The study was carried out in northern Chilean Patagonia, in an area located around 45°20′S and 73°40′W ( Figure 1 ). The marine ecosystem in this region is influenced by a main channel called Moraleda, which separates the archipelago to the west from the continental coast to the east. Fjords, channels, and numerous islands characterize this “inner sea”. Water exchange from the open ocean, glacial inputs, light regime, and freshwater discharge varied over latitudinal and longitudinal gradients, presenting a strong seasonal cycle that determines hydrographic conditions and hence productivity patterns (Aracena et al., 2011; Montero et al., 2011). Near the southern extreme of the Moraleda Channel, there is a shallow sill (60 m deep), called the “Meninea Constriction”, which partially isolates the southern part of the Moraleda Channel and Aysén Fjord from the open ocean. The sill constrains circulation, forming two basins (North and South); the southern basin is warmer, less saline, and more oxygenated than the northern one (Silva et al., 1995, Silva et al., 2000). The northern basin in its western part is influenced by water masses from the Moraleda Channel, i.e., estuarine waters on the surface, modified subantarctic waters as an intermediate water mass, and, in deeper zones, equatorial subsurface waters. In the east, the basin is influenced by waters from the Aysén Fjord. Two water masses are present during the year at the mouth of this fjord, varying their depth according to the season. At the mouth of the fjord, salty estuarine water (21–31 psu) dominates the upper 40 m of the water column, becoming deeper in winter (approximately 100 m). Below this water mass, modified subantarctic water mass dominates at different depths depending on the season (Guzman, 2004).

The study area was selected since it is the main regional fishing ground associated with marble crab (M. edwardsii). Fishers and their families reside near the fishing grounds, with travel times ranging from 10 min to several hours of navigation. Extraction occurs year-round, except for a summer break, typically from mid-December until early March, following commercial agreements with buyers.

In this context, 10 fishing grounds were sampled, with efforts to cover all seasons through an annual cycle. Zones were grouped considering oceanographic conditions associated with the Meninea Constriction and the influence of the Aysén Fjord; in this sense, the study areas were divided into the west (n = 4), east (n = 1), and south zones (n = 5, see polygons in Figure 1 ).

In the study area, the fishery was monitored in the three zones based on seasonal surveys ( Table 1 ). Due to weather conditions, the southern zone could not be sampled during the winter season. Sampling was planned daily according to fishers’ activities since the assessment was made on board fishing boats using their traps as the sample unit. Traps are the only fishing gear legally allowed for crabs (Subpesca, 2013). Traps used by the fishers were truncated cone-shaped (Nasa type) and composed of three metal rings, one in the bottom, another in the middle, and a third upper ring to which a plastic entrance tunnel is attached; the diameter of the entrance varied between 25 and 35 cm ( Figure 2 ). Most traps had escape rings with a diameter approximately of 8 cm. Traps were individually set on the seabed at a distance greater than 50 m between them. Therefore, it is assumed that there is no overlap between the effective fishing areas (Aedo and Arancibia, 2003). For each trap, the catch was recorded by the number of individuals, categorized into retained [carapace width (CW), ≥120 mm] and released (CW < 120 mm) [following the minimum legal size (MLS) of extraction, established in 120-mm CW; D. Ext. No. 9 of 1990 establishes MLS)]. This procedure allowed the estimation of catch per unit effort (CPUE) in a number of individuals. During each fishing trip, three or four traps were randomly selected to collect enough individuals (between 25 and 281 individuals per trap) to allow biological sampling, registering crab size, weight, and sex ( Table 2 ).

Since 80% of the traps used in the extraction of marble crab have lateral escape rings, soak time could have influenced escape rates and hence CPUE. Consequently, the effect of soak time on released and retained CPUE was analyzed separately.

