Abstract
Early life stages of fish are characterized by high size-selective mortality rates, with selection generally acting against smaller, slow-growing individuals. Here, we investigate, for the St. Lawrence River striped bass (Morone saxatilis) population, how size of young-of-the-year juveniles (YOYs) affected survival from the pre-wintering period until the following spring, by comparing their otolith daily growth trajectory to that of one-year-old juveniles (OYOs). Otolith growth in the first 50 d after hatch was faster in post- than in pre-winter juveniles in both years, indicating that fast-growing individuals were more likely to survive to the next spring. A larger back-calculated size at age 1 in the 2016 year class compared to that observed in 2017 also suggests interannual variability in size-selective overwinter survival. Our results indicate that the design of YOY abundance surveys aimed at predicting annual recruitment strength needs to account for the effect of size-dependent mortality until the end of the first winter of life, as high abundance of relatively small YOYs in autumn may not necessarily translate into a large number of OYOs in the following spring and thus into high recruitment.
Introduction
Fish populations are often subjected to wide fluctuations in abundance, a pattern primarily attributed to annual variability in year-class strength and recruitment success (Hjort, 1914; Houde, 2008). High and variable mortality rates in larvae and early juveniles (Anderson, 1988) have led to the consideration that these life stages constitute the primary recruitment bottleneck (Houde, 2008). Accordingly, the growth–survival paradigm (Anderson, 1988; Takasuka et al., 2017) states that faster-growing individuals within a cohort are more likely to survive the critical periods corresponding to the larval and early juvenile stages, and that strong year classes emerge when optimal environmental conditions facilitate fast growth for the majority of individuals composing a given cohort. A positive link between growth and survival is supported by three complementary mechanisms: (1) the “larval stage duration”, under which fast-growing larvae metamorphose at a younger age and thus experience a shorter larval stage with lower cumulative mortality (Houde, 1987; Leggett and Deblois, 1994); (2) the “bigger is better”, which states that larger body size at a given age is associated with reduced probability of predation mortality from organisms targeting relatively small prey (Miller et al., 1988); and (3) the “growth-selective predation”, by which faster growth at a given size generally reflects better physiological condition procuring a competitive advantage for escaping predators (Takasuka et al., 2003, 2017). Because these size-selective mechanisms target individuals through the period considered as the recruitment bottleneck, abundance indices of young-of-the-year juveniles (YOYs) sampled at the end of the first growing season are commonly used to monitor year-class strength (Bradford, 1992). Despite the importance of size-selective processes acting during the larval stage, a growing body of evidence suggests that the high and variable mortality rates responsible for year-class strength variability also extend into the juvenile stage (Fennie et al., 2020; Khamassi et al., 2020). This is especially true in temperate fish populations where mortality during the first winter may result in important decreases in YOY abundance (Sogard, 1997; Hurst, 2007). Massive winter mortality events are often size-selective and directed against the smallest individuals within the population, due partly to their higher susceptibility to predation, diseases, and energy depletion (Anderson, 1988; Sogard, 1997; Hurst, 2007).
Winter mortality in fish juveniles has been studied using different approaches, including controlled experiments (Post and Evans, 1989; Johnson and Evans, 1991; Hales Jr and Able, 2001), tracking of tagged individuals (Svåsand and Kristiansen, 1990; Willis et al., 1995), monitoring of length–frequency distributions over time (Hurst and Conover, 1998), and characterizing the survivors relative to the original population (Johnston et al., 2005; Robert et al., 2007; Khamassi et al., 2020), the latter comparing mean growth trajectories of the same year class before and after winter, using otolith daily increments among juveniles of the same annual cohort (Sogard, 1997). Because selection favouring large and fast-growing individuals is likely to occur throughout the harsh winter period, comparing the early growth trajectories or growth rate distributions of YOYs in a given year before winter to those of one-year-old juveniles (OYOs) of the same year class that have survived this critical period offers the potential to evaluate the level of winter-related selective mortality. In general, the comparison of growth rate frequency distributions between sequential life stages also allows for the investigation of the timing of the endpoint for the establishment of a given year class, which corresponds to the life stage when the magnitude and variability of mortality rates stabilize at relatively lower levels and thus year-class strength is mostly set (Robert et al., 2007; Khamassi et al., 2020).
The striped bass (Morone saxatilis) is an anadromous fish inhabiting estuaries and coastal waters along the Atlantic coast of North America, with the St. Lawrence River (SLR) striped bass population found at the northern limit of the species’ range (Scott and Crossman, 1973). Reproduction takes place during spring, when adults migrate upstream from coastal estuarine and marine waters (Scott and Crossman, 1973). Interannual variability in environmental conditions that characterize estuaries typically lead to strong variability in striped bass annual recruitment (Cooper and Polgar, 1981). The SLR striped bass population declined until its extirpation in the 1960s, due mainly to overfishing associated with the loss of juvenile habitats and pollution, during the construction of the St. Lawrence seaway (Robitaille et al., 2011). Following a successful recovery plan during which striped bass were reintroduced by stocking larvae and juveniles originating from the nearby southern Gulf of St. Lawrence population between 2002 and 2015 (Robitaille et al., 2011; Valiquette et al., 2017), natural reproduction was confirmed in Rivière du Sud near the city of Montmagny and within the Beauport Bay, in Québec City (Pelletier et al., 2011; Valiquette et al., 2017). After spawning in early June, partial migration following the larval stage has been documented with the occurrence of juvenile striped bass in a large area across the estuary (Morissette et al., 2016; Vanalderweireldt et al., 2019b).
