Optimising biodiversity protection through artificial intelligence

Over a million species face extinction, carrying with them untold options for food, medicine, fibre, shelter, ecological resilience, aesthetic and cultural values. There is therefore an urgent need to design conservation policies that maximise the protection of biodiversity and its contributions to people, within the constraints of limited budgets. Here we propose a novel framework for spatial conservation prioritisation that combines simulation models, reinforcement learning and ground validation to identify optimal policies. Our model quantifies the trade-off between the costs and benefits of area and biodiversity protection, allowing the exploration of multiple biodiversity metrics. Under a fixed budget, our model protects substantially more species from extinction than the random or naively targeted protection of areas. We find that regular biodiversity monitoring, even if simple and with a degree of inaccuracy, substantially improves biodiversity outcomes. Given the complexity of people–nature interactions and wealth of associated data, artificial intelligence holds great promise for improving the conservation of biological and ecosystem values in a rapidly changing and resource-limited world.

Spatial Conservation Prioritisation as Part of a General Land Use Planning Process

CONTEXT. Land use decisions are essential for reaching of biodiversity conservation targets. Usually, conservation is planned separately from other land use, using specialised approaches such as spatial conservation planning and prioritisation (SCP). This separation of processes makes it difficult to optimise between competing land uses or to plan for land sharing solutions. OBJECTIVES: We present a real-life planning case where SCP was integrated to regional planning process from early on. The aim is (i) to present the process and its results, and (ii) based on the experiences, to evaluate and discuss the potential and challenges of integrating SCP to a general land use planning. METHODS: We present the regional planning of the Helsinki-Uusimaa region in Finland where SCP was integrated as part of the general land use planning process between 2014 and 2018. We applied Zonation software and a diverse collection of spatial biodiversity data and carry out various spatial prioritizations guided by planners and environmental experts. We compared the priority areas to future plans (Uusmaa 2050). RESULTS: We show high spatial variation of biodiversity in Uusimaa region and SCP is able to highlight sites of high importance for biodiversity aware planning. Roughly 70% of biodiversity is outside protection by the Uusimaa 2050 plan draft.CONCLUSIONS: While SCP is relatively well-known by ecologists and nature conservationists, its concepts, framework, and tools are usually not familiar to general land use planners. Integration of SCP can yield to better decisions, but new practices require sufficient resourcing and tight collaboration between the parties.

Spatial and species-level metrics reveal global patterns of irreplaceable and imperiled gecko phylogenetic diversity

Phylogenetic Diversity (PD) is increasingly recognised as a useful tool for prioritising species and regions for conservation effort. Increased availability of spatial and phylogenetic data for reptiles now facilitates their inclusion in phylogenetically-informed conservation prioritisation efforts. Geckos are a highly divergent and diverse clade that comprises almost 20% of global reptile diversity. Their global distribution is coincident with numerous anthropogenic threats, making them worthy of conservation prioritisation. Here, we combine phylogenetic, spatial distribution and extinction risk data for geckos with global human encroachment data to identify both regions and species representing irreplaceable gecko diversity at risk from human pressure. We show that high levels of irreplaceable gecko diversity are restricted to regions under intense human pressure, such as India, Sri Lanka and the Caribbean. There is a lack of extinction risk data for the western regions of Angola and Namibia, and yet these regions harbour high levels of irreplaceable diversity. At the species level, geckos display more unique PD than other lizards and snakes and are of greater conservation concern under our metric. The PD represented by Data Deficient geckos is at comparable risk to that of Endangered species. Finally, estimates of potential gecko diversity loss increase by up to 300% when species lacking extinction risk data are included. Our analyses show that many evolutionarily unique gecko species are poorly known and are at an increased risk of extinction. Targeted research is needed to elucidate the conservation status of these species and identify conservation priorities.

Poor protection of amphibian evolutionary history reveals opportunities for global protected areas

As habitat loss is a major driver of amphibian population declines, protected areas (PAs) can play a crucial role in amphibian conservation. Documenting how well the global PA network captures the evolutionary history of amphibians can inform conservation prioritisation and action. We conducted a phylogenetic gap analysis to assess the extent to which amphibian phylogenetic diversity (PD) is unprotected by the PA network and compared this to other terrestrial vertebrate groups. 78% of amphibian species and 64% of global amphibian PD remains unprotected, which is higher than corresponding figures for squamates, mammals and birds. Amongst amphibians, salamanders were the least well protected, with 78% of PD unprotected, compared with 64% for caecilians and 63% for frogs. We identify areas that offer the greatest opportunity to capture unprotected amphibian evolutionary history. We could capture an additional 29.4% of amphibian PD, representing 40 billion years of evolutionary history, by protecting an additional 1.9% of global amphibian distributions (1.74% of global land area) and increasing the restrictions in 0.6% of amphibian distributions to match the management objectives of PAs in IUCN categories I or II. Importantly, we found that the spatial distribution of unprotected PD was correlated across all groups, indicating that expanding the PA network to conserve amphibian PD can secure imperilled vertebrate diversity more generally.

