Association of Environmental Factors in the Taiwan Strait with Distributions and Habitat Characteristics of Three Swimming Crabs

Information regarding the oceanic environment is crucial for determining species distributions and their habitat preferences. However, in studies on crustaceans, especially swimming crabs, such information remains poorly utilized, and its effects on crab communities in the Taiwan Strait (TS) has not been well documented. The purpose of this study was to understand the relationship between the catch rates of three swimming crab species and environmental factors in the TS. We fitted generalized additive models (GAMs) to logbooks and voyage data recorder data from Taiwanese crab vessels (2011–2015), developed a species distribution model, and predicted catch rates for these three swimming crab species based on the GAM output. The chlorophyll-a (Chl-a) concentration was related to the high catch rates of Chrybdis feriatus and Portunus sanguinolentus, whereas bottom temperature (BT) was related to high catch rates of Portunus pelagicus. The variance percentages for each crab species indicated that high catch rates of C. feriatus and P. sanguinolentus occurred in a Chl-a concentration > 0.5 mg/m3, whereas P. pelagicus catch rates exhibited negative correlations with BTs > 25 ◦C. The model predicted high catch rates of C. feriatus in the north of the TS during autumn and winter, whereas P. pelagicus was observed to the south during summer and autumn. P. sanguinolentus was predicted to be widely distributed around the TS and distributed further to the northern area during autumn and winter. These findings revealed that each species responds to spatiotemporal environmental variations. Understanding the distributions and habitats of these three crabs is vital in fisheries resource management and conservation planning.


Introduction
Understanding the spatial and temporal variability of key environmental variables within commercial fishing grounds is crucial in resource management to help identify species distributions and their habitat preferences. This information can be used to implement sustainable fisheries through methods, such as harvest strategy planning [1], ecosystem management [2,3], spatial management [4][5][6], and bycatch reduction [7]. Environmental variables, such as water temperature, salinity, water depth, and increased productivity, may be significant factors driving changes in population patterns [8,9], catch rates [10][11][12], reproductive and recruitment strategies [13][14][15], and spatiotemporal distribution [16,17].
The Taiwan Strait (TS) is located in the tropical to subtropical western Pacific region and is highly influenced by three currents: Kuroshio Branch Current, China Coastal Current, and South China Sea Remote Sens. 2020, 12, 2231 3 of 17

Swimming Crab Fishery Data
Data on P. sanguinolentus, P. pelagicus, and C. feriatus from 2011 to 2016 were collected from the voyage data recorder and logbooks of 170 Taiwanese crab vessels (5-200 metric tons) in the TS. The fishing data consisted of daily fishing positions for 0.1 • spatial grids including latitude and longitude, fishing date, soaking time, total catch (kg), and crab species. Swimming crabs were collected using circular crab traps with frozen mackerel as bait in the center of the pot [39].

Marine Environmental Data
Daily SST, sea surface height (SSH), BT, and sea surface salinity (SSS) marine environmental data were extracted from the Hybrid Coordinate Ocean Model and a Navy Coupled Ocean Data Assimilation. The SST model (ETOPO5) had a spatial resolution of 9 km (apdrc.soest.hawaii.edu). Bathymetry data were downloaded from the Asia Pacific Data Research Center (apdrc.soest.hawaii.edu). The SSH can be used to infer oceanic features, such as current dynamics, fronts, eddies, and convergences [45]. Monthly Chl-a data were downloaded from the NASA Aqua satellite (oceancolor.gsfc.nasa.gov), which features a sensor to detect the concentration of Chl-a in the world's oceans, among other applications. The level 3 data map image Chl-a data have a monthly temporal resolution and 4.6-km (at the equator) spatial resolution. The environmental data were then calculated monthly on a spatial grid of 0.1 • to fit the fishery data using Interface Descriptive Language version (IDL) 7.0. The environmental data were tested for intercorrelations before running the statistical models to determine the correlation coefficients among the variables. The test for intercorrelations among all environmental variable explained low correlation coefficients (r 2 < 0.25).

