Quantifying Fish Assemblages in South Australian Marine Protected Areas
Supervisor: Bronwyn Gillanders
School of Biological Sciences The University of Adelaide
Submitted in partial fulfilment of the requirements for the degree of Bachelor of Science (Honours)
I hereby declare that this work contains no material which has been accepted for the award of any other degree or diploma in any university or other tertiary institution and, to the best of my knowledge and belief, contains no material previously published or written by another person, except where due reference is made in the text.
Nina Wootton October 2017
Declaration ... 2
Abstract ... 4
INTRODUCTION ... 5
MATERIALS AND METHODS ... 8
Study sites ... 8
Baited Remote Underwater Video ... 10
Data analysis ... 12
Statistical analysis ... 12
RESULTS ... 14
Abundance and diversity analysis ... 15
Assemblage analysis ... 20
Length analysis ... 24
DISCUSSION ... 27
Effectiveness of sanctuary zones ... 27
Marine protected area design ... 28
Environmental factors influencing fish assemblages ... 30
Spatial variation influencing fish assemblages ... 32
Length of fish in sanctuary zones ... 33
Limitations in the present study ... 33
Future directions ... 35
CONCLUSION ... 36
Acknowledgements ... 38
References ... 39
Appendices ... 49
Networks of Marine Protected Areas (MPAs) and no-take sanctuary zones are increasingly being developed as a means of conserving biodiversity and protecting the ocean. This study examined the abundance, species richness and lengths of fish within a network of MPAs in South Australia’s west coast embayments within the first three years of development. Fish assemblages in the embayments were also examined to evaluate how biological and
environmental conditions influenced them. Baited Remote Underwater Video (BRUV) were used to quantify fish assemblages in a range of depths and habitats over two seasons during 2017. Total fish abundance was 76% higher inside sanctuary zones compared to outside in both Coffin Bay and Venus Bay, although not all species, seasons and sanctuary zones performed equally. The key differences occurred between fish assemblages in different seasons (January and June) and in different sanctuary zones. Ideally, MPAs should incorporate a range of different habitats over a wide depth range. This may require larger sanctuary zones, or researching locations of potential MPAs prior to their establishment, to ensure that the protected regions cover a variety of habitats and depths.
Human actions, including overfishing, pollution and maritime industries, have steadily
decreased the healthy functioning of marine and estuarine habitats (Lotze et al. 2006; Crain et al. 2009). Due to increased awareness and global concern for marine ecosystem health, there has been an escalation in the development of Marine Protected Areas (MPAs) in the last decade (Halpern et al. 2010). Of these protected areas, a small amount are no-take zones which prohibit the extraction of marine resources in entirety (Gaines et al. 2010). Various published studies have researched and evaluated the response of a range of marine organisms and ecosystems to protection (Lester et al. 2009). These include researching the type of species that respond best to protection (Mosquera et al. 2000; Barrett et al. 2007), their impact on areas surrounding the protected zone (Russ and Alcala 2011; Harrison et al.
2012b), their influence on invasive species (Burfeind et al. 2013), the ideal setup of protected reserves (Claudet et al. 2008; Edgar et al. 2014), and the amount of enforcement needed for changes of a significant level to occur (Byers and Noonburg 2007; Guidetti et al. 2008).
Marine Protected Areas are important biodiversity conservation tools which protect natural and cultural resources. Quantitative information on fish assemblages in marine protected areas is required to assess change through time or to compare no take areas to areas where fishing is permitted. Such information can be used to evaluate the effectiveness of MPAs and gain community support. Limited data exists analysing fish assemblages in MPAs in South Australian waters, yet such data are critical to evaluate their effectiveness.
Worldwide less than 4% of the ocean is protected, with Australian waters accounting for 65%
of the total global MPAs by area (Jenkins and Van Houtan 2016). Numerous studies conclude that protected areas contain higher fish diversity and abundance, with greater average size and general higher biodiversity than areas which are unprotected (Kelaher et al. 2014; Soler et al. 2015). Furthermore, there is strong evidence that there are community-wide changes surrounding MPAs, with increases in marine biota, improved larval export and recruitment benefits (Hilborn et al. 2004; Guidetti 2007; Harrison et al. 2012a). However, studies also suggest that MPAs may need to be in place for some years before these benefits are observed (Claudet et al. 2008; Edgar and Barrett 2012; Kelaher et al. 2014).
Marine Protected Areas vary in shape, size and location, and there has been limited
investigation of how such variables may affect conservation goals (Edgar and Barrett 1999).
The identification of significant scales of natural variability of fish species provides an important baseline for ecologically relevant choices on MPA design and evaluation of their effectiveness (Charton et al. 2002). These data are not available for a large number of marine species, especially at small spatial scales. The majority of studies focus either on fish
assemblages over larger ranges (Parsons et al. 2016), or on individual species and trophic groups at small spatial scales (Gillanders 1997). Large spatial fish assemblage patterns are still useful for the broad-scale zoning and planning of MPAs, however are not effective in providing the resolution needed for zoning design decisions on MPAs at a local scale (Malcolm et al. 2007). Therefore, patterns of fish assemblage at small spatial scales are useful in identifying locations that need protection and can provide data to help implement this.
Spatial-temporal variability of habitats, and both the physical and biological attributes within these habitats (e.g. temperature and depth), are important influencers on the distribution of fish species locally (Choat and Ayling 1987; Kingsford 1989; Holbrook et al. 1994). The fish assemblage will vary depending on the habitat type within the environment. The type of habitat, such as seagrass, reef, macro algae or sand, are important factors determining whether areas are inhabited and therefore affect abundance. Seagrasses and mangroves in particular are often associated with juvenile nurseries (Beck et al. 2001; Heck et al. 2003;
Dorenbosch et al. 2004), so it would be expected that fish size and composition would change in these habitats. Further physical factors, such as depth and temperature, can cause variations in fish assemblages within these habitats (Rooker and Dennis 1991; Parsons et al.
