A study of the benthos of soft bottoms, Sek Harbour, New Guinea, using numerical analysis

Abstract

Samples collected from 62 sites within c. 1½ sq miles by means of a van Veen grab contained a small number of species (animal and plant), dominated by two species of ophiuroids. Without sophisticated analyses it is possible to delimit obvious "communities" based on dominants and subdominants, for example an Amphioplus- Amphiacantha community and a soft coral-sponge community. Similarly areas of high species diversities can be identified, these conforming approximately with the recordings of species occurring in single sites.

Further analyses were made numerically after transformation of the data using the Bray and Curtis (1957) measure of affinity and the Lance and Williams (1967b) system of flexible sorting. Nine site groups were derived and ordinated by the "principal co-ordinate analysis" of Gower (1966, 1967). These groups were reduced to eight by detailed consideration of site characteristics.

A gritty area is characterized by epifauna (soft coral and sponge), a "clean" area by a tubicolous polychaete Mesochaetopterus, and six areas by the two ophiuroids Amphioplus and Amphiacantha. Amongst these the following can be distinguished from a western to eastern (seawards) direction: (a) low numbers and patchy distribution of both ophiuroids and virtually nothing else, (b) both ophiuroids with a bivalve mollusc Tellina and a stomatopod Clorida, (c) moderate numbers of both ophiuroids, (d) large numbers of Amphioplus with moderate numbers of Amphiacantha and three associated species, and (e) two areas which are essentially ecotones of the last.

The "associations" are compared in detail with others involving comparable genera. While it was expected that in these warm waters there would be a heterogeneous biota with indefinite "community patterns", as just indicated the situation was relatively simple.

Data were of such complexity to be ideally suited for a comparative study of methods of numerical analysis, and these are outlined in the Appendix. Three standard models were used, two of which were discarded. After prior transformation of the data (including elimination of species with a single occurrence and fusion of sites in proximity) the final technique proved satisfactory.