The Sablefish ( Anoplopoma fimbria ) of Č ḯ xwicən : Socioenvironmental Lessons From an Unusually Abundant Species

We analyzed sablefish (Anoplopoma fimbria) remains from Čḯxicən (pronounced ch-WHEET-son), a 2700 yearold ancestral village of the Lower Elwha Klallam Tribe in northwest Washington state, U.S.A., to improve understanding of how this species was used by Native American/First Nations peoples in the past. Though sablefish are abundant at Čḯxicən, and limited ethnographic accounts indicate they were highly prized in northwestern North America, their remains are rare in regional archaeology. We present a body-size regression model for estimating the fork length (FL) of archaeologically represented sablefish and determining which habitats they were captured from (i.e. shallow, nearshore waters as juveniles or deepwater, offshore sites as adults). FL estimates for sablefish remains from Čḯxicən indicate the site occupants exclusively targeted inshore juveniles. Comparisons of sablefish abundances over time show juvenile sablefish were reliably and sustainably harvested over the duration of the site's occupation despite major environmental perturbation from regional climate change and tectonic disturbances. However, patterns of sablefish use differ in two Čḯxicən households, suggesting access to and consumption of sablefish was socially mediated.


Introduction
Analysis of archaeological fishbone from Čḯx w icən (pronounced ch-WHEET-son), a large Native American village on the coast of Washington State, U.S.A. (Fig. 1), shows that sablefish (Anoplopoma fimbria, also commonly marketed as black cod), is the third most abundant fish taxon represented at the site (3209 NISP out of a total 44,763 NISP). Considering that sablefish are scarce or absent from all other archaeological sites in northwestern North America (Nims and Butler, 2018), we were initially perplexed that the species is so abundant at Čḯx w icən. Now that we know more about the species, we seek to understand why it is not more common elsewhere. Today, sablefish occupies nearly every North Pacific habitat over the course of its life span, from the extreme depths of continental slopes to inshore waters, where young-of-the-year are easily caught by hand-jigging (Echave et al., 2013). The species is also extremely nutritious (U.S. Department of Agriculture, 2014) and we expect sablefish would have been sought after wherever they were available.
Ethnographic and historic evidence for sablefish capture in the region is lacking, with only a few exceptions. Available records show that adult sablefish were especially prized on Haida Gwaii (Fig. 1) where Haida fished for them with specialized bentwood hooks (Blackman, 1990;Hobler, 1978;Swan, 1887). Further evidence for sablefish use among Haida was recorded in a story told by John Sky (in Swanton, 1905) that includes the expression, "Does the black cod stick you here?" According to Swanton (1905, p. 225), the question refers to the exclusive availability of 'black cod' on the west coast of Haida Gwaii, which was so highly regarded that visitors, including the story's main character, would delay departure from the region. Arima and Dewhirst (1990, p. 397) mention that groups of Nuu-chah-nulth and Ditidaht on the west coast of Vancouver Island (Fig. 1) captured 'sablefish' and 'lingcod' with lures and harpoons or dipnets from canoes. Finally, James Swan (1887) noted in the late nineteenth century that sablefish were highly valued by peoples living on the Strait of Juan de Fuca (Fig. 1), and that it was a common bycatch in European and Native American/First Nations commercial fisheries that was occasionally taken in large volumes. Swan (1887) also states that Native American/ First Nations peoples along the Strait of Juan de Fuca only desired mature sablefish, which are typically captured at depths of 100 m to 1400 m and grow to over a meter in length (Head et al., 2014;Love, 2011;Pearson and Shaw, 2004). Thus, adults were pursued with longline fishing methods as luxury items for chiefs, but only infrequently due to the difficulties associated with fishing in such deep waters (Swan, 1887).
While we have no information about traditional preferences for juvenile sablefish, young-of-the-year may have provided an attractive, low-cost alternative. Juveniles measuring 20 cm to 40 cm are commonly found in shallow bays and inlets throughout the Pacific Coast at depths of 20 m to 60 m, and they periodically become super-abundant. The occurrence of these "strong year classes" appears to correlate with primary productivity (Shotwell et al., 2014;Sogard, 2011) and have resulted in enormous commercial and recreational catches of juveniles (Cox, 1948;McFarlane and Beamish, 1983;Rutecki and Varosi, 1997a). Another important factor to consider is the high degree of spatial variability in juvenile sablefish abundances. A seven year survey of juveniles in southeast Alaska found sablefish in 11 bays and inlets, but they were only consistently present in large numbers at one location, St. John Baptist Bay (Rutecki and Varosi, 1997a). Why sablefish prefer some bays over others remains unknown, but Coutré (2014) suggests juveniles might be attracted to places with freshwater input and salmon offal, or that sablefish might be entrained in specific places as larvae.
