Introduction

Biological invasion is considered one of the most important direct drivers for the decline of global biodiversity in recent decades (Kolar and Lodge 2001; Perrings 2001; Louette 2012). Invasive species can replace or even extirpate native species via niche displacement, hybridization and exploitation (i.e. predation and parasitism), altering community structure and ecological processes (Ruiz et al. 1997; Streftaris et al. 2005). Some of these ecological impacts can further disrupt ecosystem services causing adverse economic impacts (Perrings 2001). For example, the accidental introduction of the warty comb jellyfish Mnemiopsis leidyi A. Agassiz 1865 to the Black Sea led to a dramatic drop in local fish populations, causing a loss of around USD$250 million in the fisheries industry (Ruiz et al. 1997; Anil et al. 2002). Owing to the tremendous growth of international trade, the problem of biological invasion has worsened (Ruiz et al. 2000) and become a pressing issue prompting governments around the world to formulate preventive and mitigation measures for invasive species management (McNeely 2000; Law 2006).

Despite the fact that invasive populations can be rapidly removed by biocides over a large area, biocides have a high environmental footprint and may impact non-targeted species (Wittenberg and Cork 2005; Giakoumi et al. 2019). For this reason, the biological control role of native generalist predators has been increasingly investigated to develop sustainable and environmentally-friendly approaches for bioinvasion management (Wittenberg and Cork 2005; Myers and Cory 2017). According to the “enemy release hypothesis” (Keane and Crawley 2002), however, generalist predators may facilitate the establishment of invasive species by exerting a greater predation pressure on native rather than invasive prey (Colautti et al. 2004; Veiga et al. 2011; Naddafi and Rudstam 2014). Thus, to be able to adopt native predators as a biological control agent, quantifying predator prey preferences is a key consideration to assess if predators are facilitating or deterring invasive prey (Lai 1996; Veiga et al. 2011).

Prey preferences are dependent on prey profitability, which is traditionally defined as the energy gain per unit handling time when predators prey and feed on their prey (Elner and Hughes 1978; Hughes and Seed 1981). The profitability of a prey is influenced by a combination of factors, such as species (Dudas et al. 2005; Veiga et al. 2011), size (Elner and Hughes 1978; Hughes and Seed 1981), and habitat type (Sponaugle and Lawton 1990; Eggleston et al. 1992). These factors are considered important criteria for selecting prey based on optimal foraging theory (Krebs 1980). Selecting the optimal set of prey allows predators to maximize their energy gain per unit time, and to achieve that, there are two general strategies adopted by predators depending on the physical environment: energy maximization and time minimization (Elner and Hughes 1978; Hughes and Seed 1981; Seed and Hughes 1995). Energy maximizing predators spend more time on foraging and consume prey with acceptable profitability to maximize their net energy gain, particularly in benign environments, while time minimizers devote less time on predation and obtain less energy to minimize the time exposed to danger and environmental stresses (Seed and Hughes 1995). As a result, apart from prey preference, the physical environment is another key factor which may impact the effectiveness of native predators in biological control via indirect effects on their foraging strategies.

In the intertidal zone, temperature and salinity are two key physical environmental factors determining the performance, survival and distribution of predators and prey (Kinne 1966; Garton and Stickle 1980). The metabolism of ectothermic predators generally elevates with ambient temperature, causing an increase in energy demand and allowing them to attack, handle and process prey items at faster rates (Wallace 1973; Bergman 1987; Barbeau and Scheibling 1994). Similarly, changes in salinity also causes an elevation in predators’ metabolism to maintain internal osmotic balance, leading to increases in energy demand and food consumption (Taylor et al. 1977; Curtis et al. 2010). Although such changes in temperature and salinity can drive higher predation rates, extreme variations in these factors can, conversely, cause physiological stress and impact survival of predators (McGaw and McMahon 1996; McGaw 2007). Severe environmental stress also can lower predation rates indirectly by limiting locomotion, and thus decreasing the predator–prey encounter rate (Draper and Weissburg 2022).

The black pygmy mussel Xenostrobus securis (Lamarck 1819) is native to estuarine habitats of Oceania (Wilson 1968), but invasions of this species have been documented in Southern Europe and East Asia since the last century (Lazzari and Rinaldi 1994; Sabelli 1993; De Min and Vio 1997; Kimura et al. 1999; Shirafuji and Sato 2003; Garci et al. 2007; Pascual et al. 2010), and in the Mediterranean Sea this mussel is considered as one of the 100 “worst” marine invasive species (Streftaris and Zenetos 2006). Xenostrobus securis was first observed in the Shing Mun River and inner Tolo Habour of Hong Kong in 2010 and was suggested to be introduced via the ballast water of maritime vessels (Morton and Leung 2015). Xenostrobus securis has extremely wide thermal and osmotic tolerance ranges from 5 to 32 °C (Morton and Leung 2015; Astudillo et al. 2017) and 1–31‰ (Wilson 1968), allowing the population to spread rapidly across the hydrological gradient from the eastern (oceanic) to the western (estuarine) waters of Hong Kong (Astudillo 2015; Lau et al. 2018). The growth of X. securis in Hong Kong has resulted in this species outcompeting the native mussel, Arcuatula senhousia (Benson 1842), and also increasing the mortality of the cultured oyster, Magallana hongkongensis (Lam & Morton 2003), resulting in an estimated loss of more than HKD$2.4 million (approximately USD$304 thousands) to the local oyster production industry (Lau et al. 2018).

