Quantifying relative diver effects in underwater visual censuses

Diver-based Underwater Visual Censuses (UVCs), particularly transect-based surveys, are key tools in the study of coral reef fish ecology. These techniques, however, have inherent problems that make it difficult to collect accurate numerical data. One of these problems is the diver effect (defined as the reaction of fish to a diver). Although widely recognised, its effects have yet to be quantified and the extent of taxonomic variation remains to be determined. We therefore examined relative diver effects on a reef fish assemblage on the Great Barrier Reef. Using common UVC methods, the recorded abundance of seven reef fish groups were significantly affected by the ongoing presence of SCUBA divers. Overall, the diver effect resulted in a 52% decrease in the mean number of individuals recorded, with declines of up to 70% in individual families. Although the diver effect appears to be a significant problem, UVCs remain a useful approach for quantifying spatial and temporal variation in relative fish abundances, especially if using methods that minimise the exposure of fishes to divers. Fixed distance transects using tapes or lines deployed by a second diver (or GPS-calibrated timed swims) would appear to maximise fish counts and minimise diver effects.     

Determining Methods of Underwater Visual Census for Estimating the Abundance of Coral Reef Fishes

We aimed to determine optimal methods of underwater visual census (UVC) for estimating the abundance of coral reef fishes exploited by fisheries in the tropical Pacific. Two main methods were tested using SCUBA: strip transect and stationary point counts. We assessed their relative accuracy, precision, power and efficiency (cost), and compared different census area dimensions, observer swimming speeds, and number of replicates. Twenty-five comparisons of the two UVC methods were conducted for 73 species from five families of coral reef fishes, on reefs in Australia and Fiji. Species were grouped within families based on their mobility. Few significant differences were found, either among or between strip transects and stationary point counts. The data were characterised by high variability, low precision and low power. A trend for greater accuracy in density estimates of small sedentary species with smaller census area was apparent, which probably reflects searching efficiency. Only one species group, the sedentary Acanthuridae showed differences between transects and point counts. Higher, hence presumably more accurate, density estimates were obtained with 50m×5m transects. Notably, point counts could be deployed in 70% of the time of transects. A bootstrapping procedure demonstrated a consistent improvement in precision of density estimates with increasing number of replicates, but no appreciable change in precision was found beyond 10 to 15 replicates, in all species groups and for both transects and point counts. Consequently, and because of the high variability inherent in fish density estimates, we recommend that at least 10 replicates be used to quantify the species considered here. The power calculations showed that only large differences in density will be detected with the replication levels typical of UVC surveys. Power was greatest for the roving serranids, with a decrease in density of about 50% detectable for this species group. We discuss the importance of carefully defining the behavioural attributes of species prior to selecting a UVC method. A method of temporal stratification in a count is described for censusing a range of species of varying mobilities. We suggest that fish mobility and search efficiency are key factors in optimising UVC methods.     

Visual census methods underestimate density and diversity of cryptic reef fishes

The diversity and density of small, benthic reef fishes were estimated using visual census and enclosed rotenone stations. Visual census underestimated the number of species present and the density of common species by up to 91%.     

Improving multispecies rocky reef fish censuses by counting different groups of species using different procedures

Synopsis A number of factors can influence the accuracy and precision of underwater visual transect techniques. Among these are observer swimming speed and, during multispecies surveys, the effect of counting all fishes on estimates of particular species. This paper examines the effect of these factors on population estimates of inconspicuous fishes (defined as Type 1) in a temperate reef fish assemblage near Sydney, Australia. Counting Type 1 fishes with all others yielded significantly lower estimates of species richness and abundance than when counted alone. This suggests that multispecies surveys should be split into 2 or more counts, using a census procedure that is appropriate to the group of species cencused. Further, the effect of counting all other fishes on estimates of Type 1 fishes varied according to the relative abundance of the former: their effect was lowest when abundance of other fishes was lowest. There was a negative relationship between observer speed and estimated abundance for Type 1 fishes. Survey precision of Type 1 fishes was generally improved by surveying at slower observer speeds.     

