Get access

Repeated sampling reveals differential variability in measures of species richness and community composition in planktonic protists

Authors

  • John R. Dolan,

    Corresponding author
    1. Microbial Ecology and Biogeochemistry, Laboratoire d’Océanographie de Villefranche-sur-Mer, Université Paris 6 CNRS UMR 7093, Observatoire Océanologique de Villefranche-sur-Mer, Station Zoologique, B.P. 28,F- 06230 Villefranche-sur-Mer, France
    Search for more papers by this author
  • Thorsten Stoeck

    1. University of Kaiserslautern, Ecology Department, Faculty of Biology, Erwin-Schrödinger Str. 14, D-67663 Kaiserslautern, Germany
    Search for more papers by this author

E-mail dolan@obs-vlfr.fr; Tel. (+33) 4 93 76 38 22; Fax (+33) 4 93 76 38 34.

Summary

Diversity metrics and descriptors of protistan community structure were calculated from 12 samples of 10 l each collected from the Bay of Villefranche in the NW Mediterranean Sea. Variability of the sampling was on scales of minutes and meters. The individual samples were compared with each other and compared with a pooled data set from the total volume of 120 l, considered as the ‘true’ community. We focused on a single group of planktonic protists, tintinnids, a coherent functional and phylogenetic group in which morpho-species identifications by light microscopy are unambiguous. Tintinnid abundance in the samples ranged from 217 to 321 cells of 16–21 species with the number of rare species in a sample (< 1% of abundance) positively related to species richness of the sample. Rarefaction estimates of total species richness in the 12 samples ranged from 21 ± 3.5 to 37 ± 3.6 compared with the 34 species of the pooled data set. The measures of similarity reflected the differences between samples in both the numbers and identities of the least abundant or rare species. The species abundance distribution using pooled data was best fit by a log-series or geometric distribution; eight species accounted for about 90% of total cells and most species, the remaining 22 out of 34, were ‘rare’ (concentration < 1% of total cells). Among the samples, 5 were best fit by a geometric model, 1 by a log-series distribution, 2 by a log-normal or log-series model, and 4 could not be clearly assigned a particular distribution. Our results suggest that single sample estimates of species richness are relatively robust compared with measures of taxonomic similarity and species abundance distribution. When measuring differences among populations sample variability should be considered.

Ancillary