Content uploaded by Illangarathne Arachchige Kavindu Chiranjeewa Weerarathne
Author content
All content in this area was uploaded by Illangarathne Arachchige Kavindu Chiranjeewa Weerarathne on Aug 04, 2021
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Ecological Indicators 122 (2021) 107262
Available online 23 December 2020
1470-160X/© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Sample-size requirements for accurate length-frequency distributions of
mesophotic reef shes from baited remote underwater stereo video
Illangarathne Arachchige Weerarathne, Jacquomo Monk
*
, Neville Barrett
Institute for Marine and Antarctic Studies, University of Tasmania, Private Bag 49, Hobart, Tasmania 7001, Australia
ARTICLE INFO
Keywords:
Demersal sh monitoring
Fisheries monitoring
No-take marine reserve monitoring
Size structure
Stereo BRUV
Mesophotic temperate reef
ABSTRACT
Accurate descriptions of size structure are important for adaptive management of marine sh populations subject
to anthropogenic and environmental pressures. This requires monitoring programs that can measure the length
of enough individuals within each focal sh population. This study assessed the number of length measurements
required to accurately describe the size structure in a range of common mesophotic demersal sh species
observed from baited remote underwater stereo video (stereo BRUV) sampling programs. Here, we use a
resampling approach from an empirical length dataset collected as a part of ongoing monitoring efforts to
characterize mesophotic reef sh assemblages across the continental shelf of eastern and southern Tasmania,
Australia. The results suggest that, on average, between 60 and 120 individuals length measurements are needed
from at least 20 to 2000 independent deployments to be 95% condent that samples reected the “real” size
structure of shes captured using stereo BRUVs. It is important to note that the “real” size-structure of each sh
species here is unknown but was parameterized by pooling all measurements across the stereo BRUV dataset. It
should also be noted that the exact number of length measurements differs across species, with some less
abundant species requiring substantial sampling effort. This study helps to inform initial sampling requirements
for length measurements for monitoring programs using stereo BRUVs. It provides a methodology that can aid
researchers to further rene the overall sampling effort for future sheries and marine park monitoring
applications.
1. Introduction
Size structure information provides important insights into the
reproductive potential, growth and stability of demersal reef marine sh
populations (Hixon et al., 2014; van Overzee and Rijnsdorp, 2015). A
lack of smaller size classes of sh can suggest deciencies in recruitment,
while infrequency of larger size classes might indicate mortality of
mature sh (Neumann and Allen, 2007). While there are multiple causes
to alterations in size-class structure, shing and climate change are
recognized as some of the biggest contributors (Barnett et al., 2017;
Queiros et al., 2018). Fishing causes size-selective removal of larger
individuals that truncates the size structure of sh populations (Berkeley
et al., 2004). Fishing can also inuence the trophic structure of an
ecosystem resulting in changes to size-selective predation (Mitchell
et al., 2019) and through shifts in competitive interactions within and
between species (Jenkins et al., 1999). Similarly, ocean warming has
been suggested to cause widespread declines in organism body sizes
through changes to water temperature, oxygen content and other ocean
biogeochemical properties that directly affect ecophysiology of water-
breathing organisms (Cresswell et al., 2019; Sheridan and Bickford,
2011). Ultimately, the age and size demographics of demersal reef
populations are determined by the complex interactions among these
factors because they typically experience multiple stressors simulta-
neously. Understanding these dynamics is critical to understanding the
status of shery or conservation actions (Cresswell et al., 2019) and how
demersal reef sh populations are responding to environmental stressors
(Audzijonyte et al., 2016). Therefore, accurately describing the size
structure of demersal reef sh populations is important, but it requires
capturing an adequate sample size of individuals to reliably characterize
the length-frequency distribution of individual species components.
The sample size requirements for accurately describing the size
structure of shes has been undertaken for several small-bodied sh
species (<250 mm maximum length) and evaluated using statistical
resampling of simulated and empirical data sets (Schultz et al., 2016;
Vokoun et al., 2001). For example, Vokoun et al., (2001) suggested
sample sizes of 300 – 400 individuals were suitable to describe
* Corresponding author.
E-mail address: jacquomo.monk@utas.edu.au (J. Monk).
Contents lists available at ScienceDirect
Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind
https://doi.org/10.1016/j.ecolind.2020.107262
Received 18 December 2019; Received in revised form 2 December 2020; Accepted 8 December 2020
Ecological Indicators 122 (2021) 107262
2
population size structure, and Miranda (2007) concluded that size
structure can be adequately described with smaller sample sizes for
shes (150 – 425 individuals) with limited size structure distributions (i.
e., maximum length 200 – 300 mm). However, specic assessments of
sample size requirements for describing size structure for demersal
mesophotic reef shes are virtually absent in the published literature.
For demersal marine sh populations, there are numerous options to
attain accurate size structure data. These methods can be classied as
extractive and non-extractive. Extraction methods are associated with
shing gears such as seine nets, trawls, shing lines and traps that
capture sh enabling scientists to precisely measure them directly.
While these methods are still commonly used in the sheries sector, they
are not ideal in situations where either the species or survey region are
protected (Mallet and Pelletier, 2014; Przeslawski et al., 2018; Whit-
marsh et al., 2017). Non-extractive methods often rely on underwater
visual census and remotely deployed videography related methods to
estimate the size of sh in situ. While each method has their associated
bias, stereo-video systems, such as baited remote underwater stereo
video (stereo BRUV), are increasingly being recognized (Langlois et al.,
2020) for their ability to precisely measure the size of a broad range of
shes across multiple depths and habitat types (Collins et al., 2017;
Heyns-Veale et al., 2016; Wellington et al., 2018) as well as the provi-
sion of a permanent record that can be reviewed to reduce interobserver
variability (Cappo et al., 2009). However, the often limited and xed
eld of view can limit the number of length measurements taken within
the footage. These length measurements are commonly taken at MaxN
(i.e. the maximum number of individuals for that species present in a
single frame within a deployment; Langlois et al., 2020). There is a
requirement that the individual sh needs to be clearly visible in the
eld of view of both cameras. However, it is often hard to follow in-
dividuals, particularly if in a school, and identify accurately to avoid
issues of re-measuring the same sh, and some individuals may be
obscured by others, the substrate, bait bag or diode arm. In some situ-
ations, as few as 30% of individuals can be measured (Mitchell et al.,
accepted). The reduced number of individual sh being measured im-
pacts the accuracy of the length structure data from stereo BRUVs.
