Dispersal traits may not tell us what we hope for: the case study from understory of oak-dominated forest

The post provided by David Zelený & Markéta Chudomelová

This post refers to the article Tracing the signs of local dispersal in the temperate forest understorey using spatially structured vegetation data by Markéta Chudomelová and David Zelený, published in the Journal of Vegetation Science (https://doi.org/10.1111/jvs.12835).

At the beginning was an idea to show whether the dispersal in the plant community can be linked to dispersal traits of individual species. To study plant dispersal is hard; it’s something that is surely happening, but unless we do an experimental study or very careful and time demanding observations, we don’t get much data about that. But there are some ways around. For example, we may study the current spatial distribution of plants. Where the plant grows now is a result of at least two processes: plant needs first to get there (dispersal), and the environmental conditions of the spot need to be suitable enough for plant to thrive there and hopefully also reproduce (environmental filtering). There is a statistical method, based on variance partitioning of species composition data, which is aiming to quantify these two processes, in the case that we have detail spatial information about individual species and also detail measurements of important environmental factors. If there is a spatial pattern in the community, which cannot be explained by environmental factors, then there is a good chance that this is the footprint of dispersal. Sure, the thing is much more complicated; if one does not measure all relevant environmental factors, than this unexplained variance could well be just a spatial signature of those unmeasured environmental factors. Also, additionally to dispersal, there are other processes which are not dependent on the environment but can result in spatially structured communities (like ecological drift, https://vimeo.com/245349395). But if we acknowledge all of these shortcomings, we may assume that this unexplained variation is a good proxy of dispersal limitations of species in the community. Then, we may ask: is this observed dispersal limitation related to measured dispersal traits of individual species? Such dispersal traits include seed size (smaller seeds disperse further), seed number (with more seeds there is a better chance to get further), presence of some seed appendices (which seeds can use to fly further), and also clonal growth (some plants don’t disperse by seeds, but simply send long shoots to the neighbourhood to settle at new places). There is a good reason to expect that dispersal traits will at least partly explain where the species truly establish, even if assumptions behind have also several limitations.

To understand whether dispersal related traits can help with explaining the spatial distribution of species, we sampled herb species in the understory of oak-dominated central European deciduous forest (some photos from the field sampling are included in this post). We set up a complex grid of nested subplots 2 × 2 m and 5 × 5 m within a square area of 60 × 60 m, recorded species occurring in them and also measured environmental variables which we assume to be important in this locality based on the previous study from the same area (Chudomelová, Zelený & Li 2017). Then, we focused on individual 2 × 2-m plots, for which we have information about species occurring in four neighbouring 5 × 5-m plots, and asked: Can we explain the occurrence of species in these focal plots by environmental conditions, or are they better explained by their occurrence in the neighbourhood? For this, we used analogy of variance partitioning method described above, but with the following differences: we did the analysis separately for individual species instead of the whole community, and also we did not use the explicit spatial distribution pattern, but only occurrence of the analyzed species in the neighbourhood of the focal plot. Here also come dispersal traits: we assume that for those poorly dispersing species, the environment won’t be that important, but the occurrence in the neighbourhood will be a good reason to occur also in the small target plot. Those species with easy dispersal, on the other side, will not rely on the neighbourhood that much. Previously and in slightly different context, we played with this idea also using simulated spatially structured community data, and it seemed to work well. When using our real data, however, we found very weak evidence that dispersal traits are linked to the spatial pattern of the species in the understory.

There are indeed many reasons explaining why we have not found the link between dispersal traits and species spatial distribution, and we acknowledge them in the paper.  The study design is not perfect; we have sampled a rather small area, and maybe if it was bigger, the pattern might appear. Herbs in the forest understory may act differently than plants in another context, like grasslands, and we have quite few species; maybe if we studied grasslands or had more diverse vegetation, the pattern would be more obvious. Also, the scale of the study may not fit the scale of dispersal processes. The dispersal traits have been compiled from available trait databases, not measured in situ; maybe that directly measured traits would have better explanatory power. On the other side, these are not limitations uncommon to most observational studies.

Considering that the manuscript is more like a negative study report, it was not that difficult to get it published. It took three rounds of revisions (reject-resubmit, a major revision and minor revision), with helpful and thorough suggestions from editor and reviewers. The Journal of Vegetation Science was the first journal we tried to submit it, and we were successful, which is an optimistic sign. Indeed, there are many studies with no results simply because data are poor, analytical decisions are not correct, or the main theoretical assumptions behind are flawed. We think this is not the case of our study; although the data set is small, it’s very detailed and complete. The analysis is quite simple, which makes it rather transparent, not a “black box” solution of complex algorithms. And the theory behind is sound and often used by others. True, the fact that many people are doing something or talking about something does not mean that it cannot be wrong – it often is. But we think that this is more a strength than a weakness of our study: there are good reasons to expect that it will work, but it doesn’t. We think that more negative studies showing such “it should have worked, but it doesn’t” situations are needed to show ecological patterns in a more realistic light.

Reference

• Chudomelová, M., Zelený, D., & Li, C.-F. (2017). Contrasting patterns of fine-scale herb layer species composition in temperate forests. Acta Oecologica, 80, 24-31. (https://doi.org/10.1016/j.actao.2017.02.003)