Understanding how patterns of fragmentation change across landscapes is important, but such analyses are complicated by the boundary and the modifiable areal unit problems. The cross-boundary connections procedure and moving window analysis improve our understanding but both have limitations. A new method for assessing changes in patch-based metrics across landscapes combining the cross-boundary connections procedure and a moving-window analysis is presented and applied at multiple scales across Nova Scotia. Comparisons of effective mesh size ( \({m}_{eff})\) across the province were made based on (1) reporting units (ecodistricts) and (2) a moving-window analysis. Both methods were conducted with and without the application of the cross-boundary connections procedure. \({m}_{eff}\) values increased with the application of the cross-boundary connections procedure, and in general the smaller the reporting unit or cell size, the greater the difference between the two methods, highlighting the need to consider the influence of size and shape of reporting units in these sorts of analyses. Combining the cross-boundary connections procedure and a moving window into the same analysis better captures the actual landscape structure than methods which rely on political or ecological boundaries that may be derived from forces unrelated to the pattern in question. In doing so, this analysis produces results that are more ecologically meaningful and thus can be used to make better-informed management decisions. The method presented provides a new way of analysing how patterns in patch-based metrics change across landscapes. Although \({m}_{eff}\) was used here, it could be applied to any patch-based metric that measures landscape fragmentation.