As sablefish occupy different habitats at particular stages of their life history, we aimed to improve our understanding of the ecological contexts of sablefish capture in the past by studying the population structures of archaeological sablefish. Here, we present a method for estimating the body-size of sablefish from the size of their vertebrae and apply it to the Čḯx w icən collection to explore the ecological contexts of sablefish capture. We also examine the patterns in sablefish abundances over time at Čḯx w icən and in two separate households from the site to explore the socioenvironmental factors that affect sablefish representation. Finally, we apply insights from Čḯx w icən to the question of sablefish's widespread scarcity in northwest North American archaeology.

Sablefish life history
Sablefish make two ontogenetic migrations during their early life history. First, larvae hatch from eggs on continental slopes in February or March and then migrate to the shoreline along the surface of the ocean during their first summer, developing into juvenile fish along the way (Kendall and Matarese, 1987;Mason et al., 1983;McFarlane and Beamish, 1983;Wing, 1997). Annual migrations of juveniles arrive in nearshore and inshore waters by September or October, where they reside for one or two years. Second, sablefish leave inshore waters after their second summer for the continental slopes and move progressively deeper as they mature and grow larger (Courtney and Rutecki, 2011;Maloney and Sigler, 2008;Varosi, 1997a, 1997b). These two migrations effectively separate sablefish life history into three distinct habitat regimes: offshore epipelagic (i.e. inhabiting the oceanic zone illuminated by sunlight), inshore demersal (i.e. bottom-dwelling), and offshore demersal.
Because sablefish body-size and growth rates are closely related to age, with rapid growth in the first few years of life followed by slow or zero growth after reaching maturity (Echave et al., 2012;Hanselman et al., 2015;Head et al., 2014), sablefish life-stage and habitat regime can be identified by body-size alone (Maloney and Sigler, 2008;Rutecki and Varosi, 1997b). Based on studies from southeast Alaska to California, juveniles arriving in inshore waters typically measure between 20 cm and 30 cm length (Bell and Gharrett, 1945;Edson, 1954) with average lengths of 21 cm to 23 cm in southeast Alaska (Rutecki and Varosi, 1997a). By the end of their first summer, when sablefish begin migrating to offshore waters, sablefish in the same region grow to average lengths of 35 cm to 39 cm Varosi, 1997b, 1997a). From these observations, we classify any sablefish measuring between 20 cm and 40 cm as inshore demersal juveniles, with smaller fish falling into the offshore epipelagic category (Table 1). Though some juveniles measuring up to 60 cm may reside in inshore waters for an additional year or two (Rutecki and Varosi, 1997a), we classify any larger sablefish as offshore demersal. 3. Methods and materials

Sablefish body-size regression
To estimate sablefish body-size, we calculated linear regression models relating fork length (FL) to the height of vertebral centra. We used vertebrae for our regressions because these are the only skeletal elements of sablefish represented in substantial numbers at Čḯx w icən, but this approach comes with two potential issues. First, if vertebrae from some fish survive in higher proportions than others they will be disproportionately represented in size-frequency distributions. We could control for this problem by estimating body-size with the atlas alone, but only 86 atlas specimens were identified among the 3209 sablefish specimens in the Čḯx w icən fishbone collection (2.7% NISP sablefish ). This sample is large enough to provide an overall pattern of sablefish size frequencies, but it is too small to examine intra-site variation. Instead, we calculated separate regression models for the atlas and other vertebra types and used size-frequencies for the former to test whether there are any biases in the latter.
Second, teleost vertebral columns exhibit extensive morphological variability within individual skeletons, and the dimensions of the vertebral centrum are not consistent across all vertebrae in any individual. In other words, the size of any given vertebra could be a function of both its allometric relationship to body-size and its position in the vertebral column. Our observations of sablefish vertebrae suggest that centrum size is consistent for abdominal vertebrae, but not for caudal vertebraevertebrae that possess a haemal arch (Fig. 2)which shows dramatic decreases in size towards the caudal end of the vertebral column. Therefore, we exclude caudal vertebrae from our analysis and focus on the atlas and abdominal vertebrae alone.
