its illegal to catch a fish using what method

• Loading metrics

Open Access

Peer-reviewed

Enquiry Article

A Mixed-Method Approach for Quantifying Illegal Fishing and Its Impact on an Endangered Fish Species

• Christopher Thou. Free,
• Olaf P. Jensen,
• Bud Mendsaikhan

x

• Published: Dec ane, 2015
• https://doi.org/10.1371/journal.pone.0143960

Figures

Abstract

Illegal harvest is recognized as a widespread problem in natural resource management. The use of multiple methods for quantifying illegal harvest has been widely recommended yet infrequently applied. We used a mixed-method approach to evaluate the extent, grapheme, and motivations of illegal gillnet fishing in Lake Hovsgol National Park, Mongolia and its bear on on the lake'due south fish populations, specially that of the endangered endemic Hovsgol grayling (Thymallus nigrescens). Surveys for derelict fishing gear indicate that gillnet fishing is widespread and increasing and that fishers by and large utilize iii–4 cm mesh gillnet. Interviews with resident herders and park rangers advise that many residents fish for subsistence during the spring grayling spawning migration and that some residents fish commercially year-round. Interviewed herders and rangers mostly agree that fish population sizes are decreasing simply are divided on the causes and solutions. Biological monitoring indicates that the gillnet mesh sizes used by fishers efficiently target Hovsgol grayling. Of the five species sampled in the monitoring program, only burbot (Lota lota) showed a significant decrease in population affluence from 2009–2013. Nonetheless, grayling, burbot, and roach (Rutilus rutilus) all showed pregnant declines in average body size, suggesting a negative fishing affect. Data-poor stock assessment methods advise that the line-fishing endeavour equivalent to each resident family fishing 50-chiliad of gillnet 11–15 nights per year would be sufficient to overexploit the grayling population. Results from the derelict line-fishing gear survey and interviews suggest that this level of effort is not implausible. Overall, we demonstrate the ability for a mixed-method approach to effectively depict an illegal fishery and suggest that these methods be used to appraise illegal fishing and its impacts in other protected areas.

Citation: Free CM, Jensen OP, Mendsaikhan B (2015) A Mixed-Method Arroyo for Quantifying Illegal Fishing and Its Bear upon on an Endangered Fish Species. PLoS ONE ten(12): e0143960. https://doi.org/ten.1371/journal.pone.0143960

Editor: Alexander J. Travis, Cornell Academy College of Veterinary Medicine, United states of america

Received: August 28, 2015; Accepted: Nov 11, 2015; Published: December 1, 2015

Copyright: © 2015 Gratuitous et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited

Data Availability: All relevant data are within the paper and its Supporting Information files.

Funding: This written report was supported past the National Science Foundation (NSF), Office of International Science and Engineering science (OISE), International Research Experiences for Students (IRES), https://www.nsf.gov/funding/pgm_summ.jsp?pims_id=12831, NSF-IRES OISE 1064843—OPJ; Usa Agency for International Development (USAID), Partnerships for Enhanced Engagement in Research (PEER), http://sites.nationalacademies.org/pga/peer/alphabetize.htm, USAID-PEER Project 98—BM; American Center for Mongolian Studies (ACMS), http://mongoliacenter.org/field, Field Inquiry Fellowship—CMF; Grayling Enquiry Trust (GRT), http://www.graylingresearch.org/grant, Research Grant—CMF; National Oceanic and Atmospheric Administration (NOAA), National Marine Fisheries Service (NMFS), http://seagrant.noaa.gov/FundingFellowships/NMFSSGFellowship.aspx, NMFS-Sea Grant Population Dynamics Fellowship – CMF. The funders had no role in study pattern, data collection and analysis, decision to publish, or training of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Illegal, unreported, and unregulated (IUU) fishing undermine efforts to sustainably manage fish stocks and threaten fish populations worldwide [one]. Managers must know as much as possible about the extent, character (e.1000., gear types, target/bycatch species, timing, location), and motivations of illegal fishing to effectively develop and implement regulations. However, quantifying illegal line-fishing is inherently hard: it is generally covert and significant incentives exist for informants to withhold data [two]. Furthermore, upkeep and human resources constraints often restrict efforts to monitor illegal resource utilize, peculiarly in developing countries [3]. In that location is a demand to develop inexpensive yet informative methods for quantifying illegal angling and its impacts.

Indirect observation, the utilise of signs of illegal activeness as an indicator of not-compliance, has been commonly used to narrate illegal resource use in terrestrial systems [four], but has been infrequently used in marine systems [5], and to our knowledge, has never been used in freshwater systems. In marine systems, dynamite boom craters [half-dozen,7] and derelict angling gear [8] take been used equally indicators of illegal line-fishing, simply take by and large failed to quantitatively measure non-compliance [5]. About successful quantifications of illegal fishing compare the amount of derelict fishing gear within and outside reserve boundaries [9–12], but such comparisons are of picayune use in places without reserves or where the areas outside reserves are undesirable to fishers. The total capacity for indirect observation to reveal rich and quantitative information nearly illegal line-fishing remains unexplored.

Indirect ascertainment offers several advantages over other approaches for assessing illegal fishing. It does not crave large amounts of labor, specialized equipment, or training and can exist recorded during routine enforcement patrols or biological surveys [13]. Repeated surveys can reveal spatial and temporal patterns of non-compliance [8–10,14] that can be compared to changes in fish communities to examine the effects of illegal fishing [xv]. Although indirect ascertainment mostly cannot identify specific violators or motivations for non-compliance, they can contribute to a comprehensive understanding of non-compliance when combined with other methods, such as direct questioning [9–10].

In this study, we used a mixed-method approach to evaluate the extent, grapheme, and motivations of illegal gillnet fishing in Lake Hovsgol National Park (LHNP), Mongolia and its bear upon on the lake'southward fish populations, peculiarly that of the endangered endemic Hovsgol grayling (Thymallus nigrescens). Despite the closure of the park to gillnet fishing in 1992, illegal fishing is known to persist [16–17]. Nosotros used iv complementary methods to describe this fishery and evaluate its impacts: (one) surveys for derelict fishing gear, an indirect indicator of angling activity, to evaluate how much illegal fishing is occurring, where illegal fishing is occurring, and what gear is beingness used; (ii) interviews with herders living within the park and park rangers to validate and contextualize the results of the surveys for derelict angling gear; (3) biological monitoring to identify fish species vulnerable to gillnet fishing and evaluate changes in population abundance potentially caused past fishing; and (iv) data-poor stock cess methods to estimate the effort required to overexploit the Hovsgol grayling population.

