Dramatic decline and limited recovery of a green crab (Carcinus maenas) population in the Minas Basin, Canada after the summer of 2013

Introduction

Detailed observations over time made on organisms of the same species in one or more selected locations, constituting the same putative ‘population(s)’, is a staple of studies in marine biology and ecology in general (Bertness et al., 1992; Santos & Simon, 1980; McGaw, Edgell & Kaiser, 2011; Palumbi & Pinsky, 2014). Such monitoring can provide essential information about the life history and demography of the species of interest (Hoskin et al., 2011), its interactions with other species in the same community (Seitz, Knick & Westphal, 2011), and abiotic or biotic factors influencing recruitment to its populations (Scrosati & Ellrich, 2016). When observational periods coincide with particularly disruptive events, for example natural disasters (Sato & Chiba, 2016), the introduction of new invasive species (Delaney et al., 2008), or climate-driven shifts in oceanographic regimes (Mills et al., 2013; Pinsky et al., 2013), the impacts of such events on the studied species can be illuminating. Changes in abundances and the size or reproductive structure of species’ populations following such events, as well as overall shifts in community specific composition, illustrate which ecological changes impact particular species and how (Ruth & Berghahn, 1989; Palumbi & Pinsky, 2014; Bertness et al., 1992). These observations provide information that can potentially be used to forecast longer-term impacts of such changes into the future, which is essential for management and conservation of marine species (Ruth & Berghahn, 1989; Pinsky et al., 2013; Wahle & Carloni, 2017).

When a non-native species is introduced to a new location its impacts on native biota can be substantial (e.g., Klassen & Locke, 2007; Scalici & Gherardi, 2007), so monitoring introductions and their impacts provides important ecological information. Once an introduced species establishes an invasive population, however, it takes on a distinct role within the invaded community and is essentially a part of the invaded ecology (e.g., Boudreau & Hamilton, 2012). Invaders may, however, be more sensitive to environmental perturbations in their invasive range than are native biota because invaders, unlike native biota, do not have the same evolutionary history with environmental conditions in their invaded range as do natives (Grosholz et al., 2000; Kienzle, 2015; MacDonald et al., 2018). It is thus possible that ecological shifts potentially affecting an entire biological community may first impact an invasive species before having similar effects on native taxa. Monitoring invasive as well as native populations can thus potentially serve as an early warning system for changes to the community or its ecosystem.

The Bay of Fundy is an inlet of the Gulf of Maine, located along the Atlantic coast of eastern North America, with extremely large tidal ranges (up to 16.8 m) and distinct marine communities, particularly in the Minas Basin and Chignecto Bay portions of the upper Bay (Daborn & Pennachetti, 1979; Parker, Westhead & Service, 2007). Marine communities within the Gulf of Maine-Bay of Fundy system are currently facing a number of sources of stress and change, including rapid warming of ocean waters (Mills et al., 2013; Pinsky et al., 2013) and introduction of several invasive species (Moore et al., 2014; Klassen & Locke, 2007). Particularly influential in recent years have been very warm summers, such as the 2012 ‘heat wave’, and possible associated shifts in oceanographic circulation (Mills et al., 2013). Indeed, many species in this system demonstrated low abundances or recruitment in the year 2013 (e.g., Clements, 2016; Wahle & Carloni, 2017; see also Discussion), signalling some major disruption in the system that bears further investigation. One invasive species that has become well-established within the Bay of Fundy is the European green shore crab, Carcinus maenas. This species reached the Bay of Fundy in the 1950s and became established and numerous within the upper bay (i.e., the Minas Basin) in the late 1990s (Klassen & Locke, 2007). Green crabs are aggressive consumers of a wide range of prey species as predators and scavengers (Crothers, 1968; Trussell, Ewanchuck & Bertness, 2003; Klassen & Locke, 2007; Boudreau & Hamilton, 2012) and compete with native species preying on similar food sources (Grosholz et al., 2000; Haarr & Rochette, 2012). This species thus plays significant roles within food webs of marine communities it has invaded (Wong & Dowd, 2015).

