity uncovers multiple targets of selection for

improvement in hexaploid wheat landraces and cultivars.

Proc Natl Acad Sci USA 110:8057–8062

Doebley JF, Gaut BS, Smith BD (2006) The molecular genetics

of crop domestication. Cell 127:1309–1321

Dorofeev VF, Filatenko AA, Migushova EF, Udaczin RA,

Jakubziner MM (1979) Wheat. In: Dorofeev VF, Korovina

ON (eds) Flora of cultivated plants, vol 1. Kolos, Leningrad

(St. Petersburg), p 346 (In Russian)

Dyck PL, Samborski DJ (1979) Adult-plant leaf rust resistance

in PI 250413, an introduction of common wheat. Can J

Plant Sci 59:329–332

Ellis MH, Bonnett DG, Rebetzke GJ (2007) A 192 bp allele at

the Xgwm261 locus is not always associated with the Rht8

dwarfing gene in wheat (Triticum aestivum L.). Euphytica

157:209–214

Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K,

Buckler ES, Mitchell SE (2011) A robust, simple geno-

typing-by-sequencing (GBS) approach for high diversity

Genet Resour Crop Evol

123

species. PLoS ONE 6:e19379. doi:10.1371/journal.pone.

0019379

Fischer R (2009) Farming systems of Australia: exploiting the

synergy between genetic improvement and agronomy. In:

Sadras VO, Calderini D (eds) Crop physiology: applica-

tions for genetic improvement and agronomy. Elsevier,

Amsterdam, pp 23–54

Fischer RA, Byerlee D, Edmeades GO (2014) Crop yields and

global food security: will yield increase continue to feed

the world? ACIAR Monograph No. 158 (Australian Centre

for International Agricultural Research: Canberra, ACT),

pp 92–100

Grassini P, Eskridge KM, Cassman KG (2013) Distinguishing

between yield advances and yield plateaus in historical crop

production trends. Nat Commun. doi:10.1038/ncomms3918

Hennig C (2014) fpc: Flexible procedures for clustering. R

package version 2.1-9. http://CRAN.R-project.org/

package=fpc

Henry RJ, Nevo E (2014) Exploring natural selection to guide

breeding for agriculture. Plant Biotechnol J 12:655–662

Herrera-Foessel SA, Lagudah ES, Huerta-Espino J, Hayden MJ,

Bariana HS, Singh D, Singh RP (2011) New slow-rusting

leaf rust and stripe rust resistance genes Lr67 and Yr46 in

wheat are pleiotropic or closely linked. Theor Appl Genet

122:239–249

Hickey LT, Dieters MJ, DeLacy IH, Kravchuk OY, Mares DJ,

Banks PM (2009) Grain dormancy in fixed lines of white-

grained wheat (Triticum aestivum L.) grown under con-

trolled environmental conditions. Euphytica 168:303–310

Hickey LT, Wilkinson PM, Knight CR, Godwin ID, Kravchuk

OY, Aitken EAB, Bansal UK, Bariana HS, Delacy IH,

Dieters MJ (2012) Rapid phenotyping for adult-plant

resistance to stripe rust in wheat. Plant Breed 131:54–61

Hiebert CW, Thomas JB, McCallum BD, Humphreys GD,

DePauw RM, Hayden MJ, Mago R, Schnipenkoetter W,

Hayden M (2010) An introgression on wheat chromosome

4DL in RL6077 (Thatcher*6/PI 250413) confers adult

plant resistance to stripe rust and leaf rust (Lr67). Theor

Appl Genet 121:1083–1091

Huang X, Wei X, Sang T, Zhao Q, Feng Q, Zhao Y, Li C, Zhu C,

Lu T, Zhang Z, Li M, Fan D, Guo Y, Wang A, Wang L,

Deng L, Li W, Lu Y, Weng Q, Liu K, Huang T, Zhou T,

Jing Y, Li W, Lin Z, Buckler ES, Qian Q, Zhang Q-F, Li J,

Han B (2010) Genome-wide association studies of 14

agronomic traits in rice landraces. Nat Genet 42:961–967

Ignjatović-Micić D, Ristić D, Babić V, And̄elković V, Marković

K, Vančetović J (2013) Genetic assessment of maize

landraces from former Yugoslavia. Genetika 45:405–417

Jaccard P (1908) Nouvelles recherches sur la distribution florale.