Spatial and Temporal Statistical Models of Swimming Crab Catch Rates
Monthly observed catch rates were calculated using the cumulative weight of each crab species caught from all fishing vessels for a month divided by the cumulative soaking time for all set lines in that month to adjust for fishing effort. Monthly observed catch rates were calculated for individual 0.1 • grids across the study region by using the following expression: Observed catch rate ij = ΣCatch for all vessels (kg) ij ΣSoaking time for all vessels (hour) ij , where i represents the latitude, and j represents the longitude of each 0.1 • spatial grid. We divided the observed catch rates of each species into five groups on the basis of cumulative frequency. We created a spatial distribution of catch percentages map of P. sanguinolentus, P. pelagicus, and C. feriatus caught in the TS from 2011 to 2016 between 20.5 • N and 29 • N and between 118 • E and 125.5 • E using Quantum GIS version 3.6 [46]. We constructed a seasonal mean catch rate map of the three species using IDL from 2011 to 2016 in the TS. Maps were generated for each of the four seasons: spring (March-May), summer (June-August), autumn (September-November), and winter (December-February).
To more thoroughly understand the association of observed catch rates for each crab species with the environmental factors, the longitudinal and latitudinal gravitational centers of the observed catch rates (G) were estimated using monthly longitudinal and latitudinal locations of crab vessels (L) and monthly observed catch rates [47,48]: Remote Sens. 2020, 12, 2231 4 of 17 We plotted the monthly mean trends of G presented in relation to the monthly mean values of environmental data of C. feriatus, P. pelagicus, and P. sanguinolentus from 2011 to 2016 in the TS.
We examined the effects of environmental variability on observed catch rates using GAMs [49]. The GAMs were created using R version 3.6 [50] through the "mgcv" package GAM function. The observed catch rate was the response variable, and environmental factors (SST, BT, SSS, SSH, Chl-a, and depth) were the predictor variables. The GAM is summarized as follows: In Equation (3), s(xn) is a spline smoothing function for each model covariate xn. The model with the optimal conformation was selected using a stepwise procedure that was based on the lowest value of the Akaike's information criterion (AIC), and the p-value of the final set of variables was lower than 0.05. We used the coefficient of determination (r 2 ) to evaluate the precision of the model prediction [30].

Predictions of Swimming Crab Catch Rates
The output from the selected GAM environmental factors was used to predict the catch rates of three commercial swimming crabs in the TS from 2011 to 2015. We mapped the seasonal mean of the predicted observed catch rates overlaid with the most explained variance percentages of the environmental variable by GAM of three swimming crabs in the TS. The selected GAMs were based on the data from 2011 to 2015, and the environmental data of 2016 were used to predict the catch rates of three crabs. Furthermore, we tested the prediction accuracy of high catch rates (>60% of cumulative frequency) by comparing the observed and predicted catch rates of fishery data in 2016 for the year and seasonal changes by using cumulative frequency. The root mean square error (RMSE) were used to measure how far the prediction was from the observed output, which is formulated as follows: [ log(observed high catch rates ,i ) − log(predicted high catch rates ,i ] 2 n (4)

Spatial and Temporal Distribution of Three Crab Species in the TS
The spatial distribution revealed that P. sanguinolentus was widely distributed from south to north in the TS (Figure 1), whereas C. feriatus and P. pelagicus were mostly distributed in southern and northern TS, respectively. The seasonal spatial distributions of observed catch rates illustrated that each species was distributed in different areas during different seasons ( Figure 2). During winter, C. feriatus was mainly distributed in the northern areas of TS, moved southwest during spring, and could be found in northern areas during the summer and autumn (Figure 2a-d). The distribution of P. pelagicus was more concentrated near the southern coast of Taiwan in the TS during winter and spring, and it could also be found near southern and northern Taiwan during summer and autumn (Figure 2e-h). The spatial distribution of P. sanguinolentus was widespread in the TS in all seasons, and its distribution was further northeast in spring, summer, and autumn (Figure 2i The time series of latitudinal and longitudinal G of the catch rates (Figure 3a-b) and monthly mean environmental parameters in catch positions (Figure 3c-h) for each species exhibited different trends in the TS. The monthly latitudinal and longitudinal G for C. feriatus exhibited stable seasonal variations, whereas P. pelagicus moved from the southern (>23.25°N) and western (<120.25°E) areas with an increase in the SST, BT, SSH, Chl-a, and depth and decrease in the SSS during summer and autumn. For P. sanguinolentus, when SST, SSH, and depth increased and BT, SSS, and Chl-a decreased during spring and summer, the latitudinal and longitudinal G moved to northern and eastern areas of the TS. Furthermore, in 2012-2013, the latitudinal and longitudinal G of P. sanguinolentus moved further into northern (>25.75°N) and eastern (>122.75°E) areas during spring and summer.

