Differential use of tidal delta, shoreline, and neritic habitats by natural‐ and hatchery‐origin juvenile Chinook Salmon

Abstract

Objective

Conservation and recovery efforts for depressed populations of wild Chinook Salmon Oncorhynchus tshawytscha are improved by detailed classifications of basic natural history traits for juveniles across habitats. The Skagit River system in northern Puget Sound provides a unique opportunity to study the estuarine ecology of Chinook Salmon since it is home to the healthiest remaining natural‐origin Puget Sound Chinook Salmon spawning populations, and despite major anthropogenic habitat changes, the system still has extensive areas of estuarine habitat.

Methods

We evaluated density and length of juvenile natural‐ and hatchery‐origin Chinook Salmon across tidal delta, shallow intertidal, intertidal–subtidal, and neritic habitats in the Skagit River estuary from February to November 2002.

Result

Juvenile Chinook Salmon were captured in all habitats and months sampled, and clear seasonal transitions through habitats were observed, as were habitat and seasonal differences in fish density, length, and relative abundance of hatchery‐origin versus natural‐origin fish. Natural‐origin fish showed a protracted seasonal distribution across all habitats, and the relative abundance of hatchery‐origin fish was lowest in the tidal delta and highest in the neritic zone. Mean fork length increased as fish moved downstream and offshore, and hatchery‐origin fish were consistently larger than natural‐origin fish in all habitats and months. Hatchery‐origin fish from multiple source populations were recovered in Skagit Bay, but only individuals from the Skagit River were captured in tidal delta and shallow intertidal habitats, and the likelihood of capture in Skagit Bay was inversely related to the distance from the basin of origin.

Conclusion

These results confirm that an extensive and diverse range of estuarine habitats is used by juvenile Chinook Salmon through much of the year. Contrasts between hatchery‐ and natural‐origin fish suggest that estuaries—especially shallow fringing habitats in the tidal delta and along marine shorelines—are differentially important for natural‐origin fish and that the potential for interactions between hatchery‐ and natural‐origin fish differs depending on the habitat and time of year.

Impact statement

Natural‐ and hatchery‐origin Chinook Salmon differ considerably in their use of estuarine and nearshore marine habitats, providing important context for restoration and recovery evaluation and assessment.

INTRODUCTION

By virtue of their complex life histories, anadromous species reside and migrate through multiple habitat types during their life cycle. In particular, Pacific salmon Oncorhynchus spp. rely on several habitats for foraging opportunities during their migrations, and growth in these habitats allows them to achieve a size that is critical for continued survival (Beamish and Mahnken 2001; Duffy and Beauchamp 2011; Sawyer et al. 2023). Loss or degradation of habitats used by salmon during their life cycle may significantly reduce the productivity, resilience, and persistence of salmon populations (Greene and Beechie 2004; Bottom et al. 2005b; Greene et al. 2010). However, data on habitat impacts alone are insufficient to assess threats to populations because of differential habitat use. Therefore, to advance the conservation and recovery of salmon populations, it is critical to understand how salmon use and move among habitats throughout their life cycle.

Estuaries are used by all juvenile anadromous salmon for migration, foraging, physiological transition, and refuge from predators during their transition from freshwater to marine habitats (Simenstad et al. 1982; Thorpe 1994). Estuaries are particularly important for Chinook Salmon O. tshawytscha, which use estuarine habitats more extensively than any other Pacific salmon species (Healey 1982, 1991; Simenstad et al. 1982; Thorpe 1994; Aitkin 1998; Beamer et al. 2024). Estuarine habitats are also highly degraded along the west coast of North America, including within Puget Sound (Collins et al. 2003; Collins and Sheikh 2005; Brophy et al. 2019). As such, estuaries are a primary focus of Chinook Salmon conservation and recovery efforts in the region (Bottom et al. 2005b; Shared Strategy Development Committee 2007; Puget Sound Recovery Implementation Technical Team 2015; Ellings et al. 2016). However, an incomplete understanding of juvenile Chinook Salmon migration and rearing in estuarine environments limits our ability to evaluate the condition of natural‐origin Chinook Salmon populations and to develop effective plans for their recovery. Consequently, estuarine field studies—especially in systems with remnant natural‐origin fish and estuarine habitats—are needed to provide natural history information.