Seasonal and zone effects were analyzed based on five indicators: i) released and retained CPUE, ii) proportional stock density (PSD), iii) largest 10% of the sample (Lmax), iv) sex ratio, and v) length–weight relationship. PSD relates the number of individuals in a specific size class to the total number of individuals collected. The limit size adopted for the estimation of PSD was the MLS, and therefore, PSD estimates the percentage of individuals whose size exceeds this MLS. This indicator had been proposed by Froesse (2004), describing it as “let them spawn” since it is a measure of the percentage of mature specimens in the catch and hence can show trends of the status of a fishery. Data were also explored considering the average size of the largest 10% of the sample (Lmax) and selected as a measure that is less affected by environmental effects and recruitment variability (Miethe et al., 2016). Sex ratio was estimated as the proportion of the number of males in each trap to the total number of individuals (females + males) and was estimated using only individuals over MLS. Therefore, a sex ratio greater than 0.5 indicates an increased number of males compared to females. Length–weight relationship was calculated using the following formula: T W = a C W b , where TW is body wet weight of crab in g, CW is carapace width in mm, and “a” and “b” are intercept and slope of equation, respectively. The length–weight relationship was transformed into a linear model by taking the natural logarithms of both sides of the equation (Patil and Patil, 2012), making the errors additive, and stabilizing the variances in the model. To test the isometric growth hypothesis (b = 3), a t-test was used: t = b ^ − b 0 S E b ^ , where b ^{^}, S E b ^ and df (degrees of freedom, df = n − 2) are from the linear regression, and b ₀ is the specified value in the H₀ (Maity, 2018). To compare the slopes of the regressions, first, the coefficients of the slope were estimated using the nlme package (Pinheiro et al., 2018), and then the slopes were compared using least square means (lsmeans; Lenth, 2016).

Non-parametric ANOVA with permutations (using the function aovp from lmPerm package; Wheeler and Torchiano, 2016) was used to search for differences in CPUE (released and retained) among fishing zones and seasons since the data were not normally distributed (Kolmogorov–Smirnov test, p < 0.05) and presented heterogeneity of variance (Levene’s test, p < 0.05). Post-hoc tests were performed using permutations pairwise comparisons (rcompanion package; Mangiafico, 2019). ANOVA test was used, with Tukey’s test as post-hoc analyses, for Lmax, PSD, and soak time since they presented a normal distribution and homogeneity of variance. A generalized linear model (GLM) test, assuming a quasi-binomial distribution of errors, was used to assess the sex ratio discrepancy between zones and seasons (Wilson and Hardy, 2002).

All calculations and analyses were conducted in the R programming environment (R Core Team, 2023).

The number of traps varied between boats from 26 and up to 37, depending on whether they had davits or not. A boat without a davit works with fewer traps compared to a larger boat with a davit that operates with two fishers and more traps. During the field trips, it was noticed that under good weather conditions and sufficient bait availability, fishers went daily through all their traps. In these cases, the soak time was typically 1 day. However, in instances of adverse weather conditions or limited bait availability, the soak time could extend beyond this period. The maximum soak time recorded was 31 days. Throughout the study period, all traps employed the same bait, consisting of 3 to 4 kg of Chilean mussels (Mytilus chilensis).

The seasonal surveys between autumn 2018 and summer 2019 resulted in 632 independent and validated records associated with catch per unit effort (CPUE). The soak time of these traps varied between 1 day (n = 402), 2 days (n = 126), 3 days (n = 63), 4 days (n = 36), and 31 days (n = 5).

Retained catch (CPUE ≥ 120 mm) did not show significant differences between 1 and 2 days, nor between 3 and 4 days of soak time ( Figure 3A ; Tukey’s test, p < 0.001). However, significant differences were detected at soak times of 1–2 days compared to 3–4 days (Tukey’s test; p < 0.001). The catch retained with 31 days of soak time was significantly different only from that recorded with 3–4 days (Tukey’s test; p < 0.05).

Released catch revealed significant differences at soak times of 1 and 2 days and 2 and 3 days ( Figure 3B ; Tukey’s test; p < 0.001). For all other possible comparisons, no significant differences were found (Tukey’s test; p > 0.05).

The influence of soak time on retained and released catch enabled the selection of the most suitable datasets for analyzing spatial (fishing zones) and temporal (seasons) factors concerning various biological and fishery indicators. The analysis for retained CPUE and sex ratio considered catches with 1 and 2 days of soak time (n = 528). In contrast, released CPUE and most other biological indicators were analyzed only with 1 day of soak time (n = 402, Table 2 ). For the analysis of the length–weight relationship, the information corresponding to all soak times was considered (n = 632) since the relationship between both metric variables is not affected by the rates of entry and escape to the trap.

For retained catch (considering 1 and 2 days of soak time), the highest size frequencies were between 120- and 130-mm CW (51%) even though maximum sizes reached 190-mm CW.

