Introduction
With an Exclusive Economic Zone (EEZ) of 1.37 million km2 constituting 99.7% ocean, Seychelles' EEZ is among the top 25 largest in the world and a global biodiversity hotspot with two United Nations Educational Scientific and Cultural Organization (UNESCO) World Heritage Sites of which one is a marine site (Myers et al., Reference Myers, Mittermeier, Mittermeier, Fonseca and Kent2000). Biodiversity is one of the country's most important assets that supports several major economic sectors, including its two pillars, fisheries and tourism (Bistoquet et al., Reference Bistoquet, Marguerite, Lucas, Morel, Elizabeth, Michaud and Tsuji2018). Seychelles has committed to protecting 30% of its EEZ (400,000 km2) of which half equates to 15% no-take zones. A comprehensive marine spatial plan (MSP) aiming at supporting the sustainable and long-term use and health of the Seychelles ocean waters has been developed to support this process (GoS, 2017). Using an ecosystem-based approach, the Seychelles MSP aims to be instrumental for improving ocean fisheries management, ensuring species and habitats have long-term protection, improving coastal ecosystem resilience to climate change, and fostering economic opportunities for fisheries and other ocean-related uses. By 2021, the Seychelles MSP will be the first in the western Indian Ocean, and the second largest in the world (Smith et al., Reference Smith, Tingey and Sims2018).
A fundamental part of the Seychelles MSP initiative relied on the participation of all stakeholders to gather relevant input on all ocean-related sectors, providing a large range of economic and scientific spatially resolved data as well as local knowledge. Hence, the Seychelles MSP aims to highlight knowledge gaps and provide guidance in collecting the relevant data. The sustainable exploitation of the Seychelles artisanal fishery resources, for instance, requires knowledge of the population dynamics of the various target resources. About 400 artisanal boats operate in Seychelles waters (SFA, 2018), with the majority favouring catch diversification, i.e. balancing fishing effort across a wide range of species. While such a strategy has been shown to ensure local nutritional security and protect fishing livelihoods in data-poor tropical fisheries (Robinson et al., Reference Robinson, Robinson, Gerry, Govinden, Freshwater and Graham2020), it challenges the work of the fishery scientists and managers through the need for basic biological data, and specifically length–weight relationships, for all targeted resources.
Length–weight relationships are essential information for fisheries research and management (Froese, Reference Froese2006). They are essential for stock assessment model inputs and commonly used in ecosystem models, e.g. to calculate the production over biomass ratio of different functional groups (Ricker, Reference Ricker1975; Pauly et al., Reference Pauly, Christensen and Walters2000). In particular, these relationships are used for converting fish numbers to biomass, monitoring changes in average weight, as well as for deriving the species composition of the catch in multi-species fisheries (Froese, Reference Froese2006). To reduce uncertainty when evaluating a fish stock, it is important to first reduce possible causes of variability of the parameters from length–weight analyses. Moreover, length–weight relationships provide valuable insights into fish wellbeing, growth and allometry (variation in form related to variation in size), reproductive characteristics, trophic ecology and general biology, and are used to convert growth-in-length equations to growth-in-weight. Finally, they allow for life history and morphological comparisons between different fish species, or between fish populations from different habitats and/or regions; such studies are relevant principally in regions where fisheries represent one of the most important economic activities and where fish stocks are the main food source for many traditional communities such as Seychelles (Freitas et al., Reference Freitas, Prudente, Fontoura and Montag2014).
Despite their ecological importance, the basic biology of Seychelles artisanal fish species is still poorly known. The present study provides information on the morphometrics of 39 fish species targeted by artisanal fishers in the Seychelles EEZ. The estimated length–weight relationship parameters were compared with values for the same species from different regions and oceans available in FishBase (http://www.FishBase.org). Finally, spatial differences in morphometric relationships between the Mahé Plateau, where most of the fishing activities are taking place, and three southern atoll groups (Aldabra, Farquhar and Amirantes; Figure 1) were examined for five fish species.
Map of the Seychelles islands, Western Indian Ocean.
Materials and methods
A total of 39 fish species from 10 families (Table 1) were collected between 2009 and 2020 from the Seychelles waters. Fishes were caught by the artisanal fishery using diverse small boats and gears (handlines, traps), and during scientific cruises using handlines and droplines onboard the research vessel ‘L'Amitié’ of the Seychelles Fishing Authority (SFA). Fishes were processed as soon as possible after being caught either onboard or at landing sites on Mahé Island, Farquhar or Aldabra by staff of the SFA, the Island Conservation Society and the Seychelles Island Foundation, respectively. All fishes were measured for whole weight (WT, nearest 0.1 kg) and length (total length TL and/or fork length FL, nearest 0.1 cm). When possible, the fish gutted weight (WG, nearest 0.1 kg) and gilled-gutted weight (WGG, nearest 0.1 kg) were also recorded.
