The nearest-plant method is robust and powerful enough for different survey teams to monitor change in mesic grassland species composition

Abstract

A long-term rangeland monitoring programme needs to employ a field survey  technique that is practicable, precise, powerful enough to distinguish change, not prone to worker bias, and able to distinguish real change from operator error arising from staff turnover. These criteria were used to evaluate a widely used grassland sampling technique in South Africa – the nearest-plant (with 200 points) method (NP) – against common alternatives, namely NP excluding forbs (NP-nf), the plant number scale (PNS; a cover-abundance method) and quadrat frequency (QF), using multivariate ordination and permutation tests. Four trained teams surveyed four grasslands using each method. PNS took more than twice as long as the other techniques, which were similarly rapid. Estimates of composition using NP methods were the most precise and PNS was least repeatable, with QF intermediate. Compositional differences between sites were most finely distinguished using NP-nf, followed by NP and QF. PNS was able to detect only marked differences and had the greatest potential for surveyor bias. The NP method, with or without including forbs, is therefore recommended for monitoring the species composition of mesic, dense grassland. Some suggestions for monitoring agencies on how best to use  multivariate methods are presented.

Keywords: grassland, monitoring, multivariate analysis, permutational analysis of variance, rangeland condition