Present status of the multi-gear shrimp fishery off the west coast of Sri Lanka: gear-based species diversity and selectivity

Summary

Catch statistics were monitored from well established small-scale shrimp fisheries in Negombo lagoon and the adjacent coast in western Sri Lanka, in order to evaluate resource usage, gear selectivity, and spatio-temporal dynamics of catches and CPUE. A total of 55 species, representing 35 families, including 13 shrimp species were recorded from 3546 samples obtained weekly during January 2009-April 2010, for nine types of gear in six fishing grounds. Special emphasis was on shrimp catches: four main shrimp species, Metapenaeus dobsoni, Fenneropenaeus indicus, Parapenaeopsis coromandelica and Penaeus semisulcatus, represented 82% of the total shrimp landings. Catch per unit effort (CPUE) differed among fishing grounds, months and gear types. Species diversity differed among the gear chosen. Hierarchical cluster analysis based on presence-absence of the species data of catches showed that clustering was based on habitat rather than on the fishing gear. Species composition analysed with a Detrended Correspondence Analyses over months and fishing grounds showed a distinction of trawl gear from the remainder of the gear operated in the lagoon. The information presented is of importance for evaluation of the present status of the shrimp fishery and for developing management strategies based on the types of gear.

Introduction

Small-scale fishers provide essential services to more than 180 million people living in developing countries (Zeller et al., 2006; Bene et al., 2007; Evans and Andrew, 2011). Although small-scale (artisanal) fisheries are not easily categorised, common key features are the low levels of capitalisation (Mills et al., 2011), high diversification and multitude of gear types and vessels, with dynamic patterns in their spatial and temporal usage, and the varying degree of participants, resources and ecosystem services (Berkes et al., 2001). These heterogeneous characteristics, and the numerous and scattered fishing crafts and fishing activities supplying landings directly to local markets or consumers, create challenges in managing and monitoring this sector (Wright and Richards, 1985; McClanahan et al., 1997; McClanahan and Mangi, 2001; Lleonart and Maynou, 2003; Tzanatos et al., 2005). Government agencies in developing countries do not have sufficient capacities for collecting quantitative data (Wilson et al., 2003; Pomeroy and Rivera-Guieb, 2006) or for monitoring and enforcing complex multi-species fisheries regulations.

The Negombo Lagoon, with an extended area of 62.3 km² together with the wetland system (Stromquist et al., 2000), supports year-around small-scale multi-species fisheries. Some 20 000 people are directly or indirectly dependent on the fisheries in the Negombo Lagoon and the adjacent coastal waterbodies for their livelihood. Inshore Penaeid shrimps, which use mostly the estuary and lagoon as a nursery grounds during the juvenile phase of their life cycle followed by offshore migration for subsequent maturation and reproduction (Garcia and Le Reste, 1981), are the main targeted resources of the lagoon together with fishes, crabs and squids. The multi-species nature of the lagoon fishery employs a variety of fishing gear. Outside of the lagoon, shrimp trawling is practiced in Hendala and Negombo coastal waters (Jayawardane et al., 2004; De Croos and Valtýsson, 2007).

Except for the stake seine fishery, all multi-gear shrimp fisheries in the Negombo lagoon and adjacent coastal areas have been subjected to intense fishing pressure due to an increase in fishing effort and illegal fishing methods through the nature of its open access (Pramod and Pitcher, 2006). The Negombo Lagoon and surrounding areas are also experiencing severe multiple pressures from adjoining settlements, as well as from aquaculture, agriculture, deforestation, manufacturing, trade and tourism (WCP, 1994; Bambaradeniya et al., 2002). Further, recent dredging and excavation of the lagoon for the proposed Colombo–Katunayake highway and seaplane landing port have also raised scientific concerns regarding the potential threats to the alteration of spawning and nursery grounds for many shrimp and fish species; this could potentially lead to resource depletion and jeopardisation of fishing activities, as the magnitude of the adult stock is generally determined by the success of the previous season's recruitment (Ye, 2000; Bacheler et al., 2008). Thus, knowledge of the biodiversity, size at sexual maturity, and current fishing pressures on various habitats utilised by these species during their life cycles is necessary for protection of their spawning and nursery grounds by the implementation of effective fisheries management strategies, such as the declaration of Marine Protected Areas (MPA) (Kelleher, 1999) and ecosystem-based fisheries management (EBFM) (Arkema et al., 2006).

Due to increasing global pressure on fisheries (FAO, 2009; Smith et al., 2010), concern is for the establishment of a sustainable small-scale fisheries sector, while also broadening the benefit gain for poor fishers (FAO, 2009; Mills et al., 2011). Thus, it is timely to evaluate the status of the Negombo lagoon fisheries and the adjacent coast, giving special emphasis to shrimp catches, as none of these fisheries implement any scientifically based management strategies, although community-based management approaches are being practiced successfully for some types of gear (Amarasinghe et al., 2002; Gunawardena and Steele, 2008). Scientific management of these fishery resources requires up-to-date assessment of the magnitude, variation, and distribution, both spatial and temporal, of catch and gear in order to estimate the fishing capacity (Pomeroy, 2011). In monitoring small-scale fisheries, fishery-dependent sampling methods are the most common and practical alternative (Essington et al., 2006; Worm et al., 2006; Cheung et al., 2007; Hilborn, 2007; Newton et al., 2007; Pauly, 2007). Catch per unit effort (Saltaug and Godo, 2001), gear selectivity and the overlap of resource usage can be determined if the catches are analysed based on species (Wright and Richards, 1985; Bellwood, 1988; McClanahan and Mangi, 2004).

Although the diversity, distribution, catch statistics and abundance of gear operating in the Negombo lagoon have been documented separately in various years (Amarasinghe et al., 1997, 2002; Jayawardane and Perera, 2003; De Croos and Valtýsson, 2007), a combined analysis of all shrimp fishing gear operating in the lagoon is scanty except for the 1998–1999 attempt made by Sanders et al. (2000), who considered the lagoon as one homogeneous segment. Therefore, the aim of the present study was to (i) evaluate whether the resource use of common artisanal gear in the Negombo lagoon and the adjacent coastal waters differ, giving special emphasis to shrimp catches in terms of diversity, species composition, and size compositions of the catches; (ii) analyse spatio-temporal dynamics of CPUE of shrimps with different fishing gear; and (iii) evaluate and classify the gear selectivity considering all targeted species. In addition, the present status of the fishery is evaluated by a qualitative comparison of the catch data with previously published studies. The results discussed here may be used as a baseline for the development of a sustainable fisheries management policy for the study site.

