Potential socioeconomic impacts from ocean acidification and climate change effects on Atlantic Canadian fisheries

tonnage), and for the RCP 8.5 scenario the total projected loss nearly doubles to 12,200 tonnes relative to year 2000 landings (representing a 3.3% reduction in total landings) (Fig 7). In both climate scenarios this was mainly driven by the declines in snow crab, with the RCP 8.5 sce- nario results compounded by more substantial losses in sea scallop and Stimpson’s surf clam (Fig 7). While the cumulative changes across the region were minimal, at the sub-regional scale dif- ferent patterns emerged. Latitudinal gradients appeared to be the main driver of species distri- bution with an overall northward trend apparent for most species (Fig 8; see also S1 Fig). Overall, the Maritime management area is expected to see losses for most species (except Stimpsons’ surf clam, which experienced minimal change in all scenarios). While in the Gulf and Newfoundland and Labrador management areas a mix of changes is anticipated, with overall increases slightly outweighing declines (Fig 7). Maritime management area. The three species in the Maritime management area with greatest landing weights in 2000 (American lobster, northern shrimp, and sea scallop) were all projected to decline under the quartet of exposure projections. Declines were more extreme in 2090 and under the RCP 8.5 climate scenario (Fig 7). American lobster is projected to experi- ence the most significant loss in landings (54%). Slightly lower levels of decline were projected for sea scallop (48%) and northern shrimp (42%). Importantly, the projected declines for these

commercially important species under the RCP 8.5 scenario are roughly double (or more than) the declines projected under the RCP 2.6 scenario (lobster 29%, scallop 13% and north- ern shrimp 17% in 2090, relative to 2000) (Fig 7). The remaining species contributed relatively little to the Maritime management area land- ings and presented variable changes in their respective landings. Under the RCP 2.6 scenario snow crab had the most extreme relative decline (21%) but the loss was somewhat smaller under RCP 8.5 (16%). Even so, this species contributed a comparatively small portion of the total landings to the Maritime management area in 2000 (Fig 7). Eastern blue mussel landings also declined under all treatments, but this species was nearly non-existent in the baseline data. The two species (American oyster and Stimpsons’ surf clam) that had potential gains in landings also represent comparatively low total landings for the region. American oyster land- ings in the area are negligible with only ~140 tonnes of production in 2000, thus the ~10% increase predicted under RCP 8.5 for 2090 amounts to an approximate gain of 15 tonnes (Fig 7). Stimpsons’ surf clam landings change negligibly under all treatments. Gulf management area. Care should be taken when interpreting model projections for the Gulf management area, as its semi-enclosed nature with substantial freshwater input from the Saint Lawrence River means it is highly influenced by processes that are not well con- strained by the global climate models which inform the DBEM. Given the exploratory nature

of this assessment as a starting point for anticipating future fisheries impacts in Atlantic Can- ada, the DBEM outputs were treated as a first approximation of potential future scenarios. American lobster, eastern blue mussel, American oyster, and sea scallop were all projected to increase production moderately in the Gulf management area for 2090 and under the RCP 8.5 scenario (Fig 7). Eastern blue mussel had the largest relative change, increasing 68% under RCP 8.5 by 2090 (12% under RCP 2.6) (Fig 7). Notably, the Gulf management area encom- passes nearly all of the current mussel and oyster production in Atlantic Canada (mainly in the form of aquaculture). Therefore, changes in production in this management area drives changes for the whole Atlantic Canadian region for these species. Snow crab landings in the Gulf management area appeared to decline to zero very rapidly from nearly 30,000 tonnes in 2000 (representing ~35% of total snow crab landings in all of Atlantic Canada). This occurred because in two of the three climate models (GFDL and IPSL) and hence the median dataset–the DBEM-modelled distribution in the Gulf in 2000 was OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 16 / 29

minimal (S1U–S1X Fig), and amounted to less than 1% of the total predicted landings for the whole of the Atlantic Canadian region. When the modelled projections of this small initial dis- tribution fell to zero, it resulted in relative change of -100%. The outputs based on the third cli- mate model (MPI) predicted higher initial abundances and therefore less extreme declines of snow crab in the area (reductions of 20% and 39%, for RCP 2.6 and RCP 8.5, respectively). Therefore, a mean of the three climate model impacts was adopted for the analysis of this spe- cies. Under this approach snow crab is projected to decline in the Gulf management area by 17–18% in 2050 (RCP 2.6 and RCP 8.5, respectively) and 21–39% by 2090 (RCP 2.6 and RCP 8.5, respectively) from the 30,000 tonnes landed in 2000 (Fig 7). Newfoundland and Labrador management area. The Newfoundland and Labrador management area not only encompasses the largest area of marine waters (Fig 1) but also has some of the highest landings, by weight, with over 50% of the total Atlantic Canadian catch for northern shrimp, snow crab and Stimpson’s surf clam in 2000 (Fig 7). The patterns for north- ern shrimp and Stimpson’s surf clams were similar for both scenarios and time-steps, albeit at different scales, as the shrimp landings were more than four times that of the clam. For both species under the RCP 2.6 scenario there are modest projected increases in landings by 2050 Fig 8. Relative changes in distribution for selected commercially targeted shellfish species. Northern shrimp (a

-top), American oyster (b—middle) and snow crab (c—bottom), for 2050 (left) and 2090 (right) under RCP 8.5, highlighting south to north trend of changing species distributions. Colour scales indicate relative change predicted by DBEM for each half-degree latitude by half-degree longitude cell and do not indicate absolute values. Darkest shades (in both directions) correspond to cells with low absolute values, where small absolute changes result in large relative changes (e.g., a shift of 1T to 2T results in a 100% increase, while a change of 10T to 15T only leads to a 50% increase). Changes that exceeded +100% were set to 100% to maintain coherent colour scales. See S1 Fig for baseline DBEM distributions and figures for other species and climate treatments. https://doi.org/10.1371/journal.pone.0226544.g008 OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 17 / 29