Counts of coral reef fishes by an experienced observer are not biased by the number of target species

Fishes are commercially, recreationally and functionally important inhabitants of coral reefs. Accordingly, accurate assessments of fish abundance and diversity are necessary for effective reef management. While some reef fish monitoring programmes target all fishes, many survey a subset of common and visually obvious species. Changing to counts of all species, while desirable, risks new and unwelcome biases from altered observer swimming speeds and search patterns. Here we test whether substantially increasing the number of target species in an established fish monitoring programme biases counts of the original subset, so precluding ongoing comparisons with historical data. A subset of 141 fish species have been visually surveyed along 50 × 5 m transects over 27 years throughout Australia's Great Barrier Reef. We experimentally compared counts of the subset from standard subset-only surveys and from surveys of all species (excluding small site-attached fishes that are surveyed separately) at three diverse reefs. Subset species richness and abundance, in total and of major families, and assemblage structure did not differ due to survey method. The high-level experience of the one observer appeared to overcome new biases from counting more species and the extra 2 min/transect was not logistically excessive. Surveys of all fishes recorded almost 80% more species than subset-only surveys and >130% higher total abundance on average from >150% more genera that included abundant and functionally important taxa. Overall, fish counts by an experienced observer were not biased by the number of species surveyed and counts of all species markedly improved assessments of reef fish diversity and function.     

Comparing methods for assessing mortality impacts of an aerial 1080 pest control operation on tomtits (

This study aimed to estimate the level of mortality of North Island tomtits (Petroica macrocephala toitoi) during an aerial 1080 possum poisoning operation in Tongariro Forest, New Zealand, and to evaluate transect-based alternatives to banding for monitoring tomtit populations. The operation used 12 g toxic (1080 at 0.15% weight/weight) cereal baits sown at 3 kg/ha. Transects were established at three neighbouring sites; two within the 1080 poison area, and one outside. The re-sighting of 14 out of 15 banded male tomtits at one site within the 1080 operation indicated that mortality was low. This was backed up by results from a before-after-control-impact (BACI) design to analyse density estimates from distance sampling along transects. We analysed the change in counts of territorial males before and after the operation based on the same transect surveys. This also showed little impact of poisoning on tomtits, and indicated that loss rates greater than 8.4% due to 1080 were incompatible with the data (95% one-sided confidence bound). Counts of territorial males gave a much tighter confidence bound than the banding or distance sampling results. Of the techniques applied, the counting of territorial males appears to have the most promise for providing high-precision estimates of short-term impacts, by taking full advantage of the territorial habits of male tomtits in spring. However, distance sampling shows potential for providing the basis for longer-term monitoring of tomtit populations. The transect-based approaches involved substantially fewer resources than banding for estimating short-term impacts, and offer a considerably less-intensive means of longer-term monitoring of tomtits.     

A comparison of some population density sampling techniques for biodiversity, conservation, and environmental impact studies

Twelve terrestrial and marine studies were conducted at various sites in Malaysia, Brazil, and the United States between April 1999 and February 2004. These data were analyzed using five density estimate techniques for stationary (non-motile) organisms including Stratified Random Sampling, Point-Center Quarter, Third Nearest Object, Weinberg, and Strong. The Strong method gave the most accurate density estimates of stationary animals and plants. Stratified Random Sampling ranked second best and the Third Nearest Object the third best. Belt or strip transects may be preferable but can be restrictive in some situations because of logistics and associated time constraints. Straight line measurements on reefs were 3–27% more accurate than reef slack line and reef contour measurements. Most study areas measured with the standardized Morisita index of dispersion were moderately aggregated. Results from the Third Nearest Object and Point-Center Quarter techniques indicate that the addition of more data to establish a density correction factor does not necessarily give more accurate estimates of density.     

