Assessing the size at maturity, spawning, and condition of the truncate soft-shell clam (Mya truncata) of southern Baffin Island, Nunavut, Canada

Study sites

Sampling sites were chosen through consultation with local Hunters and Trappers Associations to establish where communities collect clams during the ice-free season. M. truncata were collected at low tide from the intertidal zone at two sites within inner Frobisher Bay (hereafter FB), near the community of Iqaluit and two sites on the north shore of the Hudson Strait (hereafter HS) (i.e., south coast of Baffin Island), near the community of Kimmirut (Fig. 1). One site near Iqaluit was accessible by road (Site I1), while the other was roughly 15 km across FB and only accessible by boat (Site I2). Similarly, one site near Kimmirut was accessible by road (Site K1), and the other was located roughly 15 km from the community into the Hudson Strait (Site K2). The substrate type of the intertidal zone at all sites was sand with sparsely distributed small and large rocks (10–30 cm diameter) often with attached fucoid macroalgae. The sediment type within Frobisher Bay has been further classified as sandy mud, with increasing gravelly sandy-mud with increased proximity to the community of Iqaluit (Deering et al., 2018). No published account of the sediment classification near Kimmirut was found, but the sediment at the study sites was observed to be similar to sites near Iqaluit. In FB, 99 clams were collected from 28–30 August 2018, with weather requiring three visits to collect clams from the two FB sites. In HS, 247 clams were collected from two sites on 12 September 2018. These sample dates coincided with the new moon phase when maximum tidal amplitudes of 10.4 m and 12.1 m occurred within Iqaluit and Kimmirut coastal waters, respectively. Clams were buried 10–30 cm in the sediment and were dug out from the lower intertidal zone using garden trowels. Community members indicated clams are concentrated in the lower intertidal zone to apparently avoid ice scour during the winter months. All clams encountered in the sediment were collected, with no discrimination for size. All clams were kept in cool sea water during transport back to the community where they were immediately frozen and maintained at −20 °C for subsequent analysis.

Sex determination and gonadal staging

In the laboratory, frozen clams were thawed, and a Vernier caliper was used to measure shell length (SL; ±1.0 mm) from the anterior tip of the shell to the posterior edge. Clams were gently pried open, and a scalpel was used to sever the adductor muscles at the point of attachment to the shell. The entire animal, including the adductor muscles, was then removed from the shell. When present, a negligible sample (<0.001 g) of gonadal tissue was removed to determine sex and stage of gonadal development. Subsequently, the body of each clam, without the negligible sample of gonadal tissue, was individually dried to constant weight (±0.0001 g) at 60 °C. This analysis provided the dry body weight (DW) of each clam.

Clams that did not exhibit gonadal development (i.e., no gonadal tissue visible on the clam’s body) could not be sexed and were classified as undifferentiated and considered immature. Gonadal smears were examined at 10–40 × magnification and used to assign each clam to one of five stages of gonadal development based on the scale defined by Brousseau (1987): (1) indifferent, (2) developing, (3) ripe, (4) spawning (partially spent), and (5) spent. To aid in classification, gonadal smears of M. truncata were also compared to photographs of histological preparations of gonadal tissue of M. arenaria from Ireland (Cross et al., 2012; Cross, 2011). In the current study, M. truncata classified as indifferent possessed gonadal follicles, but no lumen or cavity was present, and no developing egg or sperm cells were found in at least three gonadal smears examined for each clam. As such, clams classified as indifferent were considered to be immature and lumped with clams classified as undifferentiated. Clams with gonads that exhibited follicular lumen and possessed egg or sperm cells developing on the periphery of the lumen were classified as developing. Ripe clams displayed more fully developed egg or sperm cells with tails and many of these gametes were free within the follicular lumen. Although many of the egg and sperm cells had been discharged in clams classified as spawning, there were still some free gametes within the follicular lumen. In spent clams, the follicular lumen was empty and there were no developing eggs or sperm on the periphery of the lumen. None of the clams examined during this study were classified as spent.

Maturity ogive

Separate male and female size at maturity ogives were constructed for clams collected across all sites by quantifying the proportion of male or female clams that were sexually mature within each five mm SL class. The sex ratio of sexually mature clams was used to estimate the sex of immature clams (i.e., undifferentiated and indifferent stages combined). Specifically, the 1.14:1 female: male sex ratio was used wherein 53% of all immature clams within each five mm SL class were deemed female and 47% were deemed male. The best fit model for each maturity ogive was selected using AICc, where models tested included binomial (logit link and probit link), Gompertz, and Weibull distributions.