Environmental Effect on Swimming Crab Catch Rates
The total variances derived using the GAMs for C. feriatus, P. pelagicus, and P. sanguinolentus were 38.1%, 24.0%, and 21.7% in 2011-2015, respectively (Table 1). Statistically, all the variables examined for the three crab species in the TS were highly substantial predictors (p < 0.01). The addition of environmental variables led to the increased variance, which was attributed to decreased AIC. Chl-a explained the most significant amount of variance for C. feriatus, followed by the depth and SST, and SSS explained the lowest amount (Table 1). Similar to C. feriatus, Chl-a, depth, and SST also explained the largest amount of variance for P. sanguinolentus, but SSH explained the least for this species (Table 1). By contrast, for P. pelagicus, BT explained the largest amount of variance, followed by the depth and SSS, and SST explained the smallest amount of variance for this species (Table 1). The seasonal mean observed catch rates overlaid on selected environmental factors based on the largest model-explained variance percentages of each crab species illustrate that both C. feriatus and P. sanguinolentus mostly aggregated in waters with a Chl-a value of > 0.5 mg/m 3 , whereas, P. pelagicus catch rates had negative correlations with BTs > 25 °C (Figure 4). The GAM results suggested that the catch rates of C. feriatus (Figure 5a-f) had positive correlations with SSTs > 20 °C and Chl-a > 0.5 mg/m 3 but had negative correlations with BTs > 15 °C, SSHs > 0.6 m, SSSs > 33 PSU, and depths > 100 m. For

Environmental Effect on Swimming Crab Catch Rates
The total variances derived using the GAMs for C. feriatus, P. pelagicus, and P. sanguinolentus were 38.1%, 24.0%, and 21.7% in 2011-2015, respectively (Table 1). Statistically, all the variables examined for the three crab species in the TS were highly substantial predictors (p < 0.01). The addition of environmental variables led to the increased variance, which was attributed to decreased AIC. Chl-a explained the most significant amount of variance for C. feriatus, followed by the depth and SST, and SSS explained the lowest amount (Table 1). Similar to C. feriatus, Chl-a, depth, and SST also explained the largest amount of variance for P. sanguinolentus, but SSH explained the least for this species (Table 1). By contrast, for P. pelagicus, BT explained the largest amount of variance, followed by the depth and SSS, and SST explained the smallest amount of variance for this species (Table 1). The seasonal mean observed catch rates overlaid on selected environmental factors based on the largest model-explained variance percentages of each crab species illustrate that both C. feriatus and P. sanguinolentus mostly aggregated in waters with a Chl-a value of > 0.5 mg/m 3 , whereas, P. pelagicus catch rates had negative correlations with BTs    Figure 6 presents images of a seasonal mean observed catch rate map overlaid on predicted catch rates for C. feriatus, P. pelagicus, and P. sanguinolentus in the TS. High catch rates of C. feriatus were predicted in the north TS between 25°N and 26°N, and these catch rates decreased during spring and coastal water of mainland China before being extended further to the northern area during autumn and winter (Figure 6k,l). Finally, the catch rates of P. sanguinolentus decreased during spring and summer in the TS.   Figure 6 presents images of a seasonal mean observed catch rate map overlaid on predicted catch rates for C. feriatus, P. pelagicus, and P. sanguinolentus in the TS. High catch rates of C. feriatus were predicted in the north TS between 25 • N and 26 • N, and these catch rates decreased during spring and summer (Figure 6a,b). A high catch rate of C. feriatus was predicted in the southwest of the region, including in coastal water of mainland China, during autumn and winter (Figure 6c,d). The predicted catch rates of C. feriatus extended to 25 • N-26 • N and 122 • E-123 • E in the TS during summer and autumn. For P. pelagicus, a high catch rate was predicted further to the south of the TS between 22 • N and 24 • N, and some fishing locations extended to the northeast at 25 • N-26 • N (Figure 6e-h). The catch distribution of P. pelagicus was mostly from the northern part of the SCS to the TS and then decreased during spring (Figure 6e). The seasonal predicted catch rates of P. sanguinolentus were widely distributed around the TS, and fishing positions also extended southwest to a region that includes coastal water of mainland China before being extended further to the northern area during autumn and winter (Figure 6k,l). Finally, the catch rates of P. sanguinolentus decreased during spring and summer in the TS.