Our reference sample of modern sablefish consisted of 11 individual fish. While minimum sample sizes of 30 to 40 individuals are preferred when performing regression, acceptable models of this type have been produced in the past with as few as 12 reference specimens (Orchard, 2001, pp. 69, 71;Reitz et al., 1987, Table 3). One sablefish was collected from St. John Baptist Bay by Dana Hanselman (Alaska Fisheries Science Center, National Oceanic and Atmospheric Administration), and an additional 10 unprovenienced individuals were provided by Dan Kamikawa (Northwest Fisheries Science Center, NOAA). FL in the reference sample ranged from 32 cm to 60 cm (x̅ = 44 cm) ( Table 2). Prior to collecting our vertebral measurements, Butler prepared two sablefish skeletons (PSU 13-2-1 and PSU 13-2-2) by warm-water maceration, and Nims extracted abdominal vertebrae from the remaining sablefish after baking them under several layers of aluminum foil at 175°C for 30 min.
Nims used Mitutoyo CD-6″CX digital calipers to record the maximum height of each vertebral centrum ( Fig. 2; Table 2; Supplementary  Table 1). Measurements for the atlas were collected three times and then averaged. Measurements for abdominal vertebrae were collected once, and then averaged across all abdominal vertebrae from each individual. All outlier heights were re-measured to check for measurement error. Comparing the heights of abdominal vertebrae across the vertebral column shows that variability is very low across all individuals ( Fig. 3; Table 2). Based on these observations, we are confident that average abdominal centrum height provides an accurate measure for regression analysis.
Using R version 3.4.2, we created linear regression models comparing FL to the atlas centrum height and average abdominal centrum height (Table 3; Fig. 4). For comparison, Nims also calculated models comparing FL to measures of centrum width and length. All six models show statistically significant (p < 0.05) and practically significant (high R 2 values) relationships between FL and centrum dimensions, but as regression models based on centrum height provided the best fit (Table 3) we only used these when estimating the FL of Čḯx w icən sablefish specimens.
One limitation of these models is that our reference set of modern sablefish only represents a fraction of the juvenile size-classes that are of interest; our smallest reference fish has a FL of 32 cm, but sablefish measure around 20 cm when they end their shoreward migration. If there is a change in the allometric relationship between vertebra size and body-size between lengths of about 20 cm and 32 cm, our regression models would make inaccurate FL predictions for smaller juveniles, potentially leading to incorrect assessments of habitat use. Fortunately, there is no reason to expect that this relationship would change over the juvenile life-stage, and our reference set shows no evidence that the relationship changes between the adult and juvenile life-stages. Therefore, we confidently assume that our model can accurately predict FL from sablefish smaller than 32 cm, though in the future we hope to expand the sample size of our reference set to include smaller juveniles, especially those smaller than 20 cm FL.

Čḯx w icən materials
The Čḯx w icən faunal collection comes from a large excavation project carried out by Larson Anthropological Archaeological Services (LAAS) and LEKT members in 2004 (Reetz et al., 2006). All excavated matrix was collected in 10 L buckets and wet-screened through nested 1″ (25.6 mm), 1/2″ (12.8 mm), and 1/4″ (6.4 mm) mesh. At least one 10 L bucket from each micro-stratum identified in the field was also screened through 1/8″ (3.2 mm) mesh and designated a 'complete' ("C") bucket. The fishbone collection used in this study was analyzed during a large-scale zooarchaeological and geoarchaeological analysis initiated by Butler, Kristine Bovy (University of Rhode Island), Sarah Campbell (Western Washington University[WWU]), Mike Etnier (WWU), and Sarah Sterling (Portland State University) that focused on animal remains recovered from two plankhouse structures, and extramural activity and midden areas at Čḯx w icən (for a full history of the Čḯx w icən mitigation project, see Butler et al., 2018a, this issue; and for overall faunal analysis and sampling, see Butler et al., 2018b, this issue).
Butler directed analysis of the fishbone assemblage at PSU from 2012 to 2015 following QC/QA protocols that are consistent with Driver's (2011) recommendations. PSU master's students Kathryn Mohlenhoff, Anthony Hofkamp, Shoshana Rosenberg, and Nims identified fish remains from "C" buckets (and other sample types), and Butler verified and often adjusted all initial identifications under magnification. For a full account of fishbone identification methods and protocols, see Butler et al. (2018c). To evaluate the reproducibility of the Čḯx w icən fishbone identifications, Nims and Butler (2017) conducted a blind reanalysis of 14 "C" buckets (140 L) from three stages of the project and found that our results are highly reproducible.
