Disentangling plant trait responses to livestock grazing from spatio‐temporal variation: the partial RLQ approach

Abstract

AbstractQuestionsWhich plant traits consistently respond to grazing in different years and across habitat‐related environmental heterogeneity? Does the proposed partial RLQ approach allow partitioning of grazing‐related environmental parameters from other environmental and temporal variations?LocationSemi‐arid savannas of central Namibia.MethodsWe recorded nine quantitative and 12 categorical traits from 87 plant species along grazing gradients in semi‐arid Namibian rangelands. We sampled from gradients in different habitat settings in 2 yr with differing total rainfall amounts. We first examined trait–environment relations with RLQ analysis. To remove confounding effects of temporal and habitat‐related environmental variation on trait performance, we introduced a novel partial RLQ analysis approach. Furthermore, we used the fourth‐corner statistic to quantify and test relations between traits, environmental factors and RLQ axes.ResultsHabitats and years had strong influences on trait patterns. After removing environmental variation caused by habitats and years, grazing became the most influential factor on trait responses. Traits negatively correlated with increasing grazing pressure were common to perennial grasses, such as long and entire leaves, anemochorous dispersal and rhizomatous growth. Positively correlated traits were those common to herbaceous, annual plants with a prostrate–creeping habit, compound leaves, high specific leaf area (SLA) and exo‐ or endozoochorous dispersal. Some previously acknowledged grazing response traits, like growth form and plant height, were strongly influenced by variations in habitats and years and showed no significant correlation with grazing pressure.ConclusionWe emphasize that some traits that respond to grazing may also vary under different habitat conditions and among years, especially in highly variable environments like semi‐arid savannas. When analysing trait–environment relations we recommend using approaches that partition environmental variation, particularly when applying broad sampling schemes at larger geographical scales.