Seasonal variability was evident across all zones ( Figure 4A ). In the eastern zone, the retained catch during autumn (average 53 ± SE 2) was higher than in winter (average 41 ± SE 4) and summer (average 43 ± SE 4) (pairwise permutation test; p < 0.05). Conversely, in the southern zone, there were no significant differences between autumn and summer (average 34 ± SE 2 and average 37 ± SE 2, respectively) (pairwise permutation test; p > 0.1). Additionally, in both seasons, the retained catch was higher than in spring (average 24 ± SE 2) (pairwise permutation test; p < 0.001). In the western zone, retained catch in autumn (average 47 ± SE 4) and winter (average 50 ± SE 2) was similar (pairwise permutation test; p > 0.05), being higher in both seasons compared to summer (average 38 ± SE 2) and spring (average 26 ± SE 1) (pairwise permutation test; p < 0.001). Retained catch in spring was the lowest of all seasons. Data for all zones were available only for summer and autumn; in summer, no differences were found between zones (aovp test; p > 0.1), but in autumn, the retained catch was lower in the south (average 34 ± SE 2) compared to the east (average 53 ± SE 2) and west zones (average 47 ± SE 4) (pairwise permutation test; p < 0.001). Retained catch in these latter zones was similar, presenting no significant differences (pairwise permutation test; p > 0.1).

For total released catch (considering 1 day of soak time), the highest size frequencies were found between 109- and 119-mm CW (72%), but individuals as small as 54-mm CW were retained in the traps. The only zone exhibiting seasonal variation in released catch was the west ( Figure 4B ), with winter showing higher values (average 77 ± SE 4) compared to the other seasons, namely, spring and summer (pairwise permutation test; p < 0.01). Data for all zones were available only for summer. When analyzing this season, no significant differences in the released catch were found between zones (pairwise permutation test; p > 0.05).

PSD (as a percentage of crabs ≥ MLS) could not be estimated for all seasons in the zones because the data considered only 1 day of soak time. However, in the east and west zones, PSD was minimal in winter, increasing toward spring and summer to a maximum in autumn ( Figure 5 ). In the eastern zone, PSD in autumn was higher (average 60 ± SE 3) than in summer (average 52 ± SE 3; pairwise permutation test, p < 0.05) and winter (average 44 ± SE 3; pairwise permutation test, p < 0.001). In the western zone, PSD could not be estimated in autumn, but spring (average 54 ± SE 2) and summer (average 59 ± SE 4) presented a higher PSD compared to winter (average 40 ± SE 2; pairwise permutation test, p < 0.001). Conversely, in the south, no significant differences were found between seasons (ANOVA test; p > 0.1), although winter could not be sampled. In summer (where all zones were sampled), no significant differences were found (ANOVA test; p > 0.1).

Lmax (top 10%) were similar between seasons in the same zone (ANOVA test; p > 0.1) and were also similar in summer when all zones could be sampled (ANOVA test; p > 0.1). Lmax varied on average between 136- and 147-mm CW, having a minimum value of 110-mm CW and a maximum of 170-mm CW. Lmax were different between sexes (ANOVA test; p < 0.001), as males were larger (average 149 ± SE 1) compared to females (average 135 ± SE 1) ( Figures 6A, B ).

The sex ratio was analyzed based on individuals exceeding the MLS using both 1 and 2 days of soak time. For most seasons and zones, males dominated this fraction of the stock ( Figure 7 ). Generally, males were predominant during winter (average 0.84 ± SE 0.05) compared to the other seasons. In contrast, in autumn in the west and south zones, females dominated the catch (average 0.34 ± SE 0.08), being statistically different from spring and summer in the south (GLM test; p < 0.05; Supplementary Material; Table S1 ) and spring and winter in the west (GLM test; p < 0.05). In summer and autumn when there were enough data to compare all zones, no significant differences were found during summer between zones (GLM test; p > 0.1), but in autumn, significant differences were found between the east zone (average 0.59 ± SE 0.03) and the south and west zones (GLM test; p < 0.05; Supplementary Material; Table S1 ).

The sample sizes for the length–weight relationship varied between 591 and 2,033 individuals for the spatial analysis (zones) and between 515 and 1,361 individuals for the temporal analysis (seasons). For each sex in the different fishing zones and seasons, after the log transformation of data, the adjustment of the length–weight relationship was significant (Student’s t-test; p < 0.05) (Supplementary data, Tables S2 and S3 ). In terms of growth type, slope b was significantly higher than 3 in males and statistically less than 3 in females, indicating positive and negative allometric growth, respectively (Student’s t-test; p < 0.05) (Supplementary data, Tables S2 and S3 ). The comparison between the slopes of the regressions determined a significant effect of the zone (ANOVA test; p < 0.01) and season (ANOVA test; p < 0.001) only in the case of males, but not for females (in both factors: ANOVA test; p > 0.05). The slope b of males collected in the eastern zone was statistically lower than the estimated for the western (Tukey’s test; p < 0.05) and southern zones (Tukey’s test; p < 0.01). In addition, the slope b of males collected in spring was statistically higher than the estimated for autumn (Tukey’s test; p < 0.001), winter (Tukey’s test; p < 0.001), and summer (Tukey’s test; p < 0.01).