Introduction

As the world population increases, the need for long-term food security of high-yielding, highly nutritious crops becomes ever more critical. Increasingly, plant breeders are turning to the wider crop gene pool to find the diversity required to cope with the changing biotic and abiotic environment while sustaining food security1,2,3. Crop wild relatives (CWR) are taxa that are closely related to crops, many with the ability to contribute beneficial traits to their related crops such as resistance to biotic and abiotic stresses and higher, more stable yields4,5,6. Although recent advances in biotechnology offer techniques for direct genetic modification of crops, conventional breeding methods offer a more accessible and reliable option for crop improvement7, particularly for breeders without access to the resources required for high-investment breeding. Thus, the most valuable taxa in conventional breeding are the priority CWR, the taxa that are most closely related to the crop and so have the highest potential to produce successful crosses8.

As interest to integrate CWR into crop breeding has increased, CWR conservation has become critical to maintaining these resources for future use9. Despite the calls for their identification and conservation in both international and regional policy instruments10,11,12,13,14, CWR conservation has historically been widely neglected8,9,15,16 and CWR research is gravely under-represented in the general conservation literature17. This deficit in both research and action may reflect a lack of awareness in the wider conservation research community or the societal value of CWR conservation and their role in underpinning provisioning ecosystem services.

The plans that do exist for the in situ conservation of CWR are generally based on distributions of all taxa in the crop genera18,19,20,21 (and recent strategies for ex situ CWR conservation by the Consultative Group on International Agricultural Research, http://gisweb.ciat.cgiar.org/GapAnalysis/). However, the difference between the best localities to conserve the priority CWR and the best localities to conserve an entire genus could be quite dramatic, especially if the crop genus is large. Using genus diversity as a surrogate for priority CWR diversity may mask the specific conservation requirements or miss important opportunities for the efficient conservation of the priority CWR and, therefore, many of the most useful resources to plant breeders. Furthermore, priority CWR conservation may be more efficiently conducted by searching within established protected areas or by building upon conservation targets for other taxonomic groups. Here we use the African cowpea (Vigna unguiculata (L.) Walp.) as a case study to explore the use of taxonomically and geographically prioritised conservation planning approaches to identify an efficient set of locations for the conservation of priority CWR.

Like all conservation efforts, CWR conservation will always be hampered by the limited availability of funds, which could explain why most current CWR conservation is limited to ex situ efforts, storing seeds in local and global seed banks. Ex situ conservation is cheaper than establishing and maintaining in situ activity and agricultural agencies have limited responsibility for wild species conservation. Furthermore, ecological conservation agencies tend to prioritise specific habitats or rare, endemic, threatened or charismatic species. Although plant resources can be safely conserved ex situ in seed banks, in situ conservation is an important compliment because it allows populations to continue natural adaptation to changing conditions22,23. In situ approaches are especially important in CWR conservation since many of their traits that are beneficial in crop breeding are relevant to adaptation to particular environmental conditions (e.g., drought tolerance, salt tolerance). The future needs of crop improvement are dynamic, so maintaining CWR diversity in situ ensures that future use value is maximized.

One way to improve the efficiency of in situ CWR conservation efforts is to focus taxonomically on the priority CWR, the taxa with the greatest potential for crop improvement. Based on breeding relationships or generic classifications, the priority CWR are 1) the wild forms of the cultivated taxa, or if breeding relationships are unknown, the infraspecific, uncultivated taxa; 2) taxa in the same taxonomic series or section as the crop; and 3) taxa in the same subgenus as the crop6. The priority CWR, should be the focus of CWR conservation and have already been identified for a number of the world's major crops8.

In addition to taxonomic prioritisation, conservation effort can be geographically focused by identifying existing protected areas where target taxa are likely to occur24. Identifying and accounting for species likely to occur in existing protected areas is a core principle of system conservation prioritisation25. An efficient set of in situ conservation sites might also include places of potential overlap with priority locations for other conservation action (e.g., important areas for endangered or critically endangered taxa from the same or different taxonomic groups26). The suitability of partnering CWR conservation with other conservation activities is relatively unexplored but may be a cost-effective way to achieve in situ conservation.