Classification of the studied fish species collected from the Seychelles waters with English (UK), French (FR) and Seychelles creole (SEY) names and FAO-ASFIS standard 3-letter codes
The parameters of the length–weight, length–length and weight–weight relationships for the studied species (sex combined) were estimated using a maximum likelihood approach with bias correction after logarithmic transformation of the following equations (Hayes et al., Reference Hayes, Brodziak and O'Gorman1995):
$$\hbox{WT} = a \times L^b$$
$$\hbox{TL} = c \times \hbox{F}\hbox{L}^d$$
$$\hbox{WT} = e \times W^f$$with WT, the whole fish weight in kg; L, the fish length in cm (Total length TL or Fork length FL); and W, the gutted (GW) or gilled-gutted weight (GGW) in kg.
The model residuals were assumed to be independent and identically distributed normal random variables with mean zero and constant variance. Assumptions of homoscedasticity and Gaussian distribution were checked through the residuals. Model fitting was performed using the lm function implemented in R version 3.6.3 (R Core Team 2020). Species for which morphometrics were collected were ones that recorded 15 or more individuals, and covered a relatively wide size range. These were selected for the estimation of the length–weight, length–length and weight–weight relationships (Jenkins & Quintana-Ascencio, Reference Jenkins and Quintana-Ascencio2020). Moreover, extreme outliers attributed to data collection error were omitted from the analyses (i.e. 8 individuals representing <0.15% of total fish number).
Second, the effect of area on length–weight relationships was tested for species occurring in spatially distant areas of the Seychelles EEZ with a sufficient number of samples. A stepwise linear regression procedure was used to test for the influence of area variable in the linear model with the function stepAIC implemented in R version 3.6.3 (R Core Team 2020). The Akaike Information Criterion (AIC) was used to evaluate the improvement of the model when adding or dropping a term. The predictor variables for log(whole fish weight) included log(fish length), area, species as main effects. In addition, an interaction between area and species was included in the model. The term log(fish length) was fitted as a continuous variable, and the terms area and species were fitted as factors.
Results and discussion
A total of 5478 fishes were collected between 2009 and 2020 from the Seychelles waters. From the 39 fish species investigated, the most represented family was Carangidae with 10 species, followed by Lutjanidae, Serranidae and Lethrinidae (seven species each). The parameters obtained from the length–weight relationships for each species are shown in Table 2 and Figure 2. Linear regressions on log-transformed data were highly significant (P < 0.001) for all species (Tables 2–4). No significant heteroscedasticity was apparent from residual plots. The coefficients of determination (r 2) ranged between 0.817 for Euthynnus affinis and 0.992 for Lutjanus sebae and Cephalopholis argus. The exponent b of the length–weight relationships ranged between 2.2356 for Sphyraena jello and 3.4989 for Euthynnus affinis and the intercept value ranged between 2.0×10−6 for Caranx sexfasciatus and Euthynnus affinis, and 1.8×10−4 for Carangoides malabaricus.
Length–weight relationships (grey lines) estimated for 36 fish species from the Seychelles waters and compared with length–weight Bayesian relationships obtained from FishBase (black lines). The species code and related names are provided in Table 1; parameters of the equations for each species are provided in Table 2. Measured weight data correspond to total weight (WT, kg); Measured length data correspond to total length (TL, cm) for Balistidae and grouper species (i.e. CNT and CFF, CFI, EEA, EFH, EFT, EWU, respectively), and to fork length (FL, cm) for the other 29 species.
Length–weight relationships for 39 fish species from the Seychelles waters. The relationships between total length (TL, cm) and total weight (WT, kg), and between fork length (FL, cm) and total weight (WT, kg) were estimated for 26 and 33 species, respectively
N is sample size; range corresponds to the minimum and maximum length (TL or FL, cm) recorded; a and b are the parameters of the equations WT = a.TLb and WT = a.FLb; SEb is the standard error of b; r2 is the coefficient of determination; G is the type of growth (I:isometry, A−:positive allometry, A + : negative allometry).
Total length vs fork length relationships for 20 fish species from the Seychelles waters
N is sample size; range corresponds to the minimum and maximum total length (TL, cm) and fork length (FL, cm) recorded; c and d are the parameters of the equation TL = c.FLd; SEd is the standard error of d; r2 is the coefficient of determination; TL/FL is the conversion factor (ratio between TL and FL).
Weight–weight relationships for 18 fish species from the Seychelles waters
The relationships between total weight (WT, kg) and gutted weight (GW, kg), and between total weight (WT, kg) and gilled-gutted weight (GGW, kg) were estimated for 11 and 14 species, respectively. N is sample size; range corresponds to the minimum and maximum total weight (WT, kg) recorded; e and f are the parameters of the equations WT = e.GWf and WT = e.GGWf; SEf is the standard error of f; r2 is the coefficient of determination; WT/GW and WT/GGW are the conversion factors (ratios between WT and GW, and between WT and GGW, respectively).
