Problems with the application of cladistics to forest fragmentation studies

Abstract

Recently, Pellens et al. (2005) applied standard parsimony methods to the analysis of species communities in forest fragments. Their proposal consists of assembling a data matrix of species presence/absence for each fragment and submitting it to a parsimony tree search using an adjacent continuous forest as outgroup. As such, fragments become analogous to terminal taxa and species to characters, with character states being the recorded presence/absence of such species in each fragment. The justification provided by Pellens et al. for such use of parsimony is that “the fragmentation effect is a matter of history: fragments are the remnants of previously continuous large forests…” so that “fragments can be merely characterized as descendants rather than remnants of an ancestrally continuous forest, since they have evolved after their isolation” (p. 9). Pellens et al. indicated that this evolution is due to “the spatial breakup of communities that become separate entities” (pp. 9–10, our italics), and assumed a strong direct link between community change and evolutionary change by stating that “…establishing relationships among communities and looking at their nestedness is not only a practical classificatory procedure but is also aimed at interpreting their evolution by descent with modification” (p. 10). Thus, parsimony would discover the pattern of nestedness among fragmented areas as well as the evolution of characters (i.e., presence/absence of species) “polarizing the changes in either extinctions or colonizations” (p. 10); i.e., species gains (0→1) and species losses (1→0). The aim of this paper is to report problems with this application using both case studies provided by Pellens et al. The first example deals with the distribution of frog species sampled by Tocher et al. (1997) at the experimentally fragmented landscape of the Biological Dynamics of Forest Fragments Project (BDFFP) site in Central Amazonia. This example is thus specially important because in the BDFFP site the fragmentation history is accurately documented. The second example deals with birds sampled in fragments of the Atlantic forest, Eastern Brazil, by Anciães and Marini (2000).