Multiple spawning run contingents and population consequences in migratory striped bass Morone saxatilis.

Multiple spawning run contingents within the same population can experience varying demographic fates that stabilize populations through the portfolio effect. Multiple spawning run contingents (aka run timing groups) are reported here for the first time for striped bass, an economically important coastal species, which is well known for plastic estuarine and shelf migration behaviors. Adult Hudson River Estuary striped bass (n = 66) were tagged and tracked with acoustic transmitters from two known spawning reaches separated by 90 km. Biotelemetry recaptures for two years demonstrated that each river reach was associated with separate contingents. Time series of individual spawning phenologies were examined via nonparametric dynamic time warping and revealed two dominant time series centroids, each associated with a separate spawning reach. The lower spawning reach contingent occurred earlier than the higher reach contingent in 2017 but not in 2018. The majority (89%) of returning adults in 2018 showed the same contingent behaviors exhibited in 2017. Spawning contingents may have been cued differently by temperatures, where warming lagged 1-week at the higher reach in comparison to the lower reach. The two contingents exhibited similar Atlantic shelf migration patterns with strong summer fidelity to Massachusetts Bay and winter migrations to the southern US Mid-Atlantic Bight. Still, in 2017, differing times of departure into nearby shelf waters likely caused the early lower reach contingent to experience substantially higher mortality than the later upper reach contingent. Anecdotal evidence suggests that higher fishing effort is exerted on the early-departing individuals as they first enter shelf fisheries. Thus, as in salmon, multiple spawning units can lead to differential demographic outcomes, potentially stabilizing overall population dynamics.

Introduction

Spawning migrations are periods of heightened vulnerability, the outcome of which drives population dynamics [1-4]. Extreme gonad provisioning and swimming expenditures cause death in some fishes. Further, the restricted and predictable migration routes of spawners lead to increased exposure to environmental degradation and catastrophe, and, in some species, severe vulnerability to predation and fishing exploitation. A key buffer against these sources of mortality is multiplicity in spawning runs [5]. Discrete spawning runs and units within runs (referenced here as run contingents and in the salmon literature as run timing groups [1]) with differential timing, routes, and endpoints will each experience different mortality regimes that are dynamic across generations. Through the stabilizing feature described initially as response diversity [6,7] and then the portfolio effect [7,8], multiple demographic outcomes among spawning runs can stabilize population dynamics.

The best studied spawning runs, those of salmons, shads, and river herrings, are overt. Such fishes push up from coastal waters into shallow and narrow confines of non-tidal freshwater ecosystems, where their abundance and schooling behaviors are often on full display [9-11]. Indeed, these oft-depicted spawning runs represent the quintessential fish migration in public understanding [12-15]. The spawning migrations that occur in coastal waters are more difficult to observe but here, too, multiple spawning run behaviors can occur in the same population. For instance, both spring and fall spawning occur in populations of Atlantic herring [16,17], Atlantic cod [18], and Atlantic sturgeon [19]. Still, more-nuanced diversity in spawning phenologies, such as those that occur within the same run, remain undescribed for coastal species.

This study documents multiple spawning-run contingents for striped bass Morone saxatilis, a ubiquitous predator in NW Atlantic shelf waters, which supports important commercial and recreational fisheries. Striped bass are moderately long-lived, late maturing, and on average spawn each year of their adult life, although skipped spawning does occur [20]. Spawning migrations, some exceeding 1000 km, occur from shelf to estuarine waters each spring [20,21]. Fisheries target and intercept spawners as they arrive and depart spawning reaches [22,23]. Typically, spawning is concentrated just above the salt front [24,25], yet in larger estuaries multiple areas of concentrated spawning occur [26]. We hypothesized that in one such estuary, the Hudson River Estuary, at least two spring spawning run contingents occur, each associated with known centers of egg production (Fig 1).

Fig 1

Map of the Hudson River, NY, USA and distribution of striped bass eggs.