The analyzed fishbone from Čḯx w icən represents seven chronological zones (CZs) of activity spanning the last 2150 years (Fig. 5;Campbell et al., 2018, this issue). In this study, we compare patterns of sablefish representation between CZ's (Table 4)and to general patterns in regional environmental history (Fig. 5) to explore whether the climate shifts and tectonic events discussed by Hutchinson et al. (2018, this issue) affected catches of sablefish. Here, we assume that local atmospheric and marine temperatures are correlated, and this assumption  Head et al. (2014) a Length at which 50% of sablefish reach maturity. shows a cooling trend in Pacific Ocean waters from 1850 to 1450 BP, followed by a warming period with peak temperatures around 850 BP, and another cooling phase that lasted from 650 to 50 BP. As these trends closely match local reconstructions for atmospheric trends, we use changes in atmospheric temperatures as a proxy for marine SST.
We also contrast patterns of sablefish representation in two households to explore whether sablefish abundances reflect social differences. Excavations at Čḯx w icən revealed remains from two separate plankhouses in Area A1 and Area A4 of the site (Fig. 6). The house in Area A4 was first established during CZ 4 (Fig. 5), and two distinct floor deposits (Floor 1 and Floor 2) date to this period of occupation. Subsequent floor levels from this house date to CZ 5 (Floor 3) and CZ 6 (Floor 4) (Campbell et al., 2018, this issue). The house in Area A1 was first established during CZ 5, after the Area A4 household formed (Fig. 5), and it also contains floor deposits that date to CZ 5 (Floor 1) and CZ 6 (Floor 2). Given the contemporaneous occupations, we compare sablefish use between the two households for CZ 5 and CZ 6 ( Table 5).
Nims measured the centrum height of all sablefish atlases and abdominal vertebrae in the Čḯx w icən fishbone collection with at least one intact centrum face. All vertebrae were recorded three times, and then averaged (Supplementary Table 2). If there was noticeable disagreement between the three measurements for any specimen, Nims

Sablefish size-frequency distributions
The FL estimates for all measurable sablefish vertebrae from Čḯx w icən fall into the demersal stage of juvenile sablefish life history between 20 cm and 40 cm FL ( Fig. 7; Table 1). A total of 41 atlas       the sablefish catch at Čḯx w icən (Table 6; Fig. 8). The mean FL of 27.7 ± 2.6 cm suggests sablefish were captured in the period between fall, when juveniles first arrive inshore, and their second summer, when southeast Alaskan sablefish approach average lengths of 30 cm (Rutecki and Varosi, 1997a). The same pattern is evident across all CZ's that have a minimum sample size of 29 NISP, which show < 1 cm of variation in mean FL for all sablefish captured across 1500 years of Čḯx w icən's occupation ( Fig. 9; Table 6).

Spatio-temporal variation in sablefish abundance
Sablefish relative abundance varies moderately throughout the occupation of Čḯx w icən (Fig. 10). Sablefish is ranked as the sixth most common fish taxon in CZ 1, and then falls to seventh most common in CZ 2, though its proportional representation does not change. From CZ 3, sablefish rank-order increases to the third most common fish and then alternates between second and third most abundant in the remaining four CZ's. While herring (Clupea pallasii) clearly dominate the Čḯx w icən fishbone record, these distributions show that juvenile sablefishalong with flatfishes (Pleuronectiformes), salmons (Oncorhynchus sp.), staghorn sculpin (Leptocottus armatus), and dogfish shark (Squalus suckleyi)were a key and reliable fish species for 2150 years (Fig. 10).