2016; Fitzgerald et al. 2017). Knowledge of the relationships between fish assemblages and habitat variability within small scale biogeographical regions would provide a more solid outline for designing MPAs.
South Australia has extensive areas of MPAs, under both Commonwealth and State
legislation (Barr and Possingham 2013; Kirkman and Shepherd 2015). The state-controlled areas are legislated under the South Australian Marine Parks Act 2007 and cover areas extending to three nautical miles from the coast (Barr and Possingham 2013). These state MPAs were developed in 2009 as part of a network of 19 marine parks with the ultimate aim
of conserving the unique marine life throughout the eight bioregions of South Australia (Kirkman 2013; Kirkman and Shepherd 2015). Within the MPAs there are zones with
different levels of protection; these include restricted access, sanctuary and habitat protection zones (Lynch 2006; Kirkman 2013). Sanctuary zones are considered areas of high
conservation value where no fishing or other disruptive activities are permitted, however enforcement of these areas did not commence until October 2014 (Scholz et al. 2017). Under the current legislation, Marine Parks Act 2007, the monitoring, evaluation and reporting of MPAs is required every 10 years (Scholz et al. 2017).
This study aims to assess MPA effectiveness by using Baited Remote Underwater Video (BRUV) to collect baseline fish assemblage data in South Australian west coast embayments.
It specifically investigates if there are any significant differences in fish abundance, species richness and size inside and outside the sanctuary zones within the MPAs. Furthermore, it investigates patterns of fish assemblages at a small spatial scale, for example metres to kilometres, and looks at assemblage variation among habitats, and at which biological and physical factors are potentially contributing to these patterns.
It can be hypothesised that with time, there will be significant differences in fish abundance, species richness and size inside and outside MPAs. However, South Australian zoning was only implemented in October 2014. Therefore, differences in abundance, richness and size may not necessarily be found, as a global meta-analysis suggested that MPAs needed to be implemented for 10 years to be effective (Edgar et al. 2014). Despite this, the project provides useful baseline information upon which to detect change in the future. BRUV data may contribute to determining the success of MPAs within areas of South Australia.
MATERIALS AND METHODS
Nineteen multiple use marine parks were established in South Australia by the South Australian Government in 2012. This includes the West Coast Bays Marine Park including Venus Bay (Figure 1) and the Thorny Passage Marine Park including Coffin Bay (Figure 2).
Within these marine parks are a number of sanctuary zones where fishing of any type is prohibited.
Figure 1. MPAs within the West Coast Bays Marine Park in South Australia, including the Venus Bay Sanctuary Zone 8 where sampling was completed. Note that Sanctuary Zone 9 does not include any subtidal habitat. Light blue area corresponds to MPAs while dark blue areas are sanctuary zones. Black GPS marks illustrate locations of BRUV deployments.
Figure 2. MPAs in the Thorny Passage Marine Park in South Australia, including Coffin Bay sanctuary zones where sampling was completed. This includes Sanctuary Zones 1, 3, 4 and 5.
Sanctuary Zones 2 and 6 were excluded from the study due to being too small and too shallow for a boat to enter for BRUV sampling. Light blue areas correspond with MPAs while the dark blue areas are sanctuary zones. Black GPS marks illustrate locations of BRUV deployments.
In total, five sanctuary zones were sampled using BRUV cameras, including one in Venus Bay (Sanctuary Zone 8) and four in Coffin Bay (Sanctuary Zones 1, 3, 4 and 5). The number of BRUV deployments and replicates varied slightly between each sanctuary zone due to logistical and video recording issues. However, with the exception of Sanctuary Zone 4, there were at least two deployments inside each sanctuary zone and a matching replicate based on depth and habitat outside each sanctuary zone (Table 1). Sampling occurred in January and June 2017.
Table 1. Summary of information relating to sampling sites for January and June sampling.
Sampling information includes: sanctuary zone, whether sampling was inside (I) or outside (O) the sanctuary zone and number of BRUV deployments.
Sanctuary zone Inside/Outside Number of BRUV deployments January
Coffin Bay 1 I 2
Coffin Bay 1 O 1
Coffin Bay 3 I 2
Coffin Bay 3 O 2
Coffin Bay 4 I 0
Coffin Bay 4 O 2
Venus Bay 8 I 2
Venus Bay 8 O 3
Coffin Bay 1 I 4
Coffin Bay 1 O 4
Coffin Bay 3 I 2
Coffin Bay 3 O 2
Coffin Bay 4 I 1
Coffin Bay 4 O 2
Coffin Bay 5 I 2
Coffin Bay 5 O 2
Venus Bay 8 I 4
Venus Bay 8 O 2
BAITED REMOTE UNDERWATER VIDEO
Stereo BRUVs were used to record fish assemblages, with two cameras mounted on each metal camera stand (Figure 3). The BRUV data were collected using GoPro Hero4
underwater digital video cameras. The cameras were placed in waterproof housings which were mounted on a metal frame. The housings sat above the substratum and faced horizontal and parallel to the ocean floor. A tube of polymerizing vinyl chloride (PVC) extended from the centre of the metal frame, with a mesh bait bag attached. The bait bag contained 500g of minced pilchards (Sardinops sagax) and sat 1200mm from the camera lens. The cameras faced forward and slightly inwards towards the bait bag. The units were deployed from a boat, with care taken to drop them into a similar depth and habitat inside and outside of the sanctuary zone. Following deployment, the boat was moved approximately 200 metres away
from the area before the next unit was dropped. This minimised the overlap of bait plumes, ensuring that double counts of fish did not occur. The cameras were left to film underwater for 60 minutes before being collected and deployed again. Four BRUV units were deployed concurrently, and following deployment the boat left the area to avoid any disturbances to fish activity. Procedures were similar to other BRUV research (Watson et al. 2005; Harvey et al. 2007; Watson et al. 2010), so that the data were comparable to other studies. Temperature was recorded using HOBO temperature loggers, which were attached to each BRUV unit.