In spite of the rapid expansion and devastating effects of X. securis, there has been little discussion on how to manage invasion by this species in Hong Kong. Astudillo et al. (2018) examined the use of the whelk Reishia clavigera (Küster 1860) as a native predator to control X. securis. Although the whelk consumed more invasive mussels (X. securis and Mytilopsis sallei (Récluz 1849)) than the native species (Brachidontes variabilis (Krauss 1848)), the consumption rate was significantly decreased when the whelk was exposed to salinities below 22‰, conditions where X. securis is still able to survive and reproduce (Astudillo et al. 2018). As a result, the invasive mussels were effectively exempted from predation by the native whelk depending on the salinity of ambient waters.

Apart from whelks, swimming crabs in the family Portunidae are another common predator of bivalves in Hong Kong. In particular, the swimming crab, Thalamita danae Stimpson 1858, which is one of the most abundant and common predatory crabs in Hong Kong, has been shown to consume bivalves as its dominant prey (Lai 1996). Hence, considering the abundance of the swimming crab in the eastern and southern waters of Hong Kong (Lai 1996) and the fact that the crab preys primarily on bivalves, this paper investigates the prey preference of T. danae on the invasive mussel X. securis under different physical conditions (thermal and salinity stresses typically found in local waters). The predation rate on, and prey profitability of both the native B. variabilis and invasive X. securis were also compared to determine the differences in prey preference of the crab. By evaluating performances and feeding behaviours in different hydrological conditions, our study assessed the potential role of T. danae as a predator of the invasive X. securis in this newly colonised area, and how environmental conditions may modify such predation, in order to inform mitigation measures for bioinvasion management in the future.

Materials and methods

Sample collection and maintenance

The invasive mussel Xenostrobus securis is found in the low intertidal to subtidal zones of Hong Kong, and is widespread among estuarine (Deep Bay, Lau et al. 2018) and sheltered shores (inner Tolo Harbour, Morton and Leung 2015; Kwun Tong typhoon shelter, Astudillo et al. 2018) where it is often sympatric with the native mussel Brachidontes variabilis in the low intertidal zone (Deep Bay, Tolo Harbour and Mirs Bay, Huang et al. 1992). In contrast, the native swimming crab, Thalamita danae, is mainly distributed along the oceanic shore of eastern and southern waters, such as Mirs Bay, Tolo Harbour, Port Shelter and Tai Tam Bay (Lai 1996) and there are no records of this crab in the western estuarine waters. The eastern oceanic coast, therefore, is the only region where the distribution of the crab and the two mussels overlap. For this study, adult crabs were collected in Starfish Bay, Hong Kong (22°26′N, 114°14′E, Fig. 1) at around 0.7–1.0 m above Chart Datum (C.D.) during low tide periods from September to December 2021 for different experiments. Starfish Bay is a sheltered sandy shore fringed by rocky, boulder substrates that are habitats for T. danae (Lai 1996). In Starfish Bay where the crab was abundant, however, both native and invasive mussels were uncommon and, therefore, mussels were collected from Island House, Hong Kong (22°26’N, 114°10’E, 6.8 km from Starfish Bay) at approximately 1.0–1.5 m above C.D. (Figs. 1 and S1).

Fig. 1
figure 1

Hong Kong showing the distributions and sampling sites of the targeted animals. Black triangles represent the native predatory crab Thalamita danae (after Lai 1996) and the grey triangle Starfish Bay (the sampling site for T. danae). Black circles represent the invasive mussel Xenostrobus securis (after Morton and Leung 2015; Astudillo et al. 2017; Lau et al. 2018) and the grey circle Island House (the sampling site for X. securis and Brachidontes variabilis)

Full size image

All collected animals were maintained in the Swire Institute of Marine Science, The University of Hong Kong until experimentation. Brachidontes variabilis and X. securis were cleaned of epibionts, sorted and held in plastic containers (19.5 cm × 12.5 cm × 11 cm each) at a density of 40 individuals per container. The mussels were maintained under aerated, natural seawater (22 ± 2 °C and 32 ± 2‰, changed twice a week) and a photoperiod of 12 h:12 h day-night cycle. The mussels were fed with a commercial algae-based soft coral food (Reef Phytoplankton, Seachem, USA) every other day. A total 42 crabs were collected (12 for tolerance limits experiment, 18 for prey preference experiments, and 12 for the handling time experiments). Collected crabs were maintained in conditions as above, except in larger containers (26 cm × 16 cm × 13 cm) at a density of 1 individual per container. Instead of a fixed thermal and salinity regime, crabs in each container were randomly assigned to one of the acclimation treatments (see below). Ample amounts of commercial dried shrimp turtle food (Sun Dried Red Shrimp, Zoo Med, USA) instead of collected mussels were offered every other day to the crabs to avoid influencing their prey preference for different mussels (Cunningham and Hughes 1984).