Shell length analysis by site, sex, and maturity

A one-way analysis of variance (ANOVA) was used to examine variation in M. truncata SL among all four sample sites (i.e., two in each of FB and HS). Tests of normality and equality of variance were performed using the Shapiro–Wilk’s normality test and the Levene median test, respectively. When assumptions of normality and equality of variance could not be met by transformation, the non-parametric Kruskal–Wallis ANOVA on ranks was used and Dunn’s post-hoc test if the result was significant. Tukey’s HSD post-hoc test was used when parametric analysis indicated significant differences.

A linear model was used to examine variation in SL among immature clams and mature male and mature female clams by community (i.e., FB and HS) using the following equation: (1)${\displaystyle {\mathrm{SL}}_{\mathrm{i}}={\mathrm{\beta }}_0+{\mathrm{\beta }}_1\mathrm{Community}\ast {\mathrm{\beta }}_2\mathrm{Sex}+\varepsilon .}$ Here, β₀ represents the intercept, β₁Community is the community where the clam was collected, either FB or HS, β₂Sex is the sex of the clam, either immature, male, or female, and ɛ is the error term. Model assumptions were validated by testing for patterns in residuals of model covariates. Tukey’s HSD post-hoc test was performed on the model to make pairwise comparisons among SL at sex at each site.

Length-based condition indices

A DW condition index was calculated for M. truncata using the following equation: (2)${\displaystyle {\mathrm{CI}}_{\mathrm{DW}}=\left(\mathrm{DW}÷\mathrm{SL}\right)\times 1000.}$ This equation multiplies (DW ÷ SL) by 1000 to avoid working with values <1. This dry weight condition index was used as opposed to wet weight indices as recommended by previous bivalve studies to avoid accounting for water content in condition calculations (Higgins, 1938; Mo & Nielson, 1992). As fish deplete their energy reserves, water content increases in muscle tissue and energy storage organs to replace lost protein and lipids, making dry weight condition indices the most accurate way to infer condition (Lucas & Beninger, 1985; Grant & Brown, 1999). Fulton’s condition factor (Ricker, 1975) was not used in this study because it is known to be length biased in the estimation of average condition in fish (Rennie & Verdon, 2008). To evaluate SL-related bias in the condition index, data from FB and HS were pooled and plotted against SL. The most suitable regression model for this relationship was selected using Akaike’s second order information criterion (AIC_c; Burnham & Anderson, 2002), where models tested included linear, quadratic, exponential, and cubic distributions, and model parameters and level of significance were calculated. Length-related bias was defined by a coefficient of determination (R²) > 0.8, and significant p-value in the linear model, which is similar to the method used by studies that examine relative weights using standard weights (Gerow, Hubert & Anderson-Sprecher, 2004).

Weight-shell length relationships

In the case of length-bias in CI_DW, DW-SL relationships were developed for M. truncata collected from each site in FB and HS. The most suitable regression model for each relationship was selected using AIC_c. Analyses of Covariance (ANCOVAs), designed to test for interactions between variables, were performed between populations in FB and HS to test for differences in relationships. When an ANCOVA indicated a significant difference in weight-SL relationships between populations, the data were plotted with the 95% confidence intervals (CIs) to identify where the difference occurred. Sites were compared within communities, i.e., sites I1 vs. I2 and K1 vs. K2, and between communities, i.e., I1 vs. K2 and I2 vs. K2, restricting the within and between community comparisons to populations that exhibited similar size ranges (i.e., K1 clams did not exceed 47 mm SL and were removed from the between community comparisons).

Data cleaning and software

A total of 8 outliers, 1 from site I1 (34 mm in SL, ripe female), 4 from site K1 (all<17 mm SL and undifferentiated), and 3 from site K2 (all<24 mm SL and undifferentiated), all with weights outside the normal range of clams collected, were removed from the data set by testing linear regressions of DW vs. log SL using the Bonferroni Outlier Test, which detects outliers that have the ability to shift the mean (Doornbos, 1981). One clam collected from Iqaluit was removed from analysis as it was 80 mm, and outside the normal SL range of clams collected. All statistical analyses were performed in R version 3.5.1 for windows (R Core Team, 2018) and the significance level was set to α = 0.05. This study was reviewed and approved by Memorial University’s Institutional Animal Care Committee (Project # 18-01-SG).