Figure 3. BRUV unit. Frame constructed of aluminium; bait arm was constructed of PVC pope and bait pouch gutter mesh; ropes were connected to a metal ring on one end and the other end was connected to a buoy which allowed the unit to be deployed and retrieved remotely; GoPro video cameras were mounted inside the underwater housings.
Videos were downloaded and analysed using SeaGIS’ EventMeasure, an event logging and 3D measuring software package designed specifically for biological information and animal behaviour in underwater movie sequences (www.seagis.com.au). For each species on each video, the maximum number of individuals observed in a single frame (maxN) was recorded (see Appendix A). MaxN is a conservative measure of relative density, which avoids
recounting of particular individuals which may revisit the bait. From each maxN frame all fish species were measured using the 3D length measurement tool. The average fish size of each species was then calculated to determine differences in fish size between sites. Species richness was recorded by identifying the total number of fish species between sites. Species were identified using online and hard copy resources (Hutchins and Swainston 1986; Gomon et al. 2008). The settings used on EventMeasure were standard practices of the Department of Environment Water and Natural Resources (DEWNR) (see Appendix B). Depth was recorded from the vessel’s depth sounder, soak time was standardised at 60 minutes, and the field of view, bias and visability were calculated following DEWNR’s standard practices (see Appendix B). Habitat was categorised as: seagrass, macro algae, sand, sponge, or broken sand. Broken sand was defined as any habitat which contained around half sand and half seagrass or seaweed. Species of questionable identification were flagged and viewed again, with the assistance of colleagues where necessary. Footage was viewed by a single observer to avoid any variation between viewers.
ABUNDANCE AND DIVERSITY ANALYSIS
Using PRIMER software Version 6 (http://www.primer-e.com/), abundance and species richness data were square root transformed and fitted to a Bray-Curtis dissimilarity
resemblance matrix (Bray and Curtis 1957). The square root transformation was applied to avoid dominance of common species and allow contribution from the rarer species. Sites (Coffin Bay and Venus Bay) were analysed both individually (e.g. inside sanctuary zone versus outside sanctuary zone) and combined. Analyses used single factor (inside versus outside sanctuary zone) permutational univariate analyses of variance (ANOVA). For all tests 9999 unrestricted permutations and Monte Carlo simulations were performed. Similar
analyses were also undertaken on dominant species and genera.
Fish assemblage data from Venus Bay, Coffin Bay and both bays combined were also square root transformed and fitted into a Bray-Curtis dissimilarity resemblance matrix. Venus Bay, Coffin Bay and combined location fish assemblage data were analysed for differences between inside and outside sanctuary zones using single factor permutational multivariate analysis of variance (MANOVA). Additional analyses also investigated differences between sanctuary zone location, depths, season and habitats. For all tests, 9999 unrestricted
permutations and Monte Carlo simulations were performed.
Post hoc pairwise tests were conducted to determine where significant differences occurred.
For the fish assemblage data, Multi-Dimensional Scaling (MDS) was used to assess both dissimilarity between deployments and identify factors causing these dissimilarities (Clarke 1993). Vectors of species were overlaid on each plot to show the species that contributed to the dissimilarities between sites. These were identified using between group similarities (SIMPER) (Clarke 1993). The stress value was used as an indication of how well the similarity matrix was represented by the non-metric multidimensional scaling plot, where stress levels are closest to zero when the data are perfectly represented (Clarke 1993).
Dominant species and genera were used to compare length frequency differences between inside and outside sanctuary zones. A Kolmogorov-Smirnov test was run on the frequency data to test if the two distributions differed, and hence quantify their statistical strength.
In total, 498 individuals from 17 species of finfish were identified from 39 BRUV deployments in 2017 (Table 2-3). The most commonly occurring species were Arripis georgianus, A. truttaceus and Sillaginodes punctata (Table 3). During January, 299 individuals from 14 species were identified in 14 BRUV deployments. The June sampling found 199 individuals from 11 species in 25 BRUV deployments (see Appendix C).
Temperature differences occurred between these two periods with the average temperature in January being 23.04C and the average temperature in June being 13.42C.
Table 2. Summary of information relating to sampling sites and environmental data for January and June sampling. Sampling information includes: sanctuary zone, whether sampling was inside (I) or outside (O) the sanctuary zone, number of BRUV deployments, average depth, and average temperature and dominant habitat.
