Functional responses of male and female European green crabs suggest potential sex-specific impacts of invasion

Animal collection

We collected European green crabs from Bedwell Bay (49.30919, −125.80489) on the west coast of Vancouver Island, British Columbia (BC), Canada, in June 2022 with the Coastal Restoration Society. Males and non-gravid intermoult females, with a notch-to-notch carapace width of 50–76 mm, free from epibionts and with two intact claws, were collected and transported to the Bamfield Marine Science Centre (BMSC), BC.

On the Pacific coast of North America, European green crabs are generalist, omnivorous predators that consume predominantly mollusks and crustaceans (Jamieson et al., 1998; Klassen & Locke, 2007). As prey species, we therefore used one of the commonest clam species in the region, i.e., the varnish clam (Nuttallia obscurata), which co-occurs with green crabs in many areas (S Dudas, pers. comm., 2022). Although these clams are not native to BC, their abundance and physical characteristics (i.e., thin shells) make them a preferred prey species for local green crabs (Molnar et al., 2008; Curtis et al., 2012). It is important to note that our results, which are based on a single prey species, could be unrepresentative of predation on other types of prey, especially when predation is not affected by physical differences between male and female crabs. Undamaged varnish clams of 25–40 mm in length (i.e., anterior to posterior edge) were collected from Robbers Passage, Barkley Sound, BC (48.89622, −125.12081).

Both crabs and clams were housed separately in indoor sea-tables supplied by flow-through unfiltered seawater at the BMSC. Female and male crabs were kept separately (∼20 crabs per sea-table) and provided with shelter, rocks, and seaweed. Female green crabs were inspected daily to ensure they did not develop eggs, to prevent introduction into Bamfield Inlet. Crabs were fed every 3–4 days with salmon and clams were fed every three days with concentrated phytoplankton (PhytoFeast). Both species were held under artificial lighting that mimicked the natural day–night cycle.

Experimental set-up

We conducted functional response experiments in June and July 2022. Prior to the start of a trial, we selected up to 12 crabs haphazardly from the holding sea-tables, isolated them, and withheld food for 48 h to standardise hunger levels (Howard et al., 2018; Howard et al., 2022). Trials with female crabs were run first, followed by trials with males to minimise the time female crabs were kept in captivity and thus the risk of accidental introduction of green crabs into Bamfield Inlet. Female green crabs (n = 36) were held for 5 to 9 days, and male green crabs (n = 36) were held for 15 to 19 days, both including starvation time. Female carapace width ranged from 50.0–76.0 mm (mean ± SD, 64.9 mm ± 5.6 mm) and males ranged from 56.0–75.0 mm (67.1 mm ± 5.0 mm). Crusher claw height for females ranged from 8.0–18.0 mm (13.8 mm ± 2.4 mm) and 12.0–25.0 mm (17.4 mm ± 3.1 mm) for males. Opaque plastic enclosures (68.1 L, 61 × 41 × 42 cm) were used for individual trials and each was supplied with an independent source of constant flowing seawater and artificial lighting. During the course of the experiment, we stopped foot traffic around the enclosures to prevent disruptions to crab foraging.

We randomly selected varnish clams from the holding sea-tables, measuring shell length and height (umbo to mouth) before placing them in the enclosures at one of six densities (1, 2, 4, 6, 10, or 16 clams per enclosure) 12 h prior to the beginning of trials. Each density by sex combination was replicated six times (n = 72). A control replicate with clams but without a crab was run with one of each of the six densities. We recorded the sex, carapace width and crusher claw height of each starved crab before introducing it to a haphazardly chosen enclosure. Crabs were allowed to feed on the clams for 8 h, starting at 08:00. At the end of a trial, crabs were removed from the enclosures, and the number and size (height and length) of clams remaining were recorded. By comparing post-trial sizes to the relevant pre-trial sizes, we were able to determine the number and sizes of clams consumed in each trial. We conducted up to 12 trials per day, which ran simultaneously. Crabs were only used once and euthanized by freezing at the end of each trial following BMSC animal care protocols. Individual clams were used only once, even if they were not consumed. Temperature and salinity were measured at the start and end of each 8-h trial using a thermometer (Fisherbrand^© 76 mm immersion thermometer) and refractometer (Tropic Eden^© PRO-1 normal seawater refractometer), respectively. Seawater temperature remained constant throughout the experimental period (10 °C ± 0.33 °C) and salinity ranged from 33–37 ppt.