We examined spatial and temporal patterns of density and individual length in juvenile natural‐ and hatchery‐origin Chinook Salmon within tidal delta, nondelta shoreline, and neritic habitats across seasons. The goal of this effort was to improve our basic understanding of how natural‐ and hatchery‐origin Chinook Salmon use estuarine habitats and to inform future management actions and recovery planning. In this study, we tracked the density and fork length of juvenile Chinook Salmon across tidal delta, shallow intertidal, intertidal–subtidal, and neritic habitats of the Skagit River estuary throughout the entire out‐migration period and compared patterns among natural‐ and hatchery‐origin fish. Understanding potential differences in habitat use between hatchery‐reared and natural‐origin populations can help to focus and/or prioritize restoration efforts and can aid the recovery of natural‐origin Chinook Salmon.

METHODS

Study area, description of estuarine habitats, and site selection

Puget Sound Chinook Salmon are among the most socioeconomically important fish species in the Pacific Northwest and are listed as threatened under the Endangered Species Act (Myers et al. 1998; Ruckelshaus et al. 2006). Recovery efforts throughout the region have spurred habitat restoration in rivers (Hall et al. 2018; Polivka and Claeson 2020) as well as estuaries (Ellings et al. 2016) and nearshore environments (Francis et al. 2022). Restoration efforts across habitats are intended to address the complex life history of Chinook Salmon and support important life history diversity within and among populations. In the Puget Sound region, the vast majority of juveniles exhibit a subyearling life history, leaving freshwater within their first year of life (Anderson and Topping 2018). A smaller percentage of fish exhibit life histories that involve rearing in freshwater for a year before migrating to sea (Zimmerman et al. 2015).

The Skagit River basin is home to the most abundant run of natural‐origin Puget Sound Chinook Salmon, supporting 6 of 22 extant population segments (of which there were 33 historically) identified in the Puget Sound Chinook Salmon Evolutionarily Significant Unit (Myers et al. 1998; Ruckelshaus et al. 2006; Figure 1, inset). The Skagit River basin produces both subyearling (freshwater age‐0) and yearling (freshwater age‐1) natural‐origin Chinook Salmon as well as hatchery‐origin groups that are used as indicator stocks for natural‐origin populations. The subyearling out‐migration from freshwater habitats is comprised of early migrating fry‐sized fish and later migrating parr‐sized fish (Zimmerman et al. 2015). The parr cohort rears for a significant length of time (weeks to months) in riverine habitats, while the fry cohort is smaller and migrates soon after emergence.

The Skagit River tidal delta and Skagit Bay are part of the larger Puget Sound fjord estuary complex located in northwestern Washington, United States (Figure 1). We divided estuarine habitats into (1) wetlands and channels within the Skagit River's tidal delta, (2) the intertidal shoreline of Skagit Bay seaward of the tidal delta, and (3) subtidal portions of Skagit Bay. The wetlands of the tidal delta extend from the riverine tidal zone, where freshwater is tidally pushed but not mixed with marine water (Day et al. 1989), through the estuarine forest transition and estuarine emergent marsh zones, where freshwater mixes with salt water (Beamer et al. 2024). Within these zones, diverse estuarine habitats are formed and maintained by tidal and riverine processes, creating a mosaic of distributaries, blind tidal channels, and vegetated wetlands. In Skagit Bay, intertidal habitats include a variety of geomorphic mudflats and beaches (Johannessen and MacLennan 2007; Shipman 2008) as well as rocky intertidal environments. In subtidal areas, habitat for juvenile salmonids occurs primarily in the neritic environment (surface waters overlaying the sublittoral zone). Due to the large output of freshwater from the Skagit River, all these areas exhibit lower salinity than is found more generally across the region (Moore et al. 2008; Figure 2).

Today, the contiguous area of the Skagit River tidal delta consists mostly of area in the vicinity of Fir Island, but it also includes a fringe of estuarine habitat extending from southern Padilla Bay to the north end of Camano Island. In 1991, the tidal delta footprint for this area was 3118 ha (Beamer et al. 2005). Prior to diking, dredging, and filling in the delta (circa 1860s), 11,483 ha of tidal delta footprint existed in the same area (Collins 2000). These estimates of tidal delta habitat area account for gains in delta habitat caused by progradation occurring between the 1860s and 1991 (Hood 2005; Hood et al. 2016). The intertidal area of Skagit Bay is 7716 ha, and the subtidal surface area is 6523 ha.