Number of BRUV deployments
Average depth (m)
Average temp (ºC)
Coffin Bay 1 I 2 4.30 21.97 Seaweed
Coffin Bay 1 O 1 6.30 22.63 Broken sand
Coffin Bay 3 I 2 1.45 23.45 Seagrass
Coffin Bay 3 O 2 2.30 23.68 Seagrass
Coffin Bay 4 I 0 - - -
Coffin Bay 4 O 2 4.60 23.14 Seagrass, sand
Venus Bay 8 I 2 1.95 22.72 Seagrass
Venus Bay 8 O 3 1.87 23.34 Seagrass
Coffin Bay 1 I 4 3.25 13.41 Seagrass, sand, sponge
Coffin Bay 1 O 4 4.60 13.38 Seagrass, sand
Coffin Bay 3 I 2 1.60 12.76 Seagrass
Coffin Bay 3 O 2 2.25 12.87 Seagrass
Coffin Bay 4 I 1 5.00 13.32 Sand
Coffin Bay 4 O 2 5.40 13.74 Sand, broken sand
Coffin Bay 5 I 2 1.60 13.20 Seagrass
Coffin Bay 5 O 2 2.50 13.42 Seagrass
Venus Bay 8 I 4 2.15 13.97 Seagrass
Venus Bay 8 O 2 1.75 13.61 Seagrass
Table 3. Summary of fish species identified from BRUV footage analysed. The number of deployments the species were found in is also indicated for each bay. Species codes indicated are used in graphical results.
Species name Code Number of deployments species were found in VENUS BAY
COFFIN BAY (N=28)
Arripis georgianus Arrgeo 6 7
Arripis truttaceus Arrtrut 6 14
Pseudocaranx wrightii Pseuwri 5 1
Sillaginodes punctata Sillpun 9 8
Pelates octolineatus Peloct 3 3
Sillago schomburgkii Sillscho 1 2
Sphyraena forsteri Sphfor 1 0
Mustelus antarcticus Musant 1 0
Sphyraena novaehollandiae Sphnov 1 0
Platycephalus spp. Plat 0 11
Acanthaluteres vittiger Acanvit 0 4
Upeneichthys vlamingii Upenvla 0 1
Trachurus novaezelandiae Tracnov 0 1
Genypterus tigerinus Genytig 0 1
Atule mate Atumat 0 1
Notolabrus parilus Notopar 0 2
Nelusetta ayraudi Nelayr 0 1
ABUNDANCE AND DIVERSITY ANALYSIS
I. VARIATION BETWEEN INSIDE AND OUTSIDE SANCTUARY ZONES FOR LOCATIONS
Total abundance was slightly higher inside sanctuary zones compared to outside for both Coffin Bay and Venus Bay locations, although no significant differences were found (P=0.842, P=0.904) (Figure 4, Table 4). The difference between fish abundance inside and outside sanctuary zones during January sampling was higher than during June sampling (Figure 5, Table 4). This difference was particularly evident in the Coffin Bay January abundance (p=0.056) (Table 4). There were no significant differences for other response variables (species richness, seasonal abundance) (Figure 4-5, Table 4-5).
0 5 10 15 20
0 1 2 3 4 5
0 0.5 1 1.5 2 2.5 3
0 10 20 30 40 50 60 70
Outside June SEASONAL ABUNDANCE
0 5 10 15 20 25 30 35
Outside June SEASONAL ABUNDANCE 0
10 20 30 40
Figure 4. Mean (± SE) total maxN and species richness of fish assemblages in Coffin Bay and Venus Bay inside and outside sanctuary zones.
Figure 5. Mean (± SE) total maxN for fish assemblages inside and outside sanctuary zones in Coffin Bay and Venus Bay in January and June 2017.
Mean (SE) total maxNMean (SE) number of species
VENUS BAY COFFIN BAY
Mean (SE) total maxN
VENUS BAY COFFIN BAY
Table 4. Single-factor permutational ANOVA results comparing fish abundance inside and outside sanctuary zones in Coffin and Venus Bays.
SITE MODEL Df MS F P
Total abundance (both locations) Inside/Outside zone 1 27.285 1.040 0.845
Residual 37 354.980
Coffin Bay total abundance Inside/Outside zone 1 17.482 0.720 0.842
Residual 26 328.710
Venus Bay total abundance Inside/Outside zone 1 7.241 0.194 0.904
Residual 9 504.180
Coffin Bay January abundance Inside/Outside zone 1 2043.4 3.883 0.056
Residual 7 526.27
Venus Bay January abundance Inside/Outside zone 1 163.190 2.046 0.292
Residual 3 79.770
Coffin Bay June abundance Inside/Outside zone 1 237.120 0.726 0.366
Residual 17 326.410
Venus Bay June abundance Inside/Outside zone 1 133.780 0.193 0.862
Residual 4 692.430
Table 5. Single-factor permutational ANOVA results comparing number of species inside and outside sanctuary zones in Coffin and Venus Bays.
SITE MODEL Df MS F P
Total species richness Inside/Outside zone 1 25.675 0.119 0.733
Residual 37 215.330
Coffin Bay total species richness Inside/Outside zone 1 45.289 0.229 0.657
Residual 26 197.640
Venus Bay total species richness Inside/Outside zone 1 1.085 0.174 1.000
Residual 9 309.830
Coffin Bay January species richness Inside/Outside zone 1 58.630 0.445 0.892
Residual 7 192.330
Venus Bay January species richness Inside/Outside zone 1 30.497 2.064 0.614
Residual 3 14.775
Coffin Bay June species richness Inside/Outside zone 1 188.930 0.913 0.280
Residual 17 206.890
Venus Bay June species richness Inside/Outside zone 1 105.870 0.244 0.863
Residual 4 433.720
Mean (SE) total maxN
0 0.5 1 1.5 2 2.5 3
0 1 2 3 4 5
0 1 2 3 4 5
A. georgianus & A. truttaceus
0 5 10 15 20
P. wrightii 0
2 4 6 8
A. georgianus & A. truttaceus
II. VARIATION BETWEEN INSIDE AND OUTSIDE SANCTUARY ZONES FOR SPECIES
Certain dominant species and genus groups showed trends of having higher abundances inside sanctuary zones, although none of the differences were significant (Figure 6, Table 6).