(10% and 6% for shrimp and clams, respectively). However, under this scenario projected pro- duction to the end of the century remains relatively unchanged, with shrimp staying at 10% above 2000 levels, and clams increasing slightly to 9% over 2000 landings. In contrast, under the RCP 8.5 scenario both species continue to increase in production through to 2090 (Fig 7). As observed in the other management areas, snow crab is also projected to decline in the Newfoundland and Labrador management area. The losses are similar between the RCP cli- mate scenarios, while the landings for 2090 declined slightly from the 2050 values, from 13– 12% in 2050 (RCP 2.6 and RCP 8.5, respectively) to 16% in 2090 (for both climate scenarios). The area is expected to see minor changes for lobster, scallop and mussel landings under the quartet of projections, with mussel production presenting an exception under RCP 8.5 at the end of the century with an expected increase of 21% by 2090 under RCP 8.5. Nevertheless, all three species represent minor fractions of the region’s landings and have much higher pro- duction values in the other two management areas. Impact assessment results Exposure. Scaling the DBEM-modelled changes in landings by species value in 2000 resulted in an array of indicator scores representing the changes in the seven assessed species (Table 2). In most quartets of exposure scores, the 2090 time-step had higher exposure than the 2050 time-step—NL was an exception to this trend under the RCP 8.5 scenario (Table 2).

The RCP 8.5 climate scenario drove increased landings for some fisheries, thus the scores cor- responding to this climate scenario were higher for both the Gulf and the Newfoundland and Labrador management areas. Conversely, they were lower in the Maritime management area (and for the region as a whole) (Table 2). The Maritime management area was projected to experience losses across its quartet of exposures with the 2090 time-step and RCP 8.5 climate scenario resulting in losses of nearly 50% of the value-scaled landings (Table 2). The scores for this management area were relevant to the NB-mar and NS-mar provincial sub-sections. Conversely, the Gulf area scored slightly positive in all scenarios, however 2090 for both climate scenarios had slightly reduced scores compared to 2050 (Table 2). The Gulf management area scores were applied to PEI and Que as well as the Gulf sub-sections of NB and NS. Notably, the Newfoundland and Labrador man- agement area scored (slightly) negative under the RCP 2.6 scenario for both time-steps, but had positive scores under the RCP 8.5 scenario, largely due to the stronger gains in shrimp for the area under the latter treatment (Table 2). The Newfoundland and Labrador management area was only applied to the province of NL. The score as calculated for the whole Atlantic region was slightly negative under the quartet of treatments, although the region as a whole was not included in the final index calculation (Table 2). Social vulnerability. The scores for each of the indicators that contributed to vulnerability

fisheries sector of the five provinces, as such the industry has a very low relevance to the econ- omy and population at large. Nova Scotia had the second lowest cumulative vulnerability score (Table 5), largely driven by strong adaptive capacity indicators (especially aquaculture and education) (Table 4). Despite having the lowest education and aquaculture indicator scores, and a consequently low adaptive capacity, NB ranked third in vulnerability due to a relatively low reliance on shellfish, and hence a low sensitivity score (Tables 3 and 4). PEI and NL were the most socially vulnerable provinces (Table 5). PEI’s vulnerability derived from having the highest sensitivity due to a comparatively small population with a rel- atively high dependence on shellfish fisheries (Table 3). Additionally, PEI had moderate scores for two of the three adaptive capacity indicators, with relatively low education and high pro- vincial unemployment (Table 4). The strong aquaculture sector in PEI was not sufficient to off- set the other indicators. NL had the lowest overall vulnerability score (Table 5). However, provincial unemployment was the only indicator where it earned the lowest provincial score (Table 4). Rather, the low overall vulnerability score was driven by generally weak scores across all the indicators, with an absence of any strong adaptive capacity or low sensitivity indicators to counterbalance the low scores. Table 3. Indicators scores for sensitivity. Columns 2–3 present absolute scores for each province while columns 4–5 present the corresponding normalised scores as

used to calculate provincial sensitivity (column 6). Higher sensitivity scores contribute to higher potential impact. The sensitivity scores were combined with the adaptive capacity scores (as indicated in Fig 4) in Table 4 to define vulnerability for each province (Table 5). Province Shellfish harvesting employment / capita(a) Shellfish value / GDP(b) Normalised shellfish employment score Normalised shellfish value score Sensitivity NB 0.013 0.008 0.21 0.18 0.19 NL 0.044 0.019 0.76 0.48 0.62 NS 0.021 0.020 0.36 0.50 0.43 PEI 0.058 0.040 1.00 1.00 1.00 Que 0.001 0.001 0.00 0.00 0.00 Data references: (a)[45–50,54,55]; (b)[20,44] https://doi.org/10.1371/journal.pone.0226544.t003 Table 4. Indicators scores for adaptive capacity. Columns 2–4 present absolute scores for each indicator. Normalised indicators were reversed (i.e., subtracted from 1.0) (columns 5–7), so that low indicator scores contributed to higher potential. The combined adaptive capacity score (last column) is the average of the reversed, normalised indicators. Adaptive capacity scores were combined with the sensitivity scores (as indicated in Fig 4) in Table 3 to define vulnerability for each province (Table 5). Province [Reciprocal] Un- employment / national un-employment(a) Post-secondary education / national post-secondary education(b) Aquaculture development / mollusc production(c) (Reversed) Normalised un- employment score (Reversed) Normalised education score (Reversed) Normalised aquaculture score (Reversed) Adaptive capacity NB 0.092 0.46 0.51 0.25 1.00 1.00 0.75

