A comparative evaluation of five common pairwise tests of species association

Co-Occurrence Patterns Do Not Predict Mutualistic Interactions Between Plant and Butterfly Species

Biotic interactions are crucial for determining the structure and dynamics of communities; however, direct measurement of these interactions can be challenging in terms of time and resources, especially when numerous species are involved. Inferring species interactions from species co-occurrence patterns is increasingly being used; however, recent studies have highlighted some limitations. To our knowledge, no attempt has been made to test the accuracy of the existing methods for detecting mutualistic interactions in terrestrial ecosystems. In this study, we compiled two literature-based, long-term datasets of interactions between butterflies and herbaceous plant species in two regions of Germany and compared them with observational abundance and presence/absence data collected within a year in the same regions. We tested how well the species associations generated by three different co-occurrence analysis methods matched those of empirically measured mutualistic associations using sensitivity and specificity analyses and compared the strength of associations. We also checked whether flower abundance data (instead of plant abundance data) increased the accuracy of the co-occurrence models and validated our results using empirical flower visitation data. The results revealed that, although all methods exhibited low sensitivity, our implementation of the Relative Interaction Intensity index with pairwise null models performed the best, followed by the probabilistic method and Spearman's rank correlation method. However, empirical data showed a significant number of interactions that were not detected using co-occurrence methods. Incorporating flower abundance data did not improve sensitivity but enhanced specificity in one region. Further analysis demonstrated incongruence between the predicted co-occurrence associations and actual interaction strengths, with many pairs exhibiting high interaction strength but low co-occurrence or vice versa. These findings underscore the complexity of ecological dynamics and highlight the limitations of current co-occurrence methods for accurately capturing species interactions.

Examining the relative influence of dispersal and competition on co-occurrence and functional trait patterns in response to disturbance

Disturbance gradients are particularly useful for understanding the relative influences of competition and dispersal. Shortly after disturbance, plant composition should be influenced more strongly by dispersal than competition; over time, this should reverse, with competition becoming more important. As such, we predicted that plant functional traits associated with high dispersal ability would be over-represented shortly after a disturbance event occurs, while those associated with high competitive ability would have increased representation as time progresses. Additionally, it has been suggested that competitive interactions may contribute to negative co-occurrence patterns; if this is the case, negative co-occurrence patterns should also increase as time-since-disturbance increases. Here, we examine how functional trait and co-occurrence patterns change over time following a herbicide-based disturbance, compared to undisturbed vegetation, in a temperate, old-field grassland dominated by herbaceous perennials. In our study system, negative co-occurrence patterns were most pronounced in disturbed plots one year after herbicide application, consistent with several lines of evidence that dispersal can strongly impact both composition and co-occurrence patterns. Over three years post-disturbance, co-occurrence patterns in disturbed plots decreased, becoming more similar to control plots. This pattern is inconsistent with the expectation that competition contributes to negative co-occurrence patterns, at least over three growing seasons. More pronounced negative co-occurrence patterns were associated with higher species evenness among plots. Functional traits related to increased dispersal (mean seed mass, and proportion of stoloniferous/rhizomatous species) and competitive ability (mean species height, and mean specific leaf area) did not differ significantly across treatments, with the exception of mean height in the third-year post-disturbance; however, the overall trajectory of this trait was inconsistent with theoretical expectations. Overall, co-occurrence patterns changed across the gradient of time-since disturbance, but not as expected; functional trait patterns (trait means, functional diversity measures) were not responsive to our experimental disturbance gradient.

A needle in the haystack? Applying species co-occurrence frameworks with fish assemblage data to identify species associations and sharpen ecological hypotheses

Different species can associate or interact in many ways, and methods exist for inferring associations and underlying mechanisms from incidence data (e.g., co-occurrence frameworks). These methods have received criticism despite their recent resurgence in the literature. However, co-occurrence frameworks for identifying nonrandomly associated species pairs (e.g., aggregated or segregated pairs) have value as heuristic tools for sharpening hypotheses concerning fish ecology. This paper provides a case study examining species co-occurrence across 33 stream fish assemblages in southeastern Oklahoma, USA, which were sampled twice (1974 and 2014). This study sought to determine (a) which species were nonrandomly associated, (b) what processes might have driven these associations and (c) how consistent patterns were across time. Associations among most pairs of species (24 species, 276 unique pairs) were not significantly different from random (>80%). Among all significant, nonrandomly associated species pairs (54 unique pairs), 78% (42 pairs) were aggregated and 22% (12 pairs) segregated. Most of these (28 pairs, 52%) were hypothesized to be driven by nonbiotic mechanisms: habitat filtering (20 pairs, 37%), dispersal limitation (two pairs, 0.4%) or both (six pairs, 11%). The remaining 26 nonrandomly associated pairs (48%) had no detectable signal of spatial or environmental factors involved with the association, therefore the potential for biotic interaction was not refuted. Only five species pairs were consistently associated across both sampling periods: stonerollers Campostoma spp. and orangebelly darter Etheostoma radiosum; red shiner Cyprinella lutrensis and bullhead minnow Pimephales vigilax; bluegill sunfish Lepomis macrochirus and redear sunfish Lepomis microlophus; redfin shiner Lythrurus umbratilis and bluntnose minnow Pimephales notatus; and bigeye shiner Notropis boops and golden shiner Notemigonus crysoleucas. Frameworks for identifying nonrandomly associated species pairs can provide insight into broader mechanisms of species assembly and point to potentially interesting species interactions (out of many possible pairs). However, this approach is best applied as a tool for sharpening hypotheses to be investigated further. Rather than a weakness, the heuristic nature is the strength of such methods, and can help guide biologists toward better questions by employing relatively cheap diversity survey data, which are often already in hand, to reduce complex interaction networks down to their nonstochastic parts which warrant further investigation.