1 Introduction

Bycatch is a severe challenge to effective marine fishery management globally (Lewison et al., 2004; Davies et al., 2009; Komoroske and Lewison, 2015; Sims and Queiroz, 2016). In contrast to commercially fished species, whose landing and effort information are commonly available, bycatch individuals are discarded during operation in most cases. Without long-term monitoring and data collection, quantitative, reliable, and standardized assessments are difficult, thereby impeding strategic intervention on specific fishing gears and activities (Crowder and Murawski, 1998; Temple et al., 2018; Savoca et al., 2020). Although bycatch is not an easily addressed issue, it has become one of the main research topics, especially nontarget vertebrates, including marine megafaunas, such as mammals, sea turtles, and seabirds (Moore et al., 2009; Lewison et al., 2014). However, invertebrate bycatch has been rarely explored.

As characteristic marine invertebrates, horseshoe crabs are commonly known to be bycatch by trawling in coastal waters (Smith et al., 2017; Meilana and Fang, 2020). However, their bycatch intensity across environmental gradients has yet to be systematically evaluated. Steele et al. (2002) assessed the efficiency of two bycatch reduction devices in otter trawling in Florida, United States. They found that the Atlantic horseshoe crab Limulus polyphemus is the most abundant among the bycatch of invertebrates. Supadminingsih et al. (2019) observed that horseshoe crab bycatch in bottom gillnetting, which is used to harvest the blue swimming crab Portunus pelagicus in Java, Indonesia, mostly occurs at water depths within 5 m. Zauki et al. (2019) also reported the bycatch occurrence of Asian horseshoe crabs in artisanal fishing in Balok, Malaysia. However, the insufficient assessment on bycatch intensity and distributional hotspots throughout environmental gradients hinders the implementation of species-specific conservation management.

Three species of Asian horseshoe crabs, i.e., tri-spine horseshoe crab Tachypleus tridentatus, the coastal horseshoe crab T. gigas, and the mangrove horseshoe crab C. rotundicauda, are distributed along the Indo-Pacific coastal water (Vestbo et al., 2018). A comparison of these species with their Atlantic counterpart L. polyphemus has suggested that the population baseline information about Asian horseshoe crabs is relatively limited, particularly in China, where most ecologyand conservation-related studies have focused on juvenile populations in intertidal nursery habitats (Wang et al., 2020). These species are mainly threatened by exploitation for Tachypleus amebocyte lysate (TAL) production, food consumption, and habitat loss through land reclamation (Laurie et al., 2019; John et al., 2020). The threatened level of T. tridentatus has been raised to endangered (EN) in the IUCN (International Union for Conservation of Nature) Red List of Threatened Species (Laurie et al., 2019), which is the highest grade among the four extant horseshoe crab species. In China, T. tridentatus and C. rotundicauda are listed as Class Ⅱ National Protected Animals issued by the Ministry of Agriculture and Rural Affairs, National Forestry and Grassland Administration of China in 2021 even though C. rotundicauda is maintained as data deficient (DD) in the IUCN Red List. Despite their increasing importance in conservation, the systematic evaluation of the perceived threats is widely lacking throughout the Indo-Pacific region, including the Chinese coastline.

Horseshoe crabs are mainly caught in not only subtidal coastal waters but also intertidal zones. Bycatch in intertidal zones remains relatively unexplored compared with that in coastal waters. With widespread tidal flats and the most abundant local populations of juvenile Asian horseshoe crabs in China (Xie et al., 2020), the northern Beibu Gulf offers a good opportunity to study horseshoe crab by- catch in intertidal zones. In this research, the bycatch of two species of Asian horseshoe crabs, namely, T. tridentatus and C. rotundicauda, by fishing gears available in the intertidal zones of the northern Beibu Gulf in Guangxi, China, was investigated. Across the environmental gradients in the intertidal zone, the density of fishing gears and the bycatch intensity of horseshoe crabs were evaluated. The spatial patterns of fishing gears and bycaught individuals were also recorded. Recommendations for prioritizing management to regulate particular fishing gears in specific places were provided to mitigate the adverse effect of bycatch on the threatened horseshoe crab populations.