NL 0.059 0.47 0.68 1.00 0.88 0.74 0.87 NS 0.098 0.53 1.14 0.11 0.00 0.00 0.04 PEI 0.077 0.49 1.05 0.60 0.62 0.14 0.45 Que 0.103 0.52 0.80 0.00 0.18 0.54 0.24 Data reference: (a)[51]; (b)[52]; (c)[20,56] https://doi.org/10.1371/journal.pone.0226544.t004 OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 19 / 29

Impact index scores. Potential impacts from OA and climate change are anticipated to be highest where high exposure and high vulnerability coincide [17,18,57] (Table 5). Since there was only a single vulnerability score for each province, while the exposure was represented by a quartet of scores, the patterns within a given province’s quartet of index scores mirrored the patterns in the exposure: 2090 generally had higher potential impacts than 2050. Under the RCP 2.6 scenario the index scores tended to be higher due to more exaggerated gains in land- ings (and hence lower potential for impact) projected under RCP 8.5. The index scores coin- ciding with exposure to the Maritime management area (NB-mar and NS-mar) were the exceptions to this and had much higher potential impacts under RCP 8.5 (Table 5). Relatively high exposure scores were only achieved for the two provinces that have fisheries in the Maritime management area (i.e., NS and NB). Within these, NS had fairly low vulnera- bility scores, while NB scored mid-range in vulnerability (Table 5). The highest single exposure score was anticipated for 2090 under the RCP 8.5 climate scenario (Tables 2 and 5). The effect of this exposure treatment was strong enough that NB-mar and NS-mar scored the two highest potential impact scores. However, the low vulnerability in NS offset the exposure for the remaining Maritime management area quartet, and the remainder of the NS-mar impact scores were ranked low to minimal potential impact (Table 5). The higher vulnerability in NB

meant the province was more directly influenced by the exposure indicator and consequently ranked high for potential impact in three of the four of exposure scenarios (Table 5). Due to the occurrence of high vulnerability and moderate exposure, the NL index quartet ranked high to moderate, with the RCP 2.6 climate scenario exposures scoring higher than the unmitigated scenario (RCP 8.5) (Table 5). The remaining provincial units (NB-gulf, NS-gulf, PEI, and Que) were all adjacent to the Gulf management area and therefore had very low normalised exposure scores (Table 5). PEI mainly ranked in the moderate potential impact category due to high vulnerability, with only the 2050 –RCP 8.5 exposure ranking as low potential for impact (Table 5). The gulf sub-divi- sions of NB and NS (NB-gulf and NS-gulf) ranked from low to least potential for impact. Out Table 5. Impact index components and scores. Components are presented as final normalised scores. The exposure scores are presented in columns 3–4 as quartets of climate scenario and future time-step. The vulnerability scores (column 5) are the average (i.e., equally weighted combination) of sensitivity (Table 3) and adaptive capacity (Table 4) as indicated by Fig 4. The final index score quartets (column 6–7) are the average (i.e., equally weighted combination) of the relevant exposures and vulnerabili- ties. The final two columns indicate rank for each unique potential impact score: 1-high, 2-moderate, 3-low, 4-minimal and 5-least potential for impact (all categories

except 3-low, represent 6 unique scores– 3-low only represents 4 unique scores). Province RCP Exposure 2050 Exposure 2090 Vulnerability Impact 2050 Impact 2090 Impact Ranking NB-gulf 2.6 0.05 0.09 0.47 0.26 0.28 4 3 8.5 0.00 0.06 0.24 0.27 4 4 NB-mar 2.6 0.32 0.42 0.40 0.44 2 1 8.5 0.48 1.00 0.48 0.74 1 1 NL 2.6 0.15 0.17 0.75 0.45 0.46 1 1 8.5 0.10 0.02 0.42 0.38 2 2 NS-gulf 2.6 0.05 0.09 0.23 0.14 0.16 5 4 8.5 0.00 0.06 0.12 0.15 5 4 NS-mar 2.6 0.32 0.42 0.28 0.33 4 3 8.5 0.48 1.00 0.36 0.62 3 1 PEI 2.6 0.05 0.09 0.73 0.39 0.41 2 2 8.5 0.00 0.06 0.36 0.39 3 2 Que 2.6 0.05 0.09 0.12 0.08 0.11 5 5 8.5 0.00 0.06 0.06 0.09 5 5 https://doi.org/10.1371/journal.pone.0226544.t005 OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 20 / 29

of the two, NB-gulf had slightly higher overall risk due to higher social vulnerability. Quebec, which had the lowest vulnerability along with the low exposures associated with the Gulf man- agement area, predictably earned the lowest index scores (Table 5). Sensitivity testing. In order to test the robustness of the framework construction (i.e., Fig 4), alternate indicator weighting and indicator aggregating steps were considered. Most of the alternate orientations mainly re-arranged the mid-level index values (i.e., the moderate to min- imal impact categories). Although there were some notable adjustments in the higher potential impact categories. In all the tested weighting orientations Que maintained the lowest scores. Weighting each individual indicator equally (i.e., setting each indicator to contribute equally to total index scores) rather than equal within each branch in the framework, caused the effect of the exposure indicator to be diluted so that the total scores were more similar to the provincial vulnerability scores. In this weighting system NL and PEI scored the highest index scores, followed by NB (S2 Fig and S4 Table). NS-mar and NS-gulf filled out the minimal and least potential impact categories. When exposure, sensitivity and adaptive capacity were weighted equally (i.e., sensitivity and adaptive capacity were kept separate, rather than combined to form vulnerability), in a framework configuration which more closely resembled the orientations used in vulnerability