Methodology

Study area and operational gear

The Negombo lagoon is an estuarine ecosystem on the western coast of Sri Lanka, connected to the sea by a single narrow opening at the northern end (Fig. 1). Except at the entrance, water within the lagoon is <2 m (mean 0.65 m) in depth. The estuary has a surface area of 35 km² and is approximately 12.5 km in length and 0.6–3.6 km in width (Devendra, 2002).

Figure 1. Location of six fishing grounds: H: Hendala, N: Negombo, LM: lagoon mouth, LP: lagoon proper, TS: transition swamp, MP: marsh proper. Dandugam Oya (a river) enters the lagoon at its southern end. H and N fishing grounds are located at the coast; other fishing grounds are in the lagoon. Grey hatched area = lagoon wetland system

Water exchange in the lagoon is influenced by oceanic tides and an inland freshwater supply. Due to a narrow inlet from the sea, the lagoon has only one third of the tide range. The main freshwater source for the lagoon and marsh is the Dandugam Oya (a river with a catchment of 727 km²) (Nagabhatla et al., 2008). During an intermediate rainy season in January and July, a pronounced salinity gradient develops in the lagoon, varying from 20–30 ppt at the mouth to <5–10 ppt at the head of the estuary (Devendra, 2002).

Although a variety of fishing gear and techniques is used throughout the lagoon, seven widespread fishing methods (hereafter ‘gear’), which contribute significantly to lagoon production, were considered for the sampling: stake net (SN), cast net (CN), fyke net (FN), brush pile/park (BP), trammel net of mesh size with 25–30 mm in inner panel (TN1), trammel net with 40–60 mm inner panel (TN5) and gillnets (GN) having 60 mm stretched mesh size. The occasional use of other fishing methods (including weirs, crab traps, and hand lines) and highly seasonal beach seine were avoided in sampling. Both SN and BP are passive gear; their comprehensive description together with mode of operation has been documented (Amarasinghe et al., 1997, 2002). The FN is a conical shape passive gear net with two broad wings and basically operated at the southern part of the lagoon. Identical small fishing craft (canoes) within the lagoon are used to operate all of the types of fishing gear. CN is operated either with or without a craft. Coastal trawling at Negombo is by sailing craft (non-motorised trawling – NMT), while at Hendala trawling gear is operated with craft having 45 hp engines (motorised trawling – MT). For various historical and community based management reasons, differences do exist in the gear operations along the lagoon and coastal waters.

Site selection and sampling design

Data were obtained by sampling small-scale fishery practiced in the Negombo lagoon and the adjacent coastal waters over a 16-month period, from January 2009 to April 2010 (Table 1). This coincided with one hydrological cycle of the lagoon, and one shrimp life cycle and fishing period. Monitoring was based on a geographical division of the lagoon into four homogeneous regions (hereafter ‘fishing grounds’) corresponding to variations in habitat characteristics (substrate type and depth) and salinity: lagoon mouth (LM); lagoon proper (LP); transition swamp (TS); and the marsh proper (MP) covering waterways in the marsh (Fig. 1). In addition, two coastal shrimp trawling grounds Negombo (N) and Hendala (H) were sampled. Trawl catches from N and H were landed at a major landing site for each fishing grounds; however, there are large numbers of landing sites within the lagoon. Based on preliminary observations, two landing/auctioneers sites were selected for continuous monitoring on sampling days, at LM and LP, and one site each at TS and MP. Landing sites were selected to ensure adequate coverage of a variety of gear and species composition at each fishing ground. At each landing site the types of gear catches/landings were recorded. Information on fishing efforts in terms of fishing units, duration of active fishing, number of fishers engaged in operations and number of hauls; number of operations per day and the mesh sizes; and operational depth and substrate types were obtained for each sampled catch, by questioning the fishermen at the time of landing. Once each month the information was cross-checked with charted commercial craft for each gear operation. Occasionally the use of multiple gear (mostly landlines with TN or GN) in a single trip was encountered; the catches were thereby separated into different gear types using the information from fishermen.

Table 1. Species and representatives of families identified in samples caught at sampling locations, west coast of Sri Lanka, Jan 2009–Apr 2010

Family	Species
Shrimps
Peneidae	Penaeus monodon (Fabricius, 1798)
Peneidae	Penaeus semisulcatus (De Haan, 1844)
Peneidae	Fenneropenaeus indicus (H. Milne-Edwards, 1837)
Peneidae	Fenneropenaeus merguiensis (De Man, 1888)
Peneidae	Metapenaeus dobsoni (Miers, 1878)
Peneidae	Parapenaeopsis coromandelica (Alcock, 1906)
Peneidae	Metapenaeus monoceros (Fabricius, 1798)
Peneidae	Metapenaeus moyebi (Kishinouye, 1896)
Peneidae	Metapenaeus elegans (De Man, 1907)
Peneidae	Metapenaeus affinis (H. Milne-Edwards, 1837)
Peneidae	Parapenaeopsis uncta (Alcock, 1905)
Teleost fishes
Sparidae	Acanthophagrus sp.
Angullidae	Anguilla sp.
Chanidae	Chanos chanos (Forsskal, 1775)
Drepanidae	Drepane sp.
Serranidae	Epinephalus sp.
Cichlidae	Etroplus sp.
Cichlidae	Oreochromis mossambicus (Peters, 1852)
Soleidae	Soles
Gerreidae	Gerres filamentosus (Cuvier, 1829)
Gobiidae	Glossogobius sp.
Clupedae	Hilsa kelee (Cuvier, 1829)
Lactaridae	Lactarius sp.
Centropomidae	Lates calcarifer (Bloch, 1796)
Lieognathidae	Leiognathus sp.
Mugilidae	Mugil cephalus (Linnaeus, 1758)
Monodactylidae	Monodactylus sp.
Pristigasteridae	Opisthopterus sp.
Ambassidae	Ambassis sp.
Scatophagidae	Scatophagus argus (Bloch, 1758)
Sciaenidae	Johnius sp.
Siganidae	Siganus sp.
Sillaginidae	Sillago sp.
Engraulididae	Thryssa sp.
Portunidae	Portunus sp. and Scylla sp. (Crabs)
Cartilaginous fishes
Rhinobatidae and Rajidae	Skates and Rays
Cuttlefishes
Loliginidae	Squids/Cuttle fish

Sampling process

After four months (Aug–Dec 2008) of pilot survey analyses of shrimp distributions, based on the method by Helle and Pennington (2004) catches from six trawlers were selected for random sampling from each landing site at H and N fishing grounds to minimise the noise observed in length-frequency distributions and to maximise the representations in sampling the natural populations (De Croos and Stefansson, 2011). The same number of landings (six) was sampled weekly from each gear type. The same sampling scenario was followed at the landing sites of the lagoon fishing grounds, and six landings were randomly selected from each gear type.

