Towards more predictable and consistent landscape metrics across spatial scales

Habitat change and fragmentation are considered key drivers of environmental change and biodiversity loss. To understand and mitigate the effects of such spatial disturbances on biological systems, it is critical to quantify changes in landscape pattern. However, the characterization of spatial patterns remains complicated in part because most widely used landscape metrics vary with the amount of usable habitat available in the landscape, and vary with the scale of the spatial data used to calculate them. In this study, we investigate the nature of the relationship between intrinsic characteristics of spatial pattern and extrinsic scale-dependent factors that affect the characterization of landscape patterns. To do so, we used techniques from modern multivariate statistics to disentangle widely used landscape metrics with respect to four landscape components: extent (E), resolution (R), percentage of suitable habitat cover (P), and spatial autocorrelation level (H). Our results highlight those metrics that are less sensitive to change in spatial scale and those that are less correlated. We found, however, significant and complex interactions between intrinsic and extrinsic characteristics of landscape patterns that will always complicate researcher's ability to isolate purely landscape pattern driven effects from the effects of changing spatial scale. As such, our study illustrates the need for a more systematic investigation of the relationship between intrinsic characteristics and extrinsic properties to accurately characterize observed landscape patterns.