This paper describes the results of efforts to monitor an established invasive population of green crabs (Fig. 1A) at Clarke Head, Nova Scotia (NS), Canada (Figs. 1B, 1C), within the Minas Basin of the upper Bay of Fundy (Fig. 1D). Green crabs first appeared at this site ca. 1999, and very quickly increased in abundance and almost completely displaced local rock crabs (Cancer irroratus) from intertidal environments at Clarke Head by 2002 (BK Quinn, pers. obs., 2000–2018). A specific monitoring protocol was established by the author in 2008 and continued thereafter. Monitoring thus unfortunately missed the early years in which green crabs first appeared and became abundant at this location. However, this relatively new population within the area was still monitored to track its health and dynamics, as doing so is potentially useful to forecast impacts of changes to invader numbers on native biota, and to use it as a potential early warning system for changes to the marine ecology within the study area. During the monitoring period (2008–2017), a substantial change to the green crab population was observed beginning in 2013 and persisting to the present. Within the present paper, the characteristics of the crab population before, during, and after this change are examined, and environmental changes potentially responsible are compared with changes in the crab population. Observations by recreational fishers and locals within the study area of biotic changes to crab bycatch and potential predators and prey of green crabs are also considered as a form of Local Ecological Knowledge (LEK; Cosham, Beazley & McCarthy, 2016) providing further avenues for research into causes of crab population changes. Findings for green crabs in the present study are also related to concurrent changes in the study region from 2013 onwards reported for other species and their potential impacts.

Material and Methods

Results

Interannual changes in overall crab densities

The density of green crabs at Clarke Head, NS observed on the intertidal zone differed significantly overall among years from 2008–2017 (ANOVA, F_9,58 = 4.461, p < 0.001; Fig. 2). Intertidal densities increased from 2008 to 2010, when peak abundances were observed, remained relatively high in 2011 and 2012, and then rapidly declined to all-time lows in 2013 (Fig. 2). Densities remained low after 2013, but did gradually increase from 2014 to 2017 (Fig. 2). Densities were also highly variable among samples in 2008–2012, but from 2013 onward they were consistently low (compare error bars in Fig. 2). Intertidal densities were highest and above-average in the years 2010–2012, and lowest and below-average from the years 2013–2016 (Fig. 2). Intertidal densities in 2010 were significantly higher than those in 2013, 2014, and 2015 (Tukey’s HSD test, p ≤ 0.034) (Fig. 2). The difference between densities in 2010 and 2012 was marginally non-significant (Tukey’s HSD test, p = 0.053), and differences among densities in all other years were not statistically significant (Tukey’s HSD test, p ≥ 0.096) (Fig. 2). Average intertidal densities in 2013 in particular were much lower (by 13.3–91.4%) than those in other years; 2013 densities were significantly lower than those in 2010 and 2012 (Dunnet’s test, both p = 0.001), but did not significantly differ from those in any other year before (2008, 2009, 2011: Dunnet’s test, p ≥ 0.172) or after this (2014–2017: Dunnet’s test, p ≥ 0.809) year (Fig. 2).

Changes in population size structure, sex ratios, and recruitment among years

Green crabs of nearly all possible sizes, from juveniles ≤5 mm CW up to adult males of 86 mm CW, were found at Clarke Head, NS over the ten years of sampling, although crabs larger than 50 mm CW were always uncommon on the intertidal (Figs. 3A–3J). In most years preceding 2013, 83.4–91.0% of the crabs observed were mature (i.e., ≥ 1–2 years old and with CW ≥ 15 mm) (Fig. 4A), a wide range of crab sizes were observed (Figs. 3A–3E), and overall average sizes of crabs were relatively typical or large for an intertidal population (28.2–52.8 mm CW; Boudreau & Hamilton, 2012) (Fig. 4B). Notably higher proportions of sampled crabs in larger size classes were observed in 2010–2012 than in other years (Figs. 3A–3E), the same years in which crab abundances were also highest (Fig. 2). In 2013, a dramatic shift occurred in which the proportion of mature crabs in the area plummeted to 58.1% (Fig. 4A), mean sizes of all crabs found decreased to 15.7 mm CW (Fig. 4B), and crabs of CW > 30 mm almost completely disappeared from the intertidal (Fig. 3F). In subsequent years (2014–2017) larger crabs (up to 77 mm CW) began to reappear, but still remained much less abundant than in previous years (Figs. 3G–3J), and the proportion of mature crabs (56.3–68.8%) and mean overall crab sizes (17.1–26.6 mm CW) remained lower than they were before 2013 (Figs. 4A, 4B). Further, even with this gradual return of larger crabs total crab abundances (densities) remained low in 2014–2016 (Fig. 2); although in 2017 intertidal density did climb to a value near the overall average (Fig. 2). Mean sizes and size distributions of male, female, and ovigerous female crabs similarly shifted toward smaller sizes in 2013–2016 than in 2008–2012, with some sign of returns to pre-2013 values occurring in 2017 (Figs. 3F–3J, 4A, 4B).

