To conserve biodiversity, we need a clear picture of how it is distributed. Easier said than done! Biodiversity is a short word for a wide concept. Even just counting the number of species in a forest can be an overwhelming task if one considers all insects, spiders, birds, mites, (ticks!), lichens, fungi, isopoda and so on. Not only, some of these groups are very difficult to identify, so sampling them all would require hundreds of hours to a team of well-trained field biologists.
Can we rely on one or few groups of species to make inferences on all the other, then? Although the use of indicators is a well-established routing, how well these indicators work in Southern European beech forest is not clearly understood. One of the open questions is: How well do indicators work at different scales? In our new article “Congruence across taxa and spatial scales: Are we asking too much of species data?”, just published in Global Ecology and Biogeography, we tried to find an answer.
More specifically, we tried to understand whether the biodiversity of one group of species correlates equally well (or bad!) to the biodiversity of other groups, when one changes either the resolution (the size of elementary sampling unit) or the extent (the geographical area included in the survey) of the study. In other words, how does cross-taxon congruence vary with spatial scale?
To answer this question, we first assembled a large dataset, encompassing biodiversity data for six groups of species (=taxa, singular: taxon): vascular plants, bryophytes, birds, epiphytic lichens, deadwood dependent beetles and wood-inhabiting fungi, collected across 354 plots in 23 forest sites situated in France, Italy and Hungary. Being the data collected for different research projects by different research groups, even just standardizing it for our study was a daunting task, but this is another story.
Changing the extent
To understand the effect of scale extent, we compared cross-taxon correlations when considering plots inside each forest (Plot grain, site extent), and when considering plots across all forests in our dataset (Plot grain, continental extent). If cross-taxon congruence was not affected by the change in extent, then one would expect the correlation coefficients between pairs of species groups (e.g. lichens vs plants, or beetles vs birds) to remain constant.
Figure 1 – Cross-taxon congruence (Pearson’s correlation coefficients) among species richness of different taxa at plot grain and continental (PG-CE) and site (PG-SE) extents in temperate European forests. (a) Effect sizes and 95% confidence intervals. (b) Comparison of the effect sizes in a coordinate system. Plots close to the grey line indicate pairs for which the congruence relationship does not change with extent. Blue and red dotted lines indicate no correlation between scales.
This was not really the case, though, at least for species richness. Not only the cross-taxon correlation were on average pretty low (Fig. 1 – left), but when changing the scale some of these became stronger, other weaker, while some switched from being positive to being negative. If you look at the right graph in Fig. 1, you will see that we are very far away from an ideal (for monitoring purposes) situation. If pairs of species groups were highly correlated, so that one was a good indicator of the other, and cross-taxon congruence was stable across scales, then most of the points (here they represent the correlation coefficient of pairs of species at the two scales) should be located at the upper-right end of the diagonal curve. As it is often the case, reality is much more complicated. And it is probably funnier so.
But what happens when instead of using the number of species (=species richness), we consider the correlation between the composition of two sampling units, i.e. we consider the actual identities of the species? We cannot forget that the species in an ecosystem are tightly linked to each other into a complex ecological network (remember the concept of food chain from your third grade). A species of insect, for instance, may feed exclusively on a given species of plant, so that we may not find the beetle, unless also the plant is there. If we consider only the number of species, we may missed such intrinsic associations. Indeed, we found that correlation between pairs of taxa were always positive when considering composition, and they increased in strength when increasing the extent of the analysis (Fig.2, up-right).
Changing the grain
Results were slightly different when focused on changing the grain of our analysis. Indeed, moving from a finer to a coarser sampling unit (i.e. when considering the whole forest as a unit, instead of each individual plot inside the forest), correlations became on average much stronger, and in this case the results did not change much when considering species richness or composition (Fig.2, bottom).
Why does cross-taxon congruence change with scale?
In our opinion, the effect of spatial scale depends on the fact that ecosystems are shaped by different processes at different scales. If we zoom in, and only consider few square meter in a forest, than we may imagine that a certain spot is better suited to a given species because it is sunnier, or maybe because it contains more dead wood than another spot. But if we are comparing two forests, then the species we will find do not depend exclusively on these fine-scale environmental patterns, but also on differences between the average climatic or soil conditions of the two forest, as well as on their different disturbance histories (maybe one was cut 40 years ago, and the other is more or less untouched).
Similarly, when shifting from local to regional scale, biogeographical processes will slowly replace ecological ones. Two forests sharing the same environmental conditions today, may well host different species because they had different histories. For instance, some species may have gone extinct in the region of the first forest during the glaciation period, or because of intense historical land-use, while the same species still occur in the second forest.
Take-home message and implications
In short, we found that cross-taxon congruence is strongly scale dependent. Generally, the spatial extent exerts a greater influence on congruence between taxa than spatial grain, and species richness is more sensitive to shifts in spatial scale than species composition. Indeed, we used an array of spatial scales and biodiversity metrics and showed that these bring complementary, and sometimes contrasting, information on multi-taxon congruence in European temperate forests. What does it mean for conservation and biodiversity monitoring?
When collecting biodiversity data, e.g. for monitoring the effects of a given management, we recommend sampling different species groups, since their diversity patterns are often not congruent. If this is not possible, then one should carefully choose the most promising indicators, and be well aware that he\she may not be getting the full picture. Our results provide some guidance on the choice (e.g. plants showed, on average, higher correlation coefficients than other groups).
Although local patterns of congruence can rarely be extrapolated to other regions, when considering fine-scale (plot grain, site extent) patterns of species composition we found consistent cross-taxon associations in many sites, especially between plants and other taxa. This is not surprising, because plants are the major structural and functional component of forest ecosystems and therefore influence forest structure and nutrient cycling. Sampling plant species composition in scattered plots across different sites may therefore be effective (or at least, provide more information compared to other groups) at summarizing the whole community composition.
Burrascano, S., R. B. De Andrade, Y. Paillet, P. Ódor, G. Antonini, C. Bouget, T. Campagnaro, F. Gosselin, P. Janssen, A. Persiani, J. Nascimbene, F. M. Sabatini, T. Sitzia, and C. Blasi. 2018. Congruence across taxa and spatial scales: Are we asking too much of species data? Global Ecology and Biogeography. https://doi.org/10.1111/geb.12766