Carapace and claw dimorphism

To assess sexual dimorphism, European green crabs were opportunistically collected and measured in August 2022 from the Tranquil River Estuary (49.207, −125.671), ∼35 km from Bedwell Bay, with help from the Coastal Restoration Society. Crabs were collected from 40 traps that were deployed for ∼24 h; of the 2,184 crabs caught, we measured 372 chosen haphazardly. Each crab was sexed and its carapace width and crusher claw height measured using calipers in the field. Forty of the crabs (n = 20 males, n = 20 females) were brought back alive to BMSC for further behavioural experiments (see below); the rest were euthanized as part of the South Coast European Green Crab Control Project.

Exploratory behaviour

We examined sex differences in exploratory behaviour by observing crabs as they moved around enclosures. Male and non-gravid female green crabs with both claws intact, a notch-to-notch carapace width between 55–75 mm, and no evidence of moulting or epibionts were collected from the Tranquil River Estuary (see above) and transported to BMSC. Crabs were housed in conditions identical to those of the functional response experiment and were starved for 48 h prior to the behavioural observations.

Exploratory behaviour experiments ran from 08:00 to noon. We suspended one GoPro Hero3 camera 40 cm above each of three enclosures (68.1 L, 61 × 41 × 42 cm), which had no substrate, cover, or flowing water. The enclosures were placed outdoors in the shade to record a clear video without glare from artificial lighting or moving water. The sex, carapace width, and crusher claw height of each crab were recorded. A dot was drawn with a permanent marker on the centre of each crab’s carapace as a reference point for tracking movement. Video recording started immediately after a crab was placed in each enclosure and three trials were run at once. We recorded 5-min videos of 16 male crabs and 16 female crabs.

For each crab, we calculated path length, time spent moving, and average moving speed. Path length was analysed using the software Tracker (V 6.0.9, 2022) by placing a tracking point on the focal crab’s carapace every 35 frames. The time spent moving was measured with a stopwatch and converted to proportion of overall observation time. Average moving speed was calculated as the path length divided by time spent moving. Because of technical issues with video recording, the sample size for females is variable (n = 12 for path length, n = 13 for proportion of time spent moving, n = 11 for average speed).

Functional response

We assessed functional response curves using a logistic regression that fit the proportion of prey eaten to prey density with the R package “frair” (Pritchard et al., 2017). Both male and female green crabs exhibited a Type II functional response (see Results). We therefore fit the data using the Rogers’ Type II model equation $N_e=N_0\left(1-e^{\left(a\left(N_{e}h-T\right)\right)}\right)$), which is the random predator equation without prey replacement (Rogers, 1972), where N_e is the number of prey consumed, N₀ is the initial prey density, a is the attack rate, h is the handling time, and T is the experimental duration (fixed to 1 for model fitting). Using AIC values corrected for small sample sizes and friar::frair_fit with maximum likelihood estimation, we confirmed that the Type II fit was best. To generate 95% confidence intervals of model parameters and to test the differences between male and female green crabs at the highest experimental density of clams (i.e.,  N₀ = 16), we used frair:frair_boot nonparametric stratified bootstrapping (n = 2,000 iterations). For each iteration of the bootstrap, we took the difference between the estimated N_e for males and for females at N₀ = 16, then examined the resulting distribution of differences. The 2.5% and 97.5% quantiles of this distribution represent the lower and upper bounds of the 95% confidence interval around the difference. If the confidence interval includes 0, the difference is considered non-significant.

We calculated the functional response ratio (FRR, i.e., attack rate/handling time) for each sex for each iteration of the bootstrap. The functional response ratio aims to integrate the various metrics of functional response experiments to predict a species’ ecological impacts, which is useful when dealing with invading species (Cuthbert et al., 2019). To determine if the FRR varied significantly between the sexes, we calculated FRR_male:FRR_female for each iteration and then examined the resulting distribution of ratios. The 2.5% and 97.5% quantiles of this distribution represent the lower and upper bounds of the 95% confidence interval around the ratio. If the confidence interval includes 0, the difference between males and females is considered non-significant.

Predator morphology

We examined the effect of crab size on predation in the functional response experiment using a generalised linear model with a binomial distribution and logit link function, with the proportion of clams eaten as a response variable, and initial prey density, sex, crusher claw height, and a sex:claw height interaction as fixed effects. We also separately ran the same model with carapace width instead of crusher claw height, as the two predictors were too correlated to be included in the same model. We then ran linear models to test for differences in mean carapace width and crusher claw height between male and female crabs in crabs from our experimental trials and from the greater population survey at Tranquil River Estuary.

Exploratory behaviour

We constructed linear models to examine differences in path length, proportion of time moving, and speed between the sexes, with carapace width included as a fixed effect and an interaction between sex and carapace width.

