5.2 Understanding habitat – habitat suitability indices
In most environmental management settings, the effects of different management options on biodiversity need to be taken into account. To support such decisions, GIS can be used to understand how a species or group of species respond to changes in habitat. Identifying species habitat preferences is a very widespread ecological application of GIS: there have been several hundred academic papers published using GIS in this way. Habitat suitability indices (HSI’s) are a measure of the suitability of habitat for a given species or group of species based on an assessment of habitat attributes. HSI’s are indices in the sense that they usually combine many different variables (such as elevation, soil type, and land cover) into a single composite measure.
HSI’s have several uses in environmental management, such as :
- To provide a complete picture of a species distribution when only incomplete data are available. Many surveys map species distributions over only limited parts of a study area, such as those that are the most accessible by road or path. HSI’s may be used to estimate the species distribution over the remaining unsurveyed portion of the study area, based on the available habitat, such as the presence of mature broadleaf woodland.
- Predicting future changes in species distribution: Many actions in environmental management potentially affect the local ecology. The introduction of a new access road, the fencing off of an area to prevent deer grazing, and the introduction of an agri-environment scheme to promote hedgerow planting all influence local habitat. Will clear-felling a stand of timber adversely affect local buzzard populations? HSI’s can be used to forecast the likely effect of these actions on a particular species and thereby assess the effect of management changes on a species. The effects of other changes not resulting from environmental management – notably climate change – can also be assessed using HSI’s.
- Targeting data collection in the field: In some instances, ecological surveys are conducted to identify a particularly important species, such as a rare orchid. A HSI can help identify the sites where the orchid is most likely to be found, based on map layers of soil type, surrounding vegetation and local climate. Using the HSI to target the field survey enables field ecologists to maximise their chances of finding as many orchids in the field as possible.
Creating Habitat Suitability Indices
HSI’s are generally created by one of two types of method:
Data-driven methods – ecological niche modelling: Usually, this involves a statistical analysis of data on a species’ current distribution. One of the most simple approaches is a so-called ‘environmental envelope’ approach. For a species such as the mopane tree (found in southern Africa), this would involve calculating the maximum and minimum rainfall for any site where the mopane trees have been recorded. The same calculation can be made for other variables, such as soil clay content or temperature, until an ‘envelope’ of maximum and minimum values for many variables has been created that describes the types of habitat occupied by the species. A great many, more sophisticated statistical approaches have also been developed, including logistic regression, auto-logistic regression, and Dempster-Shafer theory. Several specialist software packages have been developed to assess species distributions, including the DIVA-GIS package, GARP, BIOCLIM, and various packages such as Dismo and AdehabitatHS within the R software (see below).
Some of these techniques are better suited to data which describe where a species is present and also where it is absent (e.g. where field biologists have searched for the species but not found it) – presence/absence data . For example, logistic regression is a specialist form of statistical regression analysis used with presence/absence data. In logistic regression, the variable being predicted can take one of two values. Thus, in a logistic regression analysis of species distribution, the variable might indicate whether a species is present or absent. Other techniques are designed to be used with data that record when a species is present, but where a location lacking a species record might either indicate the species is absent or that the location has never been surveyed – presence only data. Examples of such techniques include the GARP software and Dempster-Shafer theory.
Which of the many methods available works best? This is a difficult question, but there have been several studies that have compared the predictive performance of different techniques. One recent example of this is by Elith et al (2006) and this suggests that some recently developed ‘novel’ techniques have greater ability to predict species distributions than some of the older and more established techniques.
There is also some evidence that it is harder to predict the ecological niche of certain types of species (e.g. more mobile or generalist species) than species that are less mobile and occupy more specialist niches. For example, Poyry et al (2008) found that it was easier to identify the habitat preferences of short-winged, less mobile butterfly species than larger-winged, more mobile butterfly species.
‘Expert’ methods: HSI’s can also be created by convening a panel of ‘expert’ ecologists and asking them to describe a species’ habitat preferences. There are many ways in which such ‘expert’ opinion may be identified (e.g. see the EMRIS web site below). One method – known as the Analytical Hierarchy Process – asks experts to rank pairs of habitat attributes relative to one another. In analysing the distribution of the klipspringer (an antelope adapted to rocky terrain), the importance of steep slopes may be ranked in relation to the frequency of fires and separately in relation to distance to roads. A formula can then be derived for creating an HSI by considering all of the different pairs of rankings. Another GIS approach towards capturing expert opinion uses fuzzy logic, a methodology which captures the inherent vagueness with which an ‘expert’ may talk of concepts such as steep slopes. Fuzzy logic enables the expert’s uncertainty as to whether steep slopes preferred by klipspringer are greater than 20 degrees, 25 degrees or 30 degrees to be captured and used within a GIS analysis.
If you are using ArcGIS Desktop, please download the attached zip file and undertake the GIS activity described in the pdf instruction sheet. For ArcGIS Pro, please download this zip file. This involves investigating the relationship between the abundance of the fulmar, a type of sea bird, and sea depth across the British Isles.
References (Essential reading for this learning object indicated by *)
For ‘data-driven’ approaches to assessing species habitat preferences, a useful starting point is the web site for the freely available DIVA-GIS software, which has been specifically designed for this purpose: http://www.diva-gis.org/
The R statistical software contains several packages for modeling species distributions, including the dismo package: https://cran.r-project.org/web/packages/dismo/dismo.pdf
Yet another popular piece of software for assessing habitat preferences is MaxEnt, which uses a maximum entropy method and is available here: https://biodiversityinformatics.amnh.org/open_source/maxent/
This article compares the performance of some of the many different techniques for ecological niche modelling:
Elith, J., Graham, C. H., Robert, P., Anderson, M. D., Ferrier, S., Guisan, A., Heijmans, R. J., Huettmann, F., Leathwick, J. R., Lehmann, A., Li, J., Lohmann, L. G., Loiselle, B. A., Manion, G., Moritz, C., Nakamura, M., Nakagawa, Y., Overton, J. M. M., Peterson, A. T., Phillips, S. J., Richardson, K., Scachetti-Pereira, R., Schapire, R. E., Soberón, J., Williams, S., Wisz, M. S., and Zimmermann, N. E. (2006) Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29 (2), 129–151.
Several articles have looked at whether it is inherently easier to predict the habitat preferences of particular types of species. For example, this study looks at the predictive accuracy of ecological niche models for different butterfly species:
Poyry, J., Luoto, M., Heikkinen, R., and Saarinen, K. (2008), Species traits are associated with the quality of bioclimatic models. Global Ecology and Biogeography 17 (3), 403-414.
Readings from practical exercise:
The sea bird data set used in the practical is available from: http://www.magic.gov.uk/
Further details of the Etopo1 elevation data set, which we use in the practical, may be found here: http://www.ngdc.noaa.gov/mgg/global/global.html
For further background on the distribution of fulmars and sea depth, see:
Huttmann, F., and Lock, A. R. (1997), A new software system for the PIROP database: data flow and an approach for a seabird-depth analysis. ICES Journal of Marine Science 54, 518–523. http://icesjms.oxfordjournals.org/cgi/reprint/54/4/518.pdf