Use of species richness estimators improves evaluation of understory plant response to logging: a study of redwood forests

The recovery of native understory plant communities after timber harvest has received a great amount of attention worldwide. However, most of these studies have ignored the potentially significant effects that differences in habitat across forest development may have on the species–area relationship, even though sampling efficacy among forest developmental stages may differ markedly as a result of equal sampling within stages. We examined vascular plant community structure in coastal redwood forests of northern California (USA) in forest patches in each of four forest stages that develop with forest age: Initiation, Closure, Mature, and Old-growth. We also used a set of traditional and novel approaches to assess sampling efficacy and guide our sampling design. We initially sampled 75 circular plots (0.04 ha) in stands within each forest stage across a 1,347 km² area. The species–area relationship suggested we had adequately sampled all stages after the first season of data collection, but non-parametric richness estimators (Incidence-based Coverage Estimator or ICE) and Michaelis–Menten means (MMMeans) rarefaction curves, indicated inadequate sampling of the older stages. Thus, we added four more samples within the Old-growth forest stage, establishing a more equitable sample of all stages. Our full data set demonstrated significant differences between plant communities in previously harvested forests (‘managed’ = Initiation, Closure, and Mature) and ‘unmanaged’ forests (Old-growth). Though several understory species were present in all stages, a suite of Old-growth indicator and restricted species suggest that recovery of some species may take long periods in managed forests. Interestingly, a separate suite of common understory species were lacking in Old-growth, and species richness in this stage was lower than in the mature managed forests. We discuss how this departure from the prediction of peak richness in Old-growth may relate to fire suppression in Old-growth forests. Lastly, analysis of our initial, smaller sampling of plots suggested that Old-growth plots were the least species rich. However, additional sampling of the Old-growth stage demonstrated that Old-growth plots were as rich as Closure plots. This has important implications for all studies concerned with measuring recovery after forest disturbances; stage-dependent habitat-mediated differences that affect sampling efficacy may not be captured by traditional sampling methods.