Low cost conservation: Fishing gear threats to marine species

Understanding conflicts between objectives of fisheries and conservation is the key to finding win-win situations for marine biodiversity and fishers. Many marine species are threatened by harmful interactions with fisheries, but the threats they face are associated with the fishing gear used. Here, we undertake a novel analysis of marine species and their gear-specific threats to evaluate conservation-fisheries trade-offs to identify areas with high competing goals. Our analysis suggests that gillnet and longline fisheries pose the greatest risk to marine species yet deliver relatively low profits, emphasizing the inefficiencies of these gears. We find that the majority of the high seas has low economic fisheries benefits with over 25% of the high seas categorized as areas of ‘conservation prioritisation’ over fisheries.

Applying genomic data to seagrass conservation

Although genomic diversity is increasingly recognised as a key component of biodiversity, it is seldom used to inform conservation planning. Estuaries and keystone species such as the southern African seagrass, Zostera capensis, are under severe anthropogenic pressure and are often poorly protected. In this study we integrated SNP data generated from populations of Z. capensis across the South African coastline into the spatial prioritisation tool Marxan. We included different measures of genomic variation to account for genomic diversity, distinctness and evolutionary potential to explore spatial planning scenarios. We investigated how conservation priority areas identified by targeting only habitat type, differed from those identified by also including genomic measures; further we assessed how different genetic diversity metrics change prioritisation outcomes. All scenarios targeting genomic variation identified unique conservation prioritisation areas compared to scenarios only targeting habitat type. As such, omitting these estuaries from regional MPA networks risks the loss of evolutionarily important populations, threatening resilience and persistence of associated estuarine communities and their ecosystem services. We also observed a high degree of overlap between prioritisation outcomes across targeted measures of genomic variation. As such, by including even single measures of genomic variation, it may be possible to sufficiently represent the evolutionary processes behind the patterns of variation, while simplifying the conservation prioritisation procedure.

Functionally unique, specialised, and endangered (FUSE) species: towards integrated metrics for the conservation prioritisation toolbox

Identifying species with disproportionate contributions to biodiversity can lead to effective conservation prioritisation. Despite well-established methods for identifying endangered species adding inordinately to evolutionary diversity, in this context functional diversity has been overlooked. Here, we compare different metrics designed to identify threatened species that contribute strongly to functional diversity. We use the diverse and threatened global marine megafauna as a case study. We found that functional contributions of species are not fully captured in a single metric. Although we found a very strong correlation between functional specialisation and distinctiveness, functional uniqueness was only moderately correlated with the other two metrics and identified a different set of top-10 species. These functional contributions were then integrated and combined with extinction risk to identify species that are both important contributors to functional diversity and endangered. For instance, the top-10 Functionally Unique Specialized and Endangered (FUSE) species contains three critically endangered, five endangered and two vulnerable species which - despite comprising only 3% of species - are among the top 10% most functionally unique and hold 15% of the global functional richness. The FUSE index was remarkably robust to different mathematical formulations. Combining one or more facets of a species contribution to functional diversity with endangerment, such as with the FUSE index, adds to the toolbox for conservation prioritisation. Nevertheless, we discuss how these new tools must be handled with care alongside other metrics and information.

Barcoding of estuarine macrophytes and phylogenetic diversity of estuaries along the South African coastline

Plant DNA barcoding serves as an effective approach to building community phylogenies and increasing our understanding of the factors that determine plant community assemblages. The aims of the study were to (i) barcode macrophytes with high estuarine fidelity and (ii) to determine the phylogenetic diversity (PD) of selected South African estuaries for conservation prioritisation. Three DNA barcoding gene regions (rbcLa, matK, and trnH-psbA) were assessed, and community phylogenies were constructed for 270 estuaries. Generally, the matK barcode had the greatest discrimination success rate of 67.4% (parsimony informative sites = 418). Closely related species formed clades that also represent estuarine habitat types. Estuaries with high phylogenetic diversity along the southeast coast were associated with a combination of mangrove and salt marsh habitats. Species richness was strongly and significantly correlated with PD (r = 0.93; p < 0.000). Based on mean pairwise distance (MPD), more temperate estuaries (56) showed significant phylogenetic clustering compared to subtropical estuaries (24) (p < 0.05). Similarly, based on mean nearest taxon distance (MNTD), significant phylogenetic clustering was highest in temperate estuaries (50) compared to subtropical estuaries (12) (p < 0.05). This suggests that the coexistence of plant species in estuaries is structured by both biotic and abiotic interactions.