Portunid Crab Distribution in the TS
The results of this study demonstrated that the observed catch rates of P. sanguinolentus were widely spread from the south to the north of the TS, whereas the observed catch rates of P. pelagicus were separated into two main areas in the south and north of the TS. However, the observed catch rates of C. feriatus were mostly distributed in the northern TS. Shaohua [51] reported that Portunidae is a dominant crab species from the central and northern TS and generally higher in number in offshore waters and lower in eastern waters. Ye et al. [52] and Ye [53] have revealed that C. feriatus is the dominant crab species in the north of TS, and P. sanguinolentus, and P. pelagicus are dominant species around the Taiwan Bank. The results in the present study also postulated that the three important commercial swimming crab species react to different types of habitats caused by environmental factors. Figure 7 indicates that C. feriatus was mainly distributed near northern Taiwan in winter and extended to the southwest during spring and could be found to the north during summer and autumn. Furthermore, a high predicted catch rate of C. feriatus was observed southwestward of TS during autumn and winter, including coastal mainland China (Figure 6c,d). Huang [54] discovered that the catch rate of C. feriatus in the southern part of the ECS was the highest during winter and lowest during summer, and the primary breeding seasons were spring and summer. Huang [54] also suggested that winter is the optimal season for C. feriatus fishing based on the distribution of their numbers, their biological characteristics, and the fishing production status. Some studies have shown that swimming crabs that inhabit marine embayment often do not leave these marine environments to spawn [55,56]. Furthermore, compared with other swimming crab species, C. feriatus can breed continually, and the broodstock is available year-round [57]. Thus, we suggested that the distribution of C. feriatus mainly in northern TS could be due to a particular factor and the preferred seasons for fishing this species being autumn and winter.
P. pelagicus was more concentrated on the southern coast of the TS and highly catch rates in the northern TS during summer and autumn (Figure 7). High P. pelagicus concentrations were observed at the south of TS during summer and autumn, and the concentration decreased during winter and spring. The highest seasonal catch rates of P. pelagicus were also recorded during summer in the Arabian Sea [38,58] and between December and May along the coast of Tanzania [59]. Similar to C. feriatus, P. pelagicus also appears to have a particular preferred distribution and habitat in the TS. Studies have revealed that female P. pelagicus spawn in estuaries. Such individuals then emigrate into coastal marine water and release their eggs [55]. Large numbers of P. pelagicus were obtained in a sandy-muddy site in South TS influenced by freshwater runoff from the Kaoping River [23]. Furthermore, juvenile crabs inhabit shallow waters and migrate to deeper water when they grow larger [60]. These observations indicate that P. pelagicus has seasonal variation because they prefer habitats near an estuarine area with sand-mud substrates and brackish waters during the juvenile stages and then migrate deeper in the TS.

Effects of Ocean Environmental Variables on the Swimming Crabs
Understanding the relationships between environmental effects and fundamental processes related to variation in catch rates is crucial. All three crab species exhibited different distribution trends in the TS based on seasons and habitats. The Chl-a concentration explained the largest variance in GAM for C. feriatus and P. sanguinolentus. In previous habitat studies, environmental factors, such as organic matter, temperature, sediment size, and salinity, have been considered to influence variations in the diversity and distribution of crab species in both time and space [11,65]. Furthermore, Signa et al. [66] determined that the density of the swimming crab Polybius henslowii is strongly related to the concentration of Chl-a, suggesting more accumulation in locations with higher production. Moreover, the highest catch rates of C. smithii near Arabian sea coasts and pelagic red crab Pleuroncodes planipes along the coastal area of Baja California were recorded in upwelling systems [67,68]. In South-East Queensland, Australia, spanner crab Ranina ranina catch rates were the highest during the seasonal favorable upwelling in spring due to increased feeding during the mating season [8]. Essential fishing ground was mostly related to the upwelling area that consisted of a high Chl-a concentration and low SST that carried large amounts of nutrients from the bottom layer to the surface layer of the ocean [69] and is a crucial factor affecting the diet of C. feriatus and P. sanguinolentus in TS.
For P. pelagicus, BT explained considerable variance and had negative correlations with BTs >25 °C in the TS. Studies have revealed that the water temperature is a factor that controls the ovarian development and spawning of P. pelagicus [10,24,58]. Moreover, Qari and Aljarari [58] mentioned that temperature is a critical environmental parameter that plays a role in limiting the distribution and activity of P. pelagicus in the Red Sea. Furthermore, in Chesapeake Bay and South-East Queensland, when the water temperature was low, the catch rates of Callinectes sapidus and R. ranina increased For P. sanguinolentus, the spatial distribution was widespread in the TS for all seasons, and the distribution were further northeast during spring, summer, and autumn ( Figure 7). The seasonal predicted catch rates of P. sanguinolentus was also widely distributed around the TS and spread southwestward of the strait, including coastal mainland China, and then extended further to the northern area during autumn and winter. Samuel and Soundarapandian [61] mentioned that the peak seasons of P. sanguinolentus catch rates were from May to August and from October to December along the southeast coast of India. Huang et al. [62] revealed that the catch rates of P. sanguinolentus considerably decreased during spring along the coastal regions of China. P. sanguinolentus distribution and habitat seems more widespread in TS probably because these species can tolerate a wide range of environmental factors. Furthermore, sea currents may also be a primary factor the wide distribution of P. sanguinolentus in TS. According to Rasheed and Mustaquim [40], most of P. sanguinolentus larvae hatch as zoea drift in the water current before inhabiting benthic habits and grow into a juvenile crab.
Throughout the year, in TS, strong Kuroshio Branch Current and SCS Current during the summer and autumn may affect the movement of P. sanguinolentus in the northern area. P. sanguinolentus typically inhabit the sandy and muddy bottom in the shallow area [26,63], whereas adult and berried females often migrate to deeper waters for spawning [24,64], which results in widespread distribution of this species.

