A conservation trade-off? Interspecific differences in seahorse responses to experimental changes in fishing effort

AQUATIC CONSERVATION: MARINE AND FRESHWATER ECOSYSTEMS

Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 468–484 (2007)

Published online 28 November 2006 in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/aqc.798

A conservation trade-off ? Interspecific differences in seahorseresponses to experimental changes in fishing effort

JANELLE M. R. CURTISa,c,*, JOAQUIM RIBEIROb, KARIM ERZINIb

and AMANDA C. J. VINCENTc

aDepartment of Biology, McGill University, Montreal, Quebec, CanadabCentro de Ciencias do Mar (CCMAR), Universidade do Algarve, Faro, Portugal

cProject Seahorse, Fisheries Centre, University of British Columbia, Vancouver, British Columbia, Canada

ABSTRACT

1. A 2-year experimental seining programme and underwater visual censuses were undertaken toquantify the direct effects of active demersal fishing on the population structure and relativeabundance of two sympatric seahorse species of conservation concern: the European long-snoutedseahorse, Hippocampus guttulatus Cuvier 1829 and the short-snouted seahorse, Hippocampushippocampus L. The influence of habitat preference on population-level responses to changes inhabitat structure following a reduction in fishing effort was also investigated.2. It was predicted that the benthic habitat would be more structurally complex after fishing ceased

and that seahorse densities would increase in response to reduced fishing mortality. Furthermore, itwas predicted that the magnitude of the increase in density would be greater for H. guttulatus thanfor H. hippocampus, because the former species prefers complex vegetated habitats while the latterspecies uses sparsely vegetated habitats.3. As predicted, the amount of habitat cover increased significantly when seining ceased, primarily

through increases in the abundance of drifting macroalgae and unattached invertebrates. Despitesimilarities in life histories, the two seahorse species responded differently in terms of magnitude anddirection to reduced fishing effort: the abundance of H. guttulatus increased significantly while H.hippocampus decreased in abundance.4. Results suggest that active demersal fishing may influence the magnitude and direction of the

responses of benthic marine fishes to exploitation through its impacts on habitat structure. Anincrease in habitat cover appeared to favour higher densities of H. guttulatus when seining effort wasreduced. By contrast, repeated seining, which maintained less complex habitats, appeared to favourgreater abundances of H. hippocampus.5. Given differences in habitat preference among benthic marine fishes subject to incidental

capture in fisheries, simultaneous attempts to manage populations of sympatric species may requirecomplementary strategies that support the persistence of diverse habitat types.Copyright # 2006 John Wiley & Sons, Ltd.

*Correspondence to: J.M.R. Curtis, Centre for Applied Conservation Research, University of British Columbia, 2424 Main Mall,Room 3004, Vancouver, British Columbia, Canada V6T 1Z4. E-mail: [email protected]

Copyright # 2006 John Wiley & Sons, Ltd.

Received 21 November 2005; Accepted 28 May 2006

KEY WORDS: seining; bycatch; population structure; habitat structure; habitat disturbance; underwater visual

census; Hippocampus; Syngnathidae

INTRODUCTION

Intense and frequent demersal fishing activities can alter benthic species abundance and composition (Engeland Kvitek, 1998; Kaiser et al., 1998; Tuck et al., 1998; Jennings et al., 2001; Cryer et al., 2002; Kaiser,2003; but see Drabsch et al., 2001). Depending on the nature, frequency and intensity of exploitation, toweddemersal fishing (e.g. trawling, seining) can reduce the structural complexity of benthic communities bycrushing or removing epifauna and smoothing sedimentary formations (Tuck et al., 1998; Watling andNorse, 1998; Turner et al., 1999; Collie et al., 2000a,b). Towed demersal fishing gears also remove andredistribute macrophytes (Meyer et al., 1999; Baum et al., 2003), which generally support a greaterabundance and diversity of marine fishes than unvegetated habitats (Heck et al., 1989; Edgar and Shaw,1995; Jenkins et al., 1997). When such fishing ceases, the physical damage to soft-sediment forms can bereversed rapidly, while the recovery of species abundance and diversity may occur over longer timescales(e.g. months to years, Kaiser et al., 1998; Tuck et al., 1998).

Fishing with towed demersal gears has the potential to cause population declines in some species withlimited resistance or resilience to exploitation, and population increases in species that exploit habitatsdisturbed by fishing (Casey and Myers, 1998; Engel and Kvitek, 1998; Kaiser et al., 1998; Kaiser, 2003).Fisheries that employ such gears capture a rich diversity of fauna and flora that vary widely in life historystrategies and habitat requirements (Alverson et al., 1994; Stobutzki et al., 2001a,b; Baum et al., 2003).Small (520 cm) benthic marine fishes often dominate bycatch in tropical shrimp trawls (Alverson et al.,1994). Because the life histories and habitat preferences of small fishes are often poorly known (Froese andPauly, 2004), it is difficult to evaluate or predict their responses to strategies for managing demersal fisheries(Walters et al., 1999; Stobutzki et al., 2001a).

Large cumulative catches of seahorses (genusHippocampus, Family Syngnathidae) reported as bycatch intrawl fisheries (e.g. Vincent, 1996; Baum et al., 2003) and high volumes in international trade (Vincent,1996; Giles et al., 2005) have raised conservation concerns for this taxon (Foster and Vincent, 2004; IUCN,2006). Limited biological information, however, restricts the ability to assess the impacts of non-selectiveexploitation on seahorse populations. Specifically, there are no available estimates of removal rates forseahorse populations subject to demersal fishing and few estimates of population abundance (Bell et al.,2003; Curtis and Vincent, 2005). Analysis of H. erectus trawl bycatch suggests that demersal fishing has thepotential to (a) differentially affect cohorts, (b) disrupt social structure by selectively capturing females, (c)reduce reproduction by disrupting pair bonds, (d) damage habitat by removing seagrasses and (e) injureand incidentally kill individuals (Baum et al., 2003). Fishery-independent data, however, were lacking tovalidate these inferences.

The two aims of this study were (1) to evaluate the direct impacts of non-selective fishing with toweddemersal gears on seahorse population structure and (2) to explore indirect effects of such exploitation onpopulations through its influence on the structure of benthic habitats. These aims were addressed using twosympatric species that differ in their habitat preferences. The long-snouted seahorse, H. guttulatus Cuvier1829, prefers structurally complex habitats with high macrophyte and invertebrate cover, while theshort-snouted seahorse, H. hippocampus L, preferentially uses sparsely vegetated habitats (Curtis andVincent, 2005) (Figure 1). Specific objectives were to (a) estimate removal rates, (b) test whether individualsdiffer in vulnerability to demersal fishing according to size, sex, or reproductive status, and (c) quantifychanges in habitat composition and seahorse population structure in response to reduced exploitation. Areduction in demersal fishing effort was predicted to result in an increase in the abundance of seagrasses,

SPECIES-SPECIFIC RESPONSES TO FISHING 469

Copyright # 2006 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 468–484 (2007)

DOI: 10.1002/aqc

macroalgae and benthic invertebrates. The densities of H. guttulatus and H. hippocampus were alsopredicted to increase in response to reductions in fishing mortality, but the magnitude of numerical increasewould be greater for H. guttulatus because of its stronger preference for complex, vegetated habitats. Thesepredictions were investigated using an experimental seining programme coupled with underwater visualcensuses in a seagrass-dominated community.

