The brighter side of climate change: How local oceanography amplified a lobster boom in the Gulf of Maine

Abstract Ocean warming can drive poleward shifts of commercially important species with potentially significant economic impacts. Nowhere are those impacts greater than in the Gulf of Maine where North America's most valuable marine species, the American lobster (Homarus americanus Milne Edwards), has thrived for decades. However, there are growing concerns that regional maritime economies will suffer as monitored shallow water young‐of‐year lobsters decline and landings shift to the northeast. We examine how the interplay of ocean warming, tidal mixing, and larval behavior results in a brighter side of climate change. Since the 1980s lobster stocks have increased fivefold. We suggest that this increase resulted from a complex interplay between lobster larvae settlement behavior, climate change, and local oceanographic conditions. Specifically, postlarval sounding behavior is confined to a thermal envelope above 12°C and below 20°C. Summer thermally stratified surface waters in southwestern regions have historically been well within the settlement thermal envelope. Although surface layers are warming fastest in this region, the steep depth‐wise temperature gradient caused thermally suitable areas for larval settlement to expand only modestly. This contrasts with the northeast where strong tidal mixing prevents thermal stratification and recent ocean warming has made an expansive area of seabed more favorable for larval settlement. Recent declines in lobster settlement densities observed at shallow monitoring sites correlate with the expanded area of thermally suitable habitat associated with warmer summers. This leads us to hypothesize that the expanded area of suitable habitat may help explain strong lobster population increases in this region over the last decade and offset potential future declines. It also suggests that the fate of fisheries in a changing climate requires understanding local interaction between life stage‐specific biological thresholds and finer scale oceanographic processes.


| INTRODUC TI ON
Global climate change is an information dense, but deceptively complicated, phrase. Most of the world's oceans are warming, but patterns in the distribution, abundance, and rate of warming varies in space, time, and in ways that can have profound effects on marine organisms (Kleisner et al., 2016;Pinsky, Worm, Fogarty, Sarmiento, & Levin, 2013). Several studies have recorded the "climate velocity" of the net poleward shifts in the distribution of marine organisms that are commercially fished (Nye, Link, Hare, & Overholtz, 2009;Pinsky et al., 2013) or ecologically important to marine ecosystems (such as coral reef fish: Vergés et al., 2014;and kelp forests: Wernberg et al., 2010). However, to assess the effects of ocean warming requires understanding how those aspects of climate change affect organisms where they live. Most studies use trends in sea surface temperature (SST) as a convenient and important measure of local warming (e.g., Pershing et al., 2015), but such remote sensing metrics may not be indicative of exposure for organisms with key life history processes occurring on or near the seabed. Such is the case of the northwest Atlantic coastal and shelf waters where thermal properties of sea temperature vary dramatically in time and space (Kavanaugh, Rheuban, Luis, & Doney, 2017).
We focus on the American lobster (Homarus americanus Milne Edwards) because it is the most valuable single-species fishery in North America, with a combined value of more than $US 1.67 billion in 2016 (DFO, 2018;NMFS, 2018), and it is also one of the best studied organisms in the world. Almost a century ago, Huntsman (1923) surmised that despite thriving lobster fisheries, the absence of juvenile lobster in the Bay of Fundy and northeastern Gulf of Maine could be attributed to the colder waters preventing larval development and settlement. Studies over the last three decades determined that the demography of this species is driven by the successful settlement of its pelagic postlarval stage on the seabed (Barshaw & Bryant-Rich, 1988;Boudreau, Bourget, & Simard, 1990;Boudreau, Simard, & Bourget, 1991;Cobb, Gulbransen, Phillips, Wang, & Syslo, 1983;Incze & Wahle, 1991;Palma, Steneck, & Wilson, 1999;Steneck & Wilson, 2001;Wahle & Steneck, 1991). More recently, field observations by Annis (2005) determined that postlarvae sounding behavior is restricted to temperatures warmer than 12°C (i.e., the "thermal threshold" sensu Annis, 2005), an observation consistent with subsequent deep-water settlement surveys (Wahle, Brown, & Hovel, 2013).
Thermal thresholds act in ways that directly impact species fitness and demographic processes. The 12°C threshold for lobster indicates a point at which colder temperatures induce lethal (e.g., irregular respiration and heartbeat; Quinn, 2017) and sublethal (increased time spent in the plankton: MacKenzie, 1988; behavioral avoidance and decreased size at molt: Annis, Wilson, Russell, & Yund, 2013) effects. Small temperature changes about this threshold correspond with large changes in survival and settlement which decouples larval supply from settlement patterns (Annis et al., 2013). Thus, the 12°C threshold can act as an ecological barrier to larval transport (Tilburg, McCartney, & Yund, 2012) and survival. Ocean warming, therefore, has the potential to significantly impact the distribution and survival of lobster larvae. This is especially the case in the Gulf of Maine where end-of-century projections suggest an expanded area and time the seabed spends above the thermal threshold .
Here, we examine the interrelationships of seabed temperatures along the oceanographically distinct Gulf of Maine and how they relate to lobster demography after larval settlement. Specifically, we suggest that the combination of thermally mediated sounding behavior of lobsters, particularly in areas susceptible to ocean stratification, decouples the easily measured SST from the more relevant bottom temperatures. This not only paints a different picture of how ocean warming may affect the distribution and abundance of the American lobster but it also alters how we interpret decades of young-of-year (YoY) settlement data that have consistently declined for approximately a decade in the Gulf of Maine. At stake is whether the single most valuable fisheries species in North America is on a trajectory of decline.