At the beginning of the study it was made certain that all enumerators (4) had a similar range of capability in identifying species; using a field guide (De Bruin et al., 1994), this allowed all lagoon and coastal landing sites to be sampled on the same day. Fish landing sites within each fishing ground were assumed to be homogeneous and data collected from the two landing sites at LM and LP were pooled for each gear. In total, 384 samples were monitored from each gear at each landing site during the study period (total of 3456 samples from all gear).

At the landing site bony fishes were identified to the genus level, and crabs, cartilaginous fishes (skates/rays) and squids/cuttlefish species were grouped in three categories due to small numbers at the genus level. A 2-L container was used to scoop through each shrimp catch then thoroughly mixed to obtain sub-samples for determining species composition and individual lengths (±0.5 mm) in the laboratory. Occasionally, when the total shrimp landing was insufficient to fill the 2-L container, the landing was used for analysis of composition and length. Even from larger catches only one sub-sample was taken without replicates using the 2-L container. According to a pilot survey analysis (De Croos and Stefansson, 2011), collecting samples from as many fishing gear as possible provides a better representation of the population rather than an increased sample size. Individuals caught together in commercial catches tend to be more similar than the individuals in the population as a whole (Pennington and Volstad, 1994), a result of animal behaviour (Arreguin-Sanchez, 1996), gear selectivity (Rozas and Minello, 1997), and vessels (Yeh and Ohta, 2002).

From each landing, total shrimp and all other species (hereafter ‘fish’) were recorded separately with fish buyers. The weights of Scatophagus argus and Monodactylus argenteus, which are collected for ornamental purposes using BP, were eye approximations given by experienced fishermen.

Catch, effort and production estimates

Throughout the study, information from fishermen at the time of landing revealed that the fishing duration is mainly governed by current patterns, wind patterns/directions (especially for NMT), and commencement of the local fish auctions/market, as these fisheries mainly cater to the daily needs of consumers for fresh fish. This is further shown by the observed narrow standard deviation of active fishing hours for each gear type (Table 2). Similarly, the number of fishermen engaged in operating each gear unit, the mesh sizes, operational depths, number of gear units and number of hauls and number of operations per day were found to be similar in range throughout the study (Table 2). Further, as the fishermen had been helping or engaged in fishing activities since childhood, they had a good command and knowledge of the gear and craft behaviours and operational skills. Thus, gear efficiencies, fishing time, gear operational skills and the number of fishermen engaged in a fishing operation were assumed to be similar with respect to each gear type. The catch per unit effort (CPUE) was calculated as the mean catch in kg per craft per day with respect to each gear type.

Table 2. Characteristics of main fishing gear, Negombo lagoon and adjacent coastal waters

Fishing gear	Fishing grounds	No of fishers	Month		Substrate type	Depth (m)	Mesh size (mm)	Active fishing duration mean ± SD	Total no. gear units in study site/s
			Range	Peak					Previous studies	Present study
^a Fishermen rarely operate as a group. ^b Sanders et al. (2000). Fishing gear: non-mechanized trawling (NMT), mechanized trawling (MT), stake net (SN), cast net (CN), fyke net (FN), brush pile (BP), trammel net 1 (TN1), trammel net 5 (TN5) and gillnets (GN), and fishing grounds; Negombo Sea (N), Hendala Sea (H), lagoon mouth (LM), marsh proper (MP), lagoon proper (LP) and transitional swamp (TS).
MT	H	3–4	year around	Apr–Sep	sandy-muddy bottom	15–18	cod end: 15–17 body: 20–24	4.1 h ± 0.32 n = 2400	95b	87
NMT	N	4	year around	May–Sept Nov–Jan	sandy-muddy bottom	12	cod end: 15–17 Body: 20–23	4.8 h ± 0.75 n = 3454	135b	115
SN	LM	2	year around		muddy bottom	2–3	10–13	9.8 h ± 1.05 n = 2324
CN	LM/ LP/TS	1a	year around	Dec–Mar	muddy flat bottom	2	20–21	5.2 h ± 1.05 n = 2454		502
BP	LP/ TS	4–5	year around	Dec–Mar	muddy bottom	1.5–2	surrounding net ~ 20–30	97 days ± 2.5 n = 2211		660
TN1	LP /TS	1–2	year around	Jan–Mar	water column	top to bottom of water column (2 m)	inner panel 25–30 outer panel 150	5.5 h ± 0.3 n = 2091		1070
TN5	LP /TS	1–2	year around	Jan–Mar	water column	top to bottom of water column (2 m)	inner panel 40–60 outer panel 200	6.0 h ± 0.8 n = 2958
GN	LP/ TS	1–2	year around	Jan–Mar	water column	top to bottom of water column (2 m)	60 mm	5.9 h ± 0.7 n = 2198		230
FN	MS	2	year around	Jan–May	muddy bottom	1.5–2		6.2 hr ± 0.3 n = 2558		12

The monthly total production (MTP) of each craft type was estimated as a product of mean catch in kg per craft-day (CPUE), mean number of fishing craft operated per day (N) and mean number of fishing days for that particular month (D), then

monthly total production (MTP) = CPUE × N × D.

The number of craft operated on each sampling day from coastal fisheries (H and N) was recorded from the logbooks maintained by vendors at the catch collection sites in order to estimate total production. However, most vendors collecting lagoon catches do not maintain logbooks. Further, due to the widely spread and loosely organised nature of the fisheries in the lagoon, it was difficult to obtain the exact number of gear units operating each month at each lagoonal fishing ground. Thus, the authors hired the services of fish vendors from the landing sites (where samples were taken) at each lagoonal fishing ground to record the number of gear units and fishing days per month. A survey conducted at the beginning of the study estimated the total number of craft engaged in fishing at each landing site. From this information the proportion of operations per gear unit was estimated for each site and then for the entire fishing ground. Annual production was calculated for 2009.

Variations in CPUE values were analysed with respect to gear, fishing grounds and the time of the year (months) using analysis of variance (anova), after applying the optimal power transformation (of CPUE + 1) using the boxcox function in r (R Development Core Team, 2010).

Gear-based selectivity and diversity of all targeted species

A hierarchical cluster analysis using the pvclust package in r (Suzuki and Shimodaira, 2009) based on distances, was calculated as 1-r² (where r is the Pearson correlation coefficient), between the presence-absence of species in different catches. The analyses were based on data collected throughout the study period, irrespective of the fishing methods and fishing grounds, done both for all species as well as one including only the shrimp species (Table 3). For each cluster in the hierarchical clustering, P-values were calculated via multiscale bootstrap, resampling 1000 times. Clusters with Approximately Unbiased (AU) P-value > 0.95, the hypothesis that ‘the cluster does not exist,’ were rejected with a significance level of 0.05.