The abundance of YOYs of the SLR striped bass population has been monitored annually since 2012 with a beach seine survey along the estuary in September (Valiquette et al., 2017; L'Italien et al., 2020). The YOY abundance survey is based on similar procedures implemented for other striped bass populations, such as in the Chesapeake Bay (Durell and Weedon, 2019; Gallagher et al., 2019) and the Hudson River (NYSDEC, 2020). An annually consistent broad range from 20 to 140 mm FL (fork length) in size distribution in September is observed within YOY SLR striped bass, but size variation is apparently much reduced in other Canadian populations (Conover, 1990; Douglas et al., 2006). The wide observed size range in juvenile striped bass suggests a potential for mortality during winter to act as a major selective force against smaller individuals (Sogard, 1997). Winter mortality could therefore constitute a significant bottleneck for the establishment of year-class strength in this population.
The main objective of this study is to characterize the strength of size-selective overwinter mortality by comparing the early growth trajectory of YOYs to that of OYOs in two consecutive year classes. To reach this objective, individual growth trajectories have been estimated from the examination of otolith daily increments of pre-wintering YOYs and of OYOs having survived through their first winter in the St. Lawrence estuary. Within a given year class, we hypothesized that OYOs were generally characterized by a faster mean early growth trajectory than pre-wintering YOYs. Such pattern would indicate size-selective winter mortality directed against smaller, slower-growing individuals. Our approach offers the opportunity to identify potential size-selective mortality related to winter, which would, in turn, allow for a better annual estimation of the proportion of YOYs that will eventually recruit to the population.
Material and methods
Study area
The St. Lawrence estuary is one of the largest estuaries in the world, connecting the Laurentian Great Lakes basin to the Gulf of St. Lawrence and the North Atlantic Ocean (Figure 1). It can be divided into three portions: (1) the fluvial estuary, a freshwater section influenced by tides; (2) the middle estuary, characterized by a wide salinity gradient ranging from 0 to 25 PSU and including an estuarine turbidity maximum (ETM); and (3) the maritime estuary, where salinity stabilizes around 30 PSU (St. Lawrence Centre, 1996). The habitat of YOY striped bass consists generally of shallow coastal areas located near the mouth of rivers draining into the fluvial and middle estuary (Valiquette et al., 2017). From the head of the estuary towards the lower end of the middle estuary, water temperature observed in summer and autumn decreases from 18–24 to 11–17°C, while salinity and dissolved oxygen increase, the latter from 5.5–8.2 to 8.2–10.5 mg l−1 (Vanalderweireldt et al., 2019b, 2020). In the ETM, located between the eastern point of Île d'Orléans and Isle-aux-Coudres, turbidity is the main varying factor and exhibits values up to 75 NTU, generally 5 to 10 times higher than in upstream and downstream areas of the estuary (Sirois and Dodson, 2000; Vanalderweireldt et al., 2019a, 2019b, 2020).
(a) Study area is located in the St. Lawrence estuary in eastern Canada, (b) at the fluvial and middle portions of the estuary. (c), (d) YOY striped bass were sampled with a beach seine in September 2016 and 2017. B.S. indicates beach seine stations where no YOY was captured. OYO striped bass from 2016 and 2017 year classes were captured between May and August of the next year, incidentally by commercial eel fishing gear, by scientific angling sampling near Île de Grâce, and by beach seine near Rivière Ouelle.
Sampling
YOY striped bass were sampled with a standardized beach seine (15-m long, 9.5-mm mesh at the wings, 6.3-mm mesh at the pocket, pulled from up to 1.4-m depths) at a total of 100 stations distributed along the fluvial and middle St. Lawrence estuary (Figure 1) in September 2016 and 2017, as part of the annual YOY abundance monitoring program carried out by the Ministère des Forêts, de la Faune et des Parcs du Québec (MFFP). A total of 179 and 158 YOYs were collected in 2016 and 2017, respectively. From these specimens, a random sub-sampling was done proportionally to the overall size–frequency distribution in each year to reflect otolith growth in YOYs (n = 49 and 48 in 2016 and 2017, respectively). As all specimens were measured after preservation in 95% ethanol, FL was corrected as FLfresh = 1.031 × FLethanol – 0.5 (G. Cervello, unpublished data).