Effects of acute temperature and salinity stresses on T. danae

The tolerance limits of T. danae were assessed under the combined stresses of temperature and salinity in conditions representative of habitats in Hong Kong (temperature: ranging from 17.5 to 31.0 °C, salinity: ranging from 5.4 to 33.2‰, EPD 2022), in order to determine the conditions used in the predation experiments. Only salinities above the lethal level (as indicated by the salinity where 50% of mortality occurred within 7 days) and temperatures below the Arrhenius breakpoint temperature (ABT) were selected, such that foraging activities are plausible without the crabs being killed by extreme environmental conditions. To achieve this, each of the 12 crabs were measured (30.1–58.2 mm carapace width) and randomly assigned to one of the four salinity treatments (5, 15, 25, and 35‰, n = 3 per treatment, all maintained at 22 °C) for a week and subsequently exposed to simulated thermal stress. Salinities were adjusted to the target levels by diluting natural seawater with Milli-Q water. The mortalities of T. danae in each treatment were checked daily to determine their survival under different salinities during the acclimation. The crabs were then exposed to an acute thermal ramp after the acclimation and their heart rates were measured to assess their physiological performances (Marshall et al. 2011; Burnett et al. 2013). Specifically, the crabs were taken out from the acclimation containers and introduced into a temperature-controlled waterbath (GP200, Grant Instruments, UK). The crabs were held separately in small plastic containers, in air, inside the waterbath, which heated the water from room temperature (25 °C) to 45 °C at a ramping rate of 1 °C per 5 min (total duration ~ 100 min) to simulate the natural rate of rock temperature increase in the intertidal zone of Hong Kong during emersion periods (after Luk 2014), in order to examine physiological responses under the maximum heating rate the crab could experience on the shore. A fine-tipped thermocouple (K-type, Omega) was inserted into the thorax coxopodite joint of an additional crab inside the waterbath to measure the crabs’ body temperatures with a digital thermometer (± 0.01 °C, TM-947SD, Lutron, Taiwan). Body temperatures of the crabs increased from 22 to 25 °C over 24 min when introduced into the containers and, during the thermal ramp increased at a rate of 0.6 °C per 5 min.

A non-invasive infrared sensor technique was adopted to monitor the heart rates of the crabs during the thermal ramp (after Depledge and Andersen 1990; see Burnett et al. 2013). A reflective optical sensor (CYN70, Vishay Semiconductors, USA) was glued (using cyanoacrylate glue) onto the crabs’ carapaces directly above the heart for real-time heart rate acquisition. An amplifier (AMP03, Newshift, Portugal) was used to amplify and filter the signals from the sensor, which were then transmitted to a PC-based oscilloscope (Picoscope 2204, Pico Technology, UK) before being recorded in the computer. The acute thermal ramp was ended when heartbeats of the crabs ceased at high temperatures. Recorded heartbeats were counted every twenty seconds to obtain heart rate expressed in hertz (beats per second, Σn = 4 salinity treatments × 3 replicates = 12 crabs in total).

Prey preference

Consumption rates and prey preference on the native B. variabilis and the invasive X. securis by T. danae were quantified for crabs acclimated under different environmental conditions. Each of 18 crabs were measured (36.6–67.5 mm carapace width) and randomly assigned to six treatments: crossing two temperatures (22 and 28 °C) and three salinities (15, 25, and 35‰, n = 3 per treatment) for a week (in acclimation containers as described above). Salinities and temperatures were adjusted to the target levels by diluting seawater with Milli-Q water and aquarium immersion heaters (Aquael, Poland) inside the acclimation containers, respectively. The crabs were fed as previously described but food supply was stopped 72 h prior to predation tests to standardize the satiation level of the crabs.

To quantify the crabs’ prey preferences, individuals were taken out from the acclimation containers and subjected to both with-choice and without-choice tests with the presence of both (with-choice) or one species (without-choice) of mussels (Underwood and Clarke 2005). The logic of comparing with- and without-choice tests is to quantify preferences unconfounded by consumption of prey when they are presented to the predator under the laboratory setting (as derived from the without-choice test, see Underwood and Clarke 2005). Control tests with only the mussel prey but not the crabs were also conducted to assess natural mortalities of the prey during the experimental period (Astudillo et al. 2018).