Comparing this pattern with regional histories of climate trends and earthquake events (Fig. 5) shows these processes had little to no longterm effect on sablefish catches at Čḯx w icən. Current research on sablefish population dynamics has revealed complex relationships between year-class strength and environmental factors (Echave et al., 2012;McFarlane et al., 1997;Shotwell et al., 2014;Sogard, 2011). On one hand, growth rates in larval and early juvenile-stage sablefish increase with sea surface temperature (SST), enhancing year-class strength (McFarlane et al., 1997;Sogard, 2011). However, Shotwell et al. (2014) concluded that increases in primary productivity associated with cooler SSTs are a more important control on sablefish recruitment. Therefore, we assume that cooler periods would enhance sablefish year-class strength and encounter rates. Using atmospheric trends as a proxy for SST (see above), sablefish abundance might be expected to increase at Čḯx w icən during the Late Antique Little Ice Age (~1600-1250 cal BP in this region [after Hutchinson et al., 2018, this issue]) and the Little Ice Age (~750-250 cal BP) (Fig. 5). These cool periods overlap with every CZ at Čḯx w icən except CZ 1, making it difficult to compare changes in these assemblages to changes in climate    regime. However, the absence of any noticeable change during CZ 4 and CZ 5, which overlap more with the Medieval Warm Period (~1250-750 cal BP) than either cool period, suggests changes in climate did not appreciably affect human capture rates of juvenile sablefish.
Meanwhile, there is no a priori reason to suspect that sablefish populations would be affected by earthquake or tsunami events for more than a season. As adults reside and spawn offshore, any short-term disruptions to inshore environments could only affect a single year-class of juveniles. Tectonic events could affect sablefish catch sizes indirectly if the fishing gear or watercraft necessary for fishing in waters 20 m to 60 m deep were lost or destroyed in a tsunami, but this would only have a long-term effect in the event that people did not replace their equipment and stopped fishing from boats. While there is clear evidence of variation in occupation intensity at Čḯx w icən following earthquake event "U" (Hutchinson et al., 2018, this issue), the consistency in sablefish representation over this period shows that village inhabitants were able to procure juvenile sablefish at the same rates as before (Fig. 10).
Despite the overall consistency in sablefish relative abundance from CZ 4 to CZ 7 when the aggregated assemblage is considered (Fig. 10), there are differences in sablefish representation over the occupation of the two houses associated with Areas A1 and A4 of Čḯx w icən (Fig. 11), suggesting social factors mediated fish use. Butler et al. (2018b, this issue) have examined all animal classes to explore whether households were autonomous or communal in resource use. They predict households that operate more communallysharing gear or access to harvest areas or captured fishwould generate faunal deposits that are more similar relative to households that operate autonomously. In CZ 5, Butler et al. (2018b, this issue) found that fish use was relatively consistent between households, suggesting a communal social structure. In CZ 6, the pattern changes, with stark differences in fish representation between houses, suggesting a more autonomous social structure.
The sablefish record between households and CZs is consistent with this broader pattern in fish representation. In the earlier CZ 5, sablefish remains are modestly represented in both houses (Fig. 11). In CZ 6, sablefish representation differs greatly between households: sablefish is ranked second after herring in the Area A4 house, but it is ranked 10th in the Area A1 house. Butler et al. (2018b, this issue) analysis found that the two households also vary greatly during CZ 6 in representation of herring (A4: high; A1: low), staghorn sculpin (A4: low; A1: high), and salmon (A4: high; A1: low). Overall, the record shows that the households operated much more independently in the later time period. Given the ethnographic link of sablefish to prestige and gifting, higher abundances at Area A4 indicate occupants of this household may have had privileged access rights to sablefish or ownership of fishing grounds where young-of-the-year congregated. While the specific social factors involved remain uncertain, the contemporaneity and shared locality of the two households demonstrates these patterns are not linked to environmental drivers.

Discussion and conclusions
Our reconstructed sablefish body-size distributions reveal that Čḯx w icən's inhabitants fished exclusively for inshore demersal juveniles that today inhabit bays and inlets at depths of 20 m to 60 m ( Fig. 8; Table 1). There is a clear unimodal distribution in sablefish FL estimates that suggest they were largely captured between late fall and spring. These juveniles are one of the most abundant fishes among a group of fish taxa that are consistently well-represented in each Čḯx w icən fishbone assemblage, and they provided a reliable resource through longterm climate shifts and short-term socio-environmental disruptions caused by tectonic events (Fig. 10).
