Many thanks to long-time friend-of-the-blog Dr Colin Khoury for this latest contribution.
Conservation gap analysis using Geographic Information System (GIS) tools relies on several sources of biological and environmental data, including in situ species occurrences and climatic and other environmental variables used to conduct species distribution modeling, as well as passport data from ex situ collections. While species distribution modeling and associated methods had been in development since at least the 1970s (see Rebelo, 1994 and Booth et al., 2013), the widespread use of these tools was not possible until such biological and environmental data were more easily and widely accessible, for example through GBIF, WorldClim, and Genesys.
Genebank scientists, often in collaboration with academic researchers, began to apply available GIS-based tools to PGRFA conservation around the turn of the century, proceeding to develop new methods, software, and datasets (for early examples, see Guarino, 1995; Greene and Guarino, 1999; Guarino et al., 2002). Global climate datasets were compiled at relatively high spatial resolution (e.g., Hijmans et al., 2005), providing key inputs for species distribution modeling. Current distribution models for plant genetic resources began to be calculated, for example for wild relatives of potatoes and peanuts, while future distributions under climate change also began to be modeled, for example for wild peanuts, potatoes, and cowpeas. Field collecting was informed through these tools, for example for wild clover and wild chile pepper expeditions.
The focus on wild relatives of food and agricultural crops was not haphazard. These species were receiving increasing conservation attention at the time in recognition of their value as genetic resources for crop breeding, and because many wild relatives were known to be threatened in their natural habitats and were underrepresented in ex situ repositories. International conservation targets for crop wild relatives had been set at the Convention on Biological Diversity (CBD) (for 2011 to 2020 and again for 2020 to 2030) and in the United Nations Sustainable Development Goals (SDGs) (for 2015 to 2030). At the same time, species distribution modeling methods had primarily been developed for wild species, i.e. taxa whose distributions are mainly driven by climatic, edaphic, and other environmental factors, rather than human preferences (which are more difficult to model), therefore the application of these methods to crop wild relatives was relatively straightforward and a logical starting point for the agricultural research community.
Programs such as DIVA-GIS and FloraMap were created to make the methods more accessible to researchers and practitioners without extensive GIS experience and computing power. Such efforts continue, for example by CAPFITOGEN.
Through an international genebank initiative called the Global Public Goods Project II, run from 2007-2010, the distributions of the wild relatives of ten CGIAR mandate crops were mapped, with priorities for further collecting for ex situ conservation identified. A major milestone of that project was the publication of a standardized, replicable gap analysis methodology for the ex situ conservation of crop wild relatives, which made use of herbarium and other biodiversity observations acquired through GBIF and other sources, as well as genebank passport data, and which embraced recent advancements in species distribution modeling methods.
A big question at the time (and still to this day) was how well these methods reflected true distributions and conservation gaps. The authors devised an evaluation tool to compare human expertise versus machine results — in this case with knowledge from one of the world’s foremost authorities on wild bean diversity compared to the gap analysis results — in a John Henry-esque battle conceived in the context of the sad truth that the world has been losing field botanists much faster than producing new ones. While these GIS-based methods can partly mitigate that loss of knowledge and expertise, gaps in high resolution, updated spatial information limit the capacity of the methods to produce fully accurate maps of species distributions and conservation gaps. A combination of gap analysis results with field-based expertise, therefore, is ideal.
This ex situ conservation gap analysis methodology was further updated and applied at a larger scale through the 2011-2021 global Crop Wild Relatives Project, which assessed ex situ conservation gaps for over 1000 wild relative taxa related to around 100 of the world’s most important food and agricultural crops, engaging many experts to evaluate the results. The collaboration with taxonomists, plant breeders, and conservationists resulted in further crop genepool-specific gap analyses, including for the wild relatives of eggplant, lettuce, pigeonpea, potato, sunflower, and sweetpotato, as well as for specific geographic areas, for example Australia.
During this time PGRFA researchers began to consider how the conservation gap analysis methods could have wider application, for example to contribute to the indicators used to measure international conservation targets on crop diversity, including the CBD’s Aichi Target 13 and SDG Target 2.5. To do so, the methods would need to expand not only taxonomically, but also in conservation scope, by developing an equivalently standardized and replicable method to assess gaps in the protection of in situ plant genetic resources. With support from the Biodiversity Indicators Partnership connected to the CBD, standardized in situ conservation gap analysis methods were developed and then integrated with the ex situ methodology, becoming an official indicator for Aichi Target 13 and covering wild relatives and other socio-economically as well as culturally valuable wild plants.
The standardized, integrated ex situ and in situ conservation gap analysis method for crop wild relatives was further applied by genebank researchers and collaborators through studies on the genepools of wild chile peppers, pumpkins, and sorghum, for crop genepools in specific regions, for example wild carrots in Tunisia, wild lettuces in the United States, and wild grapevines in the Americas, and also for many species across large geographic areas, such as via a national conservation gap analysis for the United States, covering 600 native wild relative taxa. These studies further enabled improvements in the methods, data sources, and the products, which were published in open access formats to promote further use (e.g., the GapAnalysis.R code), supporting additional gap analyses, such as for wild bananas conducted by botanic garden and academic researchers.
Conservation gap analyses using several related but different methodologies were published by PGRFA research groups during this period, including for wild cowpea, barley, grasspea, lupines, and wheat, and for wild relatives in China, Mexico, Brazil, and Northeast Africa. A global in situ conservation gap analysis of crop wild relatives was also published, leveraging the data compiled for the taxa assessed in the global ex situ gap analysis.
The botanic garden sector has also adopted and innovated on conservation gap analysis tools, with leadership from the Global Conservation Consortia for Oak and for Magnolia, and applied thus far for oaks, American beech, hickories, Kentucky coffeetree, pines, selected laurels, walnuts, yews, and magnolias. A primary focus has been on using highly curated occurrence datasets and relatively easy to generate tools (for example, calculating buffers around occurrence points rather than species distribution models to estimate potential range/habitat of target species) to produce results for many species for subsequent conservation action by Consortia partners. An online version of a gap analysis (GAMMa) tool requiring no coding experience is now in development, currently enabling gap analysis for one species at a time.
Ex situ and in situ conservation gap analysis methods for wild plants of interest to the agricultural research and botanic garden communities are now fairly well established. Further work is certainly needed to continue to refine the methods, in particular to ensure the results optimally align with direct assessments of genetic diversity (see Hanson et al., 2017 but also Hoban et al., 2020). An ongoing project led by the Morton Arboretum has been working to compare direct measures of genetic diversity for ten different wild plant species with conservation gap analysis results and may be able to provide further insights on how variables such as the germplasm buffer size can be adjusted for best fit for different types of wild plant species.
Major methodological advances are still needed for conservation gap analyses of cultivated plants, mainly due to the challenges presented by modeling the effect of human preferences. Several ex situ gap analyses for farmer landraces have been conducted, for example for pearl millet, beans, and major CGIAR crops, but the methods require substantial further development to be more robust and accessible.
A related set of GIS tools and methods have also been applied to ex situ collections of farmer varieties through the “Focused Identification of Germplasm Strategy” (see here and here), aimed at identifying traits of potential interest within existing accessions, such as resistance to specific pests and diseases such as wheat aphid, stem rust, and tolerance to abiotic stresses including drought.
As computing power increases, and as more and higher resolution spatial datasets become available, gap analysis methods should continue to improve, offering researchers a more sophisticated picture of the distribution of PGRFA diversity and of gaps in conservation of this diversity. Combined with field expertise and, ideally, field validation of the results, gap analysis methods should continue to be a key tool for PGRFA conservation planning.