OYOs were sampled between May and August 2017 and 2018, corresponding to the 2016 and 2017 year classes. They were collected from three different sources: (1) incidentally between June and August in commercial eel fishing gear distributed along the estuary; (2) during scientific angling sampling in August 2017 near Île de Grâce, upstream of the fluvial estuary (Figure 1c); and (3) from a small-scale beach seine survey conducted near Rivière Ouelle in May 2018 (Figure 1d). Eel fishing gear consisted of three fence traps with fixed nets of 19-mm mesh into which fish enter through a narrow slit and are then hampered from coming out. After analysing sizes of striped bass of all ages collected by this type of fishing gear (between 66 and 800 mm FL), we did not detect evidence of any size-based selection relative to the OYO sample used in our study (Supplementary Data S1). We have also identified no size-based selection in OYOs captured by angling relative to those captured by eel fishing gear at the same period of the year (Supplementary Data S2), nor among OYOs captured by beach seine (Supplementary Data S3). A total of 46 and 15 OYOs were captured in 2017 and 2018, respectively. These juveniles were measured fresh, except those from Île de Grâce that were measured after being preserved frozen; FL was corrected as FLfresh = 1.017 × FLfrozen– 0.52 (G. Cervello, unpublished data).
Laboratory analyses
Right sagittal otoliths were extracted and mounted on microscope slides with thermoplastic glue (CrystalbondTM 509; AremcoTM products, NY, USA). Otoliths larger than 1100–1200 μm (radius from the core towards the post-rostrum) had part of their rostrum ground to remove the typical curvature present in larger otoliths and expose more clearly the sagittal plane at the core and post-rostrum regions. Otoliths were then polished on the sagittal plane to expose daily increments, following the technique described by Secor et al. (1992). An abrasive sheet (1200 grit WetordryTM, 3MTM) and 5- and 1-μm lapping films (3MTM) were used on otoliths larger than 300–400 μm to improve readability. An optical microscope coupled to an image analysis system (LeicaTM DM2500 coupled with a DFC450 camera) was used to capture micrographs of the otoliths, at 50–400 × magnifications. Following the recommendations made by Campana (1992), we selected a measuring axis starting at the primordium towards the post-rostrum, crossing all daily increments at a 90° angle from the core (Supplementary Data S4). On that axis, the primordium radius and the width of each daily increment were measured for otoliths of both YOYs and OYOs. We assumed that the first increment is deposited at the age of 4 d, in accordance with previous studies (Houde and Morin, 1990; Vanalderweireldt et al., 2019b). We then calculated average daily increment width from ages 4 to 50 d, after which the presence of accessory primordia and growth on different axes made the measurements less reliable as proxies for fish length (Campana, 1992).
Otoliths of YOYs had their daily increments also counted beyond the otolith core to determine daily age and hatch date, although this procedure was not possible for OYOs due to the overlay of increments during winter months. Three counts were performed without knowledge of fish length or estimated age. If the counts differed by <10%, the last count was kept. Six YOY otoliths from the 2016 year class and five from 2017 were rejected due to inconsistent counts (12.2 and 10.4% rejection rates, respectively). All examined otoliths had their maximum radius measured from the core towards the post-rostrum. OYOs had their otolith radius measured at the first annulus on the same measuring axis (Supplementary Data S4). All measures and counts were performed in ImageJ using the ObjectJ plug-in (Schneider et al., 2012).
Data analyses
To investigate whether YOY median hatch date differed between year classes, we used the Mood's median test (Johnson and Evans, 1991). We have calculated average growth rate (AGR, mm d−1) for YOYs, a non-longitudinal growth measure that estimates individual mean growth from the division of length at capture by daily age. To test for interannual differences in AGR, we have used a Kruskal–Wallis test, as the homogeneity of variance was respected but the assumption of normality was not met (Quinn and Keough, 2002). Finally, Spearman's correlation was adopted to assess the strength of the association between hatch date and length at capture and between hatch date and AGR, in each year class separately (Quinn and Keough, 2002).
Following the method detailed by Chambers and Miller (1995), a repeated-measures multivariate analysis of variance (MANOVA) was used to determine if the distribution of otolith radius at age, a proxy for size at age (Hare and Cowen, 1997), changed at 5-d age intervals (referred to as age) between groups by fitting two categorical variables: (1) life stage (YOY vs. OYO); and (2) year class (2016 vs. 2017). We chose a multivariate over a univariate approach because the assumption of sphericity required for univariate repeated-measures models was not met (Sirois and Dodson, 2000) and a MANOVA approach is known to be more robust to such condition (Armstrong, 2017). This method has been widely adopted in previous studies relying on similar data (e.g. Meekan and Fortier, 1996; Searcy and Sponaugle, 2001; Robert et al., 2007; Vanalderweireldt et al., 2019b; Khamassi et al., 2020; Takahashi et al., 2022). The MANOVA assumes an unstructured covariance matrix where all variances and covariances are estimated individually (Lehman et al., 2005). To comply with MANOVA assumptions of multivariate normality and no outliers, five individuals were excluded from this analysis (one YOY and two OYOs from the 2016 year class and two YOYs from 2017). Linearity and lack of multicollinearity were tested and these assumptions were satisfied (Quinn and Keough, 2002; Lehman et al., 2005). Wilk's Λ was the chosen statistic to account for all within interactions. If the age × life stage factor yielded p-value <0.05, post-hoc F-tests were used to determine at which specific 5-d intervals the otolith radius differed between groups, at a significance level of 0.05/10 = 0.005, thus accounting for the 10 age comparisons using the Bonferroni correction (Hand and Taylor, 1987).