In the without-choice test, the crabs were provided with only one type of mussel prey: either ten native B. variabilis or ten invasive X. securis per container (Σn = 2 predator presence conditions (with or without crabs) × 2 prey × 2 temperatures × 3 salinities × 3 replicates = 72 tests in total). In the with-choice test, the same crabs were exposed to both types of mussel prey simultaneously: five native B. variabilis and five invasive X. securis per container (Σn = 2 predator presence conditions (with or without crabs) × 2 temperatures × 3 salinities × 3 replicates = 36 tests in total). Overall, ten mussel individuals were offered to the crab in each test. All predation tests were conducted in feeding arenas (26 cm × 16 cm × 13 cm each, with an opaque covering to avoid visual disturbance) with seawater at the same temperature and salinity as the crabs experienced in the acclimation containers. The tests lasted for 24 h, and the number of mussels being consumed was counted (defined as the complete absence of flesh in the shell; partial consumption was not counted). Every crab in this experiment underwent three predation tests in a randomized manner over time, including a without-choice test for B. variabilis, a without-choice test for X. securis, and a with-choice test for both mussel prey. After every test, the crabs were transferred back to the acclimation container and allowed to rest for at least three days between tests.

Handling time and mussel profitability

Predatory behaviour and handling time (breaking time plus feeding time) of the crabs were quantified to assess profitability of each type of mussel prey using timelapse videos. Each of the 12 crabs were measured (37.6–64.8 mm carapace width) and randomly assigned to acclimate in one of the four treatments crossing two temperatures (22 and 28 °C) and two salinities (25 and 35‰, n = 3 per treatment) for a week. Individuals were then taken out from the acclimation containers and provided with either one native B. variabilis or one invasive X. securis in the feeding arena setup as described above (Σn = 2 prey × 2 temperatures × 2 salinities × 3 replicates = 24 tests in total) with water temperature and salinity set as the same conditions as in the acclimation containers. Total shell lengths of the offered mussels were measured (± 0.5 mm) and to standardize the prey size of the two mussel species, only mussels with shell lengths between 15 and 20 mm were used (see Fig. S2, supplementary information for mussel size distributions on the shore). Breaking time and feeding times of the crabs preying on the mussels were recorded using a timelapse camera (T45A, Campark, China) positioned over the top of the arenas. Breaking time was defined as the period from which the crab seized the mussel prey until the mussel shell was broken, while feeding time was defined as the period from the shell being opened to the point where empty shells or shell fragments were discarded (Hughes and Seed 1981). The experiment lasted for two hours during which the crabs completed feeding and abandoned the emptied shells. The handling times for both native B. variabilis and invasive X. securis were quantified for each crab once, which was then returned to the acclimation containers and rested for at least three days between feeding measurements. Profitability of each consumed mussel was then calculated as the ratio between estimated dry flesh weight (as estimated from shell length, see supplementary information) and recorded handling time (= breaking + feeding times, Fig. S3).

Data analysis

Acclimation mortality of the crabs used in the tolerance limit experiment was tested between salinity treatments via a generalized linear model with binomial distribution (logit link function). The variation in thermal tolerance of T. danae between different salinity treatments was compared using their Arrhenius breakpoint temperature (ABT) and flatline temperature (FLT) under the acute thermal ramp. ABT is considered an indicator of rapid alteration or decline in metabolic performances under heat stress (Cossins and Bowler 1987), while FLT represent the limits of physiological tolerance when the heart ceases to function. To determine ABT for each crab, segmented regressions were conducted on an Arrhenius graph with the natural logarithmic of heart rate (ln (HR)) as the response variable and 1000/(T (temperature measured in Kelvin scale)) as the explanatory variable, and typically at least one breakpoint would be shown in each crab depending on their thermal sensitivity. Segmented regression models with 1–7 breakpoints were fitted to these graphs and the Akaike information criteria (AIC) of the fitted models were used for model comparison. For each crab, the model with the lowest AIC value was adopted after which ABT was determined as the final breakpoint (Dahlhoff and Somero 1993; Stillman and Somero 1996). FLT was estimated using the normal plot (HR (Hz) vs T (temperature measured in °C)) by extrapolating the fitted regression line after the ABT to the x-intercept (i.e. when HR = 0, Dahlhoff and Somero 1993). ABT and FLT were compared among the salinity treatments using one-way ANOVAs followed by Tukey’s HSD post-hoc tests (variances within treatments were homogenous, Levene’s tests P > 0.05 in all cases).