Data digitized from the Hudson River Long River Ichthyoplankton Survey, 1976–2012 (references to all annual reports containing the data are listed in [27]). Y-axis shows river km (R-Km) and latitude. Circles on the map display deployed telemetry receivers, and the location of New York City (NYC), the Tappan Zee Bridge (TZB), and the Poughkeepsie USGS water monitoring station (USGS) are shown. Points show yearly values of egg density and box plots represent interannual variation. Highlighted areas, upper reach (red) and lower reach (blue) represent the regions of spawner abundance monitored by NY state scientists.

To be of ecological consequence, each spawning run contingent should exhibit characteristic migration behaviors, repeat these behaviors across years, and encounter varying mortality regimes. Run contingents are defined as a cycle of directed up-estuary and down-estuary migration behaviors occurring within the Hudson River Estuary, and are classified by both phenology (timing) and spawning reach (up-estuary extent). We leave alone the question of evolutionary and conservation consequences associated with genetic differentiation between contingents [28], and rather examine ecological consequences of their behaviors following their departure. [29,30]. Through biotelemetry and time series classification using dynamic time warping, we asked; (1) Are characteristic spatiotemporal migration behaviors repeated within successive spring spawning runs of the tagged individuals? (2) Do the same individuals undertake the same contingent behavior between years? (3) Do spawning run contingents undertake the same migration behaviors during non-spawning periods? and, (4) Do mortality rates vary between spawning run contingents? Within the limits of two years of spawning phenologies, we also evaluated the influence of temperature, which has been identified as a dominant spawning cue for striped bass [31-33].

Materials and methods

Study species and system

Natal estuaries for striped bass occur from the Gulf of Mexico through the St. Lawrence Estuary. Exhibiting partial migration throughout their range [20,34,35], migratory (oceanic) striped bass are most abundant in the US Mid-Atlantic Bight [36], but are also common in the Gulf of Maine and Canadian waters [37,38]. The Hudson River, Chesapeake Bay, and Delaware Estuary together support the US Atlantic shelf stock, which is assessed and managed by the Atlantic States Fisheries Management Commission [36]. In 2019, overfishing was assessed for the stock, which currently supports the US’s single most valuable marine recreational fishery [36]. Important commercial fisheries also occur, principally in the Chesapeake Bay. The Hudson River Estuary supports its own population, which exhibits partial migration [21,39], partitioning into Upper Estuary (resident), Lower Estuary, and Oceanic (migratory) contingents that persist over years and life-times. Past research shows that partial migration shapes overall striped bass population responses to pollution, fishing, storms and management [20,40,41]. In this study, we target the Oceanic contingent through selection of larger individuals during spring [42].

The Hudson River Estuary—a fjord valley- is long (243 km), linear, and deep in comparison to other Mid-Atlantic Bight estuaries, lending itself well to biotelemetry coverage (Fig 1). Each spring, spawning fish transit through New York Harbor (river km 0) into the Peekskill-Storm King highlands region (river km 70–90), where upon salinity rapidly diminishes [43]. Snowmelt and freshets cause large variation in the salt front position during this period, which can extend nearly to New York Harbor in extremely wet years. Striped bass eggs, larvae, and juveniles have been well monitored in the Hudson River Estuary, the result of an energy utility’s agreement to conduct Hudson River longitudinal surveys beginning in 1974 [27,33]. Key spawning reaches are indicated from ichthyoplankton collections, with modes of egg density at about 90 and 120 km (Fig 1). Decades of monitoring spawning adults confirm the traditional occurrence of spawners in the 80–100 km and 160–200 km reaches [44]. Egg densities are likely displaced downriver from these spawning centers owing to down-estuary advection of eggs. Proximity to the salt front retains eggs spawned at the lower reach [25].