2 Materials and Methods 2.1 Study Site

Beibu Gulf is a semi-closed bay at the northwestern South China Sea and is surrounded by Guangxi, Guangdong, and Hainan regions/provinces of China and northern- most Vietnam. It has been viewed as the last wilderness for Asian horseshoe crabs (Brockmann and Smith, 2009; Weng et al., 2012). Seven intertidal nursery habitats along the coastline of northern Beibu Gulf, Guangxi, namely, Ronggenshan (RGS), Tieshangang (TSG), Xibeiling (XBL), Zhongsandun (ZSD), Shaluoliao (SLL), Yuzhouping (YZP), and Jiaodong (JD), which were previously demonstrated to harbor the most abundant juvenile Asian horseshoe crab populations in China (Xie et al., 2020), were chosen as the study sites (Fig.1). Each study site is located in a specific local bay or estuarine system, with different characteristic vegetations and dominant Asian horseshoe crab species (Table 1). Overall, T. tridentatus dominates the eastern habitats of northern Beibu Gulf, whereas C. rotundicauda is more abundant in the western shoreline (Xie et al., 2020).

2.2 Survey Methods

In each study site, a predefined transect line that crossed environmental gradients within the intertidal zone of tidal creeks and the tidal flat between an embankment/dominant vegetation and a low tide line or river mouth was walked through during an ebb tide to search for fishing gears and examine the bycatch of Asian horseshoe crabs. The transect lines were designed in accordance with the method described in a recent population study on juvenile horseshoe crabs in the northern Beibu Gulf (Xie et al., 2020). This design was applied to include different land- scape elements, which are important for horseshoe crab spawning, including tidal creeks in mangrove forests and intertidal flats extending to the low tide line. 'Stationary' fishing gears rather than dynamic fishing activities in the intertidal zones were mainly considered, so small fishing gears commonly operated by local fishermen for beachcombing were excluded. When a fishing gear was sighted, the type, number, and GPS coordinate (Garmin Dakota 20) were recorded. A single coordinate was assigned for pointdistributed fishing gears, whereas multiple coordinates of both ends and middle turning points were noted for fishing gears with a linear distribution. If any Asian horseshoe crab bycatch was found in the fishing gear, their species was identified in terms of the cross-section shape of the telson and morphological characteristics of the prosoma. For example, T. tridentatus has a triangular telson, whereas C. rotundicauda has a round one. The prosoma of T. tridentatus is more convex and has more spines. Alive individuals and dead/decayed bodies were included. Then, the number of ensnared horseshoe crabs was counted, and the life history stages of Tachypleus tridentatus (TT) and prosomal width (PW, mm) of each individual of these two species was measured using calipers. The prosoma remains intact even when a horseshoe crab is dead for a certain period, and the rest parts of the body (partially) have decayed or disappeared; as such, it serves as a good indicator to estimate the life history stage of the individual bycatch. Table 2 shows the range of PW and the corresponding life stages of T. tridentatus (Sekiguchi et al., 1988; Kwan et al., 2021) and C. rotundicauda (Sekiguchi et al., 1988; Hu et al., 2015; Fan et al., 2017). Once the measurement was finished, the inspected individual was removed from the fishing gear to prevent repeat counts. The field surveys were implemented in the summer of 2021 (May – July).

2.3 Types of Fishing Gears

Fishing gears are categorized into three types, namely, ground cage, stick net set, and ghost net, after exploratory field surveys on the intertidal zones of the northern Beibu Gulf. Ground cage is a linearly distributed net with a mesh size of 10 – 20 mm and a length of several meters. It has multiple rectangle metal grids (length of each side: 30 – 60 cm) inside that hold its shape as a cage (Figs.2A – 2B). Both ends of a ground cage or a set of ground cages are tied on stationary objects, such as sticks, anchors, and stones (Fig.2B). A ground cage has several openings at the lateral sides for animals to enter. Each section of a ground cage, where both ends are held by a pair of rectangular metal grids, has an internal concave net that only allows one directional entrance and prevents the animals from escaping. Eventually, most captured animals crowd in one end of the ground cage (Fig.2B). Fishermen usually deploy the ground cage along or near tidal creeks that still have water during the ebb tide (Fig.2A) to capture nonspecific species, such as fishes, crabs, and shrimps. In this study, most of the sighted ground cages were abandoned with irregular shapes, compressed metal grids, and point distribution (Fig.2C).