The intertidal population sampled was in general female-dominated (average proportion of mature crabs female ± SE = 0.580 ± 0.082, maximum = 0.700 in 2010), except in one year (2014, proportion females = 0.438) (Fig. 4A). Sex ratios in this population fluctuated considerably among years, but with no clear overall pattern except perhaps a slight decrease in the proportion of females among sampled crabs from 0.670–0.700 in the earliest years sampled (2008–2010) to 0.438–0.614 in later years (Fig. 4A). About 50% of females observed were ovigerous in all years (average proportion = 0.540 ± 0.083, range = 0.357–0.629), and this proportion fluctuated among years without any clear pattern except that ovigerous females were notably rarer in 2012 than in other years (Fig. 4A). Ovigerous females tended to be larger than non-ovigerous ones in this population, and also were often larger on-average than males (Figs. 3A–3J, 4B), though it should be noted that very large males were also often present (Figs. 3A–3J).

The abundance of juvenile (CW < 15 mm) crabs observed on the intertidal at Clarke Head also varied considerably among years, but with no clear pattern (Fig. 5A). Juvenile numbers were notably very low in 2011 and 2013, and very high in 2012 (Fig. 5A). Subtidal CPUE of large crabs was relatively high and above-average from 2008–2012 and relatively low and below-average from 2013–2017 (Fig. 5B). Subtidal CPUE thus underwent similar fluctuations to intertidal densities (Fig. 2), although the year 2008 was an exception in which intertidal densities were relatively low (Fig. 2), while subtidal CPUE was average (Fig. 5B).

Potential correlates and causes of interannual changes to green crab abundances

Of the 257 environmental variables considered as potential predictors of changes in green crab abundances across years, 98 were significantly correlated (i.e., r <  − 0.238 or r >  + 0.238, N = 68, p ≤ 0.05) with intertidal crab densities (Supplemental Information). All of the predictors among those with the top ten strongest correlations had positive and significant r-values ≥ 0.5 (Figs. 6A–6I). The strongest correlation was a positive one between the maximum SST during March of the same year and intertidal density (r =  + 0.560, N = 68, p < 0.001; Fig. 6A). Five of the remaining predictors out of those with the ten strongest correlations with intertidal density included SST parameters for the early months (February–April) of the same year as that in which densities were measured (Figs. 6B, 6C, 6E–6G; Supplemental Information), specifically the maximum February SST, minimum March SST, and the mean, SD, and maximum of April SST. All of these SST predictors were positively correlated with density, meaning that higher SSTs during late winter-early spring led to higher densities, while lower SSTs led to lower densities (Figs. 6B, 6C, 6E–6G). The top ten predictors most-strongly correlated with density also included the mean wind speed in the previous year’s October (Fig. 6D), total annual settlement in the same year (r =  + 0.520, N = 68. p < 0.001; compare Fig. 2 vs. Fig. 5A), maximum air temperature in March of the same year (Fig. 6H), and the variability (SD) of total precipitation during July of the previous year (Fig. 6I), all of which were positively correlated with density. The wind direction variable most strongly correlated with density was the average wind direction in October of the previous year, which had the 12th highest r-value overall and was positively and significantly correlated with density (Fig. 6K). The most-strongly correlated NAO index variable was that for July of the previous year, which had the 36th highest r-value overall (see Supplemental Information) and was significantly and negatively correlated with density (Fig. 6L).