Overall, nosotros demonstrate the ability for a mixed-method approach to describe an illegal gillnet fishery and suggest that these methods could be used to finer and inexpensively assess illegal fishing and its impacts in other protected areas.

Methods

Study site

Lake Hovsgol (51°05'50"N, 100°thirty'Eastward) is located in the mountains of northern Mongolia at the southern border of the Siberian taiga forest. It is the 19^th largest lake in the globe by volume (480 km³) and has a maximum depth of 262 g and surface area of 2,760 km^two [eighteen]. The lake was established as a National Park in 1992 and is by and large undeveloped. The majority of the resident population lives in two towns on the lakeshore: Hatgal (pop. 2,980) and Hankh (popular. two,460; [19]). Tourist camps line the southwestern shore and herding families alive intermittently forth the lakeshore (Fig 1). About of the park'southward ~35,000 annual visitors enter and remain in the southern portion of the park [20].

Fig i. Location of shoreline surveys for derelict line-fishing gear, fish population monitoring sites, and interviews with park rangers and resident herders in Lake Hovsgol National Park (LHNP), Mongolia.

Grey and blackness bars signal the density (# km^-ane) of derelict gillnet items observed in the 2013 (n = 10) and 2014 (northward = seven) surveys, respectively (note unlike y-axis scale for Site 10). Black site numbers indicate the 7 sites where fish population monitoring surveys were conducted in 2009 and 2011–13. Solid black lines indicate the park boundary and 17 ranger districts. Five rangers from five districts (dark gray; Hankh town limits correspond one district) were interviewed. Herders were interviewed at Sites 3 (north = 3), 4 (due north = iii), and vi (n = four). Minor white triangles indicate tourist camps, big blackness circles point boondocks centers, dotted blackness lines indicate primitive roads, and solid gray lines indicate rivers and seasonal steams.

https://doi.org/ten.1371/journal.pone.0143960.g001

Lake Hovsgol has ten fish species, the near arable of which, the Hovsgol grayling (Thymallus nigrescens), is owned to the lake and is listed every bit endangered on the Mongolian Reddish List due to climate alter and illegal fishing [16]. Hovsgol grayling are more common in littoral areas than pelagic areas and are most abundant forth the western shore [21]. A portion of the grayling population spawns in tributary streams in late spring while another portion spawns in the littoral in late summer [22]. The prevalence, fidelity, and success of these spawning strategies are unknown.

The thin literature on Mongolian fisheries suggests that commercial fishing for Hovsgol grayling, lenok (Brachymystax lenok), roach (Rutilus rutilus), perch (Perca fluviatilis), and burbot (Lota lota) removed equally much equally 200–400 tons annually before the park was established ([23]; S1 Table). Despite the ban on gillnet fishing, agile gillnets are often observed and grayling and lenok are frequently sold in Hatgal and along the southwestern shore route. Recreational claw-and-line fishing is legal within the park and is regulated through permits and season and bag limits. Subsistence angling during the spring spawning migration, though officially illegal, is generally tolerated.

Surveys for derelict line-fishing gear

We surveyed and collected derelict fishing gear at ten sites on the Lake Hovsgol shoreline in July 2013 and resurveyed vi of these sites in July 2014 (Fig 1). Although fishing gear found in the 2013 surveys could represent several years of accumulation and even pre-date the ban on gillnet fishing, gear plant in the 2014 resurveys must correspond accumulation over the preceding year, since all gear was removed from these sites during the 2013 surveys. Sites were selected equally role of a long-term fish monitoring written report [21]; though not-random, they provide excellent spatial coverage and access to points and bays on all sides of the lake. In 2013, we censused 54.nine km of shoreline (10 sites, 13 transects, 0.four–eight.5 km each, ~13% of total shoreline) for all anthropogenic debris, including derelict fishing gear, betwixt the h2o and wrack lines [24]. In 2014, nosotros recensused 31.9 km of the original transects (7 sites/transects, 1.iii–8.iii km each) for derelict line-fishing gear only. Because transect widths were variable, we report linear (km^-1) rather than areal (km^-two) debris density. Derelict fishing gear was classified into the following gillnet categories: whole internet, net fragment, float line, lead line, foam bladder, or bottle float (S1 Fig); and hook-and-line categories: rod, monofilament, lure, or bobber. Bottles, string/rope, and stakes without mesh, floats, weights, or lines were non considered line-fishing gear. Nosotros weighed each detail and measured the mesh size (knot to knot distance) of every whole gillnet or gillnet fragment.

Interviews with herders and rangers

The Rutgers Academy Internal Review Board (IRB) approved our interview protocol (Protocol E14-675) and all respondents gave informed verbal consent (written consent is problematic in quondam Soviet regions) as approved past the IRB.

Nosotros used a semi-structured questionnaire to interview ten herding families from 3 sites (Fig 1) nearly their fishing habits, angling action they observe, and status and conservation of fish in the lake (S1 Appendix). The first household at each site was selected opportunistically and additional households were recommended by this family based on proximity and availability. This "snowball sampling" method is commonly used to find respondents in isolated or hard-to-access groups [25]. There was no indication that recommendations were biased towards fishing or not-angling households. We interviewed seven male and 3 female person heads of household. Family and herd sizes ranged from 3–7 people and four–630 animals, respectively.

We used a dissimilar semi-structured questionnaire to interview five park rangers, including the head ranger, from v of 17 ranger districts (Fig 1) almost the frequency and character of illegal angling, actions taken against illegal fishers, and condition and conservation of fish in the lake (S2 Appendix). The interviewed rangers were male and had worked as rangers for 3–fifteen years. They were responsible for districts that varied in expanse (22–398 ha) and number of families (32–ane,264 families).

Biological sampling, gillnet grab efficiency, and population trends

Nosotros used fish monitoring data to estimate grab rates for gillnet mesh sizes used by fishers and to evaluate changes in fish population affluence and body size.