A stick net set is a set of long nets (mesh sizes: 10 – 30 mm) that hang on multiple sticks. The height differs significantly among sets, ranging from several meters (Fig.2D) to centimeters. The net does not always hang on sticks, but it is laid down on the ground in some instances (Fig.2E). Stick net sets are generally deposited near the low tide line (Fig.2D) or along tidal creeks (Fig.2E), where some water remains even during the lowest tide period. Similar to ground cages, stick net sets do not target specific species; instead, they capture any organisms stuck inside during the ebb tide. Both ground cages and stick net sets are listed as illegal fishing gears by the Fisheries Bureau, Ministry of Agriculture and Rural Affairs of China.

A ghost net is not a specific kind of fishing gear, but fishing gears that cannot be categorized into either a ground cage or stick net set. It is a fragmented net that may come from a gill net, stick net, and other unknown sources. The nets of a discarded ground cage, where the net and metal grids are tightly entangled, are not considered a ghost net. Apart from a few exceptional cases, the encountered ghost nets generally have a point distribution (Fig.2F).

The length of the transect line ranged from 2.39 km to 4.58 km among the study sites (Table 3). Most ground cages and ghost nets were distributed in points except for the four linearly distributed cases of ground cages in RGS and SLL and ghost nets in SLL and JD (Table 3). The distribution of the stick net set was linear, and the length ranged from 7.5 m to 671.9 m per set.

2.4 Spatial Patterns, Density of Fishing Gears, and Bycatch Intensity

The distribution and number of fishing gears and Asian horseshoe crabs bycatch were mapped on Landsat images (Path/Row: 124045 and 125045; image date: 26th and 17th June 2021 for 124045 and 125045, respectively) by using ArcMap 9.3 (Esri) to clarify their spatial patterns. Level-1 data of Landsat 8 OLI were downloaded from the EarthExplorer platform, USGS (https://earthexplorer.usgs.gov/, access date: 2nd July 2021). The band composite of RGB 753 was applied because this false-colored image could enhance the contract of different landscape elements, e.g., between land and water, among vegetated and urbanized areas (Kerr and Ostrovsky, 2003), to display the characters of the studied intertidal zones clearly.

The total number (of ground cages and ghost nets) or total set number (of stick net sets) was divided by the transect line length (km) conducted during the survey to assess the density of these three fishing gear types in a given study site. The difference in the length of the linearly distributed stick net sets (m) among study sites was examined with the Kruskal-Wallis test because the assumption of data normality was not met. The lengths of a given transect line and a stick net set were measured with the Calculate Geometry function of ArcMap 9.3 after they were mapped on the projected coordinate system of WGS 1984 UTM Zone 49N.

The total number of bycatch individuals for each species in a given fishing gear was divided by the total number of the given fishing gear in a study site to calculate the bycatch intensity of each type of fishing gear on T. tridentatus and C. rotundicauda. The total number of each horseshoe crab species was also divided by the length of the transect line in a given study site to evaluate the bycatch intensity in each habitat. The difference in the bycatch intensity among the three types of fishing gears was examined with the Kruskal-Wallis test. Pearson correlation was applied to explore the relationship between the density of fishing gears and the bycatch intensity of horseshoe crabs. The difference in the PW of T. tridentatus (excluding in TSG where only one bycatch was found) and C. rotundicauda was compared among the study sites through one- way analysis of variance (ANOVA) because data had normal distribution and equal variance and among the three fishing gears via the Kruskal-Wallis test. Additionally, the difference in the status of the horseshoe crabs (i.e., the fraction of alive individuals in the total bycatch for each species) among the fishing gears was compared using the Kruskal-Wallis test. The Dwass-Steel-Critchlow-Fligner test was applied as the post hoc pairwise comparison for the Kruskal-Wallis test. Data were statistically analyzed with Systat 13.