Bull Soc Vaud Sci Nat 44:223–270

Jaradat AA (2013) Wheat landraces: a mini review. Emir J Food

Agric 25(1):20–29

Jia J, Zhao S, Kong X, Li Y, Zhao G, He W, Appels R, Pfeifer M,

Tao Y, Zhang X, Jing R, Zhang C, Ma Y, Gao L, Gao C,

Spannagl M, Mayer KFX, Li D, Pan S, Zheng F, Hu Q, Xia

X, Li J, Liang Q, Chen J, Wicker T, Gou C, Kuang H, He G,

Luo Y, Keller B, Xia Q, Lu P, Wang J, Zou H, Zhang R, Xu

J, Gao J, Middleton C, Quan Z, Liu G, Wang J, Yang H, Liu

X, He Z, Mao L, Wang J (2013) Aegilops tauschii draft

genome sequence reveals a gene repertoire for wheat

adaptation. Nature 496:91–95

Jilal A, Grando S, Henry RJ, Lee LS, Rice N, Hill H, Baum M,

Ceccarelli S (2008) Genetic diversity of ICARDA’s

worldwide barley landrace collection. Genet Resour Crop

Evol 55:1221–1230

Jones H, Lister DL, Bower MA, Leigh FJ, Smith LMJ, Jones

MK (2008) Approaches and constraints of using existing

landraces and extant plant material to understand agricul-

tural spread in prehistory. Plant Genet Res 6:98–112

Li H, Vikram P, Singh RP, Kilian A, Carling J, Song J, Bur-

gueno-Ferreira JA, Bhavani S, Huerta-Espino J, Payne T

(2015) A high density GBS map of bread wheat and its

application for dissecting complex disease resistance traits.

BMC Genom 16:216–230. doi:10.1186/s12864-015-1424-5

Lopes MS, El-Basyoni I, Baenziger PS, Singh S, Royo C, Ozbek

K, Aktas H, Ozer E, Ozdemir F, Manickavelu A, Ban T,

Vikram P (2015) Exploiting genetic diversity from

landraces in wheat breeding for adaptation to climate

change. J Exp Bot. doi:10.1093/jxb/erv122

Maccaferri M, Zhang J, Bulli P, Abate Z, Chao S, Cantu D,

Bossolini E, Chen X, Pumphrey M, Dubcovsky J (2015) A

genome-wide association study of resistance to stripe rust

(Puccinia striiformis f. sp. tritici) in a worldwide collection

of hexaploid spring wheat (Triticum aestivum L.). G3

5:449–465

Mace ES, Tai S, Gilding EK, Li Y, Prentis PJ, Bian L, Campbell

BC, Hu W, Innes DJ, Han X, Cruickshank A, Dai C, Frère

C, Zhang H, Hunt CH, Wang X, Shatte T, Wang M, Su Z,

Li J, Lin X, Godwin ID, Jordan DR, Wang J (2013) Whole-

genome sequencing reveals untapped genetic potential in

Africa’s indigenous cereal crop sorghum. Nat Commun.

doi:10.1038/ncomms3320

Mackay MC (1995) One core collection or many? In: Hodgkin

T, Brown AHD, Van Hintum TJL, Morales AAV (eds)