METHODS

Species descriptions

H. guttulatus (widely called H. ramulosus; Lourie et al., 1999, 2004) and H. hippocampus are associated withshallow macrophyte-dominated communities in the north-eastern Atlantic Ocean and Mediterranean Sea(Boisseau, 1967; Lourie et al., 1999; Perez-Ruzafa et al., 2004; Curtis and Vincent, 2005). Both species arecharacterized by small body size, rapid growth, early age at maturity, small brood size, multiple spawningbehaviour, specialized parental care, and short lifespans (approximately 4 to 6 years) (Boisseau, 1967;Lourie et al., 1999; Curtis, 2004; Foster and Vincent, 2004; Curtis and Vincent, 2006; J. Curtis, unpublisheddata). Juveniles are planktonic for at least 8 weeks (Boisseau, 1967; Perez-Ruzafa et al., 2004) and settleinto benthic habitats at approximately 3 months of age (Boisseau, 1967; Curtis and Vincent, 2006). SettledH. guttulatus range from 65 to 215mm standard length, SL (sum of head, trunk and tail lengths), and weighfrom 0.6 to 22.5 g (wet mass, Curtis and Vincent, 2005, 2006). Settled H. hippocampus range from 45 to166mm SL and weigh 0.1 to 9.5 g (wet mass, Curtis and Vincent, 2005; J. Curtis, unpublished data). Adultsof both species exhibit high site fidelity during multiple years with home ranges averaging 20m2 and 8m2,respectively (Curtis and Vincent, 2006; J. Curtis, unpublished data). Despite differences in landscape-levelhabitat preferences (Figure 1), both species preferentially grasp holdfasts with their prehensile tails (Curtisand Vincent, 2005), presumably to maintain stability and crypsis.

Study site

This study was carried out in the Ria Formosa lagoon in southern Portugal (368 590N, 78 510W) (Figure 2).The Ria Formosa is shallow, highly productive and characterized by macrophyte communities interspersed

Figure 1. Habitat–abundance curves for sympatric European seahorses in the Ria Formosa lagoon. Equations are given for the curvesfitted to densities of Hippocampus guttulatus (solid line) and Hippocampus hippocampus (dashed line) plotted as a function of the

percentage of substrate covered by seagrasses, macroalgae and benthic invertebrates (data from Curtis and Vincent (2005)).

J.M.R. CURTIS ET AL.470


DOI: 10.1002/aqc

with sparsely vegetated sand flats in a network of channels and tidal creeks (Alberto et al., 2001; Curtis andVincent, 2005). The subtidal vegetation was dominated by seagrasses (primarily Cymodocea nodosa, butalso Zostera marina and Z. noltii) and macroalgae (primarily Ulva lactuca and Codium spp.) (Alberto et al.,2001; Curtis and Vincent, 2005). Ria Formosa was chosen as the study site because (a) there was noevidence that populations were at risk or in decline (Curtis, 2004) as suspected in many other seahorsepopulations (Vincent, 1996; Martin-Smith and Vincent, 2005), (b) high densities of H. guttulatus (0.07m�2)and H. hippocampus (0.007m�2) (Curtis and Vincent, 2005) compared with other seahorse populations(reviewed in Foster and Vincent, 2004) favoured detection of seahorse responses to experimental fishing,and (c) a study involving towed demersal fishing gears was under way (Erzini et al., 2002), providing anopportunity for addressing multiple research objectives within the same sampling framework. Moreover,use of towed demersal fishing gears for recreational or commercial purposes in the Ria Formosa isprohibited (Monteiro, 1989; Erzini et al., 2002), so fishing effort at experimental sites was known, constantand controlled.

Experimental fishing

Experimental fishing was carried out using a beach seine (25m long, 3.5m maximum height, 9mm stretchedmesh), for 2 years (October 2000 to October 2002) at a subset of sites sampled by Erzini et al. (2002). In thepresent study 12 treatment sites were experimentally fished each month (i.e. seined 12 times yr�1) during thefirst year (between October 2000 and September 2001), but not fished during the second year (betweenOctober 2001 and October 2002) (sites 1 to 12; Table 1; Figure 2). Five additional sites were included ascontrols with fishing effort kept constant between years: three fished control sites were seined each monthduring both years (sites 13 to 15; Table 1; Figure 2) and two unfished control sites were not fished in eitheryear (sites 16 and 17; Table 1; Figure 2). Additional fished and unfished control sites were planned but notsampled in 2002 for logistic reasons.

During experimental fishing, the outstretched seine was dragged perpendicular to shore forapproximately 10m while one end of the seine was held on shore and the other end was held on amotorized boat. The boat end of the seine was then driven to shore to haul the seine out of the water (Erziniet al., 2002). All seahorses captured with the seine during the first year were frozen for additional analyses,including fecundity estimates and morphometric measurements (Curtis 2004; Curtis and Vincent, 2006), butsome seahorses captured in fished control sites during the second year were haphazardly returned to the

Figure 2. Location of sample sites in the Ria Formosa coastal lagoon in southern Portugal (inset). Shown are sites where fishing effortwas reduced from 2001 to 2002 (sites 1 to 12) and sites where fishing effort was maintained constant between years (fished and unfished

control sites, 13 to 17).



DOI: 10.1002/aqc

lagoon following seining. The fates of released seahorses (i.e. survival, return to home range) were notknown. Thus fishing mortality may have been slightly reduced for some sampling events in fished controlsites during the second year.

Sampling sites were at least 100m apart and located along a gradient of habitat complexity and oceaninfluence (Erzini et al., 2002). All sites were fished or surveyed using underwater visual censuses (UVCs, seebelow) within 2.5 h of low tide. A global positioning system was used to locate sampling sites to withinapproximately 10m during experimental fishing and UVCs.

Catch data included the species, SL, sex, and reproductive status. Injuries were also noted. Specieswere identified using multiple morphological characteristics including head, snout and trunk shape, aswell as colour patterns (Lourie et al., 1999, 2004; Curtis and Vincent, 2005; Curtis, in press). Standardlength was measured as a straight line to the nearest millimetre with a plastic ruler from the cleithralring to the tip of the straightened tail, and with calipers from the tip of the snout to the cleithral ring(Lourie et al., 1999; Curtis and Vincent, 2006). H. guttulatus and H. hippocampus were consideredadults if SL exceeded 135mm and 99mm, respectively. These lengths corresponded to the maximumobserved SL of juvenile males, inferred from the presence of an immature brood pouch (Curtis andVincent, 2006; J. Curtis, unpublished data). Sex of adults was determined using the presence (male) orabsence (female) of a mature brood pouch. Reproductive status was determined by dissecting the male’sbrood pouch (full of eggs or empty) or by visually inspecting the female’s trunk girth (swollen inferred tomean hydrated eggs, concave inferred to mean immature or spent ovaries) (Vincent and Sadler, 1995;Curtis, 2004; Curtis and Vincent, 2006).