Capture and tagging

During April and May 2016, 100 Hudson River striped bass were captured within the two spawning reaches (Fig 1) and received coded acoustic transmitters (Lotek Wireless, Inc.; model MM-MR-16-50; 8 cm, 35 g, 2.5 year expected battery life; continuous 60s transmission delay at 69 kHz, 7s delay from April-June at 76 kHz). All fish were > 68 cm total length (TL) and assumed to be members of the migratory Oceanic contingent [21,40]. As the two reaches differ in extent and bathymetry, they required that sampling take place during non-overlapping times and with different gear. Within the lower reach, 50 fish were captured in shallow water through electroshocking and tagged April 20–26. As waters warmed, fish moved into deeper waters and became inaccessible via electrofishing after April 27. Another 50 fish were captured in deeper water in the upper reach with a 152 m haul seine deployed by boat and tagged May 5–19.

Surgeries in both regions were conducted onboard a small vessel in a portable electro-immobilization unit [45] following procedures under an approved protocol by the University of Maryland Center for Environmental Science IACUC (#F-CBL-16-05). Fish selected for tagging were immediately transferred to a flow-through tank or in-river live well. Holding time both pre- and post- surgery was adequate to both ensure healthy condition (<5 minutes each) and to minimize holding stress on the fish. In preparation for surgery, fish were inclined on a surgery sling with a tank containing ambient freshwater so that their head and gills were immersed and abdomen exposed for surgery. Anesthesia using electronarcosis was immediate (<30 seconds); introduced voltage was adjusted to induce anesthesia but generally ranged between 15 and 20 volts. Sex was determined either by expressed gametes or confirmed during surgery. Length and weight measures were taken, and then sterilized transmitters were implanted through a 1–2 cm incision slightly lateral to the linea alba and mid-distance between the vent and pectoral fins. Incisions were closed with a series of simple surgical knots and sterilized. Fish were held in pre- and post-surgery recovery pens.

Biotelemetry

During the spring 2016–2018 study period, 22–62 telemetry receivers per year (Vemco VR2W ©) were deployed throughout the Hudson River Estuary between river km 43 and 245 (Fig 1). The two study spawning reaches were well bracketed by both down- and up-estuary receivers. At deployment sites above river km 100, the detection range of receivers (600 m) allowed bank-to-bank coverage by single receivers. This was not the case for those deployed below river km 100 where the estuary broadens (Haverstraw Bay). Receivers were hung on buoys, where they logged detections during deployment periods April-October each year. Receivers were checked every 2–3 months and detection data downloaded. Outside the Hudson River Estuary, deployments of acoustic telemetry arrays occurred throughout the US NW Atlantic shelf waters, supporting evaluations of broad scale coastal migrations. The Atlantic Cooperative Telemetry Network [20] is an online portal that facilitated the return of transmitter detections from colleagues within the Network willing to share data. Consistent array deployments occurred 2016–2018 and supported depictions of seasonal shelf migrations, ordered by latitude. For depiction purposes receivers were combined for Maine (ME; estuary and shelf waters), Massachusetts (MA; principally Massachusetts and Cape Cod Bays), Long Island Sound (LIS), NY Bight (NYB; combined New York and New Jersey shelf waters), Delaware (DE; estuary and shelf waters), Maryland shelf (MD), Virginia shelf (VA), and Chesapeake Bay (CH; estuary).

Spawning run contingents

The design of the biotelemetry study sought to sample equivalent numbers, sex ratio, and size range between the two hypothesized spawning reach contingents. In addition to being a spawning area, the lower reach represents an area of staging for up-estuary-spawning fish. Due to temperature constraints mentioned above, the lower reach was sampled first. Therefore, an equivalent number of spawners in each reach were targeted in year 2016, recognizing that without a priori information on how fish visited each area it was likely that lower reach sampling may have captured spawners destined for the upper reach. To correct for this bias, we classified spawners during the subsequent spring spawning run of 2017. Those individuals were then compared with their 2016 capture location and subsequent 2018 spawning run behaviors.