The Rutgers University Animate being Intendance and Facilities Committee canonical our fish sampling protocol (Protocol 11–005). Permission to conduct field research (Allow half-dozen/445) was granted by the Mongolian Ministry of Environment and Dark-green Development (MEGD). In July 2009 and 2011–thirteen, nosotros prepare 2 monofilament horizontal gillnets at seven of the x surveyed sites (Fig i). Both gillnets were ii m deep and xx 1000 long with 4 m panels of 2.54, 3.81, 5.08, half-dozen.35, and 7.62 cm bar mesh. They were set at to the lowest degree 100 m autonomously, perpendicular to shore, using a stationary bottom set in h2o < ten m deep, and were fished overnight (viii.5–10.5 hour) at each location. Captured fish were identified and measured to the nearest millimeter in full length. Weights for fish without weight measurements were estimated using length-weight parameters derived from our data (S2 Fig).

Vulnerability of fish to gillnets can vary depending on species, torso size, and mesh size. We calculated catch-per-unit of measurement-of-effort (CPUE) for each gillnet panel in terms of count and biomass (#/kg 10 g^-1 night^-1) to make up one's mind species-specific and overall take hold of rates for each mesh size. We also calculated the species-specific CPUE of each gillnet set in terms of count and biomass (#/kg night^-one) and used linear mixed furnishings models to examine changes in species-specific abundance from 2009–13 while accounting for sampling site as a random result on the model intercepts. Decreases in trunk size tin can exist a useful indicator of fishing impacts when changes in abundance cannot be accurately assessed [26]. Therefore, nosotros also used linear mixed furnishings models to examine changes in species-specific trunk size (length/weight) from 2009–xiii. P-values were generated through likelihood ratio tests of the total models and goose egg 'intercept just' models. All analyses were performed in R version iii.2.0 [27] and mixed furnishings models were fit using the lme4 package [28].

Potential population level impacts on Hovsgol grayling

We used methods commonly used in data-poor fisheries management to estimate the maximum sustainable yield (MSY) for Hovsgol grayling and evaluate the likelihood that illegal gillnet fishing could approach or exceed this threshold. Angling at a rate greater than that which results in MSY is a mutual definition of overfishing [29].

Meta-analyses have shown that fish life history traits tin be used to judge natural mortality rates [30], which tin can in turn exist used to estimate F_MSY [31], the angling bloodshed charge per unit resulting in MSY. We estimated the Hovsgol grayling natural mortality charge per unit (1000) using iii dissever life history invariant approaches (Table 1) and applied the Zhou et al. [31] method to judge F_MSY every bit 0.87*Thou. We used a length-converted catch curve analysis [32] to calculate full mortality (total mortality = fishing bloodshed + natural mortality) to place an upper limit on possible natural mortality rates and gauge electric current fishing mortality rates. More than details on the mortality estimation methods are provided in S3 Appendix.

We then calculated MSY for each F_MSY estimate using the Ahrenstorff et al. [21] hydroacoustic biomass judge for Hovsgol grayling (4.4 ± 0.nine kg ha^-one) and estimated the number of nights of gillnet line-fishing required to attain each MSY assuming fishers use 50-g gillnets with 2.54-cm mesh, the optimal mesh size for targeting grayling (~fifteen kg grayling night^-one; encounter Gillnet take hold of efficiency results). Finally, nosotros estimated the number of fishers required to achieve each MSY assuming fishers utilise l-k of gillnet 100 nights yr^-1. These assumptions seem reasonable given the number of nets used by observed and cocky-reported fishers and reports that angling continues throughout the winter (encounter Interviews with herders and rangers results).

Results

Surveys for derelict angling gear

A total of 220 (5.78 kg) and 281 (3.82 kg) pieces of derelict fishing gear were collected in the 2013 and 2014 surveys, respectively. Fishing gear comprised 25% of the total weight of plastic debris observed during the 2013 surveys [24]. Derelict gillnet material, the majority of line-fishing gear found in both years (Fig two), was plant in all but 2 2013 transects and all 2014 transects (Fig 1). Cream floats were the near abundant gillnet debris items past count, probable due to their ability to separate from nets and disperse widely; gillnet fragments were the almost abundant gillnet debris items by weight, likely due to their large size and heavy lead lines. Gillnet fragments ranged from 2–8 cm in mesh size with 3–4 cm mesh being the most mutual past both count and weight (Fig 2). All six active gillnets observed had 3.0 cm mesh. The density of derelict gillnet textile varied among transects, only in both years, Site 7, the almost remote and difficult to admission site, had the lowest density of gillnet fabric and Site x (Har Us), the primary location of the spring spawning migration fishery, had the highest density of gillnet textile. The density of derelict gillnet material in resurveyed sites was college in 2014 than 2013 at all but Site vii suggesting that illegal line-fishing may be increasing (Fig ane).

Fig two. Average density of derelict fishing gear past category (GN = gillnet fabric) and derelict gillnet fragments by mesh size in count and weight.

Confined betoken average densities amid the 2013 (night gray, n_sites = x, n_transects = fourteen) and 2014 shoreline transects (light grey, n_sites/north_transects = 7) weighted by transect length. Panels A and B betoken density in count (# km^-i) and Panels C and D point density in weight (g km^-ane). Note variable y-centrality scales.

https://doi.org/10.1371/journal.pone.0143960.g002

Interviews with herders

All of the interviewed herding families (n = x) reported fishing and observing others angling (S4 Table). Families on the eastern shore reported fishing with gillnets repeatedly throughout the year and during the spring grayling spawning migration. They also reported observing commercial gillnet fishers from Hatgal during the winter and during the spring spawning migration, and they reported finding enforcement ineffective. In contrast, families on the northwestern shore reported angling with rods or by hand only one time per spring spawning migration. They reported no commercial fishing action and plant enforcement effective. All of the families reported that Russian visitors fish recreationally year-circular but particularly in winter with ice fishing rods and gillnets (S4 Table).

All of the families reported fishing primarily for Hovsgol grayling and primarily for household consumption; merely ane family from the eastern shore reported selling fish (S4 Table). Families reported angling primarily during the spawning migration because (1) grayling soup is healthy after the long winter; (2) fish are more abundant and easier to take hold of than whatsoever other time; (iii) herders are besides busy to fish, or they live away from the lake, the rest of the twelvemonth; (iv) cooking grayling soup interferes with milk production, their principal food source; and (5) eating grayling allows them to delay the slaughtering of herd animals until they accept had fourth dimension to fatten.

