In Sri Lanka lagoonal small-scale fisheries (SSF) are particularly abundant, with over 100 lagoons along its 1,340 km coastline (Silva et al., 2013). These fisheries support the livelihoods of 2% of Sri Lanka's population (some 500,000 people; Samarakoon and Samarawickrama, 2012), providing up to 40% of livelihoods in some locations (Ranasinghe, 2010). Sri Lankan lagoonal fisheries saw a near tripling in its fisher numbers between 1986 and 2011 along with a general move from subsistence to a commercial enterprise (IUCN, 2012). Associated income is estimated to be more than five billion rupees (39.6 million USD; Samarakoon and Samarawickrama, 2012), with shrimp exports alone valued at nearly three billion rupees (19.5 million USD; MFARD, 2015).

This study was conducted in Puttalam lagoon (Figure 1), in the Puttalam district on the north-western coast of Sri Lanka (IUCN, 2012). When joined with Dutch Bay and Portugal Bay, the lagoon forms Sri Lanka's most considerable brackish water body (IUCN, 2003b) covering an area of 46,000 ha characterized by open water, mudflats, mangroves, seagrass, and salt marsh (IUCN, 2003a; Ranasinghe, 2010).

The Puttalam district contains 16 Divisional Secretariat Divisions (DSD's), where DSD's are administrative sub-units governed by a “divisional secretary” (Department of Census Statistics, 2011). There are 4 DSD's around the lagoon, Puttalam, Kalpitiya, Mundel, and Wanathavilluwa, but only populations in Puttalam, Kalpitiya, and Wanathavilluwa depend directly on the lagoon for food and livelihoods (Ranasinghe, 2010).

Over 15,000 fishermen target the lagoon, 90% of whom depend on fishing as their sole source of income (IUCN, 2012). The southwest monsoon sees an additional influx of fishers from coastal reef fisheries, which are too rough to be fished from June to September (IUCN, 2012), these fishers tend to operate larger boats targeting different species to fishers who operate in the lagoon all year round. The most common gear types used within the lagoon are variations of trammel nets (set and drift) which are used to catch finfish (74%) and shellfish (crustaceans) (26%), however, professional fishers also use illegal gear such as fyke nets and trawl nets. While trawls are generally not considered SSF, within the Puttalam Lagoon these are small and assisted not by motor but by wind and operated from traditional craft with sails.

A scoping exercise was conducted around the lagoon to identify a representative landing site. A total of 42 landing sites were identified (generally 1 for every 1 km of lagoon coast). The majority of landing sites were utilized by seagrass fishers operating non-motorized traditional craft (NTRB), given that seagrass fringes the coasts of the lagoon. Only a small number, closest to Puttalam town, were characteristically different—made up of much larger boats suitable for fishing outside of the lagoon. The catch survey was conducted at the landing site of a multi-gear seagrass associated fishery in Kalpitya North (Figure 1) being characteristic of the majority of fishing communities around Puttalam Lagoon.

The entire catch from 63 distinct fishing trips was recorded at the point of landing over 7 days in August 2015. Catch composition was recorded to species level together with the size of the first 20 individuals of each species as they were removed from fishing gear. For trammel and trawl nets, all fishers brought nets ashore to remove catch. Similarly, fishers using fyke nets placed all of the catch into a crate and brought it to shore before sorting. Gear type, mesh size, hours spent fishing, preferred fishing habitat, transport type, target species, and total catch weight were also recorded. Species that were discarded, and thus deemed by-catch, were recorded as fishers removed them from their nets. The authors were present for two trips to retrieve catches and confirmed that no sorting of the catch occurred at sea due to (1) time and (2) size of craft. Species were grouped into three categories, 1 = always discarded, 2 = discarded if small, and 3 = always kept (Appendix A in Supplementary Material). To determine the juvenile composition of the catch, size at maturity data was collated from the literature. Where length at maturity data was not available (Froese and Pauly, 2016) it was defined by one-third of the maximum length of each species (Harmelin-Vivien et al., 1985).

Over the same period as the landing survey, a survey was conducted at two local fishmongers (Wardi, Kalpitiya North and Razeek, Kalpitiya Centre). Wardi was the only fishmonger near the landing site at Kalpitiya North, and all fishers operating at this site would sell here. Market owners were additionally asked to record which species customers requested on each day as well as the total weight of each species sold per day.