Both A. georgianus and A. truttaceus had higher abundances inside sanctuary zones at both locations (Figure 6), however there was considerable variability both inside and outside sanctuary areas. Similarly, P. wrightii and S. punctata both had higher abundances inside the sanctuary zone in Venus Bay (Figure 6). The only dominant genus that had higher abundance outside the sanctuary zone was Platycephalus spp. in Coffin Bay (Figure 6).
Figure 6. Mean (± SE) total maxN of numerically-dominant genera groups and species inside and outside sanctuary zones in Venus Bay and Coffin Bay. Only highly abundant species are included.
Mean (SE) total maxNMean (SE) total maxN
Table 6. Single-factor permutational ANOVA results comparing fish abundances inside and outside sanctuary zones in Coffin Bay and Venus Bay for numerically-dominant genera and species.
SITE MODEL Df MS F P
Venus bay A. truttaceus and A.
Inside/outside zone 1 321.19 0.2956 0.6093
Residual 9 1086.7
Coffin bay A. truttaceus and A.
Inside/outside zone 1 319.79 0.4048 0.5657
Residual 26 789.88
Venus bay P. wrightii Inside/outside zone 1 306.6 0.2946 0.7079
Residual 9 1048.8
Coffin bay Platycephalus spp. Inside/outside zone 1 32.489 0.7887 0.9116
Residual 26 557.52
Venus bay S. punctata Inside/outside zone 1 56.962 0.1322 0.7263
Residual 9 430.85
I. ASSEMBLAGE VARIATION BETWEEN INSIDE AND OUTSIDE SANCTUARY ZONES
The structure of fish assemblages between inside and outside sanctuary zones at both locations showed no significant differences (P=0.9330) (Figure 7, Table 7).
Figure 7. MDS ordination of fish assemblages represented as centroids for each site within sanctuary zone (black circles) and outside sanctuary zone (white circles), where: (a) is combined locations, (b) is Coffin Bay, and (c) is Venus Bay. Fish assemblages are Bray- Curtis similarity measures following square root transformations.
Table 7. Single factor permutational MANOVA results comparing fish assemblages inside and outside sanctuary zones in Coffin Bay and Venus Bay. Combined represents both Coffin Bay and Venus Bay data.
SITE MODEL Df MS F P
Combined Inside/Outside zone 1 642.69 0.2930 0.9330
Residual 37 2197.20
Coffin bay assemblage Inside/Outside zone 1 71.48 0.5210 0.9790
Residual 9 1855.80
Venus bay assemblage Inside/Outside zone 1 819.19 0.3640 0.9040
Residual 26 2252.70
II. FACTORS CONTRIBUTING TO ASSEMBLAGE VARIATION
The structure of fish assemblages at both locations was significantly different depending on environmental factors and geographic locations. Fish assemblages varied by depth (F5,33
=2.371, P=0.0004) (Figure 8, Table 8). Post hoc pairwise analysis revealed fish assemblages differed between some but not all the depths (Appendix D). Most of the significant
differences were between the larger depth differences, e.g. between 1m and 6m (P=0.007) (Appendix D).
Fish assemblage structure also varied by habitat (F4,34=1.918, P=0.0041) (Figure 9, Table 8).
Post hoc pairwise analysis showed that significant differences occurred only between seagrass and broken sand habitats (P=0.0108) (Appendix D).
There were further significant differences in fish assemblages between January and June sampling (F1,37=6.256, P=0.0002) (Figure 10, Table 8). Although insignificant, trends were also seen for A. georgianus, P. wrightii, S. punctata and P. octolineatus, which all had higher abundances in January than in June. Several other groups (A. truttaceus and Platycephalus spp.) had higher abundances in June than in January.
The different sanctuary zone sites also showed significant differences in fish assemblages (F4,34=4.564, P=0.0001) (Figure 11, Table 8). Post hoc pairwise analysis showed significant differences between all sanctuary zones except for sanctuary zone 8 (Venus Bay) and sanctuary zone 3 (Coffin Bay) (Appendix D).
The species which contributed most to the assemblage differences for all factors were A.
truttaceus, A. georgianus, P. wrightii, S. punctata and P. octolineatus (Figure 8-11).
Figure 8. MDS ordination of fish assemblages from both Coffin Bay and Venus Bay
separated into depth bins of one metre. Overlay vectors of species contributing to differences among depths (codes are shown in Table 3). Fish assemblages used Bray-Curtis similarity measures following square root transformations. The stress value is shown on the plot.
Figure 9. MDS ordination of fish assemblages from both Coffin Bay and Venus Bay separated by habitat type. Overlay vectors of species contributing to differences in sites (codes are shown in Table 3). Fish assemblages used Bray-Curtis similarity measures following square root transformations. The stress value is shown on the plot.
Figure 10. MDS ordination of fish assemblages from both Coffin Bay and Venus Bay separated by sampling time (January and June). Overlay vectors of species contributing to differences between seasons (codes are shown in Table 3). Fish assemblages used Bray- Curtis similarity measures following square root transformations. The stress value is shown on the plot.
Figure 11. MDS ordination of fish assemblages from both Coffin Bay and Venus Bay separated by sanctuary area. Sanctuary areas 1, 3, 4 and 5 are located in Coffin Bay and sanctuary area 8 in Venus Bay. Overlay vector of species contributing to differences in sites (codes are shown in Table 3). Fish assemblages used Bray-Curtis similarity measures following square root transformations.
0 1 2 3 4 5 6 7 8 9
Arrgeo Arrtru Pseuwri Sillpun Peloct Plat
Mean (SE) number of species
Table 8. Single factor permutational MANOVA results comparing fish assemblages with different factors (depth, habitat, season and sanctuary zone (by location)).