Nearly all of the interviewed herders stated that fish population sizes have decreased dramatically (S4 Table). Many recalled that migrating fish were one time so numerous that the rivers appeared to "exist but fish and no water." Most of the herders too asserted that fish torso sizes have decreased and that large lenok and burbot take become specially rare (S4 Table). The herders stated that "local people should protect the lake and fish" but offered few concrete ideas for achieving this objective (S4 Table).

Interviews with rangers

The rangers reported that recreational, commercial, and subsistence line-fishing all occur in LHNP (S5 Tabular array). The rangers agreed that the majority of recreational fishers are non-local Mongolians or foreigners who fish with rods primarily in summertime simply likewise through the ice in wintertime. The rangers reported that recreational fishers are more often than not permitted and compliant with the law. All just one ranger reported that local Mongolians apply gillnets to target Hovsgol grayling and lenok for subsistence or commercial purposes (S5 Table). The rangers reported that subsistence fishers fish almost exclusively at river mouths during the spring spawning migration and that commercial fishers come up predominantly from Hatgal due to that town'southward proximity to the developed southwestern shore and the city of Mörön. The rangers asserted that the boondocks of Hankh is too remote and undeveloped for commercial fishing to exist viable. The rangers reported that commercial gillnet fishing occurs year-round and that fishing when the lake is freezing, thawing, or entirely frozen may even exist preferred (S5 Table).

The rangers were divided on the status of fish in the lake: three rangers reported that fish population sizes are decreasing and two rangers reported that they are increasing (S5 Table). The rangers who reported fish population sizes to be decreasing reported that lenok have get especially rare. The bulk of rangers reported that fish body sizes have not inverse (S5 Table). The rangers were also divided on the best approach to conservation. The head ranger asserted that the native Not bad Cormorant (Phalacrocorax carbo) population is the principal threat to fish and that their population must be controlled. Another ranger suggested that grayling die naturally after the spring spawning migration (an assertion that is not supported by the scientific literature) and that these migrations must therefore exist prevented. The remaining rangers emphasized the importance of improved enforcement during the spawning migration (S5 Table).

The rangers offered a detailed description of fishing at Har Us mineral spring (Site 10), the primary location of the spring grayling spawning migration fishery. Mineral springs are culturally important to Mongolians and visiting this spring in May-June is a longstanding social tradition. Rangers are instructed not to enforce the gillnet ban on fishers at Har Usa during this time. The rangers reported that over 570 people visited the spring in 2013 and set up a total of lx–100 nets per solar day with an average grab of l–70 grayling per net. They estimated that three,600 grayling were caught per day during peak migration (Jun vii–12) and i,000–1,500 grayling per mean solar day from May thirty-Jun half-dozen and Jun 13–24. Based on this report, we estimate that the Har U.s. fishery removes ~33,000 fish annually.

Gillnet catch efficiency and population trends

The 2.54-cm mesh in our survey gillnets maximized total nightly catch by numbers considering it maximized the catch of the abundant Hovsgol grayling (S3 Fig). The 3.81- and 5.08-cm mesh sizes showed similar catch rates and maximized total nightly catch by biomass considering they maximized the catch of larger-bodied lenok and burbot (Fig 3); still, the median nightly take hold of biomass of the two.54-cm mesh was comparable to those of the 3.08- and v.81-cm mesh and the 2.54-cm mesh captured fish during every gillnet ready, while the larger mesh sizes were oft observed empty.

Fig 3. Catch-per-unit-of-endeavor (CPUE; kg 10 yard^-1 dark^-i) by mesh size for the five nigh abundant species in gillnet catches and the sum of their weight.

Data from the two 5-panel sequential mesh gillnets used at seven sites in 2009 and 2011–2013 (14 sets yr^-1, 56 sets total). Boxplots point median (heavy blackness line), interquartile range (IQR; box), i.5 times the IQR (whiskers), and farthermost values (open up circles). Note variable y-axis scales.

https://doi.org/x.1371/periodical.pone.0143960.g003

Analysis of the biological monitoring data identified significant reductions in body size for 3 species over the sampling period (2009–xiii), but a significant alter in CPUE for merely i species. Linear mixed effects regression on species-specific CPUE indicates that merely burbot population affluence decreased significantly from 2009–thirteen (Fig iv; S4 Fig). Linear mixed effects regression on body size indicates that grayling, roach, and burbot body size decreased significantly from 2009–xiii (Fig v; S5 Fig). The abundance and torso size of other species remained abiding.

Fig four. Trends in the abundance of the five near arable fish species in gillnet catches from 2009–2013.

Points bespeak the CPUE (kg net^-one night^-1) of each v-panel sequential mesh gillnet set (2 nets site^-ane x 7 sites year^-i = xiv sets year^-1). Dark lines indicate linear mixed effects regressions fit to the catch data, greyness shading indicates the conviction interval for each regression, and dashed lines indicate the prediction interval for the data. P-values are indicated in the upper right corner of each panel. Points are jittered around year for display. Note variable y-centrality scales.

https://doi.org/10.1371/journal.pone.0143960.g004

Fig 5. Trends in the body size of the five most abundant fish species in gillnet catches from 2009–2013.

Points indicate the weight (kg) of every fish caught in gillnet sets that yr (two nets site^-1 x seven sites year^-one = 14 sets twelvemonth^-1). Nighttime lines indicate linear mixed furnishings regressions fit to the catch data, grey shading indicates the confidence interval for each regression, and dashed lines bespeak the prediction interval for the information. P-values are indicated in the upper right corner of each panel. Points are jittered around year for display. Annotation variable y-axis scales.

https://doi.org/x.1371/periodical.pone.0143960.g005

Potential population level impacts on Hovsgol grayling

Estimates of Hovsgol grayling natural mortality (M) ranged 0.25–0.37 (Table 1). A total bloodshed estimate of 0.42 (S7 Fig) implies fishing moralities of 0.06–0.15, all of which are less than their associated F_MSY estimates (Table 1). The F_MSY estimates imply MSY values of ~255–331 metric tons yr^-1, which could be achieved in ~17,000–22,000 nights of fishing with 50-m optimal mesh gillnets (Tabular array 1). Although these estimates seem big for a low-density resident population, they could be accomplished past 170–220 fishers using 50-1000 of optimal mesh gillnet 100 nights year^-1 (roughly twice per week). With an estimated permanent population of 5,440 in LHNP and average family size of 3.6 people [19], this try could be attained if xi.3–14.6% of families participated in the fishery (Table 1). Alternatively, this effort could be attained if every family living in the park fished with fifty-grand of optimal mesh gillnet 11.3–14.vi nights per yr.