Otolith radius at first annulus of OYOs, a proxy for size at age 1, was compared with a one-way ANOVA to determine if there were differences between year classes. For this life stage, we have also back-calculated FL at age 1 from otolith radius at the first annulus with the biological intercept (BI) method (Campana and Jones, 1992) to compare it with values obtained in previous studies. FL at BI was defined as 4.75 mm for the age of 4 d, based on previous studies (Houde and Morin, 1990; Vanalderweireldt et al., 2019b). A linear relationship between otolith radius and fish size was previously confirmed (Supplementary Data S5). Statistical analyses were performed using Python 3.6.9 packages numpy and scipy (Virtanen et al., 2020) and SAS Institute JMPTM 13 software (SAS Institute Inc., 2016).
Results
YOY size distributions, hatch date, and AGR
In September 2016 and 2017, YOY size distributions were wide and bimodal (Figure 2). Mean YOY FL ± SD [min, max] was 51 ± 27 [25, 121] mm in 2016 and 67 ± 24 [14, 126] mm in 2017, whereas medians differed more, with a value of 40 mm in 2016 and 70 mm in 2017. In 2016, a dominant mode centred on the 30-mm-size class was observed when FL was partitioned into 10-mm bins, while a secondary mode was detected between 70 and 90 mm (Figure 2a). In contrast, the primary mode in 2017 was observed near the length of 70 mm and the secondary mode was centred on 50 mm (Figure 2b).
Size–frequency distributions of YOY striped bass captured in September (a) 2016 and (b) 2017 prior to subsampling. Size classes are assigned each 10 mm in FL, centred in 15, 25, 35 mm, and so on.
Hatch date distributions of YOY striped bass captured in September were characterized by a similar hatch date range in both years, occurring between 15 May and 24 June in 2016 and between 22 May and 24 June in 2017 (Figure 3; Mood's median test: χ2 = 1.16, p = 0.28). Even though some juveniles hatched up to 7 d earlier in 2016 relative to 2017, median hatch date in 2017 was 3 d earlier than in 2016 (Figure 3). FL at capture in September (Spearman's ρ = −0.70 in 2016, ρ = −0.73 in 2017, all p-values < 0.001) and AGR (ρ = −0.63 in 2016, ρ = −0.55 in 2017, all p-values < 0.001) were negatively correlated to hatch date.
Relative frequency within year class of estimated hatch date of YOY striped bass captured in September 2016 and 2017. M designates the median hatch dates.
AGRs of YOYs ranged between 0.28 and 0.99 mm d−1 in 2016 and between 0.48 and 1.16 mm d−1 in 2017. Mean AGR was 12.5% greater in 2017 (0.72 mm d−1) than in 2016 (0.64 mm d−1), although the statistical support for this observed difference was weak (Kruskal–Wallis test: H = 3.27, p = 0.07).
Differences in growth trajectory between YOYs and OYOs
Otolith radius at 5-d age intervals statistically differed for all within interactions (Figure 4; Table 1; MANOVA: Wilk's Λ27, 412 = 0.25, p < 0.001). As expected from cumulative otolith radius, the age factor exhibits strong statistical differences (Table 1; MANOVA: F9, 141 = 41.24, p < 0.001). The age × life stage interaction indicated significant differences in otolith growth between pre-wintering YOYs and OYOs (Figure 4; MANOVA: F9, 141 = 1.99, p < 0.001). Interannual differences calculated over all individuals through the age × year class interaction had, on the other hand, relatively low statistical support (Table 1; MANOVA: F9, 141 = 0.12, p = 0.07). When the life stage factor was considered, however, the triple interaction age × life stage × year class indicated a statistically significant difference (Table 1; MANOVA: F9, 141 = 0.13, p = 0.033). Following post-hoc analyses, in 2016, OYOs showed larger otolith radius than YOYs as early as the age of 25 d, while in 2017, differences could be detected starting at the age of 20 d (Table 2; F-tests, p < 0.005).
Otolith radius (µm) at 5-d age intervals of YOY (circles) and OYO (diamonds) striped bass from 2016 (empty black) and 2017 (solid grey) year classes. Error bars designate two standard errors.
