Reef fish size and abundance data from underwater visual census surveys in Fiji These data were collected during one phase of a larger study of reef fish community structure and fishing impacts in Fiji. Habitat cover data were collected at the same time as the fish census data and are provided as a separate file. Data collection is described in detail in the following publication: Jennings, S. and Polunin, N.V.C. (1997) Impacts of predator depletion by fishing on the biomass and diversity of non-target reef fish communities. Coral Reefs, 16, 71-82. These data have been used for analyses reported in the following publications: Jennings, S. and Polunin, N.V.C. (1997) Impacts of predator depletion by fishing on the biomass and diversity of non-target reef fish communities. Coral Reefs, 16, 71-82. Graham, N., Dulvy, N.K., Jennings, S. and Polunin, N.V.C. (2005) Size spectra as indicators of fishing effects on coral reef fish communities. Coral Reefs, 24, 118–124. Description and structure of data files: Data file name: "Fiji_reef_fish_Jennings_Polunin_1997.csv" Data set description: The main data file comprises 25536 length estimates for 144 species of reef fishes recorded by underwater visual census in 10 adjoining areas spanning the western coast of Kadavu Island, Fiji. Data structure: A .csv file comprising 25537 rows by 6 columns, including one row of headers (variable names). Variable name Variable definition Type(units) Range of values area numeric code for study area Integer 1-10 site numeric code for site within study area Integer 1-7 replicate numeric code for replicate within site Integer 1-12 species latin binomial species name Character (144 species names) aphia_id aphia id code for species Integer 208914-279174 length fork length of individual fish Integer(cm) 8-90 Authentication after download: The dataset comprises 25537 rows by 6 columns, including one row of headers (variable names). The sums of the 25536 records in the column "area" should be 137733 and the sum of the 25536 records in column "length" should be 467308. Data file name: "Fiji_habitat_Jennings_Polunin_1997.csv" Data set description: The habitat data file comprises mean habitat descriptors for 7 sites in 10 adjoining areas spanning the western coast coast of Kadavu Island, Fiji. The mean values were calculated from 12 replicate habitat descriptions within each site. Mean depth by replicate was taken as the mean of the minimum and maximum depth within each replicate. Mean depth by site (provided in dataset) was the mean of the replicate means within sites. Mean values by site are provided for habitat descriptors because the original 'habitat description by replicate' data file was not located. Data structure: A .csv file comprising 71 rows by 8 columns, including one row of headers (variable names). Variable name Variable definition Type (units or %) Range of values area numeric code for study area Integer 1-10 site numeric code for site within study area Integer 1-7 coral_m mean cover living massive corals Numeric (%) 1.1-11.9 coral_O mean cover other living hard corals Numeric (%) 13.3-53.5 rock mean cover of rock and dead coral Numeric (%) 0-12.2 rubble_sand mean cover of coral rubble, sand and "other" Numeric (%) 41.0-81.7 rugosity mean habitat complexity, 6 point scale Numeric 2.9-3.7 depth mean depth Numeric (m) 5.7-6.5 Authentication after download: The dataset comprises 71 rows by 8 columns, including one row of headers (variable names). The sums of the 70 records in the column "area" should be 385 and the sum of the 70 records in the column "depth" should be 421.1. Methods of data collection and processing: Data collection was conducted within ten areas (selected to match areas of traditional fishing grounds or qoliqoli) on the western coast of Kadavu Island in Fiji. Quantitative estimates of the abundance and size of reef fishes were made using an underwater visual census (uvc) point count technique based on the method developed by Samoilys and Carlos (1992). A point count technique was used because net movements of target fishes along the reef front (frequently a response to prevailing current) would not bias abundance estimates as they would if transect techniques were used (Watson et al. 1995). Validation of the uvc point count methodology, and the benefits and disadvantages of the technique, were discussed by Samoilys (1992), Samoilys and Carlos (1992) and Jennings and Polunin (1995). All members of the families Chaetodontidae (butterflyfishes), Labridae (wrasses), Lutjanidae (snappers), Mullidae (goatfishes), Scaridae (parrotfishes) and the sub-family Epinephelinae (groupers and coral trout) which could be reliably identified were selected for study. Lethrinidae (emperors) were not included because the census methodology was shown to be not appropriate for these species in Fiji (Jennings and Polunin, 1995). Within each area ("area" in dataset) abundance estimates were made at seven sites. The sites were selected by dividing all areas of reef front (well-developed reef front sloping directly to a depth of at least 10 m, openly exposed to the open ocean, with low siltation rates and estimated mean living coral cover in excess of 25%: as confirmed by direct underwater observation) in each area into 100 m sections (on aerial photographs from the Australian Aerial Mapping 1994 survey conducted on behalf of the Land and Survey Department, Government of Fiji) and randomly selecting seven of these sites ("sites" in dataset). At each site, the abundance and size of fishes estimated to be 8 cm or more fork length was estimated within 12 adjacent replicate census areas of 7 m radius ("replicate" in dataset) by counting each fish and making an estimate of its fork length to the nearest 1 cm. The 12 areas were censused in a random sequence and each census area was centred as close as practically feasible to the 6 m depth contour (on the reef slope). Boundaries of each census area were estimated from as far above or to one side of the census area as the prevailing visibility permitted and counts of the most wary fishes began immediately. Species in each census area were recorded sequentially, the most active species being recorded first. When a count for one species was complete, all further movements of that species in or out of the census area were disregarded. The time required to complete a count was not standardised since this was dependent on the number and diversity of fish in the census area and the