Climatic adaptation and ecological descriptors of wild beans from Mexico

Abstract Despite its economic, social, biological, and cultural importance, wild forms of the genus Phaseolus are not well represented in germplasm banks, and they are at great risk due to changes in land use as well as climate change. To improve our understanding of the potential geographical distribution of wild beans (Phaseolus spp.) from Mexico and support in situ and ex situ conservation programs, we determined the climatic adaptation ranges of 29 species and two subspecies of Phaseolus collected throughout Mexico. Based on five biotic and 117 abiotic variables obtained from different databases—WorldClim, Global‐Aridity, and Global‐PET—we performed principal component and cluster analyses. Germplasm was distributed among 12 climatic types from a possible 28. The general climatic ranges were as follows: 8–3,083 m above sea level; 12.07–26.96°C annual mean temperature; 10.33–202.68 mm annual precipitation; 9.33–16.56 W/m2 of net radiation; 11.68–14.23 hr photoperiod; 0.06–1.57 aridity index; and 10–1,728 mm/month of annual potential evapotranspiration. Most descriptive variables (25) clustered species into two groups: One included germplasm from semihot climates, and the other included germplasm from temperate climates. Species clustering showed 45% to 54% coincidence with species previously grouped using molecular data. The species P. filiformis, P. purpusii, and P. maculatus were found at low‐humidity locations; these species could be used to improve our understanding of the extreme aridity adaptation mechanisms used by wild beans to avoid or tolerate climate change as well as to introgress favorable alleles into new cultivars adapted to hot, dry environments.


| INTRODUC TI ON
Mexico is a center of origin, diversity, and domestication for many crops of global importance, including beans (Phaseolus spp.). The Phaseolus genus includes 70-80 species distributed in the Americas, mainly in the Mesoamerican region (central and southern Mexico and Central America) (Gepts, 2014). The genus Phaseolus has five domesticated species: P. vulgaris L. (common bean), P. coccineus L.
Climate change is of concern to the scientific community due to the negative impacts on crop production worldwide (González-Eguiarte et al., 2011;Medina-García et al., 2016). All predictive scenarios of climate change include region-specific changes in precipitation as well as increases in temperature and pest and disease incidence (Porch et al., 2013). Wild relatives of crops represent a primary resource for genetic improvement to ensure food security in the face of accelerated population growth and climate change (Debouck, 2000;Gepts, 2014;Porch et al., 2013).
The genetic variability of domesticated Phaseolus spp. is well represented in germplasm banks; however, a lack of seeds due to livestock, agriculture, forestry, climate change, or urbanization, among other factors is common (Acosta-Díaz et al., 2014;Maxted & Kell, 2009;Ramírez-Villegas et al., 2010). Bean breeders need access to new genotypes for use in the generation of cultivars to satisfy food demand under variable climate or production conditions. Wild parents and domesticated germplasm represent a valuable but underutilized resource (Brozynska, Furtado, & Henry, 2016;Machida-Hirano et al., 2014).
Ecological descriptors are environmental data from collection sites obtained after the standardization of map and layer construction using geographical information systems (GIS) (Cuervo-Robayo et al., 2014;Suárez-Venero, Soto-Carreño, Garea-Llanos, & Solano-Ojeda, 2015;Wang et al., 2015). One classification system based on GIS aids in the development of a conservation strategy by enabling the retro-classification of germplasm collections, facilitating efforts to focus on further exploration, and research in those regions with high probabilities of the presence of specific species or genotypes. This classification system also facilitates the selection of areas for conservation and restoration as well as the prediction of responses to climate change (Elith & Franklin, 2013;Porfirio et al., 2014;Ramírez-Villegas et al., 2014;Wang et al., 2015). López-Soto et al. (2005) characterized climatic distribution types of 25 Phaseolus species throughout Mexico and defined environmental intervals where each species grew by itself. However, climate change and population growth accelerate the natural habitat losses and affect species and/or ecosystems diversity (Delgado-Salinas & Gama-López, 2015;Maxted, Hawkes, Ford-Lloyd, & Williams, 1997).
The goals of the present work were (1) to determine the climatic adaptation of 29 Phaseolus species from Mexico (particularly those species that represent poorly studied genetic reservoirs), (2) describe their potential geographical distribution, and (3) evaluate differences among the species based on climatic/ecological adaptation descriptors and their comparisons with previously reported genetic descriptors.