3 Results 3.1 Density and Spatial Pattern of Fishing Gears

Fishing gears differed in their densities within any given site, and sites differed in the density of each type of fishing gear (Table 3). Although the three fishing gears had similar density throughout the study sites (1.0 – 7.1 cages, 0 – 5.7 sets, and 0.9 – 6.2 nets per kilometer of the transect line for the ground cage, stick net set, and ghost net, respectively), the density of ground cages was generally higher than two other fishing gears in four of the seven habitats (i.e., RGS, XBL, ZSD, and JD). The stick net set in SLL was not sighted during the field survey. The length of the stick net sets was not significantly different among habitats (Kruskal-Wallis, H = 6.35, p = 0.27).

Both ground cages and stick net sets (green rectangles and light blue lines, respectively, in Fig.3) had a similar distribution along the fringe of mangrove forests, along or near the tidal creeks, and along the low tide line close to estuaries. No obvious spatial pattern was observed in ghost nets (purple triangles in Fig.3).

3.2 Bycatch Intensity, Spatial Distribution, Size, and Life History Stage

The bycatch intensity of each species of Asian horseshoe crabs differed among the study sites and the three types of fishing gears (Table 4). Tachypleus tridentatus was more intensive in RGS (18.4 individuals per km transect) and ZSD (14.2 individuals per km transect), but C. rotundicauda was to a greater extent in YZP and JD (22.7 and 13.9 individuals per kilometer of the transect line in YZP and JD, respectively). In most cases, the bycatch intensity of each site matched the dominant species of juvenile horseshoe crab (Table 1); however, in XBL and SLL, where T. tridentatus dominated the juvenile population, the intensity of C. rotundicauda was higher than that of T. tridentatus (Table 4). The linear correlation between the density of fishing gears and the bycatch intensity of horseshoe crabs was not significant among all types of fishing gears (Pearson r = 0.34, 0.39, and 0.25; and p = 0.31, 0.3, and 0.46 for ground cages, stick net sets, and ghost nets, respectively).

The ground cage and stick net set caught more T. tridentatus and C. rotundicauda than the ghost net throughout their life stages (Table 2), although the difference in the bycatch intensity among the fishing gears was not significant (Kruskal-Wallis, H = 1.96, p = 0.16). Although the overall trend showed that higher proportions of individuals were caught in stick net sets during the four life stages, the fraction of bycatch in ground cages increased throughout the development of T. tridentatus (Table 2). During the life stages of the 9th – 10th instars and late instars, more C. rotundicauda was caught in stick net sets, whereas the proportion of adults caught in ground cages was higher than that in stick net sets (Table 2).

The main threat differed among habitats (Table 4). In RGS, ground cages (1.8 individuals per ground cage) caught more T. tridentatus per unit than stick net sets (1.3 individuals per stick net set). By comparison, stick net sets were the major threat for T. tridentatus in ZSD (2.3 individuals per stick net set). In the bycatch hotspots of C. rotundicauda, ground cages had more bycatches in YZP (8.6 individuals per ground cage). Conversely, stick net sets were the main threat in JD (9.3 individuals per stick net set). Overall, the intensity of C. rotundicauda was slightly higher (1.2 – 22.7 individuals per kilometer of the transect line) than that of T. tridentatus (0.3 – 18.4 individuals per kilometer of the transect line) among habitats.

Tachypleus tridentatus did not display a clear spatial pattern, although some of its bycatch occurred near the low tide line in RGS (Fig.3A) and ZSD (Fig.3D) or along the tidal creek in SLL (Fig.3E). For C. rotundicauda, most individuals were found along the river (Fig.3C), tidal creeks (Figs.3D – 3G), mangrove fringe (Fig.3F), and the boundary between the low tide line and estuary (Fig.3G).