Correction for bias in visual transect censuses of coral reef fishes

Transect techniques for censusing reef fishes, and the sources of bias inherent in them are considered. A technique, derived from aeraly survey methods, is demonstrated to correct a bias in density estimates due to the width of the transect being censused. This bias is sufficient on a transect 1 m wide to underestimate density by 11.1–26.7% for five species or species groups examined. The bias is still greater on wider transects. Because this bias varies in degree among species, comparisons among species should not be made using uncorrected transect data. Comments are made on other probable sources of bias in transect data, and on ways of minimising bias when making visual transect censuses.     

Quantitative video sampling of coral reef benthos: large-scale application

The feasibility of reliably estimating percent cover of coral reef benthos by video techniques is examined. Video belt transects were recorded within study areas on Davies and John Brewer Reefs in the central Great Barrier Reef during September 1988. Two years later in September 1990, the study area at Davies Reef was resampled concurrently by video and line intercept transects. Percent cover data of major coral growth forms and non-biotic physionomic attributes were extracted from video footage by scoring the identity of items located at even or random spaced points along the transect. A cost-benefit analysis which compared increases in precision with increases in sampling effort suggested that the optimum regime for analyzing 200 m long video transects was five subsamples of 110 random points or one sample of 550. This regime resolved both spatial variability at large and smaller spatial scales (between study areas and among transects) and temporal change (within a single study area over a two year period) for total live coral and individual growth forms. The strengths of the technique lie in its cost-savings in filed expenses, and in the production of a permanent visual record. The limitations of the technique lie in reduced taxonomic resolution when compared with hands on field techniques. The results suggest that for broad taxonomic categories of coral reef benthos reliable estimates of relative abundance can be obtained by video techniques.     

Community Structure of Fishes and Habitat Complexity on a Tropical Rocky Shore

The relationship between the variables of reef fish community structure (fish richness, fish diversity and total number of fishes) and those of habitat complexity (total surface area, substratum diversity, topographic complexity, number of holes, percent cover of turf algae, zoanthids, millepores, massive corals, bare rock, encrusting calcareous algae, urchins, other sessile organisms and Sargassum) were examined on three different rocky shores (FA, FB and FT) at Arraial do Cabo, a tropical region located on Brazilian southeastern coast (23° S, 42° W). Fish abundance and vertical distribution were assessed by a visual census technique using strip transects. Percentage cover of benthic organisms and other substratum types were calculated by replicated transects using a chain laid down on the substratum. Topographic complexity was determined by the chain link method and number of holes were estimated by direct counts on replicated transects. More than 91 fishes belonging to 37 families were seen in all study sites during one year of visual census effort. FA and FB sites had similar fish community structure and habitat complexity characteristics, while FT showed different habitat characteristics and higher fish diversity and richness, plus total number of fishes. Vertical distribution of fishes along the rocky shores studied seemed to be predictable and was determined by factors such as feeding habits and behavior, availability of refuges and social interactions. The habitat variables that best explained the higher diversity and number of fishes observed in FT site were total surface area of rocky shores and the abundance of benthic sessile invertebrates; these conditions were typical of rocky shores more exposed to wave surge.     

Laboratory sources of error for algal community attributes during sample preparation and counting

Applied algal studies typically require enumeration of preserved cells. As applications of algal assessments proliferate, understanding sources of variability inherent in the methods by which abundance and species composition data are obtained becomes even more important for precision of measurements. We performed replicate counts of diatoms on permanently fixed coverglasses and all algae in Palmer–Maloney chambers to assess precision and accuracy of measurements derived from common counting methods. We counted diatoms and all algae with transects and random fields. Variability estimates (precision) of diatom density, species diversity, and species composition on permanent coverglasses were low between replicate subsamples and between replicate transects. However, average density estimates of diatoms settled on coverglasses determined with transect methods were 42–52% greater than density estimates made with random fields. This bias was due to a predictable, nonrandom distribution of diatoms on the coverglass with few diatoms near edges. Despite bias in density when counting diatoms along coverglass transects, no bias was observed in estimates of species composition. Estimates of density and taxa richness of all-algae in Palmer–Maloney chambers also had low variability among multiple transects and high similarity in species composition between transects. In addition, counting method in Palmer–Maloney chambers did not affect estimates of algal cell density, taxa richness, and species composition, which suggested that counting units were distributed randomly in the chambers. Thus, most sources of variability in sample preparation and analysis are small; however, transect counts should not be used to estimate cell density, and sufficient numbers of random fields must be counted to account for edge effects on cell distribution with material settled on permanently fixed coverglasses.     

