By Ines Standfuß, Deutsches Zentrum für Luft- und Raumfahrt

Animal habitats are not only subject to alterations by regular phenomena like seasonal variations, but are increasingly exposed to human intervention. The most prominent example is likely human-induced climate change that is modifying environmental conditions worldwide. In addition to climate change, which affects entire regions mainly in the medium and long term, human activities can have an immediate impact on the small-scale landscape mosaic. The direct alteration of habitat patches through agricultural management is an example of how this can have both positive and negative impacts on wildlife. The removal of vegetation through mowing/harvest on the one hand, can increase abundance and accessibility of prey items in the short term. On the other hand, many nests and thus, the offspring of ground-nesting birds, are lost yearly through the employment of heavy machinery.

Understanding how such small-scale environmental drivers influence animals’ habitat use, selection and distribution as well as their individual fitness and population demographics is crucial for animal ecology and conservation. There is hence, an urgent need to characterize information on small-scale vegetation dynamics in ecologically meaningful ways. Unfortunately, to date these temporal variations remain mostly overlooked in ecological studies and if yet considered, are often not tailored to the behavioral needs of the studied species.

Remote sensing time series, like multi-temporal Landsat-7/-8 imagery, have proven suitable for mapping small-scale vegetation dynamics. These datasets allow for deriving intra-annual NDVI profiles of fine-scale habitat features like individual fields and reflect their vegetation phenology. Additionally, they enable to identify sudden declines in NDVI that are often associated with agricultural practices such as harvest/mowing. As an established indicator for leaf-unfolding/loss of canopy structure, the so-called half maximum (Figure 1) marks the points between local minima/maxima in NDVI profiles. An index like this is therefore highly promising to identify resource variations on field-level – i.e. in this case, periods with good or poor prey accessibility for species relying on short vegetation – over time.

Figure 1: Schematic representation of an intra-annual NDVI profile and the half-maximum which should enable to distinguish between periods with good or poor prey accessibility for species who rely on short vegetation to forage.

Using the white stork as test case, we investigate if time series are in fact suitable to characterize such small-scale variations. Storks are breeding increasingly in landscapes dominated by agricultural land use. Here, vegetation becomes too tall during summer and thus, can impede or even prevent access to their prey. Hence, the birds are known to forage mainly in fields with vegetation in the early growth phase or after agricultural practices have taken place. We process one year of Landsat time series and derive intra-annual vegetation dynamics that we assume are related to differences in storks’ prey accessibility: the half-maximum on field-level. Information on used and unused fields, foraging duration and daily visits per field are obtained from GPS-telemetry data of 18 breeding white storks from a German population. We combine these two data sets and explore storks’ habitat use and compare habitat use with habitat availability. Additionally, we test temporal predictors distinguishing resource variations – i.e. good or poor prey accessibility – in habitat selection models.

Our results show that the studied storks prefer to forage in fields that we classified as having a good prey accessibility. The birds visited these habitats more frequently and partially longer than those with supposedly poor prey accessibility. Compared to their relative availability, storks further used habitats with good prey accessibility in larger shares than fields characterized by poor prey accessibility. Finally, our habitat selection models indicate that storks preferably select habitats closer to their nest – typical for central place foragers – and also those with supposedly good prey accessibility (Figure 2 A/B). 

The important outcome of our study is, that time series hold indeed potential to map ecologically relevant small-scale vegetation dynamics. The application of such data should therefore no longer remain limited to studies of large-scale phenomena, such as vegetation phenology along migration routes. Rather, their potential to characterize various kinds of small-scale resource variations should be explored further, to uncover how human actions directly affect various aspects of animal behavior and the survival of species eventually.

Figure 2: Predicted probabilities of habitat selection of two tested predictors used for habitat selection modelling of storks. The predictors characterize the decrease in habitat quality with increasing distance from the nest (A) and distinguish variations prey accessibility over time (B) on field level.

Read the full paper: Time series enable the characterization of small-scale vegetation dynamics that influence fine-scale animal behavior – an example from white storks’ foraging behavior. Standfuß et al.