FACTOR MODEL Df MS F P
Depth 1m, 2m, 3m, 4m, 5m, 6m 5 4332.00 2.371 0.0004
Residual 33 1826.60
4 3771.30 1.918 0.0041
Residual 34 1966.30
Season January/June 1 11851.00 6.256 0.0002
Residual 37 1894.30
Sanctuary zone (by location) 1/3/4/5/8 4 7156.90 4.564 0.0001
Residual 34 1568.00
Figure 12. Mean number of numerically-dominant species and genus groups in January (light grey) and June (dark grey) sampling periods.
The length frequency of dominant species and genus groups varied depending on location (either Coffin Bay or Venus Bay) and species, although no differences in frequencies were significant (Figure 13, Table 9). Generally, there was a trend of a higher percentage of smaller fish inside sanctuary zones (Figure 13). This is evident for Platycephalus spp. in Coffin Bay, A. georgianus in both Coffin Bay and Venus Bay, A. truttaceus in Coffin Bay, and S. punctata in both locations (Figure 13).
0 20 40 60
0 50 100 150 200 250 300 350 A. truttaceus
VB inside VB outside 0
20 40 60 80
0 50 100 150 200 250
P. wrightii n=32
VB inside VB outside
0 20 40 60
0 50 100 150 200 250 300
A. georgianus n=38
VB inside VB outside
0 10 20 30 40 50
0 50 100 150 200 250 300 350 400 450 500 550 600
S. punctata n=37
VB & CB Inside VB & CB Outside
0 20 40 60 80
0 50 100 150 200 250
P. octolineatus n=25
CB Inside CB Outside Percentage (%) Percentage (%) Percentage (%) Percentage (%)
0 20 40 60 80
0 50 100 150 200 250 300 350 A. truttaceus
CB inside CB outside 0
20 40 60 80
0 50 100 150 200 250 300
A. georgianus n=37
CB inside CB outside
Figure 13. Length frequency of numerically-dominant species and genus groups inside (light grey) and outside (dark grey) sanctuary zones in Coffin Bay and Venus Bay. Where n is the sample size.
0 10 20 30 40 50
0 50 100 150 200 250 300 350 Platycephalus spp.
CB Inside CB Outside
Table 9. Kolmogorov-Smirnov values and p-values of the differences in distribution of the length data between inside and outside the sanctuary zones.
Location Species/genus Kolmogorov-Smirnov value P-value
Venus Bay Pseudocaranx wrightii 0.167 1.000
Coffin Bay Platycephalus spp. 0.500 0.270
Venus Bay Arripis georgianus 0.286 0.938
Coffin Bay Arripis georgianus 0.143 1.000
Venus Bay Arripis truttaceus 0.125 1.000
Coffin Bay Arripis truttaceus 0.375 0.627
Venus Bay Sillaginodes punctata 0.154 0.998
Coffin Bay Pelates octolineatus 0.333 0.893
EFFECTIVENESS OF SANCTUARY ZONES
South Australia’s recently developed sanctuary zones within marine parks are showing promising signs of increasing fish abundances and diversity. There were no significant differences between abundance, species richness and size of specific fish species between inside and outside sanctuary zones. Although not statistically significant, there were still clear trends showing that sanctuary zones in both Venus Bay and Coffin Bay have higher total fish abundances inside rather than outside. When combined, there were 76% more fish in terms of abundance inside the sanctuary zones compared to outside. In a broad sense, these are
positive signs for the potential future success of these sanctuary zones.
When abundance and diversity data were separated by location, there were no significant differences inside Sanctuary Zone 8, the only underwater sanctuary zone in Venus Bay, or the four sanctuary zones analysed in Coffin Bay. Despite the recent establishment of South Australia’s protected areas, there were positive trends in both locations’ sanctuary zones.
There were 90% more fish inside the Venus Bay sanctuary zone compared to outside, and 58% more fish inside the Coffin Bay sanctuary zones. In both locations, the key abundance difference occurred in January. In Venus Bay there was almost three times the amount of fish inside the sanctuary zone compared to outside. In Coffin Bay there were 278% more fish inside protected zones, with the significantly lower p-value most likely a result of less variation and a higher sampling size. These results support that these sanctuary zones are functioning as they should, and are particularly encouraging considering their recent establishment.
Despite abundance showing positive signs of increasing, there were no differences in the species richness of fishes between inside and outside sanctuary zones at both locations. This suggests specific species are the cause of high differences in abundance data. Within Venus Bay, four main species dominated, including A. georgianus, A. truttaceus, P. wrightii and S.
punctata. These four all have higher average abundances inside Sanctuary Zone 8 compared to outside, although none of them were found to be significant. The large variation within the abundance data may have contributed to the lack of significance. For example, although there were 220% more P. wrightii inside Sanctuary Zone 8 than in fished areas, the comparison
was not significant due to high amounts of variation of total abundance between inside and outside. This could be in part due to P. wrightii being a schooling species, meaning that if a large school swam past the BRUV the data could be unrealistically high. The low degrees of freedom due to lack of replicates in this analysis may also reduce power to detect differences.
Fish assemblages within sanctuary zones do not differ from areas outside sanctuary zones at both locations. Fish assemblages are biotic indicators of overall ecosystem health and productivity (Bell 1983). Over time, it is expected that protected areas will improve in environmental health by increasing the structural complexity of habitats, which in turn will cause an increase in fish abundance and richness (Bell 1983). These results suggest the recent establishment of parks has not yet allowed sufficient time for the ocean floor habitat to improve to a level where fish assemblages are differing. Furthermore, the difference in assemblage structure may not be evident due to the close proximity between BRUV deployments, and the similarities in habitats and depths between inside and outside zones.
