Identifying where species occur is an important but challenging aspect of many conservation projects. Species occurrence can be thought of in several ways. At the most basic level, it describes the geographic area in which a species can be found. This is the range map that everyone has seen in one field guide or another. While useful for understanding what species could be in a given area, this coarse data does not account for a species’ habitat requirements or behavior. There may, in fact, be large portions of the range where a species will never be found because the habitat is unsuitable. At the other end of the spectrum, species occurrences can be visualized as points on a map where individuals have been observed. This offers a better understanding of where species actually occur but can be heavily influenced by sampling bias or a species’ cryptic nature. Areas may have no presence points simply because nobody has ever searched in that area. For most species, it is likely impossible to ever effectively survey the entire range or even a portion of the range, creating an obvious need for a tool that allows scientists to predict where a species will occur over large geographic areas.
Conservation biologists commonly approach the complex challenge of identifying where species occur by building a model that uses available data to predict species distributions. There are many approaches to this process, but it generally follows a similar pathway. Species occurrence points are obtained from all legitimate sources for the area of interest. These observations can come from natural history collections, other scientists, state and federal wildlife databases, and through the general public. The invention of apps like HerpMapper and iNaturalist have made it easier than ever for anyone to record observations of wildlife that can be used for applications like this. Once observations are compiled, they can then be overlaid on top of GIS layers that represent the environmental differences across the area. These layers are laid out in a large grid (see below). Environmental data can vary depending on what is important for a particular species, but commonly include things like landcover type (forests, urban, farmland, etc.), percent canopy cover, soil type, and elevation. A statistical model is then created from these datasets that predicts where a species occurs across the study area based on the habitat values in locations where the species has already been documented.
This approach makes intuitive sense and is based on the idea that each species has a set of environmental and biological parameters that govern where it can live. This is commonly referred to as the niche. For most species, we have insufficient data to model their actual niche at the landscape scale. The above modeling process is therefore a useful substitute for approximating where suitable habitat occurs. This type of model is generally referred to as a species distribution model (SDM) or habitat suitability model. SDMs have been commonly used in species conservation for many years, and they are often one of the first things created for a species to aid in range-wide conservation planning. SDM outputs are generally presented as scores ranging from 0–1 for each cell of a grid that covers the study area. Higher values indicate a higher relative suitability to other points based on the environmental data at the original set of species occurrence points. For example, if most of the species observation come from locations with sandy soils then the model output will identify these areas as having a higher suitability than areas with other soil types. See Crawford et al. (2020) for an excellent example of SDMs applied to some of the Southeast’s rarest herpetofauna.
SDMs are an incredibly useful tool for several reasons. They facilitate the visualization of potentially suitable habitat at large spatial scales, which can be critical for identifying areas that should be protected or identifying potential wildlife corridors between protected areas. Model outputs can also be used to examine how individuals could move across the landscape and identify populations that are likely connected by dispersal events. Biologists can use the output to identify potential study sites or areas where additional survey effort is needed. Are there locations with no occurrence points but that have potential habitat? This type of modeling approach can also be applied to areas outside of a species current distribution, which is particularly relevant for understanding the potential distribution of an invasive species (Crall et al. 2013).
It is ultimately impossible to know exactly where species occur across their entire range, and distribution (SDM results provide a useful prediction of a species’ distributions that can inform on-the-ground conservation.), of course, change over time. SDM results provide a useful prediction of a species’ distributions that can inform on-the-ground conservation. It is important to remember that the outputs are just predictions, and there are often limitations with the data available to create these models. Many rare species have few occurrence points available, and we may simply not have adequate survey data to build a good model. The quality of the available environmental data varies significantly depending on the study area and variable of interest. It can be challenging to identify data sources that adequately describe a species’ hypothesized relationship to its environment. Without perfect data, model results should often be taken with at least a small grain of salt. However, their usefulness as a conservation tool that can be applied to a variety of different situations makes SDMs an important part of the conservation toolbox.
One of the projects that I am currently working on is building a range-wide SDM for Eastern Indigo Snakes. Current habitat models for indigo snakes only cover the state of Florida and are of limited usefulness when attempting to understand conservation questions across the range. Indigo snakes present an interesting challenge for this type of effort because their ecology changes moving from north to south. In southern Florida, indigo snakes are not reliant on tortoise burrows and occupy habitat types not available in northern Florida and Georgia. This type of ecological complexity must be incorporated in the model for its predictions to be as accurate as possible. Indigo snakes also occur in a landscape that is experiencing rapid environmental change in many locations. Careful consideration must therefore be given to which occurrence points are included in our habitat model (i.e., do points from 15 years ago still represent current habitat conditions).
The goals of this project are twofold. First, as mentioned above, the SDM will provide an excellent tool for visualizing potential indigo snake habitat across the entire range. The model results will be used to examine connectivity between different populations and assess how populations can be better connected through land protection and management. Second, we will use the results of the habitat model to inform future population viability modeling for indigo snakes. Snake populations are notoriously difficult to study because of their low detectability. A map of potential habitat will allow us to start some of this population viability work by delineating where potential populations exist and estimating how many snakes could be in each population. We will then be able to examine different management strategies to better understand how they could impact populations over the long-term. This is an exciting project for indigo snake conservation, and I will share the results of our efforts as we work through this project over the next couple of years!
Crall, A.W., C. S. Jarnevich, B. Panke, N. Young, M. Renz, and J. Morisette. 2013. Using habitat suitability models to target invasive plant species surveys. Ecological Applications 23: 60–72.
Crawford, Brian A., John C. Maerz, and Clinton T. Moore. 2020. Expert-informed habitat suitability analysis for at-risk species assessment and conservation planning. Journal of Fish and Wildlife Management. In-Press.