Stereo BRUVs are now accepted as a standard operating protocol for
monitoring demersal reef shes on mesophotic reefs throughout
Australia, particularly in the network of recently established Australian
Marine Parks (Przeslawski et al., 2019) and increasingly world-wide
(Langlois et al., 2020). Accordingly, the aims of this study were two-
fold: 1) to assess the number of length measurements required from
stereo BRUVs to accurately describe the size structure of typical
demersal sh populations and the key species they are composed from
and, 2) calculate how many individual stereo BRUV deployments would
typically be required in a survey program to reach this number of length
measurements. The datasets utilized were from a range of sampling
programs undertaken on temperate mesophotic rocky reef systems in
shelf waters of eastern and southern Tasmania, Australia, and were
examined via statistical resampling procedures. By focusing the analyses
on a suite of sheries important sh species that are often applied as
indicators for quantifying population structure change (e.g., Hill et al.,
2018), we provide further renement around optimal sample sizes
required in monitoring programs designed for both conservation and
sheries management within this region.
2. Materials and methods
2.1. Study region
This study used stereo BRUV deployments collected from six loca-
tions, including; Tasman Fracture Marine Park, South and South-east
Cape, the Friars, Butlers Point and Flinders Marine Park. Combined,
these locations provide coverage representing ~500 km of coastline
from the south-west to north-east of Tasmania (Fig. 1). Locations
covered a range of depths from ~30 to 180 m and consisted of pre-
dominantly sessile invertebrate dominated reef habitats. The southern
locations fall in and around the sanctuary zone (IUCN II) of the Tasman
Fracture Marine Park which has been in place since 2007. This region is
known for its’ unique sessile invertebrate communities which appear to
be associated with the exposure to regular south ocean fronts and the
Zeehan Current (Monk et al., 2016). The Friars and Butlers Point
Fig. 1. Location of stereo BRUV deployments used in resampling procedures for assessing sample size requirements for size class structure.
I. Arachchige Weerarathne et al.
Ecological Indicators 122 (2021) 107262
3
represent locations that remain open to commercial and recreational
shing pressures, while the northern locations fall within the multiple-
use zone (IUCN IV) of the Flinders Marine Park which received low to
moderate commercial shing effort (~2000 kg/year) on the continental
shelf before its declaration in 2007. Within the Flinders Marine Park,
demersal trawling was concentrated on the outer shelf, while hook, line
and gillnet shing were more dispersed across the shelf (Pitcher et al.,
2016). Since 2007, demersal trawling has been prohibited, but some
hook and line and recreational shing is allowed on the shelf of the
Marine Park (Williams et al., 2013).
2.2. Sampling design
A total of 251 stereo BRUV deployments were collected during the
austral summer/autumn months between 2012 and 2015 (Fig. 1). The
exact sampling designs for each location are detailed in Monk et al.
(2016), Lyle et al. (2017) and Hill et al. (2018). Each stereo BRUV
deployment represents a 60 min soak time following the standard
operating procedure described in Langlois et al. (2018). Stereo camera
pairs had 700 mm separation with a camera incidence angle of 8◦
mounted to a galvanized steel frame allowing a 500 mm height clear-
ance above benthos. Bait arm length was 1.5 m in front of the cameras
and the wire mesh bait holders contained ~1 kg of crushed pilchards
(Sardinops neopilchardus). Deployments in depths >50 m were arti-
cially illuminated using seven Royal Blue CREE XLamps XP-E LEDs
(delivering a radiant ux of 350–425 mW at wavelength ranging from
450 to 465 nm). Blue lights were used as they avoid potential behav-
ioural biases (Fitzpatrick et al., 2013). The stereo BRUVs were deployed
during daylight hours avoiding 1 h after sunrise and before sunset.
Concurrent deployments were separated by a minimum of 250 m to
avoid potential confounding following Langlois et al. (2018).
2.3. Video annotation and length measurements
All individual shes were identied to their lowest taxonomic level,
with their relative abundance estimated using the maximum number of
sh occurring in any one frame for each species (MaxN; Ellis and
Demartini, 1995). Only sh within a standardized 4 m eld of view of
the bait bag were annotated and measured. The length of all sh species
was recorded for as many individuals as possible occurring within
frames adjacent to MaxN as some individuals were obscured by other
sh. Calibrations, annotations and measurements were performed using
methods outlined in Langlois et al. (2020) with calibrations completed
in software Cal (www.seagis.com.au), and annotations and
Table 1
Distribution, ecology and sheries information for focal species.
