Santo André Lagoon is a coastal lagoon in the northeastern Atlantic located in the southwest of mainland Portugal (38°05’47.57’’N; 8°47’23.69’’W, see Fig. 1). This lagoon is part of the Natural Reserve of Lagoa de Santo André and Sancha and it is under the protection of the Ramsar Convention on Wetlands and Natura 2000 (Birds Directive - Special Protection Area and Habitats Directive - Site of Community Importance). The lagoon has an area of 150 ha (Cancela da Fonseca et al. 2001), with an average annual depth of 1.8 m (Cancela da Fonseca et al. 1989). The high nutrient levels and shallow depth of this coastal lagoon provide ideal conditions for high rates of primary production (Duarte et al. 2002).

A sand bar separates the water body from the ocean during most of the year. The connection to the sea is artificially established, usually in March, through a man-made channel that typically remains open for a few weeks to a month (Cancela da Fonseca et al. 1989). Outside this period, the lagoon may sporadically be in contact with the sea, especially when waves surpass the dunes, usually in autumn and winter (Félix et al. 2015). Water renewal occurs when the communication between the lagoon and the sea is re-established, which is essential to maintain the functioning and productivity of the system. In addition to preventing eutrophication processes, this connection also allows the recruitment to the lagoon of several marine animals and their larval stages, as well as plant species (Costa et al. 1985, Bernardo et al. 1988, Félix et al. 2015). The lagoon can be colonized by both marine and freshwater species, but in most cases the species survive only for short periods due to the lagoon’s highly fluctuating environment (Cancela da Fonseca et al. 1989, Garside et al. 2014).

The lagoon fauna is dominated by macroinvertebrates, fish species and aquatic birds (Cancela da Fonseca et al. 2001), some of which are exploited by the fishing community and have high economic value. The European eel supports the most important fishery in the lagoon, with an estimated 4 to 8 t of this species being harvested annually (Correia et al. 2019).

Sampling (experimental fishing) was conducted monthly at five locations from January to December 2019. To cover the habitat diversity, the five sampling sites were selected on the basis of the distance between them and the substrate type. Sediments were then grouped as Mud <63 µm; 0.063 mm ≤Muddy sand/Sandy mud <0.2 mm; 0.2 mm ≤Sand <2 mm; 2 mm ≤Gravel, following the classification system adopted in the sediment charts of the Portuguese Navy Hydrographic Institute (Moita 1985). Site A is the closest to the sea and has a sandy bottom; site B is located in the middle of the lagoon, with mud and sand; site C is the upstream area, with a muddy and sandy bottom; site D in the southern area has a sandy bottom and is an even more closed water body inside the lagoon; and site E, in the northern area, has a muddy bottom and is an important fishery area (Fig. 1).

At each location, four unbaited winged fyke nets with the same dimensions as those used by fishermen (2 m bag length, 5 m wing length, 1.5 m wing height and 18 mm mesh size in cod-end) were set during the afternoon and removed the next morning. On each sampling occasion, four environmental variables (Table 1) were measured using a multiparametric probe (YSI II Water Quality Sonde).

For each fyke net, the total number of crabs caught were counted and weighed. A random subsample of 20 individuals per fyke net was retained and transported to the laboratory and frozen until processing, to collect individual data: carapace width, measured between the tips of their fifth anterior lateral spines (±1 mm accuracy); weight (precision±0.01 g); sex, determined by analysis of the abdomen (V-shaped abdomen in males and wider and U-shaped abdomen in females); and ovigerous females, depending on whether visible eggs were present.

Data on crab catches from the artisanal fishery were obtained from three volunteer fishermen (≈8% of the total number of fishermen with a fishing licence) during the 2019 fishing season, which was divided into two fishing periods. The first fishing period ran from 1 January to 17 March and the second from 16 July to 30 September. Logbooks were distributed to fishermen who volunteered to record the number and/or total weight of crabs caught per day, as well as the number of fyke nets used daily.

