The potential of trait data to increase the availability of bioindicators: A case study using plant conservatism...

Biological indicators are commonly used to evaluate ecosystem condition. However, their use is often constrained by the availability of information with which to assign species-specific indicator values, which reflect species' responses to the environmental conditions being evaluated by the indicator. As these responses are driven by underlying traits, and trait data for numerous species are available in publicly accessible databases, one possible approach to approximating missing bioindicator values is through traits. We used the Floristic Quality Assessment (FQA) framework and its component indicator of disturbance sensitivity, species-specific ecological conservatism scores (C-scores), as a study system to test the potential of this approach. We tested the consistency of relationships between trait values and expert-assigned C-scores and the trait-based predictability of C-scores across five regions. Furthermore, as a proof-of-concept exercise, we used a multi-trait model to try to reconstruct C-scores, and compared the model predictions to expert-assigned scores. Out of 20 traits tested, there was evidence of regional consistency for germination rate, growth rate, propagation type, dispersal unit, and leaf nitrogen. However, the individual traits showed low predictability (R²  = 0.1-0.2) for C-scores, and a multi-trait model produced substantial classification errors; in many cases, >50% of species were misclassified. The mismatches may largely be explained by the inability to generalize regionally varying C-scores from geographically neutral/naive trait data stored in databases, and the synthetic nature of C-scores. Based on these results, we recommend possible next steps for expanding the availability of species-based bioindication frameworks such as the FQA. These steps include increasing the availability of geographic and environmental data in trait databases, incorporating data about intraspecific trait variability into these databases, conducting hypothesis-driven investigations into trait-indicator relationships, and having regional experts review our results to determine if there are patterns in the species that were correctly or incorrectly classified.