| Database construction
The database matrix included information about the collection site of each accession throughout Mexico, including genus, species, subspecies, variety, state, county, latitude, longitude, and elevation. The geographic coordinates of each collection site were projected using datum WGS 1984, after which we obtained the value of each variable from 1950 to 2000. Subsequently, we calculated the climatic ranges for each species as well as the ecological descriptors based on climatic range. Environmental information was obtained using DIVA-GIS software, version 7.1.7 (Hijmans et al., 2004;

| Data analysis
Differences across all environmental variables among species were calculated and compared using STATISTICA version 8.0 (StatSoft, 2007) using species as the class variable. Relationships among species were determined by calculating similarity indices using ecological descriptor data (Appendix S2) based on the maximum and minimum values of each parameter. Environmental variables were subjected to principal component analysis (PCA), after which cluster analysis was performed using the complete linkage method and Euclidean distances to identify taxon relationships. The fit of the dendrogram was measured using the k-means clustering algorithm with maximized initial cluster distances.

| RE SULTS
The geographical distribution of collection sites of the 29 Phaseolus species is shown in Figure 1, which includes a broad dispersion of germplasm across regions and climate types of Mexico (12 of 28 climatic variables, Appendix S2).
The collection sites demonstrate the wide eco-geographical range in elevation and climate variables observed among and within species of the genus, indicating that Phaseolus spp. has developed climatic adaptability.

| Climatic correspondence
The Phaseolus species are more frequently found in subtropical and tropical climates ranging from arid to humid conditions. The subtropical temperate subhumid climate includes the highest number of species and accessions (11 and 20, respectively), followed by subtropical arid temperate (9 and 11, respectively), subtropical subhumid semihot (8 and 20, respectively), and tropical hot-subhumid (5 and 11, respectively) climates. Accessions of P. acutifolius exhibited the highest frequency of accessions and were observed in the following seven climatic types: subtropical semiarid hot, semihot, and temperate; subtropical subhumid temperate and semihot; subtropical arid semihot; and tropical semiarid hot. In terms of the frequency of accessions in different climate types, the species were ordered as follows: P. coccineus (subtropical arid temperate; subtropical semiarid semihot and temperate; and subtropical subhumid temperate and semihot), P. vulgaris (subtropical subhumid hot, semihot, and temperate; tropical subhumid semihot and hot), and P. lunatus (tropical subhumid hot; tropical humid hot and tropical arid very hot).

| Classification
The Both the PCA and cluster analysis (Figure 3)   Mesoamerica. Conservation of these species is a priority because of their importance in breeding programs (Andueza-Noh et al., 2016;De Ron et al., 2015;Debouck, 2000;Rodriguez et al., 2015;Shi & Lai, 2015;Singh, Singh, & Dutta, 2014 to the low number of samples analyzed, although they provide valuable information for future and more intensive collections (Gil & Lobo, 2012;Pliscoff & Fuentes-Castillo, 2011;Ramírez-Villegas et al., 2010;Russell et al., 2016). A major trait of wild beans, however, is that the seeds are tolerant to less stringent storage conditions; they are capable of withstanding high temperatures and high relative humidity without germinating (dormancy). Therefore, seeds of wild beans can be stored for The data suggest that species with a reduced range in elevation or distribution in the highest elevations (P. coccineus subsp. coccineus, P. gladiolatus, P. palmeri, and P. pedicellatus) will face adaptation obstacles under climate change, with a consequent loss of genetic diversity (Hill, Griffiths, & Thomas, 2011). In addition, the data may predict the presence of proper germplasm for breeding based on elevation adaptation (Porch et al., 2013).