Scale-dependent spatial variability of coral assemblages along the Florida Reef Tract

 Coral reef communities of the western Atlantic have changed over the past two to three decades, but the magnitude and causes of this change remain controversial. Part of the problem is that small-scale patterns observed on individual reefs have been erroneously extrapolated to landscape and geographic scales. Understanding how reef coral assemblages vary through space is an essential prerequisite to devising sampling strategies to track the dynamics of coral reefs through time. In this paper we quantify variation in the cover of hard corals in spur-and-groove habitats (13–19 m depth) at spatial scales spanning five orders of magnitude along the Florida Reef Tract. A videographic sampling program was conducted to estimate variances in coral cover at the following hierarchical levels and corresponding spatial scales: (1) among transects within sites (0.01- to 0.1-km scale), (2) among sites within reefs (0.5- to 2-km scale), (3) among reefs within sectors of the reef tract (10- to 20-km scale), and (4) among sectors of the reef tract (50- to 100-km scale). Coral cover displayed low variability among transects within sites and among sites within reefs. This means that transects from a site adequately represented the variability of the spur-and-groove habitat of the reef as a whole. Variability among reefs within sectors was highly significant, compared with marginally significant variability among sectors. Estimates from an individual reef, therefore, did not adequately characterize nearby reefs, nor did those estimates sufficiently represent variability at the scale of the sector. The structure and composition of coral reef communities is probably determined by the interaction of multiple forcing functions operating on a variety of scales. Hierarchical analyses of coral assemblages from other geographic locations have detected high variability at scales different from those in the present study. A multiscale analysis should, therefore, precede any management decisions regarding large reef systems such as the Florida Reef Tract. Accepted: 19 July 1999     

Biased underwater visual census biomass estimates for target-species in tropical reef fisheries

Visual census biomass estimates are compared with yields of three emperor (Lethrinidae) genera from shallow water fished reef habitats and we demonstrate that visual census can consistently underestimate the exploitable biomass of Lethrinus spp.     

Improvement of fish length estimates for underwater visual census of reef fish biomass

Accuracy and precision are of great importance in the assessment of reef fish biomass when conducting an underwater visual census (UVC). Quantification and subsequent correction of the bias is required in order to standardize the estimates and correct for underwater distortion. To optimize the UVC, the observer should receive length‐measurement training in order to obtain in situ‐measurements that are as accurate and precise as possible. The objective of this study was to quantify the bias of fish length measurements made by divers with and without training in order to enhance reef fish biomass estimates. Adaptation of the diver to estimate fish lengths was analysed as a part of reef fish biomass monitoring in the Karimunjawa National Park, a national marine sanctuary in the Java Sea, Indonesia. Two divers practiced estimating a variety of fish in a natural environment by using styrofoam models attached to strings and sinkers. Analyses showed that by training the diver, his/her accuracy and precision improved substantially. Proving its reliability, an underwater visual census (UVC) becomes a useful and reliable method to assess the biomass of reef fishes.     