The prey preference of the crabs on the native and invasive mussels was analysed following Underwood and Clarke (2005). Chi-squared tests were adopted to test for differences between observed and expected crab consumptions of each mussel species based on with- and without-choice tests. In the without-choice tests, the effects of salinity (fixed, three levels), temperature (fixed, two levels), crab size (fixed, continuous), mussel species (fixed, two levels), and their interactions on the consumption rate of the crabs (= proportion of mussels being consumed, as the response variable) were analysed with a linear mixed model using the lme4 package (Bates et al. 2015) in R. The effect of crab size was assessed by comparing models with or without crab size, and the model was adopted based on whether crab size significantly improved the model fit. This was a balanced repeated measures design, in which each crab underwent a without-choice test for B. variabilis, a without-choice test for X. securis, and a with-choice test for both mussel species. Due to the repeated use of the same crab between treatments, the crab identity was treated as a random factor (measurements using the same crabs were considered independent, however, due to at least three days of recovery time). Using the same set of explanatory variables (salinity, temperature, crab size, mussel species and their interactions), the breaking, feeding and handling times of the crabs on, as well as the profitability (estimated by dry weight/ handling time) of both mussel prey, were analysed using generalized linear mixed models with gamma distribution (log link function, via the glmmTMB package, Brooks et al. 2017). Gamma distribution was used since more variable responses were observed during data exploration when the mean responses were higher. This model was also a balanced repeated measures design, in which each crab was used once in each of the B. variabilis and X. securis handling time tests. Goodness-of-fit of the model was evaluated based on standardized residual diagnostics (using the package DHARMa in R, Florian 2022).

Results

Tolerance limits

The mortality rate of Thalamita danae acclimated in 5‰ salinity was significantly higher than in 15‰ salinity or above (Generalized linear model with binomial distribution, χ23 = 12.17, P < 0.01, Fig. 2). No mortality was noted in 35‰ during the seven days acclimation and > 75% of crabs in 25 and 15‰ survived throughout the acclimation. All the crabs exposed to 5‰, however, died within 48 h. Heart rates of crabs acclimated in 35 and 25‰ increased steadily with temperature following the log-linear pattern until the ABT was reached, after which the heart rate dropped abruptly to flatline. In contrast, the heart rates of all crabs acclimated to 15‰ depressed twice before reaching the final breakpoint under the acute thermal ramp (Fig. S4). Neither ABT nor FLT, however, differed between salinity treatments (ABT: one-way ANOVA, F2,6 = 1.44, P > 0.05; FLT: F2,6 = 0.84, P > 0.05, Fig. 3). On average, heart rates of the crabs increased with rising body temperatures but started to decline at 34.6 °C (ABT) and reached zero at 39.7 °C (FLT).

Fig. 2
figure 2

Average survival duration of the crab, Thalamita danae, under four different salinity treatments (5, 15, 25, 35‰) within a period of 7-days. Error bar =  + SD (n = 4)

Full size image
Fig. 3
figure 3

Average values of a Arrhenius breakpoint temperature (ABT) and b flatline temperature (FLT) of the heart rate of the crab, Thalamita danae, under an acute thermal ramp following three different salinity treatments (15, 25, 35‰). Error bar =  + SD (n = 3)

Full size image

Prey preference

All mussels survived throughout the experiment in the control tests (i.e. in the absence of crabs). Thalamita danae predated and fed on both native Brachidontes variabilis and invasive Xenostrobus securis when offered these prey. The proportion of each mussel prey consumed during the without-choice tests was, however, similar to that of with-choice tests in nearly all 18 comparisons, indicating that the crab did not show any preference towards either the native or invasive mussels (Table 1). The only significant difference between observed and expected consumption was shown in one individual held at 15‰ and 28 °C (Table 1). For prey consumption in the without-choice tests, the consumption rate did not vary with crab size and, as a result, a more simple model involving only temperature, salinity, prey type and crab identity was adopted (see Table S1). No significant variation in the consumption rate between the native and the invasive mussels was found in the without-choice tests (linear mixed model, χ22 = 2.32, P > 0.05, see Table S1 for results). The proportion of mussels consumed, however, decreased significantly when the crabs were exposed to 15‰ (χ22 = 18.30, P < 0.01). The crabs consumed more than 55% of mussel prey under salinities of 25 and 35‰, whereas only around 6% of mussel prey were eaten when exposed to 15‰ (Fig. 4). There was a significant interaction among mussel species, temperature, and salinity in without-choice tests (χ22 = 6.25, P < 0.05, see Table S2 for post-hoc test results). Water temperature, however, did not significantly influence the consumption rate of the crabs (see Table S1).

Table 1 Chi-squared tests showing the prey preference of the crab, Thalamita danae, for either the native mussel, Brachidontes variabilis, or the invasive mussel, Xenostrobus securis, under different temperature and salinity treatments
Full size table
Fig. 4
figure 4

Number of mussels consumed by various-sized crabs in without-choice tests for a period of 24 h after the crabs were acclimated to one of the combinations of two temperatures (22 and 28 °C) and three salinities (15, 25, and 35‰). The colour and shape of a point represent the temperature and salinity that the crab experienced, respectively (n = 3)