Underwater visual census

Sample sites were surveyed using standard underwater visual census (UVC) techniques (English et al.,1994; Samoilys, 1997). All treatment and control sites were surveyed using UVCs once each year except

Table 1. Experimental treatments, underwater visual census (UVC) survey dates and months of catch data used to (A) test forintraspecific differences in vulnerability to the seines and (B) calculate removal rates in 2001. Experimental seining began in October2000. (Reduced fishing effort ¼ seined monthly during first year, not seined during second year; fished control ¼ seined monthly in

both years; unfished control ¼ not seined in either year)

Site Experimental treatment 2001 UVC 2002 UVC A B

1 Reduced fishing efforta 14 Aug Aug–Sept Aug2 Reduced fishing efforta 13 Sept Sept–Oct Sept3 Reduced fishing efforta 13 Sept Sept–Oct Sept4 Reduced fishing effort 30 Aug 30 July Sept–Oct Sept5 Reduced fishing effort 30 Aug 29 July Sept–Oct Sept6 Reduced fishing effort 14 Aug 24 July Sept–Oct Sept7 Reduced fishing effort 18 Sept 28 June Oct–Nov Oct8 Reduced fishing effort 28 Aug 23 July Sept–Oct Sept9 Reduced fishing effort 28 Aug 31 July Sept–Oct Sept10 Reduced fishing effort 4 Sept 29 July Sept–Oct Sept11 Reduced fishing effort 3 Sept 26 July Sept–Oct Sept12 Reduced fishing effort 3 Sept 31 July Sept–Oct Sept13 Control (Fished) 13 Aug 21 June Aug–Sept Aug14 Control (Fished) 4 Sept 14 June Sept–Oct Sept15 Control (Fished) 19 Sept 20 June Sept–Oct Sept16 Control (Unfished) 20 Aug 2 July17 Control (Unfished) 20 Aug 26 June

aNo UVC in 2002 for logistic reasons.



DOI: 10.1002/aqc

for three of the treatment sites, for logistic reasons (sites 1 to 3; Table 1). In 2001, UVCs were carriedout in August or September (i.e. after 11 or 12 seine hauls). In 2002, UVCs were carried out in Juneor July (i.e. 9 or 10 months after seining ceased within treatment sites) (Table 1). On average, UVCswere carried out 13.6 days (�7:3 SD, ranging from 1 to 24 days) following the previous month’s seiningin 2001, and 25 days (�3:2 SD, ranging from 20 to 26 days) following the previous month’s seining in2002.

UVC data came from three belt transects 30m long and 2m wide with a total search area of 180m2 persampling site (Figure 3). The starting points and direction of transects extending from a measuring taperunning perpendicular to shore were randomly selected prior to surveys, but constrained to within 30m ofshore to ensure safety from boat traffic in the narrow channels. The species, trunk length (cleithral ring tothe last trunk ring), sex, reproductive status, and injuries of seahorses on transects were recordedunderwater. Trunk lengths were measured with a plastic ruler and converted to SL using equationsdeveloped for these species (Curtis and Vincent, 2005, 2006). Reproductive status was visually assessed(Vincent and Sadler, 1995; Curtis and Vincent, 2005). All seahorses encountered during UVCs werereleased immediately to capture locations after measuring. Few (51%) seahorses swam away fromobservers during searches (Curtis and Vincent, 2005). Therefore UVCs were not time-constrained and thebenthic habitat was thoroughly searched. The same experienced observer (J.M.R.C.) participated in allUVCs thereby further reducing observer bias.

During UVCs, the three belt transects were used to calculate the percentage cover of seagrasses,macroalgae and conspicuous epifauna (>2.5 cm in height, width or length), using the line intercept method(English et al., 1994; Samoilys, 1997; Curtis and Vincent, 2005). A dimensionless index of habitatcomplexity, Ct/At (Bartholomew et al., 2000), was calculated, where Ct/At was defined as the proportion ofthe transect line that overlay macrophytes or invertebrates.

Statistical analyses

Vulnerability to seining

Intraspecific differences in vulnerability (by size, sex, or reproductive status) were examined with pairedtests (t-tests for size comparisons, Wilcoxon signed ranks tests for other aspects of population structure) by

Figure 3. Schematic of seine area (light grey) and UVC transects (dark grey) at sampling sites. An anchor was set at approximately30m from the point on shore closest to the GPS coordinates. A measuring tape laid out from the anchor to shore served threepurposes: (a) to locate the randomly selected starting points of the UVC transects, (b) to ensure that starting points of transects on thesame side of the measuring tape were at least 5m apart to prevent overlap in search areas and (c) to allow scuba divers to navigatesafely under low visibility conditions in the narrow channels. Transects extended beyond the seine area to allow for error in the GPScoordinates (approximately � 10m). The area accounting for this source of uncertainty is bounded by the dotted lines. The 30-mtransects also allowed us to sample seahorses whose home ranges overlapped with the swept area. Most individuals probably stayedwithin 25m (Curtis and Vincent, 2006) of the fished site (i.e. within area encompassed by dashed lines). Tidal currents flowed

perpendicular to shore except at slack tide.



DOI: 10.1002/aqc

comparing, within each site, the mean and maximum SL, sex ratios, and proportions of reproductivelyactive males and females observed during UVCs (UVC data) with corresponding data from individualscaptured subsequently with the seine (catch data). These paired tests included 2001 UVC and 2001 catchdata from sites 1 to 15 (Table 1); data collected in 2002 were excluded from this analysis because of thehaphazard release of some individuals. Catch data from the first two months of seining following the 2001UVCs in these 15 sites were pooled to increase the power of paired tests between UVC and catch data(Table 1). In all comparisons of size structure (including those described below), only adults were usedbecause the recruitment of juveniles is seasonal and begins in late summer or early autumn (Curtis andVincent, 2006).

Some paired t-tests and Wilcoxon signed ranks tests for evaluating intraspecific variation in vulnerabilitywere not carried out owing to small sample sizes in the catch or UVC data. Therefore an additional set ofcomparisons was carried out. UVC data and catch data collected from the 15 sites in 2001 were pooled,respectively, to test for overall differences in population structure using either Kolmogorov–Smirnov tests(for differences in length frequency distributions) or chi-squared tests.