Protected areas of ocean, particularly in the form of sanctuary zones, are integral to the health and maintenance of natural ocean ecosystems. Numerous studies conclude that MPAs contain higher fish diversity and abundance, with greater average fish size and general higher
biodiversity than areas outside MPAs (Russ et al. 2008; Kelaher et al. 2014; Soler et al.
2015). Despite this, it was still expected that the results from this study would not show drastic improvements inside sanctuary zones due to the enforcement of the South Australian MPA’s zoning only commencing recently, in October 2014. Other studies investigating similar concepts in recently established MPAs have found that a three year period since reserve establishment may not be sufficient in generating clear cut trends in fish population recoveries (Edgar and Barrett 2012; Kelaher et al. 2014). Furthermore, a global meta-analysis suggested that the positive effects of MPA success are linked to the amount of time that has passed since the MPA was established, with reserves established for more than ten years having higher species diversity and abundance (Claudet et al. 2008; Edgar et al. 2014).
MARINE PROTECTED AREA DESIGN
When interpreting positive abundance trends within MPAs, it is important to consider if the protected areas were purposefully located in areas where fish naturally occur in higher numbers (Kelaher et al. 2014). The West Coast Bays Marine Park, of which Venus Bay is a part, and the Thorny Passage Marine Park of which Coffin Bay is a part, were created in
areas according to the guidelines recommended by DEWNR (Baker 2004). The development of the MPAs used existing data, resources and knowledge to better understand the marine habitats and hence protected areas were placed in ecologically important locations (Baker 2004). The recommendations also state that modifications of the existing network of MPAs may occur as the knowledge on the functioning, distribution and environmental impacts of South Australia’s unique marine biota is broadened (Department of Environment, Water and Natural Resources 2012). One of the most effective ways to monitor the effectiveness of the protection is to sample the abundances of particular key species over time. Hence, this study provides baseline data which can be built on in future years to continue the monitoring in these regions. In turn, this information can provide rare knowledge on these regions, potentially influencing the design and locations of MPAs and their zoning in the west coast embayment regions.
While individual MPAs provide some conservation benefits, the design of a MPA is also essential to its effectiveness. Previous research has shown that the five key features:
enforcement of regulations, full protection, large size (>100km2), longevity (>10 years since setup), and isolation, are essential for MPA success (Edgar et al. 2014; Halpern 2014).
Although it is hard to monitor the enforcement of west coast MPAs, research shows that protected areas with boundaries that exist only in principle but have limited enforcement dramatically lessen the MPAs success (Mora et al. 2006; Guidetti et al. 2008). Furthermore, areas which are only partially protected, such as all of the MPAs in the west coast
embayments excluding the sanctuary zones, have significantly less effectiveness than no-take regions (Edgar et al. 2014). South Australia’s MPAs cover an area of 26,655 square
kilometres, although some of the sanctuary zones within, including those tested in this study, are less than the 100km2 needed for success (Department of Environment, Water and Natural Resources 2012). These factors are important when designing MPAs in the future and
ensuring as many as possible are included is key to the future success of the protected areas.
Understanding the distribution of fish assemblages at a wide range of different spatial scales is an essential step towards discovering important underlying ecological processes and factors that affect fish assemblages. This knowledge, in turn, can be used for the selection and design of MPAs to ensure that the appropriate areas are covered. This includes taking environmental
and biological factors into consideration, so that the ways in which they are interacting can be accounted for.
ENVIRONMENTAL FACTORS INFLUENCING FISH ASSEMBLAGES
The results showed that fish assemblages significantly differed between different habitats and depths, and particularly seasonally. Species of fish associated with temperate zones have a wide range of biological characteristics that affect the way in which they respond to the environment, hence their variation in spatial distribution. Particular species prefer to live in certain depths, habitats and migrate and recruit in different seasons. It is for these reasons that it is important to cover a wide range of depths and habitats when designing sanctuary zones within MPAs. Although important in MPA success, the range of depths and habitats raise issues relating to variability within the data and make it is hard to distinguish between the factors which are causing these assemblage differences. Despite this, there are still some obvious fish assemblage differences between environmental factors.
Significant fish assemblage differences occurred between different depths. Although, this was expected between shallower and deeper areas (Friedlander and Parrish 1998; Connell and Lincoln-Smith 1999; Hyndes et al. 1999), in this study depths only varied between shallow zones (one to six metres), so these results were unexpected. Because of similarities in depth range, it is hard to know if the results were in fact related to depth, or rather a spatial
difference associated with marine sanctuary zones. For example, in the samples taken inside and outside Sanctuary Zone 1, all of the BRUV deployments were in the 4-6 metre range, while in Sanctuary Zone 8, BRUV deployments were all in 1-2 metres. Furthermore, certain depths were more abundant than other depths, which may have skewed the significance of the data slightly. Despite this, there were assemblage differences across depth ranges, which warrants further investigation. As such, the study design could be modified in the future to distinguish spatial (sanctuary zone) and depth related patterns in fish assemblages.
As well as assemblage differences between depths there were also significant differences between different habitats. This was particularly evident between seagrass and sand based habitats. Seagrass is known to act as important habitat for juvenile fish (Aaron et al. 2006;
Nagelkerken et al. 2012), As such, these results are most likely attributable to juvenile fish
species being within the seagrass regions. Habitats are being used more commonly in MPA planning as they are useful surrogates for biodiversity (Ward et al. 1999; Harman et al.
2003). When planning the design of an MPA, including a variety of habitats is recommended, with all habitats represented within the protected area (Kelleher and Kenchington 1991;
Roberts et al. 2000). This is beneficial to preserving both the habitats themselves, and the fish that use them (Rosenberg et al. 2000).