assessments such as [11,14], NL and PEI again ranked higher due to their weak scores in the sensitivity and adaptive capacity (S2 Fig and S4 Table). In contrast, other risk assessment frameworks have used hazard, exposure and vulnerability as three equally weighted compo- nents of risk [15,17]. However, in this assessment, the hazard was effectively incorporated into the exposure term. To test the effect of including an equally weighted hazard component, the exposure scores were weighted twice as heavily as vulnerability. Under this weighting system NS-mar and NB-mar scored higher with some of their scores moving up a rank. Conversely, NL along with the NS-gulf and NB-gulf scored relatively lower (S2 Fig and S4 Table). Since the indices resulting from the alternate orientations emphasised either the vulnerability or the exposure indicators, the equal weighting system used throughout the analysis presented the most neutral approach. Discussion Fisheries changes in Atlantic Canada Temperature-driven changes are expected to shift marine species’ ranges poleward [e.g., 32,40,58]. Six of the seven species considered in this study are currently near the middle or even at the southern extent of their historic natural range in Atlantic Canada (i.e., American Lobster, northern shrimp, snow crab, sea scallop, eastern blue oyster and Stimpson’s surf calm [59]). It was, therefore, unsurprising that the DBEM predicted changes in species distributions in the study region resulted in the southernmost management area (i.e., the Maritime manage-

ment area) experiencing declines for most species under the quartet of exposure scenarios (Fig 7). Meanwhile, the Gulf along with the Newfoundland and Labrador management areas were predicted to see modest increases in landings for most species under most of the treatments (Fig 7). Snow crab, which requires very cold bottom waters [60], was predicted to see substantial declines in all scenarios and management areas (Fig 8C). This was consistent with other recent assessments of marine species distributions in Atlantic Canadian waters under warming ocean temperatures [61]. Due to its significant contribution (both in value and tonnage) to overall fisheries landings in Atlantic Canada, the region-wide anticipated decline in future snow crab landings impacted the cumulative fisheries changes across the management areas (Fig 7). OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 21 / 29

American oyster was the only assessed species that is currently near the northern limits of its distribution in Atlantic Canadian waters. At present, landings for the fishery are relatively small (Figs 5 and 7), but overall, the region is expected to become better suited for the species. With most of its current production deriving from aquaculture, the industry could be in a position to take advantage of the improving habitat suitability in the region. Enhanced aqua- culture production of oyster could help to partially compensate for losses anticipated of other species, especially in the southern reaches of the region. However, current American oyster production rates are orders of magnitude lower than most of the other assessed species, so any mitigation potential for losses are likely to be highly localised. Potential impacts and social vulnerability The communities in NB that rely on fisheries in the Maritime management area were found to be most at risk to OA and climate change. Though, the bulk of NB shellfish production occurs in the Gulf management area and is therefore not exposed to the changes in the Maritime management area. Nonetheless, the sub-section of the province reliant on the Maritime man- agement area should be considered as an area for pro-active responses to potential changes in fisheries driven by OA and climate change. Conversely, the NB-gulf sub-section did not appear to be a high risk area, and scored minimal potential for impact in all of four of the exposure

scenarios. The contrasting potential impact levels within the province may present an opportu- nity to locally mitigate impacts in the Maritime management area dependent communities, as production and harvesting activity could be shifted to the potentially less exposed management area. Such shifts would certainly have to take the potential consequences of increasing fishing capacity in the Gulf management area into account. The within province gradient in future resource accessibility may present a microcosm of the Atlantic Region as a whole, where declines in one location are, to an extent, counterbalanced by gains in another. The provinces of PEI and NL were also at generally higher risk from OA and climate change, mostly ranking in the high to moderate potential impact categories. The potential impacts in these provinces were predominantly driven by social vulnerability rather than changes in access to the fisheries themselves. Many of the vulnerability factors (such as educa- tion and unemployment) occur at local scales and are therefore more immediately actionable by provincial decision-makers. As such, these provinces may be able to more directly pursue social and economic shifts to reduce risk from OA and climate change. Overall, NS scored a single high potential impact score (in 2090 under RCP 8.5) for the Maritime management area; otherwise the province ranked low to least potential for impact. In spite of being highly exposed in communities which depend on the Maritime management

area, Nova Scotia’s relatively low vulnerability suggests that the province will generally be more capable of responding to the changes in shellfish production driven by OA and climate change than other provinces in the region. However, it is possible that communities within the province have much less capacity to adapt than the provincial statistics imply. As with most of the other provinces in the region, NS has a relatively high rural population and many small coastal communities with a high dependence on fisheries [23,24]. Given the relatively strong declines anticipated for the Maritime management area, a more localized investigation may be warranted for this province. Lastly, Que appears to be largely unthreatened by OA and climate change impacts on fisher- ies. The province as a whole is not vulnerable to changes in shellfish production, and under the DBEM projections the comparatively small shellfish harvest was not highly exposed to change. OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 22 / 29

Responding to risk To reduce exposure to OA and climate change a global scale response is required. That said, some localised factors that amplify OA (e.g., eutrophication) can be acted upon at more con- fined scales [62]. In order to reduce impacts from OA and climate change in Atlantic Canada (and the rest of the planet), global efforts need to be made to reduce carbon emissions [e.g., 1,39,63]. Local policies can, and should, be enacted to reduce emissions and to contribute to reductions in atmospheric CO2 concentrations. However, action directly responding to antici- pated changes in resource availability will also be important for mitigate the impacts on coastal communities and economies. Factors affecting social vulnerability can be acted upon locally by regional managers and decision-makers to reduce risk. Although this will not reduce the ecological impacts caused by OA and climate change, it can help to reduce the impacts felt by human communities. In the case of vulnerability related to Atlantic Canadian shellfish fisheries, there are multiple opportu- nities for mitigating actions. While education levels were near the national rate for most of the provinces (Table 4), efforts to improve education rates could alleviate some of the social vul- nerability in the region by opening more employment opportunities to the populations. The education statistic considered here was a broad indicator of overall education in the provinces, however, more targeted education programs regarding the future of the affected fisheries (for