Discussion

Noesis of illegal angling in Lake Hovsgol National Park (LHNP) has been anecdotal and express in its usefulness to managers, only with a mixed-method approach, we accept empirically described the extent, graphic symbol, and motivations of illegal angling and its potential bear upon on the lake's fish populations.

Our mixed-method approach reveals a fuller understanding of illegal line-fishing in LHNP than using a unmarried method alone. Each method validates, contextualizes, and builds upon the others to construct a consistent story for a complex fishery: (1) surveys for derelict line-fishing gear quantitatively describe the extent, location, and methods of angling: gillnet fishing is widespread and increasing and fishers more often than not use 3–4 cm mesh gillnet; (2) interviews with herders and park rangers contextualize these results by qualitatively describing the motivations of fishers, character of fishing, and status of fish in the lake: many residents gillnet fish for subsistence during the spring grayling spawning migration, some residents gillnet fish commercially year-circular, and fish population sizes are decreasing; (3) biological monitoring documents the vulnerability of fish to gillnets besides as population-level trends in fish affluence and body size: the gillnet mesh sizes used by fishers efficiently target Hovsgol grayling and grayling, burbot, and roach exhibit negative population-level trends; and (4) data-poor stock assessment analyses demonstrate that plausible levels of fishing attempt by Lake Hovsgol residents using gillnets have the capacity to consequence in overexploitation of the Hovsgol grayling population. Though seemingly intuitive, the use of multiple methods to quantify and characterize illegal resource use has been rare and should exist more than widely used by conservation scientists and resource managers [4,v].

Our surveys for derelict fishing gear are an improvement to previous studies because we use repeated surveys to measure re-accumulation rates and biological monitoring data to evaluate the vulnerability of fish to the gear observed in surveys. The majority of studies take focused on comparing the density of derelict gear within and outside marine reserves for quantifying non-compliance and fail to measure or study accumulation rates (e.yard., [9–11]). A few studies have measured the accumulation rates of derelict gear among habitat types to inform cleanup efforts but have non used the results to understand non-compliance (east.thou., [8,33,34]). Only Williamson et al. [14] and the nowadays study have linked these objectives and used both the density and re-accumulation charge per unit of derelict fishing gear to evaluate temporal and spatial trends in non-compliance. By measuring re-accumulation, we show non simply that the observed gillnet was used recently and does not pre-appointment the ban on gillnet fishing, just also that gillnet line-fishing is becoming increasingly common. Neither Williamson et al. [14] or our written report properly control for the influence of habitat characteristics (east.yard., shore/lesser cover or wind/wave exposure) on aggregating and future studies must consider these covariates when identifying hotspots of illegal fishing.

Although our interview method likely underestimates the charge per unit of non-compliance [35,36], information technology provides a relative description of the frequency of illegal fishing and important information most the motivations for non-compliance, which cannot be gained using other respondent-based approaches [iv]. The biases and limitations of straight questioning (DQ) can be reduced when researchers take long-standing relationships with the customs [37,38] or past interviewing multiple stakeholders [37,39]. In our study, this likely contributes to the discrepancy in personal fishing habits reported past herders on the eastern and western shores. Whereas eastern shore herders, with whom we have long partnerships, reported frequent gillnet employ, western shore herders reported fishing by hook and line or by hand only. Although this may reflect real geographic differences, it may also reflect social desirability bias [40], as western shore herders might exist less comfortable revealing sensitive information to united states of america. In our study, this bias is partially corrected by interviewing multiple stakeholders and by inquiring well-nigh observed illegal behavior [37,39]. For case, herders were more likely than park rangers to characterize enforcement as ineffective and park rangers were more likely than herders to draw illegal angling. Similarly, although some respondents were likely to underreport personal fishing, they may not be every bit likely to underreport observed fishing past others.

Because of these biases, recent papers promote the randomized response (RRT; [41]) and item count techniques (ICT; [42]) over DQ for quantifying not-compliance [36,43–45], but we argue that DQ more hands and fully reveals the motivations for non-compliance [4], which is essential information for successful management [46]. RRT and ICT incentivize honest responses virtually illegal behavior by protecting anonymity and generally generate more accurate estimates of the proportion of the sample population engaging in illegal beliefs [35,36]; however, these approaches require large sample sizes and foreclose researchers from implicitly discerning motivations for not-compliance past linking behaviors with covariates or from explicitly inquiring well-nigh the motivations for non-compliance [45]. DQ, on the other manus, allows researchers to ask well-nigh the motivations for non-compliance, importance of natural resource to culture or livelihood, and desire for changes to management rules. Managers must consider the socioeconomic functions of resource use and DQ should remain in the conservation science toolbox.

Although the population-level impacts observed in our biological monitoring data cannot necessarily be attributed to illegal fishing, they bespeak the importance of improving fisheries management in LHNP, particularly given the feasibility for gillnet fishers to overexploit the Hovsgol grayling population, as indicated past the data-poor stock assessment analysis. These calculations represent a simplification of population dynamics fabricated necessary past the lack of fourth dimension serial of fishery removals or estimates of biological parameters needed for more than complex information-poor assessment methods [47]. Still, our indirect estimates of G for Hovsgol grayling are like to direct estimates of M for Arctic grayling (T. arcticus), a close relative (0.29 average; S3 Table). Furthermore, all of our MSY estimates indicate that overexploitation is possible even with only a small percentage of the population participating in the fishery using gillnets, an inexpensive and widely available fishing gear. The threat of overexploitation is not unrealistic given that grayling, as a taxonomic grouping, can exist susceptible to anthropogenic influences equally has been seen with the extirpation of many Due north American Arctic grayling populations in Montana and Wyoming [48]. Salmonids are vulnerable to exploitation and other disruptions during their spring spawning migrations [49] and managers must carefully consider the value and touch of the spring spawning migration fishery.

The results of our mixed-method approach signal that illegal fishing is a problem in Lake Hovsgol simply that fish as well serve an important socioeconomic function. An effective management system will need to incorporate the needs of local people as well as address the synergistic pressures of climate modify, h2o pollution, increasing tourism, and invasive species on LHNP'due south fish populations. In the terminal xl years, regional air temperatures have increased 2.i°C [l], a rate of warming more than three times faster than the global average [51], which has prompted the drying of many of Lake Hovsgol's previously reliable streams and loss of grayling spawning habitat [16,17]. Increasing tourism may result in increased fishing force per unit area, habitat destruction, water pollution, and invasive species introductions without proper management. Lake Hovsgol is already heavily polluted with household trash and will merely become more polluted with additional strains on its inadequate waste direction system [24]. Although no invasive species have established to date, the successful introduction of a new fish or aquatic plant species could alter this otherwise intact ecosystem [52].

