Skip to content

Home > News > Towards an objective definition of optimal habitat for small mammal (...)

Towards an objective definition of optimal habitat for small mammal species


How to define objectively the optimal habitat of small mammal species ? This is the question addressed in an article published in PLoS One by researchers from the French-British-Chinese network, which began working together in 1994.

Marston, C., Raoul, F., Rowland, C., Quéré, J.-P., Feng, X., Lin, R., Giraudoux, P., 2023. Mapping small mammal optimal habitats using satellite-derived proxy variables and species distribution models. PLOS ONE 18, e0289209.

Small mammals play an important role in terrestrial ecosystems, dispersing seeds, modifying soil structure and vegetation composition. They are also the prey of many predators, and are therefore often keystone species in these communities. Some species undergo population peaks (outbreaks), which can have a major impact on agricultural production and the transmission of pathogens (see e.g. Arvicola terrestris in the Jura massif).

Montane forest and meadows near Narati, Tian Shan, ChinaIt has been shown that those population peaks are more likely in landscape with larger proportion of favourable habitat, also named ’optimal habitat’, in a landscape. To predict outbreaks and the risks associated with them, it is therefore essential to be able to objectively define the optimal habitat for each species of concern. There are a number of methodological difficulties with this definition. When long-term studies cannot be carried out (emergency, lack of funding, etc.), data are based on snapshot sampling. Thus, multi-annual variations in abundance (cyclic or not) may lead to find that a species is absent from its optimal habitat (e.g. during a phase of low density), and conversely, during population peaks, the presence of a species may be observed in unfavourable habitats, due to the spillover of populations from genuinely favourable habitats. As a result, the optimal habitat for each species is more often than not defined by ’expert opinion’, which can be a source of subjectivity.

Random forest predicted abundance of Microtus obscurus for the Narati study areaThe aim of this study is to use measurements derived from satellite data to define the optimal habitats for species sampled by heterogeneous means (specific traps, index transects), which is often required by field constraints. Satellite data quantifying the characteristics and dynamics of the landscape according to their spatial and temporal variations in reflectance (particularly seasonal) were obtained from GoogleEarth. The data on small mammals were obtained in the Narati area, Tian Shan, China, in September 2006 and in the Sary Mogul area, Kyrgyzstan, in September 2014 during sampling campaigns carried out as part of a French-Sino-British cooperation funded by the US Institutes of Public Health and the Wellcome trust. Statistically significant satellite variables were selected using the "boruta" algorithm, and their association with the distribution of various small mammal species was modelled using "random forests".

General view of the study area of Sary Mogul, Kirgyzstan

Although this approach does not completely overcome all the possible biases mentioned above, it does define optimal habitat by a combination of dynamic landscape variables accessible from satellites. This means that we can move from a definition of optimal habitat based on "expert opinion" to one based on objective measurement of the distribution of small mammal species themselves. The study shows that there are significant differences between the two types of estimation.


  • Patrick Giraudoux, professor emeritus of ecology, Chrono-environment, France
  • Christopher Marston, Ecological Remote Sensing Specialist at UK Center for Ecology and Hydrology, Lancaster, UK.

Part of the crew and more at Narati, 23 September 2006