Two hundred household surveys were carried out across the 3 DSD's that surround Puttalam Lagoon (August–September 2015; Unsworth and Cullen, 2010; Table 1). Households were randomly selected, and respondents were given project information before asking for consent to take part. Verbal consent was obtained. Respondents were interviewed to ascertain information on fishing characteristics (if fishers) and household fish consumption.

Respondents provided their preferred five choices of fish and invertebrates for consumption. If the respondents classified themselves as fishermen (either part time or full time), they were asked how they disposed of by-catch. Due to multiple language uses around the lagoon (notably Sinhala and Tamil), not all fish names could be translated into Latin at the species level. Although some individuals were identified to species level (following the landing survey), others corresponded to family or order.

Differences in the total number of species caught, the magnitude of by-catch and the number of target species caught across gear types were analyzed using SPSS v. 23. Data were tested for normality using a Shapiro–Wilk test. Where data were not normally distributed, transformations were attempted. Where transformations were not possible, non-parametric tests were used. One Way ANOVA was used to test for differences in the number of species caught and the prevalence of by-catch across gear types and Kruskal–Wallis used to test for differences in target species caught across gear types (Underwood, 1997). A T-test was performed to investigate differences in catch composition and magnitude of by-catch (number of non-target species caught) between target type (classified as finfish or shrimp) and a Mann–Whitney U was used to test for differences in the number of target species caught between target types (finfish and shrimp; Stewart et al., 2010). Additionally, a Kruskal–Wallis test was used to test for differences in life stage of by-catch (adult or juvenile) across gear types and a Mann–Whitney U-test was used to test for differences in life stages of by-catch (adult and juvenile) between target type (finfish and shrimp).

Analysis of the differences between fish species assemblages across gear types and between target types was carried out using Primer v6 (Clarke and Gorley, 2006). Multi-dimensional scaling (MDS) plots were generated with superimposed Bray Curtis similarity clusters at the 40% similarity level. ANOSIM (analysis of similarities) and pairwise tests were used to test the similarity between a priori defined groups of samples (i.e., gear type). SIMPER analysis (Similarity Percentages) was carried out to test for species contributions to the Bray Curtis similarity between and within a priori defined groups (Clarke, 1993).

Shrimp (Penaeus spp.) was the target for over 80% of the fishing trips observed. Shrimp fishers used a combination of gear types both legal (trammel net) and illegal (trawl net, fyke net; Table 2). All fishers used non-motorized traditional craft (NTRB) and trawls were assisted using wind power. Fishers using set trammel nets for finfish were only targeting Mullet (Mugilidae spp.), rabbitfish (Siganus spp.), and the green chromide (Etroplus suratensis).

All fishers at the landing site preferred fishing on seagrass over any other habitats and every fishing trip observed was within shallow seagrass (1–2 m). Of the 200 household interviews conducted, 111 were fishers. Seventy-seven percent of fishers operating around the entire lagoon preferred seagrass over any other habitats. Seventeen percent preferred to fish in deeper waters off the coast (identified around the town of Puttalam), and the remaining fisheries utilized mud or sand toward the middle of the lagoon (3–4 m). There was no preference for mangrove or coral among fishers.

Sixty-two fish species from 35 families and six invertebrate species from two families were recorded (Appendix A in Suplementary Material). One species of aquatic snake, the little file snake (Acrochordus granulatus) was present in two catches but was excluded from further analysis as it is classed as least concern (Sanders et al., 2010). Of the other recorded species, both near threatened (Dasyatis zugei, Gymnura poecilura, and Epinephelus malabaricus) and vulnerable species (Hippocampus histrix and Hippocampus kuda) were present (IUCN, 2016).

Of the species recorded across catches, 51.5% were routinely discarded, whereas 35.3% were discarded if they were too small. Only 13.2% of species were kept regardless of size. These were six species of invertebrate (Metapenaeus dobsoni, Penaeus indicus, P. monodon, P. semisulcatus, Portunus pelagicus, Scylla serrata) and three species of mullet (Chelon macrolepis, C. parsia, Mugli cephalus). Unwanted species (dead or alive) were indiscriminately discarded onto land and quantitative interview data confirmed that observations of this discarding reflect a general trend with over 50% of fishers confirming this practice (Figure 2). An additional 5% of fishers stated that, if non-target species were alive, it would be killed (in the instance of snakes) and left on land. Twenty-two percent of respondents stated that they would return unwanted species to the lagoon (making no specification of mortality status), and 17% said that they returned live species to the lagoon. Less than 2% of respondents would utilize non-target species.