Fishing, historically uncommon in Mongolia's pastoralist civilisation, may be gaining prevalence as a new source of food, income, or recreation, particularly every bit climatic change makes herding more than difficult [53] and urban Mongolians acquire more globalized tastes in food and leisure [54]. At the same time, Mongolia aims to protect 30% of the country past 2030, more doubling the area currently under protection [55]. These trends forecast connected conflicts between economic and conservation objectives and the mode in which these conflicts are resolved or ignored in the iconic LHNP could shape future protected surface area management in the country.

Supporting Information

S3 Fig. Catch-per-unit-of-effort (CPUE; # 10 thousand^-1 night^-1) by mesh size for the five virtually abundant species in gillnet catches and the sum of their catch.

Data from the ii 5-panel sequential mesh gillnets used at vii sites in 2009 and 2011–2013 (14 sets yr^-1 x 4 yr = 56 sets total). Boxplots point median (heavy blackness line), interquartile range (IQR; box), 1.5 times the IQR (whiskers), and extreme values (open circles). Notation variable y-centrality scales.

https://doi.org/10.1371/journal.pone.0143960.s008

(PNG)

S4 Fig. Trends in the abundance of the five most abundant fish species in gillnet catches from 2009–2013.

Points indicate the CPUE (# net^-1 nighttime^-1) of each 5-panel sequential mesh gillnet fix (ii nets site^-1 x 7 sites yr^-one = 14 sets yr^-1). Night lines indicate linear mixed effects regressions fit to the catch data, greyness shading indicates the confidence interval for each regression, and dashed lines indicate the prediction interval for the data. Points are jittered effectually year for display. P-values are indicated in the upper right corner of each panel. Note variable y-centrality scales.

https://doi.org/ten.1371/journal.pone.0143960.s009

(PNG)

S5 Fig. Trends in the torso size of the five most abundant fish species in gillnet catches from 2009–2013.

Points indicate the full length (mm) of every fish caught in gillnet sets that year (2 nets site^-ane ten vii sites yr^-i = 14 sets twelvemonth^-ane). Nighttime lines indicate linear mixed effects regressions fit to the catch information, grey shading indicates the confidence interval for each regression, and dashed lines indicate the prediction interval for the data. P-values are indicated in the upper right corner of each panel. Points are jittered around twelvemonth for display. Annotation variable y-axis scales.

https://doi.org/10.1371/periodical.pone.0143960.s010

(PNG)

S6 Fig. Estimates of the life history characteristics used to summate natural bloodshed (Grand) for Hovsgol grayling.

See Table 1 for M interpretation methods and results. In (A), L _inf , K, and t _max were estimated from anile otoliths and a von Bertalanffy growth model (black line) fit through the observed historic period-size relationship and origin (Tsogotsaikhan et al. in review). In (B), GSI was estimated as the mean gonadosomatic index (GSI) for all observed grayling (Jensen, unpublished data). In (B), the black line indicates a linear regression fit and the grey shading indicates the conviction interval for the regression. Life history characteristics are marked and labeled in both panels.

https://doi.org/10.1371/journal.pone.0143960.s011

(PNG)

S7 Fig. The (A) length and (B) length-converted age structure of the Hovsgol grayling population.

The length strucutre was observed in the Ahrenstorff et al. (2012) hydroacoustic surveys. In (B), the solid blackness line indicates a linear regression fit to the log-transformed trailing arm of the age structure. The dashed black lines point the confidence interval for the regression. Z is equal to the negative gradient of the regression.

https://doi.org/x.1371/journal.pone.0143960.s012

(PNG)

Acknowledgments

We are grateful for the devoted work of our many field assistants. We too give thanks Steven Alexander, Talia Young, and the members of the Jensen lab for their helpful feedback on the manuscript. The Taimen Conservation Fund, Mongol Ecology Center, Sweetwater Travel, and Hovsgol Travel provided field support and aid. This piece of work was supported past a NSF-IRES grant (OISE 1064843) to OPJ, a USAID-PEER grant (Project 98) to BM, and American Middle for Mongolian Studies and Grayling Research Trust grants to CMF. CMF is funded by a NMFS-Sea Grant Population Dynamics Fellowship.

Author Contributions

Conceived and designed the experiments: BM CMF OPJ. Performed the experiments: BM CMF OPJ. Analyzed the data: CMF OPJ. Wrote the newspaper: CMF OPJ. Conducted and translated interviews: BM.