Growth trajectories were analysed with a repeated-measures MANOVA on otolith radius at 5-d age intervals as dependent variables, and age, life stage (YOY or OYO), and year class (2016 or 2017) as independent factors.
| Value | DF | DFe | Probability > F | |
|---|---|---|---|---|
| All within interactions | Wilk's Λ = 0.25 | 27 | 412 | p < 0.001 |
| Age | F = 41.24 | 9 | 141 | p < 0.001 |
| Age × life stage | F = 1.99 | 9 | 141 | p < 0.001 |
| Age × year class | F = 0.12 | 9 | 141 | p = 0.071 |
| Age × life stage × year class | F = 0.13 | 9 | 141 | p = 0.033 |
| Value | DF | DFe | Probability > F | |
|---|---|---|---|---|
| All within interactions | Wilk's Λ = 0.25 | 27 | 412 | p < 0.001 |
| Age | F = 41.24 | 9 | 141 | p < 0.001 |
| Age × life stage | F = 1.99 | 9 | 141 | p < 0.001 |
| Age × year class | F = 0.12 | 9 | 141 | p = 0.071 |
| Age × life stage × year class | F = 0.13 | 9 | 141 | p = 0.033 |
Results from within-subject interactions are shown.
Growth trajectories were analysed with a repeated-measures MANOVA on otolith radius at 5-d age intervals as dependent variables, and age, life stage (YOY or OYO), and year class (2016 or 2017) as independent factors.
| Value | DF | DFe | Probability > F | |
|---|---|---|---|---|
| All within interactions | Wilk's Λ = 0.25 | 27 | 412 | p < 0.001 |
| Age | F = 41.24 | 9 | 141 | p < 0.001 |
| Age × life stage | F = 1.99 | 9 | 141 | p < 0.001 |
| Age × year class | F = 0.12 | 9 | 141 | p = 0.071 |
| Age × life stage × year class | F = 0.13 | 9 | 141 | p = 0.033 |
| Value | DF | DFe | Probability > F | |
|---|---|---|---|---|
| All within interactions | Wilk's Λ = 0.25 | 27 | 412 | p < 0.001 |
| Age | F = 41.24 | 9 | 141 | p < 0.001 |
| Age × life stage | F = 1.99 | 9 | 141 | p < 0.001 |
| Age × year class | F = 0.12 | 9 | 141 | p = 0.071 |
| Age × life stage × year class | F = 0.13 | 9 | 141 | p = 0.033 |
Results from within-subject interactions are shown.
Otolith radius at 5-d age intervals of YOYs vs. OYOs were tested within each year class with post-hoc F-tests.
| 5 d | 10 d | 15 d | 20 d | 25 d | 30 d | 35 d | 40 d | 45 d | 50 d | ||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2016 | F1, 90 | 2.20 | 2.96 | 0.54 | 3.98 | 20.09 | 38.94 | 55.75 | 79.71 | 102.13 | 124.88 |
| p-value | 0.24 | 0.19 | 0.46 | 0.05 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
| 2017 | F1, 59 | 2.54 | 1.12 | 0.04 | 10.81 | 29.25 | 51.86 | 83.31 | 105.28 | 131.21 | 127.25 |
| p-value | 0.12 | 0.20 | 0.84 | 0.002 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 |
| 5 d | 10 d | 15 d | 20 d | 25 d | 30 d | 35 d | 40 d | 45 d | 50 d | ||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2016 | F1, 90 | 2.20 | 2.96 | 0.54 | 3.98 | 20.09 | 38.94 | 55.75 | 79.71 | 102.13 | 124.88 |
| p-value | 0.24 | 0.19 | 0.46 | 0.05 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
| 2017 | F1, 59 | 2.54 | 1.12 | 0.04 | 10.81 | 29.25 | 51.86 | 83.31 | 105.28 | 131.21 | 127.25 |
| p-value | 0.12 | 0.20 | 0.84 | 0.002 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 |
Otolith radius differed statistically between pre-wintering YOYs and surviving OYOs when the p-value was smaller than a significance level of 0.05/10 = 0.005 (Bonferroni correction for 10 multiple comparisons).
Otolith radius at 5-d age intervals of YOYs vs. OYOs were tested within each year class with post-hoc F-tests.
| 5 d | 10 d | 15 d | 20 d | 25 d | 30 d | 35 d | 40 d | 45 d | 50 d | ||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2016 | F1, 90 | 2.20 | 2.96 | 0.54 | 3.98 | 20.09 | 38.94 | 55.75 | 79.71 | 102.13 | 124.88 |
| p-value | 0.24 | 0.19 | 0.46 | 0.05 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
| 2017 | F1, 59 | 2.54 | 1.12 | 0.04 | 10.81 | 29.25 | 51.86 | 83.31 | 105.28 | 131.21 | 127.25 |
| p-value | 0.12 | 0.20 | 0.84 | 0.002 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 |
| 5 d | 10 d | 15 d | 20 d | 25 d | 30 d | 35 d | 40 d | 45 d | 50 d | ||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2016 | F1, 90 | 2.20 | 2.96 | 0.54 | 3.98 | 20.09 | 38.94 | 55.75 | 79.71 | 102.13 | 124.88 |
| p-value | 0.24 | 0.19 | 0.46 | 0.05 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
| 2017 | F1, 59 | 2.54 | 1.12 | 0.04 | 10.81 | 29.25 | 51.86 | 83.31 | 105.28 | 131.21 | 127.25 |
| p-value | 0.12 | 0.20 | 0.84 | 0.002 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 |
Otolith radius differed statistically between pre-wintering YOYs and surviving OYOs when the p-value was smaller than a significance level of 0.05/10 = 0.005 (Bonferroni correction for 10 multiple comparisons).