Core collections of plant genetic resources. Wiley,

Chichester, pp 199–210

Mackay MC, Street K (2004) Focused identification of germ-

plasm strategy–FIGS. In: Black CK, Panozzo JF, Rebetzke

GJ (eds) Proceedings of the 54th Australian cereal chemistry

conference and the 11th wheat breeders’ assembly. Royal

Australian Chemical Institute, Melbourne, pp 138–141

McIntosh RA, Hart GE, Devos KM, Gale MD, Rogers WJ

(1998) Catalogue of gene symbols for wheat. In: Pro-

ceedings of the 9th international wheat genetic symposium

5, Univ. Extension Press. Univ. Saskatchewan, Saskatoon

Mitrofanova OP (2012) Wheat genetic resources in Russia: current

status and prebreeding studies. Russ J Genet 2:277–285

Montilla-Bascón G, Sánchez-Martı́n J, Rispail N, Rubiales D,

Mur L, Langdon T, Griffiths I, Howarth C, Prats E (2013)

Genetic diversity and population structure among oat cul-

tivars and landraces. Plant Mol Biol Report 31:1305–1314

Motley TJ (2006) Crop plants: past present future. In: Motley

TJ, Zerega N, Cross H (eds) Darwin’s harvest: new

approaches to the origins, evolution and conservation of

crops. Columbia University Press, New York, pp 1–27

Newton AC, Akar T, Baresel JP, Bebeli PJ, Bettencourt E,

Bladenopoulos KV, Czembor JH, Fasoula DA, Katsiotis A,

Koutis K, Koutsika-Sotiriou M, Kovacs G, Larsson H,

Pinheiro de Carvalho MAA, Rubiales D, Russell J, Dos

Santos TMM, Vaz Patto MC (2010) Cereal landraces for

sustainable agriculture: a review. Agron Sustain Dev

30:237–269

Genet Resour Crop Evol

123

Nielsen NH, Backes G, Stougaard J, Andersen SU, Jahoor A

(2014) Genetic diversity and population structure analysis

of European hexaploid bread wheat (Triticum aestivum L.)

varieties. PLoS ONE 9(4):e94000. doi:10.1371/journal.

pone.0094000

Pasam RK, Sharma R, Walther A, Ozkan H, Graner A, Kilian B

(2014) Genetic diversity and population structure in a

legacy collection of spring barley landraces adapted to a

wide range of climates. PLoS ONE 9:e116164. doi:10.

1371/journal.pone.0116164

Pineda-Hidalgo KV, Méndez-Marroquı́n KP, Alvarez EV,

Chávez-Ontiveros J, Sánchez-Peña P, Garzón-Tiznado JA,

Vega-Garcı́a MO, López-Valenzuela JA (2013) Microsatel-

lite-based genetic diversity among accessions of maize

landraces from Sinaloa in Mexico. Hereditas 150:53–59

Poland JA, Rife TW (2012) Genotyping-by-sequencing for

plant breeding and genetics. Plant Genome 5:92–102

R Core Team (2014) R: a language and environment for sta-

tistical computing. R Foundation for Statistical Comput-

ing, Vienna. http://www.R-project.org/

Ray DK, Mueller ND, West PC, Foley JA (2013) Yield trends

are insufficient to double global crop production by 2050.

PLoS ONE 8:e66428. doi:10.1371/journal.pone.0066428

Reif JC, Hamrit S, Heckenberger M, Schipprack W, Peter

Maurer H, Bohn M, Melchinger AE (2005) Genetic

structure and diversity of European flint maize populations

determined with SSR analyses of individuals and bulks.

Theor Appl Genet 111:906–913

Reynolds A, Richards G, De la Iglesia B, Rayward-Smith V

(1992) Clustering rules: a comparison of partitioning and

hierarchical clustering algorithms. J Math Model Algo-

rithm 5:475–504. doi:10.1007/s10852-005-9022-1

Salse J, Chague V, Bolot S, Magdelenat G, Huneau C, Pont C,

Belcram H, Couloux A, Gardais S, Evrard A, Segurens B,

Charles M, Ravel C, Samain S, Charmet G, Boudet N,

Chalhoub B (2008) New insights into the origin of the B

genome of hexaploid wheat: evolutionary relationships at

the SPA genomic region with the S genome of the diploid

relative Aegilops speltoides. BMC Genom 9:555–567.

