Part 2 - Development of protocols to improve accuracy in baited
video techniques used to detect effects of zoning
Research Report
Mike Cappo, Glenn De'ath, Marcus Stowar, Charlotte Johannson and
Peter Doherty
Australian Institute of Marine Science, Townsville
ISBN 9781921359378
Published October 2009
MTSRF Project 4.8.2 -
Influence of the GBR Zoning Plan on inshore habitats and
biodiversity, of which fish and corals are indicators: Reefs
and shoals
Executive Summary
Baited video techniques (BRUVS) are a useful, non-destructive
approach to measure the effects of zoning in marine parks. The use
of stereo pairs allows extremely accurate and precise fish
measurments.
However, the technique is so novel that it lacks protocols in
data collection. It can therefore be argued that bias introduced by
the performance of various tape readers could reduce the
'signal-to-noise' ratio in comparing diversity, abundance and fish
lengths between zones with the BRUVS technique.
By repeating readings in a robust sampling design, we
demonstrated remarkable consistency amongst three readers and two
readings in counting highly-prized red emperor, coral trout and
red-throat emperor in the field of view of the BRUVS. Almost all
the variation in those species could be attributed to the
tapes.
However, three minor sources of error emerged across the entire
suite of 257 species recorded in the test. Firstly, a 'new' reader
added the sightings of a considerable number of the same, small
taxa multiple times as 'new species' whereas the other readers did
identify them only once to species level. This inflated the
richness (but not the abundance) recorded by the 'new' reader.
Secondly, there were obvious differences amongst readers in
distinguishing species with very similar appearance. Species pairs
in the genera Cantheschenia, Heniochus, Naso,
Acanthurus, Pseudolabrus, Parapercis and
Pomacentrus were notable sources of variation in
identification.
Finally, we found that the 'new' and 'veteran' readers were
inadvertently switching choices in identification, rather than
missing sightings, between readings for some of these 'difficult'
species pairs. In contrast, the 'experienced' reader showed
remarkable consistency.
These three errors accounted for, at most, only one-third of the
variation due to the differences amongst tapes. At this level they
would not affect the signal-to-noise ratio for any of the 'target',
'bycatch' and 'unfished' groups of fish in assessing effects of
zoning.
We recommended the best protocols to assure the quality of BRUVS
data collection would:
-
Organise the 'experienced' reader to process the first few tapes
in the collection of replicates for a sampling site, then allow
'new' readers to complete the remaining tapes by referring to, and
learning from, the names and images stored for that site;
We developed a 'rule' and a protocol for allocating measurement
effort that maximised opportunities for measurement of the full
range of fish sizes in the field of view of stereo-video BRUVS.
Smaller individuals were arriving earlier than larger conspecifics
in the case of coral trout, red emperor and perhaps Venus tuskfish.
There were clear differences between southern 'shoals' open and
closed to fishing in the shape of length frequency plots. These
could not be immediately attributed to removals by fishing in the
absence of temporal data to quantify shoal-specific differences in
recruitment and growth.