The removal rate, q, of H. guttulatus and H. hippocampus was estimated within each site as:

q ¼Dcatch

DUVC

DUVC was the density of individuals observed during the UVC. Dcatch was the number of H. guttulatus orH. hippocampus captured per square metre seined during the first month following the UVC, assuming aseine area of 300m2 (Erzini et al., 2002). The potential influence of habitat structure (i.e. Ct/At) or seahorsedensity on q was evaluated using Spearman rank correlations.

Habitat responses to reduced fishing effort

UVC data from 2001 and 2002 were employed to evaluate the influence of a reduction in seining effort onhabitat structure. Differences in the percentage cover of seagrasses, macroalgae and invertebrates, Ct/At,and species richness between years were compared with two sets of paired tests (Wilcoxon signed rankstests). The first set of paired tests was applied to the nine treatment sites surveyed with UVC in both years(sites 4 to 12; Table 1). The second set of paired tests was applied to the five control sites (sites 13 to 17;Table 1).

Population responses to reduced fishing effort

The population responses of H. guttulatus and H. hippocampus to a reduction in fishing effort wereevaluated with UVC data by comparing changes in population structure (density, mean and maximum SL,sex ratio, proportion of reproductively active males and females, proportion of individuals with injuries)within sites from 2001 to 2002. As above, two sets of paired tests were used to evaluate differences betweenyears in response to a reduction in seining effort: the first set of paired tests was applied to the ninetreatment sites and the second set was applied to the five control sites. Some paired tests were not carriedout because of small sample sizes in the 2001 or 2002 UVC data. Therefore data were pooled amongtreatment sites within years, respectively, and among control sites within years, respectively, to test foroverall differences in population structure using either Kolmogorov–Smirnov tests (for differences in lengthfrequency distributions) or chi-squared tests.

All statistical analyses were carried out using SPSS 10.0.7 (SPSS Inc.). Means are reported with standarderrors (SE) except where stated otherwise.



DOI: 10.1002/aqc

RESULTS


In general, vulnerability to seining did not differ among individuals according to SL, sex or reproductivestatus (Tables 2 and 3). However, when 2001 UVC data were pooled among sites and compared with thepooled 2001 catch data, the proportion of male H. guttulatus with full pouches and proportion of femaleH. hippocampus with hydrated eggs were significantly greater in the UVC data than in the catch data(Table 3). The proportion of individuals with injuries (50.02) did not differ between the UVC and catchdata. None of the injuries observed in the catch data (primarily missing tail tips) appeared to be recent orsustained during seining.

H. guttulatus was more likely than H. hippocampus to be captured by the seine. The removal rate of H.guttulatus averaged 0.078� 0.085 (ranging from 0 to 0.254), while on average, H. hippocampus wereremoved at a rate of 0.007� 0.026 (ranging from 0 to 0.092). Removal rates were not related to Ct/At (H.guttulatus rs ¼ �0:006; p ¼ 0:986; H. hippocampus rs ¼ 0:223; p ¼ 0:463) or density (H. guttulatus rs ¼0:169; p ¼ 0:542; H. hippocampus rs ¼ �0:044; p ¼ 0:819) for either species. The number of individualscaptured per seine haul (catch per unit effort, CPUE), however, was significantly correlated with density forboth H. guttulatus ðrs ¼ 0:629; p ¼ 0:012Þ and H. hippocampus ðrs ¼ 0:634; p ¼ 0:011Þ:

Table 2. Intraspecific differences in vulnerability to seines. Paired comparisons are of Hippocampus guttulatus population structurebetween 2001 UVC data and 2001 catch data (paired t-tests for size structure where n is the degrees of freedom; Wilcoxon signed rankstests for all other comparisons where n is the number of sites included in the analyses). Insufficient data were available to test for

differences in Hippocampus hippocampus population structure using paired t-tests or Wilcoxon signed ranks tests

Population parameter Statistic n p

Mean size 1.608 9 0.142Max size 2.061 9 0.069a

Sex ratio 0.514 6 0.600Brooding males �1.521 7 0.128Hydrated females �1.604 4 0.109Injuries �1.095 10 0.273

aMaximum size usually greater in UVCs.

Table 3. Intraspecific vulnerability of Hippocampus guttulatus and Hippocampus hippocampus to seining. Size structure, sex ratio, theproportion of males with full pouches, the proportion of females with hydrated eggs and proportion of individuals with injuries werecompared between 2001 UVC data and 2001 catch data (all individuals pooled within UVC and catch data, respectively). Chi-squared

tests were used, except for a Kolmogov–Smirnov test for differences in length–frequency distributions

Population parameter Hippocampus guttulatus Hippocampus hippocampus

Statistic n p Statistic n p

Size distribution 0.190 270 0.090 0.390 74 0.181Sex ratio 2.705 384 0.100 0.932 76 0.334Brooding males 19.742 198 0.0001a 2.600 43 0.529Hydrated females 2.636 183 0.104 7.044 35 0.008a

Injuries 0.311 375 0.576 0.341 81 0.559

aLower proportion of brooding Hippocampus guttulatus males and Hippocampus hippocampus females with hydrated eggs in catchdata.



DOI: 10.1002/aqc


Changes in seining effort had significant impacts on the structure of the benthic habitat. As predicted, therewas a significant increase in Ct/At when fishing effort ceased within treatment sitesðn ¼ 9;Z ¼ 2:199; p ¼ 0:028) (Figure 4), resulting in an approximate doubling of covered habitat from2001 to 2002. The increase in Ct/At in treatment sites was primarily attributable to a significant increase inthe percentage cover of drifting or mobile invertebrates ðZ ¼ 2:240; n ¼ 9; p ¼ 0:025) as well as an increasein the percentage cover of attached and drifting macroalgae (Figure 5). The percentage cover of seagrassesdecreased on average in treatment sites, but changes in seagrass and macroalgae abundance were notstatistically significant (all jZj51:345; all p > 0:180) (Figure 5). Species richness ranged from 3 to 28 speciesduring UVCs, but did not change when seining ceased (Z ¼ 0:604; n ¼ 9; p ¼ 0:546).

Figure 4. Changes in the proportion of substrate covered by all macrophytes and invertebrates (Ct/At, mean� SE) in control sitesðn ¼ 5Þ and in response to a reduction in experimental fishing effort (treatment sites, n ¼ 9) from 2001 to 2002.

Figure 5. Changes in the composition of seagrasses, macrophytes and invertebrates in treatment sites surveyed with UVC techniques in2001 (black bars) and 2002 (grey bars). Macrophytes and invertebrates with holdfasts are plotted separately from those withoutholdfasts to demonstrate the influence of attachment capability on responses to a reduction in seining effort from 2001 to 2002 (only

taxa with 53% cover are listed).