References

1. 1. Agnew DJ, Pearce J, Pramod Thousand, Peatman T, Watson R, Beddington JR, et al. Estimating the worldwide extent of illegal angling. PLoS One. 2009; 4(2): e4570. pmid:19240812
   • View Commodity
   • PubMed/NCBI
   • Google Scholar
2. 2. Renzetti CM, Lee RM. Researching sensitive topics. Newbury Park, CA: Sage Publications; 1993.
3. 3. James AN, Gaston KJ, Balmford A. Balancing the Earth's accounts. Nature. 1999; 401(6751): 323–324. pmid:16862091
   • View Article
   • PubMed/NCBI
   • Google Scholar
4. 4. Gavin MC, Solomon JN, Bare SG. Measuring and monitoring illegal use of natural resources. Conserv Biol. 2010; 24(1): 89–100. pmid:20015259
   • View Article
   • PubMed/NCBI
   • Google Scholar
5. five. Bergseth BJ, Russ GR, Cinner JE. Measuring and monitoring compliance in no-take marine reserves. Fish Fish. 2013; xvi(2) 240–258.
   • View Article
   • Google Scholar
6. vi. Guard 1000, Masaiganah M. Dynamite angling in southern Tanzania, geographical variation, intensity of utilise and possible solutions. Mar Pollut Bull. 1997; 34(10): 758–762.
   • View Article
   • Google Scholar
7. 7. Crawford BR, Siahainenia A, Rotinsulu C, Sukmara A. Compliance and enforcement of community-based coastal resource management regulations in North Sulawesi, Republic of indonesia. Coast Manage. 2004; 32(i): 39–50.
   • View Article
   • Google Scholar
8. 8. Chiappone G, Swanson DW, Miller SL, Dienes H. Spatial distribution of lost angling gear on fished and protected offshore reefs in the Florida Keys National Marine Sanctuary. Caribb J Sci. 2004; 40(3): 312–326.
   • View Article
   • Google Scholar
9. 9. Cinner JE, Marnane MJ, McClanahan TR. Conservation and customs benefits from traditional coral reef direction at Ahus Isle, Papua New Guinea. Conserv Biol. 2005; 19(half dozen): 1714–1723.
   • View Article
   • Google Scholar
10. 10. Cinner JE, Marnane MJ, McClanahan TR, Almany GR. Periodic closures every bit adaptive coral reef management in the Indo-Pacific. Ecol Soc. 2006; xi(1): 31.
    • View Article
    • Google Scholar
11. 11. McClanahan TR, Marnane MJ, Cinner JE, Kiene WE. A comparison of marine protected areas and alternative approaches to coral-reef management. Curr Biol. 2006; sixteen(fourteen): 1408–1413. pmid:16860739
    • View Article
    • PubMed/NCBI
    • Google Scholar
12. 12. McClanahan TR, Graham N, Wilson SK, Letourneur Y, Fisher R. Effects of fisheries closure size, age, and history of compliance on coral reef fish communities in the western Indian Bounding main. Mar Ecol Prog Ser. 2009; 396: 99–109.
    • View Commodity
    • Google Scholar
13. 13. Bleher B, Uster D, Bergsdorf T. Assessment of threat status and direction effectiveness in Kakamega Wood, Kenya. Biodivers Conserv. 2006; 15(iv): 1159–1177.
    • View Commodity
    • Google Scholar
14. 14. Williamson DH, Ceccarelli DM, Evans RD, Hill JK, Russ GR. Derelict fishing line provides a useful proxy for estimating levels of non-compliance with no-take marine reserves. PLoS One. 2014; 9(12): e114395. pmid:25545154
    • View Article
    • PubMed/NCBI
    • Google Scholar
15. 15. Jachmann H. Monitoring police-enforcement performance in nine protected areas in Ghana. Biol Conserv. 2008; 141(1): 89–99.
    • View Article
    • Google Scholar
16. 16. Ocock J, Baasanjav Grand, Baillie JEM, Erdenebat M, Kottelat M, Mendsaikhan B, et al. Mongolian Cherry-red List of Fishes. London, Britain: Zoological Gild of London; 2006a.
17. 17. Ocock J, Baasanjav G, Baillie JEM, Erdenebat M, Kottelat M, Mendsaikhan B, et al. Summary Conservation Action Plans for Mongolian Fishes. London, United kingdom of great britain and northern ireland: Zoological Society of London; 2006b.
18. 18. Goulden CE, Sitnikova T, Gelhaus J, Boldgiv B. The geology, biodiversity, and ecology of Lake Hovsgol (Mongolia). Leiden, Netherlands: Backhuys; 2006.
19. 19. NSOM. Mongolian Statistical Yearbook 2012. Ulaanbaatar, Mongolia: National Statistical Office of Mongolia (NSOM); 2012.
20. 20. MEGD. Lake Hovsgol National Park: General Management Plan Foundation Document. Ulaanbaatar, Mongolia: Ministry building of Environment and Green Development (MEGD); 2013.
21. 21. Ahrenstorff TD, Jensen OP, Weidel BC, Mendsaikhan B, Hrabik TR. Abundance, spatial distribution, and nutrition of endangered Hovsgol grayling (Thymallus nigrescens). Environ Biol Fishes. 2012; 94(2): 465–476.
    • View Article
    • Google Scholar
22. 22. Sideleva VG. Fish fauna of Lake Hovsgol and Selenga River in comparison with ichthyofauna of Lake Baikal. In: Goulden CE, Sitnikova T, Gelhaus J, Boldgiv B, editors. The geology, biodiversity, and ecology of Lake Hovsgol (Mongolia). Leiden, Netherlands: Backhuys; 2006. pp 357–378.
23. 23. Dulmaa A. Fish and fisheries in Mongolia. In: Petr T, editor. Fish and fisheries at college latitudes: Asia. FAO Fisheries Technical Paper 385. Rome, Italy: FAO; 1999.
24. 24. Costless CM, Jensen OP, Mason SA, Eriksen Chiliad, Williamson NJ, Boldgiv B. High-levels of microplastic pollution in a big, remote, mountain lake. Mar Pollut Bull. 2014; 85(1): 156–163. pmid:24973278
    • View Article
    • PubMed/NCBI
    • Google Scholar
25. 25. Atkinson R, Flintstone J. Accessing hidden and hard-to-reach populations: Snowball research strategies. Soc Res Update. 2001; 33(one): 1–4.
    • View Article
    • Google Scholar
26. 26. McClenachan L. Documenting loss of large trophy fish from the Florida Keys with historical photographs. Conserv Biol. 2009; 23(3): 636–643. pmid:19183214
    • View Commodity
    • PubMed/NCBI
    • Google Scholar
27. 27. R Core Squad. R: A language and surroundings for statistical computing. R Foundation for Statistical Computing: Vienna, Austria. 2015. Bachelor: http://www.R-project.org/.
28. 28. Bates D, Maechler 1000, Bolker B, Walker S. lme4: Linear mixed-effects models using Eigen and S4. R package version i.1–8. 2015. Available: http://CRAN.R-project.org/package=lme4.
29. 29. Hilborn R, Walters CJ. Quantitative Fisheries Stock Assessment: Choice, Dynamics and Doubtfulness. Dordrecht, Netherlands: Springer Scientific discipline+Business organisation Media; 1992.
30. 30. Kenchington TJ. Natural bloodshed estimators for information‐limited fisheries. Fish Fish. 2014; xv(4): 533–562.