Fish collection

Four different gear types were used for fish collection: fyke traps for tidal delta blind channels; a small beach seine for shallow intertidal shorelines; a larger beach seine for deeper shorelines (intertidal–subtidal fringe); and a Kodiak trawl, or “surface trawl,” for neritic sites (Table 1). The sampling period for each habitat/gear type was intended to encompass the full seasonal distribution of juvenile Chinook Salmon in each habitat type (tidal delta, intertidal shallow, intertidal–subtidal, and neritic). Sampling took place from February through August twice per month on spring tide series in the tidal delta. Beach seining was conducted twice per month on neap tide series in intertidal shallow and intertidal–subtidal habitats from February through October. Sampling with the surface trawl was performed monthly from March through November; however, in February, we sampled twice with the surface trawl to determine whether early migrating Chinook Salmon fry were present in neritic areas of Skagit Bay.

For each sampling method, we calculated a gear‐specific set area and recapture efficiency (RE) to estimate fish density (Table 1). Set area ranged between 0.01 and 1.00 ha but remained consistent within each method across the sampling season (Figure S5). The REs for the fyke trap and beach seines were estimated using standard mark–recapture techniques and were adjusted for habitat attributes that were unique to each method. Recapture efficiencies for the Kodiak trawl were taken from Wilder and Ingram (2006). Total catch was expanded based on RE and then was divided by set area to obtain an estimated density. For a detailed description of methods, set area, and RE estimates, see the Supplemental Information (available in the online version of this article).

In the tidal delta, we sampled seven blind tidal channel sites using fyke‐trap methods (Figure 1). These sites were selected to represent the three estuarine wetland zones (estuarine emergent, estuarine scrub–shrub, and riverine tidal) that are present within the Skagit River tidal delta and its major areas: north fork, south fork, and the bay fringing area between each fork. We sampled 17 shallow intertidal sites by using a small beach seine, and six sites spanning the intertidal–subtidal fringe were sampled by using a large beach seine in Skagit Bay. Objectives for selection of beach seine sites were broad spatial coverage alongshore and beach type diversity. We were unable to sample the rocky or bedrock beaches that account for 22% of the shoreline length in the study area. In subtidal neritic habitat, we sampled a total of 12 surface trawl sites. Objectives for neritic habitat site selection were proximity to beach seine sites, baywide spatial coverage alongshore (at depths just sufficient for the fishing gear to be deployed without hitting the bottom), and accessibility by boats and fishing gear at any tidal stage.

Fish processing

For all fish capture methods, fork lengths (mm) were measured on up to 20 individual Chinook Salmon from each set. All Chinook Salmon were visually examined for clipped adipose fins and were checked for coded wire tags (CWTs) by using a handheld detector wand (Northwest Marine Technologies Inc.). Chinook Salmon with adipose fin clips and/or CWTs were considered marked, while all other Chinook Salmon were considered unmarked. Marked Chinook Salmon were assumed to be of hatchery origin, whereas unmarked Chinook Salmon were assumed to be of natural origin.

Hatchery mark rates, origin, release date, and release size data were from two sources: the Regional Mark Information System (RMIS; http://www.rmpc.org) and the Alaska Department of Fish and Game's CWT Laboratory (http://tagotoweb.adfg.state.ak.us/CWT/reports/cwtrelease.asp?). For hatchery‐origin Chinook Salmon originating from the Skagit River, 99.9% of the 764,570 fish that were released (yearlings and subyearlings combined) could be identified as hatchery‐origin fish based on the presence of an adipose fin clip or a CWT (Appendix Table A.1). Mark rates for other regional hatcheries were generally greater than 90% except for the Fraser River, Nooksack River, and Tulalip Bay, with mark rates of 12.3, 50.4, and 14.5%, respectively. Thermal marking of otoliths was used in two basins (Nooksack River and Tulalip Bay) for some hatchery release groups that did not have CWTs or adipose fin clips. However, otolith marks are not useful for field determination of natural‐ or hatchery‐origin fish and were not used in this study. Up to 10 coded‐wire‐tagged Chinook Salmon per set were sacrificed to identify their hatchery of origin.

The vast majority of hatchery releases in Puget Sound consist of the subyearling life history type (Table A.1; RMIS). Although yearling releases do occur, they are less widespread and abundances are far lower than subyearling releases, especially within the study area (Table A.1). Hatchery releases of summer/fall Chinook Salmon typically occur between mid‐May and mid‐June. Within the Skagit River, spring Chinook Salmon are typically released in mid to late April. Release distance from the Skagit River estuary was measured as the most likely migration pathway (by water) from the mouth of the hatchery release river to the center of the Skagit River estuary study area. We standardized the recoveries of coded‐wire‐tagged juvenile Chinook Salmon in the Skagit River estuary to examine whether the distance from the river of origin for hatchery fish influenced the frequency of encountering a particular hatchery‐origin fish in the Skagit River estuary. Hatchery release data from the read CWTs were expanded to the results for all 688 fish with CWTs. We then divided these estimates by the total number of hatchery fish that were released from each basin to calculate the percentage of hatchery‐origin fish that were caught in the Skagit River estuary from each basin. The total number of fish released from each basin included unclipped–untagged, clipped–untagged, clipped–tagged, and unclipped–tagged individuals.