The PWs of T. tridentatus (ANOVA, F = 0.74, p = 0.6) and C. rotundicauda (ANOVA, F = 1.38, p = 0.25) were not significantly different among the study sites. Most of them were adults or at late instar stages (Fig.4). Although T. tridentatus bycatch had more diverse life stages, which included the 10th – 12th instars, late instars, and adults (Fig.4A), almost all C. rotundicauda specimens were adults, except some 9th – 10th instars and late instars in JD (Fig.4B). The PW of C. rotundicauda differed significantly among the three fishing gears (Kruskal-Wallis, H = 38.7, p < 0.001). In particular, the largest individuals were found in ground cages (125.1 mm ± 17.6 mm), followed by those in stick net sets (105.6 mm ± 22.1 mm) and ghost nets (88.4 mm ± 18.2 mm; Dwass-Steel-Critchlow-Fligner test, p < 0.001 for all pairwise comparisons). Conversely, the PW of T. tridentatus did not significantly differ among the fishing gears (Kruskal-Wallis, H = 2.8, p = 0.25). As for the status of bycatch individuals, the percentage of alive individuals did not significantly vary among the three fishing gears (Kruskal-Wallis, H = 1.87, p = 0.39).

4 Discussion

In the seven studied intertidal zones along the coastline of northern Beibu Gulf, the density of the fishing gears (i.e., ground cages, stick net sets, and ghost nets) had similar ranges, but the number of ground cages per kilometer of the transect line was more than that of the other two gears. Ground cages and stick net sets were mainly distributed along tidal creeks, mangrove fringe, and near the low tide line. They were identified as the major threat for the two Asian horseshoe crab species. Consistent with the spatial distribution pattern of these gears, the occurrence of C. rotundicauda bycatch was mostly observed near tidal creeks, mangrove fringe, and low tide line, but the T. tridentatus bycatch had no clear spatial pattern. The majority of the bycatch individuals of both species were late instars and adults.

This study focused on bycatch by stationary fishing gears, i.e., fishing gears remaining on the intertidal zone during ebb tides, so the information about bycatch from other fishing activities was not available. The status of fishing gears, such as ground cages with different extending levels, stick net sets hung on sticks or laid down on the ground, and diverse gear/mesh sizes, may affect the results of bycatch. In our study, the potential effects of fishing gear status on horseshoe crab bycatch were not considered, but they should be further explored. Additionally, the majority of the recorded ground cages were not installed on any object, although some of them were partially buried in sediment or entangled in stick net sets. The position of these seemingly abandoned ground cages and fragmented ghost nets might vary throughout the tidal cycles. The intertidal landscape that affects local tidal flow might further complicate the distribution of discarded fishing gears. These factors could interact simultaneously; as such, the general spatial pattern of horseshoe crabs bycatch could not be easily concluded. Although our focus and applied methods had limitations, the evaluated density of different fishing gears and the by-catch intensity of these two species of Asian horseshoe crabs helped reveal the main threats. They also provided a basis for developing management solutions to solve issues on horseshoe crab bycatch in the intertidal zones of the northern Beibu Gulf.

Among the three types of fishing gears, ground cages and stick net sets showed clear spatial patterns along or near tidal creeks, mangrove fringe, and the low tide line at the boundary with estuaries. By contrast, ghost nets had no distribution pattern. The spatial distribution of fishing gears reflects the general ways by which fishermen use them. For instance, our observations implied that ground cages and stick net sets are commonly used to capture nonspecific targets, such as fishes, crabs, and shrimps. Landscape elements, such as tidal creeks and low tide lines, retain a thin water layer during ebb tides and may serve as the pathway for organisms to explore the increasingly immersed underwater area. Animals that move to the intertidal zone during flooding tides may get caught more easily by ground cages and stick net sets near tidal creeks and low tide lines. Therefore, nets of these fishing gears should stay at least partially immersed in estuarine water during ebb tides to ensure the accessibility and freshness of harvested organisms.

