Older but not wider. Impacts of molt frequency variation on age and growth dynamics of Newfoundland & Labrador snow crab (Chionoecetes opilio) toward corroborating gastric mill 2 age estimations

Aging specimen and survey data from a heterogeneous stock range are combined to investigate effects of 17 molt frequency variation on age and growth dynamics and efficacy of gastric mill age band estimations in 18 snow crab. A central assumption that bands form independent of molting is examined through analyses of 19 molt frequency variation with an overall result of support for the hypothesis that gastric mill bands form 20 independent of molting. This is based on an applied review approach, demonstrating blind age estimation 21 results within and across population units featuring different rates of molt frequency interruptions in context 22 of literature-based expected outcomes of age and growth dynamics. The comprehensive analyses overall 23 supports the ability of gastric mill bands to track chronological ages, but non-fully explained outcomes of growth delays into sexual maturity stages in males and potential estimation error in 1-3 year old crab 25 preclude a conclusion that absolute ages are consistently tracked throughout ontogeny. The results suggest gastric mill bands continue to form during both skip-molts and after terminal molt. Despite uncertainties, the research highlights that males originating from populations where skip-molting is prominent are older 28 reaching fishery size than previously thought (9-10 years), with tentative average estimates of 10-12 years. The study presents novel observations of size-at-age and age-at-maturity in Newfoundland & Labrador 30 (NL) snow crab and broadens life history knowledge for the species. Along with corroborating gastric

supports the ability of gastric mill bands to track chronological ages, but non-fully explained outcomes of 24 growth delays into sexual maturity stages in males and potential estimation error in 1-3 year old crab 25 preclude a conclusion that absolute ages are consistently tracked throughout ontogeny. The results suggest 26 gastric mill bands continue to form during both skip-molts and after terminal molt. Despite uncertainties, 27 the research highlights that males originating from populations where skip-molting is prominent are older 28 reaching fishery size than previously thought (9-10 years), with tentative average estimates of 10-12 years.

29
The study presents novel observations of size-at-age and age-at-maturity in Newfoundland  The demonstration by Kilada et al. (2012) of the potential ability to directly age crustaceans by counting 43 bands in cuticular structures is arguably one of the most important developments to benefit assessment of 44 crustacean resources in recent decades. Age-structured population models based on counting band 45 signatures within various hard structures of animals are commonly used in assessments of fish species (i.e., 46 Campana, 2001). Age-based models normally allow greater precision for tracking and projecting cohort-47 specific demographics within populations than size frequency or surplus production models commonly used 48 in assessments of crustaceans (i.e., Hartnoll, 2001;Vogt, 2012). However, to-date, application of age band 49 counts into the betterment of crustacean stock assessments has been limited (Kilada and Driscoll, 2017).

51
The reliability of band counts in the gastric mill ("stomach teeth") or other parts of crustaceans to assess 52 age has been highly debated. There is both support (i.e., Kilada  In many crustaceans, molting schedules are temporally inconsistent over ontogeny, thus for band count 64 techniques to track absolute ages a key assumption is that cuticular band formation (whatever the 5) age-at-maturity is likely to be highest in male populations where skip-molting is most common due 130 to expected consistency in minima on instar stages where terminal molt will first occur; 131 6) age-at-maturity should be higher in males than females due to expected earlier commitment to 144 Mullowney et al., 2018). Of particular relevance is the terminal molt, which stops growth (Conan and 145 Comeau, 1986). Terminal molt occurs over a broad size range, particularly in males, in all global stocks. In 151 during any given molting period. A "regular-molt" entails a crab remaining morphometrically immature 152 (sub-adult) after molting, a "skip-molt" entails a crab not molting at a time it might be expected to do so, 153 and the aforementioned "terminal-molt" involves a crab developing morphometric features defining it as 154 an adult upon its final molt (Conan and Comeau, 1986). Shell condition represents a subjective proxy for 155 time elapsed since molt, and shells advanced beyond a new-shell stage (intermediate and old shell) are used 156 to identify skip-molters (for sub-adult males) or adults that did not recently molt. Notwithstanding morphometrically mature at the same time (i.e. a "puberty/maturity" molt). Males, however, become 164 sexually mature prior to the terminal molt (i.e. a "puberty/regular" molt). Herein, sub-adult females are 165 termed immature and sub-adult males are termed either immature (pre-pubescent [not sexually mature]) or 166 adolescent (pubescent [sexually mature]). Size-at-maturity and age-at-maturity both refer to the terminal 167 molt stage to define maturity. 168 169 Instar refers to the number of molts undertaken and is not equivalent to chronological years. This reflects 170 differing inter-molt periods between instar stages. In males, the puberty molt most commonly occurs at 171 instar VIII or IX (Kon and Honma, 1978;Sainte-Marie et al., 1995) and terminal molt at instars X-XIII 172 (Comeau et al., 1998 (Ernst et al., 2005;Orensanz et al., 2007), and the Seas of Japan and Okhotsk (Ito, 1970;Kon, 1980 Mullowney and Baker (2021) showed that over the past twenty-five years males throughout different 232 portions of the NL stock range had a 25-40% probability of being terminally-molted at 80mm CW versus 233 a 50-75% probability of being terminally-molted at 100mm CW. These sizes most closely conform to 234 instars XI and XII ( Figure 1)