DOI: 10.1002/aqc

Habitat structure remained constant between years within control sites. No differences in Ct/At

ðZ ¼ 0:542; n ¼ 5; p ¼ 0:588Þ (Figure 4), the percentage cover of invertebrates ðZ ¼ 1:604; n ¼ 5; p ¼ 0:109),the percentage cover of macrophytes (all jZj51:345; all p > 0:180) or species richness ðZ ¼ �1:342; n ¼ 5;p ¼ 0:180) were observed between years. Fished and unfished control sites did not differ in Ct/At or thepercentage cover of seagrasses, macroalgae or invertebrates during 2001 or 2002 UVCs (Mann–Whitneytests, all U52:0; n ¼ 5; all p50:8).


The numerical responses of H. guttulatus and H. hippocampus differed in both magnitude and directionwhen seining ceased. The density of H. guttulatus increased significantly by an average of 270% withintreatment sites from 2001 to 2002 ðZ ¼ �1:96; n ¼ 9; p ¼ 0:05Þ (Figure 6). By contrast, the density ofH. hippocampus decreased significantly by an average of 65% within treatment sites ðZ ¼ �1:97; n ¼9; p ¼ 0:049Þ: The density of both H. guttulatus (Z ¼ 0:674; n ¼ 5; p ¼ 0:5) and H. hippocampus (Z ¼ 0:00;n ¼ 5; p ¼ 1:000) remained constant within control sites from 2001 to 2002. When data from treatment sitesand control sites were pooled, respectively, within years, similar differences in relative frequencies wereobserved: the relative frequency of H. guttulatus significantly increased in response to a reduction in fishingeffort (w2 ¼ 28:43; df ¼ 1; p50:0001), but did not change in control sites (w2 ¼ 0:03; df ¼ 1; p ¼ 0:862).

The magnitude of increase in H. guttulatus abundance (density in 2002 divided by density in 2001) withintreatment sites was significantly and positively correlated with the magnitude of change in Ct/At (rs ¼ 0:786;n ¼ 9; p ¼ 0:012). The change in abundance of H. hippocampus was not correlated with the magnitude ofchange in any habitat component (all p > 0:209). Fished and unfished controls did not differ in initial andfinal densities of H. guttulatus (Mann–Whitney tests, all U41:0; n ¼ 5; all p50:4) or H. hippocampus(Mann–Whitney tests, all U52:0; n ¼ 5; all p50:8).

Figure 6. Mean (� SE) changes in Hippocampus guttulatus and Hippocampus hippocampus densities in control sites ðn ¼ 5Þ and inresponse to a reduction in fishing effort (n ¼ 9 treatment sites) from 2001 to 2002.



DOI: 10.1002/aqc

In paired tests, SL, sex ratios, proportions of brooding males, proportions of hydrated females andproportions of injured H. guttulatus did not change in response to reduced fishing effort in treatment sites,nor were there differences between years in control sites (Table 4). Similarly, H. hippocampus populationstructure did not differ between years within treatment sites (insufficient data for control sites), when pairedtests were employed. Increasing power by pooling all individuals within treatment sites, or within control

Table 4. Paired comparisons are of size structure, sex ratio, proportion of males with full pouches, proportion of females withhydrated eggs, and proportion of individuals with injuries between years in treatment and control sites. All paired tests were Wilcoxonsigned ranks tests, except comparisons of mean and maximum sizes, which were tested with paired t-tests. The number of sites includedin the comparison (or degrees of freedom for t-tests) is given by n. There were insufficient data to test for changes in Hippocampus

hippocampus population structure within control sites using paired tests



Treatment SitesMean size 1.404 6 0.210 0.244 4 0.820Max size �1.041 6 0.338 1.165 4 0.309Sex ratio 0.105 7 0.917 �1.461 5 0.144Brooding males 1.782 6 0.075 �0.535 4 0.593Hydrated females �0.944 6 0.345 �0.447 5 0.655Injuries �0.730 8 0.465 �1.342 5 0.180

Control sitesMean size 0.193 4 0.856Max size �1.542 4 0.193Sex ratio 0.674 5 0.500Brooding males 0.148 5 0.138Hydrated females �0.535 4 0.593Injuries 0.757 4 0.491

Table 5. Population-level responses to reduced fishing effort. Size structure, sex ratio, the proportion of males with full pouches, theproportion of females with hydrated eggs and the proportion of individuals with injuries were compared between years with individualspooled among treatment and control sites, respectively (chi-squared tests, except for Kolmogov–Smirnov tests for length–frequencydistributions). Tests indicated by an asterisk were significant after a sequential Bonferroni correction was used to adjust p-values for

multiple tests



Treatment sitesSize distribution 0.154 220 0.187 0.605 51 0.858Sex ratio 0.035 292 0.859 0.071 67 0.789Brooding males 10.358 149 0.001* 1.933 41 0.164Hydrated females 24.054 143 50.0001* 0.036 29 0.847Injuries 2.291 384 0.130 0.053 68 0.817

Control sitesSize distribution 0.144 267 0.175 0.996 14 0.274Sex ratio 0.160 318 0.689 0.126 23 0.721Brooding males 6.2101 162 0.012y 0.185 19 0.666Hydrated females 12.90 153 0.0003* 0.000 10 1.000Injuries 1.804 269 0.179 0.203 25 0.651

*yGreater proportion of brooding males or females with hydrated eggs in 2002.



DOI: 10.1002/aqc

sites, did not reveal significant differences in SL, sex ratios or the proportions of injured individuals betweenyears for either species (Table 5). However, when all individuals within treatment sites were pooled betweenyears, the proportions of reproductively active male and female H. guttulatus increased significantly afterfishing ceased. Because the proportions of reproductively active males and females also increased betweenyears in the control sites (Table 5), these comparisons were probably confounded by seasonal variation inreproductive activity (Curtis, 2004; Curtis and Vincent, 2006) (Figure 7).

DISCUSSION

Most studies investigating the impacts of non-selective, towed demersal fishing gears focus on heavy gearsdragged over large areas (e.g. Engel and Kvitek, 1998; Kaiser et al., 1998; Cryer et al., 2002; Baum et al.,2003). Our results suggest that even repeated use of light demersal fishing gears can influence the structureof non-target species under exploitation, both directly through fishing mortality and indirectly throughhabitat alteration. Although both seahorse species currently studied were vulnerable to capture, a reductionin seining effort resulted in an increase in the abundance of H. guttulatus, but a decrease in the abundanceof the sympatric H. hippocampus. Interspecific differences in numerical responses were correlated withchanges in preferred habitat.

Figure 7. (A) Seasonal pattern in reproductive activity of male Hippocampus guttulatus (solid line; data from Curtis and Vincent(2006)) and Hippocampus hippocampus (dashed line; J. Curtis, unpublished data). The time periods during which 2001 UVC data, 2001catch data and 2002 UVC data were collected are indicated with arrows (see Table 1 for details). (B) No seasonal patterns in density(circles), mean size (diamonds), sex ratio (triangles), or proportion of individuals with injuries (squares) were observed for eitherHippocampus guttulatus (solid symbols) or Hippocampus hippocampus (open symbols) during a mark–recapture study (data from 2001

to 2003 pooled by month from Curtis (2004) and Curtis and Vincent (2006)).