We use the African cowpea as a case study to demonstrate how to efficiently identify conservation targets for priority CWR using both taxonomic and geographic prioritisation. In sub-Saharan Africa, cowpea is one of the major grain legume and protein sources for subsistence farmers, valued for its high protein content, drought tolerance and nitrogen fixing abilities18. Sub-Saharan Africa has the highest concentration of the most undernourished countries in the world27, which makes the region an important target of the Millennium Development Goals in reducing the number of undernourished people by half28. Cowpea is one of the twenty most produced food sources in Western Africa27, where many of the cowpea CWR are found18.

Expanding and improving cowpea cultivation could help alleviate undernourishment in Western and sub-Saharan Africa. The conservation of the priority cowpea CWR is an important target for achieving this goal. Even though they offer the opportunity to increase production as well as biotic and abiotic resistance18, CWR have not yet been widely exploited in cowpea breeding. The range of cowpea CWR diversity has not been adequately conserved for later use and is threatened by the by-products of human population expansion and urban sprawl continually threaten the maintenance of that diversity (i.e., irrigation practices and soil salinisation, deforestation, agricultural expansion and the introduction of invasive species)18. Cowpea CWR populations are further threatened by climate change, which stands to deplete or change the patterns of natural ranges and habitats29,30. African cowpea is also a practical case study for demonstrating taxonomically-prioritised conservation plans for CWR since the priority CWR have already been identified8.

In this study we identify the best locations to find and conserve the priority African cowpea CWR based on georeferenced herbarium and germplasm records. Of the over 100 taxa in the African cowpea genus, only 14 are considered priority wild relatives of cowpea8, so the difference in best locations to conserve the entire genus versus the priority CWR could be quite large. Using the African cowpea genus as a case study, we first test the extent to which the best conservation sites identified by the known records of the entire genus adequately represent the conservation needs of the priority cowpea CWR. To design more practical conservation goals, we also identify where priority cowpea CWR in Africa are likely to occur in existing protected areas and the extent to which they occur in places that are likely to be the focus of more traditional in situ conservation actions, in this case, focusing on the conservation of endangered mammal species. Finally, we use our findings to propose a series of localities for the in situ conservation of priority cowpea CWR in Africa. Although widely used to prioritise conservation efforts, these conservation planning techniques have not been applied to CWR conservation. We believe this case study provides a useful framework to identify efficient conservation targets that taxonomically and geographically prioritise the most important wild agricultural resources.

Methods

We started with a database of all collection localities of African Vigna taxa compiled from herbarium specimens and gene bank accessions18. We removed accessions collected before 1920 as older collection sites are less likely to reflect current distributions of taxa and since there was a general increase in Vigna collection after this time18. The resulting database included point locality records for 111 African Vigna taxa, including 13 of the 14 priority CWR taxa. We applied a 25 km radius buffer to each point locality record to account for possible error in the georeferencing of records in the database.

To explore whether the entire Vigna genus is a reasonable surrogate for identifying the best in situ conservation localities for the 13 priority CWR species, we first divided the continent of Africa into 100 km2 planning units and then summarised the occurrence of all Vigna species within each planning unit. We used the reserve design software Marxan31 to identify the set of planning units that most efficiently represented at least one occurrence of all species in the Vigna genus, excluding the 13 priority CWR. Surrogates, especially based on point data, are likely to perform better using a complementarity approach than using a hotspot approach32. Performance is based on how well the surrogate represents the other species of interest. Within Marxan, we identified priority sites using both a basic Richness heuristic and the more optimal Simulated Annealing algorithm33. Under the Richness heuristic, planning units that are likely to contain multiple Vigna taxa are selected preferentially. We repeated the Richness and Simulated Annealing prioritisations 100 times and identified the best set of sites to protect all non-priority cowpea CWR taxa. We also identified high priority sites, which were those sites selected in the best set of sites in at least 75 of the 100 runs. We then calculated the extent to which both the best set of sites and the set of high priority sites captured the 13 priority cowpea CWR, which were not considered in the selection of those sites. This approach to surrogacy testing is sometimes referred to as the selection method and is the most widely adopted approach34.

Prioritisation analyses that are based on buffered point data, rather than the full distribution of each species, are sensitive to the size of the planning units used in the analysis. To test sensitivity to this effect, we repeated the analysis described above with 200 km2 planning units and with 50 km2 planning units.

To identify existing protected areas across the African continent that are likely to contain at least one of the 13 priority cowpea CWR, we overlapped the buffered locality records for priority species with data obtained from the World Database on Protected Areas35.

