Pairs – a FORTRAN program for studying pair-ise species associations in ecological matrices
Community assembly is often studied in terms of nestedness (Patterson and Atmar 1986, Wright et al. 1998, Ulrich and Gotelli 2007a) and non-random patterns of species co-occurrences (Diamond 1975, Gotelli 2000, 2001 Ulrich and Gotelli 2007b) and the respective standard software is widely used: The Nestedness Temperature Calculator (Atmar and Patterson 1995), EcoSim (Gotelli and Entsminger 2002, and Nestedness (Ulrich 2006).The present program Pairs extends these approaches and implements beside standard nestedness and co-occurrence metrics a new metric to study pairwise species associations: The software implements:
- the discrepancy metric of Brualdi and Sanderson (1999)
- the species combinations score (Pielou and Pielou 1968)
- the C-score (Stone and Roberts 1990)
- the checkerboard score (Gotelli 2000)
- the Soerensen metric
- the togetherness score (Stone and Roberts 1992)
- the species absences score (Stone and Roberts 1992)
- the variance test (Schluter 1984)
- a pairwise correlation test
Pairs not only studies matrix wide patterns. It uses a Bayesian approach to detect non-random associations of pairs of species. The number of species pairs of a matrix is S(S-1/2. Hence even for medium sized matrices many ‘significantly non-random’ species pairs are expected at the 1% or 5% error level. Pairs chooses only those scores for further analysis where the number of observed instances is either larger than the mean expectation (Bayes M criterion) or larger than the upper 5% or 1% confidence limit (Bayes CL criterion).
The manual is a pdf-file with about 150 kB, Pairs is a single program file of about 460 kB.