DOI: 10.1002/aqc


There was little evidence that size, sex or reproductive status influenced the vulnerability of individuals toseining. This is an unusual finding suggesting that catch data collected in a similar manner (with small-meshedseines dragged over soft-bottom communities) could provide useful indices of seahorse population structurein situ. Fisheries-dependent data are usually biased with respect to the structure of exploited populationsbecause of intraspecific differences in vulnerability to fishing gears (Hilborn and Walters, 1992; King, 1995).Larger individuals are selectively captured in many fisheries because they are preferentially targeted andbecause smaller individuals escape through the mesh of nets and traps (King, 1995). Size structure of catchdata may also be biased because of the spatial distribution of fishing effort relative to that of different cohorts(Hilborn and Walters, 1992; King, 1995; Baum et al., 2003). In this study, both settled juveniles and adultswere longer and wider than the 9mm stretched mesh of the seine; therefore the seine probably did not exertsize-selectivity on the catch data. Because settled juvenile and adultH. guttulatus and H. hippocampus occupysimilar habitats (Curtis and Vincent, 2005) it is unlikely that the spatial distribution of seining influenced theselife-history stages differentially, as inferred for H. erectus subject to trawling (Baum et al., 2003).

The sex ratio in catch data would be expected to differ from that observed in situ if behavioural differencesmade one sex more likely to be captured than the other. No such evidence was found; similar sex ratios in thecatch and UVC data were consistent with similar patterns in distribution, habitat preference and activity ofmale and female H. guttulatus and H. hippocampus (Curtis and Vincent, 2005, 2006). Catch data maytherefore provide a useful index of European seahorse sex ratios in situ. The sex-selectivity of shrimp trawlsinferred by Baum et al. (2003) may have derived from their methodology: the authors used the size of thesmallest male with evidence of a brood pouch as the benchmark for distinguishing between juvenile and adultH. erectus (versus size of the largest juvenile as used in this study), making the authors more likely to identifyjuvenile males as females, and therefore to calculate female-biased sex-ratios in their trawl bycatch.

The proportion of reproductively active individuals in catch data would be expected to differ from thatobserved in situ if behavioural differences influenced the vulnerability of reproducing and non-reproducingindividuals to the fishing gear. In two of six comparisons there was a greater proportion of reproductivelyactive individuals in situ than in the catch data. These differences, however, probably reflect seasonalvariation in reproductive activity, rather than a lower vulnerability of reproductively active individuals toseining. The catch data included in these analyses were collected towards the end of the reproductive seasonand up to two months after the UVCs were carried out. The proportion of male seahorses reproducingdeclines naturally from approximately 0.75 in August to 0.15 in October (Figure 7), information that onlybecame available after our study was carried out. To further investigate the influence of reproductiveactivity on vulnerability to non-selective demersal fishing, UVCs in similar studies should be followed withexperimental fishing in close temporal sequence.

Although the removal rates of H. guttulatus and H. hippocampus were low with the seine (maximum of0.25 and 0.1, respectively), frequent and sustained demersal fishing could potentially lead to localpopulation extirpations if immigration of benthic juveniles or adults was insufficient to offset fishingmortality. A significant correlation between CPUE and densities observed in situ suggests that catch dataobtained in a similar manner to this study may also provide useful indices of relative abundances forrecruited (benthic) juvenile and adult European seahorses in soft-bottom communities (though not ofplanktonic juveniles). Abundance indices would be useful for monitoring the population trends of thesefishes of conservation concern (Santos et al., 1995; IUCN, 2006), given the significant logistic challenges ofcarrying out UVCs over large areas (Samoilys, 1997).


Seining appeared to have a stronger influence on benthic organisms with poor attachment capabilities. Despitea high frequency of seining (12yr�1) and presence of seagrass leaves in the seine catches, no significant changes



DOI: 10.1002/aqc

in the percentage cover of seagrasses were observed in either treatment or control sites. Similarly, no significantincreases in the percentage cover of most species of macroalgae were observed when seining ceased. Bycontrast, the percentage cover of Ulva lactuca, a macroalgal species that does not usually anchor to a holdfast,increased by almost 350% when seining ceased. Similarly, the percentage cover of invertebrates with poorattachment capabilities (e.g. colonies of the bryozoan Zoobotryon verticillatum and sea urchins) were morelikely to increase when fishing ceased than species with secure holdfasts (e.g. tunicates, tubeworms andsponges). Poor attachment capability probably influenced the abundance of these species by making themmore vulnerable to being swept up by the seine as it was being dragged along the bottom. Seahorses, with theirspecialized prehensile tails and ability to grasp holdfasts, are probably less vulnerable to capture using similartowed demersal fishing gears than fishes that lack the ability to grasp holdfasts or anchor themselves to thebottom. These morphological and behavioural features of seahorses may account for the relatively lowremoval rates of H. guttulatus and H. hippocampus observed in this study.


Frequent seining did not appear to influence size structure, alter sex ratio, disrupt reproduction, or be theprimary cause of injuries for either H. guttulatus or H. hippocampus. Apart from significant effects ondensity, population structure remained constant whether seining effort was reduced or sustained from 2001to 2002. Heavier trawling gears are likely to be more disruptive (e.g. greater removal rates, more injuries)than the light weight seine employed in this study. Approximately 14% of injured H. erectus appeared tohave sustained wounds } primarily broken tails } while being captured in beam trawls (Baum et al.,2003). The frequency of demersal trawling in areas where significant biological impacts have been observedcan range from as little as 0.2 to 6.5 yr�1 (Jennings et al., 2001). Given that the majority of seahorses ininternational trade are captured as trawl bycatch (Vincent, 1996), future fisheries-independent studies arewarranted to estimate the removal rates and population-level impacts of seahorses exploited with heaviertowed demersal fishing gears over wider spatial scales. Seahorses may be more vulnerable to injuries,mortality and disruption of reproduction in habitats that are disturbed by heavy trawls deployed for longerperiods and over greater areas (Baum et al., 2003).

Although both H. guttulatus and H. hippocampus have similar life histories, their numerical responsesdiffered in both magnitude and direction when seining effort was reduced. Given interspecific differences inhabitat preference and the effects of seining on habitat structure, a reduction in seining effort appeared tohave resulted in greater habitat cover and favoured higher densities of H. guttulatus, while repeated seining,which maintained more open habitats, favoured greater abundances of H. hippocampus. It is unclearwhether numerical responses to reduced fishing effort arose from changes in local recruitment dynamics (i.e.habitat selection of juveniles recruiting to the lagoon), or from the redistribution of individuals within andsurrounding the treatment sites (Kramer and Chapman, 1999; Rodwell et al., 2003). Rapid juvenile growth(Curtis and Vincent, 2006) coupled with a lack of mark–recapture data made it difficult to distinguishbetween new recruits, resident adults and immigrating adults from the UVC data. Preliminary data,however, suggest that changes in density probably resulted from new recruitment of juveniles based onhabitat selection. Tagged H. guttulatus and H. hippocampus adults monitored at a different site within RiaFormosa exhibited high site fidelity and displaced home ranges between years by an average of 2m(Curtis and Vincent, 2006; J. Curtis, unpublished data) despite apparently suitable and unoccupied habitatsurrounding the focal study site.