doi:10.1186/1471-2164-9-555

Schierhorn F, Faramarzi M, Prishchepov AV, Koch FJ, Müller

D (2014) Quantifying yield gaps in wheat production in

Russia. Environ Res Lett 9:084017. doi:10.1088/1748-

9326/9/8/084017

Shewry P (2009) Wheat. J Exp Bot 60:1537–1553

Singh RP, Hodson DP, Huerta-Espino J, Jin Y, Bhavani S, Njau

P, Herrera-Foessel S, Singh PK, Singh S, Govindan V

(2011) The emergence of Ug99 races of the stem rust

fungus is a threat to world wheat production. Annu Rev

Phytopathol 49:465–481

Smale M, Reynolds MP, Warburton M, Skovmand B, Tretho-

wan R, Singh RP, Ortiz-Monasterio I, Crossa J (2002)

Dimensions of diversity in modern spring bread wheat in

developing countries from 1965. Crop Sci 42:1766–1779

Sthapit J, Newcomb M, Bonman JM, Chen X, See DR (2014)

Genetic diversity for stripe rust resistance in wheat

landraces and identification of accessions with resistance to

stem rust and stripe rust. Crop Sci 54:2131–2139

Sukumaran S, Dreisigacker S, Lopes M, Chavez P, Reynolds

MP (2015) Genome-wide association study for grain yield

and related traits in an elite spring wheat population grown

in temperate irrigated environments. Theor Appl Genet

128:353–363. doi:10.1007/s00122-014-2435-3

Tyryshkin LG, Tyryshkina NA (2003) Resistance to diseases in

wheat collection samples and somaclonal variants. Czech J

Genet Plant 39:21–23

Vavilov NI (1926) Studies on the origin of cultivated plants.

Institut Botanique Appliqué et d’ Amélioration des Plantes,

Leningrad

Voorrips RE (2002) MapChart, software for the graphical pre-

sentation of linkage maps and QTLs. J Hered 93:77–78

Voss-Fels K, Frisch M, Qian L, Kontowski S, Friedt W, Gott-

wald S, Snowdon RJ (2015) Subgenomic diversity patterns

caused by directional selection in bread wheat gene pools.

Plant Genome. doi:10.3835/plantgenome2015.03.0013

Wang S, Wong D, Forrest K, Allen A, Chao S, Huang BE,

Maccaferri M, Salvi S, Milner SG, Cattivelli L, Mas-

trangelo AM, Whan A, Stephen S, Barker G, Wieseke R,

Plieske J, International Wheat Genome Sequencing C,

Lillemo M, Mather D, Appels R, Dolferus R, Brown-

Guedira G, Korol A, Akhunova AR, Feuillet C, Salse J,

Morgante M, Pozniak C, Luo M-C, Dvorak J, Morell M,

Dubcovsky J, Ganal M, Tuberosa R, Lawley C, Mikoulitch

I, Cavanagh C, Edwards KJ, Hayden M, Akhunov E (2014)

Characterization of polyploid wheat genomic diversity

using a high-density 90000 single nucleotide polymor-

phism array. Plant Biotechnol J 12:787–796

White J, Law J, MacKay I, Chalmers K, Smith J, Kilian A,

Powell W (2008) The genetic diversity of UK, US and

Australian cultivars of Triticum aestivum measured by

DArT markers and considered by genome. Theor Appl

Genet 116:439–453. doi:10.1007/s00122-007-0681-3

Worland AJ, Korzun V, Röder MS, Ganal MW, Law CN (1998)

Genetic analysis of the dwarfing gene Rht8 in wheat. Part

II. The distribution and adaptive significance of allelic

variants at the Rht8 locus of wheat as revealed by

microsatellite screening. Theor Appl Genet 96:1110–1120

Genet Resour Crop Evol

123