These patterns provide an intriguing contrast with available ethnohistoric information about traditional sablefish use, which largely relate to mature sablefish. Swan's (1887) observations specifically highlighted the importance of adult sablefish in the Salish Sea and cited their depth as a limiting factor in sablefish consumption. Discussions of sablefish at Haida Gwaii also imply that sablefish were primarily captured as adults with long-line trawls measuring up to 350 m in length (Hobler, 1978;Swan, 1887). However, Arima and Dewhirst's (1990) account of fishing for sablefish with lures from canoes may indicate that sablefish were being captured as juveniles from western Vancouver Island as lures would only be effective for attracting demersal fishes in relatively shallow waters, which would not be inhabited by adults. Considering the ethnohistoric and the archaeological evidence together, it is clear that Native American/First Nations peoples in northwestern North America took sablefish using a variety of methods that intersected with a number of stages in sablefish life history. Importantly, our work highlights problems with over-reliance on ethnographic records in interpreting archaeological assemblages (for other examples, see Moss, 1993;Orchard and Wigen, 2016).
particularly favor, or indirectly through greater access to watercraft, necessary fishing gear, labor, or other resources. Alternatively, if Swan's (1887) observation that sablefish were a chiefly luxury held true at Čḯx w icən, the higher frequencies of sablefish in Area A4 may reflect higher rates of gifting and/or tribute to the house's inhabitants.
While we are not in a position to answer our original question about why sablefish are so scarce in northwest North American archaeology, the insights from Čḯx w icən provide much needed context that can help guide future research on this topic. Importantly, the consistency in sablefish abundance at Čḯx w icən indicates the species can be productively and sustainably harvested at local scales for millennia, despite major shifts in climate and tectonic disruption. This suggests the absence of sablefish is probably not related to environmental perturbations, as sablefish and their juveniles are apparently able to tolerate a wide range of marine conditions. But if juvenile populations were as variable in the past as they are today, environmental factors could still play a role in determining the archaeological distributions of sablefish remains through their controls on year-class-strength and influencing where sablefish congregate. Unfortunately, these relationships remain poorly understood.
Our research also confirms our assumption that, at least in some contexts, sablefish were considered an attractive resource and some people may have had privileged access to the species. This could lead to high degrees of intra-site variation in sablefish deposition, affecting our ability to recognize sablefish archaeologically at sites where time and budget constraints prevent the implementation of an intensive sampling-strategy. However, sablefish at Čḯx w icən are widely distributed throughout the site, and the number of sablefish remains identified from Area A1 alone greatly outnumber the total number of sablefish remains recorded at all other northwestern North American archaeological sites despite the marked differences in relative abundance that we documented between the two houses. We doubt that intra-site variation in sablefish frequencies could be so extreme that sablefish would only be recovered from a single context, and it seems unlikely that intra-site variability in sablefish distribution accounts for the rarity of sablefish in most archaeological sites.
Finally, though the archaeological record presents an uninterrupted history of sablefish procurement by the inhabitants of Čḯx w icən, juvenile sablefish are not known to congregate in or around Čḯx w icən today. Most juvenile sablefish studies have focused their efforts on regions north of Vancouver Island (Kennedy and Smith, 1972;Mason et al., 1983;McFarlane and Beamish, 1983;Rutecki and Varosi, 1997a). It is not clear when juvenile sablefish stopped congregating in the vicinity of Čḯx w icən, or why they are no longer abundant, but future work on these questions could help resolve the factors that lead juveniles to their preferred nurseries. Combining such information with reconstructions of coastal environmental conditions could help predict where sablefish would or would not have congregated in the past, allowing us to test hypotheses about whether the archaeological scarcity of sablefish is related to the distribution of sablefish populations in the past.
Like other focused studies of individual fish taxa (e.g. herring - McKechnie et al., 2014;halibut -Orchard and Wigen, 2016;lampreys -Smith and Butler, 2008), we found that paying close attending to particular species like sablefish can deepen our understanding of, and our appreciation for, the people who skillfully managed these food webs in the past.
Supplementary data to this article can be found online at https:// doi.org/10.1016/j.jasrep.2018.06.028. loaned comparative fish skeletons and/or assisted with identifications. Dan Kamikawa (NOAA) and Dana Hanselman (Alaska Fisheries Science Center, NOAA), collected the sablefish used in the study; Jessica Miller (Oregon State University) and Ed Backus (Collaborative Fisheries Associates) delivered them to Butler. Former PSU students Tony Hofkamp, Kathryn Mohlenhoff, Emily Rocha, Shoshanna Rosenberg, and Laura Syvertson assisted in the lab. Bruce Floyd (University of Auckland) provided invaluable statistical advice. Kris Bovy and two anonymous reviewers provided valuable comments on an earlier version of this manuscript. We alone are responsible for any errors. Work was funded by the National Science Foundation [grant numbers 1219468, 1353610, and 1663789].