Incorporating evolutionary and threat processes into crop wild relatives conservation

Attaining food and nutrition security under global change is a significant challenge. Crop wild relatives (CWR) are crucial as a reservoir of genes that can enhance crop adaptation to changing environmental conditions. CWR possess greater genetic diversity than crops, offering useful adaptations to various biotic and abiotic stresses. Mesoamerican wild relatives of important crops (maize, beans, chili, pumpkins) are particularly vulnerable, with up to 35% of taxa threatened with extinction. Anthropogenic activities and habitat modification pose major threats. Innovative conservation solutions are needed to maintain range-wide genetic diversity, essential for evolutionary resilience. Systematic conservation planning can help locate and manage key areas to protect biodiversity, but most assessments focus on taxa representation rather than persistence, neglecting genetic diversity below the species level. CWR genetic diversity depends on demographic history, population structure, and natural selection, influenced by climatic and geologic history. Centers of domestication, often rich in CWR diversity, are frequently located in topographically heterogeneous tropical regions where long-term population persistence and historical isolation contribute to complex genetic differentiation patterns. Conservation planning must consider both current and historical evolutionary drivers.

Conservation planning for CWR has increased, including identifying priority sites for in situ protection and ex situ collections. However, most planning approaches use the "minimum set cover problem," which may not fully represent the genetic diversity spectrum. This is particularly relevant in centers of origin and domestication like Mexico, where genetic diversity is shaped by complex historical and environmental factors, and where crops and CWR coexist and interact. Existing approaches often fall short in representing areas where population differentiation is expected due to historical processes, such as range shifts during Pleistocene glacial fluctuations or range subdivision into naturally isolated populations. These processes are strong drivers of population structure in tropical and topographically complex areas. Population structure can lead to locally restricted alleles, both neutral and adaptive, and contribute to speciation.

This study introduces a framework to assess conservation areas based on hierarchical prioritization of the landscape to maximize the representation of genetic diversity even in the absence of genomic information. The systematic conservation planning accounts for: (i) Evolutionary processes: historical and environmental drivers of genetic diversity. (ii) Threat processes: taxa-specific tolerance to human-modified habitats and extinction risk status. The methodology involved several steps: 1. **Data Collection:** Collation and cleaning of data on taxa occurrences, extinction risk, and other relevant factors. 2. **Species Distribution Modeling (SDM):** Generation of SDMs for 116 taxa with sufficient occurrence records using MaxEnt. 3. **Proxies of Genetic Differentiation (PGD):** Creation of PGDs to represent potential genetic differentiation within a taxon's distribution. This involved subdividing Holdridge life zones based on phylogeographic studies of various taxa to account for both environmental and historical drivers of differentiation. A total of 102 PGDs were generated for Mexico. This approach was tested using available genomic data from *Zea mays* subsp. *parviglumis*, demonstrating its effectiveness in representing genetic variation. 4. **Systematic Conservation Planning:** Integration of SDMs subdivided by PGDs, taxon-specific habitat preferences, occurrence records for taxa without SDMs, and IUCN threat categories into a systematic conservation planning analysis using Zonation software. Different scenarios were compared, including various combinations of SDM, PGDs, and Holdridge life zones. 5. **Conservation Area Identification:** Based on the best-performing scenario (SDM subdivided by PGDs), a hierarchical landscape priority map was generated. A 20% area threshold was selected based on performance curves and Aichi Target 11, proposing key areas for CWR in situ conservation in Mexico.

The study used an inventory of 224 Mesoamerican CWR taxa related to nine crops. Based on SDMs, areas of high taxa richness were identified along the Trans-Mexican Volcanic Belt and in the montane areas of Oaxaca and Chiapas. Given the scarcity of high-resolution genetic data for all taxa, the study employed PGDs as surrogates of potential genetic differentiation. The analysis included 116 SDMs subdivided by 102 PGDs, resulting in 5004 input layers for the systematic conservation planning analysis. The approach that maximized the representation of intraspecific diversity, as given by the proportion of PGD areas averaged for each taxon, used a dataset where each SDM was subdivided by the genetic proxies. The combination of SDM and proxies resulted in 5004 input layers or conservation features; each conservation feature represented a part of a taxon range occurring in a proxy of genetic differentiation. Different scenarios were compared, including using only SDMs, SDMs with Holdridge life zones, SDMs with PGDs, and a combination of SDMs and PGDs. The scenario that combined SDMs with PGDs maximized the representation of intraspecific diversity. The identified conservation areas are located in diverse regions of Mexico, including montane areas, coastal regions, and arid and semi-arid zones. A significant portion (almost half) of these areas overlap with indigenous communities, highlighting the biocultural importance of these regions. Approximately 11% of the proposed areas are within federal protected areas, and one-third are in areas voluntarily designated for conservation. Performance curves demonstrated that representing at least 50% of all conservation features required sustainably managing 80% of Mexico's terrestrial surface. Additional analyses considering taxa habitat preferences were performed, leading to identification of areas best suited for different CWR types. The study also showed that the conservation of ecological and evolutionary processes shaping biodiversity at all levels cannot be secured in a small fraction of the territory and with few individuals.

The findings address the research question by providing a spatially explicit framework for incorporating genetic diversity into CWR conservation planning. The significance of the results lies in the development of a novel methodology that addresses the lack of genomic data for many CWR taxa. This approach allows for a more comprehensive representation of genetic diversity in conservation planning, which is crucial for maintaining evolutionary resilience. The hierarchical prioritization of the landscape offers a broader perspective, not only for identifying areas for area-based conservation but also for implementing sustainable development policies in agricultural landscapes. The results support the development of National Strategic Action Plans, inform public policy to mitigate threats, and suggest directions for future research, including germplasm exploration and collection. The integration of planning outputs into cross-sectoral policies can extend in situ conservation beyond protected areas. This is particularly important in regions that are centers of origin and domestication of crops, where sustainably managed landscapes can contribute to halting biodiversity loss and the provision of ecosystem services.

This study presents a novel framework for incorporating evolutionary and threat processes into crop wild relative conservation. The methodology, using proxies of genetic differentiation, allows for the efficient representation of genetic diversity in systematic conservation planning even in the absence of genomic data. The approach was successfully applied to Mesoamerican CWR in Mexico, identifying key areas for conservation. Future research should focus on refining the proxies of genetic differentiation as more genetic data become available and exploring the application of this framework to other taxa and regions.

A major limitation is the lack of high-resolution genetic data to fully validate the accuracy of the proxies of genetic differentiation. The methodology is computationally intensive, requiring a computing cluster to handle the large datasets. The model relies on existing knowledge of phylogeographic patterns, and the accuracy of the proxies will depend on the quality and completeness of this information. Furthermore, the study focuses primarily on in situ conservation and doesn't fully address the complexities of ex situ conservation strategies. Future research is needed to further refine the methodology and address these limitations.