Additional interesting predictor variables showed correlations with lower, but still significant values (i.e., r between ± 0.238–0.5). For example, most SST and air temperature variables for the winter, spring, and summer months of the same year were positively correlated with intertidal density (see Supplemental Information), meaning that colder winters, springs, and summers resulted in lower crab numbers; however, several mean and minimum SSTs and air temperatures during the summer and early fall of the previous year (July–October) were significantly negatively correlated with density (see Supplemental Information), meaning that warmer summers resulted in lower intertidal densities in the following year. SSTs were generally better predictors of intertidal density than air temperatures were, while precipitation and especially wind variables were overall less well-correlated with densities than these (Figs. 6A–6C, 6E–6H, 6J; Supplemental Information). Crab density in a given year was also significantly correlated to the value of the North Atlantic Oscillation (NAO) index in certain months within the summer or early autumn of the same or previous year, but not in other seasons (Fig. 6L; see Supplemental Information). Specifically, higher or more positive values of the NAO index in the preceding year’s July and August and the same year’s June and August were associated with lower crab densities in a given year, while higher NAO index values in the preceding year’s September were associated with higher crab densities (see Supplemental Information); all other periods’ NAO index values were not significantly correlated with intertidal densities.

Stepwise multiple regression resulted in six models to describe changes in intertidal crab densities across years based on up to seven potential predictors, which represented the best individual predictors (i.e., highest r-values) of each type of variable examined (Table 3; Supplemental Information). The best model (#6) explained 40.7% of the interannual variation in intertidal density by relating it to two environmental predictor variables: total annual potential settlement and maximum March SST, both in the same year (Table 3). Both predictors included in the best model were not significantly correlated with one another, although some of those included in other non-best models were (see Supplemental Information). The variables in the best model therefore provided perhaps the best explanation of observed changes to intertidal crab densities. However, all models were much better than a null (#0, intercept-only) model (evidence ratio of all models versus the null = 10,823.771–693,128.818) and had comparable r²-values (0.369–0.407) to each other (Table 3). Also, although the best model (#6) carried most of the model weight (Δ_i = 0, w_i = 0.355), the second- and third-best models may also have contained additional useful information (Table 3). Specifically, the second-best model (#5), containing the same predictors as the best model in addition to the mean wind direction in October of the previous year, had an AIC_C value almost the same as that of the best model (Δ_i = 0.114) and a model weight of 0.335 (Table 3). The third-best model (#4), containing the same predictors as model #4 plus the NAO index value in July of the previous year, also had a Δ_i-value <2 (Δ_i = 0.997) and carried some weight (w_i = 0.215; Table 3). Therefore these other variables may also have had some influence on interannual fluctuations in this crab population.

Discussion

The present study presents a ten-year dataset obtained by monitoring a single, established population of invasive green crabs (Carcinus maenas) in the Minas Basin of the upper Bay of Fundy, Atlantic Canada. Monitoring of densities, size structure, sex ratios, and reproductive ratios of intertidal crabs revealed a marked wholesale decline in this population occurring within the year 2013. This decline in 2013 was associated with a shift in the population’s demographic structure towards smaller-sized and fewer sexually-mature or reproductive (i.e., ovigerous) crabs, and these changes have largely persisted through subsequent years. These changes to the crab population were significantly correlated with various environmental variables, suggesting some change to the physical conditions experienced by crabs in the area during or after 2013. Although correlations do not equal causation, the links established herein between interannual changes in crab biology and their physical environment are highly suggestive, and provide directions that can be pursued by future research.

