Demographics and vulnerability of a unique Australian fish, the weedy seadragon

Памер0.69 Mb.
  1   2   3

Demographics and vulnerability of a unique Australian fish, the weedy seadragon Phyllopteryx taeniolatus

Jaime Sanchez-Camara1, Keith Martin-Smith2, David J. Booth3, Juan Fritschi1 & Xavier Turon4*

1 Aquadec Aquariums S.L.18600 Motril, Granada, Spain
2 Australian Antarctic Division, Australian Government Department of Environment, Water, Heritage and the Arts, Kingston, Tasmania 7050, Australia
3 Department of Environmental Sciences. University of Technology, Sydney. PO Box 123 Broadway 2007, NSW, Australia

4 Dept. of Aquatic Ecology. Center for Advanced Studies of Blanes (CEAB, CSIC), 17300 Blanes (Girona), Spain (email: )

The weedy seadragon Phyllopteryx taeniolatus is a vulnerable and endemic Australian fish and also an icon and flagship species for marine conservation. However, little is known about its population dynamics, which hinders the establishment of conservation policies. We have previously demonstrated seadragons to be highly site-attached, so we estimated population densities, growth and survival of weedy seadragons using mark-recapture techniques at five sites in New South Wales (NSW, 34º S) and Tasmania (TAS 43º S), near the northern and southeastern limit of distribution for the species, over a seven year period. Population densities varied from ca. 10 to 70 seadragons*Ha-1 depending on site and year. There was a significant decline in the number of weedy seadragon sightings per unit area searched in two out of three study sites near Sydney, NSW, from 2001 to 2007. There was also a decline at one of the two sites surveyed in the lower Derwent Estuary, TAS, in 2009 compared to 2003 and 2004. Survival rates at NSW sites ranged from 0.62 to 0.65 yr-1 and were higher at TAS sites ranging from 0.71 to 0.77 yr-1. Birth occurred approximately three months later and seadragons exhibited significant slower growth in Tasmania (L*k = 31.02) compared to New South Wales (Sydney) (L*k = 55.15). This study is the first population assessment of seadragons over ecologically relevant spatial and temporal scales, and shows differences in the dynamics of populations at different latitudes. It also shows declines in some populations at widely separated sites. Determining whether these declines are natural inter-annual fluctuations or whether they are caused by environmental or habitat changes must be a priority for conservation.

Keywords: mark recapture analyses, conservation, growth, survival rate, population dynamics, endemic Australian fish


In the world´s oceans, human impacts have been substantial, leading to concerns about the extinction of marine taxa (Dulvy et al. 2003). These concerns have created the need to identify which species are at risk of extinction (Dulvy et al. 2004, Hutchings 2001) and design proper management strategies (Beissinger & Westphal 1998, King & Mcfarlane 2003). Extinction risk can be assessed based on knowledge of life history and ecology, examining changes in abundance, and using demographic analysis (Dulvy et al. 2004). Such assessment has been considerably aided by the use of individual tagging and monitoring. Burnham et al. (1987) and Lebreton et al. (1992) provide guidelines for capture-mark-recapture (CMR) data analysis and these methods are now generally used to estimate a wealth of demographic parameters such as abundance, recruitment, survival rates, immigration rates and population growth rates in fish, birds, mammals and amphibians. These demographic data are particularly important for flagship species, those used in conservation to raise public awareness, action and funding (Zacharias & Roff 2001, Caro et al. 2004).