| Study area
The Gulf of Maine ( Figure 1a) is a semienclosed continental shelf sea that is bounded by Cape Cod, Massachusetts, USA, and Nova Scotia, Canada ( Figure 1a). It has supported some of the most productive fisheries in the world; most notably groundfish (e.g., Atlantic cod; Alexander et al., 2009), ocean scallop (MDMR, 2018NMFS, 2018), and lobster (DFO, 2018;MDMR, 2018;NMFS, 2018). Currently, the American lobster supports the most valuable fishery in the United States and Canada, and 90% of the US production comes from the Gulf of Maine (NMFS, 2018). The region's high biological productivity (Bigelow, 1926;O'Reilly & Busch, 1984) derives from nutrient rich, deep slope water (Townsend, Thomas, Mayer, Thomas, & Quinlan, 2006) and Scotian Shelf water (Townsend, 1997(Townsend, , 1998 flowing into the Gulf of Maine near Nova Scotia. After these waters enter the Gulf of Maine, the Eastern Maine Coastal Current is generated by cyclonic baroclinic circulation directing water toward the Bay of Fundy (Brooks, 1985; between New Brunswick and Nova Scotia), intense mixing due to some of the world's largest and strongest tides (Townsend et al., 2006), flow along the coast of eastern Maine, and an offshore plume just prior to Penobscot Bay (Bisagni, Gifford, & Ruhsam, 1996;Brooks & Townsend, 1989;Pettigrew et al., 1998;Townsend, Christensen, Stevenson, Graham, & Chenoweth, 1987).

Southwestward of Penobscot Bay, the Western Maine Coastal
Current is generated by freshwater discharge from rivers aided by winds and cyclonic baroclinic circulation (Bigelow, 1927;Brooks, 1985;Fong, Geyer, & Signell, 1997;Franks & Anderson, 1992). These two current systems act to divide the coast of the Gulf of Maine into northeastern and southwestern sectors, each having unique physical characteristics; colder and well mixed to the northeast, warmer and stratified to the southwest.

| Lobster landings
The State of Maine lobster fishery has one of the longest time series of lobster landings in the world dating back to 1880 (MDMR, 2018). To the extent that landings (integrated over several years) are a reliable predictor of lobster population size (Steneck & Wilson, 2001), these historical data help us identify population trends over the past century. Maine landings were used to identify the time frame over which landings have changed. To compare landing trends from areas with very different coastline lengths along the coast of the Gulf of Maine, landings over this time frame were standardized per unit length of coastline (e.g., Steneck & Wilson, 2001). Specifically, we incorporated landing

| Characterizing thermal habitat
Many recent studies necessarily use SST to characterize temperature exposure of benthic organisms. In this study, we used numerical model output that assimilates SST but has the ad-  (Chen, Beardsley, & Cowles, 2006). NECOFS is well suited to simulate geophysical marine environments characterized by complex coastlines due to its flexibility in mass conservation, triangular grid geometric flexibility, and computational efficiency (Chen et al., 2006). In this study, the NECOFS FVCOM-G3 grid was employed. The unstructured NECOFS FVCOM-G3 grid provides a horizontal resolution ranging between ~20 m inshore and ~10 km at the model openboundary (Chen et al., 2006). Hourly bottom temperature data are modeled at 48,451 nodes and demonstrate a high reliability to predict bottom water temperatures (Li, Tanaka, Chen, Brady, & Thomas, 2017).