244
Growth patterns follow a near-linear process that reflects overall decreasing relative but increasing absolute 245 growth increments from one instar to the next. Sainte-Marie et al. (1995) estimated percentage growth 246 increments to decrease from 60.5% to 34.3% from instars I-VII, and from 32.6% to 17.9% from instars 247 VIII-XIV. The rich literature on growth increments in snow crab conforms well with the tabulated instar 248 size ranges and modal centres presented in (Ito, 1970) and Sainte-Marie et al. (1995) in the Sea of Japan 249 and the nGSL respectively, with the latter used herein for quantitative comparisons (Figure 1).

251
The phenomenon of similar size-at-instar across and within both Atlantic and Pacific stocks holds 252 particularly well at smallest sizes (Yamamoto et al., 1998a

259
Size-at-instar is similar across sexes during early ontogeny but diverges in mid-life stages, with the 260 divergence particularly obvious upon sexual maturation in females. A historic field-based report showed 261 tight across-sex coupling in size-at-instar occurred at all comparative instars (Ito, 1970), but recent lab- Importantly, high population densities of large males appear to promote skip-molting over terminal-molting 298 for adolescent males not undertaking a regular molt, as shown in Mullowney and Baker (2021) and inferred from previous studies (Hébert et al., 2002;2017). This outcome promotes overall greater sizes within 300 populations through an overall greater number of molts occurring over ontogeny as the "stop point" for 301 molting is prolonged. Moreover, the regulatory influences of large male density on growth are likely further 302 intertwined with mating success in adolescent males, which in-turn plausibly affect molt-type "decisions" 303 at any given instar upon which one of the three molt-type outcomes (regular, skip, terminal) are possible. The majority of the Atlantic Canadian snow crab stock occurs in NL waters, with NL consistently being 327 the world's largest single-source of market supplies for three decades. NL snow crab is an ideal stock upon 328 which to examine the efficacy of direct age determination due to heterogeneous bathymetric, thermal, and 329 population density attributes across a large stock range. In particular, this region provides a strong basis to 330 examine the key assumption that gastric mill bands form independent of molting. This is because skip-  For males, given the rankings of skip-molting rates range from lowest to highest as 2J3K, 3LNO, and 3Ps, 390 we expect 3Ps to have lowest size-at-age and 2J3K to have highest size-at-age. However, opposing "greater 391 forces" of higher ambient temperatures in 2J3K versus higher population densities of large males in 3LNO 392 could affect how results from either SD relate to one another and deviate from 3Ps. In comparisons to 393 adolescent males in the nGSL, the higher levels of skip-molting in NL populations should lead to lower 394 size-at-age in all NL groups, with greatest difference in SD 3Ps and likelihood for least difference in SD 395 2J3K. For age-at-maturity, we expect highest age-at-maturity in SD 3Ps due to the slower growing 396 conditions, assuming males would strive to achieve a relatively high instar stage (associated with a 397 relatively large size) before committing to terminal molt. With the least common incidence of skip-molting 398 occurring in 2J3K, we expect age-at-maturity to be lowest in males that SD. However, again, competing 399 forces of temperature versus population density could affect the extent to which SD 2J3K and SD 3LNO 400 have lower age-at-maturity than SD 3Ps, particularly in light of modest sample sizes used in this study.

