Limits...
Overestimating fish counts by non-instantaneous visual censuses: consequences for population and community descriptions.

Ward-Paige C, Mills Flemming J, Lotze HK - PLoS ONE (2010)

Bottom Line: Several studies have demonstrated the effectiveness of protected areas for increasing fish abundance or provided insight into the natural abundance and structure of reef fish communities in remote areas.Based on our results, we urge that published literature state explicitly whether instantaneous counts were made and that survey procedures be accounted for when non-instantaneous counts are used.Because such studies are used to make important management and conservation decisions, incorrect estimates of animal abundance and biomass have serious and significant implications.

View Article: PubMed Central - PubMed

Affiliation: Department of Biology, Dalhousie University, Halifax, Nova Scotia, Canada. globalshark@gmail.com

ABSTRACT

Background: Increasingly, underwater visual censuses (UVC) are used to assess fish populations. Several studies have demonstrated the effectiveness of protected areas for increasing fish abundance or provided insight into the natural abundance and structure of reef fish communities in remote areas. Recently, high apex predator densities (>100,000 individuals x km(-2)) and biomasses (>4 tonnes x ha(-1)) have been reported for some remote islands suggesting the occurrence of inverted trophic biomass pyramids. However, few studies have critically evaluated the methods used for sampling conspicuous and highly mobile fish such as sharks. Ideally, UVC are done instantaneously, however, researchers often count animals that enter the survey area after the survey has started, thus performing non-instantaneous UVC.

Methodology/principal findings: We developed a simulation model to evaluate counts obtained by divers deploying non-instantaneous belt-transect and stationary-point-count techniques. We assessed how fish speed and survey procedure (visibility, diver speed, survey time and dimensions) affect observed fish counts. Results indicate that the bias caused by fish speed alone is huge, while survey procedures had varying effects. Because the fastest fishes tend to be the largest, the bias would have significant implications on their biomass contribution. Therefore, caution is needed when describing abundance, biomass, and community structure based on non-instantaneous UVC, especially for highly mobile species such as sharks.

Conclusions/significance: Based on our results, we urge that published literature state explicitly whether instantaneous counts were made and that survey procedures be accounted for when non-instantaneous counts are used. Using published density and biomass values of communities that include sharks we explore the effect of this bias and suggest that further investigation may be needed to determine pristine shark abundances and the existence of inverted biomass pyramids. Because such studies are used to make important management and conservation decisions, incorrect estimates of animal abundance and biomass have serious and significant implications.

Show MeSH
Diagram of the visual sampling field for the belt-transect diver (left) and the stationary-point-count diver (right).Symbols: at is the total area surveyed by the belt-transect diver; vat is the area visually surveyed by the belt-transect diver in one time step; tw is the width of the belt-transect; v is the visibility distance; as is the total area surveyed by the point count diver; vas is the area visually surveyed by the point count diver in one time step.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2908695&req=5

pone-0011722-g002: Diagram of the visual sampling field for the belt-transect diver (left) and the stationary-point-count diver (right).Symbols: at is the total area surveyed by the belt-transect diver; vat is the area visually surveyed by the belt-transect diver in one time step; tw is the width of the belt-transect; v is the visibility distance; as is the total area surveyed by the point count diver; vas is the area visually surveyed by the point count diver in one time step.

Mentions: Our model AnimDens was written in R [26] to simulate divers counting fish while deploying the belt-transect and stationary-point-count UVC techniques (File S1). Two experiments were used to explore the effect of fish mobility and survey procedure on the difference between observed and expected counts. Simulations were run across a range of fish speeds and survey procedures to determine the effect of these parameters on observed counts. Figure 1 shows sample runs for fish that remained still, that moved at 0.5 m·s−1, and at 1.0 m·s−1. For simplicity, the model assumed a sample area that was featureless, flat and 1 m deep. For each simulation, a diver from each of the two census methods was placed in the centre of the sample area and each had an orientation of 90° (each facing in the same direction) at initial time, t0. The sample area was populated with fish with a random distribution and random initial orientation. At t0 the number of fish observed and recorded by each diver was a function of the distance and the angle between the diver and each fish (those located within view of the diver). For the belt-transect diver, the distance was set to maximum visibility v and an angle of ±90° but only to a distance of transect width (tw) to the right and left of the diver's location (Fig. 2A). Because the belt-transect diver sampled an area directly in front of them the angle was set to 180°. For the stationary-point-count diver, all fish within ±80° of the diver's main orientation were counted (Fig. 2B). Fish that reached the area boundaries were allowed to leave and return (i.e. not reflected back into the sample area). Note that the simulated divers did not recount fish they already recorded (as if they were all numbered), as divers strive to do in the field [17], [21].