There are several important aspects of the monitoring program that could be improved in the future and may have led to some uncertainty regarding specific numbers (e.g., crab densities, water temperatures) in the present study. Most importantly, the use of daytime intertidal quadrats to simplify the monitoring of the population has the potential to have missed important trends and to be inconsistent with results of other studies of crabs in the study region that used more standard methodologies (e.g., Fukui or pitfall traps; MacDonald et al., 2018). Crabs would almost certainly have been more abundant and present over other areas of the intertidal (not just in shelters among reefs) if sampling had been done at night, when crabs are more active and likely to stay on the intertidal due to lower air temperatures, water stress, and visibility to terrestrial predators compared to daytime (Crothers, 1968; Klassen & Locke, 2007). LEK observations reported by local fishers of reduced crab activities during nighttime worm harvesting activities in and after 2013 do suggest that, even if sampling had been done at night, the reduction in crab numbers in 2013 would have been observed; although this is unconfirmed at present. The use of simplified intertidal quadrats instead of standard Fukui traps in this study means that the densities quantified herein are not directly comparable to those collected in other studies in the region (e.g., MacDonald et al., 2018), which does impair the possibility of direct comparisons across studies. Crabs also undergo tidal migrations (Hunter & Naylor, 1993), and because sampling focused mainly on intertidal crabs no information was obtained regarding what proportion of the crab population left the intertidal zone for the subtidal on ebb tides, and whether this changed among years. Declines in the CPUE of large (>50 mm CW) subtidal crabs caught during flounder fishing trips in and after 2013 were noted, which lent support to the conclusion that this decline was a real event that impacted the entire local crab population, and not just the result of crabs migrating away from the intertidal for some reason more in later years (Hunter & Naylor, 1993). However, because the catchability of green crabs by flounder fishing hooks was not formally quantified (though it did seem to be high, see Methods) and the proportion of crabs leaving the intertidal at low tides was unknown, these results are not without some degree of uncertainty. Future monitoring work within this population should attempt to quantify these variables, as well as attempt to sample crab densities at night and using Fukui traps, to confirm the conclusions drawn herein and improve the quality of future data obtained. A more thorough and formal analysis of spatial and temporal variations in crab numbers at this site (among tidal elevations (Scrosati, Grant & Brewster, 2012), sampling dates within a year, etc.) would likely also be worthwhile. Direct monitoring of environmental variables at the site, such as temperatures (air at low tide only, and water at high tide), could also improve future analyses of the drivers of population changes. However, while specific numbers may not have been assessed with perfect accuracy, the magnitude of the change observed at and after 2013 as captured over this ten-year monitoring program is unequivocally a real and important result.

The lack of substantial recovery of this population to its former characteristics from 2014 to 2017 could be a result either of slow growth due to reduced reproductive capacity (i.e., loss of larger, mature individuals), and/or due to the cause of the initial decline still being in play within the environment inhabited by these crabs. Better understanding of the potential causes of this decline is needed to fully examine the scope and implications of this change. However, the negative implications of the observed loss of breeding stock to this population’s future growth and stability is obvious, particularly with the loss of larger individuals that are thought to have considerably greater size-specific-fecundity in decapod crustaceans than smaller breeders (Somers, 1991; McGaw, Edgell & Kaiser, 2011). The ability of crabs to feed on shelled prey is also strongly size-dependent, such that smaller crabs have access to a more limited suite of prey than larger one (Elner & Hughes, 1978), so the shift in population size structure may also have implications to crab foraging ecology and to prey species and their communities in the region. Even in 2017, when observed overall crab densities appear to have begun approaching ‘normal’ levels for this study site and period and some large crabs were present on the intertidal, juvenile crabs still made up a substantial portion of the population and average sizes were smaller than in pre-2013 conditions. Whether this population will fully recover and how long it will require to do so are open questions, but because crabs interact strongly with many species (e.g., Klassen & Locke, 2007; Boudreau & Hamilton, 2012) if recovery is slow or does not occur the ecology of the study area could change considerably; thus monitoring should and will continue.

It is important to note that intertidal crab densities at Clarke Head were comparably low in 2008 to those seen in and after 2013, and these low densities were associated with some similar environmental factors to those putatively involved in the 2013 decline (e.g., oceanographic currents leading to low predicted settlement). However, lower abundances in 2008 were not associated with differences in the size or reproductive structure of this population or low subtidal CPUE versus other years, whereas in 2013 larger, sexually mature crabs become extremely scarce and subtidal CPUE was very low. One could postulate some similar factors having been in play in 2007–2008 to those in 2012–2013, but with additional factors from 2012–2013 and thereafter (discussed below) leading to particularly high losses of larger, breeding crabs. As a result, the population recovered quickly in 2009 from 2008 declines, but as of 2017 had not fully-recovered from the 2013 collapse.