Table 3. Estimated total production (tonnes) of each gear operated at different fishing grounds, and percentage composition of shrimps and fishes

Gear		Total	% Shrimps composition
Fishing gear: non-mechanized trawling (NMT), mechanized trawling (MT), stake net (SN), cast net (CN), fyke net (FN), brush pile (BP), trammel net 1 (TN1), trammel net 5 (TN5) and gillnets (GN), and fishing grounds; Negombo Sea (N), Hendala Sea (H), lagoon mouth (LM), marsh proper (MP), lagoon proper (LP) and transitional swamp (TS).
Sea		420	64
MT	N	196	63
NMT	H	224	65
Lagoon		1185	30
SN	LM	162	72
CN	LM	10	50
	LP	15	53
	TS	7	57
BP	LP	156	20
	TS	168	22
TN1	LP	248	37
	TS	177	34
TN5	LP	131	1
	TS	57	1
GN	LP	24	2
	TS	26	3
FN	MP	2	38
Total		1604	39

Percentage composition and the Shannon's diversity index (Table 4) were calculated similarly for two categories: for all species and for shrimps, and separately for each gear type with respect to fishing grounds:

Table 4. Calculated Shannon's index ± 95% confidence intervals (CI) by bootstrapping for all species and shrimps with respect to fishing method and fishing method-fishing grounds

Fishing method	Fishing ground	Shannon–Wiener index ± CI
		All species	Shrimps
Fishing gear: non-mechanized trawling (NMT), mechanized trawling (MT), stake net (SN), cast net (CN), fyke net (FN), brush pile (BP), trammel net 1 (TN1), trammel net 5 (TN5) and gillnets (GN), and fishing grounds; Negombo Sea (N), Hendala Sea (H), lagoon mouth (LM), marsh proper (MP), lagoon proper (LP) and transitional swamp (TS).
MT	H	1.966 ± 0.048	0.937 ± 0.024
NMT	N	1.802 ± 0.036	0.754 ± 0.021
SN	LM	2.396 ± 0.029	1.595 ± 0.020
CN		2.312 ± 0.026	1.385 ± 0.028
	LM	2.370 ± 0.029	1.576 ± 0.024
	LP	2.138 ± 0.034	1.256 ± 0.036
	TS	1.978 ± 0.033	0.896 ± 0.034
BP		2.599 ± 0.017	1.005 ± 0.040
	LP	2.582 ± 0.025	0.822 ± 0.024
	TS	2.584 ± 0.020	1.124 ± 0.059
TN1		2.640 ± 0.012	1.530 ± 0.017
	LP	2.627 ± 0.014	1.540 ± 0.023
	TS	2.623 ± 0.021	1.502 ± 0.024
TN5		2.342 ± 0.011	1.702 ± 0.063
	LP	2.355 ± 0.016	1.774 ± 0.054
	TS	2.285 ± 0.013	1.507 ± 0.115
GN		2.479 ± 0.012	2.109 ± 0.025
	LP	2.497 ± 0.016	2.067 ± 0.064
	TS	2.419 ± 0.018	1.952 ± 0.039
FN	LS	1.941 ± 0.029	1.111 ± 0.034

where, p_i is the relative abundance of each species, (n_i/N) is the number of individuals in species i; N is the total number of all individuals and S is the number of species. Standard deviation was calculated based on 1000 bootstrap samples.

Lengths of three species (F. indicus, M. dobsoni and P. semisulcatus), all harvested with lagoon gear, were compared among fishing grounds separately for each fishing gear with the Kruskal–Wallis test, and multiple comparisons were conducted using the Nemenyi and Dunn test (Zar, 2009).

Exploration of the associations and degree of similarity in the gear selectivity was determined with a Detrended Correspondence Analysis (DCA) of the percentage composition of species in each sample, using the vegan package in R (Oksanen et al., 2010). The DCA was analysed for two categories: one for all species and one for only shrimps. The groupings of the samples were summarised with respect to gear type. In addition, the temporal species-gear selectivity changes were considered (averaged over each month) and also how they varied among fishing grounds.

Results

Main species by gear types

The catch data of nine dominant fishing methods comprised a total of 55 species (31 bony fish species, 13 shrimps, three crabs, three squids/cuttlefishes and five skates/rays) representing 34 families. The most abundant 26 species and 11 shrimps species, which accounted for more than 5% in total samples of any gear, were considered for further analysis (Table 1).

Among the shrimps, four species dominated catches, contributing 82% of the total shrimp landings in the lagoon and coast: Fenneropenaeus indicus, M. dobsoni and P. semisulcatus were present in both lagoon and coast samples while P. coromandelica was found only in coastal trawl samples (in both NMT and MT). Among the other shrimps, Metapenaeus moyebi, M. elegans and F. merguiensis were the most common. M. moyebi, M. elegans and M. monoceros were reported only in the lagoon, while the P. uncta were observed only in NMT samples. Two shrimp species, P. cornuta and P. maxillipedo, found in coastal trawl catches were disregarded due to low accountability and with the latter species encountered only in MT catches. Leiognathus spp. followed by Mugil spp. and Latus spp. were the most common fish species captured at both the lagoon and coastal fishing grounds by all fishing methods. Oreochromis mossanbicus and Etroplus spp. were common at MP.

Fishing gear differed in terms of the number of fishermen engaged in a fishing operation, the peak month of operation, depth and type of fishing substrate, but among the same gear types these characteristics were found to be in a similar range (Table 2). All fishing methods were operated using a craft, except the cast net operations especially operated at shallow lagoon areas of LP and TS that counted for 3–5% of the total CN operations. Craft were also used to construct and harvest from the passive gear, BP, SN and FN.

Catch variations in dominant shrimp species

CPUE variations

For NMT, MT, SN and CN, the estimated mean CPUE values (kg per landing) of shrimps were higher than the values reported for fish; however, for BP, TN1, TN5 and GN, the CPUEs of fish were higher than for the shrimps (Fig. 2). Evident by the whiskers of the boxplots, the CPUE of SN fluctuated more than for the other gear. Variable catches were also evident in MT and NMT. The mean shrimp CPUE of CN decreased from the lagoon mouth towards the interior of the lagoon, while catches from BP, TN1, TN5 and GN remained in a similar range (Fig. 2a). According to the species-wise composition analysis the decline of catches in CN towards the interior of the lagoon (LM and TS) was mainly due to the decline of P. semisulcatus catches. Both F. indicus and M. dobsoni catches declined when moving into the lagoon from LM to TS.