Environmental variables

Temperature and salinity measurements were taken using an electronic meter (YSI Model 30) at all sampling events. At surface trawl sites, water was drawn by the primary vessel's deck hose from a depth of approximately 1.2 m to measure temperature and salinity. Water temperature and salinity were measured just under the water surface near the midpoint of beach seine set areas and just upstream of the traps at fyke‐trap sites. Water depth was measured at the maximum depth within the beach seine set area, just upstream of the trap at the start of trapping (high tide) for fyke‐trap sites, and near the midpoint of the tow for surface trawl sites.

Statistical analysis

We used general linear models and generalized linear models to evaluate (1) the influence of habitat type and month on the densities, lengths, and relative abundances of hatchery‐ and natural‐origin Chinook Salmon, respectively; and (2) the influence of habitat type, month, and fish origin (natural or hatchery) on individual fish length. Models including all or subsets of our selected predictor variables were compared using Akaike's information criterion (AIC; Akaike 1974; Anderson et al. 2000; Burnham and Anderson 2002; Dayton 2003). The AIC scores are improved (i.e., made smaller) by better model fits and are penalized for additional parameters. In general, models with AIC values differing by 2 units or less relative to the AIC of the best model are considered to have substantial empirical support, models with AIC differences of 4–7 are still plausible but much less likely, and models with AIC differences greater than 10 have essentially no support (Burnham and Anderson 2002). In addition, we calculated AIC model weights and the proportion of deviance explained for each model to aid interpretation of our model selection process. We used standard model diagnostics to ensure proper model fit and specification.

An additional factor for origin was included in models for length, as densities for hatchery‐ and natural‐origin fish were modeled separately. Interactions between habitat and month were included in the models of density, proportion natural origin, and length because of the migratory behavior and ontogenetic habitat shifts characteristic of juvenile salmon.

RESULTS

Sampling effort and environmental variables

Overall, 1195 individual net sets were successfully completed during the study (Table 2). Both surface water temperature and salinity varied by month and among the four habitat types (Figure 2). Temperature showed clear seasonal changes in all habitats; salinity also demonstrated seasonal changes except at tidal delta sites. Mean water temperature was between 4°C and 16°C across habitats over the sampling period and steadily increased from February to August. The tidal delta sites had the widest range in average temperature (4.5–15.6°C) and the steepest seasonal increase. Mean salinity was between 2 and 5 practical salinity units (PSU) in the tidal delta sites and was fairly consistent throughout the sampling period. The other three habitat types had much higher mean salinity values overall, ranging from approximately 18 to 28 PSU over the sampling period. The seasonal pattern in mean salinity at these more marine sites corresponded to flow in the Skagit River. Salinity declined from spring into June before again increasing into fall. Declines in salinity from spring to summer were associated with increased freshwater flow from spring snowmelt. Water depth at sampling sites changed little over the season; overall, depth averaged 0.9, 0.7, 3.2, and 8.6 m for tidal delta, intertidal shallow, intertidal–subtidal, and neritic habitats, respectively.

Densities of natural‐ and hatchery‐origin Chinook Salmon

Juvenile natural‐origin Chinook Salmon were caught during all months sampled and showed a clear seasonal transition through the habitats (Figure 3). Out‐migration of natural‐origin fish was well underway in three of the four habitats by the start of the sampling period in February, but only six Chinook Salmon were caught in neritic habitats during February–April. Peak densities of natural‐origin fish occurred early (February–May) in tidal delta and intertidal shallow habitats, but density did not peak until August in neritic habitats. Densities of natural‐origin Chinook Salmon were consistently higher in the tidal delta in all months through June compared to other habitats before densities in neritic habitats increased (i.e., during late summer). In contrast, hatchery‐origin fish were not caught until May and differences in density across habitats were lower, especially in the intertidal shallow habitats (Figure 4). Peak densities of hatchery‐origin fish occurred in June for most habitats except the neritic habitats, where the peak occurred in July.