both increasing and decreasing abundances) could greatly improve the adaptability of the har- vesters and communities which rely on them and help to promote sustainable long-term har- vests [64]. Similarly, addressing provincial unemployment rates, which were above the national levels in all of the provinces, could greatly reduce vulnerability to potential lost employment in declining fisheries. Improvements in either (or both) of these indicators should also improve provincial adaptive capacity regarding a range of potential climate change driven impacts. The third adaptive capacity indicator, aquaculture, is more case specific but also presents a direct way to respond to OA impacts on shellfish production in the region. Current aquacul- ture production of shellfish in Atlantic Canada is mainly dependent on wild populations for recruitment and hatchery production in the region is limited [27]. However, investment in hatchery infrastructure could greatly improve the region’s ability to maintain production of mollusc species in the face of falling pH or irregular natural recruitment. Adaptation of hatch- ery procedures is already being used to mitigate low pH events on the Pacific Coast of Canada and the United States; the expertise developed there could be leveraged to support Atlantic production [8,53]. It is worth noting that while improving shellfish aquaculture may help to maintain production levels, it is not a direct substitute for loss of livelihood from decreased capture fisheries. Fish harvesters may be unlikely to actively pursue a transition to aquaculture

production. Furthermore, in terms of actual employment aquaculture is a more efficient method of production, by tonnage of product, and typically requires fewer people to produce equivalent harvests. Nonetheless, as a factor to reduce overall impacts from changes in fisher- ies, aquaculture (with hatcheries) is an opportunity worth further consideration in the region. The indicators related to sensitivity are somewhat more difficult to address because they are more directly tied to the fishing industry and are ingrained in the social and cultural identity of the region. Furthermore, DFO acknowledges that overcapacity already exists in many Atlan- tic Canadian fisheries [23]. Incrementally reducing overall fishing effort could help to limit sensitivity to changes in the fisheries and support overall conservation targets [23,65]. To be a viable option, this would first require that alternative employment opportunities existed. Diversifying harvests is another mechanism that can be enacted to dampen losses in any one fishery [23]. Diversity of mollusc species harvested has previously been used as a component OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 23 / 29

of adaptive capacity in their assessment of vulnerability to OA in the United States [14]. How- ever, many of the commercial harvesters in Atlantic Canada already target multiple species [23]. While this may ultimately indicate that the provinces are less sensitive to changes than this assessment implies, it also means that this aspect will be difficult to improve upon as a method to reduce vulnerability for the region. Developing other industries and factors sup- porting provincial economies could reduce the relative importance of fisheries in the region and thereby reduce sensitivity related to changing fisheries. A finer scale assessment is also potentially highly relevant for addressing potential impacts in Atlantic Canada. Much of the region is made up of small communities with much more localised economic and social factors than is represented by the provincial scale statistics [23,24]. In some counties, employment in shellfish fisheries is much higher than the provincial rates imply–especially when additional steps of the supply chain are included. For example, lobster fishing area (LFA) 34 is by far the most productive area for American lobster in the Maritime management area [66]. Following an approximation that counties rely most heavily on adjacent management areas, this implies that counties such as Digby, Yarmouth and Shel- burne could be much more sensitive to changes in that fishery than other NS counties. More- over, LFA 34 corresponds to the area with some of the most substantial DBEM projected

losses for lobster landings (S1E–S1H Fig). Future assessments at a finer social scale might be necessary to identify these potentially more at-risk communities. Following the conclusions of this study, NB and NS would be strong candidates for future fine-scale assessment, as these provinces were most exposed to changes in fisheries, thus highly vulnerable communities in these provinces could be among the most at risk in the region. Relatedly, future assessments could include more detailed economic models to better account for how markets may respond changes in the industry. More broadly, future research into socioeconomic implications of OA and climate change will benefit substantially from better understanding species and ecosystem responses to OA and other climate change stressors, especially including how responses to these stressors will interact. Conclusions could also be improved with more detailed biophysical models. To this end improving the resolution and incorporating finer scale ocean currents and processes into models, such as the DBEM used here, would allow for more nuanced estimates of social impacts and more specific predictions of where and how fisheries will potentially be most affected. Consideration of additional changes to ecosystem functions, such as increased rates of harmful algal blooms, will also benefit future estimates of potential impacts on fisheries. Additionally, the OA impact level in the iteration of the DBEM used in this research was based on the mean effect from two meta-analyses [3,4]. The OA driven losses could be much

more severe than anticipated by the model, potentially to the point where OA might not be entirely overwhelmed by temperature driven patterns in abundance. Conversely, the OA impacts could also be much less severe and the benefits for many species in the region from thermal loading could be more pronounced. Recent efforts have explicitly tested different impact levels if OA in the DBEM [34]. Thus future socioeconomic assessments using DBEM projections may be able to examine a broader range of OA impact levels. Similarly, the OA effects in this assessment assume a linear impact from OA (i.e., as pH changes there was a lin- ear relationship to the affected life-history traits–growth and survival). This treatment does not allow for adaptation, where species might be able to limit the effects of OA through biolog- ical mechanisms such as acclimatisation (including behavioural responses), parental effects (i.e., epigenetic effects) or evolution [e.g., 67]. However, including a simulation of adaptation in the DBEM (where the pH level is applied in relation to the previous time-step rather than the baseline level) appears to lead to negligible effects from OA in the model outputs [34]. OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 24 / 29

