Post provided by Veronica F. Frans

Species distribution models (SDMs) are popular methods for identifying important habitats for species, but what does it take to translate SDMs into conservation practice? In this post, Veronica Frans discusses the applications of iSDMdb as featured in the paper, “Integrated SDM database: enhancing the relevance and utility of species distribution models in conservation management”, recently published in Methods in Ecology and Evolution.

Unlocking the black box of SDMs

There are a lot of questions that wildlife rangers, managers, and decision-makers can ask from SDMs when trying to prepare for species’ range shifts or expansions due to climatic or anthropogenic pressures or conservation success. Yet, results are complex, often coming out of a “black box“, and difficult to interpret, which creates hurdles for fully unleashing the benefits of SDMs for practitioners.

Perhaps managers want to focus on size requirements, land tenure types, or other kinds of spatial factors to designate suitable sites instead of an SDMs’ traditional, continuous predictions. SDMs harbour uncertainty that managers need to account for, such as critiquing predictions with wide confidence intervals or other variations. Managers also need to understand that SDMs can stretch a little too far, for example when the environmental conditions where a species is currently located are extrapolated in novel conditions elsewhere. SDMs can also inform on where habitat suitability is dampened due to the quality of certain predictors (limiting factors), but their final assessment can be quite overwhelming. Managers may only want to focus on the most important factor to target for their next habitat restoration project, but how can they sort through hundreds of model runs?

Ultimately, what can managers do if SDMs cannot account for novel habitat features (e.g. roads, new land cover types) in the model projection range but they still need to somehow consider them because they pose potential threats to the species?

SDMs and SDM frameworks have been increasingly developing over the past decades and are plentiful, but few translate into a format that easily extracts such essential information to suit practitioners’ needs.

iSDMdb framework structure. A more detailed version of this framework and a description of the iSDMdb data fields can be found in Figure 2 and Table 2 of Frans et al. 2021.

With hopes to fill this important gap in conservation application, our paper in Methods in Ecology and Evolution introduces the integrated SDM database (iSDMdb). The iSDMdb is an extended habitat suitability assessment that uses SDM results to create a spatial database of predicted sites that contains additional, user-friendly, informative data fields. These data fields synthesise and summarise SDM predictions and uncertainty, human impacts, restoration features, novel preferences in novel spaces, and management priorities. We created these data fields using a mix of methodologies in addition to traditional SDMs (e.g., multi-state SDMs, multivariate environmental similarity surfaces, multi-criteria decision analysis). We included these within the iSDMdb based on our engagements with wildlife rangers, managers, and decision-makers regarding conservation efforts for our case study species, the New Zealand sea lion (Phocarctos hookeri).

A case study for conservation success

New Zealand sea lions are an endangered species that is currently showing signs for optimism, as their once-extirpated population has been returning to New Zealand’s mainland over the past 30 years. However, their return to the mainland poses concerns as they have multiple distinct habitat requirements within their breeding season, including sandy beaches and coastal forests. The requirement to move inland leaves these sea lions prone to human interactions that can potentially be dangerous, demonstrated by reports of sea lions being killed on roads by cars. Habitat features in the current breeding colonies in the sub-Antarctic islands differ from the mainland areas that they’re recolonising, as roads pose a new threat for them and they’ve also had to adapt to commercial non-native pine forests.

The New Zealand sea lion, our case study species, is the world’s rarest sea lion and can be found up to 2 km inland in forests. As they recolonise the mainland, conservation planning needs to include new adaptations, such as their use of commercial pine forests. Photo credit: Jim Fyfe

The iSDMdb allowed us to identify 395 suitable sites for these sea lions, showing potential for their populations to one day fill their historic range. However, with the caveats of human impacts and novel habitat features on the mainland to consider, the iSDMdb’s additional assessments beyond a traditional SDM indicated that nearly 90% of the sites contained human impacts such as roads and fences that would limit the availability and real-world suitability of these predicted areas. With the iSDMdb serving as a descriptive tool to facilitate discussions and planning, such information can now be put into the hands of rangers, managers, and decision-makers so they can ultimately decide which locations they want to consider and ground-truth.

We also offer the iSDMdb in a variety of formats to increase accessibility and suit various application needs: shapefiles for Geographic Information System (GIS) users, R scripts for advanced and larger-scaled evaluations, printed maps for overviews, spreadsheets for manually sorting and searching through predicted sites, and an interactive HTML map for non-GIS users and/or public access.

Example screenshot of the interactive map version of the iSDMdb. End-users can click on the predicted sites to access the iSDMdb’s simplified data fields for site-by-site evaluation. This interactive map is publicly accessible here. Source: Frans et al. 2021

Accessible to end-users

As a practical end-user, New Zealand Department of Conservation Science Advisor, Dr Laura Boren, offers her opinion on the iSDMdb’s utility: “This sort of work is really helpful to see where we might need to have increased management or community engagement in the future and allows us to be better prepared for it rather than having to respond reactively,” she reports. “We have the example with the New Zealand fur seal / kekeno, which has been successful recolonising, and to have had something like this that could predict likely spots of recolonisation would have been a really helpful resource.”

Road signs can help protect New Zealand sea lions by warning motorists of their presence as they recolonise the mainland. The iSDMdb can help identify locations for such proactive measures. Photo credit: Amélie Augé

As predictive models, the intention of SDMs is to guide proactive efforts. The iSDMdb provides an enhanced guiding tool by pairing predictions with expert knowledge and novel local impacts that sort the wheat from the chaff of predicted suitable areas. This way, only areas of highest potential or highest concern remain as recommended targets for conservation management to allocate sparse resources into efficient decision-making and awareness-building where it is needed most. To this end, the New Zealand sea lion stands exemplary for a wide range of use cases for the iSDMdb, such as in species conservation programs, recolonisation management, translocation efforts, climate change mitigation, and invasive species risk assessments.

The iSDMdb is a flexible tool and can be created using any SDM algorithm or other habitat suitability assessment method, and additional supplemental assessment types can be integrated or excluded as needed. We provide reusable R code to create an iSDMdb for your own species in Appendix S1-S8 of the article, GitHub, the Dryad Repository, and Zenodo.

To find out more about the iSDMdb, read our Methods in Ecology and Evolution article, ‘Integrated SDM database: Enhancing the relevance and utility of species distribution models in conservation management’.