Figure 2. Study period catch per unit efforts (CPUE) (kg per craft-day) of (a) shrimps and (b) other species caught at six fishing grounds (FG): Negombo Sea (N), Hendala Sea (H), lagoon mouth (LM), marsh proper (MP), lagoon proper (LP) and transitional swamp (TS) by nine fishing gear: non-mechanized trawling (NMT), mechanized trawling (MT), stake net (SN), cast net (CN), fyke net (FN), brush pile (BP), trammel net 1 (TN1), trammel net 5 (TN5) and gillnets (GN). Boxplots show distribution of CPUE values; boxes span interquartile range (IQR), thick line = median, whiskers drawn at 1.5*IQR distance from box, and circles = outliers. Dot on each box = mean. [Correction added on 24 September 2012, after first online publication: the missing data on the x-axis have been inserted.]

Observed CPUE values for the main shrimp species F. indicus, M. dobsoni, P. semisulcatus and P. coromandelica varied with respect to months, fishing grounds and gear (P < 0.001). High catch rates were observed at the coast and in the lagoon at different time of the year (Fig. 3). January–May 2009, the CPUE values for all lagoon gear were higher in these four main shrimp species than in other months. The same trend was observed in January–April the following year. At the coast, highest shrimp CPUE values were reported from May to October, but at the same time the lagoon catches were low.

Figure 3. Variation in mean monthly CPUE (kg per craft-day) of four species caught by gear at six fishing grounds, Jan 2009–Apr 2010. Monthly sampling of 24 gear from each gear type at each fishing grounds and respective mean values plotted. CPUE axis not equal in all panels.

Total production and fishing effort

Estimated total annual production from all gear was 1604 tonnes (t) whereby shrimp contributed 625 t (Table 3). Lagoon gear produced a higher shrimp production (355 t) than the trawling gear (269 t), a result of catching most of the shrimp life stages in the lagoon using a large number of seven gear types (Fig. 4). From all gear the highest annual production was from the TN1 for both shrimp and fishes; 58% of these were at the LP grounds. The number of TN1 gear units operating in the LP fishing grounds was four times higher than at the TS grounds. The second highest shrimp production was given by trawl gear. Fish and shrimp production by cast nets was in a similar range. For BP, fish production was four times higher than the shrimp production. The number of gear units operated at the TS was 30% less than that at the LP.

Figure 4. Composition of shrimp species harvested per fishing grounds (fg); Negombo Sea (N), Hendala Sea (H), lagoon mouth (LM), marsh proper (MP), lagoon proper (LP) and transitional swamp (TS) and fishing gear: non-mechanized trawling (NMT), mechanized trawling (MT), stake net (SN), cast net (CN), fyke net (FN), brush pile (BP), trammel net 1 (TN1), trammel net 5 (TN5) and gillnets (GN). Numbers represent: P. monodon (1), P. semisulcatus (2), F. indicus (3), F. merguiensis (4), M. dobsoni (5), P. coromandelica (6), M. monoceros (7), M. moyebi (8), M. elegans (9), M. affinis (10), P. uncta (11).

Length variations

No differences were observed in mean lengths of F. indicus, M. dobsoni and P. semisulcatus among the lagoonal fishing grounds with respect to different fishing gear (Table 5). In all three species the SN and BP select a broader range of length classes than the other gear (Fig. 5), and comparatively smaller shrimps were vulnerable to the SN catches. The CN, TN and BP select a similar range of length classes, while the largest shrimp, M. dobsoni, were noted in the trawl (MT and NMT) catches.

Figure 5. Variation of length–frequency distribution, males (M) and females (F) of three species by six gear types: non-mechanized trawling (NMT), mechanized trawling (MT), stake net (SN), cast net (CN), fyke net (FN), brush pile (BP), trammel net 1 (TN1) from all fishing grounds, Jan 2009–Apr 2010. In each panel n = respective sample size.

Table 5. Average carapace length ± SE of P. semisulcatus, F. indicus and M. dobsoni from three fishing gear (CN – cast net, BP – cast net, TN1-trammel net 1 and SN – stake net) at their respective fishing grounds for females (F) and males (M); lagoon mouth (LM), lagoon proper (LP), transitional swamp (TS)

Location species	Sex	CN				BP			TN1			SN
		LM	LP	TS	P-value	LP	TS	P-value	LP	TS	P-value	LM
Length range is given in parentheses. P-values = probabilities from KruskalWallis test on length differences between fishing grounds. Average length of F. indicus males and M. dobsoni females caught with CN-gear varied significantly (P < 0.05) among fishing grounds after multiple comparisons (Nemenyi and Dunn test), as denoted with a, b, c superscripts.
P. semisulcatus	F	3.36 ± 0.03 (1.2–4.6) n = 404	3.34 ± 0.03 (1.6–4.6) n = 453	3.35 ± 0.02 (1.6–4.6) n = 450	0.85	3.13 ± 0.02 (1.4–4.8) n = 531	3.21 ± 0.03 (1.4–4.8) n = 627	0.07	3.51 ± 0.02 (2.2–5.0) n = 497	5.52 ± 0.01 (2.2–5.0) n = 540	0.81	2.95 ± 0.02 (1.2–5.0) n = 314
	M	3.33 ± 0.03 (1.6–4.4) n = 494	3.30 ± 0.02 (1.6–4.4) n = 525	3.33 ± 0.02 (1.6–4.4) n = 540	0.68	3.01 ± 0.02 (1.3–4.8) n = 561	3.05 ± 0.02 (1.4–4.8) n = 478	0.27	3.41 ± 0.01 (2.4–4.2) n = 451	3.43 ± 0.01 (2.4–4.6) n = 511	0. 07	2.91 ± 0.01 (0.8–4.7) n = 322
F. indicus	F	3.77 ± 0.02 (1.8–4.4) n = 1175	3.89 ± 0.02 (2.4–4.8) n = 879	3.88 ± 0.03 (2.2–5.4) n = 566	0.05	3.78 ± 0.02 (2.1–5.3) n = 521	3.73 ± 0.02 (1.9–5.3) n = 709	0.29	4.08 ± 0.01 (2.5–6.0) n = 529	4.14 ± 0.01 (2.5–6.2) n = 503	0.12	3.68 ± 0.02 (2.0–5.5) n = 430
	M	3.35 ± 0.02^a (2.0–4.2) n = 471	3.62 ± 0.01^a (3.0–4.2) n = 491	3.87 ± 0.02^b (2.4–5.4) n = 1387	0.0001	3.70 ± 0.02 (1.9–5.3) n = 425	3.73 ± 0.02 (1.9–5.3) n = 494	0.28	3.96 ± 0.02 (1.9–5.5) n = 356	3.98 ± 0.01 (1.9–5.5) n = 308	0.48	3.58 ± 0.01 (1.8–5.3) n = 392
M. dobsoni	F	1.74 ± 0.01^a (0.8–2.8) n = 481	1.42 ± 0.01^b (0.8–2.6) n = 528	1.85 ± 0.01^c (0.8–2.8) n = 1308	0.0001	1.78 ± 0.02 (0.8–2.4) n = 559	1.75 ± 0.01 (0.8–2.4) n = 1648	0.44	1.85 ± 0.01 (1.0–2.6) n = 654	1.88 ± 0.01 (1.0–2.6) n = 839	0.07	1.9 ± 0.02 (0.8–3.1) n = 410
	M	1.73 ± 0.01 (0.8–2.0) n = 1112	1.70 ± 0.01 (0.8–2.8) n = 795	1.75 ± 0.01 (0.8–2.8) n = 587	0.14	1.56 ± 0.00 (0.7–2.2) n = 1653	1.57 ± 0.01 (0.7–2.2) n = 1245	0.10	1.76 ± 0.02^a (0.7–2.2) n = 715	1.71 ± 0.01^b (0.7–2.2) n = 717	0.0005	1.74 ± 0.02 (0.8–3.0) n = 436