Full size image

Handling time and mussel profitability

Crab spent nearly twice as long to handle the native than the invasive mussel (generalized linear mixed model, χ21 = 8.12, P < 0.01, Fig. 5c). This discrepancy was attributed to differences in both the breaking (χ21 = 6.87, P < 0.01, Fig. 5a) and feeding (χ21 = 27.04, P < 0.01, Fig. 5b) times. As a result, the invasive mussel had a significantly greater profitability to the crabs, producing approximately 0.03 mg/s higher rate of energy gain than the native mussel (χ21 = 71.43, P < 0.01, Fig. 5d). The breaking time of the crabs increased significantly when they were exposed at 28 °C (χ21 = 4.21, P < 0.05, Fig. 5a). The feeding time increased substantially when crabs were exposed to 25‰ (χ21 = 18.53, P < 0.01, Fig. 5b). The mussels, therefore, presented significantly lower profitability to the crabs under hypo-osmotic stress (χ21 = 3.93, P < 0.05, Fig. 5d). Significant interactions between crab size and the three fixed factors (species, temperature and salinity) were, however, found on the feeding time, in which larger crabs consumed the native mussel at a faster rate, especially under 28 °C or 35‰ (see Table S3 for the results of generalized linear mixed model, Fig. 5b). Correspondingly, the profitability of native mussels was more dependent on crab size as compared to invasive mussels (see Table S4 for results of generalized linear mixed model, Fig. 5d).

Fig. 5
figure 5

a Breaking time, b feeding time, c handling time and d profitability of various-sized crabs, Thalamita danae on both native and invasive mussels under the combination of two temperatures (22 and 28 °C) and two salinities (25 and 35‰). The colour and shape of a point represent the temperature and salinity that the crab experienced, respectively (n = 3)

Full size image

Discussion

Physiological tolerance limits of Thalamita danae

The thermal and salinity tolerance limits of Thalamita danae were around 35 °C and 15‰, which are similar to those of other portunoid crabs (e.g. T. crenata, Rüppell 1830; Kannupandi et al. 1997; Azra et al. 2020; Liocarcinus depurator, Linnaeus 1758; Hopkin et al. 2006; and Necora puber, Linnaeus 1767; Hopkin et al. 2006). Such tolerance levels appear to explain the distribution of T. danae on the more oceanic east coast of Hong Kong (Lai 1996, see also Fig. 1), but not the estuarine west coast where salinity can drop to 5.4‰ during the summer due to monsoon rains and freshwater discharge from the Pearl River (Morton and Morton 1983; EPD 2022). The invasive mussel Xenostrobus securis, in contrast, is euryhaline and has been found in both eastern and western waters of Hong Kong (Astudillo 2015; Morton and Leung 2015; Lau et al. 2018). Given these environmental conditions, T. danae does not appear to be an appropriate biological control agent for X. securis in estuarine environments due to its low survival.

In contrast to previous studies on other crab species, thermal tolerance limits of T. danae as measured by ABT and FLT did not decline when salinity decreased from 35 to 15‰. In contrast, Tagatz (1969) found that hypo-osmotic stress significantly reduced thermal tolerance of the blue crab Callinectes sapidus Rathbun 1896, which is less tolerant to high temperatures when exposed to water at 6.8‰ as compared to 34‰. Such trade-off was apparently driven by the increased energy demand for osmoregulation under hypo-osmotic conditions (Todd and Dehnel 1960; Taylor et al. 1977), which reduces the physiological ability to cope with thermal stress and eventually leads to lower thermal tolerances (Todd and Dehnel 1960; Rome et al. 2005). This trade-off, however, does not seem to occur in T. danae, possibly because of differences in energy allocation. The heart rates of the crabs reduced when they were exposed to low salinity (15‰), and this reduction in metabolism may conserve energy to support physiological functions at high temperatures without compromising the crab’s thermal limits. Such a trait is potentially important for intertidal and shallow water species such as T. danae, where environmental fluctuations in salinity can be rapid and contingent upon local weather conditions. As such, being able to adjust physiological rates and energy allocation under such variable conditions may offer an adaptive advantage when compared to pelagic and/ or stenohaline conditions.

When exposed to low salinities, T. danae demonstrated a unique heart rate profile under an acute thermal ramp (heart rate dropped and increased again before the final breakpoint, in contrast to the typical unimodal profiles for crabs in 25 and 35‰). Such depressed heart rates upon thermal stress might result from metabolic depression which has been reported in the littorinid, Echinolittorina malaccana (Philippi, 1847), oyster, Isognomon nucleus (Lamarck 1819) and decapods such as Austrothelphusa transversa (von Martens 1868) (MacMillen nd Greenaway 1978; Alekseev and Starobogatov 1996; Marshall et al. 2011; Hui et al. 2020). In those cases the suppression of aerobic metabolism has been suggested to minimize energy expenditure during or when anticipating adverse environmental conditions (Marshall and McQuaid 1991; Guppy and Withers 1999). In the case of T. danae, heart rate was depressed at low salinity (15‰). Whilst hyposaline stress may directly impact circulatory functions leading to bradycardia (Uglow 1973; McGaw and McMahon 1996), energy conservation may offer a possible explanation for such metabolic depression at hyposaline conditions (Guppy and Withers 1999), as short term fluctuations in salinity are common in the intertidal and subtidal zones (Firth and Williams 2009; Williams et al. 2011).