The syngnathid fish weedy seadragon (Phyllopteryx taeniolatus) is a flagship species restricted to the seaweed beds of southern Australia. They are popular and charismatic organisms due to their size, colorful appearance, and unusual life history characteristics including male brooding of eggs. In addition, they might: (1) be a threatened species on its way to being endangered because of life history including low fecundity and small home ranges (Sanchez-Camara and Booth 2004, Sanchez-Camara et al. 2005) and its endemic nature (2) be a good indicator of the condition of the habitat because of their dependence on the seaweed beds (Sanchez-Camara et al. 2006) and because they are highly sensitive to changes in temperature, salinity and water quality (Koldewey 2005). Although the species is well protected by legislation (Connolly 2006), the habitat of P. taeniolatus has been adversely affected by shoreline alterations and by pollution. The degradation is worst near major urban centers such as Sydney, raising concerns about the current state of seadragon populations (Connolly 2006). In Tasmania, the increase of sea urchins Centrostephanus rodgersii, due to harvesting of rock lobster and climate change, reduces the abundance of kelp and other macroalgae through the formation of urchin barrens (Ling & Johnson 2009, Ling et al. 2009) and this reduction in macroalgae could adversely affect seadragons. The abundance of seadragons might have been affected by significant habitat degradation and loss in other areas of Australia as well. These concerns have led to a recent change in the status of P. taeniolatus in the IUCN Red List of Threatened species from Data Deficient to Near Threatened (Connolly 2006) which means that it is likely to qualify for a threatened category in the near future. In fact, it is precisely the lack of trend data that prevents seadragons from meeting any of the threatened categories.

In previous studies, the authors have provided data on the life history of P. taeniolatus, including population structure, site fidelity, movement patterns, reproductive ecology, recruitment and growth, habitat use and behaviour (Sanchez-Camara and Booth 2004, Sanchez-Camara et al. 2005, Sanchez-Camara et al. 2006). However, these investigations were performed on the time frame of ca. 1 year, which hindered the assessment of demographic trends. The aim of the present study was to build on previous results using longer term capture-mark-recapture (CMR) data, collected from 2001 to 2009, and compare demographic characteristics between 2 widely separated locations. These findings should represent the basis for management policies in order to prevent this “Near Threatened species” from becoming a truly endangered species.


Study Species

The weedy seadragon (Phyllopteryx taeniolatus) is a syngnathid fish endemic to southern Australian waters. It is found along much of the southern Australian coastline, from Port Stephens, New South Wales (NSW) (32° 38’S) southwards to Actaeon Island, Tasmania (TAS) (43° 32’S) and westwards through Victoria (VIC) and South Australia (SA) to Geraldton, Western Australia (WA) (28°46’S) (Pogonoski et al. 2002). Weedy seadragons have elongate, non-prehensile tails and can grow up to 45 cm in length. Males are responsible for the offspring that they fertilise and incubate on the outside of the tail with each egg partly embedded in the skin (Kuiter 2000). They live among the larger algae on exposed reefs and feed mainly on mysid crustaceans (Edgar 2000).

Populations studied in NSW lived in resident groups throughout the year with males, females and juveniles occupying broadly overlapping home ranges. They breed from approximately June to January with a peak of brooding males at the end of the season. Males breed once or twice per breeding season with gestation periods of 30 to 38 days (Sanchez-Camara & Booth 2004). Juvenile individuals appear in the residence areas from November to March measuring ca. 15 cm (Sanchez-Camara et al. 2005).

Study Sites

Most of this study was carried out during 2004 and 2007 at three sites around Sydney (site 1, 34º 00’ 04” S 151 º 13’ 24”E; site 2, 34º 00’ 07” S 151º 13’ 06”E; site 3, 33º 53’ 33” S 151º 16’ 56”E), NSW (Figure 1a). In these same sites previous studies were conducted during 2001 and 2002 (see Sanchez-Camara & Booth 2004), and we take advantage of this previous information to analyze longer-term trends. Sites 1 and 2 are at Kurnell, in Botany Bay, an increasingly industrialized bay with Sydney’s airport and major port facility on the shoreline. Potential threats to the kelp habitats and site water quality have come from construction and operation of a new desalination plant and several pipelines across nearby seagrass, as well as expansion of Australia’s busiest seaport 5 km across Botany Bay. Site 3 is located in North Bondi, an exposed rocky reef only a few kilometers from the Sydney city centre. Additional occasional surveys were conducted around Sydney searching for tagged seadragons outside the tagging areas. In order to compare seadragon populations from different latitudinal ranges, fieldwork was also conducted in 2003, 2004 and 2009 in Tasmania (TAS), at Kingston Beach (Site 4, 42º 59’ 03” S 147º 19’ 34”E) and Blackman’s Bay (Site 5, 43º 00’ 02” S 147º 19’ 41”E, Figure 1b).

