Using species abundance and phylogeny conjointly to approach vegetation classification. A case-study on Macaronesia’s woody vegetation

Prepared by Jorge Capelo

Summer aspect of Euphorbia berthelotii community, without leaves. La Gomera, the Canaries. Photo credit: Jorge Capelo.

Classification aims to identify distinct and coherent groupings within vegetation. Modern approaches use a statistical analysis of data obtained from plot surveys in the field, most importantly species identity and abundance. Although a variety of classifications methods have been employed, almost all share a common analytical data flow: a calculation of similarities among plots based on shared species and a subsequent plot classification based on such similarity. Two underlying assumptions justify using species co-occurrence in vegetation classification: first, species co-occur in space because they were either filtered by abiotic conditions or, second, because they were conditioned by biological interactions, such as competition or facilitation. While vegetation scientists are well aware of the importance of these two factors, they have overlooked a third important aspect in classification: plants can share a habitat because of a common evolutionary history, i.e. similarity within and across vegetation groups can be also the result of shared ancestry.  In this paper, I asked how vegetation classification accounting for phylogeny compares to a traditional classification approach, in which only ecological similarity is used.

My case-study was the woody vegetation of Macaronesia, i.e. the set of the Azores, Madeira, Canaries and Cape Verde archipelagos. From a biogeographical standpoint, Macaronesia has been considered a coherent unit, as islands share a similar or analogous flora and vegetation types and are quite distinct in this respect from the African and European mainland. The floristic similarities reflect the common biogeographical history of intricate plant dispersion pathways among islands in short evolutionary time frames. Reproductive isolation of small samples of genetical variation – many times a single seed – of newcomer plants lead to rapid differentiation of new species in islands. This process has given rise to species that are distinct across island or archipelago but are still quite similar morphologically and with respect to their environmental conditions (they ‘conserve their ecological niche’). Such taxa are known to science as vicariant species. Using a similar analogy, we can also define vicariant plant communities (= communities made up of vicariant species). Macaronesia has many vicariant vegetation types and hence provides a suitable study subject.

Autumn aspect of Euphorbia tuckeyana community, with leaves. Fogo Island, Cape Verde. Photo credit: Eduardo Dias.

When I used a conventional co-occurrence-only approach, Macaronesian vegetation was split into several units that lacked biogeographic coherence. This is because in the dataset, species are taxonomically distinct across each archipelago. Further, as I calculate similarity using also phylogeny, names in the plots are not just neutral tags of presence/absence (or quantity): the name also carries an implicit biological similarity issuing from the species’ position in the Tree of Life. The approach was able to detect biogeographically vicariant vegetation types, made up by taxonomically distinct but phylogenetically similar species in each archipelago.

Phylogeny-informed vegetation classification is of paramount importance when we classify plots with many vicariant species. Greater a priori phylogenetic similarity weights more in the final similarity when vicariants are frequent. By contrast, it has less weight when species have not undergone adaptive radiation (i.e. a rapid intense geographic speciation).

My comparison of a phylogeny-informed vs. a conventional approach suggests that Macaronesia may arise as coherent biogeographical unit, unified by the sharing of vicariant vegetation types that are, nonetheless, made up of clear-cut not-shared distinct species.

This is a plain language summary for the paper of Jorge Capelo published in the Journal of Vegetation Science (