Gear and fishing grounds

Hierarchical cluster analysis based on present-absent data showed seven clearly separated clusters for all species data (Fig. 6a); only two clusters were observed for shrimp data, P < 0.05 (Fig. 6b).

Figure 6. Cluster dendrogram with au/bp values (%) based on a) all species and b) only shrimps. Values at edges of clustering are P-values (%) calculated via multi-scale bootstrap of 1000 resampling. Values on left side in red = au (approximately unbiased) P-values, and values on right side in green = bp (bootstrap probability) values. Clusters with au larger than 95% highlighted by rectangles.

The clustering of species was based on habitat and occurrence rather than the selectivity of fishing methods. Cluster 1 represents six marine species including a shrimp species, P. coromandelica, which is found in trawl catches (NMT and MT); cluster 7 represents seven brackish water species together with an Ambassis sp. that migrates from coastal to brackish waters for spawning. Low catch rates of Ambassis sp. were observed with other fishing gear except for the SN. Two species, Monodactylus sp. and S. argus, which clustered together in cluster 4 were found only in the BP catches. The other clusters (2, 3, 5 and 6) represent shrimp species together with fish species found in both coast and lagoon areas by all fishing methods. The clustering pattern of shrimps (Fig. 6b) seems not to be affected by excluding the other species (6a).

Gear selectivity by species, season and space

According to DCA, variations in gear selectivity (catch compositions of different fishing gear) were observed for all species together (Fig. 7a) and only for shrimps (Fig. 7d). The main differences in catch compositions were between the trawl gear (NMT and MT) and the others along the DCA1 axis, while the gear FN differed from the rest along the second DCA2 axis. The differences between gear types were consistent over the season (Fig. 7b) and fishing grounds (Fig. 7c), but to a lesser degree for the shrimps (Fig. 7e and f).

Figure 7. Detrended Correspondence Analysis plots of nine gear and species composition for all species (a, b, c) and shrimps (d, e, f). ‘a’ and ‘d’ = species, ‘b’ and ‘e’ = monthly catch compositions, ‘c’ and ‘f’ = catch compositions at ports. Panel a) numbers refer to different species, shrimp in bold (1–11) other species in gray (12–37). Species names for corresponding numbers: P. monodon (1), P. semisulcatus (2), F. indicus (3), F. merguiensis (4), M. dobsoni (5), P. coromandelica (6), M. monoceros (7), M. moyebi (8), M. elegans (9), M. affinis (10), P. uncta (11), Acanthophagrus sp., (12), Anguilla sp. (13) C. chanos (14), Drepane sp. (15), Epinephalus sp. (16), Etroplus sp. (17), O. mossambicus (18), Soles (19), G. filamentosus (20), Glossogobius sp. (21), H. kelee (22), Lactarius sp. (23), L. calcarifer (24), Leiognathus sp. (25), M. cephalus (26), Monodactylus sp. (27), Opisthopterus sp. (28), Ambassis sp. (29), S. argus (30), Johnius sp. (31), Siganus sp. (32), Sillago sp. (33), Thryssa sp. (34), Portunus sp. and Scylla sp. (35), skates and rays (36) and squids/cuttlefish (37). Cluster of open circles = species 18, 22, 37, cluster of solid circles = species 33, 34 and 36. Gear types denoted as non-mechanized trawling (NMT), mechanized trawling (MT), stake net (SN), cast net (CN), fyke net (FN), brush pile (BP), trammel net 1 (TN1), trammel net 5 (TN5) and gillnets (GN). NMT and MT are overlapping in all panels.

Similarity in selectivity between NMT and MT was mainly due to catches of P. coromandelica (6), P. uncta (11) and Johnius sp. (31).

A similar range of selectivity observed in the TN5 and GN for all species (6b) was due to Mugil cephalus (26), Epinephalus sp. (16), Siganus sp. (32) and Portunus sp. and Scylla sp. (35), while selectivity of these two gear types was more separated along the second DCA2 axis when analysing the shrimp catches (7e).

Selectivity analysis among fishing grounds shows larger separation of shrimp catches (7f) than all species catch compositions (7c) along the first DCA1 axis.

Discussion

Resource use of common artisanal gear in the Negombo lagoon and adjacent coastal waters differs in diversity, species and size composition of the catches; there are also differences in spatio-temporal dynamics of shrimp CPUE. This is the first study to examine the various segments of the Negombo lagoon and the adjacent coast together to determine how artisanal fishing gear select the fishery resources in the lagoon in a special and temporal context. The current exploitation rates shows signs of higher pressure on the lagoon resources.

Resource use by common artisanal gear

The artisanal shrimp fishery along the western coast of Sri Lanka is complex, due to the multi-gear and multi-species context of both shrimp and fish. Compositions of shrimps caught with different types of gear seem to be a result of both gear selectivity and species' behavioural characteristics. Nine of 13 shrimp species spawn in coastal waters and have an essential estuarine phase to complete their life cycles; P. coromandelica and P. uncta are purely marine species, whereas M. elegans and M. moyebi are able to complete their lifecycles within the lagoon (De Bruin et al., 1994). The coastal and lagoon catch statistics of this study were in conformity with the reported compositions of the shrimps by Sanders et al. (2000) during 1998–1999, except for M. elegans and M. moyebi, which were in low numbers, comparatively lower in NMT catches, and completely absent in MT catches. Absence of M. elegans, M. moyebi, M. affinis and M. monoceros in MT but present in NMT could be due to either gear selectivity or the absence of these species in the Hendela fishing grounds, which is located further from the lagoon mouth than the Negombo fishing grounds.