Natural‐origin fish also showed a much broader seasonal distribution than hatchery‐origin fish in all habitats (Figure 3). Natural‐origin fish were present from February through August in three of the four habitats sampled. Natural‐origin fish slowly transitioned from shallow habitats to intertidal–subtidal and neritic habitats beginning in April and June, respectively. By August, the natural‐origin fish had transitioned primarily to neritic habitats. The presence of hatchery‐origin fish was much more constricted, indicating a rapid progression within and through tidal delta, intertidal shallow, and intertidal–subtidal habitats (Figure 4).

The proportion of natural‐origin Chinook Salmon varied by habitat and month (Figure 5). Natural‐origin fish were predominant in most months across all habitat types, and hatchery‐origin fish did not occur before May. Hatchery‐origin fish were generally rare within the tidal delta and intertidal shallow habitats, but they occurred in higher proportions in the intertidal–subtidal and neritic habitats. However, although the proportion of natural‐origin fish decreased through time in the intertidal–subtidal habitats, the proportion increased during the same period in neritic habitats.

Model selection based on AIC indicated that both habitat type and sampling month influenced the density and proportions of hatchery‐ and natural‐origin fish (Table 3). Models that included the interaction between habitat and month had the most explanatory power and provided the best fit to the data (Figures 3–5). Month was included in the top‐three candidate models, indicating strong temporal trends in the density and relative abundance of natural‐ and hatchery‐origin fish.

Chinook Salmon length

Chinook Salmon fork length generally increased in all habitats over the season, ranging from approximately 40 mm for the few fish captured during winter and early spring to over 120 mm for those captured in the fall (Figure 6). One obvious exception to this pattern was the abrupt increase and subsequent decrease in April and May, which tended to be larger in hatchery‐origin fish than in natural‐origin fish, with more narrow size distributions.

The proportion of fish representing different size‐classes differed among habitats and between natural‐ and hatchery‐origin fish (Figure 7). The smallest size‐class (<50 mm fork length) was only represented among natural‐origin fish and was overwhelmingly distributed within the tidal delta, intertidal shallow, and intertidal–subtidal habitats. Nearly all natural‐origin Chinook Salmon in the tidal delta were less than 75 mm fork length, whereas 50% of the fish captured in neritic habitats exceeded 100 mm fork length. There was a relatively clear size transition out of the tidal delta and intertidal shallow habitats and into the neritic habitats at 75–100 mm for natural‐origin fish. Size‐classes of hatchery‐origin fish showed less separation among habitats, although the smallest size‐classes still occurred predominantly in the tidal delta and intertidal shallow habitats. The distribution of size‐classes for hatchery‐origin fish in neritic habitats was consistent with the patterns for natural‐origin fish.

Statistical analysis of length data indicated that the model including habitat, month, origin, and the month × origin interaction was overwhelmingly the best model among those examined (Table 4). Models that individually considered factors of habitat, month, and origin generally performed very poorly, indicating the importance of these predictors in combination, but the rank importance of individual variables was greatest for month, followed by habitat and then origin.

Origin of coded‐wire‐tagged fish

Over the entire sampling period, we caught 688 juvenile Chinook Salmon that were marked with CWTs, and we read 38% of the tags to determine hatchery origin (Table 5). In tidal delta habitat, 120 fish with CWTs were caught at one fyke‐trap site on a single day (June 12, 2002) after a Skagit River hatchery release.

Over 85% of the coded‐wire‐tagged juvenile Chinook Salmon caught in the Skagit River estuary were from hatchery releases originating in the Skagit River watershed (Figure 8). The remaining juvenile Chinook Salmon with CWTs were from hatchery releases originating in seven different watersheds outside of the Skagit River watershed (Table A.1). The most common non‐Skagit River hatchery‐origin fish were from the Samish River, located 50 km (by water) north of the Skagit River tidal delta (Figure 1, inset). Three other systems with similar numbers of juvenile hatchery‐origin Chinook Salmon captured in Skagit Bay were the Nooksack, Stillaguamish, and Snohomish rivers (including Tulalip Bay, located just north of the Snohomish River estuary).

We caught 17,112 unmarked (presumed natural‐origin) juvenile Chinook Salmon in the Skagit River tidal delta. Only hatchery‐origin fish from the Skagit River were identified via CWTs in tidal delta habitat, and only 644 unmarked fish (those with no external fin mark and no CWT present) were estimated to have been released from Skagit River hatcheries in 2002. Given this information, we presumed that the likelihood of misidentifying natural‐origin fish in the tidal delta was low. Although unmarked releases from Samish Hatchery occurred during the study period, other studies have isolated their presence to Swinomish Channel (E. M. Beamer, unpublished data). Results were different for Skagit Bay. During 2002, we caught 3502 juvenile Chinook Salmon, assumed to be of natural origin, in Skagit Bay by using beach seines and surface trawls. Based on estimates of unmarked releases originating from basins known to contribute to the Skagit Bay juvenile Chinook Salmon population, we estimated that 102 (2.9%) of the 3502 assumed natural‐origin fish that we caught were actually of hatchery origin.