Family Scientic name Distribution and ecology Fisheries
Australian distribution Habitat Depth
range
Trophic
ecology
Fishery Target/
Incidental
Cheilodactylidae Nemadactylus
macropterus
Southern Australia (Perth to Coffs
Harbour)
Soft sediment 0–400 m Benthic
invertivore
Commercial/
Recreational
Target
Chirodactylus
spectabilis
South-eastern Australia (Victor
Harbour to Sydney)
Hard bottom 0–50 m Benthic
invertivore
Commercial/
Recreational
Target
Nemadactylus
douglasii
South-eastern Australia
(Melbourne to Brisbane)
Hard bottom 0–200 m Benthic
invertivore
Commercial/
Recreational
Target
Cyttidae Cyttus australis South-eastern Australia (Ceduna to
Coffs Harbour)
Benthopelagic/Soft
sediment
20–250 m Benthic
invertivore
Commercial/
Recreational
Target
Dinolestidae Dinolestes lewini Southern Australia (Perth to
Sydney)
Benthopelagic 0–70 m Higher
carnivore
Commercial/
Recreational
Target
Notolabrus tetricus South-eastern Australia (Adelaide
to Sydney)
Mixed bottom 0–60 m Benthic
invertivore
Commercial/
Recreational
Target/
Incidental
Latridae Latris lineata Southern Australia (Albany to
Sydney)
Hard bottom 0–400 m Benthic
invertivore
Commercial/
Recreational
Target
Monacanthidae Meuschenia scaber Southern Australia (Margaret
River to Port Macquarie)
Hard bottom 0–200 m Benthic
invertivore
Bycatch Incidental
Thamnaconus degeni Southern Australia (Esperance to
Bairnsdale)
Mixed bottom 5–80 m Benthic
invertivore
Commercial/
Recreational
Target/
Incidental
Meuschenia australis South-eastern Australia (Victor
Harbour to Bairnsdale)
Hard bottom 0–30 m Browsing
herbivore
Commercial/
Recreational
Target/
Incidental
Meuschenia freycineti Southern Australia (Perth to Coffs
Harbour)
Mixed bottom 0–50 m Browsing
herbivore
Commercial/
Recreational
Target/
Incidental
Acanthaluteres
vittiger
Southern Australia (Perth to
Flinders Island)
Mixed bottom 0–40 m Browsing
herbivore
Commercial/
Recreational
Target/
Incidental
Eubalichthys gunnii South-eastern Australia (Adelaide
to Flinders Island)
Hard bottom 5–50 m Browsing
herbivore
Commercial/
Recreational
Target/
Incidental
Mordiae Morid cods South-eastern Australia (Adelaide
to Coffs Harbour)
Mixed bottom 10–350 m Higher
carnivore
Commercial/
Recreational
Target/
Incidental
Mullidae Upeneichthys
vlamingii
Southern Australia (Geraldton to
Sydney)
Soft sediment 0–200 m Benthic
invertivore
Commercial/
Recreational
Target
Neosebastidae Neosebastes
scorpaenoides
South-eastern Australia (Esperance
to Sydney)
Hard bottom 0–200 m Benthic
invertivore
Recreational Target/
Incidental
Platycephalidae Platycephalus
bassensis
Southern Australia (Perth to Coffs
Harbour)
Soft sediment 0–100 m Higher
carnivore
Commercial/
Recreational
Target
Platycephalus
richardsoni
South-eastern Australia (Victor
Harbour to Coffs Harbour)
Soft sediment 20–450 m Higher
carnivore
Commercial/
Recreational
Target
Scyliorhinidae Cephaloscyllium
laticeps
Southern Australia (Esperance to
Bairnsdale)
Mixed bottom 0–80 m Higher
carnivore
Bycatch Incidental
Sebastidae Helicolenus percoides South-eastern Australia (Portland
to Coffs Harbour)
Mixed bottom 10–400 m Higher
carnivore
Commercial/
Recreational
Target/
Incidental
Squalidae Squalus spp Australia wide (Port headland to
Townsville)
Soft sediment 0–600 m Higher
carnivore
Commercial Target/
Incidental
Triakidae Mustelus antarcticus Southern Australia (Perth to
Sydney)
Soft sediment 0–350 m Higher
carnivore
Commercial Target/
Incidental
I. Arachchige Weerarathne et al.
Ecological Indicators 122 (2021) 107262
4
measurements were done in the software EventMeasure Stereo (www.
seagis.com.au).
2.4. Selection of potential focal species
From the 94 potential species, 22 demersal sh species were chosen
based on their numerical abundance and their relevance to the shing
sector (Tables 1 and S1). Many of these species have large distributions
spanning temperate Australia and cover a wide range of maximum
lengths (i.e., 200–2000 mm). Some are targeted by both commercial and
recreational shers using various shing gears, some of which are pro-
hibited in multiple-use zone of the Flinders Marine Park (Table 1).
2.5. Statistical analysis
2.5.1. Number of length measurements
A resampling approach developed by Schultz et al. (2016) was
adapted to evaluate the number of length measurements required to
obtain a representative size-structure distribution for each of the focal
sh species from the stereo BRUVs. This approach involved a random
subsample, with replacement, of a range of individual lengths from the
entire pooled stereo BRUV dataset. For this study subsample sizes
ranged from 5 to 500 individual sh, at 5-sh intervals (i.e., 5, 15, 20…
500) with minimum and maximum lengths parameterized as smallest
and largest sh measured for that species. The subsample sizes represent
the approximate lower and upper limit of the number of sh encoun-
tered in a school that could be captured in a stereo BRUV eld of view
with moderate sampling effort.
To evaluate the accuracy of characterizing a size-structure distribu-
tion using different subsample sizes, the size frequencies of each sub-
sample was compared to the original pooled sample to estimate mean
absolute difference (MAD) in relative frequency in each length interval
using the following equation:
MAD =1
rr
i=11
bb
j=1pyj−pyij(1)
where p(yj) was the proportion of individuals in the jth interval from the
original pooled sample, p(
yij)was the proportion of individuals in the jth
interval and the ith subsample, b was the number of length intervals, and
r the number of iterations. Additional resampling was used to examine
the effect of size interval on MAD estimates. The width of individual size
class interval can change the interpretation of size-class distributions (e.
g., Langlois et al., 2012; Vokoun et al., 2001). For each subsample size,
1000 iterations were generated to calculate estimated MAD and
compute 95th percentile for each of three size class intervals of 10, 20,
and 50 mm. These size class intervals represent commonly used size
class bins for marine shes. For eample, Anderson and Neumann (1996)
suggests 10 mm intervals for sh that reach 300 mm, 20 mm intervals
for sh that reach 600 mm, and 50 mm intervals for sh that reach 1500
mm maximum length. The estimated MAD and the upper 95th percentile
of resamples for each subsample size were plotted. The interpretation of
these plots is straight forward, providing the relative accuracy of a given
sample size. For example, the sample size at which the 95th percentile
declines to below a MAD of 5% is interpreted as the number of length
measurements needed to yield sample size-class distribution across all
length bins within 5% of the size-class distribution of the original pooled
dataset 95% of the time. It is important to note that the true size-
structure of the sh population is unknown. However, Langlois et al.