Spatial-temporal variability of green crab abundance was assessed using a permutational analysis of variance (PERMANOVA) (Anderson et al. 2008), in which a two-way fixed-effect crossed design was performed (factors: sampling month and sampling site). A square root transformation was applied to the variables, and the Bray-Curtis similarity coefficient was used as a resemblance measure. A posteriori pairwise tests were conducted when significant interactions between factors were identified. These procedures were performed using PRIMER® v6 (Clarke and Gorley 2006).

The influence of environmental factors on the abundance of green crab (response variable) was assessed with a generalized linear model (GLM, McCullagh and Nelder 1989) performed in R environment (R Core Team 2021) using the “MASS” software package (Venables and Ripley 2002). In addition to the environmental variables measured on each sampling occasion (Table 1), the remaining explanatory variables included in the analysis were sampling month (January-December), sediment type and total organic matter content (%) (unpublished data). Before modelling, a Spearman correlation test was performed between the predictors, and a high correlation coefficient (ρ>0.7, Dormann et al. 2013) was identified between total dissolved solids and salinity. Thus, to avoid redundancy, total dissolved solids was removed from subsequent analysis. The Akaike information criterion (Sakamoto et al. 1986) was chosen for the selection of the best model. Subsequently, an ANOVA was performed with the chi-squared (χ²) test to verify the significance of the variables in the model.

Laboratory data were used for the biological characterization of the green crab population. Regarding carapace width, after verifying the normality and homoscedasticity of the dataset, a two-way ANOVA was conducted to examine the effect of sampling month and sampling site on carapace width. A Tukey HSD post hoc test was performed for multiple comparisons between effects. The statistical tests were conducted in IBM® SPSS® Statistics 26 (IBM Corp. Released 2019).

The gender data of the individuals were analysed by sampling month and by sampling site using a χ² test. Data on ovigerous and non-ovigerous females were also analysed by sampling month and by sampling site using a χ² test. These statistical tests were conducted using IBM® SPSS® Statistics 26 (IBM Corp. Released 2019). To estimate the size at which 50% of the green crab females were sexually mature in the lagoon, a logistic regression model was used, and the maturity ogive was calculated using the “sizeMat” software package (Torrejon-Magallanes 2020) in R environment (R Core Team 2021).

Data from fishing logbooks and experimental fishing were used to estimate the total weight of green crabs harvested annually (kg/ha/year). The catch per unit effort (CPUE), was determined as the average ± standard deviation of the estimated CPUE for each fisherman and the average obtained from the experimental fishery, expressed as kg of green crabs per fyke net per fishing day. The total annual green crab catch in the Santo André Lagoon was estimated by multiplying the average CPUE obtained by the total number of fishing days, by the total number of fishermen with a fishing licence, and by the number of fyke nets they used in each fishing period.

A total of 15063 green crabs were collected over the study period: 6740 at site A, 1995 at site B, 904 at site C, 3123 at site D and 2301 at site E (Fig. 2). The highest number (1944 crabs) were collected at site A in January and the lowest number (14 crabs) at site C in February. January and December were the months with the highest catches (2819 and 2116, respectively) and October was the month with the lowest (326 crabs) (Fig. 2). The highest catches were at site A from January to May plus December and at site D from July to September.

Regarding the spatial-temporal variability, PERMANOVA tests showed that the abundance of green crab was significantly different for both factors, sampling month (F=6.0738, df=11, ρ<0.05) and sampling site (F=15.4270, df=4, ρ<0.05), and for the interaction between factors (F=2.3634, df=44, ρ<0.05). Pairwise tests revealed significant differences between October and all months, except for February and March (Supplementary material Table S1). Pairwise tests for each pair of sampling sites revealed significant differences between all sampling sites, except for site E, which did not differ from sites B and D (Supplementary material Table S2).

The GLM developed to understand the relationship between abundance and environmental factors showed a predictive performance of 37.18%. The variables that best explained the species abundance inside the lagoon were sediment type (27.26%), temperature (6.15%), salinity (3.06%) and depth (0.71%). The abundance increased in deeper sandy areas and decreased in the remaining sediment types and with higher water temperature and salinity. The summary of the model coefficients is presented in Table 2.

