Observer-driven pseudoturnover in vegetation monitoring is context-dependent but does not affect ecological inference
Prepared by Steffen Boch, Helen Küchler, Meinrad Küchler, Angéline Bedolla, Klaus T. Ecker, Ulrich H. Graf, Tobias Moser, Rolf Holderegger & Ariel Bergamini
To document vegetation and environmental changes across time, resurveys of vegetation plots, as well as monitoring programs investigating plots on a regular basis, are suitable instruments. Such resurvey data become increasingly important for interpreting temporal changes in biodiversity, ecology and conservation management. However, resurveys are prone to several shortcomings that can affect data and subsequently result in misleading conclusions or management recommendations. Keeping such errors consistently low and finding reliable estimates of temporal vegetation and ecological changes is therefore essential in resurvey studies. For instance, the relocation error, i.e. shifts in plot position and in the included vegetation, can be avoided using permanently marked plots. In contrast, observer errors, including different results of the same observer at different times or results from two or more observers, cannot be fully avoided. Observer errors are affected by personal biases such as species knowledge and the experience in conducting vegetation surveys (an error that might decrease with time in long-term projects), but also mental fatigue and physical stress. For example, species lists of vegetation plots compiled on the same day by different observers usually differ from each other to a certain degree. This well-studied phenomenon describing observer differences between species lists from the same plot is called pseudoturnover, i.e., the amount of shared and differing taxa. This error can be largely attributed to the overlooking of species but also, to a lesser extent, to misidentification. This suggests that commonly used measures to characterize temporal vegetation changes, such as species numbers, are highly affected by observer differences.
But how large is the pseudoturnover in vegetation surveys, which factors are affecting it, can it be reduced and are there alternative measures that are robust against pseudoturnover and therefore provide reliable estimates of temporal vegetation and ecological changes? To fill these knowledge gaps, we conducted the present study. Over seven years, we double-surveyed a total of 224 plots that were marked once in the field and then sampled by two observers independently on the same day. By this, we excluded several error sources, such as relocation error, plot size differences, seasonal changes in vegetation composition, and phenological differences. Using the lists of recorded vascular plant species, we then calculated the commonly used mean ecological indicator values for each survey and the pseudoturnover between species lists of the two observers from the same plot.
We found that pseudoturnover was on average 29% when raw species lists were compared, a value that is within the reported range of pseudoturnover in other studies. However, by applying simple aggregation steps to the species list, pseudoturnover was reduced to 17%. Pseudoturnover further varied among habitat types and declined over the years, indicating a training effect among observers. Most overlooked taxa, responsible for pseudoturnover, had low cover values. Notably, mean ecological indicator values were robust against observer differences.
To minimize pseudoturnover, we therefore suggest continuous training of observers and species-list aggregation prior to analysis. We further conclude that mean ecological indicator values can provide reliable estimates of temporal vegetation and ecological changes.
This is a plain language summary of the paper of Boch et al. published in Applied Vegetation Science (https://doi.org/10.1111/avsc.12669).