A DESCRIPTION OF THE CORAL STRUCTURE OF TUTIA REEF (PANGANYIKA TERRITORY, EAST AFRICA), AND ITS FISH FUNA

The origin and structure of Tutia Reef, a coral reef off the Tanganyika coast, is described. It is concluded that the reef is formed in part from the cutting back flat of the older reef 14 ft above the level of the present reef, and in part from seaward growth of the reef.
Based on lines of quadrats down three sides of the reef an assessment of the cover of living coral on the reef is made, and the distribution of the dominant corals of the reef described. It is suggested that, contrary to the views of Crossland (1902, 1903), many East African coral reefs are undergoing seaward growth.
The distribution of the fishes on the reef is described, and a correlation is shown with the distribution of living coral on the reef. Many species of fishes have restricted distribution on the reef.
The feeding relationships of the fishes are discussed, and the total fish biomass divided into separate feeding categories. Tutia Reef has a low biomass of herbivorous fishes compared with the results from Eniwetok Atoll obtained by Odum & Odum (1955). This is considered to be due to the fact that Tutia Reef exhibits characteristics of an outer reef, deriving much of its energy from plankton, not those of an enclosed and almost autotrophic logoon.     

The line transect method for estimating densities of large mammals in a tropical deciduous forest: An evaluation of models and field experiments

We have evaluated techniques of estimating animal density through direct counts using line transects during 1988–92 in the tropical deciduous forests of Mudumalui Sanctuary in southern India for four species of large herbivorous mammals, namely, chital (Axis axis). sambar (Cervus unicolor). Asian elephant (Elephas maximus) and gaur (Bos gaurus) Density estimates derived from the Fourier Series and the Half-Normal models consistently had the lowest coefficient of variation. These two models also generated similar mean density estimates. For the Fourier Series estimator, appropriate cut-off widths for analyzing line transect data for the four species are suggested. Grouping data into various distance classes did not produce any appreciable differences in estimates of mean density or their variances, although model fit is generally better when data arc placed in fewer groups. The sampling effort needed to achieve a desired precision (coefficient of variation) in the density estimate is derived. A sampling effort of 800 km of transects returned a 10% coefficient of variation on estimate for ehital; for the other species a higher effort was needed to achieve this level of precision. There was no statistically significant relationship between detectability of a group and the size of the group for any species. Density estimates along roads were generally significantly different from those in the interior of the forest, indicating that road-side counts many not be appropriate for most species.     

Recruitment rates in forest plots: Gf estimates using growth rates and size distributions

1 In censuses of tree populations in permanent plots, short census intervals and small population size lead to uncertainty in the observed recruitment rate of a minimum size. Increasing the census interval, however, underestimates the rate because of unrecorded 'recruit and die' events.
2 We propose a new Gf procedure for estimation of recruitment rates. Recruitment rate per area is obtained by multiplication of the density in the smallest size class (f) and the average size growth rate in that class (G) divided by the width of the class. This procedure is valid when the size distribution of the population examined is continuous with size.
3 When tree size structure is negative‐exponentially distributed, as is often the case in natural rain forest populations, the Gf estimate of the recruitment rate for a given size class was least biased close to the midpoint size of this class.
4 Gf estimates agreed well with census estimates of recruitment rate from permanent plots in rain forests. A tendency for Gf estimates to be larger than census estimates disappeared when census estimates were corrected for mortality after recruitment.
5 The effects of plot size, census interval and variation in growth rate on estimates of recruitment rate were simulated using model populations. Small plot size caused substantially more among‐plot deviation for the census count of recruitment events than for the Gf estimate. The census recruitment rate also showed larger variation among plots for shorter intervals than the Gf estimate, which was independent of census interval. The Gf estimates were therefore more accurate than census counts in many situations. More than several tens of trees were needed in a size class to allow a reliable Gf estimates.     