(2012) has shown that size structure information captured at MaxN from
stereo BRUVs is similar to other sampling methods such as line and trap
data.
2.5.2. Number of stereo BRUV deployments
The proportion of sh measurable at MaxN and the 95th percentile
MAD values were used to calculate the number of stereo BRUV de-
ployments required to reach the minimum length measurement sample
size to adequately describe species/population size structure. This was
calculated using the following equation:
n=MAD95th
p(y) × x(m)(2)
where, for each species, n is the estimated number of stereo BRUV
samples, MAD95th is the 95th percentile MAD value, p(y) was the pro-
portion of individuals measurable, and x(m)was the mean MaxN
calculated across all 251 stereo BRUV samples.
3. Results
Ninety-four species were recorded across the surveyed population
(S1), with 22 retained for analysis due to their relevance to sheries.
Meuschenia scaber, Nemadactylus macropterus and Thamnaconus degeni
were the most abundant with mean MaxN per stereo BRUV drop of 5.4,
4.5 and 3.1, respectively (Table 2). Helicolenus percoides was the next
most abundant with a mean MaxN of 1.9, followed by Dinolestes lewini
(1.5), Morid cods (1.1) and Notolabrus tetricus (0.9; Table 2). Latris lin-
eata, Cephaloscyllium laticeps and Cyttus australis were moderately
abundant with between mean MaxN of between 0.5 and 0.6 (Table 2).
Meuschenia australis, Meuschenia freycineti, Acanthaluteres vittiger, Neo-
sebastes scorpaenoides, Eubalichthys gunnii, Upeneichthys vlamingii, Chiro-
dactylus spectabilis, Mustelus antarcticus, Nemadactylus douglasii,
Platycephalus bassensis, Platycephalus richardsoni and Squalus spp were
the least abundant with a mean MaxN of <0.5 per drop (Table 2).
Like the proportion of length measurements achieved at MaxN, the
range in species length and the proportion measured varied considerably
between species, ranging from 31 mm Cyttus australis to 1707 mm for a
Mustelus antarcticus (Tables 2; S1). Platycephalus bassensis, Squalus spp,
Notolabrus tetricus, Platycephalus richardsoni and Mustelus antarcticus
were the most measurable species with >80% of individuals at MaxN
being able to be measured (Table 2). Cyttus australis, Dinolestes lewini,
Thamnaconus degeni, Eubalichthys gunnii and Chirodactylus spectabilis
were the least measurable species with <50% of individuals being
measured. Complete summaries of proportions measured, and lengths
are provided in Tables 2 and S1, respectively.
As a rst step to examining length-frequency distributions in indi-
vidual species, the inuence of using differing interval widths to
represent size class distributions was examined. However, when visually
comparing the length-frequency plots for the 10-, 30- and 50-mm size
class intervals, similar size structure distributions were observed for
most of the 22 sh species examined (Fig. 2). For example, all three size
class intervals for the length-frequency plot for commercially and rec-
reational prized Latris lineata suggests a peak in individuals around 470
mm (Fig. 2). Despite similar size structure distributions between size
class intervals, larger intervals exhibited greater MAD values for a given
length measurement sample size for most species (Fig. 3). However,
regardless of size class interval, a consistent asymptotic decrease in MAD
estimates was observed for all species (Fig. 3). While the sh species
exhibited different MAD values across the three size class bins, the
largest decreases in MAD, thus increase in accuracy, occurred within the
rst 50 individuals sampled, with most species reaching an acceptable
reduction in MAD (to 5%) with less than 100 individual length mea-
surements (Fig. 3, Table 2). For all species, the length measurement
sample size required for the upper 95th percentile of the error to be
below 5% MAD, increased with the size class interval used, with a
minimum number of length measurements of 30 to 105 individuals
required for the 10-mm size class bins, 55–130 individuals for 30-mm
size class bins, and 80–135 individuals for 50-mm size class bins (Fig. 3).
Using the minimum number of individuals required to be measured
to achieve a reliable MAD (below 5% and within the 95% percentile),
and accounting for the average proportion of individuals able to be
I. Arachchige Weerarathne et al.
Ecological Indicators 122 (2021) 107262
5
reliably measured per stereo BRUV drop, we estimate that across our 22
focal species, the minimum number of stereo BRUV deployment per
species ranged between 20 for Nemadactylus macropterus to 1569 for
Squalus spp. based on 10-mm length-frequency bins (Table 2). Slightly
more stereo-BRUV deployments were estimated to be required for larger
size class intervals, ranging from 37 to 1961 and 39 to 2118 for the 30-
and 50-mm size class bins, respectively (Table 2).
4. Discussion
This study found the minimum number of individual length mea-
surements required to accurately describe the size-class structure of
mesophotic reef shes from stereo BRUVs varied greatly between spe-
cies, ranging from 30 to 135 individuals. The results also suggested that,
while the size class interval used to generate length-frequency distri-
butions did not appear to overly inuence the overall pattern, slightly
more individual length measurements are required to adequately
describe the size class distribution when using larger intervals. From
these analyses, it is recommended that, generally, stereo BRUV moni-
toring programs should aim to collect measurements from at least 60,
110, or 120 individuals for length-frequency analyses when using 10-,
30-, and 50-mm size class intervals, respectively, to be 95% condent
that samples reected the “real” size class distribution of the species of
interest. This should be, on average, achievable from 430, 680 and 730
stereo-BRUV deployments for length-frequency analyses using 10-, 30-,
and 50-mm size class intervals, respectively. Though, it should be noted
that the exact number of individual measurements changes between
species, with species that exhibit higher encounter probabilities (such as
Notolabrus tetricus and Nemadactylus macropterus) generally required
fewer number of stereo BRUV deployments.