| The American Lobster Settlement Index
The American Lobster Settlement Index (ALSI) is an annual monitoring program that quantitates settlement densities of YoY lobster across New England and Atlantic Canada. ALSI was initiated in 1989 at a few sampling locations in midcoast Maine (Incze & Wahle, 1991) and has since expanded to more than 100 sites ranging from Rhode Island, USA, to Newfoundland, CAN sampled either by diver-based suction sampling (Incze & Wahle, 1991;Wahle & Steneck, 1991) Figure S1).

| Thermal habitat and YoY settlement
We reasoned that if the area of thermally suitable seabed varied in time, interannual variability in YoY density alone could misrepresent the true time trend in lobster year class strength. One consequence of expanding habitat could be that larval settlement spreads over a larger area with correspondingly reduced densities while extrapolated abundances may be stable or even increasing. We refer to this as the "thermally mediated dilution hypothesis." As a partial test of that hypothesis, we evaluated the relationship between the area of seabed warmer than 12°C ( (Carloni, Wahle, Geoghegan, & Bjorkstedt, 2018) and demersal fish abundance (Wahle, Bergeron, et al., 2013) effects on natural mortality of pre-and postlarvae, spatiotemporal variability of competent larvae and suitable thermal habitat, proximity to land boundaries (Steneck & Wilson, 2001), sediment availability and use, and fixed station sampling that may underrepresent annual lobster recruitment (Li, Cao, Chang, Wilson, & Chen, 2015).
We estimated the depth over which lobster settlement is relatively homogenous by reanalyzing data from Wahle, Bergeron, et al. (2013). Our analysis focused on data from the northeastern Gulf of Maine where temperature was uniform across depths and was unlikely to influence depth-wise patterns of settlement. We standardized annual settlement densities over the three depth strata (0-25, 25-50, and 50-100 m) to shallow settlement densities over the study period (2007)(2008). We then performed a one-way ANOVA and Tukey-Kramer post hoc test to determine the depth at which settlement density significantly deviated from shallow settlement densities. Based on this analysis, we chose 50 m as the limit for the relatively uniform distribution of YoY lobster over suitable thermal habitats ( Figure S2).

| Extrapolating YoY settlement over thermally suitable habitats
On the strength of the inverse relationship between YoY density and area of thermally suitable habitat, we generated two additional indices of YoY abundance, which extrapolate YoY density over areas of a thermally suitable seabed in different ways. The first assumes uniform settlement across depths shallower than 50 m (Wahle, Bergeron, et al., 2013;see above). In this case, extrapolated recruitment, R, was taken as the product of YoY settlement density, ρ , from shallow diver-sampled locations and the area, A, of thermally suitable sea bed subject to temperatures >12°C: The second scales YoY settlement density across three depth strata  Figure S3). Since relative settlement success is not known for alternative habitat types (Chassé & Miller, 2010), we assumed binary recruitment success: whereby the proportion recruited onto rocky habitat is 1 and nonrock habitat is 0. Maine. Soon thereafter, the northeastern coastal fishery entered a phase of geometric growth, greatly surpassing landings in the southwest even as they continued to grow, so that by 2015 the northeastern landings per unit coastline were 80% higher than in the southwest (Figure 1c).

| RE SULTS
Northeast-southwest differences in thermal regime, water column structure, and ocean warming are largely driven by differences  Table S3).
Modeled surface-to-bottom temperature differences are further corroborated by in situ buoy measurements that are significantly linearly related ( Figure S4). Water column temperature differences are  (Table S1). Similar inverse relationships between surface and bottom temperatures to settlement density were found, indicating that periods of warmer ocean temperatures correspond with periods of low settlement density (Table S2) (Tukey, 1977); however, similar extreme recruitment events are common in marine populations (e.g., Hjort, 1914). In