Before classifying 2017 spawning run contingents, the mean daily river kilometer of each fish’s observed detections was calculated. Gaps in the daily time series were interpolated with a 4-day exponential-weighted moving average using the imputeTS package in R [46]. Per-fish, this resulted in 23.9 ± 9.9 and 26.0 ± 6.1 measured observations (mean ± standard deviation) from river entry to exit in 2017 and 2018, respectively, and 14.0 ± 7.8 and 8.4 ± 8.5 imputed values. The ends of each series were catenated to equal length with the river kilometer of the New Jersey-New York border (river km 35) to maintain information on date and penalize alignment of run phenologies that occurred far apart in time. This point was south of the lowest receiver (river km 40) near Tappan Zee Bridge (river km 43), the designated start and end point for each spawning phenology.

Individual spawning phenologies were categorized by clustering around median centroids (k-medoids [47]) utilizing dynamic time warping as a time series dissimilarity measure. Dynamic time warping, a machine learning algorithm most often applied in speech recognition classification, was chosen due to its suitability in matching phenomena that are offset in time and magnitude [48,49]. Starting from the beginning of the time series, dynamic time warping iteratively finds which points are matched with the least cost [50]. This allows matching of multiple-to-one or one-to-multiple points (Fig 2), and reports the cost of “warping” the two time series to be alike in this manner. Compared to Euclidean matching, which necessitates one-to-one matching in time and would calculate dissimilarity between spawning run contingents based on the daily river kilometer distance between each fish, dynamic time warping measures how much an entire spawning phenology trajectory must be stretched or compressed in time to match that of another fish. As such, spawning run phenologies that are alike in shape, extent, and destination, but offset in time are clustered together [51,52]; Fig 2). The iterative clustering procedure, utilizing dynamic time warping and performed with the dtw [53] and dtwclust [54] packages in R, solved for maximum within-cluster similarity and minimum between-cluster similarity. A synthetic median centroid was extracted from each of two clusters to represent lower- and upper-reach spawning.

Fig 2

Visualization of matching two spawning phenologies via standard Euclidian (top panel) and dynamic time warping (bottom panel) methods.

Single fish spawning phenologies are depicted as light brown (left axis units) and dark brown (right axis units). Time series are vertically offset for ease of visualization. Gray lines display daily positions matched using the respective algorithms. The arrival of each fish (subscript 1, 2) in the river (A) or spawning reach (B), and subsequent exits from the spawning ground (C) and river (D) are shown. Note that dynamic time warping allows these spawning phenologies, offset in time, to be matched, while the Euclidean method matches only those points that co-occur in time.

To investigate consistency of run contingent categorization for individual fish across years, the tracks of returning fish in 2018 were clustered onto the 2017 centroids and the cross-classification between the two years compared. Independence between cross-classified frequencies between years was tested using a Chi-square test.

Spawning run phenology

Temperature was examined as a possible driver for the phenology of 2017-defined run contingents, under the premise that the 2017 classifications best represented the two run contingents. Flow was also examined for its influence on temperature. All classified individuals were included in 2017, and only those individuals correctly classified included for 2018. Daily mean water temperature (°C) and discharge (m3s-1) records encompassing the spawning season were extracted from USGS Poughkeepsie water monitoring station (river km 122; Fig 1) and compared to the cumulative presence of fish in each run. Runs entered the spawning reaches in a series of pulses (see Results) not well characterized by weighted-means that are commonly used to describe spawning phenology [55,56]. Rather, time of entry and exit were characterized by median dates and experienced temperature by the 50th percentile occurrence for each run contingent.

Mortality between spawning run contingents

Attrition of tagged fish was followed over the two years after release, with the intent focused on evaluating differential mortality between spawning run contingents following their identification in spring 2017. Mortality was analyzed for the period June 2017-December 2018 to limit any bias associated with the tagging procedure. The last detection date for each individual separated their period at large (alive) and their assumed loss from the sample (death). The number of extant individuals was summed for each date. Sums were log-transformed and regressed against days-at-large (date) to estimate daily instantaneous loss rates (Z). Post-spawn survival was modeled for the clusters identified in 2017 separately for each year using Kaplan-Meier estimation. Differences in survival between the clusters was tested using the Peto & Peto modification of the Gehan-Wilcoxon test. Survival analyses were conducted using the survival package in R [57].