At sites 1 to 3 we used permanent transects parallel to shore and 350 meters in length, following the reef edge that were set up in the study of 2001 and 2002 (see Sanchez-Camara & Booth 2004 for details). Estimated area covered was 1 Ha for each site (see Sanchez-Camara et al. 2006). At sites 4 and 5, transects parallel to the shore at the reef/sand interface were surveyed and site dimensions estimated using a GPS unit into a waterproof case that was towed on the surface. We covered an area similar in size to that of sites 1 to 3.

Sampling and tagging protocol

All tagging and observations were conducted using SCUBA (following Sanchez-Camara & Booth 2004). On each dive at the Sydney sites, the transect was followed from one end to the other with constant short incursions into the rocky reef and the sand flat. Seadragons were tagged with VIFE injectors (Visual Implant Fluorescent Elastomer®, Northwest Marine Technologies, Inc. Shaw Island, WA, USA). The tagging was done in situ by gently restraining the animal with one hand and injecting the tagging material with the other. Additional individuals, especially recruits, were identified by their pattern of appendages, allowing estimation of recruitment and movement patterns. However, this method was not suitable for long term study and these individuals were not considered in the mark-recapture analysis.

On each dive, sighted seadragons were checked for tags using an UV torch. Natural marks and appendages of all fish seen were noted to later corroborate tag identification and to avoid double-counting of untagged fish on the same dive. Seadragons were measured (Ls) following Sanchez-Cámara et al. 2005.

Data Analyses

In the three NSW sites (sites 1-3), we analyzed temporal changes in the number of sightings per hectare searched at each dive, assuming that the site area (transect width ca. 28 m) was fully searched when covering the transect both ways, one way covering the kelp-sand limit and the other way covering the kelp area. The time spent on each survey was dependent on the level of complexity of the habitat, mainly related to kelp development, so we were confident that our ability to find seadragons was equivalent in the different sites and among years. We used ANOVA to compare the sightings/area of all surveys at each site using the different time periods as the categorical variable. To this end, we pooled data from 2001-2002 to allow homogeneous intervals between time points, and compared them to data from 2004 and from 2007. A two-way factor ANOVA with time period and sex (male, female, juvenile, uncertain) as factors were also used to ascertain whether differences in time, if any, were consistent across sexes. For Tasmania, we pooled the data for 2003-2004 and compared the average number of sightings per area in this period with those of 2009 using a t-test.

Normality and homogeneity of variances of the data were assessed by Kolmogorov-Smirnov and Bartlett tests, respectively. When our untransformed data did not fulfill these assumptions (detailed in Results), rank transformed data were used in the analyses (Potvin & Roff 1993). All analyses were performed with Sigmastat v3.1.

Population size estimates using the Schnabel and the Schumacher-Eschmeyer method (extension of the Petersen estimator to a series of samples, based on the proportion of marked animals present in the sample) were calculated for the different sites and sample periods (Krebs 1999). This method assumes that the populations are closed over the time period concerned, which we considered a reasonable approximation over these relatively short time periods. These estimates were translated to density estimates by dividing by the area surveyed at each site.

With the mark-recapture data we estimated apparent survival () and encounter probability (p) of seadragons at each site using a Cormack-Jolly-Seber (CJS) model of open populations as implemented in program MARK (White & Burnham 1999). The CJS model was chosen because, while the more general Jolly-Seber model can be used to estimate population size and the number of new individuals entering the population, it can be difficult to avoid bias in these estimates because of individual heterogeneity (White et al. 1982). Other models, including the robust design (Kendall et al. 1997), were investigated but the data structure of the encounter history matrix did not allow robust estimation of life history parameters. A set of candidate models were generated for each site including the full models (i.e, dependent on time and group), and reduced models with time and group-independent apparent survival .) and encounter probability p(.), time-dependent apparent survival t) and encounter probability p(t) and group-specific (male, female, juvenile) apparent survival g) and encounter probability p(g). Bootstrap goodness-of-fit tests were implemented with the MARK program. Where these tests indicated significant extra binomial variation, the variance inflation factor ĉ was estimated and used to adjust the model outputs (see Cooch & White 2006). The Akaike Information Criterion (AIC) for each model was compared to indicate the degree of support. The apparent survival and encounter probabilities were estimated for all models with model likelihoods >0.20 (Cooch & White 2006).