Shell length

Overall, 337 truncate soft-shell clams were examined: 115 male, 131 female, and 91 immature (i.e., undifferentiated (82) and indifferent (9) stages combined). There were more female truncate soft-shell clams collected at sites I1 and K1 while the sex ratios were equal or nearly equal at sites I2 and K2 (Table 1). Analysis indicated the sex of animals examined was independent of site (Chi-squared =3.77, df = 3, p = 0.29).

Shell length differed significantly among sites where the average shell length of clams collected at site K1(30 mm ± 1.2 SE) was significantly lower than site K2 (47.8 mm ± 1.2) and both sites in FB (I1 44.5 mm ± 1.6, I2 46.4 mm ± 0.6) (Tukey’s HSD all p < 0.001) (Fig. 2). Analysis revealed that the mean SL of immature clams did not differ significantly between FB and HS (32.5 mm ± 1.5, 24.4 mm ± 0.6, respectively) (Tukey’s HSD p = 0.09; Fig. 3). However, the mean SL of immature clams from each community was significantly lower than mature males and females from each community (p < 0.01). The mean SL of females from FB (50.8 mm ± 1.4) did not differ significantly from females (53 mm ± 1.4) or males (50.6 mm ± 1.3) from HS (p = 0.9, p = 1.0, respectively). Males from FB (43.6 mm ± 1.4) were significantly smaller than females from FB and males and females from HS (p = 0.02, p < 0.01, p < 0.01, respectively; Fig. 3).

Stage of gonadal development

In Frobisher Bay, 11.3% of the truncate soft-shell clams examined were classified as undifferentiated, 2.1% were in the indifferent phase of gonadal development, 8.2% were developing, 77.4% were ripe, 1.0% were spawning, and 0% were spent. In the Hudson Strait, 30.5% of clams were undifferentiated, 4.1% indifferent, 30.0% developing, 34.9% ripe, 0.5% spawning, and 0% were spent. Overall, ripe individuals accounted for 89.3% of the mature clams examined from FB and 53.5% of the mature clams from HS.

Maturity ogive

A logistic regression with a binomial distribution and logit link provided the best fit to both the male and female maturity data based on AICc scores (Table 2). This model is represented by the formula log(p/1-p) = α + βX where p represents the probability of being mature, X is shell length, and α and β are constants.

The majority (90.1%) of the 91 truncate soft-shell clams classified as immature did not possess gonads while 9.9% possessed gonads in the early stage of differentiation (i.e., indifferent). The immature clams exhibited a range in SL of 11–40 mm. In females, the SL at first attainment of sexual maturity was 28 mm, SL at 50% attainment of sexual maturity (L₅₀) was 32 mm, and all females exhibiting a SL ≥42 mm were sexually mature (Fig. 4A). For the males, SL at first attainment of sexual maturity was 23 mm, SL at L₅₀ was 31 mm, and all males exhibiting a SL ≥42 mm were sexually mature (Fig. 4C).

Body condition

The exponential model provided the best fit to the pooled CI_DW-SL relationship (Table 3). Pooled data across all sites showed significant positive relationships and a high coefficient of determination between CI_DW and SL (n = 337, R² = 0.837, p < 2.2⁻¹⁶), precluding its utility to compare average condition among the populations studied (Gerow, Hubert & Anderson-Sprecher, 2004; Rennie & Verdon, 2008).

The exponential model was the best model to use for comparisons of the DW-SL relationships between sites within FB and HS (Table 4), and all DW-SL relationships were highly correlated (Table 5). ANCOVA revealed the only significant difference in DW-SL relationships occurred between sites in FB (slope t-value =2.965, p = 0.004, intercept t-value = −1.864, p = 0.070). Plots of the line of best fit with 95% CIs indicate clams within the 49 to 63 mm SL range from site I2 exhibited a significantly higher dry weight at SL than clams from site I1 (Fig. 5).

Among the two FB sites and site K2, ANCOVA analysis found no significant differences between DW-SL relationships between site I1 and K2 (slope t-value =−0.243, p = 0.808, intercept t-value = −0.142, p = 0.887). However, subsequent pair-wise comparisons between the two sites in FB and site K2 revealed that the DW-SL relationships of sites I2 and K2 differed significantly (Estimate = 0.600, S.E. = 0.120, df = 224, t-ratio = 4.99, p < 0.001). Figure 6 shows that clams ≥49 mm SL from site I2 had higher DW than similar sized clams from site K2.