IMPLICATIONS FOR MANAGEMENT AND CONSERVATION

Differences in life history and ecology are likely to influence species-specific population-level responses ofmarine fishes to changes in fishing effort, including full spatial closures to fishing (Mosqueira et al., 2000).



DOI: 10.1002/aqc

Given the extensive habitat damage and challenges of managing multiple species ranging widely inlife-history strategies, the use of marine protected areas is a widely supported ecosystem-based strategy forpromoting the recovery of benthic communities and their associated fish assemblages (Walters et al., 1999),including seahorses (Martin-Smith et al., 2004). Differential responses to towed demersal fishing means thatthere may be important trade-offs to consider when implementing marine protected areas to manage thedirect effects of exploitation on mortality and population structure, and the indirect effects of fishingdisturbance on benthic habitats. In this study, our results suggest that management actions that promote anincrease in habitat complexity may benefit H. guttulatus, but lead to declines of H. hippocampus unless themanagement strategy also provides for the maintenance of more open habitats. Given that both species areof conservation concern (Santos et al., 1995; Foster and Vincent, 2004) and potentially subject to a varietyof non-selective towed demersal fishing gears (Vincent, 1996), this is an important trade-off to considerwhen developing conservation strategies for these species. In general, simultaneous attempts to protectincidentally exploited populations of sympatric species may involve different management strategies thatsustain a diversity of habitat types.

ACKNOWLEDGEMENTS

This is a contribution from Project Seahorse. We gratefully acknowledge field and laboratory assistance from ProjectSeahorse research assistants and volunteers: K. Bigney, H. Balasubramanian, B. Gunn, S. Lemieux, C.-M. Lesage,J. Nadeau, S. Overington, A.M. Santos, S.V. Santos, M. Veillette and K. Wieckowski. We thank those involved in theseining surveys, particularly G. Carvalho, P. Feijoo, M. Silva and J. Soares, as well as members of the Coastal FisheriesResearch Group (CFRG) at the Universidade do Algarve: L. Bentes, R. Coelho, C. Correia, J. Gonc-alves, P. Lino andP. Monteiro. Members of the Parque Natural da Ria Formosa provided valuable logistical support, as did F. Fiesta.We also thank K. Martin-Smith and anonymous reviewers for valuable comments on this manuscript. This work wasfunded in part by Guylian Chocolates, Belgium, and the Commission of the European Communities (Project reference:DG XIV C1/99/061). J.M.R.C. held a Natural Sciences and Engineering Research Council of Canada (NSERC)Post-Graduate Scholarship, a Fonds de Recherches sur la Nature et les Technologies Quebec (FCAR) doctoral award,and a JW McConnell McGill Major Fellowship. The John G. Shedd Aquarium in Chicago kindly supported A.V.’soperational costs through its partnership for marine conservation with Project Seahorse. Research was carried out inaccordance with standards set out by McGill University’s Animal Care Committee and the Parque Natural da RiaFormosa.

REFERENCES

Alberto F, Mata L, Santos R. 2001. Genetic homogeneity in the seagrass Cymodocea nodosa at its northern Atlanticlimit revealed through RAPD. Marine Ecology Progress Series 221: 299–301.

Alverson DL, Freeberg MH, Pope JG, Murawski SA. 1994. A global assessment of fisheries bycatch and discards. FAOFisheries Technical Paper No. 339, Food and Agriculture Organization, Rome.

Bartholomew A, Diaz RJ, Cicchetti G. 2000. New dimensionless indices of structural habitat complexity: predicted andactual effects on a predator’s foraging success. Marine Ecology Progress Series 206: 45–58.

Baum JK, Meeuwig JJ, Vincent ACJ. 2003. Bycatch of lined seahorses (Hippocampus erectus) in a Gulf of Mexicoshrimp trawl fishery. Fishery Bulletin 101: 721–731.

Bell E, Lockyear J, McPherson J, Marsden AD, Vincent ACJ. 2003. First field studies of an Endangered South Africanseahorse, Hippocampus capensis. Environmental Biology of Fishes 67: 35–46.

Boisseau J. 1967. Regulations hormonales de l’incubation chez un vertebre male: recherches sur la reproduction del’Hippocampe. PhD thesis, l’Universite de Bordeaux, France.

Casey JM, Myers RA. 1998. Near extinction of a large widely distributed fish. Science 281: 690–692.Collie JS, Escanero GA, Valentine PC. 2000a. Photographic evaluation of the impacts of bottom fishing on benthicepifauna. ICES Journal of Marine Science 57: 987–1001.

Collie JS, Hall SJ, Kaiser MJ, Poiner IR. 2000b. A quantitative analysis of fishing impacts on shelf-sea benthos. Journalof Animal Ecology 69: 785–798.



DOI: 10.1002/aqc

Cryer M, Hartill B, O’Shea S. 2002. Modification of marine benthos by trawling: toward a generalization for the deepocean? Ecological Applications 12: 1824–1839.

Curtis JMR. 2004. Life history, ecology and conservation of European seahorses. Doctoral thesis, McGill University,Montreal, Canada.

Curtis JMR. In press. A case of mistaken identity: skin filaments are unreliable for identifying Hippocampus guttulatusand Hippocampus hippocampus. Journal of Fish Biology.

Curtis JMR, Vincent ACJ. 2005. Distribution of sympatric seahorse species along a gradient of habitat complexity in aseagrass-dominated community. Marine Ecology Progress Series 291: 81–91.

Curtis JMR, Vincent ACJ. 2006. Life history of an unusual marine fish: survival, growth and movement patterns of theEuropean long-snouted seahorse, Hippocampus guttulatus (Cuvier 1829). Journal of Fish Biology 68: 707–733. DOI:10.1111/j.1095-8649.2006.00952.x.

Drabsch SL, Tanner JE, Connell SD. 2001. Limited infaunal response to experimental trawling in previously untrawledareas. ICES Journal of Marine Science 58: 1261–1271.

Edgar GJ, Shaw CU. 1995. The production and trophic ecology of shallow-water fish assemblages in southernAustralia I. Species richness, size-structure and production of fishes in Western Port, Victoria. Journal ofExperimental Marine Biology and Ecology 194: 53–81.

Engel J, Kvitek R. 1998. Effects of otter trawling on a benthic community in Monterey Bay National Marine Sanctuary.Conservation Biology 12: 1204–1214.

