Study site and field data collection
Permanent 1 m2 photoquadrats were established on Heron Reef in 1962/63, using 9 mm diameter mild steel (rebar) pegs, which were replaced over time. From the 1990’s, replacement pegs were stainless steel for greater longevity. Four sites were established, the protected (south) crest, inner flat, exposed (north) crest and exposed pools. Co-ordinates for each site are presented in Table 1, the layout shown in Fig. 2, and sites have been well described previously5,6. At each census, a 1 m2 frame divided into a 5 × 5 grid using string was placed over the pegs, and the quadrat photographed from directly above at low tide. From 1963 until 2003, a 35 mm camera and colour slide film were used. The camera was attached to a tripod affixed to the 1 m2 frame, and captured around 2/3 of the quadrat. The frame (and camera) were then rotated 180 degrees to capture the remainder of the quadrat. After 2003, a hand-held digital camera was used, with the entire quadrat being captured in a single image. Concurrent with each census, mud maps of each quadrat were hand drawn in the field, and all colonies identified in situ by someone with expertise in coral taxonomy.
Quadrat layouts for each of the four sites respectively, noting that the north crest and north ridge have been treated as a single north crest site in previous publications. Underlining indicates original 1962/63 quadrats. Other quadrats were added in or after 2008, as indicated in the text. Contiguous quadrats are pictured bordering each other. Spacing between separate quadrats or groups of quadrats is not shown to scale. Note that up until 2005, NRNW was known as NR. The acronyms in each quadrat represent its name.
At the protected (south) crest, a set of six contiguous quadrats were established in 1963 in a 2 × 3 arrangement parallel to the waterline, and about 420 m southeast of the island. This site is exposed at low tide, and was photographed once all water had drained off it. Images of quadrats A, C & E (the shoreward row) from 1963 to 2012 have been fully processed, and the data have been through QA/QC. Data after 2012 exist as images only. These quadrats form the basis of previous analyses1,4,5,6 for this site. Photographs are available for quadrats B, D & F, but apart from 2003–2010, have not been processed. In 2010, an additional two quadrats were established either side of the original six, leading to a 2 × 5 arrangement. Again, only imagery is available for these additional quadrats.
At the inner flat, two pairs of contiguous quadrats were established in 1962, 44 m apart, about 70 m south of the island. This site is covered by ~10 cm of water at low tide, so could only be photographed on a still day. Imagery for this site is only available to 2012, after which the marker stakes appear to have been removed in a cleanup of the area. Images for one quadrat in each pair have been processed, but have not been subject to full QA/QC.
At the exposed (north) crest main site, a set of four contiguous quadrats was established about 1100 m northeast of the island in 1963. An additional single quadrat (north ridge) was established 326 m to the east. Images from 1963 to 2012 have been fully processed, and the data have been through QA/QC. Data after 2012 exist as images only. In 2005, the single north ridge quadrat was expanded to 4 m2, and in 2008, both subsites were expanded to six quadrats in a 2 × 3 arrangement. These additional quadrats have been digitised up to 2012, but have not been through full QA/QC.
The exposed pools are two individual quadrats about 5 m apart about 30 m north of the eastern (north ridge) exposed crest site. These are on the edge of a natural pool, and range from ~5–50 cm deep at low tide, and so could only be photographed on a calm day. Imagery for this site is only available until 2005, after which the marker stakes could not be relocated. Images from 1963 to 1998 have been processed, but have not been through full QA/QC.
Retrieval of coral composition data from the photoquadrats
Processing of the images involved scanning the colour slides to produce digital images, and then orthorectifying each image to a 1 m2 basemap in ArcGIS (ESRI Ltd). The corners of the frame, and the holes for the string grid, were used as control points for the orthorectification. For images that originated as colour slides, each half of the quadrat was individually orthorectified to the same basemap, producing a single image of the entire quadrat (see Fig. 3). While contiguous quadrats were orthorectified individually, they were done so against a basemap containing all quadrats in the group, meaning that the resulting images can be easily merged to create a single image of the group. The outlines of all visible coral colonies (>~1 cm2), and other benthic organisms such as algae and clams, were then digitised in ArcGIS to create a single shapefile for each quadrat for each year. Each colony was represented as an individual feature within the shapefile, and was assigned a unique colony number and species based on the mud maps drawn in the field. Colony numbers were consistent across years, allowing individual colonies to be tracked over time. If a colony underwent fission, the original colony number was retained for each, with the addition of a unique identifier after a decimal point. For example, if colony 35 split in two, the resultant colonies were identified as 35.1 and 35.2. If 35.2 later split again, the resultant colonies were identified as 35.2.1 and 35.2.2. If the colony overlapped the edge of the quadrat, only the area within the quadrat was digitised, and a flag was applied to indicate that only part of the colony was included (edgestatus = 1 in the data). Upon completion of digitisation, ArcGIS was used to calculate the area and perimeter of all colonies. While multiple census were conducted in 1963, 1971 and 1983, only a single census in each year has been processed. There are currently no plans to undertake further digitisation or QA/QC of this data set.
Example orthorectified and stitched (prior to 2001) images from the NCNE quadrat, showing the effects of a cyclone that removed all colonies in 1972, and slow recovery over subsequent decades.
Source: Ecology - nature.com