Results of this study showing population declines of a benthic marine invertebrate in 2013 agree with findings of other studies and species in the Bay of Fundy-Gulf of Maine region. Kienzle (2015) and MacDonald et al. (2018) noted a decline in abundances of green crabs inhabiting sites in Chignecto Bay, a nearby but distinct branch of the upper Bay of Fundy to the Minas Basin (Fig. 1), from 2013 to 2014. Monitoring of early benthic recruitment (i.e., settlement) of American lobster (Homarus americanus) juveniles by the American Lobster Settlement Index (ALSI) recorded record-low recruitment of juveniles in the year 2013 throughout most of the Bay of Fundy, Gulf of Maine, and southwestern Nova Scotia (Wahle & Carloni, 2017). Lobster recruitment since 2013 in these areas has remained low compared to its pre-2013 values and appears to have been even lower than in 2013 in the year 2016 (Wahle & Carloni, 2017). Potential settlement of lobster larvae in the Gulf of Maine-Bay of Fundy system and the Scotian Shelf has also been predicted by a bio-physical dispersal model of the species’ range (Quinn, Chassé & Rochette, 2017) to have been extremely low in 2013 (BK Quinn, J Chassé, and R Rochette, pers. comm., 2017). Benthic recruitment of soft-shell clam (Mya arenaria) settlers monitored on mudflats in the Bay of Fundy was also extremely low in 2013 compared to earlier years (Clements, 2016). Along with the present study, these various observations point to the occurrence of some sort of large-scale oceanographic event in the region of the Bay of Fundy during or around the year 2013. This event appears to have negatively impacted numerous marine species in the region, and as of 2017 many species’ populations have still not completely recovered. If such an event occurred, what were the causes or mechanisms? Results of the present study and postulated causes of declines reported in others may provide some hints, as discussed below.

Recent climate change has had important impacts on thermal variability within the atmosphere and the ocean, including warmer summers and periodically colder, harsher winters in temperate regions such as Atlantic Canada (Petrie, 2007; Mills et al., 2013; Kienzle, 2015; MacDonald et al., 2018). When extremes of seasonal temperatures coincide due to such climatic changes, the effects on biota can be considerable (e.g. Quinn, 2016). Excessively hot summers in recent years may have played a role in observed declines because heat stress leads to decreased performance and survival in decapod crustaceans, especially in larger-bodied benthic life stages that tend to have lower heat tolerances (e.g., Byrne, 2011). The Gulf of Maine is now widely acknowledged to have experienced an ‘ocean heat wave’ in 2012, which led to negative heat stress effects on many species, and changes to phenology (e.g., moulting by lobsters) of many others (Mills et al., 2013). This heat wave likely contributed to results of the present study, as higher temperatures during the summer months of a given year were correlated with lower crab densities in the following year, with 2012, the year preceding 2013, having one of the hottest summers recorded (see Supplemental Information). Of course, crab densities in the monitored population were actually quite high during the summer of 2012, implying that rather than being stressed crabs were experiencing temperatures close to optimal for their survival and growth. Although it is important to note that ovigerous females were less frequently encountered on the intertidal in 2012 compared to other years, which may mean that this very warm summer had negative impacts on the survival or reproduction of females. Negative effects of the 2012 heat wave on crab abundances in 2013 may alternatively have occurred via some indirect means, such as effects on prey or predator species, altered phenology, or perhaps accumulated sub-lethal stress effects from the previous year if supra-optimal temperatures had occurred.

Cold winters and springs may also have contributed to crab declines in 2013 onward, since crab survival at all life stages—but especially overwintering survival of young-of-the-year—is adversely affected by low (<6–10 °C) temperatures (Berrill, 1982; Klassen & Locke, 2007). Therefore, Kienzle (2015) and MacDonald et al. (2018) postulated that especially harsh, cold conditions during the winter from 2013–2014 likely contributed to the declines they observed in green crab numbers in Chignecto Bay. In the present study, crab intertidal densities in a given year were also found to be positively correlated with mean temperatures during the preceding winter and early spring months (i.e., colder winter and spring = fewer crabs); while the data examined herein do not suggest that the winter of 2013 was particularly cold (at least for the studied part of the Minas Basin), it was colder than that of preceding years, and the winters of 2014 and 2015 were notably cold (see Figs. 6A–6C, 6E–6H, 6J). However, it is possible that the combined negative effects of the very warm summer of 2012 on larger adult life stages and the somewhat cold winter of 2013 on earlier life stages could have led to some of the declines observed; this disparity in thermal extremes may also explain why numbers of larger crabs, which would have overwintered and experienced both extreme temperature conditions, were observed to have declined relatively more than smaller crabs in 2013.