Results

Biotelemetry returns

Tagged fish in the upper reach (n = 50; 71–104 cm, mean 88.6 cm TL; 3.5–17 kg, mean 8.6 kg) were significantly larger than those in the lower reach, although sizes broadly overlapped (n = 50; 68–99 cm, mean 83.4 cm TL; 2.8–12.1 kg, mean 7.0 kg) (Fig 3; Welch’s t-test: p = 0.002, n = 100; p = 0.003, n = 97, respectively). Females were 11.0 cm and 3.2 kg larger than males at both reaches (Welch’s t-test, combined reaches: p<0.001, n = 100; p<0.001, n = 97, respectively). In both reaches, more females were captured and tagged than males (lower reach: 28 female, 22 male; upper reach: 36 female, 14 male), and sex ratio did not vary between reaches (Chi-squared: p = 0.10, n = 100).

Fig 3

Distributions (box whisker plots) of total length (cm; left panel) and weight (kg; right panel) of Hudson River striped bass receiving transmitters by tagging region and sex.

Horizontal dashed lines represent the Hudson River no-take slot limit between 71.1 and 101.6 cm Total Length.

Telemetry returns in the two months following tagging indicated that all but one fish—which stopped being logged shortly after surgery—migrated downriver to regions below the Tappan Zee Bridge (river km 43); seven others were not detected after leaving the Hudson River that year. Sixty-six fish completed the spawning run the following spring (2017), while 40 fish completed it in 2018. One tagged fish from each region skipped spawning in 2017 (i.e., did not return to Hudson River spawning reaches), but did return during spring 2018. Four other fish, two tagged in each Hudson spawning region, either strayed or were originally strays from the Chesapeake and Delaware estuaries and did not return in subsequent years.

In 2017, time series clustering of returning tagged striped bass produced two distinct median behaviors (Fig 4). Centroid 1 was the most frequent behavior (45/66) and is clearly oriented towards the upper reach, with individual phenologies showing repeated up-estuary excursions but tending not to retreat to regions below river km 150. The median period of the first classified run contingent, as defined by centroid 1, is April 22 –May 30, with individual phenologies ranging between April 4 and June 24 (Table 1). Centroid 2 represented the minority of fish (21/66 individuals), exhibiting a modal behavior centered on the lower reach. Similar to Centroid 1, individual excursions occur up-estuary, with the retreat migrations delimited at c. river km 95. Centroid 2 represented an earlier run contingent behavior, with the median period April 18-May 21. Sex ratios were again skewed towards females, more so for the upper reach contingent (lower reach: 11 female; 10 male; upper reach: 29 female, 16 male), but did not signicantly differ (Chi-squared: p = 0.51; n = 66).

Fig 4

Dynamic time warping centroid identification of Hudson River spawning run contingents.

Centroids are fitted to 2017 spawning run behaviors in both 2017 and 2018 (bold black lines; note centroid phenology is duplicated for both years). Individual phenologies are shown as gray and red lines, the latter indicating mis-classified individuals. Colored rectangles identify the river kilometers of the upper (red) and lower (blue) reaches.

Table 1

First, last, and median dates of entry above, and exit below, Tappan Zee Bridge (river km 43) for the 2017 and 2018 spawning seasons.