One major trait for selection and breeding in
No species were found at locations with maximum temperatures lower than the annual mean of all Phaseolus species (19.5°C), although some species were found at sites with minimum temperatures higher than general mean temperatures. For example, P. lunatus . The most optimistic climate change scenario will involve an increase in temperature. Therefore, we suggest the necessity to study adaptation patterns and ranges to predict the future distribution of species; the reduced range of temperatures will reduce adaptation capacity (Hill et al., 2011;Porch et al., 2013). In addition, as high temperatures are associated with an increase in pest and disease incidence, we can efficiently breed beans for increased resistance to these adverse factors if we know the environmental conditions that act as selection pressure sites (Abberton et al., 2015;Miklas, Kelly, Beebe, & Blair, 2006).

2010) as well as physiological and biochemical changes in root tissue
can occur to provide tolerance to dehydration (Sánchez-Urdaneta et al., 2003).
The PCA and cluster analysis provided similar results. In addition, species grouped by climatic and ecological variables were 45% to 54% identical to those of the molecular groups (ITS and trnK sequences) described by Delgado-Salinas et al. (2006). Differences in grouping by the two strategies reinforce the necessity to complement eco-geographical data with phenotypic and/or genotypic information, as morphologic, biogeographic, or ecologic distinctness can be detected even in phylogenetically related species (Delgado-Salinas & Gama-López, 2015).
Wild germplasm is the primary genetic resource for plant breeding (Castañeda-Álvarez et al., 2016;Piñero et al., 2009 The use of environmental information in breeding programs has been promoted by the use of methodologies that can predict the presence of specific traits in germplasm growing at specific locations with consequent savings of time and cost (Cortés et al., 2013;Parra-Quijano et al., 2012;Rodriguez et al., 2015;Song et al., 2015;Thormann et al., 2015;von Wettberg, Marques, & Murren, 2016). However, conservation programs are not considered important by the government and have therefore not received economic support, resulting in the low representation of wild crop relatives in germplasm banks (Castañeda-Álvarez et al., 2016;Maxted et al., 2016). In addition, it is necessary to know the conditions of in situ conservation to conduct long-term monitoring because of the advantages of low maintenance costs and the dynamic evolution of populations that this type of conservation offers (Acosta-Díaz et al., 2015;Ramírez-Villegas et al., 2014;Smýkal et al., 2015).
Furthermore, we must consider the threats to wild germplasm caused by agriculture, urbanization, invasive species, contamination, mining, and climate change. Climate change is of concern to the scientific community due to the negative impacts on crop pro- ties, responsibility that it is inherited from generation to generation (Boege, 2008;Pretty et al., 2009). We suggest that it is imperative to legitimize and strengthen community property to support genetic resource management and conservation (Maffi, 2005;Naughton-Treves & Wendland, 2014). Finally, we propose future work to model the current potential distribution of Phaseolus spp. in Mexico and to evaluate the impact of climate change on their future distribution.
Such information would aid in decision making to implement conservation strategies for vulnerable genetic resources, especially those around the proposed critical area of domestication.

| CON CLUS IONS
Wild crop relatives represent a primary genetic resource in crop improvement to ensure food security in the face of accelerated population growth and climate change. The eco-geographical analysis of germplasm collection sites in Mexico revealed the broad dispersion and distribution of wild Phaseolus based on elevation, mean annual temperature, precipitation, and photoperiod patterns. Our results confirm the broad climatic variability adaptation of Phaseolus and represent the potential geographical distribution of these species.
The Phaseolus species studied were abundant in climates with arid to humid conditions, especially in subtropical and tropical environments. The highest species diversity was found in subtropical temperate subhumid climate types. P. acutifolius was the most frequently observed species and was found in seven climatic types.
Knowledge of the climatic distribution supported by geographical information systems will allow us to generate maps and establish potential areas of distribution, adaptation, and location of wild Phaseolus germplasm in Mexico. These data will assist the planning of future collection expeditions and allow efficient strategies to acquire, manage, and support in situ conservation of wild bean genetic resources.

ACK N OWLED G M ENTS
The first author is grateful to Consejo Nacional de Ciencia y