Conclusion The findings of this assessment contrast with previous socioeconomic analyses of OA effects on fisheries. In previous assessments of OA impacts [11–17], projected changes in fisheries have been presented almost exclusively as declines in potential landings with subsequent impacts on societies and economies. However, the majority of these studies investigated OA impacts in isolation from other facets of climate change and did not account for potential multi-stressor impacts. This study demonstrates that accounting for additional environmental factors can allow for different, and potentially more representative, narratives to emerge. Future management decisions and mitigation plans could be better informed through analyses which account for a more complete range of future effects on resource accessibility. Atlantic Canada is a region with an exceptionally high dependence on OA-susceptible fish- eries. However, the findings presented here suggest that OA-driven declines in the region will be minor compared to temperature-driven changes in future landings (Fig 6). Nonetheless, fisheries resources in Atlantic Canada are expected to see some notable redistributions under OA and climate change over the coming century. As observed in global-scale studies this is expected to result in ‘winners and losers’ with respect to access to future fisheries [e.g., 40]. It may be tempting to view the ‘winners’ as gaining access to new production. However, it is essential to recall that perceived increases come at a cost to other regions [68]. Additionally,

while OA effects may appear to be locally overwhelmed by temperature driven increases this should be considered in context and interpreted as limiting potential gains, and exacerbating declines. Within the Atlantic Canadian region, management plans should be developed that take cli- mate change and shifting species distributions into account. These plans should specifically address future allocation of resources when long-held access to certain high value fisheries (e.g., American lobster) might transition into new jurisdictions. Moreover, these consider- ations need to be extended to cooperative management of marine resources between nations. For example, net gains in Atlantic Canada may come at a cost to US states to the south. Management plans seeking to account for changing access to resources will benefit from assessments, such as in this study, which address how and where biological changes are likely to affect human communities, and where efforts are needed to mitigate these impacts. Supporting information S1 Note. Additional details for DBEM. (PDF) S2 Note. Details inclusion of ‘inshore’ and ‘offshore’ fleets and special cases regarding data availability. (PDF) S1 Table. DBEM projected relative change from individual climate models for each climate scenario and both future time-steps for each species. (ODS) S2 Table. data sources for crew size estimates. (PDF) S3 Table. Annual landing value (year 2000 dollars) and weights (tonnes) for marine species in Atlantic Canada. Values for DFO fisheries species groups (finfish, groundfish and shellfish)

are presented, as well as values for individual shellfish species. (ODS) OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 25 / 29

S1 Fig. DBEM projections for OA and RCP treatments. Each species is presented in four fig- ures representing: RCP 2.6 ‘with O,’ RCP 2.6 ‘without OA,’ RCP 8.5 ‘with OA’ and RCP 8.5 ‘without OA,’ respectively. Species are presented as follows: a-d = American oyster; e-h Ameri- can lobster; i-l sea scallop; m-p eastern blue mussel; q-t northern shrimp; u-x snow crab; and y-ab Stimpsons’ surf clam. Each figure includes: top left–species graphic; top right–relative change for management areas. Scale is the same for each figure; therefore, species which extend beyond 25% change in either direction go beyond the figure axes (note data separates NAFO subareas that border DFO management areas (i.e. 4R and 4Vn), line colours corre- spond to distribution map in bottom left). Bottom left–DBEM predicted distribution for base- line year 2000 (averaged 1991–2010). Management borders outline colour corresponds to relative change in top right panel. Bottom middle–relative change in 2050 (average 2041– 2060) compared to 2000 (average 1991–2010). Bottom right–relative change for 2090 (average 2081–3000) against 2000 (average 1991–2010). Note that colour scales presented in bottom- middle and bottom-right panels are relative changes within individual data cell. An artefact of this was that cells which had zero presence in 2000 but gained presence in future time steps had an infinite increase in catch potential; therefore, all increases above 100% were presented as 100%. Relatedly, the vast majority of the cells that experience 100% increases or decreases

represent small baseline inputs (as per the bottom right panel). (ODS) S2 Fig. Alternate framework orientations for sensitivity testing. Letters correspond to sensi- tivity testing risk outcomes in S4 Table (TIF) S4 Table. Risk index results for 3 alternative weighting schemes. Framework orientations schematics presented in S2 Fig. (ODS) Acknowledgments TJBW would like to thank Michael Gardner for sharing his time and expertise regarding the structure and operation of Atlantic Canadian fisheries. Author Contributions Conceptualization: Tyler J. B. Wilson, Sarah R. Cooley, Peter H. Tyedmers. Data curation: Tyler J. B. Wilson, Travis C. Tai, William W. L. Cheung. Funding acquisition: Tyler J. B. Wilson, Travis C. Tai, William W. L. Cheung, Peter H. Tyedmers. Investigation: Tyler J. B. Wilson. Methodology: Tyler J. B. Wilson, Sarah R. Cooley, Travis C. Tai, William W. L. Cheung, Peter H. Tyedmers. Supervision: Peter H. Tyedmers. Visualization: Tyler J. B. Wilson, Sarah R. Cooley, Peter H. Tyedmers. Writing – original draft: Tyler J. B. Wilson. Writing – review & editing: Tyler J. B. Wilson, Sarah R. Cooley, Travis C. Tai, William W. L. Cheung, Peter H. Tyedmers. OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 26 / 29

References 1. Doney SC, Fabry VJ, Feely RA, Kleypas JA. Ocean acidification: The other CO2 problem. Annu Rev Mar Sci. 2009; 1: 169–192. https://doi.org/10.1146/annurev.marine.010908.163834 PMID: 21141034 2. Orr JC, Fabry VJ, Aumont O, Bopp L, Doney SC, Feely RA, et al. Anthropogenic ocean acidification over the twenty-first century and its impact on calcifying organisms. Nature. 2005; 437: 681–686. https://doi.org/10.1038/nature04095 PMID: 16193043 3. Kroeker KJ, Kordas RL, Crim R, Hendriks IE, Ramajo L, Singh G, et al. Impacts of ocean acidification on marine organisms: quantifying sensitivities and interaction with warming. Glob Change Biol. 2013; 19: 1884–1896. https://doi.org/10.1111/gcb.12179 PMID: 23505245 4. Kroeker KJ, Kordas RL, Crim RN, Singh GG. Meta-analysis reveals negative yet variable effects of ocean acidification on marine organisms. Ecol Lett. 2010; 13: 1419–1434. https://doi.org/10.1111/j. 1461-0248.2010.01518.x PMID: 20958904 5. Ries JB, Cohen AL, McCorkle DC. Marine calcifiers exhibit mixed responses to CO2-induced ocean acidification. Geology. 2009; 37: 1131–1134. https://doi.org/10.1130/G30210A.1 6. Gazeau F, Quiblier C, Jansen JM, Gattuso J-P, Middelburg JJ, Heip CHR. Impact of elevated CO2 on shellfish calcification. Geophys Res Lett. 2007; 34: n/a–n/a. https://doi.org/10.1029/2006GL028554 7. Waldbusser GG, Hales B, Langdon CJ, Haley BA, Schrader P, Brunner EL, et al. Saturation-state sen- sitivity of marine bivalve larvae to ocean acidification. Nat Clim Change. 2014; 5. https://doi.org/10. 1038/nclimate2479