Overestimating fish counts by non-instantaneous visual censuses: consequences for population and community descriptions.

Ward-Paige C, Mills Flemming J, Lotze HK - PLoS ONE (2010)

Diagram of the visual sampling field for the belt-transect diver (left) and the stationary-point-count diver (right).Symbols: at is the total area surveyed by the belt-transect diver; vat is the area visually surveyed by the belt-transect diver in one time step; tw is the width of the belt-transect; v is the visibility distance; as is the total area surveyed by the point count diver; vas is the area visually surveyed by the point count diver in one time step.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2908695&req=5

pone-0011722-g002: Diagram of the visual sampling field for the belt-transect diver (left) and the stationary-point-count diver (right).Symbols: at is the total area surveyed by the belt-transect diver; vat is the area visually surveyed by the belt-transect diver in one time step; tw is the width of the belt-transect; v is the visibility distance; as is the total area surveyed by the point count diver; vas is the area visually surveyed by the point count diver in one time step.
Mentions: Our model AnimDens was written in R [26] to simulate divers counting fish while deploying the belt-transect and stationary-point-count UVC techniques (File S1). Two experiments were used to explore the effect of fish mobility and survey procedure on the difference between observed and expected counts. Simulations were run across a range of fish speeds and survey procedures to determine the effect of these parameters on observed counts. Figure 1 shows sample runs for fish that remained still, that moved at 0.5 m·s−1, and at 1.0 m·s−1. For simplicity, the model assumed a sample area that was featureless, flat and 1 m deep. For each simulation, a diver from each of the two census methods was placed in the centre of the sample area and each had an orientation of 90° (each facing in the same direction) at initial time, t0. The sample area was populated with fish with a random distribution and random initial orientation. At t0 the number of fish observed and recorded by each diver was a function of the distance and the angle between the diver and each fish (those located within view of the diver). For the belt-transect diver, the distance was set to maximum visibility v and an angle of ±90° but only to a distance of transect width (tw) to the right and left of the diver's location (Fig. 2A). Because the belt-transect diver sampled an area directly in front of them the angle was set to 180°. For the stationary-point-count diver, all fish within ±80° of the diver's main orientation were counted (Fig. 2B). Fish that reached the area boundaries were allowed to leave and return (i.e. not reflected back into the sample area). Note that the simulated divers did not recount fish they already recorded (as if they were all numbered), as divers strive to do in the field [17], [21].

Bottom Line: Several studies have demonstrated the effectiveness of protected areas for increasing fish abundance or provided insight into the natural abundance and structure of reef fish communities in remote areas.Based on our results, we urge that published literature state explicitly whether instantaneous counts were made and that survey procedures be accounted for when non-instantaneous counts are used.Because such studies are used to make important management and conservation decisions, incorrect estimates of animal abundance and biomass have serious and significant implications.

View Article: PubMed Central - PubMed

Affiliation: Department of Biology, Dalhousie University, Halifax, Nova Scotia, Canada. globalshark@gmail.com

ABSTRACT

Background: Increasingly, underwater visual censuses (UVC) are used to assess fish populations. Several studies have demonstrated the effectiveness of protected areas for increasing fish abundance or provided insight into the natural abundance and structure of reef fish communities in remote areas. Recently, high apex predator densities (>100,000 individuals x km(-2)) and biomasses (>4 tonnes x ha(-1)) have been reported for some remote islands suggesting the occurrence of inverted trophic biomass pyramids. However, few studies have critically evaluated the methods used for sampling conspicuous and highly mobile fish such as sharks. Ideally, UVC are done instantaneously, however, researchers often count animals that enter the survey area after the survey has started, thus performing non-instantaneous UVC.

Methodology/principal findings: We developed a simulation model to evaluate counts obtained by divers deploying non-instantaneous belt-transect and stationary-point-count techniques. We assessed how fish speed and survey procedure (visibility, diver speed, survey time and dimensions) affect observed fish counts. Results indicate that the bias caused by fish speed alone is huge, while survey procedures had varying effects. Because the fastest fishes tend to be the largest, the bias would have significant implications on their biomass contribution. Therefore, caution is needed when describing abundance, biomass, and community structure based on non-instantaneous UVC, especially for highly mobile species such as sharks.

Conclusions/significance: Based on our results, we urge that published literature state explicitly whether instantaneous counts were made and that survey procedures be accounted for when non-instantaneous counts are used. Using published density and biomass values of communities that include sharks we explore the effect of this bias and suggest that further investigation may be needed to determine pristine shark abundances and the existence of inverted biomass pyramids. Because such studies are used to make important management and conservation decisions, incorrect estimates of animal abundance and biomass have serious and significant implications.

Show MeSH