All non‐Skagit River hatchery‐origin fish identified via CWT were caught in beach seines from deeper intertidal–subtidal habitat or by surface trawl in neritic habitat (Figure 8). Clear temporal patterns in the presence of hatchery‐origin fish from a particular hatchery were not apparent in any of the four habitat types except that hatchery Chinook Salmon of any origin were not present in Skagit Bay until May, largely due to the timing of hatchery releases (Table A.1).

Standardized CWT data from juvenile Chinook Salmon recovered in the Skagit River estuary showed that the distance from the natal river for hatchery‐origin fish influenced the frequency of encountering a particular hatchery‐origin fish in the Skagit River estuary. Distance from the river of origin to the capture location was negatively correlated with the percentage of hatchery‐origin fish caught (Figure 9).

DISCUSSION

In this paper, we highlight extended residency in multiple estuary habitats by predominantly subyearling, natural‐origin juvenile Chinook Salmon, in contrast to individuals of the same life history but of hatchery origin from the same watershed. These patterns are consistent with other studies on this topic (Reimers 1973; Levings et al. 1986; Johnson et al. 2011; Weitkamp et al. 2014; Chalifour et al. 2020; Greene et al. 2021), but our study is novel in rigorously documenting such strong differences in density and size structure. The striking results were observable primarily due to frequent and extensive surveys; high marking rates of hatchery‐origin fish, allowing us to reliably differentiate hatchery‐ and natural‐origin individuals; and a suite of intact estuary habitats that were relatively isolated from an influx of nonsystem migrants.

Patterns in habitat‐specific density and size

Densities of natural‐origin fish were considerably higher than those of hatchery‐origin fish and showed a more protracted period of residence in three of the four habitats sampled. Natural‐origin fish inhabited multiple habitats for a greater period of time, suggesting a much greater reliance on estuarine habitats than their hatchery counterparts. This dynamic likely reflects the life history diversity of Chinook Salmon populations responding to dynamic conditions, as reflected in both density‐independent (dynamic abiotic variables; Greene et al. 2005; Beamer et al. 2024) and density‐dependent (e.g., competition for food or space favoring earlier natural‐origin colonizers in different habitats) mechanisms. Although we do not definitively demonstrate the potential mechanisms in this study, other studies of salmon using different habitats have highlighted some of these possibilities (Reimers 1973; Greene and Beechie 2004; Beamer et al. 2016; Greene et al. 2021). Densities for both natural‐ and hatchery‐origin fish declined substantially as fish moved from the tidal delta to neritic habitats—presumably the consequence of migration, mortality, and an increasing area of available habitat.

Variability in size among life history types (e.g., subyearling vs. yearling migrants) and origin can influence habitat use during early life stages (Chamberlin et al. 2022). Differences in size distributions between natural‐ and hatchery‐origin subyearling fish were evident and reflected in the observed patterns of habitat use. The smallest size‐classes (<75 mm) were overwhelmingly natural‐origin fish and were the primary inhabitants of both tidal delta blind channel and shallow intertidal habitats. The lack of hatchery‐origin fish within the smaller size‐classes can be explained by standard hatchery practices whereby fish are released at larger sizes, which presumably influences their specific habitat use patterns. Although these larger hatchery‐origin fish were observed within the tidal delta and shallow intertidal habitats, the extent or duration of use within the tidal delta blind channel and shallow intertidal habitats was considerably shorter than that of their natural‐origin counterparts.

The combination of abundance and size data implies that tidal delta and shallow intertidal habitats are important rearing areas for subyearling individuals but are largely bypassed by older out‐migrants that have extensively reared and obtained larger size in rivers or at hatcheries (e.g., yearlings). This pattern is mirrored in other salmonids (Healey 1982; Simenstad et al. 1982; Dawley et al. 1986) as well as in other diadromous species (reviewed by Potter et al. 2015). It also suggests that size trajectories and the sequence of habitats used may depend in large part upon habitat types formerly occupied at smaller sizes or during earlier periods of the migration (Gamble et al. 2018). In Chinook Salmon, we expect several emergent patterns of habitat use, including limited tidal delta and shallow intertidal rearing for larger sizes (>75 mm) at out‐migration from rivers, growth that depends upon the size achieved in earlier stages, and low densities in deeper water habitats when shallow‐water habitats are accessible. We expect that habitat dependencies like these exemplify carryover effects that can influence size trajectories, growth dynamics, maturation, and subsequent survival in marine waters (Duffy and Beauchamp 2011; Clarke et al. 2016; Chasco et al. 2021; Wilson et al. 2021).