To explore the extent to which conservation priorities driven by more charismatic species, the conventional focus of in situ conservation in Africa, could also include priority cowpea CWR species, we obtained the distributions of all 159 African mammal species listed as Endangered or Critically Endangered by the IUCN. As with the Vigna data, we summarised the occurrence of these 159 species within 100 km2 planning units. We used the procedures described above to identify the best set of sites to represent at least one occurrence of each mammal species. We then calculated the extent to which these sites captured known occurrences of the 13 priority cowpea CWR.

Finally, we identified the best sites for in situ conservation of the 13 priority cowpea CWR in Africa. Again we used Marxan as described above but only considered the known occurrences of the 13 priority CWR species.

Results

Performance of Vigna genus as a surrogate for priority CWR species

The best conservation solution for the entire Vigna genus, not including the priority CWR, contained 32 planning units across Africa (Figure 1) and captured 12 of the 13 priority cowpea CWR. The only species not represented was Vigna monantha, known only from the coast of Somalia.

Figure 1
figure 1

The set of sites that most efficiently captures the Vigna genus excluding the priority crop wild relatives (CWR) (squares), compared with the set of sites identified specifically to represent the priority CWR (blue circles).

The eight highest priority planning units from the Vigna genus set (identified in black) capture 10 of the 13 priority Vigna CWR. Eight priority CWR-specific planning units capture all 13 priority Vigna CWR taxa. Content for the maps was generated in Marxan 1.8.10 and the map images were generated in ArcGIS 1038.

Full size image

When we consider only the highest priority sites, those planning units selected more than 75% of the time, 10 of the 13 priority cowpea CWR are still captured (Figure 1). Vignamonantha was still not represented, along with Vigna keraudrenii, which is endemic to Madagascar and Vigna unguiculata subsp. baoulensis, known from only two records in Côte d'Ivoire.

Using a simulated annealing algorithm rather than a simple richness heuristic delivered only a marginal improvement, reducing the total number of planning units required from 32 to 30. However, the best set of sites based on simulated annealing captured only 9 of the 13 priority CWR, 3 less than the richness heuristic. This result is explained by the fact that simulated annealing locates the most complementary set of sites to represent the genus, excluding the priority CWR species.

Doubling the planning unit size to 200 km2 made little difference, with the best solution still capturing 11 of the 13 priority CWR, with Vigna unguiculata subsp. baoulensis missing in addition to Vigna monantha. The larger planning units were also less efficient, requiring 27 planning units or 1.08 million square km to represent the Vigna genus without the priority CWR, compared with 320,000 square kilometres for the 100 km2 planning units. Halving the planning unit size to 50 km × 50 km proved the most efficient, still only requiring 32 planning units or 80,000 square km to represent the genera. The best set of 50 km2 planning units also captured 12 of the 13 priority CWR, only missing Vigna monantha. This suggests that priority sites to conserve the Vigna genus in situ are extremely consistent and conservation efforts can be precisely targeted.

Protected area analysis

We found that 182 established protected areas across Africa are likely to contain at least one priority cowpea CWR (Figure 2, Table S1). In total, 9 of the 13 priority cowpea CWR are likely to be found in an existing protected area. Of these protected areas, 63 are likely to harbour multiple priority taxa (Table S2). Based on collection records, however, Vigna monantha, Vigna unguiculata subsp. alba, Vigna unguiculata subsp. aduensis and Vigna unguiculata subsp. letouzeyi are not likely to exist within any established protected areas.

Figure 2
figure 2

Established protected areas where priority Vigna crop wild relatives may be found (green areas) and the best conservation solution to represent all Endangered and Critically Endangered mammal species in Africa (grey rectangles).

9 of the 13 priority CWR taxa are likely to be found in the established protected areas shown (see Supplementary tables S1 and S2 for details) and the set of sites that capture Africa's Endangered and Critically Endangered mammal species also capture 10 of the 13 priority cowpea CWR. Content for the maps was generated in Marxan version 1.8.10 and the map images were generated in ArcGIS 1038.

