Niche modelling of marsh plants based on occurrence and abundance data

Abstract

The information of species' response (optimum or critical limits along environmental gradients) is a key to understanding ecological questions and to design management plans. A large number of plots (762) from 70 transects of 13 wetland sites in Northeast China were sampled along flooding gradient from marsh to wet meadow. Species response (abundance and occurrence) to flooding were modelled with Generalized Additive Models for 21 dominant plant species. We found that 20 of 21 species showed a significant response to flooding for the occurrence and abundance models, and four types of response were found: monotonically increasing, monotonically decreasing, skewed unimodal and symmetric unimodal. The species with monotonically increasing response have the deepest flooding optimum and widest niche width, followed by those with unimodal curve, and the monotonically decreasing ones have the smallest values. The optima and niche width (whether based on occurrence or abundance models) both significantly correlated with the frequency, but not with mean abundance. Abundance models outperformed occurrence models based on goodness of fit. The abundance models predicted a rather sharp shift from dominance of helophytes (Carex pseudo-curaica and C. lasiocarpa) to wet meadow species (Calamagrostis angustifolia and Carex appendiculata) if water levels drop from about 10cm above soil surface to below the surface. The defined optima and niche width based on the abundance models can be applied to better instruct restoration management. Given the time required to collect abundance data, an efficient strategy could be to monitor occurrence in many plots and abundance in a subset of these.

Keywords: Distribution; Generalized Additive Models (GAM); Herbaceous marsh; Niche width; Optimum; Species response curve.