The discarding of by-catch was frequent at the landing site surveyed. Levels of discarding were high and varied between gear types and target types (Figure 3). The number of non-target species that were generally discarded per fishing trip was <2 within Fyke net, set trammel net and drift trammel net catches. However, around three species were routinely discarded from trawl catches. For individuals per trip, these numbers were much higher. Generally, fishers targeting finfish discarded much less non-target individuals (16) than fishers targeting shrimp (Fyke = 47; drift trammel = 47; trawl = 64).

When the total number of species and individuals discarded at the site over the study period is taken into account, the magnitude of this issue is realized (Figure 3). Over the study period (7 days), at total of 2,752 individuals were discarded from 63 catches. Proportioally, fishers targeting shrimp accounted for the majority of these discards with discards from fkye net catches being highest (1,040 individuals) and discards from set trammel net catches being lower (191). Compared to discarded species, the total number of individuals (774) that were retained throughout the study was low. There was little difference between the number of species retained per fishing trip across gears and target types, indicative of a general low preference for non-target species.

Across gear types, there were significant differences in the total number of species caught [F_{(3, 59)} = 4.49, p < 0.05] and the number of target species caught [H₍₃₎ = 17.831, p < 0.001]. The magnitude of by-catch (number of non-target species caught) [F_{(3, 59)} = 6.545, p < 0.001] also differed significantly across gear types. Wind-powered trawl nets had the highest total catch and by-catch rates per trip, catching on average 14.30 ± 1.34 species, where 11.90 ± 1.33 species were considered by-catch, amounting to over 80% of species caught. Drifting trammel nets caught a total of 13.56 ± 0.92 species where 11.32 ± 0.89 were considered by-catch, amounting to over 80% of species caught. Set trammel nets consistently had the lowest total catch and lowest by-catch. On average, 9.92 ± 0.81 species where caught using this gear type and 6.67 ± 0.77 species were considered by-catch. This gear type also had the highest target catch per trip (3.25 ± 0.13) in comparison to the other three gears (all <3 target species per trip; Figure 4).

Target species had a strong influence on catch composition. The total number of species caught when shrimp were targeted (12.76 ± 5.12), was significantly higher than when finfish were targeted [9.92 ± 8.11; t₍₆₁₎ = −2.495, p < 0.05]. Significant differences in the amount of by-catch were observed between target types [t₍₆₁₎ = −3.267, p < 0.05] where those targeting shrimp (10.37 ± 0.52) catch a higher number of unwanted species than those targeting finfish (6.67 ± 0.77). Similarly, the number of target species caught differed with target type. Fishers targeting shrimp (2.39 ± 0.07) caught significantly fewer target species than those targeting finfish [3.25 ± 0.13; U₍₆₁₎ = 99, z = −4.045, p < 0.001].

Target catch also had a strong influence on the species that were caught, retained and discarded. The common silver biddy (Gerres oyena) was present in nearly 100% of catches from fishers targeting shrimp. Over 50% of these were juvenile and always discarded (Figure 5). In total, a recorded 577 individuals were discarded over the survey period by shrimp fishers. The streaked spinefoot (Siganus javus) was present in over 80% of shrimp catches, where nearly all were juvenile. Of the 10 by-catch species most common in shrimp catches, individuals from five of these species were always discarded and individuals from the five other species kept if large enough.

Fishers targeting finfish discarded fewer species, but the proportion of juveniles were high in six of the 10 species most present in catches (Figure 6). For example, only juvenile S. javus were present, and were third most common fish across catches and were routinely discarded if they were small. G. oyena was also frequent in finfish catches and always discarded regardless of size. In total, a recorded 76 individuals were discarded. Unlike shrimp catches, G. oyena were present mainly as adults in finfish catches. Only the most common species caught in catches, C. macrolepis, was always kept, and the eight-remaining species were retained only if they were large enough.

Differences in average catch weight across gear types were not reflective of the number of individuals caught. Finfish fishers, who only used set trammel nets, routinely caught the least species and had the highest total catch weight per trip (12.7 ± 3.7 kg), made up entirely of finfish (Figure 7). Shrimp fishers caught a higher number of species per trip than finfish fishers, but landed lower biomass, characteristically landing similar weights of both finfish and shrimp. Fishers using drifting trammel nets caught more finfish (2.1 ± 0.8 kg), in terms of weight than shrimp (1.0 ± 1.0 kg), whereas those (Figure 7). Concerning weight, around 67% of drifting trammel net catches, and around 44% of fyke and trawl catches, were considered by-catch.

