Systematic monitoring of heathy woodlands in a Mediterranean climate--a practical assessment of methods.

Practical and useful vegetation monitoring methods are needed, and data compatibility and validation of remotely sensed data are desirable. Methods have not been adequately tested for heathy woodlands. We tested the feasibility of detecting species composition shifts in remnant woodland in South Australia, comparing historical (1986) plot data with temporal replicates (2010). We compared the uniformity of species composition among spatially scattered versus spatially clustered plots. At two sites, we compared visual and point-intercept estimation of cover and species diversity. Species composition (presence/absence) shifted between 1986 and 2010. Species that significantly shifted in frequency had low cover. Observations of decreasing species were consistent with predictions from temperature response curves (generalised additive models) for climate change over the period. However, long-term trends could not be distinguished from medium-term dynamics or short-term changes in visibility from this dataset. Difficulties were highlighted in assessing compositional change using historical baselines established for a different purpose in terms of spatial sampling and accuracy of replicate plots, differences in standard plot methods and verification of species identifications. Spatially clustered replicate plots were more similar in species composition than spatially scattered plots, improving change detection potential but decreasing area of inference. Visual surveys detected more species than point-intercepts. Visual cover estimates differed little from point-intercepts although underestimating cover in some instances relative to intercepts. Point-intercepts provide more precise cover estimates of dominant species but took longer and were difficult in steep, heathy terrain. A decision tree based on costs and benefits is presented assessing monitoring options based on data presented. The appropriate method is a function of available resources, the need for precise cover estimates versus adequate species detection, replication and practical considerations such as access and terrain.