The number of individual length measurements found in our study
agrees with previous research. Anderson and Neumann (1996) recom-
mended that at least 100 individual sh should be measured to describe
a length-frequency distribution to meet sheries management objec-
tives. Similarly, Gilliland (1987) compared the various length
measurement requirements to examine the size class structure of the
freshwater sh, Micropterus salmoides, sampled using electroshing and
concluded that 150 individual measurements were adequate. More
recent work of Kritzer et al. (2001), Vokoun et al. (2001), Miranda
(2007) and Schultz et al. (2016) suggest that between 100 and 1200
individual measurements are required to accurately capture the size-
class structure of a range of different (predominately) freshwater sh
species. However, none of these studies explored how many sample
replicates (i.e., stereo BRUV deployments) would be required to attain
these individual measurements.
For most species, it would take a high sampling effort of ≫200 stereo
BRUV deployments to attain an adequate number of length measure-
ments. In most monitoring programs having to deploy this number of
stereo BRUVs would be an unreasonable expectation due to economic
and logistical costs. It should be noted that our analysis did not account
for heterogeneous encounter probabilities that are common for many
shes due to habitat/depth afnities (Katsanevakis et al., 2012), sam-
pling gear (Langlois et al., 2012; Monk et al., 2012) and size/age classes
utilizing different habitats or protection zones (Ortiz and Tissot, 2008).
In many cases, the absence of an individual, or a particular size, from a
dataset may be reective of imperfect detection due to sampling bias
rather than true absence (Monk, 2014). Realistically, the number of
length measurements and deployments suggested in this study are likely
to be an overestimate and are only applicable to sh populations asso-
ciated with mesophotic temperate reef habitats. More rened estimates
could be achieved by accounting for these heterogeneous encounter
rates associated with habitat, and perhaps researchers may be able to
increase the number of length measurements obtained without the need
to increase (even a decrease in) the number of stereo BRUV deployments
suggested here.
It is also important to note that the accuracy of the length-frequency
distribution used to estimate the number of length measurements cannot
be resolved by the analysis used here. We acknowledge that stereo
BRUVs have been criticized for potential biases associated with limiting
measurements to individuals at MaxN, which may reduce the “tails” in
Table 2
Mean abundance (MaxN), proportion measured, 95th percentile mean absolute difference (MAD 95th) values and estimated number of deployments required for the
focal species based on the pooled 251 stereo BRUV dataset. A complete list of species including summed abundance (MaxN) and length summaries are provided in S1
Table.
Family Scientic Mean MaxN per
drop
Proportion measured at
MaxN
No of sh required for reliable
estimate (MAD 95th)
Estimated no of deployments for
accurate length-frequency data
10 mm
bin
30 mm
bin
50 mm
bin
10 mm
bin
30 mm
bin
50 mm
bin
Cheilodactylidae Nemadactylus
macropterus
3.07 0.65 40 90 100 20 45 51
Chirodactylus spectabilis 0.17 0.48 55 105 115 690 1318 1443
Nemadactylus douglasii 0.10 0.68 50 100 115 738 1476 1698
Cyttidae Cyttus australis 0.52 0.24 65 110 120 526 891 972
Dinolestidae Dinolestes lewini 1.48 0.26 65 115 115 166 295 295
Notolabrus tetricus 0.94 0.92 60 115 135 69 133 156
Latridae Latris lineata 0.75 0.63 45 95 110 94 199 230
Monacanthidae Meuschenia scaber 5.43 0.52 60 105 110 21 37 39
Thamnaconus degeni 4.48 0.40 55 100 100 31 56 56
Meuschenia australis 0.49 0.65 75 120 135 238 381 429
Meuschenia freycineti 0.40 0.60 55 110 120 226 453 494
Acanthaluteres vittiger 0.36 0.62 95 130 110 426 583 493
Eubalichthys gunnii 0.27 0.46 85 125 135 688 1012 1093
Mordiae Morid cods 1.09 0.64 45 100 125 65 144 180
Mullidae Upeneichthys vlamingii 0.21 0.58 90 125 120 729 1012 972
Neosebastidae Neosebastes
scorpaenoides
0.28 0.77 105 120 100 479 548 456
Platycephalidae Platycephalus bassensis 0.09 0.96 85 120 125 970 1369 1426
Platycephalus
richardsoni
0.07 0.89 80 120 135 1255 1883 2118
Scyliorhinidae Cephaloscyllium laticeps 0.58 0.57 35 75 105 107 230 321
Sebastidae Helicolenus percoides 1.86 0.50 60 115 125 65 124 135
Squalidae Squalus spp 0.07 0.94 100 125 130 1569 1961 2039
Triakidae Mustelus antarcticus 0.10 0.80 30 55 80 377 690 1004
I. Arachchige Weerarathne et al.
Ecological Indicators 122 (2021) 107262
6
Fig. 2. Length-frequency distributions showing the effects of different size class intervals.
I. Arachchige Weerarathne et al.
Ecological Indicators 122 (2021) 107262
7
Fig. 3. Mean absolute difference (MAD) in relative size class bins (darker shading lines) and upper 95th percentiles (lighter shaded lines) between subsamples and
original pooled stereo BRUV samples for various subsample sizes for each species. Line format represents different size class intervals (red solid line, 50 mm; orange
dash line, 30 mm; black dotted line, 10 mm). The blue line indicates the 5% MAD threshold. Also, note that only the rst 150 of the possible 500 measurements are
shown for clarity as all species reached 5% MAD by this value. (For interpretation of the references to colour in this gure legend, the reader is referred to the web
version of this article.)