English S, Wilkinson C, Baker V. 1994. Survey Manual for Tropical Marine Resources. ASEAN-Australia ScienceProject, Australian Institute of Marine Science: Townsville, Australia.

Erzini K, Bentes L, Coelho R, Correia C, Lino PG, Monteiro P, Ribeiro J, Goncalves JMS. 2002. Recruitment of seabreams (Sparidae) and other commercially important species in the Algarve (southern Portugal), Final Report.Commission of the European Communities DG XIV/C/1.

Foster SJ, Vincent ACJ. 2004. The life history and ecology of seahorses: implications for conservation andmanagement. Journal of Fish Biology 65: 1–61.

Froese R, Pauly D (eds). 2004. FishBase. World Wide Web electronic publication. www.fishbase.org, Version 05/2004.Giles BG, Ky TS, Hoang DH, Vincent ACJ. 2005. The catch and trade of seahorses in Vietnam. Biodiversity andConservation. Published online. DOI: 1007/s10531-005-2432-6.

Heck KL, Able KW, Fahay MP, Roman CT. 1989. Fishes and decapod crustaceans of Cape Cod eelgrass meadows –species composition, seasonal abundance patterns and comparison with unvegetated substrates. Estuaries 12: 59–65.

Hilborn R, Walters CJ. 1992. Quantitative Fisheries Stock Assessment: Choice, Dynamics and Uncertainty. Kluwer:Massachussetts, USA.

IUCN (World Conservation Union). 2006. 2006 IUCN Red List of Threatened Species. hwww.redlist.orgi. [AccessedJuly 2006.]

Jenkins GP, May HMA, Wheatley MJ, Holloway MG. 1997. Comparison of fish assemblages associated with seagrassand adjacent unvegetated habitats of Port Phillip Bay and Corner Inlet, Victoria, Australia, with emphasis oncommercial species. Estuarine, Coastal and Shelf Science 44: 569–588.

Jennings S, Pinnegar JK, Polunin NVC, Warr KJ. 2001. Impacts of trawling disturbance on the trophic structure ofbenthic invertebrate communities. Marine Ecology Progress Series 213: 127–142.

Kaiser MJ. 2003. Detecting the effects of fishing on seabed community diversity: importance of scale and sample size.Conservation Biology 17: 512–520.

Kaiser MJ, Edwards DB, Armstrong PJ, Radford K, Lough NEL, Flatt RP, Jones HD. 1998. Changes in megafaunalbenthic communities in different habitats after trawling disturbance. ICES Journal of Marine Science 55: 353–361.

King M. 1995. Fisheries Biology, Assessment and Management. Fishing News Books: Oxford.Kramer DL, Chapman MR. 1999. Implications of fish home range size and relocation for marine reserve function.Environmental Biology of Fishes 55: 65–79.

Lourie SA, Vincent ACJ, Hall HJ. 1999. Seahorses: An Identification Guide to the World’s Species and theirConservation. Project Seahorse: London.

Lourie SA, Foster SJ, Cooper EWT, Vincent ACJ. 2004. A Guide to the Identification of Seahorses. Project Seahorseand TRAFFIC North America: Washington, DC. Available from www.projectseahorse.org (accessed June 2005).

Martin-Smith KM, Vincent ACJ. 2005. Seahorse declines in the Derwent Estuary, Tasmania, in the absence of fishingpressure. Biological Conservation 123: 533–545.

Martin-Smith KM, Samoilys MA, Meeuwig JJ, Vincent ACJ. 2004. Collaborative development of management optionsfor an artisanal fishery for seahorses in the central Philippines. Ocean and Coastal Management 47: 165–193.

Meyer DL, Fonseca MS, Murphey PL, McMichael RH, Byerly MM, LaCroix MW, Whitfield PE, Thayer GW. 1999.Effects of live-bait shrimp trawling on seagrass beds and fish bycatch in Tampa Bay, Florida. Fishery Bulletin 97:193–199.



DOI: 10.1002/aqc

Monteiro C. 1989. La faune ichthyologique de la Ria Formosa; organisation spatio-temporelle. PhD thesis, UniversiteMontpellier, France.

Mosqueira I, Cote IM, Jennings S, Reynolds JD. 2000. Conservation benefits of marine reserves for fish populations.Animal Conservation 3: 321–332.

Perez-Ruzafa A, Quispe-Becerra JI, Garcıa-Charton JA, Marcos C. 2004. Composition, structure and distribution ofthe ichthyoplankton in a Mediterranean coastal lagoon. Journal of Fish Biology 64: 202–218.

Rodwell LD, Barbier EB, Roberts CM, McClanahan TR. 2003. The importance of habitat quality for marinereserve–fishery linkages. Canadian Journal of Fisheries and Aquatic Sciences 60: 171–181.

Samoilys MA. 1997. Underwater visual census surveys. In Manual for Assessing Fish Stocks on Pacific Coral Reefs,Samoilys MA (ed.). Queensland Department of Primary Industries: Townsville, Australia; 16–29.

Santos RS, Hawkins S, Monteiro LR, Alves M, Isidro EJ. 1995. Marine research, resources and conservation in theAzores. Aquatic Conservation: Marine and Freshwater Ecosystems 5: 311–354.

Stobutzki I, Miller M, Brewer D. 2001a. Sustainability of fishery bycatch: a process for assessing highly diverse andnumerous bycatch. Environmental Conservation 28: 167–181.

Stobutzki IC, Miller M, Jones P, Salini JP. 2001b. Bycatch diversity and variation in a tropical Australian penaeidfishery; the implications for monitoring. Fisheries Research 53: 283–301.

Tuck ID, Hall SJ, Robertson MR, Armstrong E, Basford DJ. 1998. Effects of physical trawling disturbance in apreviously unfished sheltered Scottish sea loch. Marine Ecology Progress Series 162: 227–242.

Turner S, Thrush S, Hewitt J, Cummings V, Funnell G. 1999. Fishing impacts and the degradation or loss of habitatstructure. Fisheries Management and Ecology 6: 401–420.

Vincent ACJ. 1996. The International Trade in Seahorses. TRAFFIC International: Cambridge, UK.Vincent ACJ, Sadler LM. 1995. Faithful pairbonds in wild seahorses, Hippocampus whitei. Animal Behaviour 50:1557–1569.

Walters C, Pauly D, Christensen V. 1999. Ecospace: prediction of mesoscale spatial patterns in trophic relationships ofexploited ecosystems, with emphasis on the impacts of marine protected areas. Ecosystems 2: 539–554.

Watling L, Norse EA. 1998. Disturbance of the seabed by mobile fishing gear: a comparison to forest clearcutting.Conservation Biology 12: 1180–1997.



DOI: 10.1002/aqc

Documents

A conservation trade-off? Interspecific differences in seahorse responses to experimental changes in fishing effort