How Landscape Ecology Utilizes Spatial Analysis: A Journal Article Review
Purpose
Mendes et al. (2016) were interested in understanding the processes that occur in heterogeneous landscapes at different focal scales and how these processes influence bat richness and activity in the Baixo Vouga Lagunar (BVL) region in Spain. They were specifically interested in how the spatial heterogeneity affects dispersal patterns and foraging behaviours as resource availability and suitability varies across habitat patches over different scales.
Method/Spatial Analysis
The authors gathered primary data through walking surveys within the BVL region, a heterogeneous landscape “with a mosaic of natural, semi-natural and human-altered landscapes” (p. 207). The authors sampled bat activity acoustically and divided the 12 found bat species into three guilds based on their eco-morphological traits (one guild was dropped due to low species count). Insect data was collected simultaneously to infer food availability.
In order to perform analysis, Mendes et al. (2016) included 12 landscape predictors and 12 local predictors (using previous research to support some of these predictors) and used a hierarchical partitioning analysis (HPA) to statistically evaluate these predictors to explain the variation in species richness, total bat activity and guild-activity.
To compute the landscape variables, the authors used ArcGIS 10.0 and ArcMap to create three different buffers (1.5, 3 and 6km radii) around the midpoint of three transects intended to cover the known home ranges of the detected species. They used secondary data supplemented by their own field incursions to divide up the study area into six land classes. They then calculated land class percentages within each transect in ArcMap. They used ArcGIS to calculate total edges between differing adjacent patch types and watercourse lengths. To calculate landscape metrics, the authors used Fragstats 4.1 to calculate patch density, Shannon diversity index, area-weighted mean patch shape area and contagion. They used these metrics to analyze how fragmented the study area was, the various complexity of different patch habitats and the connectivity of similar patches. The authors used local independent variables such as weather, food availability and habitat related predictors which they directly measured within each sampling transect. They also included the normalized difference vegetation index (NDMI) and habitat type.
The authors inputted these variables into their statistical analysis to identify which variables best predicted bat activity and species richness at each scale. They used a maximum of three variables holding the highest z-score (strongest independent contribution) and their model construction was made using Akaike’s Information corrected for small sample sizes (AICc).
Results
Using the HPA results and model selection, the authors determined that depending on the species and scale, bats responded to different factors. At the landscape level, fragmentation and edge length were the most influencing. A finer scale revealed keystone habitat and vital resources were the limiting factors. At a local scale, prey availability and temperature were the main factors influencing bat activity.
I rate this paper 9/10 because I believe the approach used is appropriate to address the authors’ research question. It allowed the authors to statistically determine which aspects of the landscape affected bat activity the most at different scales. This multi-scale approach allowed the authors to identify and measure dependent and independent variables to explain bat activity and richness which can perhaps be applied to conservation policies in the future. While restricted to bat activity, it at least highlights the necessity to take into account how a species is affected at multiple scales.
Reference List
Mendes, E. S., Fonseca, C., Marques, S. F., Maia, D., & Ramos Pereira, M. J. (2016). Bat richness and activity in heterogeneous landscapes: Guild-specific and scale-dependent? Landscape Ecology, 32(2), 295-311. doi:10.1007/s10980-016-0444-0