Although heart rates of the crabs increased with temperature in air, handling time was longer for crabs feeding in warm (28 °C) as compared to cooler water (22 °C). The decline in activity rate at 28 °C may indicate the behavioural thermal limits of the crabs, which are ~ 7 °C lower than the mean ABT at 34.6 °C in T. danae. The decoupling of physiological performance and behaviour could be driven by the behavioural partitioning of the crabs that feed primarily when immersed by the tide and become less active during tidal emersion. In intertidal gastropods such partitioning across tidal phases has been suggested to result in a decoupling of thermal performance curves, where behavioural performance has lower thermal limits and peak at temperatures close to seawater temperature (Monaco et al. 2007; Marshall et al. 2011). The consumption rates of the crabs, however, remained similar between 22 and 28 °C, suggesting that despite the differences in handling time the crabs are consuming a similar number of prey to fulfill their energy demands under different water temperatures. Water temperature, therefore, appears to play a less important role than salinity in determining the predation pressure of the crabs on the mussels.

Prey preference of Thalamita danae

Vulnerability to crab predation was similar between the invasive and native mussels despite the significantly longer handling time for consuming the native mussel. The variation in handling time may be explained by a substantial difference in the shell morphology of the two species; the invasive mussel having a thinner subcylindrical shell (average shell thickness index = 0.59 mm) with straight dorsal and ventral margins (Astudillo et al. 2018; Garci et al. 2007), while the native mussel has a thicker triangular shell (shell thickness index = 0.72 mm) with a globular and inflated posterior margin (Astudillo et al. 2018; Morton et al. 2020, see Fig. S1). The crab, therefore, should need a greater compressive force to break the native mussel shells (Boulding 1984; Dudas et al. 2005). The round shell shape also makes handling of native mussels more difficult as they can easily slip from the crabs’ chelipeds (Dudas et al. 2005). As a consequence, the crab may shift from the common shell crushing method to the more time-consuming edge chipping method when feeding on native mussels (Fig. S5, Elner and Hughes 1978; Hughes and Seed 1981; Seed and Hughes 1995) and, therefore, the profitability of preying on the invasive mussel will be higher than the native mussel.

Previous studies indicated that crab size is another critical factor affecting the selection of prey items (Elner and Hughes 1978; Hughes and Seed 1981). The chela of larger crabs is bigger and more powerful, allowing them to open more resistant prey in a shorter time than smaller crabs (Hughes and Seed 1981; Seed and Hughes 1995). Hence, the size of mussels consumed should increase with crab size (Enderlein et al. 2003). There was, however, no crab size effect on mussel consumption or breaking time. The effects of crab size were only detected in their feeding time, where smaller crabs spent more time eating native mussels. This finding may be explained by the accumulation of shell fragments in the foregut (Hill 1976; Ap Rheinallt and Hughes 1985), which were ingested accidentally during mussel consumption. The clearance of such shell fragments is very slow, and the foregut of small crabs fills up quickly when feeding on prey with thick shells (Hill 1976). The average shell thickness index of the native mussel is 0.13 mm greater than that of the invasive mussel (Astudillo et al. 2018). More time, therefore, will be required for small crabs to eat the entire native mussel. This effect was enhanced when the crabs were exposed to high temperature or low salinity when food digestion poses extra physiological demands to the crab (McGaw 2006a, b). These demands can be fulfilled by increasing air uptake which, however, facilitates the exchange of water and ions, resulting in a high death rate (McGaw 2005, 2006b). To avoid mortality, the metabolic performance of crabs declines by reducing respiration and cardiac output when exposed to high temperature or low salinity, leading to a decrease in the foregut clearance rate (Draper et al. 2022).

Prey size also contributes to mussel selection of crabs. Seed and Hughes (1995), in a laboratory experiment, showed that crabs exhibit a strong preference for small mussels (10–15 mm), and more than one-third of the tested crabs failed to feed on mussels > 20 mm (Seed and Hughes 1995), probably due to long handling times and thus reduced profitability (Elner and Hughes 1978; Enderlein et al. 2003). Although mussel size was standardized in our experiments, on the shore, mussel consumption by crabs could be substantially influenced by prey size due to the co-existence of various-sized mussels (Elner and Hughes 1978; Huges and Seed 1981). The mean total length of the invasive X. securis (16.8 mm) is larger than the native Brachidontes variabilis (14.0 mm) at the Island House site (Fig. S2), which may allow for size escapes of the invasive species and result in greater consumption of the smaller native mussels by the crabs if prey size overrides species identity in determining profitability.