8. Washington State. Ocean Acidification: From Knowledge to Action, Washington State’s Strategic Response. Washington State Blue Ribbon Panel on Ocean Acidification; 2012. Report No.: 12-01-15. Available: http://www.ecy.wa.gov/water/marine/oceanacidification.html 9. FAO. The state of world fisheries and aquaculture 2016. Rome: Food and Agriculture Organization of the United Nations; 2016 p. 200. Available: http://www.fao.org/fishery/sofia/en 10. Smith MD, Roheim CA, Crowder LB, Halpern BS, Turnipseed M, Anderson JL, et al. Sustainability and global seafood. Science. 2010; 327: 784. https://doi.org/10.1126/science.1185345 PMID: 20150469 11. Cooley S, Lucey N, Kite-Powell H, Doney SC. Nutrition and income from molluscs today imply vulnera- bility to ocean acidification tomorrow. Fish Fish. 2012; 13: 182–215. https://doi.org/10.1111/j.1467- 2979.2011.00424.x 12. Narita D, Rehdanz K, Tol RSJ. Economic costs of ocean acidification: a look into the impacts on global shellfish production. Clim Change. 2012; 113: 1049–1063. https://doi.org/10.1007/s10584-011-0383-3 13. Narita D, Rehdanz K. Economic impact of ocean acidification on shellfish production in Europe. J Envi- ron Plan Manag. 2016; 1–19. https://doi.org/10.1080/09640568.2016.1162705 14. Ekstrom JA, Suatoni L, Cooley SR, Pendleton LH, Waldbusser GG, Cinner JE, et al. Vulnerability and adaptation of US shellfisheries to ocean acidification. Nat Clim Change. 2015; 5: 207–214. 15. Heinrich L, Krause T. Fishing in acid waters: A vulnerability assessment of the Norwegian fishing indus-

try in the face of increasing ocean acidification. Integr Environ Assess Manag. 2016; n/a–n/a. https:// doi.org/10.1002/ieam.1843 PMID: 27563756 16. Moore C. Welfare impacts of ocean acidification: an integrated assessment model of the us mollusk fishery. Washington DC: EPA; 2011 p. 35. Report No.: 11–06. Available: https://www.epa.gov/ environmental-economics/working-paper-welfare-impacts-ocean-acidification-integrated-assessment 17. Mathis JT, Cooley SR, Lucey N, Colt S, Ekstrom J, Hurst T, et al. Ocean acidification risk assessment for Alaska’s fishery sector. Prog Oceanogr. 2015; 136: 71–91. https://doi.org/10.1016/j.pocean.2014. 07.001 18. Cardona OD, van Aalst MK, Birkmann J, McGregor G, Perez R, Pulwarty RS, et al. Determinants of risk: exposure and vulnerability. In: Field CB, Barros V, Stocker TF, Qin D, Dokken DJ, Ebi KL, et al., editors. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge, UK, and New York, NY, USA: Cambridge University Press; 2012. pp. 65–108. 19. DFO. Canada’s fisheries fast facts 2016. Fisheries and Oceans Canada; 2017 p. 6. Available: http:// www.dfo-mpo.gc.ca/stats/facts-Info-16-eng.htm 20. DFO. Seafisheries landings. 2017. Available: http://www.dfo-mpo.gc.ca/stats/commercial/sea- maritimes-eng.htm 21. Divovich E, Belhabib D, Zeller D, Pauly D. Eastern Canada, “a fishery with no clean hands” Marine fish- eries catch reconstruction from 1950 to 2010. University of British Columbia; 2015. Report No.: 2015– 56.

22. DFC. Federal support to provinces and territories. In: Department of Finance Canada [Internet]. 2017 [cited 10 Nov 2017]. Available: https://www.fin.gc.ca/fedprov/mtp-eng.asp OA and climate change impacts on Atlantic Canadian fisheries PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 27 / 29

23. DFO. Atlantic fisheries policy review—A policy framework for the management of fisheries on Canada’s Atlantic Coast. Ottawa: Fisheries and Oceans Canada; 2004. Available: http://www.dfo-mpo.gc.ca/ reports-rapports/regs/afpr-rppa/framework-cadre-eng.htm 24. Statistics Canada. 2011 Census. Statistics Canada; 2011. Available: https://www.statcan.gc.ca/tables- tableaux/sum-som/l01/cst01/demo62d-eng.htm 25. NAFO. Northwest Atlantic Fisheries Organization. [cited 10 Nov 2017]. Available: https://www.nafo.int/ About-us 26. Bank of Canada. Inflation calculator. 2017 [cited 27 Nov 2017]. Available: http://www.bankofcanada.ca/ rates/related/inflation-calculator/ 27. CAIA. Canadian Aquaculture Industry Alliance. 2018. Available: http://www.aquaculture.ca/ aquaculture-species-index 28. Crain CM, Kroeker K, Halpern BS. Interactive and cumulative effects of multiple human stressors in marine systems. Ecol Lett. 2008; 11: 1304–1315. https://doi.org/10.1111/j.1461-0248.2008.01253.x PMID: 19046359 29. Ghedini G, Connell SD. Moving ocean acidification research beyond a simple science: Investigating ecological change and their stabilizers. Food Webs. 2017. https://doi.org/10.1016/j.fooweb.2017.03. 003 30. Kroeker KJ, Kordas RL, Harley CDG. Embracing interactions in ocean acidification research: confront- ing multiple stressor scenarios and context dependence. Biol Lett. 2017; 13. https://doi.org/10.1098/ rsbl.2016.0802 PMID: 28356409 31. Cheung WWL, Dunne J, Sarmiento JL, Pauly D. Integrating ecophysiology and plankton dynamics into