Effects of Ocean Environmental Variables on the Swimming Crabs
Understanding the relationships between environmental effects and fundamental processes related to variation in catch rates is crucial. All three crab species exhibited different distribution trends in the TS based on seasons and habitats. The Chl-a concentration explained the largest variance in GAM for C. feriatus and P. sanguinolentus. In previous habitat studies, environmental factors, such as organic matter, temperature, sediment size, and salinity, have been considered to influence variations in the diversity and distribution of crab species in both time and space [11,65]. Furthermore, Signa et al. [66] determined that the density of the swimming crab Polybius henslowii is strongly related to the concentration of Chl-a, suggesting more accumulation in locations with higher production. Moreover, the highest catch rates of C. smithii near Arabian sea coasts and pelagic red crab Pleuroncodes planipes along the coastal area of Baja California were recorded in upwelling systems [67,68]. In South-East Queensland, Australia, spanner crab Ranina ranina catch rates were the highest during the seasonal favorable upwelling in spring due to increased feeding during the mating season [8]. Essential fishing ground was mostly related to the upwelling area that consisted of a high Chl-a concentration and low SST that carried large amounts of nutrients from the bottom layer to the surface layer of the ocean [69] and is a crucial factor affecting the diet of C. feriatus and P. sanguinolentus in TS.
For P. pelagicus, BT explained considerable variance and had negative correlations with BTs >25 • C in the TS. Studies have revealed that the water temperature is a factor that controls the ovarian development and spawning of P. pelagicus [10,24,58]. Moreover, Qari and Aljarari [58] mentioned that temperature is a critical environmental parameter that plays a role in limiting the distribution and activity of P. pelagicus in the Red Sea. Furthermore, in Chesapeake Bay and South-East Queensland, when the water temperature was low, the catch rates of Callinectes sapidus and R. ranina increased considerably [8,70]. Andrade et al. [12] determined that in continental shelf waters off the Southeast Region of Brazil, neither SST nor BT had significant effects on swimming crabs, whereas, the spatial distributions of C. bimaculata, C. japonica, and P. trituberculatus in Haizhou Bay, China, were mostly affected by BT [30]. These differences could be because of the regions where the studies were conducted. The TS is located in a temperate region, with large variations in temperature throughout the year. The temperature not only affects the distribution of crabs but is also a critical factor in their reproduction, growth, and life stage.
Salinity appears to be one of the most critical environmental factors affecting the reproductive cycle, with higher salinities being favorable for breeding for portunid crabs [71]. Moreover, bottom salinity significantly influenced the distribution of C. bimaculata, indicating that the preferred salinity range was 29-31 in Haizhou Bay [30]. P. sanguinolentus females mostly inhabit deeper waters and prefer higher salinity compared with male crabs [72]. Salinity differences also affected the catch rates and habitat of C. danae and C. ornatus, which are commonly caught in areas with low salinity and estuary regions off the northeast coast of Brazil [73]. However, salinity explained among the lowest variance in GAM of the three important commercial swimming crabs in the TS. Jan et al. [74] discovered that bottom and surface sea salinity in TS varied. The bottom area recorded higher salinity throughout the year. The fishery data in the present study were collected from a crab trap, and the sub-surface salinity should be investigated in a future study. The depth is a crucial environmental factor shaping the community structure of swimming crabs in bottom environments, and changes in the structure of benthic communities may be related to the depth incline [11,33]. The composition of benthic crustacean communities near southwestern Taiwan was influenced by the water depth [35]. The species composition and distribution of crabs in the southern ECS were mostly related to the water depth and were maximum at a depth of 100-120 m, followed by 80-100 m [52]. Moreover, at the northern continental shelf in the SCS, the largest aggregation of crab species was mainly distributed at depths of 10-60 m with a peak in depths near 10 m [25]. Our GAM finding indicated that higher catch rates of C. feriatus, P. pelagicus, and P. sanguinolentus were observed at depths of approximately 100 m in the TS. Huang [54] also revealed that C. feriatus was mainly distributed in areas where the depth was less than 80 m in the southern part of ECS.
In the present study, P. sanguinolentus was distributed in the water of various depths around the north and southwest parts of the TS. Hsueh and Hung [24] suggested that the breeding of P. sanguinolentus is associated with high temperature and deeper waters areas near southern TS. P. sanguinolentus juveniles were often found in high concentrations in estuaries and inshore waters; adults were more abundant in deep waters; and females were abundant at depths of 40-80 m [22,72,75]. P. sanguinolentus females were abundant at a depth of 80 m, whereas males prefer depths of 40-60 m in the north of TS [39]. Thus, we conclude that the distribution of P. sanguinolentus in the TS moves toward the north of Taiwan and into deeper waters during spring, summer, and autumn, which could be because mature crabs breed during these seasons and require deeper and warmer areas.
Notably, swimming crabs adapt to their habitat and distribution of space and time. Because the TS is located in the East Asian monsoon region, wind direction changes with the seasons causes a different current that affects the marine environment and species [76]. The availability of light as a source of energy for photosynthesis, mineral nutrients, and temperature, which influence their metabolic rate, play a crucial role in regulating Chl-a in the oceans [77]. In addition, the high salinity and temperature with low nutrients from the warm SCS current and Kuroshio Branch Current flows in summertime and the strong northeastern winds push the fresh, cold, nutrient-rich China Coastal Current southward along the western part of the TS [78]. This causes a complex bottom topography, and the several current systems in TS result in changes in essential variables, such as Chl-a and BT, which affect these crabs.
The spatial and temporal variables in GAM analyses is critical because GAM is used to determine whether changes in catch rates are related to these environmental variables or some other spatial and temporal variables that are unaccounted for. However, most higher amount of variance in GAM analyses were attributed to the spatial and temporal variables. Lower variance was observed for environmental variables [8,30,31,48]. The environmental variables change over time and positions and their effect should be considered in fishery regulations [9,45]. Furthermore, the geographic information probably quite well known by fishers and fishery managers, and our seasonal catch rate distributions of three crabs in Figures 1-4 also could support the similar results. Our results suggested that models of catch rates that incorporate relevant environmental variables can be used to infer possible responses in the distribution of swimming crabs. Climate variability leads to changes in the fishing location. Further studies should investigate the annual variations to explore future changes in distributions for the swimming crabs.