Apart from the prey species identity and predator size, anti-predatory responses of mussels, like byssal attachment and aggregation, will also influence predators’ prey selection and consumption on the shore. Cheung et al. (2004) found that the length and number of byssus threads increase under predation risk, which will enhance the mussels attachment strength and reduce the chance of them being removed from the substratum (Lin 1991). Reimer and Tedengren (1997) also reported that mussels growing in a clump have, overall, less exposed surface area which limits the handling strategies used by crabs, consequently reducing the efficiency of mussel consumption. The formation of clumps, however, depends heavily on the density of mussels (Kobak 2001; Bertolini et al. 2017). A high density of mussels will result in a complex aggregation with better protection against predation (Bertolini et al. 2017). The native B. variabilis and invasive X. securis exist in mixed patches and form multi-layered reefs with the local oyster Saccostrea cuccullata (Born 1778) at the study site (Fig. S6). Such mixed aggregations may further complicate the clump structure and alter predation by the crabs. Although we were unable to directly measure predation by T. danae on the two mussel species on the shore, we adopted a comparative approach to assess the handling time of crabs on each mussel species following Hughes (1979). Quantifying differences in handling time is key in determining the variation in profitability among prey items, an important consideration when predicting the diet of the crabs. The shorter handling time required and higher profitability of X. securis suggest that this mussel is likely to be one of the prey items of the crab. This paper, therefore, provides estimates of foraging parameters under a simplified scenario. Other location-specific information such as predator and prey densities, prey recognition, and consumption of aggregated individuals must, however, be further considered when extrapolating our laboratory results and considering using T. danae as a native biological control agent on the shore.

Predation under environmental stresses

The decline in predation rate when the crabs were exposed to low salinity (15‰) is consistent with previous data (Taylor et al. 1977; McGaw and McMahon 1996; Witman and Grange 1998). The substantial reduction in consumption and survival rate under hypo-osmotic stress lowers the potential for the crab to control the invasive mussel populations. Feeding and food processing pose additional physiological demands on crabs under hypo-osmotic conditions (McGaw 2006a, b). Increasing oxygen uptake by more frequent respiration to support these demands, however, can lead to higher rates of water and ion exchanges across respiratory surface which may be lethal in osmo-conforming crabs under hypo-osmotic conditions (McGaw 2005, 2006b).

Such conflicting processes between feeding and stress responses have been shown to be mediated by hormonal controls. Curtis et al. (2010) reported that hypo-osmotic stress stimulates the sinus gland in crabs to produce a putative hormone, feeding inhibition factor (FIF), which suppresses hunger and stimulates satiety, resulting in a loss of appetite (Curtis et al. 2010). In addition, hypo-osmotic stress modulates the stomatogastric ganglion, which lowers the peristalsis rate in the foregut and stimulates food regurgitation, in order to lessen food materials entering the midgut (Morris and Maynard 1970; Rezer and Moulins 1983). These stress responses substantially decrease food intake (McGaw 2006a; Curtis et al. 2010), and when coupled with the reduced survival of the crabs in hypo-osmotic conditions, further reduce the feasibility of the crab to be a biological control agent for the invasive mussels in estuarine environments. The invasive X. securis is distributed widely from the eastern oceanic zone to the western estuarine zone of Hong Kong. Both Tolo Harbour and Deep Bay have a similar range of water temperatures (17.5 to 31.0 °C) but lower salinities, however, are recorded in Deep Bay with an average of 17.7‰ between 2018 and 2022, as compared to Tolo Harbour, which has a mean level of 30.5‰ (EPD 2022). Due to the high freshwater discharge from the Pearl River, the salinity of inner Deep Bay even drops to below 10‰ during the wet season (EPD 2022). As such, the crab could be used as a biological control agent for the invasive mussels currently present in Tolo Harbour and other coastal habitats with normal salinities, but not in Deep Bay where salinity is low and beyond the tolerance limits of the crabs.

Although temperature and salinity are the main driving forces for physiological responses in intertidal animals, a variety of environmental factors, like water flow and habitat heterogeneity, also have significant influences on crab predation (for instance by influencing searching time, Sponaugle and Lawton 1990). Whilst this study has examined the joint influence of temperature and salinity stress in the laboratory, further research should be undertaken to test the predation of crabs under natural conditions in the field, in order to verify our experimental findings and the feasibility of using T. danae as a native biological control agent.

Conclusion

The predatory crab Thalamita danae exerts similar predation pressure on the invasive mussel, Xenostrobus securis and the native mussel, Brachidontes variabilis of the same size, suggesting that the crab may deter the establishment of X. securis in Hong Kong. Variations in the physical environment, however, strongly affect the predation pressure of the crabs on the invasive mussels, with survival and consumption rate of the crabs significantly decreasing when exposed to low salinity. Hence, although the invasive mussels may be preyed upon by T. danae in regions with water salinities above 15‰, both this and previous studies show that local predators (swimming crabs and whelks) are unable to effectively prey on the invasive mussels in more estuarine environments (Astudillo et al. 2018). There is, therefore, a pressing need to investigate the biological control potential of other predators in estuarine and freshwater habitats to manage invasive mussel populations in these environments.