Dataset: inverts - pooled

To assess changes in abundance of invertebrates and evaluate their relationship to the cover of corals and macroalgae, data were contrasted among times in biennial increments. Each of the six sites consisted of a single transect parallel to the 7-9 m depth contour, along which photoquadrats (0.25 m^2) have been sampled since 1992. A 20-m transect was used prior to 2001 when ~17 photoquadrats were recorded on 35-mm film, but thereafter, a 40-m transect has been used along which ~40 photoquadrats were recorded using digital photography. The photoquadrats were positioned randomly along transects every year, and initially were recorded with a Nikonos V camera fitted with a 28-mm lens, strobes, and slide film (Kodachrome 64). Since 2001, digital photography with either 3.3 megapixel (Nikon Cool Pix 990) or 6.1 megapixel (Nikon D70) resolution has been employed. All cameras were mounted on a framer to hold them perpendicular to the reef.

To census the population density of invertebrates, photoquadrats were screened to identify tractable taxa based on size (>=3-cm diameter) and appearance. These organisms were identified to class, genus, and species wherever possible, although species resolution sometimes proved unreliable. The initial screening revealed a consensus group of 30 taxa (Table S1). As these taxa varied in abundance, and many were rarely seen in photoquadrats (<4 individuals/year), a subset of the initial group was created by selecting taxa that were relatively abundant (e.g., found in an average of 12% of the quadrats censused each year). This subset consisted of 14 species from 14 genera and four classes (See supplemental file: table S1), and was used to evaluate temporal trends in populations at a lower taxonomic level.

Organisms were counted on images enlarged by projection, or on computermonitors. Unitary organisms were easily counted, but colonial taxa were defined as individuals based on areas of autonomous tissue. Therefore, colonial taxa with several portions of unconnected tissue were scored as multiple individuals, even though they might share a common genotype through fission. Organismswere only scored if >50% of their area occurred within the photoquadrat, and when they occurred in adjacent photoquadrats, were scored only once.

In August 2007, benthic invertebrates were counted in situ at all sites to compare abundances with those generated from photoquadrats in the same year. In situ countswere completed during the day in quadrats (1x1 m) placed in a similar manner to the photoquadrats. Although these surveys revealed greater species richness than was visible in the photoquadrats, the enumeration was restricted to those taxa that previously had been identified as tractable. To compare the two techniques, mean abundances of the invertebrates were calculated by class, standardized to a common area, and contrasted by t-tests using quadrats as statistical replicates. Where significant differences were detected between invertebrate abundances originating from in situ and photographic techniques, the mean abundance by the photoquadrat technique were expressed as a percentage of the mean value from the field counts to evaluate the bias in using photoquadrats to quantify invertebrates (See supplemental file: figure S1).