Full size image

Overlap with priorities for conservation of Endangered and Critically Endangered mammal species

The best conservation solution to represent all Endangered and Critically Endangered mammal species in Africa (Figure 2) required 61 planning units and captured 10 of the 13 priority cowpea CWR. The three species not captured were Vigna monantha, V. unguiculata subsp. aduensis, V. unguiculata subsp. unguiculata var. spontanea. Of these, V. unguiculata subsp. aduensis is recorded from Ethiopia in the planning unit adjacent to a mammal priority site and V. unguiculata subsp. unguiculata var. spontanea is a widely distributed species, also with recorded occurrences in planning units adjacent to mammal priorities, particularly in South Africa. These results were indifferent to the use of simulated annealing or a richness heuristic.

Best sites for priority cowpea CWR species

Based on collection records, at least 8 in situ conservation sites are required to represent all 13 priority CWR from the African Vigna genus (Figure 1). Both the number and general location of the required sites were insensitive to changes in planning unit size or solution method. The 8 priority sites encompassed the following localities: the tri-border region of Côte d'Ivoire, Guinea and Liberia; northern Ethiopia; coastal Somalia; the tri-border region of the Democratic Republic of Congo (DRC), Central African Republic and Congo; the coastal border of DRC and Angola; the border of Zimbabwe and Mozambique; and central Madagascar (Fig. 1). Together, these eight sites also capture 27 of the 98 non-priority African Vigna taxa.

Discussion

By identifying the priority conservation sites for all taxa in the African cowpea genus, we were also able to capture 12 of the 13 of the most important CWR analysed in this study. This result suggests that the whole cowpea genus acts as a reasonable surrogate for identifying conservation sites for priority CWR. However, for other crop genera, the whole genus may not be a suitable surrogate for the priority CWR. Because CWR are vital resources for future crop improvement and sustainability, it is important to conduct a similar analysis on other important crop genera to ensure that the conservation needs of priority CWR are identified. While this study focuses on identifying conservation targets for the priority CWR, an important aspect of the methodology is exploring both genus-wide and priority CWR-specific conservation needs, as focusing solely on the priority CWR may overlook the unique needs of non-priority taxa.

Based on occurrence data for 13 of the 14 priority cowpea CWR, we identified over 100 established protected areas where priority cowpea CWR may be located and conserved, many of which are likely to harbour multiple priority taxa (see Tables S1 and S2). Locating and initiating conservation activity within protected areas is likely to be an efficient and cost-effective strategy for protecting priority CWR. We also found that three African cowpea CWR are not found in established protected areas, so extra effort will be needed to locate and conserve these priority taxa. Although we use cowpea as a case study, the conservation goals for other CWR may also be met by building on established conservation initiatives36.

Additionally, we found that a number of the best localities for Endangered and Critically Endangered mammal conservation overlap with occurrence data for many of the priority cowpea CWR. This suggests that there is potential for CWR conservation efforts to build on or partner with active conservation activities focused on other species. An example of such partnership is the inclusion of conservation goals for fynbos vegetation in South African conservation efforts originally motivated by fauna conservation37.

Our analysis was based on herbarium specimen and seed collections, not on modelled distributions. Modelling expected distributions of the priority CWR would be useful in conservation planning to reduce the extent to which site selection is biased by sampling effort. However, distribution modelling would only be possible for a small subset of cowpea CWR taxa, since some taxa are known from fewer than 5 records. It is an unfortunate paradox that it is these lesser-known species that would benefit most from distribution modelling and yet there would be little confidence in the modelled distributions. The results presented here represent conservative but reliable geographic targets for initial, cost-effective conservation of the priority cowpea CWR.

Although we find that the African cowpea genus acts as an adequate surrogate for CWR conservation planning, this may not be the case for all CWR. It would be useful to analyse multiple crop genera using similar methodologies to investigate the generality of the results presented here. Where possible, identifying the priority CWR and using their distributions as a guide for conservation effort is likely to lead to the most efficient conservation of priority CWR. Furthermore, we recommend fully exploring the conservation needs of the entire genus and the priority CWR to ensure that the specific needs of threatened or remote taxa are identified.

Although in situ conservation of critical crop resources is an economic and humanitarian priority12,28, it has been largely ignored by the conservation movement. Here we demonstrate that the African cowpea CWR distribution overlaps with both established protected areas and existing conservation initiatives that target other taxonomic groups. This finding suggests that much in situ CWR conservation can be accomplished by utilising existing resources, yet will have a huge benefit to humanity.