Long-term monitoring of ecological communities is an integral part of land management and biodiversity conservation. Monitoring design influences the patterns captured by collected data. As a result, sampling extent, resolution (grain size), and sample size (density) are often considered in monitoring efforts. Within these parameters, however, subsequent methodological choices can also influence resulting compositional patterns through effects on taxa detection.
We explored how sampling methodology affects the conclusions of vegetation monitoring, focusing on metrics of vegetation community diversity and change. Data were drawn from 48 permanently-marked transects in a sagebrush-steppe ecosystem in the Columbia Basin of Washington state, USA. One hundred points along each transect were monitored five times over a decade, collecting compositional data at three nested scales: point-intercept, 15 x 15 cm quadrat, and 60 x 60 cm quadrat. We compared patterns in diversity and compositional change over time as a function of these differences in sampling methodology, while holding sampling extent and density constant.
Results/Conclusions
Compositional patterns differed strongly depending on sampling methodology. Species accumulation curves indicated saturation at different levels: both quadrat datasets captured greater diversity than point intercept. This disparity was due to better detection of less abundant taxa and certain growth forms in quadrats. The functional profiles of the communities also varied with sampling methodology. For example, all of the additional species detected using 15 x 15 cm quadrats compared to point-intercept sampling were forbs, a life form that contributes most of the plant diversity in this system. Compositional changes over time also differed depending on sampling method. Notably, communities sampled using point-intercept were misleadingly depicted as stable, whereas communities sampled using quadrats responded to variation in weather and disturbances.
Overall, our results indicate that lower sampling intensities run the risk of biasing compositional datasets. Much greater intensities of point-intercept sampling would be required to detect many of the forbs and other important taxa that were detected in the quadrats. Low-intensity sampling methodologies can provide reasonable estimates for some metrics, but can misrepresent community sensitivity to agents of change. We recommend that sampling methodologies be considered relative to the scale of the organisms that land managers are interested in preserving or restoring.