We calculated a growth curve following the well-known von Bertalanffy (VB) equation (von Bertalanffy 1938):

Lt= L-(L- Lo)*e-k*t

where Lt is the size (in our case, standard length in cm) of an individual of age t (years), L is the maximal size attained at adulthood, Lo is the size at birth and k is a growth rate parameter. In a previous study (Sanchez-Camara et al. 2005) we provided a VB growth function calculated on data from recruits and juveniles at the Sydney sites (1-3) during 2001-2002. Now we have gathered a larger dataset by considering all individuals that could be assigned to a given cohort in all sites over the study years, and including all measures available for them, so we have a larger number of individuals and a much longer temporal frame to estimate more accurately the parameters of the VB function. We calculated the growth functions for Sydney and for Tasmania separately.

The VB function has 3 parameters (L∞, Lo, and k). In addition, we need to estimate a date of birth to determine the exact age. The four parameters were estimated simultaneously for the Sydney populations, for which we had more data; while for Tasmania some parameters had to be set to pre-established values to obtain reliable estimates for the remaining parameters (see Results). The growth function parameters that maximized the fit of our data to the equation were estimated using a non-linear iterative procedure (NONLIN module of the program Systat v. 11 with least squares estimation and using a quasi-Newton algorithm).


Population abundances and temporal trends

Overall, 210 dives and 456 diving hours were performed across all sites and years. At NSW sites, number of dives each year (2001, 2002, 2004, 2007) were 40, 16, 22, 6 for Site 1; 22, 12, 22, 5 for Site 2 and 1, 10, 16, 1 for Site 3. At TAS sites, number of dives each year (2003, 2004, 2009) were 5, 7, 5 for Site 4 and 1, 3, 16 for Site 5. Thirteen more dives and 25 diving hours were performed at the additional sites. A total of 223 seadragons were tagged from 2001 to 2009 at the study sites, 147 at the NSW sites and 76 at the TAS sites. Surveys reported 845 sightings of tagged individuals. Additionally, a number of recruits (settled young of the year) were individually identified by appendages due to the difficulty of tagging small specimens. The cohorts of those recruits will be identified with the notation Y00-01 (i.e, young from the season 2000-2001), Y01-02, etc. indicating the breeding season in which the animals were born.

There was a temporal trend across years of decreasing numbers of seadragons seen over the study at sites 1, 2 and 4 (Figure 2, a, b and d) but not at sites 3 and 5 (Figure 2, c and e). At Site 1, the ANOVA analyses showed a significant change over time (F(2,76) = 6.973, p = 0.002, rank-transformed data). The post-hoc comparisons (SNK, Student-Newman-Keuls method) detected significant differences among all time periods except for the comparison 2001-02 vs 2004. Similarly, at Site 2 the overall time period effect was significant (F(2,56) = 12.642, p < 0.001, rank-transformed data) and the SNK revealed that the only non-significant difference was between 2004 and 2007. In contrast, the time period did not have a significant effect on the number of sightings at Site 3 (F(2,26) = 1.854, p = 0.177). Thus, a significant decline in population abundance was substantiated in the Botany Bay sites but not in the more exposed Site 3. At Tasmanian sites, there was a significant decline in numbers in 2009 compared to 2003-2004 at Site 4 (t-test, t = 3.504 df = 15, p =0.003) but not at Site 5 (t-test, t = -0.222, df = 18, p =0.826).