Otolith radius at first annulus and OYO back-calculated size distributions
Mean otolith radius at first annulus in OYOs from the 2016 year class (2558 ± 171 μm) was 20% larger than that of individuals from 2017 (2130 ± 222 μm; Figure 5a; one-way ANOVA: F1, 59 = 60.63, p < 0.001). Back-calculated length at age 1 from otoliths of OYOs showed different mean lengths between the 2016 and 2017 year classes, with respective values of 139 ± 171 and 111 ± 13 mm (Figure 5b, c; one-way ANOVA: F1, 59 = 66.42, p < 0.001). Minimum and maximum back-calculated lengths at age 1 were 115 and 159 mm in 2016, and 91 and 130 mm in 2017. In OYOs from the 2017 year class, a bimodal pattern was observable in the distribution of back-calculated length at age 1 (Figure 5c).
(a) Otolith radius at first annulus (µm) of OYO striped bass from the 2016 and 2017 year classes. Boxplots show median, quartiles, and maximum and minimum; triangles designate the mean value. Different capital letters (A, B) indicate statistically different means (one-way ANOVA, p < 0.05). (b), (c) Size–frequency distributions (back-calculated at age 1 from otoliths) of OYO striped bass from the 2016 and 2017 year classes. Size classes are assigned each 10 mm in FL, centred in 95, 105, 115 mm, and so on.
Discussion
The growth–survival paradigm in fisheries science underscores the role of fast growth in driving high survival rates during early life stages of fish (Houde, 1987; Anderson, 1988; Takasuka et al., 2017). Interannual variability in growth-dependent mortality is one of the primary sources responsible for the fluctuations in year-class strength and recruitment experienced by fish populations. Using striped bass as a model species, the main objective of this study was to test the hypothesis that size-selective mortality extends into the juvenile stage. In both year classes studied, OYO survivors exhibited faster mean early growth than pre-wintering YOYs, indicative of strong size-selective mortality during their first winter. Growth trajectories differed earlier in age between YOYs and OYOs in the 2017 year class than in 2016. The larger mean otolith radius at first annulus observed in OYOs from 2016 compared to 2017 suggests the existence of relatively high interannual variability in size-selective mortality during the juvenile stage until age 1.
Sampling methods and the measure of size-selective mortality
When estimating size-selective mortality through the comparison of an initial population and its survivors after a given period, a central premise of this method is that samples from both life stages originate from the same reference group. Hatch date could not be estimated in otoliths of OYOs because of the stacking of daily increments during the first annulus formation in winter, preventing the direct confirmation that OYOs shared the same hatch date ranges with YOYs. Our examination of YOY hatch date distributions indicated that hatching occurred over a period of 30–40 d in both years. This period corresponds well with knowledge of SLR striped bass spawning season, which is known to occur over a relatively short 1-month period between May and June, with a peak in reproductive activity in early June (Vanalderweireldt et al., 2019b; L'Italien et al., 2020). This result makes us confident that (1) YOY samples were representative for both 2016 and 2017 year classes; and (2) given the absence of apparent size selectivity by OYO capture methodologies, these older juveniles also correspond to the same reference groups from which YOYs originated.
Growth–survival paradigm and selection for fast growth
The growth–survival paradigm highlights the importance of the larval stage for recruitment variability, by considering that mortality processes during that stage are the primary source of variability in year-class strength (Houde, 1987; Anderson, 1988; Takasuka et al., 2017). This conceptual framework states that a positive relationship exists between mean growth rate observed within a larval cohort and larval survival, and subsequent year-class strength. Investigations on the effects of size-selective mechanisms in post-larval juveniles are scarce due to the difficulty of obtaining representative samples of a year class between the pelagic larval and the late juvenile stages. However, the susceptibility of a year class to these mechanisms is likely to extend past the early juvenile stage, and can therefore contribute to generate variability in year-class strength that goes unexplained (Sissenwine, 1984; Sogard, 1997). Indeed, large size variation in fish juveniles may be responsible for a greater potential of size-selective mortality during the juvenile stage compared to the larval stage, especially during the first winter of life (Sogard, 1997). Biological processes during the early juvenile stage play a greater role in regulating fish population dynamics than historically thought, and its understanding is key to adequately assess annual recruitment variability and predict with increased accuracy the occurrence of exceptionally strong or weak year classes (Bradford, 1992; Sogard, 1997).