Other environmental parameters found in this study to be related to lower crab abundances included greater wind speeds from a more southeasterly (i.e., towards a more northwestern direction, of higher degree values relative to north) in the preceding autumn (October) and greater variability in precipitation during the early summer (July) of the previous year (for more details, see ‘Results’, especially Table 3 and Figs. 6D, 6I, 6K). The exact roles of these factors and other wind- and precipitation-related predictors (see Supplemental Information) are complex, as different impacts in different seasons may affect one or more different life stages. Winds and precipitation, for example, may signal storm frequency, with more frequent storms being expected to have negative impacts on both planktonic larvae and benthic adults and juveniles (e.g., Jury, Howell & Watson III, 1995; Moksnes et al., 2014). Wind speed and direction impact surface currents and water circulation, so during the summer and autumn these can impact whether and how many larvae are retained versus dispersed (Bertness, Gaines & Wahle, 1996), thus affecting settlement and recruitment to benthic populations (Crothers, 1967; Moksnes et al., 2014). Reduced salinity resulting from high precipitation can adversely affect physiological health and survival of larvae in the surface waters directly and benthic life stages indirectly via runoff (Nagaraj, 1993; McGaw, Reiber & Guadagnoli, 1999; Klassen & Locke, 2007). The positive correlations found in this study of crab abundance with summer precipitation and autumn winds are somewhat unexpected. It is conceivable that these atmospheric forces alter circulation at or beneath the sea surface in such a way that is actually beneficial to crabs—for example, by directing settling larvae shoreward or forcing saline water to flow into the estuary in response to wind-induced surface currents. Indeed, given the orientation of Clarke Head (Fig. 1C) more southeasterly winds in the autumn, blowing towards the northwest, would potentially help to direct larvae in surface currents toward the shore at this site and facilitate greater settlement there, while winds in the opposite direction could impede larval transport toward the coast. The strong positive correlation between model-predicted settlement—essentially a proxy for oceanography—and crab abundances lends some support to this supposition. However, this requires further study of the interaction of weather and oceanography in the study area to be confirmed and clarified.

It is interesting that in years in which oceanographic conditions leading to higher potential settlement occurred the abundance of crabs older than one year (>15–20 mm CW; Crothers, 1967; Crothers, 1968; Klassen & Locke, 2007) was also higher, and conversely when there was less settlement older crab abundances were also low. Clearly decreased settlement cannot explain low mature crab abundances in years such as 2013. However, these results do offer the intriguing suggestion that oceanographic conditions favouring settlement of larvae in the studied region also favour survival of older crabs. Given that the study site is located in a relatively shallow, well-mixed estuary, it is conceivable that this might be the case.

If indeed some large-scale oceanographic event was responsible for crab declines in 2013 and declines of other species in the region, one of the environmental variables examined which would be expected to have had an impact was the value of the NAO index. The winter NAO in particular has been found to be strongly associated with climate and ocean characteristics (Hurrell et al., 2003), such that in years with highly negative values of the winter NAO index bottom waters in the Bay of Fundy-Gulf of Maine system (including the southwest NS Shelf) have relatively low salinity and are colder than average, whereas in years with strong positive winter NAO index values waters in this region become warmer and more saline (Petrie, 2007). Negative winter NAO index-years also tend to have more frequent storms and stronger offshore-directed currents in this region than positive years (Hurrell et al., 2003; Petrie, 2007). The year 2013 had somewhat negative NAO index values during the winter months (see Supplemental Information), which may mean that conditions in the Bay of Fundy were unfavourable for survival of benthic marine species and retention and recruitment of their planktonic larvae; observations cited for various species in this region certainly suggest that NAO is negatively correlated to abundance and population recruitment (Kienzle, 2015; Clements, 2016; Wahle & Carloni, 2017). However, even greater negative values of the NAO index occurred in the winter of 2010, when summer crab densities observed herein were at their highest. Indeed, while significant correlations between crab densities and NAO index values for the summer months were found in this study, most were negative, such that more negative NAO indices were associated with higher crab abundances. However, NAO index values in winter months and other seasons were not significantly correlated with crab densities. This is unexpected, but does agree with other correlations (e.g., the positive relationship between precipitation and wind speeds during spring and summer months and crab densities; see Supplemental Information). Clearly the relationship between the NAO index and localized recruitment is thus not as simple or direct in all cases as might be supposed. Perhaps the Minas Basin, being relatively isolated from the rest of the Bay of Fundy-Gulf of Maine system, is less affected by NAO-associated shifts in regional oceanography or climatology. Future studies should further examine the relationship between the NAO index and recruitment in other marine species in the region to clarify these complexities.

