segRDA: An r package for performing piecewise redundancy analysis

Abstract

Redundancy analysis is widely used to correlate species matrix against environmental descriptors. This analysis assumes that the response variables have the same continuous linear trend over the whole gradient, which is often not the case. To overcome this issue, we propose the piecewise redundancy analysis (pwRDA). The pwRDA allows the relationship between the response and explanatory variables to break into pieces. The breakpoint between communities is assessed by the split-moving-window (SMW) approach. Both analyses are within segRDA package for the R environment. We illustrate the use of segRDA using simulated and empirical examples with a varying degree of discontinuity among communities. SMW results provide consistent estimates for both, number and location of community breakpoints along the gradient. In communities with a sharp transition, the pwRDA models were significantly more informative than the traditional RDA model. For smooth transitions between communities, the pwRDA did not differ from the traditional RDA. The combined use of SMW and pwRDA allows one to identify breakpoints and transition zones among ecological communities along environmental gradients. This approach is particularly relevant when species–habitat associations may differ among communities.