Size-related mortality of YOYs during the first winter of life has been reported in several freshwater (Miranda and Hubbard, 1994; Smith and Griffith, 1994; Heermann et al., 2009), anadromous (Johnston et al., 2005; Moss et al., 2005), and marine fish populations (Schultz et al., 1998; Hales Jr and Able, 2001; Robert et al., 2007; Anderson and Scharf, 2014), including other striped bass populations in North America, e.g. in the Hudson River (Hurst and Conover, 1998) and in the Chesapeake Bay (Martino and Houde, 2012). In the southern Gulf of St. Lawrence, the striped bass population that spawns in the Miramichi River, from which the reintroduced SLR population originates, YOY size-selective winter mortality is also presumed to occur (Chaput and Robichaud, 1995; Robichaud-LeBlanc et al., 1998; Douglas et al., 2006). Relying on several studies on striped bass populations along the Atlantic coast of North America, Conover (1990) found that back-calculated sizes at age 1 in adults of various ages are consistently larger compared to sizes measured in YOYs at the end of the growing season, providing further support for this biological pattern.
In the SLR striped bass population, OYOs showed a faster mean early growth than pre-wintering YOYs in both year classes studied, with differences in growth trajectory observed before the age of 25 d. This selective mortality against slow growers draws attention to early life processes like larval and early juvenile daily growth that will influence survival later until age 1, including the first autumn and winter of life. Variability in mortality of juvenile fish is thought to be mainly driven by predation, with a potentially strong selectivity favouring fast-growing larger individuals (Sogard, 1997; Van der Veer et al., 1997). In the Hudson River estuary, Hurst and Conover (2003) observed a seasonally diverse growth pattern in YOY striped bass, with an apparent priority for growth in terms of length during the summer that shifts to structural mass and energy reserves towards the end of the growing season, as winter approaches. Such growth strategy is likely determined by a size-dependent predation risk combined to a lower risk of starvation during summer (Hurst and Conover, 2003). During winter, starvation and depletion of energy reserves may take over as the main sources of mortality among temperate fish populations, with a size dependence attributed to differences in energy stores relative to body length (Shuter and Post, 1990; Schultz and Conover, 1999). Energy losses through the usage of lipid energy reserves during winter in YOY Hudson River striped bass were positively correlated to the energy levels at the onset of winter, which, in turn, were correlated to fish length attained at the moment of the autumn shift from favouring growth in length to allocating energy in lipid reserves (Hurst and Conover, 2003). Overwintering YOYs that reached larger sizes during the growing season have thus a substantial advantage in starvation resilience, due to the allometry of their energy reserves allocation relative to lean tissue, which is favoured by a larger body size (Hurst and Conover, 2003). The proposed mechanisms reported in Hudson River striped bass may be even more prevalent in the SLR population, where the contrast between summer and winter conditions is greater, especially given the shorter summer when individuals must maximize growth, and the longer winter, which increases reliance on stored energy.
In the two SLR striped bass year classes observed in this study, pre-wintering YOYs showed similar early growth trajectory between years. Conover (1990) found that YOY striped bass at the end of the growing season exhibited little variation in size across the latitudinal range of the species, with mean lengths obtained from several studies between 76 and 106 mm (Supplementary Data S6). When compared with other populations north of the Hudson River (Supplementary Data S6), maximum size observed in YOY SLR striped bass in September is within the expected range, but mean and particularly minimum sizes are relatively small, with the frequent occurrence of 20–30 mm individuals that have not been observed in other populations.
A broad range in AGR was also observed within pre-wintering YOYs. In the present study, larger YOYs (>110 mm FL) exhibited AGRs ranging from 0.85 to 1.16 mm d−1, while their smaller counterparts (<40 mm FL) had their AGRs ranging from 0.28 to 0.55 mm d−1. We found negative correlations between hatch date and both size at capture in September and AGR, suggesting that earlier hatching within a given cohort results in larger pre-wintering size, conferring a survival advantage during winter for early hatchers. In a recent study focussing on the 2014 year class of the SLR striped bass, Vanalderweireldt et al. (2019b) observed AGRs of YOYs in September ranging between 0.6 and 0.9 mm d−1. Conover (1990) also calculated an AGR of 0.9 mm d−1 for YOYs at the end of the growing season from studies on Canadian striped bass populations. The broad range in AGR and size consistently documented in YOYs of the re-established SLR striped bass population in September draws attention to potentially low size-selective pressure during the first weeks of life (Vanalderweireldt et al., 2019b), signaling that size-selective predation and density-dependent mechanisms that drive mortality during the first growing season occur at lower rates than in other striped bass populations, including in the SLR population prior to extirpation and re-establishment (Conover, 1990). It is possible that the carrying capacity for YOYs in the first growing season of this reintroduced population has not been reached yet. During autumn and winter, size-dependent selection pressure increases and may thus constitute a more important bottleneck for this population, as demonstrated through the characterization of survivors after the first winter.
Over the species distribution range in North America, Conover (1990) found a wide variation of mean back-calculated sizes at age 1 (80 to 160 mm; Supplementary Data S7). Interestingly, in the extirpated SLR population, relatively small back-calculated sizes at age 1, down to 69 mm, were observed (Magnin and Beaulieu, 1967; Supplementary Data S7). In the Miramichi River, it is suggested that few YOY striped bass measuring <100 mm survive their first winter (Douglas et al., 2006). Back-calculated lengths at age 1 in OYOs in our study are within the range observed in other northern striped bass populations and suggest an interannual variability in size-dependent YOY winter survival, as mean otolith radius at first annulus was significantly larger in survivors from the 2016 year class compared to those from 2017.