Variability in Abundance of Temperate Reef Fishes Estimated by Visual Census

Identifying sources of sampling variation and quantifying their magnitude is critical to the interpretation of ecological field data. Yet, most monitoring programs of reef fish populations based on underwater visual censuses (UVC) consider only a few of the factors that may influence fish counts, such as the diver or census methodology. Recent studies, however, have drawn attention to a broader range of processes that introduce variability at different temporal scales. This study analyzes the magnitude of different sources of variation in UVCs of temperate reef fishes off Patagonia (Argentina). The variability associated with time-of-day, tidal state, and time elapsed between censuses (minutes, days, weeks and months) was quantified for censuses conducted on the five most conspicuous and common species: Pinguipes brasilianus, Pseudopercis semifasciata, Sebastes oculatus, Acanthistius patachonicus and Nemadactylus bergi. Variance components corresponding to spatial heterogeneity and to the different temporal scales were estimated using nested random models. The levels of variability estimated for the different species were related to their life history attributes and behavior. Neither time-of-day nor tidal state had a significant effect on counts, except for the influence of tide on P. brasilianus. Spatial heterogeneity was the dominant source of variance in all but one species. Among the temporal scales, the intra-annual variation was the highest component for most species due to marked seasonal fluctuations in abundance, followed by the weekly and the instantaneous variation; the daily component was not significant. The variability between censuses conducted at different tidal levels and time-of-day was similar in magnitude to the instantaneous variation, reinforcing the conclusion that stochastic variation at very short time scales is non-negligible and should be taken into account in the design of monitoring programs and experiments. The present study provides baseline information to design and interpret results from visual census programs in temperate reefs.     

A global coral reef probability map generated using convolutional neural networks

Coral reef research and management efforts can be improved when supported by reef maps providing local-scale details across global extents. However, such maps are difficult to generate due to the broad geographic range of coral reefs, the complexities of relating satellite imagery to geomorphic or ecological realities, and other challenges. However, reef extent maps are one of the most commonly used and most valuable data products from the perspective of reef scientists and managers. Here, we used convolutional neural networks to generate a globally consistent coral reef probability map—a probabilistic estimate of the geospatial extent of reef ecosystems—to facilitate scientific, conservation, and management efforts. We combined a global mosaic of high spatial resolution Planet Dove satellite imagery with regional Millennium Coral Reef Mapping Project reef extents to build training, validation, and application datasets. These datasets trained our reef extent prediction model, a neural network with a dense-unet architecture followed by a random forest classifier, which was used to produce a global coral reef probability map. Based on this probability map, we generated a global coral reef extent map from a 60% threshold of reef probability (reef: probability ≥ 60%, non-reef: probability < 60%). Our findings provide a proof-of-concept method for global reef extent estimates using a consistent and readily updateable methodology that leverages modern deep learning approaches to support downstream users. These maps are openly-available through the Allen Coral Atlas.

     

Animal Density and Track Counts: Understanding the Nature of Observations Based on Animal Movements

Counting animals to estimate their population sizes is often essential for their management and conservation. Since practitioners frequently rely on indirect observations of animals, it is important to better understand the relationship between such indirect indices and animal abundance. The Formozov-Malyshev-Pereleshin (FMP) formula provides a theoretical foundation for understanding the relationship between animal track counts and the true density of species. Although this analytical method potentially has universal applicability wherever animals are readily detectable by their tracks, it has long been unique to Russia and remains widely underappreciated. In this paper, we provide a test of the FMP formula by isolating the influence of animal travel path tortuosity (i.e., convolutedness) on track counts. We employed simulations using virtual and empirical data, in addition to a field test comparing FMP estimates with independent estimates from line transect distance sampling. We verify that track counts (total intersections between animals and transects) are determined entirely by density and daily movement distances. Hence, the FMP estimator is theoretically robust against potential biases from specific shapes or patterns of animal movement paths if transects are randomly situated with respect to those movements (i.e., the transects do not influence animals’ movements). However, detectability (the detection probability of individual animals) is not determined simply by daily travel distance but also by tortuosity, so ensuring that all intersections with transects are counted regardless of the number of individual animals that made them becomes critical for an accurate density estimate. Additionally, although tortuosity has no bearing on mean track encounter rates, it does affect encounter rate variance and therefore estimate precision. We discuss how these fundamental principles made explicit by the FMP formula have widespread implications for methods of assessing animal abundance that rely on indirect observations.