A total of 16 species (41% of the total number of studied species) showed isometric growth (Table 2), implying that there is no change of body shape as the fish grows and that weight increases as the third power of length (i.e. b = 3). Moreover, 14 and 9 species (total 23 species, 59%) showed a negative allometric growth (A−; the fish becomes slenderer as it becomes longer with b < 3) or a positive allometric growth (A+; the fish becomes relatively stouter or deeper-bodied as it increases in length with b > 3), respectively.
Moreover, the relationships between fork length vs total length, total weight vs gutted weight, and total weight vs gilled-gutted weight, and the related conversion factors are provided for 20, 11 and 14 species, respectively (Tables 3 and 4, Figures 3 and 4).
Weight–weight relationships estimated for 18 fish species from the Seychelles waters. The species code and related names are provided in Table 1; parameters of the equations for each species are provided in Table 4. Dark grey lines represent total weight (WT, kg) vs the gutted weight (GW, kg); Light grey lines represent total weight (WT, kg) vs gilled-gutted weight (GGW, kg).
Of the 39 species, we reported updated maximum lengths and weights and subsequently more robust and comprehensive length–weight relationships for two species, namely the Carangidae Uraspis secunda and the Lethrinidae Lethrinus variegatus, that were not considered accurate in the FishBase database (Froese & Pauly, Reference Froese and Pauly2020). Moreover, we reported species-specific length–weight relationships for two species (the Balistidae Canthidermis maculata and the Lethrinidae Lethrinus crocineus), that were estimated at the sub-family and genus levels in Fishbase, respectively. Of the remaining species, 33 species showed comparable length–weight relationships between this study and FishBase (Figure 2), and three species, namely the Carangidae Gnathanodon speciosus, the Lutjanidae Lutjanus gibbus and the Serranidae Variola louti, were higher than the upper 95% confidence interval bounds of the FishBase length–weight Bayesian relationships.
Five species met the criteria for wide spatial distribution and high numbers of individuals collected from the different areas. For all species, the best model included the factors species and area only, while the interaction area:species had no effect on the AIC and was thus removed. Significant differences in the length–weight relationship among areas were observed for two species only (Epinephelus multinotatus and Lethrinus nebulosus), with individuals from the Mahé Plateau being bigger than those from the southern atoll groups for a given size (P < 0.001; Figure 5). Spatial differences in intraspecific morphometrics are possible due to the effects of spatial differences in food availability, and/or life history characteristics. However, the absolute differences were small and adding the factor area to the model resulted in a low reduction of the AIC and associated residual sum of squares. We thus conclude that the regression models, based on the pooled data, would be adequate for estimating body weight of the species concerned across the Seychelles EEZ.
Length–weight relationships estimated for five fish species from four different areas in the Seychelles EEZ: Mahé Plateau, Farquhar group, Amirantes group and Aldabra group (see Figure 1). The species code and related names are provided in Table 1. Measured weight data correspond to total weight (WT, kg); Measured length data correspond to Total length (TL, cm) for EWU, and to Fork length (FL, cm) for the other four species.
Conclusion
This study presents information on morphometric relationships for 39 ecologically and economically important fish species from the Seychelles waters. Such information is essential for determining accurate fisheries data such as biomass estimates, and thus contributes to the improvement of fish stock assessments and fisheries research management in the Seychelles and neighbouring countries.
Authors' contribution
Conceptualization: NB, Funding acquisition: NB, RG, Data acquisition: NB, AS, MM, TC, CS, AJB, Statistical analysis: NB, EC, Writing – 1st draft: NB, Review & editing: all co-authors.
Acknowledgements
The authors would like to thank all staff from Seychelles Fishing Authority (SFA; Fred Mondon, Robert Dookley, Yashim Marday, Christian Decommardmond, Gerard Ernesta, Achille Pascal and Rahim Woodcock), Island Conservation Society (ICS; Licia Calabrese, Annabelle Cupidon and Jean-Claude Camille) and Seychelles Islands Foundation (SIF) for their assistance in the field; the Islands Development Company, Desroches and Farquhar Foundations for logistical and financial support; and the SFA Research Management Team, and SIF and ICS Head Office Management Team for their support in coordinating and facilitating the administration procedures associated with this project. We are also grateful to the two anonymous reviewers for their valuable comments that helped improve this article.
Financial support
This study is part of the SEYFISH project (‘Nutrients and contaminants in Seychelles fisheries resources’) and is a contribution to the Mahé Plateau Trap and Line Fishery Co-Management Plan led by the Seychelles Fishing Authority (SFA), with the financial support of the Seychelles Government and the European Fisheries Partnership Agreement (EU-FPA).