Recruitment endpoint and management implications
Our results highlight the importance of size-selective mortality until the end of the first year of life for the SLR striped bass population, and thus indicate that the endpoint for year-class strength determination is most likely set at the end of the first winter. Even though the temporal resolution of our sampling precludes us from assessing the possibility that the endpoint could correspond to the end of the first growing season around November, our results strongly suggest that year-class strength determination extends beyond the larval and post-larval juvenile stages, contrary to the prediction of the growth–survival paradigm that strong growth-dependent mortality responsible for recruitment variability primarily occurs before metamorphosis. Winter survival for YOYs of a northern estuarine population like the SLR striped bass is likely affected by a sum of effects stretching from early larval growth and dispersion to pre-winter energy allocation (Hurst and Conover, 2003) and winter severity conditions (Hurst and Conover, 1998; Lankford and Targett, 2001). In the Hudson River striped bass population, by monitoring eight different year classes and bottom water temperature during winter, Hurst and Conover (1998) associated the occurrence of colder and longer winters with lower post-winter OYO abundance and eventually weaker recruitment. In the northernmost striped bass population in the St. Lawrence estuary, winter severity should be considered as an important potential driver of year-class strength.
YOY abundance monitoring surveys are typically conducted in late summer for several striped bass stocks in North America (Durell and Weedon, 2019; Gallagher et al., 2019; NYSDEC, 2020). In the case of the SLR population, our results demonstrate that an autumnal YOY abundance index obtained in mid-September may not provide a reliable proxy for recruitment. Generally, the assessment of YOY size-dependent winter mortality and its effects on the abundance and size distribution of a given year class can be highly informative from a fisheries management perspective (Stige et al., 2019) and thus failure to consider winter mortality may represent a crucial source of bias in year-class strength estimation. We propose three complementary research avenues to refine our estimation of the timing of the endpoint of SLR striped bass recruitment and potentially other temperate fish populations with monitored YOY abundance: (1) considering a later, pre-winter sampling (e.g. November) to characterize size-selective mortality processes occurring during the pre-wintering period, coupled with a post-winter sampling of OYOs (e.g. Khamassi et al., 2020); (2) extending the present study over a 5- to 10-year period to better appreciate the interannual variability that exists in the timing of size-selective mortality during the first year of life; and (3) implementing a winter survey to examine to what extent condition and size–frequency distribution depend on bottom water temperatures in wintering areas, with the purpose of providing direct information on the relationship between winter severity and YOY survival. Such information would help to better estimate the contribution of annual cohorts to the stock and to underpin fisheries management decisions based on the species biology.
Supplementary material
The following supplementary material is available at ICESJMS online: Supplementary Data S1. Size–frequency distribution of striped bass captured by commercial eel fishing gears; Supplementary Data S2. Size–frequency distribution of OYOs captured by angling, compared to OYOs captured by eel fisheries in August; Supplementary Data S3. Size–frequency distribution of striped bass captured by beach seine between 2013 and 2018; Supplementary Data S4. Striped bass otolith photos and measurement procedures; Supplementary Data S5. Relationship between length at capture and otolith total radius, and between length at capture and otolith radius at 50 d for YOY striped bass; Supplementary Data S6. Mean FL of pre-wintering YOY striped bass from different populations in North America, as reviewed in the literature; Supplementary Data S7. Mean back-calculated FL at age 1 of OYO striped bass from different populations in North America, as reviewed in the literature.
Authors’ contributions
HAP designed the study, conducted laboratory and statistical analyses, analysed the data, interpreted the results, and wrote the manuscript. DR designed the study, analysed the data, interpreted the results, and wrote the manuscript. JM designed the study, conducted sampling and fieldwork, analysed the data, interpreted the results, and wrote the manuscript. PS designed the study, analysed the data, interpreted the results, wrote the manuscript, and funded the research.
Conflict of interest
The authors have no conflicts of interest to declare.
Funding
This project was funded by the Ministère des Forêts, de la Faune et des Parcs du Québec (MFFP) through a grant to the Chaire de recherche sur les espèces aquatiques exploitées (PS) (Research Chair on Exploited Aquatic Species, CREAE-UQAC). DR was supported by the Canada Research Chairs programme.
Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.
Acknowledgements
The authors would like to thank Anne-Lise Fortin, William Fortin, and Stevens Gagné from UQAC for their unaccountable help during lab work; Anne-Marie Pelletier, Eliane Valiquette, Denise Deschamps, Léon L'Italien, Philippe Brodeur, Ariel Arsenault, Jérôme Barbeau, Caroline Brûlé, and Catherine Ouellet from the Ministère des Forêts, de la Faune et des Parcs du Québec (MFFP) for the supervision on the young-of-the-year juvenile striped bass monitoring programme and the sampling of one-year-old juvenile striped bass; Olivier Morissette and Gérald Chaput for their contributions and ideas; and four anonymous reviewers for their insightful comments.