I. Arachchige Weerarathne et al.
Ecological Indicators 122 (2021) 107262
8
the real length-frequency distribution (Harvey et al., 2013; Langlois
et al., 2012). However, we argue that all sh sampling methods are size-
selective and provide a biased representation of the real length distri-
bution. We also did not account for situations when sampling habitats
that do not attract all lengths equally (such as monitoring a species with
known ontogenetic habitat preferences). As such we do caution readers
that our results relate to temperate water mesophotic shes captured
using stereo BRUVs. However, the approach used here provides a tem-
plate that could be used to further rene monitoring programs focusing
on different species in other marine regions and habitats.
While researchers may have differing levels of acceptable accuracy,
our results do highlight that a substantial number of deployments may
be needed in situations when a high degree of accuracy is required (such
as sheries stock assessments). In such circumstances where a high level
of accuracy is required, it is important to consider the ramications of
under-sampling length-frequency distribution. Previous research, such
as Harvey et al. (2002), have illustrated scenarios in which imprecise
sh lengths produce inaccurate biomass estimates and may provide an
insight into the potential effects of under-sampling length-frequency
distributions. As stereo BRUVs continue to be used for length/biomass
related metrics in monitoring programs, it is important to get a better
understanding of these potential effects. While previous studies have
focused on understanding the effort required to accurately sample
abundance or lumped size categories (such as the number of legal-sized
sh) from stereo BRUVs (e.g., Hill et al., 2018; Langlois et al., 2012), our
study helps to inform initial sampling requirements for length mea-
surements for monitoring programs using stereo BRUVs. However, more
research is required to understand the sampling effort required to
accurately capture length-frequency data from stereo BRUVs across a
broader range of habitats and species. The approach used here provides
a means to further rene length/biomass related monitoring programs
and helps to initially inform sample size requirements for length mea-
surements for monitoring using stereo BRUVs for sheries and marine
park monitoring applications, which are increasingly using size-based
indicators as descriptors associated with shing pressure in both
temperate (e.g., Stuart-Smith et al., 2017) and coral reef habitats (e.g.,
Robinson et al., 2017).
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Acknowledgements
This work was supported by FRDC (project number: 2014-012) and
the National Environmental Science Program’s Marine Biodiversity
Hub. The Marine Biodiversity Hub, a collaborative partnership sup-
ported through funding from the Australian Government’s National
Environmental Science Program (NESP). NESP Marine Biodiversity Hub
partners include the University of Tasmania; CSIRO, Geoscience
Australia, Australian Institute of Marine Science, Museum Victoria,
Charles Darwin University, The University of Western Australia, Inte-
grated Marine Observing System, NSW Ofce of Environment and
Heritage, and NSW Department of Primary Industries. Justin Hulls,
Bastien Taormina, Lainey James and Richard Zavalas for participating in
eldwork and assisting in stereo BRUV annotations. Nicole Hill and
Jeremy Lyle are thanked for their input in the coordination of the FRDC
project. Author contributions: IAW- Analysis, Writing (Original draft
preparation). JM - Conceptualization, Methodology, Data curation,
Analysis, Supervision, Writing (Reviewing and Editing). NB - Method-
ology, Funding acquisition, Supervision, Writing (Reviewing and
Editing).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.ecolind.2020.107262.
References
Anderson, R.O., Neumann, R.M., 1996. Length, weight, and associated structural indices.
In: Murphy, B.R., Willis, D.W. (Eds.), Fisheries Techniques, second ed. American
Fisheries Society, Bethesda, Maryland.
Audzijonyte, A., Fulton, E., Haddon, M., Helidoniotis, F., Hobday, A.J., Kuparinen, A.,
Morrongiello, J., Smith, A.D.M., Upston, J., Waples, R.S., 2016. Trends and
management implications of human-inuenced life-history changes in marine
ectotherms. Fish Fish. 17, 1005–1028.
Barnett, L.A.K., Branch, T.A., Ranasinghe, R.A., Essington, T.E., 2017. Old-growth shes
become scarce under shing. Curr. Biol. 27, 2843–2848.
Berkeley, S.A., Hixon, M.A., Larson, R.J., Love, M.S., 2004. Fisheries sustainability via
protection of age structure and spatial distribution of sh populations. Fisheries 29,
23–32.
Cappo, M., De’ath, G., Stowar, M., Johansson, C., Doherty, P., 2009. The inuence of
zoning (closure to shing) on sh communities of the deep shoals and reef bases of
the southern Great Barrier Reef Marine Park. Reef and Rainforest Research Centre
Limited, Cairns, Australia.
Collins, D.L., Langlois, T.J., Bond, T., Holmes, T.H., Harvey, E.S., Fisher, R., McLean, D.
L., 2017. A novel stereo-video method to investigate sh–habitat relationships.
Methods Ecol. Evol. 8, 116–125.
Cresswell, A.K., Langlois, T.J., Wilson, S.K., Claudet, J., Thomson, D.P., Renton, M.,
Fulton, C.J., Fisher, R., Vanderklift, M.A., Babcock, R.C., Stuart-Smith, R.D.,
Haywood, M.D.E., Depczynski, M., Westera, M., Ayling, A.M., Fitzpatrick, B.,
Halford, A.R., McLean, D.L., Pillans, R.D., Cheal, A.J., Tinkler, P., Edgar, G.J.,
Graham, N.A.J., Harvey, E.S., Holmes, T.H., 2019. Disentangling the response of
shes to recreational shing over 30 years within a fringing coral reef reserve
network. Biol. Conserv. 237, 514–524.