The taxonomic composition of by-catch was similar between gear types, with all discards being finfish. Additionally, the age and life stage of by-catch as a whole did not differ significantly across gear types, with all gear types having a similar proportion of juveniles [H₍₃₎ = 3.586, p = 0.310] and adult [H₍₃₎ = 4.573, p = 0.206]. Wind operated trawl nets caught the most juveniles, where, on average 65.6 ± 42.5% of by-catch was juvenile. By-catch from fyke nets included 61.8 ± 44.3% juveniles. Set and drifting trammel nets had an adult by-catch of 53.2 ± 47.4 and 51.9 ± 46.2%, respectively. Similarly, there was no difference in the proportion of juveniles [U₍₂₃₉₎ = 4344.5, z = −1.889, p = 0.059] and adults [U₍₂₃₉₎ = 4762.5, z = −0.910, p = 0.363] between target species. However, fishers targeting finfish caught more adults than those targeting shrimp.

S. javus and G. oyena were the most common species caught in gear types. S. javus was present in 83.3% of set trammel net catches, 86.4% of fyke net catches, 68.4% of drifting trammel net catches, and 90% of wind-powered trawls. G. oyena occurred in 75% of set trammel net catches, 95.5% of fyke net catches and all drifting trammel net and trawl net catches. Across all gear types, the majority of S. javus individuals caught were juvenile, with <1% mature individuals. Fish from the Leiognathidae family were also common within gears targeting shrimp, and mostly present as juveniles. The giant catfish (Netuma thalassina) was frequently landed, more so in gears targeting shrimp. No adults of this species were recorded. This was true for other key species such as the pink ear emperor (Lethrinus lentjan), the blackspot snapper (Lutjanus fulviflamma) and the greasy grouper (E. malabaricus), which, although less abundant, were only present as juveniles. Fish from the Terapontidae family were mostly present as adults. Other Siganus spp., the white-spotted spinefoot (Siganus canaliculatus) and the bronze-lined spinefoot (Siganus insomnis) were present, but were less frequent and caught as adults (Figure 5).

Species assemblages of the total catch differed for gear type. Assemblages separated into distinct groupings within an nMDS ordination plot (Figure 8). Differences were significant (ANOSIM: R = 0.73, p < 0.01) with all pairwise comparisons showing similar significant differences (P < 0.01). Similarity within drift Trammel net landings (SIMPER) was driven mostly by G. oyena, P. indicus, and P. semisulcatus (top three species comprising 45% of the similarity). Similarities within set Trammel net landings were driven by C. macrolepis, E. suratensis and S. javus (58% of the similarity). Within Fyke net landings similarity was driven by P. semisulcatus, G. oyena, and S. javus (comprising 45% of the similarity). Within trawl net landings this similarity was the result of G. oyena, P. semisulcatus, and M. dobsoni (top three species accounting for 40% of the similarity).

Market surveys revealed that fish from three families (Mugilidae, Siganidae, and Ariidae) were routinely sold and considered “high value.” Six other species were recorded but were less common. Additional interviews with market owners suggested that fish from the Mugilidae and Siganidae families were requested every day, in addition to E. suratensis, yet the latter species was not frequently observed during the survey. Four species of shrimp, M. dobsoni, P. indicus, P. semisulcatus, and P. monodon, and two species of crab, the blue swimmer crab (P. pelagicus) and the mud crab (S. serrata), were recorded in markets and considered “high value.” However, only M. dobsoni, P. pelagicus, and S. serrata were in demand at the household level and P. indicus, P. semisulcatus, and P. monodon were sold for export via a broker in Puttalam.

Fish from the mullet family were most commonly observed, occurring in 80% of market surveys. These species were also the most abundant in terms of biomass with an average daily stock of 47.3 ± 65.8 kg. Fish from the Siganidae and Ariidae families were also commonly found in 70 and 60% of surveys respectively but regular stock weight of these was far less at 11.1 ± 7.3 and 3.8 ± 2.1 kg, respectively.

Shrimp was the most important invertebrate for over 65% of respondents. Fish from the Mugilidae family were the preferred food fish for 37% of interview respondents, 20% of respondents stated they were the second most important and a further 13% stating that they were the third most important food fish. Fish from the Siganidae family were noted as the second most important fish, with 12% of respondents stating that it was their first preference, 15% stating it as their second and a further 10% stating it as their third most important food fish. Highly discarded species, G. oyena and fish from the Leiognathidae family, were recorded as having some importance for food, with 7% of respondents stating that G. oyena was third most important in their diet, and 2% and 3% of respondents stating that Leiognathidae sp. were second and third most important within their diet, respectively.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.