complexity of habitat to be searched. In practice, counts took 6 to 13 minutes per replicate. Following the count, the diameter of the census area was measured twice (once parallel with, and once perpendicular to, the reef slope) to determine the accuracy with which the dimensions of the census area had been estimated. The depth at the centre of the census area was also recorded to confirm that the area was centred on the 6 m depth contour. All counts were conducted by the same observer (SJ), during daylight hours, from September 1995 to January 1996. Habitat was described within the perimeter of each replicate census area. When a fish count was complete, the percentage cover (based on plan view) of living massive coral, other living coral, rock and "rubble, sand and 'other'" was estimated by eye, the minimum and maximum depths in the census area were recorded and the topographic complexity of the substrate was described using the six point scale of Polunin and Roberts (1993) where 1 represents low complexity and 6 represents high complexity. Additional details of the fish and habitat census methods are provided in Jennings and Polunin (1997). The locations of survey areas are shown in Figure 1 of Jennings and Polunin (1997). The integer codes for "area" in the dataset correspond with the following areas reported in Jennings and Polunin (1997). Area code Area 1 Natusara 2 Ko Ono 3 Cokovata (Lomanikoro, Nakasaleka and Nakaugasele) 4 Yale 5 Namoce 6 Naqolotini and Naocovonu 7 Suesue 8 Drue 9 Namuana and Boutalevu and Tavuki 10 Yawe Quality assurance: Accuracy of fish length estimation was maintained by practice and testing at intervals throughout the study period (with 57 lengths 2 cm diameter white plastic tube cut to lengths from 8 to 65 cm in 1 cm increments and threaded onto a 1 cm diameter rope in a haphazard sequence). The tests revealed that length estimates were within 11.4% of the actual length and that the mean error associated with estimation was 4.0% for actual lengths of 8-65 cm (n. tests = 71). Estimated lengths were not consistently shorter or longer than actual lengths and no corrections were applied. Estimated positions of points on the census area boundary were consistently within 0.6 m of their correct position as determined following deployment of the measuring line. Records of fish which had been wrongly assigned to a position inside or outside the census boundary were excluded from, or included in, the data set at the time of the count. This was a relatively simple operation as there were few borderline decisions in each count. The reliability of habitat cover estimates was not assessed as thoroughly as the reliability of the fish length estimation. Habitat cover estimation was not practiced as frequently. Comparisons between visual cover estimates and line intercept cover estimates made at other sites in Fiji implied that 90% of cover estimates would fall within ±0.16 of the line intercept estimate. Visual estimates of rugosity by experienced uvc divers have been shown to be closely correlated with other metrics and measurements of habitat complexity (Wilson et al. 2007), but the strength of such relationships was not assessed directly in this Fijian study. After all fish census data were entered (transcribed from the plastic paper used for underwater recording) the compiled data were back-checked, row by row, for consistency with the paper records and any errors corrected. Species length distributions were also checked to ensure that length estimates were transcribed correctly from the platic paper. The taxonomic classification of all censused species was updated for consistency with the current (1 October 2016) taxonomic classification in WoRMS before these data were submitted to Dryad, and an aphia ID has been added to every species' record to simplify future updates of latin binomial names. References in methods: Jennings, S. and Polunin, N.V.C. (1995) Biased underwater visual census biomass estimates for target species in tropical reef fisheries. Journal of Fish Biology, 47, 733-736. Jennings, S. and Polunin, N.V.C. (1997) Impacts of predator depletion by fishing on the biomass and diversity of non-target reef fish communities. Coral Reefs, 16, 71-82. Polunin, N.V.C. and Roberts, C.M. (1993) Greater biomass and value of target coral-reef fishes in two small Caribbean marine reserves. Marine Ecology Progress Series, 100, 167-176. Samoilys, M. (1992) Review of the underwater visual census method developed by the QDPI/ACIAR project: Visual assessment of reef fish stocks. Conference and Workshop Series QC92006, Queensland Department of Primary Industries, Cairns. Samoilys, M. and Carlos, G.M. (1992) Development of an underwater visual census method for assessing shallow water reef ̃sh stocks in the south-west Pacific. Queensland Department of Primary Industries, Cairns. Watson, R.A., Carlos, G.M, and Samoilys, M.A. (1995) Bias introduced by the non-random movement of fish in visual transect surveys. Ecological Modelling, 77, 205-214. Wilson, S.K., Graham, N.A.J. and Polunin, N.V.C. (2007). Appraisal of visual assessments of habitat complexity and benthic composition on coral reefs. Marine Biology, 151, 1069-1076. Acknowledgements: I am grateful to the Fijian Government for support and permissions when conducting this work, to Keresi Fuoro of the Native Lands Trust Board for her assistance with identifying qoliqoli boundaries and Esaroma Ledua of Fiji Fisheries Division and Viliame Katonivere of Fijian Affairs for introductory letters in support of the project. The villagers of Dravuni, Nakasaleka, Rakiraki,Daku, Naivakarauniniu, Naikorokoro, Drue, Tavuki and Nalotu gave much assistance by providing accommodation and logistical support and Viliame Niavalarua, Salamisa Baselala, Panapasa Pone and Mareka Lavelawa provided boat transportation and diving support.I thank Robin South, Veikila Vuki, Milika Naqasima-Sobey, Nani Bulai and other staff at the Marine Studies Programme of the University of the South Pacifc for supporting this research project and for their logistical and technical help. The research which generated these data was supported by the Natural Environment Research Council (UK), University of the South Pacific (Fiji) and the Overseas Development Administration (now Department of International Development) (UK). Contact: Simon Jennings, School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich, NR4 7TJ, United Kingdom. Email: simon.jennings@uea.ac.uk