Ellis, D.M., Demartini, E.E., 1995. Evaluation of a video camera technique for indexing
abundances of juvenile pink snapper, Pristipomoides lamentosus, and other Hawaiian
insular shelf shes. Fish. Bull. 93, 67–77.
Fitzpatrick, C., McLean, D., Harvey, E.S., 2013. Using articial illumination to survey
nocturnal reef sh. Fish. Res. 146, 41–50.
Gilliland, E., 1987. Evaluation of Oklahoma’s standardized electroshing in calculating
population structure indices, in: Proceedings of the Annual Conference Southeastern
Association of Fish and Wildlife Agencies, pp. 277–287.
Harvey, E.S., Mclean, D.L., Frusher, S., Haywood, M.D.D., Newman, S.J., Williams, A.,
2013. The use of BRUVs as a tool for assessing marine sheries and ecosystems: A
review of the hurdles and potential. Report to Fisheries Research and Development
Corporation (Project No 2010/002). ISBN: 978-1-74052-265-6.
Harvey, E., Fletcher, D., Shortis, M., 2002. Estimation of reef sh length by divers and by
stereo-video: a rst comparison of the accuracy and precision in the eld on living
sh under operational conditions. Fish. Res. 57, 255–265.
Heyns-Veale, E.R., Bernard, A.T.F., Richoux, N.B., Parker, D., Langlois, T.J., Harvey, E.S.,
G¨
otz, A., 2016. Depth and habitat determine assemblage structure of South Africa’s
warm-temperate reef sh. Mar. Biol. 163, 1–17.
Hill, N.A., Barrett, N., Ford, J.H., Peel, D., Foster, S., Lawrence, E., Monk, J., Althaus, F.,
Hayes, K.R., 2018. Developing indicators and a baseline for monitoring demersal sh
in data-poor, offshore marine parks using probabilistic sampling. Ecol. Ind. 89,
610–621.
Hixon, M.A., Johnson, D.W., Sogard, S.M., 2014. BOFFFFs: on the importance of
conserving old-growth age structure in shery populations. ICES J. Mar. Sci. 71,
2171–2185.
Jenkins, T.M., Diehl, S., Kratz, K.W., Cooper, S.D., 1999. Effects of population density on
individual growth of brown trout in streams. Ecology 80, 941–956.
Katsanevakis, S., Weber, A., Pipitone, C., Leopold, M., Cronin, M., Scheidat, M., Doyle, T.
K., Buhl-Mortensen, L., Buhl-Mortensen, P., D’Anna, G., de Boois, I., Dalpadado, P.,
Damalas, D., Fiorentino, F., Garofalo, G., Giacalone, V.M., Hawley, K.L., Issaris, Y.,
Jansen, J., Knight, C.M., Knittweis, L., Kroencke, I., Mirto, S., Muxika, I., Reiss, H.,
Skjoldal, H.R., Voege, S., 2012. Monitoring marine populations and communities:
methods dealing with imperfect detectability. Aquatic Biol. 16, 31–52.
Kritzer, J.P., Davies, C.R., Mapstone, B.D., 2001. Characterizing sh populations: effects
of sample size and population structure on the precision of demographic parameter
estimates. Can. J. Fish. Aquat. Sci. 58, 1557–1568.
Langlois, T., Williams, J., Monk, J., Bouchet, P., Currey, L., Goetze, J., Harasti, D.,
Huveneers, C., Ierodiaconou, D., Malcolm, H., Whitmarsh, S., 2018. Marine sampling
eld manual for benthic stereo BRUVs (Baited Remote Underwater Videos). in: Field
Manuals for Marine Sampling to Monitor Australian Waters, Przeslawski R, Foster S
(Eds). Report to National Environmental Science Programme., pp. 82-104.
Langlois, T.J., Fitzpatrick, B.R., Fairclough, D.V., Wakeeld, C.B., Hesp, A., McLean, D.L.,
Harvey, E.S., Meeuwig, J.J., 2012. Similarities between line shing and baited
stereo-video estimations of length-frequency: novel application of kernel density
estimates. PLoS One 7, e45973.
Langlois, T., Goetze, J., Bond, T., Monk, J., Abesamis, R.A., Asher, J., Barrett, N.,
Bernard, A.T.F., Bouchet, P.J., Birt, M.J., Cappo, M., Currey-Randall, L.M.,
Driessen, D., Fairclough, D.V., Fullwood, L.A.F., Gibbons, B.A., Harasti, D.,
Heupel, M.R., Hicks, J., Holmes, T.H., Huveneers, C., Ierodiaconou, D., Jordan, A.,
Knott, N.A., Lindeld, S., Malcolm, H.A., McLean, D., Meekan, M., Miller, D.,
Mitchell, P.J., Newman, S.J., Radford, B., Rolim, F.A., Saunders, B.J., Stowar, M.,
I. Arachchige Weerarathne et al.
Ecological Indicators 122 (2021) 107262
9
Smith, A.N.H., Travers, M.J., Wakeeld, C.B., Whitmarsh, S.K., Williams, J.,
Harvey, E.S., 2020. A eld and video annotation guide for baited remote underwater
stereo-video surveys of demersal sh assemblages. Methods Ecol. Evol. 11,
1401–1409.
Lyle, J.M., Hill, N., Barrett, N.S., Lucieer, V., Thomson, R., Hulls, J., Ewing, G.P., 2017.
Tasmania’s coastal reefs: deep reef habitats and signicance for nsh production
and biodiversity. Report to Fisheries Research and Development Corporation
(Project No. 2014/012). Institute for Marine and Antarctic Studies, University of
Tasmania, Hobart, p. 102.
Mallet, D., Pelletier, D., 2014. Underwater video techniques for observing coastal marine
biodiversity: a review of sixty years of publications (1952–2012). Fish. Res. 154,
44–62.
Miranda, L.E., 2007. Approximate sample sizes required to estimate length distributions.