Bycatch intensity differed among the study sites and various types of fishing gears. The bycatch intensity of ground cages and stick net sets for these two Asian horseshoe crab species was stronger than that of ghost nets; that is, the proportions of bycatch individuals in ground cages and stick net sets throughout the life stages were higher than those in ghost nets. Generally, C. rotundicauda suffered from a more intensified bycatch pressure than T. tridentatus (1.2– 22.7 and 0.3 – 18.4 individuals per kilometer of the transect line for C. rotundicauda and T. tridentatus, respectively). Although the bycatch hotspots of T. tridentatus were unlikely found in this study, this result implied that the local population of T. tridentatus might be in a poorer state with smaller population size and lower density than C. rotundicauda. In contrast to DD (data deficient)-listed C. rotundicauda, T. tridentatus is upgraded to EN (endangered) in the IUCN Red List (Laurie et al., 2019). Experts from the Indo-Pacific region perceived an overall decline in the local population size of both species throughout their geographical range (Wang et al., 2020). In the northern Beibu Gulf, most residents of coastal communities in Guangxi also recognized a decrease in the local population of T. tridentatus (Fu et al., 2019; Liao et al., 2019). For the threatened Asian horseshoe crab populations in the northern Beibu Gulf, especially the severely depleting T. tridentatus, by- catch by stationary fishing gears in intertidal zones pose a remarkable challenge to an effective conservation.

The intensity of T. tridentatus bycatch in RGS and ZSD, which are the critical nursery habitats of T. tridentatus (Xie et al., 2020), was higher than those in other sites. Similarly, more bycatch of C. rotundicauda was noted in YZP and JD, which was dominated by juvenile C. rotundicauda (Xie et al., 2020). In XBL and SLL, which harbored abundant T. tridentatus juveniles (Xie et al., 2020), the number of C. rotundicauda bycatch was more than that of T. tridentatus. For these two species, the majority of them were late instars and adults. The phenomenon that more C. rotundicauda adults were found in juvenile T. tridentatus-dominated nursery habitats indicated that the 'spawning corridor' function of intertidal nursery habitat for Asian horseshoe crabs had been overlooked. For Asian horseshoe crabs, the baselines of juvenile populations and nursery habitats in intertidal zones are more abundant than those of spawning habitats and adult populations, particularly in China (Wang et al., 2020). Most studies have focused on the population structure of juveniles and the characteristics of nursery habitats in intertidal zones (Chiu and Morton, 2003; Hsieh and Chen, 2009; Chen et al., 2015; Koyama et al., 2020; Meilana et al., 2021). Contrary to North American L. polyphemus, whose spawning activities (Widener and Barlow, 1999; Smith et al., 2002; Cheng et al., 2015) and habitat characteristics have been well studied (Weber and Carter, 2009; Landi et al., 2015), Asian horseshoe crabs have been rarely described in terms of the function of intertidal zones in spawning (Shuster and Sekiguchi, 2009; Mohamad et al., 2019). Our results showed that most by-catch of T. tridentatus and C. rotundicauda were adults, highlighting the importance of intertidal zones in their spawning activities.

Similar to the spatial pattern of major threats from ground cages and stick net sets, the distribution of C. rotundicauda bycatch was mainly found along or near the tidal creeks, mangrove fringe, and low tide line. Conversely, the spatial distribution of T. tridentatus bycatch did not show a clear pattern, but it seemed to scatter across the area between the embankment/vegetation and low tide line. The density of the three fishing gears and bycatch intensity had no significant correlation, indicating that the spatial distribution of fishing gears could be more influential than their density in determining the bycatch intensity of horseshoe crabs. Most bycatch individuals were adults, so spawning activity might play an important role in determining the reported spatial distribution of bycatch. Therefore, T. tridentatus and C. rotundicauda likely had different spawning behaviors and preferences. In the absence of systematic field surveys, adults of C. rotundicauda were widely observed alone or in amplexed mating pair in tidal creeks during the ebb tides in summer. This result implied that C. rotundicauda might move to the high tide line for spawning even within the period of the ebb tide. The pathway to high tide spawning grounds is restricted in the low tide line and tidal creeks that cross mangroves and tidal flats; thus C. rotundicauda was caught in fishing gears close to these landscape elements. As for T. tridentatus, we never observe any alive adults during the ebb tides in our long experience of field surveys, thus its spawning activity remains nearly unknown. They probably moved to spawning habitats during flood tides, where the intertidal zone was increasingly immersed under water. Mating pairs of T. tridentatus could approach the high tide area through more diverse microhabitats in the intertidal zone; as a result, the bycatch patterns were more scattered. Additionally, the proportion of these two species in ground cages increased from the juvenile stages to the adult stage. Ground cages intended to catch C. rotundicauda with a significantly larger PW than the other two fishing gears. This finding suggested that the design of ground cages met the instinctive and burying behavior of horseshoe crabs. Therefore, ground cages were likely an 'attractor' for large individuals. However, future management endeavors should focus on the potential detrimental effect of ground cages and clarify the behavior of adults.