Of the 3898 specimens analysed in the laboratory over the study period, 2288 (58.70%) were males and 1610 (41.30%) were females. The carapace width of the largest green crab measured was 73.60 mm and that of the smallest 21.24 mm. The carapace width of males ranged from 21.24 to 71.46 mm and that of females from 26.68 to 73.60 mm. The average carapace width of the whole green crab population was 45.50 mm (Figs 3 and 4), with 40.79 mm for females and 48.81 mm for males (Fig. 5). The largest green crabs were predominant in the summer months (Fig. 3). The results of the ANOVA showed that there were statistically significant differences in carapace width between sampling months (F=90.07, df=11, ρ<0.05) and between sampling sites (F=18.00, df=4, ρ<0.05) and a statistically significant interaction between the effects of both factors (F=4.66, df=44, ρ<0.05). The Tukey HSD post hoc test revealed that the smaller carapace width of green crabs caught in January was significantly different (ρ<0.05) when compared with the measures obtained for the remaining months (Fig. 3). Regarding sampling sites, the Tukey HSD test showed that the carapace width was significantly different between site C and the other sampling sites (ρ<0.05) and between site D and the other sampling sites (ρ<0.05). There was no statistically significant difference in carapace width between sites A and B (ρ=0.10), A and E (ρ=0.54) or B and E (ρ=0.87).

The population sex ratio varied significantly, in relation to both the sampling month (χ²=752.06, df=11, ρ<0.05), and the sampling sites (χ²=37.36, df=4, ρ<0.05). The highest female-biased sex ratios occurred in January and December, with respectively 79.13% and 80.05%, followed by February, with 63.16%. From March to November, catches were dominated by males, with August standing out with the highest proportion of males (92.79%), while March was the month closest to a balanced distribution between sexes, with 45.51% females and 54.48% males (Fig. 6). Overall, the sex ratio was biased towards males at all sampling sites except site A, where the sex ratio was closer to a 1:1 balance (48.75% females; 51.25% males) (Fig. 6).

The ratio between ovigerous and non-ovigerous females varied significantly with the sampling month (χ²=143.13, df=11, ρ<0.05) but not with the sampling site (χ²=7.19, df=4, ρ=0.13). Non-ovigerous females (n=1430; 88.82%) predominated over ovigerous females (n=180; 11.18%) in all months and sites sampled. The highest catches of ovigerous females occurred in January and December (23.29% and 22.74%, respectively), whereas in June, August and September no ovigerous females were caught. All sampling sites followed the same pattern, with an average of 11.59% ovigerous females per site.

Over the study period, the size of ovigerous females varied between 26.68 and 56.22 mm (mean, 41.07 mm) and that of non-ovigerous females between 27.05 and 73.60 mm (mean, 40.75 mm). In Santo André Lagoon, 50% of females attained sexual maturity when they reached an average carapace width of 45.11 mm (Fig. 7).

In 2019, for the first fishing period, each fisherman caught on average 2.92±3.23 kg/fyke net/fishing day, ranging from 0.14 kg/fyke net/fishing day in February to 12.86 kg/fyke net/fishing day in January. In the second fishing period, each fisherman caught on average 1.60±0.39 kg/fyke net/fishing day, ranging from 0.16 kg/fyke net/fishing day in September to 2.75 kg/fyke net/fishing day in August (Fig. 8).

The average obtained with experimental fishing throughout the entire year, using the same gear as the fishermen, was 1.24±1.44 kg/fyke net/fishing day, ranging from 0.04 kg/fyke net/fishing day in February to 6.97 kg/fyke net/fishing day in December (Fig. 8).

The average daily catch from the two commercial fishing periods and the average daily catch from the experimental fishing were used to estimate the total amount of green crab harvested in the lagoon in 2019. Considering the whole fishing period allowed (155 days), the entire fishing community (39 licensed fishermen) and the area of the lagoon (150 ha), the total green crab catches in 2019 was estimated to range between 281 and 503 t. As there is no catch limit imposed by the legislation, these values suggest that fishermen can harvest 1.87 to 3.35 t/ha (1873 to 3354 kg/ha) of green crab annually.