309
9 Studying and Monitoring Aggregating Species
aggregation has been surveyed with precision. If numbers vary, with individuals
moving, hiding or behaving unnaturally, this implies reduced precision for surveys
and such data viewed more sceptically. Hiding can even be sex-specifi c (Sadovy
et al. 1994 ) . Other factors, such as water visibility, frustrate determining fi sh
numbers in even small aggregations.
Accuracy decreases with increasing aggregation size (i.e. number of fi sh), or if
the species concerned is shy or otherwise diffi cult to survey. What sort of accuracy
can be expected for various types of surveys? For fi sh numbers over about 100 individuals accuracy drops rapidly with increasing numbers, errors amplifi ed by movements of individuals and the group as a whole. At some point there is a major
disconnect between the ability of a human observer(s) to determine the numbers of
fi sh present by direct observation and actual numbers.
If many thousands of fi sh are aggregated for spawning, how can the accuracy of
estimated numbers be determined (e.g. Fig. 9.8 , Chaps. 12.6, 12.9, 12.10, 12.11 ). At
present there is no way to do that, although referring to examples from the literature
on counting birds in fl ocks may be useful when the task is to count large numbers of
massed fi sh. How can the differences between 5,000 versus 6,000 fi sh swimming in
a densely packed aggregation be determined by divers in the water? It should just be
accepted that using present visual methods estimates of fi sh numbers in large aggregations are inaccurate, that estimates of fi sh numbers become increasingly unreliable with increasing numbers and that this information should be treated as more
qualitative than quantitative. Extreme qualitative changes (say from 1,000 to 200
fi sh) in fi sh numbers can be detected by direct observational surveys, but the actual
values cannot. But by defi nition, estimates cannot be accurate and there is little basis
to say the “accuracy of estimates” can be improved. Estimates can be made more
precise through improved or multiple methods, ideally so estimated numbers can
have the same level of precision day to day. However accuracy remains elusive and
other factors such as inter-diver variation in counts should also be considered.
Training in estimating fi sh numbers can help to an extent, mostly by looking
at underwater photographs or videos of fi sh groups (without knowing the actual
number visible) and estimating the numbers present, and then comparing with the
actual numbers of fi sh visible. In this manner the observers can get an idea of what
100, or 500 or 1,000 fi sh look like underwater. Using multiple observers and
arriving at consensus values for fi sh numbers may be helpful, but the basic lack of
accuracy still persists. Estimates should not be treated later as accurate data. Such
qualitative data, while useful, does not identify absolute trends in population
numbers in aggregations (Heyman et al. 2005 ) and there is still no way to assess
accuracy levels, although Heyman and Kjerfve ( 2008 ) state, using two teams of
observers, “trends in abundance were captured with accuracy within 10% since the
maximum numbers ..... were obtained on the same day by both teams”. Where
there were major differences between team counts, this was attributed to one team
having seen a school of fi sh which the other missed. Doing statistical analyses on
such estimates is inappropriate.
Single line transects through aggregation areas have been used to obtain quantitative data on fi sh numbers although multiple, often parallel transects are most likely
9 Studying and Monitoring Aggregating Species
aggregation has been surveyed with precision. If numbers vary, with individuals
moving, hiding or behaving unnaturally, this implies reduced precision for surveys
and such data viewed more sceptically. Hiding can even be sex-specifi c (Sadovy
et al. 1994 ) . Other factors, such as water visibility, frustrate determining fi sh
numbers in even small aggregations.
Accuracy decreases with increasing aggregation size (i.e. number of fi sh), or if
the species concerned is shy or otherwise diffi cult to survey. What sort of accuracy
can be expected for various types of surveys? For fi sh numbers over about 100 individuals accuracy drops rapidly with increasing numbers, errors amplifi ed by movements of individuals and the group as a whole. At some point there is a major
disconnect between the ability of a human observer(s) to determine the numbers of
fi sh present by direct observation and actual numbers.
If many thousands of fi sh are aggregated for spawning, how can the accuracy of
estimated numbers be determined (e.g. Fig. 9.8 , Chaps. 12.6, 12.9, 12.10, 12.11 ). At
present there is no way to do that, although referring to examples from the literature
on counting birds in fl ocks may be useful when the task is to count large numbers of
massed fi sh. How can the differences between 5,000 versus 6,000 fi sh swimming in
a densely packed aggregation be determined by divers in the water? It should just be
accepted that using present visual methods estimates of fi sh numbers in large aggregations are inaccurate, that estimates of fi sh numbers become increasingly unreliable with increasing numbers and that this information should be treated as more
qualitative than quantitative. Extreme qualitative changes (say from 1,000 to 200
fi sh) in fi sh numbers can be detected by direct observational surveys, but the actual
values cannot. But by defi nition, estimates cannot be accurate and there is little basis
to say the “accuracy of estimates” can be improved. Estimates can be made more
precise through improved or multiple methods, ideally so estimated numbers can
have the same level of precision day to day. However accuracy remains elusive and
other factors such as inter-diver variation in counts should also be considered.
Training in estimating fi sh numbers can help to an extent, mostly by looking
at underwater photographs or videos of fi sh groups (without knowing the actual
number visible) and estimating the numbers present, and then comparing with the
actual numbers of fi sh visible. In this manner the observers can get an idea of what
100, or 500 or 1,000 fi sh look like underwater. Using multiple observers and
arriving at consensus values for fi sh numbers may be helpful, but the basic lack of
accuracy still persists. Estimates should not be treated later as accurate data. Such
qualitative data, while useful, does not identify absolute trends in population
numbers in aggregations (Heyman et al. 2005 ) and there is still no way to assess
accuracy levels, although Heyman and Kjerfve ( 2008 ) state, using two teams of
observers, “trends in abundance were captured with accuracy within 10% since the
maximum numbers ..... were obtained on the same day by both teams”. Where
there were major differences between team counts, this was attributed to one team
having seen a school of fi sh which the other missed. Doing statistical analyses on
such estimates is inappropriate.
Single line transects through aggregation areas have been used to obtain quantitative data on fi sh numbers although multiple, often parallel transects are most likely