402
For females, with little to no incidence of skip-molting, we expect little difference is size-at-age across SDs.

403
Further, given conformity to a limited number of instars at terminal molt, we expect little difference in age-404 at-maturity across SDs. However, it is possible smaller molt increments in cold conditions (not measured 405 in this study) could lead to lowest [highest] size-at-age in SDs 3Ps [2J3K] respectively. It is also possible 406 unmeasured incidence of skip-molting at early instars could affect the outcome in the same way, with 407 coldest [warmest] 3Ps [2J3K] having lowest and highest size-at-age respectively due to contrasting thermal 408 regimes.

410
Across sexes, there should be no differences in size-at-age in small crab where no divergence in sex-specific 411 maturity processes occur within any SD. In mid-large crab of either sex, after stages where sex-specific 412 divergences in maturity processes occur, size-at-age is expected to be higher in males in all SDs. Age-at-413 maturity should be higher in males than females in all SDs because they normally undergo terminal molt at

712
The GLM on male CW expectedly showed age to be a significant predictor of size (

732
Small sample sizes and a more limited size range likely affected model-estimated patterns of growth in 733 females moreso than in males, but the female models nevertheless also showed more similar growth in SDs 734 2J3K and 3LNO than in 3Ps. The relatively large estimated size in estimated age 2 females in 3LNO 735 (~35mm CW) almost certainly reflects insufficient data (n = 6 females ≤2 years old), and no females above 736 estimated age 8 were present in the Division 3Ps data ( Figure 6). Nevertheless, female size-at-age was 737 similar in SDs 2J3K and 3LNO over estimated ages 6-12, during which time sizes gradually increased from 738 about 40 to 60mm CW. In contrast, female growth rates in SD 3Ps progressively increased over the limited age range examined and showed lower estimated size-at-age than in the other two SDs over the estimated 740 ages 1-5, before reaching a similar level of about 40mm CW at estimated age 6.

750
The GLM on female CW similarly showed age to be a significant predictor of size (Table 1, p<0.0001) and 751 pairwise comparisons among SDs were consistent in showing significant differences in estimated size-at-752 age between SD 3Ps versus both 2J3K (p<0.01) and 3LNO (p<0.001), but no difference between 2J3K and 753 3LNO (p=0.79) ( Table 1). The odds ratios of size differences over the age spectrum for SD 2J3K and 3LNO 754 crab relative to 3Ps crab were both about 1.3, and there were significant differences in mean sizes of 3Ps 755 crab with the other two SDs by age 6 (estimated age 5.8 years) (Figure 6).

757
Across-sex comparisons of size-at-age were affected by sample size limitations in all SDs, which likely 758 affected model fits, including limited (SD 2J3K) or no (SD 3Ps) samples of mature females and scarcity of 759 youngest (immature) crab of both sexes (estimated ages 1-3, SD 3LNO). Nonetheless, with both maturity 760 groups pooled the data were deemed sufficient to qualitatively interpret across-sex differences over the 761 entire age spectrum of SD 2HJ, beyond age 3 in SD 3LNO, and below age 8 in SD 3Ps. Overall, it is 762 apparent that across-sex divergence in size-at-age began after estimated age 4 in SDs 2J3K and 3LNO, with 763 the growth rate curves in females more restricted than that of males and gradually dampening thereafter ( Figure 5) and presumably reflecting the earlier maturation of females. By estimated age 14, males were 765 about 50mm CW larger than females on-average (110 vs. 60mm CW) in both SDs 2J3K and 3LNO. The 766 immature females in 3Ps were only marginally smaller at any given size than their male counterparts, all of 767 which were also immature/adolescent, over the entire estimated age 1-7 range. Overall, despite sample size 768 limitations, it is apparent that female size-at-age is lower than that of males for part of the age spectrum 769 within at least some SDs (SDs 2J3K and 3LNO). 770 771