| D ISCUSS I ON
We illustrated that lobster landings have been increasing for decades with the greatest expansion in northeastern Maine (Figure 1c). This is consistent with studies of climate velocities attributed to rising ocean temperatures (Pinsky et al., 2013) and contrasting rates of landings increase in the Gulf of Maine. However, often, the mechanisms driving these changes in distribution are unknown or difficult to quantitate. SST, used in most studies of climate-fisheries interactions (e.g., Pershing et al., 2015), offer generalized trends in ocean conditions. It is only when we couple these trends to biological thresholds that we get closer to process-level understanding and improve our predictive capability. We considered warming in the context of thermally mediated lobster settlement behavior (Annis, 2005). It is influenced by the strong tidal mixing in the northeast region of the Gulf of Maine (Figure 2), and we offer a mechanistic explanation for an economically important species that has been shown to have settlement-driven demography (e.g., Incze et al., 1997;Palma et al., 1999;Wahle, Bergeron, et al., 2013;. We demonstrated that the impacts of climate change are spatially variable in the Gulf of Maine due to local oceanographic conditions that change the susceptibility of nearshore areas to warming. Northeastern Gulf of Maine warming is largely uncoupled to synoptic conditions due to strong tidal mixing (Mountain et al., 1996), and is most likely driven by offshore advective components of heat flux.

However, interannual variability in surface heat flux in southwestern
Gulf of Maine is highly correlated with local heat flux due to the importance of stratification in this region (Figure 2).  (Table S2), but these declines may have been offset by expanded thermal habitats ( Figure 3) and increased depth of larval sounding behavior (Annis, 2005). To the extent that trends in recruitment predict trends in landings, these results suggest an alternate future for the American lobster fishery, which has been predicted to be on the verge of an imminent decline due to the decline in the density of observed settlers (Le Bris et al., 2018;Oppenheim, 2016). Our two methods of estimating newly settled lobster abundance suggest relative stasis   Bris et al., 2018;Sherman et al., 2015) facilitated by an increased area of suitable habitat offshore (Tanaka & Chen, 2016). Since the abundance of stage-one larvae has been shown to be significantly correlated with the abundance of spawning stock biomass (Carloni et al., 2018)  upper threshold and has been associated with physiological stress, rising prevalence of epizootic disease (e.g., Castro & Angell, 2000;Glenn & Pugh, 2006), and with mass mortalities (Pearce & Balcom, 2005). The widespread mortality in southern New England contributed to the reduced reproductive potential and settlement (Wahle, Dellinger, Olszewski, & Jekielek, 2015). Southwestern-most parts of the Gulf of Maine occasionally exceed 20°C, but these periods are relatively short in comparison to southern New England where regions can exceed this threshold for over 2 months (Wahle, Bergeron, et al., 2013). Northeastern parts of the Gulf of Maine rarely exceed this upper threshold, and the relatively stable thermal regime may buffer this part of the Gulf of Maine from lobster declines related to excessive future warming. Poleward migration of other taxa including predatory fishes will have unknown impacts on American lobsters (Nye et al., 2009;Wahle, Bergeron, et al., 2013). Nevertheless, our understanding of the role of climate change on marine organism distribution requires a better understanding of important details of both the oceanographic and organism attributes.
Clearly, the role of climate change on the distribution and abundance of organisms is steadily gaining attention in all sectors from the public to managers and policy makers. Rates of poleward migration or "climate velocity" of commercially important species have been well documented (Nye et al., 2009;Pinsky & Fogarty, 2012) and generally conform to well-established patterns of thermogeography driving biogeography (Adey & Steneck, 2001;Hutchins, 1947). However, beyond simple patterns of temperature and organism distribution are complex processes and mechanisms (such as a thermal envelope) driving those patterns. Our proposed mechanistic explanation of climate velocities incorporates warming in the context of realized essential habitat (e.g., nursery habitat) and biological thresholds for a commercially important species. When seabed temperatures exceed the thermal threshold, the probability of lobster settlement onto previously cooler and uninhabitable areas increased. However, contrasting stratified and unstratified water masses differentially affects seabed warming and thus lobster recruitment.
Global models of climate change will continue to provide a valuable coarse-scale view of the impact of ocean warming on world fisheries, but they do not capture finer scale coastal processes that surely affect how fishery productivity will play out on a local scale.
We demonstrate here that local oceanography, coupled with nonlinear species-specific responses to warming, may have led to a demographic expansion disproportionate to the relatively small change in temperature. By virtue of local differences in vertical mixing, the Gulf of Maine has different susceptibilities to offshore warming that can lead to differences in future productivity. In the case of the American lobster, we argue that ocean warming drove a northeastward population surge in the Gulf resulting from an expansion of the area of seabed across a biologically important thermal threshold for larval settlement.
We suggest this northeastward expansion largely contributed to the historic sixfold increase in lobster harvests that has elevated the fishery to its high national ranking in value. We, therefore, argue that the fate of fisheries in a changing climate requires a better understanding of interactions between local oceanography and species traits.

ACK N OWLED G EM ENTS
This activity is supported by National Science Foundation award