Potential system‐specific limitations

Our results highlighting temporal progressions through habitats and size‐dependent residence are likely common patterns across systems and anadromous populations (Reimers 1973; Levy and Northcote 1982; Bottom et al. 2005a; Roegner et al. 2012), but it is worth considering how system‐specific life history diversity and abundance may modify the comparison. Our findings largely reflect subyearling migrants and may be less relevant for systems that lack similar life history diversity or that are dominated by yearling migrants and/or hatchery‐origin populations (Quinn 2005). Likewise, populations at lower levels of abundance, especially for subyearling life history types, may experience reduced competition for space or food in estuary habitats, resulting in extended estuarine residence of larger natural‐origin individuals or even hatchery‐origin fish (Hayes et al. 2019). However, where estuarine habitat is reduced relative to population size, we may expect more rapid density‐dependent habitat transitions or mortality (see Greene et al. 2021).

Additionally, for estuaries that are subject to the intermingling of many spawning populations (e.g., the Columbia River estuary) and especially where population abundances differ, isolation of strong habitat‐specific residence patterns may be more difficult (see Burke 2004; Bottom et al. 2005b). Most of the nonnatal stocks/populations (both natural and hatchery origin) in Skagit Bay originate from areas so distant that individual fish have likely achieved sizes that are too large to be present in the tidal delta and shallow intertidal habitats (Figures 8 and 9). Additionally, population status differences in natural‐origin stock abundance compared to the Skagit River could inhibit the detection of nonnatal stocks in the Skagit River estuary. For example, the nearest nonnatal, natural‐origin populations originate from the Nooksack, Stillaguamish, and Snohomish rivers, which have recent average annual out‐migrations of 0.28 million (2005–2015), 0.19 million (2003–2020), and 0.58 million (2002–2020) subyearlings, respectively, compared to the average annual out‐migration of 3.89 million subyearlings from the Skagit River (1994–2020; Hall et al. 2018).

Potential applications

Hatchery management

Our findings indicate considerable differences in habitat use patterns between hatchery‐ and natural‐origin stocks, suggesting that the use of hatchery‐origin fish as information analogs for natural‐origin fish may not be justified, at least during the juvenile estuarine portion of the Chinook Salmon life cycle. This would be especially true if significant mortality or reduced survival potential in later life stages is initiated within the estuarine life stage. For example, studies of survival (Magnusson and Hilborn 2003; Ruff et al. 2017), movement patterns (Adlerstein et al. 2007; Chamberlin et al. 2011), and growth (Duffy and Beauchamp 2011) have been based on tagged hatchery‐origin fish, which are characterized by large releases of similarly sized—and relatively large—individuals within short periods of time. These studies may have limited relevance for (1) natural‐origin stocks that utilize multiple habitats for different periods or (2) populations that are dominated by yearling life histories, thereby reducing their co‐occurrence with pulsed environmental impacts. Multiple studies have estimated survival through rivers, estuaries, and marine habitats based on the recovery of tagged hatchery‐origin individuals (Baker et al. 1995; Magnusson and Hilborn 2003; Holt et al. 2009; Chasco et al. 2021). Differences in size distributions as well as habitat use and duration between natural‐ and hatchery‐origin fish reduce the relevance of these findings and may result in biased estimates if applied to natural‐origin populations. For example, the conclusions of Duffy and Beauchamp (2011) regarding high survival for individuals surpassing a length of over 120 mm in neritic habitat by July would apply only to the outliers of natural‐ and hatchery‐origin Skagit River fish in 2002. In contrast to the inference of low survival for the majority of Skagit River out‐migrants is recent work by Campbell et al. (2023), who used otolith chemistry methods and found that a high proportion of returning adult Chinook Salmon consisted of individuals that out‐migrated as fry (i.e., smaller fish), a life history type that was observed in this study. Finally, Ruggerone and Goetz (2004) observed lower survival of coded‐wire‐tagged Chinook Salmon every other year and argued that biennial out‐migrations of Pink Salmon O. gorbuscha resulted in competition and lower recruitment. However, this pattern was absent from annual estimates of natural‐origin productivity (recruits per spawner) for Skagit River Chinook Salmon (Greene et al. 2005).