Similarly, lower CPUE of P. uncta by one fourth in MT than in NMT, and by one fifth in Sanders et al. (2000) study, may be due to differences in gear selectivity or lower abundance at Hendala. Sanders et al. (2000) recorded two species, P. latisulcatus and P. canaliculatus, not encountered in this study; their not recorded P. maxillipedo, however, was caught in MT catches in low numbers.

Spatial-temporal dynamics of CPUE of shrimps

The observed fluctuations of the coastal shrimp catches and the peak CPUE durations in this study are consistent with the findings of De Bruin et al. (1994), Jayawardane et al. (2004), and De Croos and Valtýsson (2007). The peak fishing season coincides with the southwest monsoonal period in Sri Lanka. Seasonality in shrimp catches has also been reported from Jaffna lagoon in northern Sri Lanka (Chitravadivelu, 1990), India (Rajendran and Kathiresan, 1999; Iwasaki and Shaw, 2008) and many other parts of the world (Fischer and Bianchi, 1984; Potter et al., 1986), although the peak production periods are different.

The reason for the observed seasonality of CPUE values in the lagoon and coastal gear can be explained by a mass migration of shrimps towards the spawning grounds in coastal waters from the lagoon with the onset of the southwest monsoon (May–Sept.). Monsoonal rainfall lowers the salinity and may cause osmotic stress and further affect other climatic and physiochemical factors such as water level, currents, and temperature fluctuations (Garcia and Le Reste, 1981; Luchmann et al., 2008). This results in high shrimp production and relative abundance in coastal trawling gear in May–September, while lagoon production is high from January to May. In the Negombo trawling grounds, high CPUE values were observed in the southwest monsoonal period, with a peak period in January; whether this is related to the processes in the lagoon is unknown.

The trawling gear NMT and MT had considerably higher CPUE values with respect to other gear types. The percentage mean annual contributions of shrimps estimated for MT (66%) and NMT (63%) were comparable to the values reported by Jayawardane et al. (2004) in 1998–1999. According to De Croos and Valtýsson (2007), the mean proportion of shrimp contributions from the same waters for 1997–2004 was 55% (NMT) and 44% (MT); however, in many parts of the world, bycatch dominates the shrimp trawl catches (Kennelly, 2007). Metapenaeus dobsoni and P. coromondalica had the highest contributions to the total trawl catches; a similar observation was reported in previous studies (Sanders et al., 2000; Jayawardane et al., 2004), moreover contribution percentages of these two species have remained in a similar range in all studies. However, in comparison to Jayawardane et al. (2004), percentage contributions of F. indicus and P. semisulcatus to the total trawl catches were reduced from 5 to 1%. The mean annual shrimps CPUE of MT and NMT (13 and 12 kg landing⁻¹) was slightly higher than the values reported (11–10 kg landing⁻¹) by Jayawardane et al. (2004) but the fishing effort, in terms of number of boats operated, was reduced by 8–10% in both MT and NMT. Although the average CPUE is slightly higher in this study compared to that of 1998–1999 (Sanders et al., 2000), the total trawl catch (MT and NMT) has been reduced by 2 t. This could be due to a lower number of gear units operated and less competition, or to lower species abundance.

Shrimps contributed 70% to the total catch of SN, lower than the values (~82%) reported in 1997 and 1999 (Amarasinghe et al., 1997; Jayawardane and Perera, 2003). Although M. dobsoni contributed the highest percentage to the total catches of SN, the composition fluctuated from 57% in the earliest study by Amarasinghe et al. (1997) to 22% (Jayawardane and Perera, 2003), and to 38% in the present study. The contribution of F. indicus with 19% in the present study is higher than the 14% reported by Jayawardane and Perera (2003). High catch rates of M. dobsoni are partially due to the high number of individuals. Moreover, Amarasinghe et al. (1997) reported that M. dobsoni attained adult size before migrating back to the sea for spawning. Although mean annual CPUE was in a similar range, the total shrimp production was lower, with 20 t of shrimps and 40 t of fish than the reports of Jayawardane and Perera (2003).

Estimated total production caught by trammel nets (TN1 + TN5) was half the value (304 t) reported by Sanders et al. (2000); the CPUE was lower in shrimps by 0.4 kg landing⁻¹ but higher in fish by 3 kg landing⁻¹. Use of the TN has spread widely within the past decade in the lagoon and may have caused the observed reduction in mean annual CPUE, and might be an indication of reduced abundance due to higher exploitation. Similarly, the mean annual shrimp CPUE (in LM + LP + TS) is less (2.5 kg landing⁻¹) for the CN catches compared to Sanders et al. (2000) (3.7 kg landing⁻¹). Although there are no previous records on the total production of CN and number of gear units operating in the lagoon, it is believed that most of the cast net fishers have shifted to trammel nets in the past decade, but TN production does not appear to have increased. The highest CPUE of CN and TN observed at LP could be the high abundance of shrimps due to their habitat preference. Although habitat preference of shrimps with respect to vegetation, sediment characters, salinity and dissolved oxygen has been studied in many waters (Henderson et al., 2006; Fransozo et al., 2009) such information for the study site is scanty. Decreased numbers of FN and GN gear units in past few decades (personal communication from fishermen) might have caused their slightly higher CPUE (LP + TS) compared to Sanders et al. (2000) due to reduced competition. The mean annual shrimp compositions in BP catches are much higher (18%) than the values (5%) reported by Sanders et al. (2000) for 1998–1999. However, for the same years, Amarasinghe et al. (2002) reported a 91% fish catch while the remainder was from both Penaeid shrimps and crabs.