The spatial pattern of fishing gears and the bycatch intensity of Asian horseshoe crabs in the intertidal zone indicated the need for managing ground cages and stick net sets, which are listed as illegal fishing gears (issued in 2013 by the Fisheries Bureau, Ministry of Agriculture and Rural Affairs of China). Under the limited manpower of law enforcement, we strongly recommend targeting bycatch hotspots, including tidal creeks, mangrove fringe, and low tide lines, to regulate and remove ground cages and stick net sets regularly. Given that T. tridentatus is severely endangered (Laurie et al., 2019) and the local population of C. rotundicauda is potentially declining (Wang et al., 2020), the suggested management strategy can help mitigate the threat posed by stationary fishing gears and ensure the ecological function of intertidal zones as spawning corridors and nursery habitats for the threatened Asian horseshoe crabs.

In addition to bycatch from these three types of fishing gears in intertidal zones, the intentional capture of Asian horseshoe crabs in subtidal coastal areas in Chinese waters remains largely unknown. For local communities in Guangxi, horseshoe crabs are considered traditional seafood, particularly the EN-listed T. tridentatus (Fu et al., 2019; Liao et al., 2019). Another source of threat is the direct harvest of horseshoe crabs for TAL (Tachypleus amebocyte lysate) production. Although our observations suggested that the awareness of horseshoe crab conservation has increased since T. tridentatus and C. rotundicauda were upgraded to Class Ⅱ National Protected Animals by the Chinese government in early 2021, potential needs for consumption and TAL production may enhance illegal exploitation through smuggling. Besides the explicit stress due to overharvesting, implicit factors such as habitat loss and degradation can also affect the local populations of Asian horseshoe crabs (John et al., 2018; Laurie et al., 2019). Critical intertidal nurseries and spawning habitats have likely suffered from extensive land reclamation in the coastal waters of China (Tian et al., 2016). Therefore, future studies should target explicit and potential stresses to quantify and assess their impacts on Asian horseshoe crab populations systematically. An integrated method of field survey, questionnaire inquiry, and satellite remote sensing should be developed to explore the long-term changes in population size, distribution, and habitat status (Vihervaara et al., 2017), and help design, prioritize, and implement effective management interventions.

5 Conclusions

Among the three types of fishing gears, ground cages and stick net sets were the main threats for the two Asian horseshoe crab species, namely, T. tridentatus and C. rotundicauda, in the intertidal zone of northern Beibu Gulf, Guangxi, China, during summer. The majority of horseshoe crabs bycatch were large individuals at the life stages of late instars and adults. The ground cages and stick net sets along/near the tidal creeks, mangrove fringe, and the boundary between the low tide line and nearby estuary imposed a strong bycatch intensity on C. rotundicauda. However, no clear spatial pattern was observed in T. tridentatus. Our study suggested that regularly regulating and removing ground cages and stick net sets in the bycatch hotspots can help ensure the functionality of the intertidal zone as a spawning corridor and nursery habitat for horseshoe crabs. Combined with records on critical information about stationary fishing gears and the bycatch of horseshoe crabs, our method of a transect line that crossed environmental gradients in the intertidal zone can serve as a standard for systematically evaluating the density of fishing gears and the bycatch intensity of endangered Asian horseshoe crabs and other non-targeted species.

Acknowledgements

This study was supported by the Basic Research Fund of Guangxi Academy of Sciences (No. 2020YBJ706), the National Natural Science Foundation of China (No. 3206 0129), the Guangxi Natural Science Foundation (No. 2018 GXNSFBA281071), the Guangxi Science and Technology Base and Talent Project (No. 2021AC19355), the Guangxi BaGui Youth Scholar Program, and the Guangxi Recruitment Program of 100 Global Experts. We would like to thank Dr. D. Christopher Rogers for his kind help in improving the English expression of the manuscript.