Crab populations could alternatively have declined due to biotic changes in the studied region, such as those cited by local recreational fishers concerned with other species potentially interacting with crabs. Increased presence in the area of two finfishes known to feed on benthic crustaceans such as smaller crabs (Klein-MacPhee, 2002; Munroe, 2002) in 2012–2013 may have played a part in crab declines. Cunner (Tautogolabrus adspersus) were largely absent from the inner Bay of Fundy prior to 2012 (Munroe, 2002), but then in 2012 they began to penetrate deeper into the bay (Woodard, 2018). Cunner numbers in the Bay of Fundy have since been lower than those in 2012, but nonzero (Woodard, 2018). Longhorn sculpin (Myoxocephalus octodecemspinosus) are native to the Minas Basin (Klein-MacPhee, 2002; Parker, Westhead & Service, 2007), but from 2013 onward they have become extremely abundant in the Minas Basin. The extent to which these fish species prey on green crabs and their impact on crab populations needs to be better quantified to determine whether predation by them could be responsible for crab declines. Declines to populations of species preyed on by green crabs in the area, such as the decline in sandworm (Alitta virens) stocks reported by recreational harvesters, may also be related to crab declines. Green crabs will prey on sandworms (Crothers, 1968; Klassen & Locke, 2007), so it is quite possible that predation by the invasive crab population on these worms in combination with worm harvest (potentially overharvest) by humans (Miller, 2009) could have contributed to the worm population’s collapse (though this needs further study). However, green crabs are quite capable of feeding on a wide variety of prey species also present off of Clarke Head, including blue mussels (Mytilus spp.), barnacles (Semibalanus balanoides), periwinkles (Littorina spp.), dogwhelks (Nucella lapillus), soft-shell clams (M. arenaria), and many others (Klassen & Locke, 2007). All of these species were present and quite abundant throughout the 2008–2017 period (BK Quinn, pers. obs., 2013–2017), so even if crabs had relied heavily on sandworms as a food source they should have been readily able to switch their prey to any of the various other species present (Huntingford & Taylor, 1997) when the worm stock declined in 2013. While unconfirmed, these observations by locals provide interesting directions for future research to pursue.

Some additional or alternative localized factors that may also have been impacting crab populations within the Minas Basin during or since 2013 could include: development of tidal power within the Minas Channel, the only means of connectivity between the Minas Basin and the remainder of the Bay of Fundy (Parker, Westhead & Service, 2007; Copping et al., 2016); increasing eutrophication in recent years within the adjacent estuary, causing harmful algal blooms (‘red tides’) that cause anoxia and produce toxins potentially harmful to crustacean development (e.g., copepod: Miralto et al., 1999); effects of ocean acidification on crab exoskeleton formation and moulting (Clements, 2016; Miller et al., 2016); changes in the phenology (timing) of planktonic productivity on which larval diet and survival depends (Scrosati & Ellrich, 2016; Wahle & Carloni, 2017); or others. However, the impact of all of these listed factors requires further directed study before a link to crab declines can be drawn.

Conclusions

To conclude, there is much evidence that is suggestive that some sort of physical oceanographic and/or climatic event occurred in or around 2013 and impacted marine biota in the Bay of Fundy and Gulf of Maine, and this should certainly be investigated further. Results of the present study strongly suggest that the observed decline of the green crab population at Clarke Head was at least partially due to this event. There are important demographic and ecological implications to other benthic species in the Bay of Fundy-Gulf of Maine region if they have been or are still being impacted similarly to this green crab population by an oceanographic event in or beginning in 2013. If invasive species are indeed more vulnerable to climatic changes than native ones (e.g., Kienzle, 2015; MacDonald et al., 2018), then these observations of green crab declines should serve as a warning of potential changes to other species in the study region, such as native rock crabs or American lobsters. Continued monitoring of this population and further study of other species will be needed to confirm the causes of these observed shifts, their impacts, and whether populations will continue to be negatively impacted.

Supplemental Information