Comparisons of spatial-temporal differences in CPUE are justifiable, as no modifications of NMT, MT and SN gear, which have led to changes in fishing effort with time, have been recorded. Also most of these fisheries have been operating for more than 100 years (Amarasinghe et al., 2002; Gunawardena and Steele, 2008) and good knowledge of the fishing location, fish behaviour and gear operations can thus be assumed. Such a comparison for BP is problematic, as gear and effort standardisation require assumptions on the fishing method and operation, which are difficult with the available data. In Negombo lagoon, the use of light sources (lamps, fire torches, etc.) as devices to attract shrimps in SN and in some FN and CN operations, the composition and density of mangrove branches used in BP constructions, exact area of BP, and the magnitude of the much needed gear operational skills especially for CN operations are really hard to measure and standardise but could definitely have an effect on the gear selectivity and efficiency. A lack of reliable direct data from fisheries for estimating actual fishing effort and catch (including discards and bycatch) is a common challenge among all fisheries in the world (McCluskey and Lewison, 2008; Anticamara et al., 2011). This challenge can be overcome by collecting data on detailed gear information, time spent for fishing and searching, catch rates and special and temporal information across the entire fishery (McCluskey and Lewison, 2008) by the use of on-board electronic logbooks (Gallaway et al., 2003; Cole et al., 2006). However, such fully controlled surveys are expensive and may not be feasible for small-scale fisheries; thus a combination of direct observations (Lynch, 2006), interview data taken at the landing site (accurate memory) and logbooks data provide reasonable estimates of catch and effort (Cotter and Pilling, 2007; Bishop et al., 2008). But even with a strong sample design the quantification of variables may confound effort estimates, such as skipper skill and information shared among skippers, repetitive fishing in the same area, and technological advances (Hilborn, 1985; Fonteneau et al., 2004; Eigaard et al., 2011). Further, accurate catch estimates are challenging especially in small scale fisheries where illegal, unregulated and unreported (IUU) fishing are abundant (Mills et al., 2011). Walters (2003) and Maunder et al. (2006) have also shown that the standardised CPUE are not always representing the abundance. Nevertheless, comparisons of this study are pervasive as they represent the resource utilisation and are thus important for understanding the harvested resource and its management (cf. McClanahan et al., 2010).

The observed higher annual CPUE for some gear other than in the Sanders et al. (2000) findings may be due to reduced competition among gear as a result of the reduced number of gear units. According to Hilborn and Walters (1992) and Harley et al. (2001), CPUE calculated from commercial data might not be proportional to the abundance due to behaviour and experience of fishermen, but Bellido et al. (2001) suggested that CPUE could be used as an index of abundance. At any rate, comparison of 2–3 years of data limits the conclusions on exploitation patterns, as the observed CPUE changes could also be due to an annual fluctuation of environmental factors that may influence recruitment, survival and/or growth of the species (Miyahara et al., 2005).

The estimated 200 t less shrimp than during 1998/9 from the same waters (Sanders et al., 2000) could be an indicator of population decline due to increased fishing pressure by the higher number of TN usage together with habitat alteration. Moreover, comparison with the studies of Sanders et al. (2000) and Jayawardane et al. (2004), shows a reduced number of P. uncta individuals and complete absence of P. latisulcatus and P. canaliculatus in this study despite 2–3 times larger samples. Further, relatively smaller length at maturity (L₅₀), in P. coromandelica (De Croos and Pálsson, 2011) and M. dobsoni (De Croos et al., 2011) at Hendala than at Negombo, as well as lesser mtDNA diversity in P. coromandelica (De Croos and Pálsson, 2011), F. indicus (De Croos and Pálsson, 2010), M. dobsoni and P. semisulcatus (unpublished) may reflect intensive/selective fishing at Hendala. Lagoons and estuaries in developing countries are, in general, heavily exploited by artisanal fisheries (Lae et al., 2004), as lagoons are an important source of fish to the local people and fisheries an important component of their economy (Blaber, 2000). Although commercial fisheries are not stabilised in the Negombo lagoon area, most artisanal fisheries operating large numbers of gear units may be responsible for an increased fishing effort due to an open access nature, which could lead to overexploitation as seen in many parts of the world (Blaber, 2000; Arellanop-Torres et al., 2006). It should be noted that the open access nature has been restricted by a community based management system in Negombo lagoon for SN (Amarasinghe et al., 1997; Gunawardena and Steele, 2008), and BP (Amarasinghe et al., 2002) fisheries. Overall, the fishery is showing some signs of the high fishing pressure on shrimp resources.

Gear selectivity

The selectivity of gear is associated with the length of the shrimp. The SNs are operated at the lagoon mouth, targeting migratory shrimp as revealed by the length-frequency distributions in this study. Trawl gear select larger, mature shrimp that have migrated to the coast for spawning.

There were differences in gear catch compositions. The trawl gear showed a high dominance of marine species and lower diversity in its catches. Comparatively high variability in monthly catch compositions of shrimps (Fig. 7e) might be due to their migratory behaviour between the lagoon and coast at different months of the year, but in all species analyses (7b) such dominance in catch compositions is not prominent. The diversity indexes calculated for all species within the lagoon are in a similar range for all gear except FN, perhaps due to the knowledge accumulated over the past 100–250 years for some gear (Gunawardena and Steele, 2008). Fishermen seem to be aware of the natural pattern of fish availability and operate their gear to optimise the yield (Amarasinghe et al., 2002). Thus, despite the structural, operational and spatial differences, all lagoon gear seem to be targeting all available resources, resulting in a similar range of catch compositions as seen in Fig 7b. Although the catch compositions and size frequencies are determined partly by the gear, as described by Wright and Richards (1985) and Gobert (1994), gear selectivity needs to be studied in combination with traditional fishing (see also Ruddle, 1996). Moreover, the shrimp migratory pattern along the lagoon, the proportion of mixing of species at different fishing grounds (especially at Hendala and Negombo) and the existence of any stock structures will be important in developing a sound management plan. Gear-based management may be one of the best alternatives in managing multi-gear, multi species fishery (McClanahan and Mangi, 2004; McClanahan and Cinner, 2008; Davies et al., 2009) as it is less likely to threaten fishers' livelihoods (Cinner et al., 2009). This requires baseline information on selectivity of resources and catch compositions of gear used in fisheries. Thus, the identification of fishing gear that preferentially target species is of great management importance (Cinner et al., 2009) to utilise most of the available species in a sustainable manner (McClanahan and Mangi, 2004). The information presented in this study can contribute to the development of fisheries management decisions, allowing resources to be captured without promoting resource and size overlap, in addition to improving rationalisation of future sampling protocols.

Acknowledgements

This research was fully supported by a grant from the Icelandic International Development Agency (ICEIDA-Sri Lanka) and the United Nations Universities, Fisheries Training Programme (UNU-FTP) in Iceland. We thank Prof. Gunnar Stefansson for his suggestions in designing the pilot sample survey used in the study. Support received from Dr. Tumi Tomasson, (County Director, ICEIDA- Sri Lanka) is gratefully acknowledged. RMGN Thilakarathna, P. Dimuthu, R. Kumara, Sarath, Raja, Ranjith and the fisher communities are acknowledged for support in the field.