776
The GLM tests on <40mm CW crab confirmed that there was no significant difference in size-at-age 777 between the sexes in this size grouping (Table 1, p=0.17), which was not expected to be associated with 778 terminal molt in either sex. The lack of difference between the sexes reflected insignificant differences 779 (p>0.05) in all sex-SD pairings (Table 2). Despite lack of significance, females were overall slightly larger 780 than males in this size group of crab (odds ratio 1.1 F:M, Figure 6), which likely reflected some unusually 781 large females at young ages in SD 3LNO in particular ( Figure 5). In contrast to the GLM tests on ≤40mm CW, the results of the >40mm CW test on size-at-age variation between the sexes showed there were 783 significant differences (Table 1,   Age-at-maturity ogives in males were characterized by a high level of variance across the entire age 826 spectrum in all SDs, but it was nonetheless ostensible that the onset of morphometric maturity on-average 827 occurred later in SD 3Ps than in SDs 2J3K and 3LNO (Figure 7). The GLM expectedly showed that age 828 was a significant predictor of maturity rates in males (p<0.0001, Table 1). Overall, maturity occurred over 829 a relatively broad and largely overlapping estimated age range in the 3 SDs: 6-14 estimated age for SD 830 2J3K, 7-16 for SD 3LNO, and 9-14 for SD 3Ps (Figure 7). Age at 50% maturity (mat50) in males occurred 831 at estimated age 9 in SDs 2J3K and 3LNO and estimated age 11 in SD 3Ps.

844
Although the p=0.08 (Table 1) produced by the GLM test on age between sex-specific differences in age-845 at-maturity in SD 3LNO was not significant at the conventional p<0.05 level, it was suggestive of a 846 difference in the process and was consistent with post-hoc tests which showed a higher proportion of 847 females (0.74) were expected to be mature at age 9 (point estimate age 8.6) than males (0.47).

849
The investigation of differences in age between maturity groups for same-size crab showed mean ages of

872
Modal size estimates for instars IV to VIII from the nGSL overall compared well with modes observed in 873 ≤45mm CW size frequency data from Campelen trawl surveys (1995-2020) for both sexes and in our three 874 SDs of the NL shelf (Figure 9). This supported application of modal instar sizes for nGSL males to the aged 875 male crab from NL ( Figure 10).

899
The higher incidence of skip-molting in SD 3Ps males that would contribute to lower size-at-age found in 900 the aging component of this study was evident in the field data, particularly across instars IX to XIII. Skip-901 molting was overall most pronounced at instars X-XI, with annual instar-specific rates of skip-molting often 902 nearing or exceeding 50% of the animals (28% on-average). In contrast, annual instar-specific incidence of 903 'control' population after age 3 suggests that the estimated ages were 'real', and that the 1-3 year old NL 1050 crab were in fact substantially larger than same-age individuals in nGSL population, which suggests that 1051 the former underwent more rapid molting than the latter. We undertook an exploratory post-hoc analysis to

1113
Consequently, skip-molters at young instars go undetected in the trawl-survey data using the criteria that 1114 they are immature/adolescent males with a shell condition advanced beyond new-shell. Indeed, the post-1115 hoc analysis on recent thermal distribution of crab ( Figure 12) revealed that in some SDs and years smallest 1116 crab were captured in waters warm enough to promote rapid molting, but it also revealed that in some SDs 1117 and years, particularly in SD 3Ps, these earliest instar crab would be captured in cold temperatures (i.e. 0-1118 1°C) sufficient to induce common incidence of skip-molting. Despite its shortcomings, this study is comprehensive in its comparative analyses and breadth of novel age-1146 based outcomes that are consistent with field data and literature on snow crab growth dynamics. The basis 1147 to conclude that bands in the gastric mill form independent of molting, and that the aging method is able to 1148 detect this, is strong. The study also presents compelling evidence that absolute ages are being tracked, at

1176
Availability of data and material: All data used in this study and pre-processing details are available upon 1177 request to DM. The aging specimen data are a small dataset and require minimal pre-processing information