Trans. Am. Fish. Soc. 136, 409–415.
Mitchell, P., Bolam, S., Close, H., Garcia, C., Monk, J., Alliji, K., accepted. Distribution of
demersal sh assemblages along the west coast of St. Lucia: implications for planning
no-take marine reserves. Aquatic Conservation: Freshwater and Marine Ecosystems.
Mitchell, J.D., McLean, D.L., Collin, S.P., Langlois, T.J., 2019. Shark depredation and
behavioural interactions with shing gear in a recreational shery in Western
Australia. Mar. Ecol. Prog. Ser. 616, 107–122.
Monk, J., 2014. How long should we ignore imperfect detection of species in the marine
environment when modelling their distribution? Fish Fish. 15, 352–358.
Monk, J., Ierodiaconou, D., Harvey, E., Rattray, A., Versace, V.L., 2012. Are we
predicting the actual or apparent distribution of temperate marine shes? PLoS One
7, e34558.
Monk, J., Barrett, N.S., Hulls, J., James, L., Hosack, G.R., Oh, E., Martin, T., Edwards, S.,
Nau, A., Heaney, B., Foster, S.D., 2016. Seaoor Biota, Rock Lobster and Demersal
Fish Assemblages of the Tasman Fracture Commonwealth Marine Reserve Region:
Determining the Inuence of the Shelf Sanctuary Zone on Population Demographics.
The University of Tasmania, p. 97. Report to National Environmental Research
Programme.
Neumann, R.M., Allen, M.S., 2007. Size structure, in: Guy C.S., M.L., B. (Eds.), Analysis
and interpretation of freshwater sheries data. American Fisheries Society, Bethesda,
Maryland, USA, pp. 375-421.
Ortiz, D.M., Tissot, B.N., 2008. Ontogenetic patterns of habitat use by reef-sh in a
marine protected area network: a multi-scaled remote sensing and in situ approach.
Mar. Ecol. Prog. Ser. 365, 217–232.
Pitcher, C.R., Williams, A., Ellis, N., Althaus, F., Mcleod, I., Bustamante, R., Kenyon, R.,
Fuller, M., 2016. Implications of current spatial management measures for AFMA
eras for habitats. Report to Fisheries Research and Development Corporation
(Project No 2014/204). CSIRO Oceans & Atmosphere, Brisbane, QLD, p. 50.
Przeslawski, R., Foster, S., Monk, J., Langlois, T.J., Lucieer, V., Stuart-Smith, R., 2018.
Comparative assessment of seaoor sampling platforms. Report to National
Environmental Science Programme. Geoscience Australia., p. 57.
Przeslawski, R., Foster, S., Monk, J., Barrett, N., Bouchet, P., Carroll, A., Langlois, T.,
Lucieer, V., Williams, J., Bax, N., 2019. A suite of eld manuals for marine sampling
to monitor Australian waters. Front. Marine Sci. 6.
Queiros, A.M., Fernandes, J., Genevier, L., Lynam, C.P., 2018. Climate change alters sh
community size-structure, requiring adaptive policy targets. Fish Fish. 19, 613–621.
Robinson, J.P.W., Williams, I.D., Edwards, A.M., McPherson, J., Yeager, L., Vigliola, L.,
Brainard, R.E., Baum, J.K., 2017. Fishing degrades size structure of coral reef sh
communities. Glob. Change Biol. 23, 1009–1022.
Schultz, L.D., Mayeld, M.P., Whitlock, S.L., 2016. Sample sizes needed to describe
length-frequency of small-bodied shes: an example using larval pacic lamprey.
J. Fish Wildlife Manage. 7, 315–322.
Sheridan, J.A., Bickford, D., 2011. Shrinking body size as an ecological response to
climate change. Nat. Clim. Change 1, 401–406.
Stuart-Smith, R.D., Edgar, G.J., Barrett, N.S., Bates, A.E., Baker, S.C., Bax, N.J.,
Becerro, M.A., Berkhout, J., Blanchard, J.L., Brock, D.J., Clark, G.F., Cooper, A.T.,
Davis, T.R., Day, P.B., Duffy, J.E., Holmes, T.H., Howe, S.A., Jordan, A.,
Kininmonth, S., Knott, N.A., Lefcheck, J.S., Ling, S.D., Parr, A., Strain, E.,
Sweatman, H., Thomson, R., 2017. Assessing national biodiversity trends for rocky
and coral reefs through the integration of citizen science and scientic monitoring
programs. Bioscience 67, 134–146.
van Overzee, H.M.J., Rijnsdorp, A.D., 2015. Effects of shing during the spawning
period: implications for sustainable management. Rev. Fish Biol. Fish. 25, 65–83.
Vokoun, J.C., Rabeni, C.F., Stanovick, J.S., 2001. Sample-size requirements for
evaluating population size structure. North Am. J. Fish. Manage. 21, 660–665.
Wellington, C.M., Harvey, E.S., Wakeeld, C.B., Langlois, T.J., Williams, A., White, W.T.,
Newman, S.J., 2018. Peak in biomass driven by larger-bodied meso-predators in
demersal sh communities between shelf and slope habitats at the head of a
submarine canyon in the south-eastern Indian Ocean. Cont. Shelf. Res. 167, 55–64.
Whitmarsh, S.K., Fairweather, P.G., Huveneers, C., 2017. What is Big BRUVver up to?
Methods and uses of baited underwater video. Rev. Fish Biol. Fish. 27, 53–73.
Williams, A., Althaus, F., Smith, A.D.M., Daley, R., Barker, B.A., Fuller, M., 2013.
Developing and Applying a Spatially-Based Seascape Analysis (the “habitat proxy”
method) to Inform Management of Gulper Sharks: Compendium of CSIRO Discussion
Papers. CSIRO Marine and Atmospheric Research, Hobart, Tasmania, p. 220.
I. Arachchige Weerarathne et al.