As mentioned above there are some sampling issues, most likely due to lack of haphazard spacing among sites, therefore results could be due to spatial variation. The results showed that most sanctuary zones had a particular habitat that dominated the region. Sanctuary Zones 3, 5 and 8, for example, had all BRUV deployments in seagrass. Therefore, the results could be due to differences in sanctuary zones mentioned below, rather than differences in habitats.
Despite this, there were still significant differences between these seagrass-based sanctuary zones, indicating that spatial patterns are greater than habitat patterns. Furthermore, the dominance of seagrass as a habitat, and the lack of other habitat samples such as macro algae, could have caused some issues with statistical power of the analysis. The statistical power of an analysis is increased with sample size, hence in our study where there is just one site with macro algae, the sample size is very low for this particular analysis. Some changes to the study design could be incorporated to increase the statistical power, such as ensuring that each sanctuary zone has a wide variety of habitats and depths tested. Despite this, the
differences between fish assemblages among habitats are interesting, and supported by a wide range of literature (Guidetti 2000; Gratwicke and Speight 2005), and confirm the necessity for sanctuary zones to be located in a range of habitats.
Seasonally, fish assemblages are expected to undergo changes, cycling consistently among years (Wright 1988; Hyndes et al. 1999). The results from this study showed clear differences in fish assemblages between January and June sampling periods. The seasonal differences may be attributable to immigration and emigration of different fish species including recruitment throughout the year (Ansari et al. 1995; Potter et al. 1997). Particular nursery species could be moving out to deeper waters or fish could be migrating to spawning locations (Hyndes et al. 1999). The water temperature differed between these sampling
periods, so lower abundances of fish during winter sampling could also be attributed to temperature changes.
When specific species and genera were investigated there were clear differences in
abundances between seasons, which supports the idea that fluctuations relate to the life cycles and emigrations and immigrations of particular species. P. octolineatus, for example, was only recorded during January sampling, with 51 individuals sited in January compared to zero in June. This is expected as previous research found that P. octolineatus migrate from
seagrass nursery areas into deeper waters to mature and spawn during spring (Potter et al.
1983; Veale et al. 2015). Therefore, by January, it is expected that the abundance of juvenile P. octolineatus would be high in seagrass meadows, where most of the sanctuary zones are located. It is likely that other species follow similar life cycles and trends.
SPATIAL VARIATION INFLUENCING FISH ASSEMBLAGES SANCTUARY ZONE SITE
Variation in fish assemblages at a range of different spatial scales is expected for temperate reef fishes (Anderson and Millar 2004; García-Charton et al. 2004; Gladstone 2007).
Significant spatial variation in fish assemblages occurred between individual sanctuary zones.
This variation occurred at a scale of 1-10 kilometres. Spatial variation on a small scale may be related to variation in structure of the habitat (Connell and Jones 1991; Willis and
Anderson 2003), depth, recruitment (Connell and Jones 1991; Smith et al. 1991), local larval accumulation and retention (Warner et al. 2000), or other influences.
The design and layout of MPAs are often limited due to social and economic factors. This is particularly seen when the size of protected areas is limited to a fraction of the bioregion whose biodiversity they are intended to represent. Often, they are not large enough to be self- sustaining as their size is smaller than the dispersal distance of key species (Halpern 2003;
Claudet et al. 2008). This relates to potential issues with the design of the sanctuary zones within South Australian MPAs. The results of this study show that over very small distances there is still variation related to species composition. An easy solution would be creating larger reserves, however due to socio-economic pressures this is unlikely to have community support. An alternative is to create more connectivity between the reserves. In theory, this allows each reserve to contribute and receive a sufficient amount of adults and larvae from connected reserves. Furthermore, connectivity between protected areas is important for
population dynamics and genetics of marine organisms (Palumbi 2003; Cowen et al. 2006).
This would potentially ensure the protection of species and increase the success of sanctuary zones to replenish a wide range of fish species.
LENGTH OF FISH IN SANCTUARY ZONES
While increasing the abundance and diversity of fish, individuals of commercially fished species in MPAs are thought to increase in size. Protected areas provide a safe area for these large fish, which are usually crucial for reproduction, the offspring of which often spills over to areas outside the reserve. Furthermore, protected areas also provide a healthy habitat for recruitment and larval export to occur, with strong evidence of community wide benefits flowing from MPAs to unprotected areas (Hilborn et al. 2004; Guidetti 2007; Harrison et al.
Some size differentiation occurred between inside and outside sanctuary zones. In particular, some species showed trends of higher abundances of smaller sized species inside the
sanctuary zones, indicating that recruitment and larval export maybe occurring within the protected areas. This is seen in Coffin Bay for A. truttaceus and Platycephalus spp., and in both bays for A. georgianus. Despite the patterns seen within these results, it is important to keep in mind that the sample sizes of all species were quite small, with all species having a total abundance of 43 or less. Furthermore, the sanctuary zones within both embayments had only been enforced for less than three years when the data were collected, which is most likely not enough time for the fish populations and habitats to recover to a detectable level.
LIMITATIONS IN THE PRESENT STUDY
Similar to most studies, limitations occurred throughout phases of this study. As mentioned above, there are issues relating to the time frame since the South Australian MPAs were established, and the enforcement of the sanctuary zones within them. Previous studies have indicated that three years is insufficient time to see obvious changes in fish abundance and diversity, as well as habitat recovery (Edgar and Barrett 2012; Kelaher et al. 2014). Despite this, the preliminary results of this study are promising and provide important baseline data that can be used in the future to compare how much the MPAs have improved.