projected maximum fisheries catch potential under climate change in the Northeast Atlantic. ICES J Mar Sci J Cons. 2011; 68: 1008–1018. https://doi.org/10.1093/icesjms/fsr012 32. Cheung WWL, Lam V, Pauly D. Modelling present and climate-shifted distribution of marine fishes and invertebrates. University of British Columbia; 2008 p. 72. Report No.: 16(3). Available: http://hdl.handle. net/2429/40936 33. Cheung WWL, Reygondeau G, Fro¨ licher TL. Large benefits to marine fisheries of meeting the 1.5˚C global warming target. Science. 2016; 354: 1591. https://doi.org/10.1126/science.aag2331 PMID: 28008069 34. Tai TC, Harley CDG, Cheung WWL. Comparing model parameterizations of the biophysical impacts of ocean acidification to identify limitations and uncertainties. Ecol Model. 2018; 385: 1–11. https://doi.org/ 10.1016/j.ecolmodel.2018.07.007 35. Portner HO, Lannig G. Chapter 4 Oxygen and Capacity Limited Thermal Tolerance. In: Richards JG, Farrell AP, Brauner CJ, editors. Fish Physiology. Academic Press; 2009. pp. 143–191. Available: http:// www.sciencedirect.com/science/article/pii/S1546509808000046 36. Pauly D, Cheung WWL. Sound physiological knowledge and principles in modeling shrinking of fishes under climate change. Glob Change Biol. 2017; 24: e15–e26. https://doi.org/10.1111/gcb.13831 PMID: 28833977 37. Lam VWY, Cheung W, Sumaila UR. Marine capture fisheries in the Arctic: winners or losers under cli- mate change and ocean acidification? Fish Fish. 2014; n/a–n/a. https://doi.org/10.1111/faf.12106

38. van Vuuren DP, Edmonds J, Kainuma M, Riahi K, Thomson A, Hibbard K, et al. The representative con- centration pathways: an overview. Clim Change. 2011; 109: 5. https://doi.org/10.1007/s10584-011- 0148-z 39. Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assess- ment Report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland: IPCC; 2014 p. 151. Available: http://www.ipcc.ch/pdf/assessment-report/ar5/syr/SYR_AR5_FINAL_full_wcover.pdf 40. Cheung WWL, Lam V, Sarmiento JL, Kearney K, Watson R, Zeller D, et al. Large-scale redistribution of maximum fisheries catch potential in the global ocean under climate change. Glob Change Biol. 2010; 16: 24–35. https://doi.org/10.1111/j.1365-2486.2009.01995.x 41. M_Map. M_Map: A mapping package for Matlab. 2014. Available: https://www.eoas.ubc.ca/~rich/map. html 42. Hajkowicz S. Multi-attributed environmental index construction. Ecol Econ. 2006; 57: 122–139. https:// doi.org/10.1016/j.ecolecon.2005.03.023 43. OECD. Handbook on Constructing Composite Indicators: Methodology and User Guide. Organisation for Economic Co-operation and Development; 2008. Available: https://doi.org/10.1787/ 9789264043466-en 44. Statistics Canada. Table 379–0030—Gross domestic product (GDP) at basic prices, by North American Industry Classification System (NAICS), provinces and territories, annual (dollars). 2017. Report No.: 379–0030. Available: http://www5.statcan.gc.ca/cansim/a26?lang=eng&id=3790030 OA and climate change impacts on Atlantic Canadian fisheries

PLOS ONE | https://doi.org/10.1371/journal.pone.0226544 January 10, 2020 28 / 29

45. DFO. Atlantic region licences. 2017. Available: http://www.dfo-mpo.gc.ca/stats/commercial/licences- permis/licences-permis-atl-eng.htm 46. Brander L, Burke L. Rights based vs. competitive fishing of sea scallops in Nova Scotia. 1994 p. 14. Available: ftp://ftp.fao.org/docrep/fao/010/a1497e/a1497e11.pdf 47. DFO. Costs and earnings survey 2004: Atlantic region report. Ottawa: Fisheries and Oceans Canada; 2007 p. 72. Available: http://publications.gc.ca/collections/collection_2012/mpo-dfo/Fs23-501-2004-eng.pdf 48. Gardner Pinfold Consulting. Overview of the Atlantic snow crab industry. The Atlantic Council of Fisher- ies an Aquaculture Ministers; 2006 p. 66. 49. Gardner Pinfold Consulting. Profile of the Atlantic shrimp industry. The Atlantic Council of Fisheries an Aquaculture Ministers; 2006 p. 50. 50. Stevens G, Robert G, Burke L, Poullioux E, Roussel D, Wilson JR. The evolution of management in Cana- da’s offshore scallop fishery. p. 14. Available: ftp://ftp.fao.org/docrep/fao/010/a1497e/a1497e11.pdf 51. Statistics Canada. Table 282–0002—Labour force survey estimates (LFS), by sex and detailed age group, annual (persons unless otherwise noted). 2017. Report No.: 282–0002. Available: http://www5. statcan.gc.ca/cansim/a26?lang=eng&id=2820002 52. Statistics Canada. Table 282–0004—Labour force survey estimates (LFS), by educational attainment, sex and age group, annual (persons unless otherwise noted). 2017. Report No.: 282–0004. Available: http://www5.statcan.gc.ca/cansim/a26?lang=eng&id=2820004