The number of males and females seen per area searched are shown for the 3 Sydney sites (Figure 3, a to c). A trend of decline in the number of males, but not of females, can be observed at site 2. However, no significant interaction was found between year and sex (male, female, juvenile) (Site 1, F(9,300) = 1.776, p = 0.072; Site 2, F(9,220) = 1.810, p = 0.068, rank-transformed data; Site 3, F(9,96) = 1.1946, p = 0.308). The factor sex was significant at all sites (Site 1; F(3,320) = 41.184, p < 0.001; Site 2; F(3,220) = 40.652, p < 0.001; Site 3, F(3,96) = 6.522, p < 0.001); in all cases there were significantly fewer sightings of juveniles compared to males and females (Student-Newman-Keuls test). Due to the lower number of observations and the fact that the sex was not determined for some seadragons in 2003-4, this analysis could not be done for Tasmanian sites.

Population estimates using the Schnabel and Schmacher-Eschmeyer methods are shown in Table 1. Similar trends in abundance over time to those found with the number of sightings were observed. At site 1 the number of seadragons per ha decreased from 45-49 in 2001-02, depending on the period and method used, to 30-34 (2004, 2007). The same trend occurred at Site 2, with seadragons per ha decreasing from ca. 56-69 to 28-35. At site 3 estimated population sizes were 10 to 15 seadragons with no clear temporal trend. At site 4 estimated population sizes decreased over the study period from 32-34 to 24-27 and at site 5 population estimates were 15 to 32 with an increasing trend (Table 1).

Survival and encounter rates

Summaries of the most likely models explaining the patterns of mark-recapture are given in Table 2. At Site 1 the best supported model was (g) p(g) with higher apparent survival and lower encounter probabilities for juveniles compared to adults. Males had slightly lower survival rate and probability of encounter compared to females. At Sites 2, 3 and 4 the best supported models were (.) p(.) with no differences in apparent survival or encounter probabilities among any of the groups. At Site 5, both, the (g) p(.), with potential differences in apparent survival between males and females, and (.) p(.) models were well supported. We suggest that the model (.) p(.) is the correct model for this site but that fewer marked animals and less diving effort compared to other sites biased the results. Apparent survival rates were higher for Tasmanian sites (Site 4 (.) = 0.770, Site 5 (.) = 0.713) than for Sydney sites (Site 1, (female) = 0.577; Site 1, (male) = 0.473; Site 2, (.)= 0.650; Site 3, (.)= 0.642) with the exception of the individuals tagged as juveniles in 2001(Y00-01) at Site 1 (Site 1, (juvenile) = 0.822). Encounter probabilities were higher at Site 3 compared to the other sites likely because the better visibility and the lower density of the “kelp bed” facilitated seadragon sightings. Goodness-of-fit tests for the mark-recapture data revealed that there was no additional binomial variation due to transience at any of the sites (p > 0.20).

Site fidelity and longevity

Seadragons showed similar persistence at all sites. One third (33 %) of the seadragons tagged at sites 1 and 2 (NSW) from July to November 2001 were seen in 2004. Seventy four percent of seadragons tagged at Site 4 (TAS) during 2003 were seen in December 2004 although none of the 7 tagged at Site 5 in 2003 were seen again. Nevertheless, at least for young life-history stages, relocation to new sites cannot be dismissed; another seadragon initially tagged as a juvenile in 2001 at Site 1 and never seen there again was found at Bare Island (33º59’29”S 151º13’46”E, 1.3 km away, on the other side of the bay) in December 2004. No other tagged seadragon was seen during the surveys made in the Sydney area outside the study sites.