Although we found differences in the timing and abundance of hatchery‐ and natural‐origin juvenile Chinook Salmon cohorts, there was not a complete habitat compartmentalization between the two cohorts during their estuarine migration. To the extent that competitive interactions depend upon encounter rates (Chapman 1966; Fausch 1984), differential use of various estuarine habitats by natural‐ and hatchery‐origin fish suggests an increasing potential for ecological interactions as fish move downstream and offshore. Neritic habitats are the most likely area for competitive interactions, given that this habitat type has the greatest overlap of natural‐ and hatchery‐origin densities for the longest period (Rice et al. 2011). However, interactions between natural‐ and hatchery‐origin cohorts are not limited to these habitats and may include fish from other basins or regions (Hayes et al. 2019). Other hatchery–natural interactions that depend upon encounter rates (e.g., disease transmission; Rhodes et al. 2006) are also most likely to occur in neritic habitats.

The patterns that we highlight herein emphasize the need for field studies and monitoring to understand hatchery–natural interactions in the estuary, as many uncertainties remain in addressing the outcome of these interactions for both natural‐ and hatchery‐origin stocks (Weber and Fausch 2003; Pearsons 2008; Satterthwaite and Carlson 2015; Greene et al. 2021).

Recovering natural‐origin stocks

One implication of these results is that further declines in the extent, quality, and connectivity of estuarine habitats will likely have adverse effects on population survival and viability attributes (Magnusson and Hilborn 2003; Greene et al. 2010; Hodgson et al. 2020). Conversely, increasing the extent, quality, and connectivity of estuarine habitats may have considerable positive influences on these population attributes (David et al. 2014; Davis et al. 2018; Woo et al. 2018). This is especially true for natural‐origin populations that exhibit greater life history diversity and rely more heavily on estuarine environments (Healey 1982; Levings et al. 1986; Sawyer et al. 2023).

The portfolio concept is well documented in several species and populations (Greene et al. 2010; Moore et al. 2010; Schindler et al. 2010; Carlson and Satterthwaite 2011; Siple and Francis 2016) and highlights the importance of life history diversity for maintaining population productivity in the face of environmental variation. We can apply the same idea to spreading risk across multiple habitats. The idea that a diverse habitat mosaic provides increased opportunities for individuals within a population (Moore et al. 2015; Davis et al. 2019; Greene et al. 2021) points to the importance of building “habitat portfolios” into restoration plans. Such plans could allow for the expression of different life history types, thereby buffering populations from environmental variation (Greene et al. 2010).

Monitoring and assessment

More practically, our findings suggest several implications for improving the monitoring of status and trends. Our results focused on using multiple gear types to sample across habitats and months. Broad spatiotemporal coverage was critical for revealing both the extended habitat use by natural‐origin individuals and the contracted habitat use by hatchery‐origin fish. Imagine how our conclusions would have differed if sampling had been limited to the peak out‐migration period—a common practice of many monitoring programs. This effort comes at significant cost but offers much better precision to identify population status and trends and priorities for habitat‐based recovery planning. The results also point to the need for good marking techniques for hatchery‐origin fish. A variety of techniques exists, but not all result in markings that can be readily identified during surveys of juvenile life stages. Techniques that allow for rapid identification of hatchery‐origin individuals in the field, including emerging inexpensive genetic techniques that require small amounts of tissue, will allow for better effectiveness monitoring of hatchery programs and natural‐origin populations during juvenile life stages.

ACKNOWLEDGMENTS

This research was funded in part by the Pacific Salmon Commission and the U.S. Environmental Protection Agency under assistance agreement PA‐01J27601 through the Northwest Indian Fisheries Commission. The contents of this article do not necessarily reflect the views and policies of the U.S. Environmental Protection Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use. We thank Martin Liermann for his thoughtful input on initial versions of the manuscript. Lastly, we thank the two anonymous reviewers for their insightful comments and suggestions.

CONFLICT OF INTEREST STATEMENT

The authors declare that there are no competing interests.

DATA AVAILABILITY STATEMENT

Data used in this study were shared by the Tribes, and they exercise sovereign authority over it. Requests for data not supplied in tables, figures, or supplemental text may be made by formal request to the coauthors.

ETHICS STATEMENT

No ethical guidelines were applicable to this study.

REFERENCES

APPENDIX A

SUMMARY OF JUVENILE HATCHERY CHINOOK SALMON ABUNDANCES