Conclusions
In summary, our results confirmed that the distributions of C. feriatus, P. pelagicus, and P. sanguinolentus were related to spatiotemporal environmental variations in the TS. The Chl-a concentration accounted for significant catch rates for C. feriatus and P. sanguinolentus. BT accounted for the high catch rates for P. pelagicus. The high catch rates of C. feriatus and P. sanguinolentus occurred at a Chl-a concentration > 0.5 mg/m 3 , whereas P. pelagicus catch rates had negative correlations with BTs > 25 • C. The model predicted high catch rates of C. feriatus in the north during autumn and winter, whereas P. pelagicus was observed further to the south during summer and autumn in TS. The predicted catch rate of P. sanguinolentus was widely distributed around the TS for all seasons and distributed further to the northern area during autumn and winter. Based on this information, sustainable crab fisheries can be implemented in the future through harvest strategy planning, ecosystem management, spatiotemporal management, and bycatch reduction. We suggest that other environmental factors, such as bottom salinity, sediment type, and organic matter content, should be added in future modeling to improve predictions. This is because the natural habitat of crabs is often at the bottom of the sea, and a species' habitat considerably affects its catch rates. Moreover, recording the carapace size and sex of swimming crabs during the study period is essential for future studies to conduct a more comprehensive investigation of the influence of environmental factors on crab habitats and distributions.