There were considerable differences in persistence between males, females and juveniles in NSW. Fifty percent of the juveniles initially tagged in 2001 were later seen in November-December 2004 while this number decreased to 31% for females and only 7% for males (Figure 4). These females and males were tagged in 2001 at over 33 cm in length and an expected age over 2 years. Therefore, they were more than 5 years old when they were subsequently observed in December 2004. However, none of the seadragons tagged in 2001, including the youngest ones (belonging to Y00-01 cohort), were seen in December 2007, despite the site fidelity exhibited from 2001 to 2004. This suggests that the longevity is ca. 6 years for Sydney seadragons.
At Tasmanian sites, 2 adults tagged in 2004 and 1 adult tagged in 2003 were encountered in 2009. The latter corresponds to the seadragon with the maximum period of time from the first encounter to the last in our study. It was first seen on September 12, 2003 with a size of 31.5 cm and last seen on February 27, 2009. We are confident that the animal corresponded, at the latest, to the cohort Y00-01 because two smaller-size age-classes could be identified in 2003 ranging from 26.0 to 29.6 cm (breeding season 2001-2002) and from 18.7 to 20.5 cm (breeding season 2002-2003). Therefore, it is the oldest seadragon we can report in this study, with an estimated age over 8 years.


Fifteen recruits from the breeding season 2000-2001(Y00-01), 10 from 2001-2002 (Y01-02), 22 from 2003-2004 (Y03-04) and 1 from 2004-2005 (Y04-05) could be identified at the NSW sites. Two recruits from 2001-2002 (Y01-02) and 4 from 2002-2003 (Y02-03) were also identified and measured at TAS sites. We used all measures made on these identified recruits in subsequent observations (Figure 5) to estimate a growth curve. In the Sydney area the number of observations allowed us to estimate simultaneously all parameters iteratively. Although the birth date likely varied among individuals and years, in order to be able to use all observations we had to include in the equation the estimate of a common birth date that maximized the fit of the growth function to the data. This estimate was that, on average, the seadragons in the Sydney area were born the 24 October. The estimate of the growth rate k obtained was 1.52 yr-1 (95% confidence interval = 1.27-1.78); Lo was estimated at 3.33 cm (0.087-6.57), and the L estimate was 36.28 cm (35.15-37.42). The r2 of the regression was 0.936. Thus, the VB growth curve obtained was:

Lt = 36.28 - (36.28 - 3.33)*e(-1.52 * t)

The value estimated for Lo (size at birth) was reassuringly close to the 3.2 cm actually measured in aquarium by Forsgren & Lowe (2006). The asymptotic size (L), on the other hand, was almost coincident with the overall mean adult (individuals over two years old and 33 cm whose sex could be visually determined) size that we recorded in the area (Ls = 36.29 ± 0.34 cm; mean ± SE). For TAS sites, due to much fewer data available, we could not estimate all parameters simultaneously and reliably (lack of convergence and/or stability in the estimates). In particular, we lacked aged individuals older than 3 years, which could result in an underestimation of L∞. We therefore decided to use 3.2 cm (from Forsgren & Lowe 2006) as Lo value, and to use as L the mean of adult sizes observed in the area (34.07 ± 0.39 cm), while estimating the remaining parameters (k and common birth date). The estimate of k obtained was 0.91 yr-1 (95 % confidence interval = 0.70-1.14; r2 of the regression = 0.915) (Fig. 5). The estimation that maximizes our fit of the growth curve is that, on average, birth date was January 24.

Каталог: bitstream -> 10261
10261 -> Springer-Verlag
10261 -> R. B. Elias T. Resendes E. Dias
10261 -> Private forest reserves can aid in preserving
10261 -> Mammalia: Chiroptera: Vespertilionidae from the Canary islands
10261 -> Effect of chill storage under different icing conditions on sensory and physical properties of canned farmed salmon (Oncorhynchus kisutch) Alicia Rodríguez 1, Nicolás Carriles 1, and Santiago P. Aubourg 2
10261 -> Table 1 Sample inventory, feeding and migratory behaviour of the 14 bird species under study
10261 -> Biological and behavioral effects of a kaolin particle film on larvae and adults of Chrysoperla carnea (Stephens)
10261 -> Brief report Absence of haematozoa on colonial White Storks Ciconia ciconia throughout their distribution range in Spain
10261 -> Supplementary On-line Material Appendix S1
10261 -> Origina L article

Поделитесь с Вашими друзьями:
  1   2   3

База данных защищена авторским правом © 2022
звярнуцца да адміністрацыі

    Галоўная старонка