Modelling population-level estimates of consumption efficiency

To examine the potential bias in estimation of consumption efficiency (i.e., proportion of clams eaten) when considering only male crabs relative to varying proportions of female crabs in a population, we used the generalised linear model described in the Predator morphology section and bootstrapped estimates of male and female consumption efficiency to generate confidence intervals around the proportional differences. For each iteration of the bootstrap, we examined the model predictions for the proportion of clams consumed by an average sized female (i.e., based on the mean crusher claw height observed at Tranquil River Estuary; 11.48 mm, 95% CI: 11.15 mm–11.81 mm, n = 197) and an average sized male (mean crusher claw height: 16.71, 95% CI: 16.19 mm–17.23 mm, n = 175). We then calculated the proportional difference between each of these estimates and the original model predictions for an average sized male (e.g.,  $\frac{propclamsconsume{d}_{femalecrab,bootstrapiteration1}-propclamsconsume{d}_{malecrab,originalmodel}}{propclamsconsume{d}_{malecrab,originalmodel}}$). We then simulated a series of hypothetical populations ranging in sex ratios from entirely male to entirely female. For each iteration of the bootstrap, we calculated the difference between an all-male population using the original model, and the hypothetical mixed-sex population using the bootstrapped iteration of the model. We took the 2.5% and 97.5% quantiles of the resulting distribution as the lower and upper bounds of the 95% confidence interval around these estimates.

For all analyses, we used R (Version 4.1.2; R Core Team, 2020). For all (generalised) linear models presented in this paper, model fit was assessed using the ‘DHARMa’ package (Hartig, 2022), which includes tests for assessing the distribution of residuals, dispersion, outliers, and homogeneity of variance.

Predator morphology

The proportion of clams consumed by green crabs decreased with increasing prey density (coefficient = −0.76, p < 0.001). Female green crab consumption rates increased with increasing crusher claw height (coefficient = 0.59, p = 0.006), while crusher claw height had no effect on male green crab consumption rates (interaction between sex and crusher claw height = −0.57, p = 0.04; Fig. 3). There was no effect of green crab sex on the proportion of clams consumed at mean crusher claw height (coefficient = 0.29, p = 0.24). The results were qualitatively similar when we considered carapace width instead of crusher claw height, although in that case, there was a significant effect of sex (coefficient = 0.46, p = 0.03), with males consuming a greater proportion of clams than females at the mean carapace width, but no interaction between carapace width and sex (coefficient = −0.32, p = 0.16) (Fig. S2).

For the green crabs used in the functional response experiment, there was no significant difference in carapace width between the two sexes, although male green crabs tended to be slightly larger (2.22 mm, or 3.4%, p = 0.078) than females. This is not surprising since we selected crabs from a limited size range. However, male green crabs had significantly larger crusher claws (3.58 mm, or 25.9%, p < 0.001). The differences in carapace and crusher claw sizes observed in our experiment were magnified in the larger sample (372 green crabs: 197 females and 175 males) with unrestricted size range from the Tranquil River. Male carapace was, on average, 16% larger (9.26 mm, p < 0.001), and crusher claws 45% larger (5.22 mm, p < 0.001) than those of females.

Prey morphology

At the beginning of the trials, males were offered slightly smaller clams on average compared to female crabs (i.e., 2.85 mm or 8.5%) as a result of our random assignment of prey (post-hoc pairwise contrast p < 0.0001; Fig. 4). Neither male nor female crabs preferentially consumed smaller or larger prey than they were offered (post-hoc pairwise contrasts p = 0.17 for females, p = 0.72 for males; Fig. 4).

Predator exploratory behaviour

There were no detectable differences in exploration behaviour between male and female green crabs. Path length did not vary between the sexes (coefficient = 760.38, p = 0.74; Fig. 5A) or with carapace width (coefficient = −12.74, p = 0.66), and there was no interaction between sex and carapace width (coefficient = −8.51, p = 0.81). Male and female green crabs were in movement for a similar proportion of time (coefficient = −0.38, p = 0.81; Fig. 5B), with no effect of carapace width (coefficient = −0.03, p = 0.21) or interaction between predictors (coefficient = 0.01, p = 0.71). Similarly, there was no difference in speed between female and male green crabs (coefficient = 1.41, p = 0.84; Fig. 5C), as well as no effect of carapace width (coefficient = 0.03, p = 0.75) or interaction between sex and carapace width (coefficient = −0.02, p = 0.83). The results were similar when we considered crusher claw height instead of carapace width.