EFFECTS OF NUMERICAL AND TAXONOMIC RESOLUTION ON COMPOSITIONAL VARIATION REVEAL THE NATURE OF ENVIRONMENTAL GRADIENTS AND LINEAGE-LEVEL ECOLOGICAL CONSERVATISM

Differences in the numerical (abundance to presence-absence) and taxonomic resolution can affect the nature of the relation between taxa and environmental gradients, and thus our inference about the ecological and evolutionary assembly of metacommunities. The relationship between taxon distribution and environmental gradients is captured by the ratio of variation in community composition explained and unexplained by gradients. These two components of variation can be independent and are related to compositional variation occurring (1) within habitats (generated by patchiness) and (2) between habitats (generated by niche specialization). We hypothesize (1) that an increase in compositional segregation along an environmental gradient with decreasing numerical resolution can reflect a relatively high habitat specialization, and (2) that an increase in compositional segregation of habitats at higher taxonomic levels can reflect high within-lineage niche heritability and conservatism. We evaluate these hypotheses with molluscan community data from the Miocene of the Central Paratethys at three taxonomic levels (species, genera and families), using gradients in bathymetry and salinity. We find that segregation of communities from different habitats is maximized with abundance data along bathymetric gradients and with presence-absence data along salinity gradients. This occurs because (1) species along bathymetric gradients are less specialized than along salinity gradients, and (2) within-habitat variation in composition tends to decline with decreasing numerical resolution. The between-habitat segregation is maximized at genus and family levels along bathymetric gradients whereas it is maximized at species level along salinity gradient. This difference is generated by a higher degree of within-genus and within-family conservatism in habitat affinity along bathymetric gradients than along salinity gradients. We think that it is misleading to use species-level data as a single reference for assessments of numerical and taxonomic sufficiency in capturing environmental gradients. Species-level data do not capture phylogenetic relatedness and thus are silent about similarities in niche attributes.