    • View Article
    • Google Scholar
31. 31. Zhou S, Yin Southward, Thorson JT, Smith ADM, Fuller K. Linking fishing bloodshed reference points to life history traits: an empirical written report. Can J Fish Aquat Sci. 2012; 69: 1292–1301.
    • View Article
    • Google Scholar
32. 32. Pauly D, Morgan GR. Length-based methods in fisheries research. ICLARM Conference Proceedings 13. Manila, Philippines: International Heart for Living Aquatic Resources Direction; 1987.
33. 33. Donohue MJ, Boland RC, Sramek CM, Antonelis GA. Derelict line-fishing gear in the Northwestern Hawaiian Islands: diving surveys and debris removal in 1999 confirm threat to coral reef ecosystems. Mar Pollut Balderdash. 2001; 42(12): 1301–1312. pmid:11827117
    • View Article
    • PubMed/NCBI
    • Google Scholar
34. 34. Bauer LJ, Kendall MS, Jeffrey CF. Incidence of marine debris and its relationships with benthic features in Gray'due south Reef National Marine Sanctuary, Southeast USA. Mar Pollut Bull. 2008; 56(iii): 402–413. pmid:18096190
    • View Commodity
    • PubMed/NCBI
    • Google Scholar
35. 35. Solomon J, Jacobson SK, Wald KD, Gavin M. Estimating illegal resource utilize at a Ugandan park with the Randomized Response Technique. Hum Dimens Wildl. 2007; 12(2): 75–88.
    • View Commodity
    • Google Scholar
36. 36. Thomas As, Gavin MC, Milfont TL. Estimating non-compliance among recreational fishers: Insights into factors affecting the usefulness of the Randomized Response and Item Count Techniques. Biol Conserv. 2014; 189: 24–32.
    • View Article
    • Google Scholar
37. 37. Mann BQ. Quantification of illicit fish harvesting in the Lake St Lucia game reserve, Southward Africa. Biol Conserv. 1995; 74(2): 107–113.
    • View Article
    • Google Scholar
38. 38. Wolter F, Preisendörfer P. Asking sensitive questions an evaluation of the randomized response technique versus direct questioning using individual validation data. Sociol Methods Res. 2013; 42(three): 321–353.
    • View Commodity
    • Google Scholar
39. 39. Jupiter SD, Egli DP. Ecosystem-based direction in Fiji: successes and challenges after five years of implementation. J Mar Biol. 2010; 2011: 940765.
    • View Commodity
    • Google Scholar
40. 40. Fisher RJ. Social desirability bias and the validity of indirect questioning. J Consum Res. 1993; 20: 303–315.
    • View Article
    • Google Scholar
41. 41. Warner SL. Randomized response: a survey technique for eliminating evasive answer bias. J Am Stat Assoc. 1965; 60(309) 63–69. pmid:12261830
    • View Article
    • PubMed/NCBI
    • Google Scholar
42. 42. Miller JD. A new survey technique for studying deviant beliefs. Washington, D.C: George Washington University; 1984.
43. 43. Bare SG, Gavin MC. The randomized response technique as a tool for estimating not-compliance rates in fisheries: a case report of illegal red abalone (Haliotis rufescens) fishing in Northern California. Environ Conserv. 2009; 36(2): 112–119.
    • View Commodity
    • Google Scholar
44. 44. St John FAV, Edwards-Jones Thou, Gibbons JM, Jones JPG. Testing novel methods for assessing rule breaking in conservation. Biol Conserv. 2010; 143(4): 1025–1030.
    • View Article
    • Google Scholar
45. 45. Nuno A, St John FA. How to ask sensitive questions in conservation: A review of specialized questioning techniques. Biol Conserv. 2014; 189: five–15.
    • View Article
    • Google Scholar
46. 46. Keane A, Jones JP, Edwards-Jones Chiliad, Milner-Gulland EJ. The sleeping policeman: understanding problems of enforcement and compliance in conservation. Anim Conserv. 2008; eleven(2): 75–82.
    • View Article
    • Google Scholar
47. 47. Jensen OP, Gilroy DJ, Hogan Z, Allen BC, Hrabik TR, Weidel BC, et al. Evaluating recreational fisheries for an endangered species: a case written report of taimen, Hucho taimen, in Mongolia. Can J Fish Aquat Sci. 2009; 66(ten): 1707–1718.
    • View Article
    • Google Scholar
48. 48. Northcote TG. Comparative biological science and management of Arctic and European grayling (Salmonidae, Thymallus). Rev Fish Biol Fish. 1995; v(2): 141–194.
    • View Article
    • Google Scholar
49. 49. Roberts BC, White RG. Effects of angler wading on survival of trout eggs and pre-emergent fry. North Am J Fish Manage. 1992; 12(3): 450–459.
    • View Article
    • Google Scholar
50. 50. Dagvadorj D, Natsagdorj L, Dorjpurev J, Namkhainyam B. MARCC: Mongolia cess report on climate change 2009. Ulaanbaatar, Mongolia: Ministry of Nature, Surroundings, and Tourism; 2009.
51. 51. IPCC. Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Written report of the Intergovernmental Console on Climatic change. Eds: Stocker TF, Qin D, Plattner GK, Tignor Thou, Allen SK, Boschung J, Nauels A, Xia Y, Bex 5, Midgley PM. Cambridge, UK: Cambridge University Press; 2013.
52. 52. Young T, Jensen OP, Weidel BC, Chandra S. Natural trophic variability in a large, oligotrophic, near-pristine lake. J Cracking Lakes Res. 2015; 41(ii): 463–472.
    • View Article
    • Google Scholar
53. 53. Batima P, Natsagdorj L, Batnasan North. Vulnerability of Mongolia'due south pastoralists to climate extremes and changes. In: Leary N, Conde C, Kulkarni J, Nyong A, Pulhin J, editors. Climate Change and Vulnerability and Adaptation. London, UK: Earthscan; 2013. pp. 67–87.
54. 54. Bruun O, Odgaard O. Mongolia in transition: old patterns, new challenges. Abingdon, Britain: Nordic Institute of Asian Studies; 1996.
55. 55. Myagmarsuren D. Special Protected Areas of Mongolia. Ulaanbaatar, Mongolia: Munkhyn Useg Co, Ltd; 2008.
56. 56. Then AY, Hoenig JM, Hall NG, Hewitt DA. Evaluating the predictive functioning of empirical estimators of natural mortality charge per unit using information on over 200 fish species. ICES J Mar Sci. 2014; 72(i): 82–92.
    • View Article
    • Google Scholar
57. 57. Gunderson D. Merchandise-off between reproductive effort and adult survival in oviparous and viviparous fishes. Can J Fish Aquat Sci. 1997; 54(5): 990–998.
    • View Article
    • Google Scholar

starnercareekeres.blogspot.com

Source: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0143960