Year	Centroid	Reach	First Entry	50% Incidence	Last Entry	First Exit	50% Exit	Last Exit
2017	1	Upper	April 4	April 22	May 3	May 19	May 30	June 24
	2	Lower	April 12	April 18	April 28	May 2	May 21	June 4
2018	1	Upper	April 5	April 27	May 11	May 11	May 28	June 5
	2	Lower	April 5	April 18	May 1	May 18	May 27	June 26

The 2017 and 2018 clustering procedures provided similar individual classications with all but 4 of the 38 individuals that returned in both years (two of the 40 fish returning in 2018 had not returned in 2017) correctly classified (11% misclassification; Chi-square test: P<0.001; Figs 4 and 5). All misclassified individuals in 2018 occurred for the lower reach contingent (4 of 12 individuals; Fig 5C). In 2018, upper reach fish were again more frequent (n = 26) in comparison to lower reach individuals (n = 12). Median periods were April 20-May 27 for lower reach and April 27-May 28 for the upper reach.

Fig 5

Cross-classification of regions at tagging and as determined by 2017 and 2018 clustering of spawning run contingents.

(A) 2017 spawning run centroids v. 2016 tagging locations; (B) 2018 spawning run centroids v. 2016 tagging locations; (C) 2017 spawning run centroids v. 2018 spawning run centroids.

Centroids classified for 2017 and 2018 identified upper and lower reach fish, whose membership was tested against individuals tagged in upper and lower reach samples in 2016. Chi-square tests showed that tagging location influenced identified centroids for both 2017 and 2018 (P<0.01). Thus, tagging locations in 2016 were reasonably selective for upper and lower reach contingents (Fig 5). Most fish (89%) tagged in the upper reach in 2016 were classified (in 2017) as Centroid 1, while only 55% of fish tagged in the lower reach in 2016 were classified as Centroid 2 in keeping with the prediction that some fish sampled in the lower reach were intercepted enroute to upper reach spawning. In 2018, 78% (31/40) of classified 2018 fish aligned with 2016 capture and release locations, and were similarly better classified for upper reach (88%) than lower reach (63%) samples.

Surface water temperature in both years increased from 4°C to 23°C as fish moved up-estuary during the spawning season (Fig 6). In 2017, temperature increased more rapidly during April than in 2018, approaching 13 °C by April 28, then leveling off during early May in association with increased discharge. In 2018, temperatures did not approach 13 °C until May 13 and showed steady increases throughout spring.

Fig 6

Environmental conditions encountered in 2017 and 2018 by lower (blue) and upper (red) reach contingents.

Run contingents were classified according to 2017 returns. Top panels are cumulative frequency distributions on each contingent (lower: blue; upper: red), initiated as they pass Tappan Zee Bridge (river km 43). Monitored water temperature and discharge are daily means derived from Poughkeepsie USGS station data (river km 122).

The upper reach contingent lagged behind the lower reach contingent, measured as 50% incidence, more so in 2018 than in 2017 (Fig 6). Different responses by contingents to temperature may be indicated for the upper reach contingent, which shifted its timing earlier during the warmer year (April 22, 2017) in comparison to the cooler year (April 27, 2018). In contrast, the lower reach contingent exhibited the same date of 50% incidence between years (April 18). With the exception of the lower reach contingent in 2017, a few individuals entered the river in early April when mid-river water temperatures were 5–6 °C. In 2017, mean temperatures at 50% incidence were 10.1 and 11.8 °C, respectively for the lower and upper reach contingents. In 2018, 50% incidence occurred at substantially cooler mean temperatures of 6.1 (lower reach contingent) and 8.0 °C (upper reach contingent).

Coastal migrations

Detections in the three years following tagging (2016–2018) indicated that the two classified spawning run contingents undertook similar shelf migrations from southern Maine to southern Virginia (Fig 7). In 2016, we reclassified individuals to spawning run contingent on the basis of the 2017 centroid analysis rather than tagging location and observed that the lower reach contingent departed the Hudson River estuary 6 days earlier than the upper reach contingent, similar to what was observed in 2017 when the lower reach contingent departed 9 days earlier on average. The opposite occurred in 2018, however, with the upper reach contingent departing 2 days earlier than the lower reach contingent on average. A single lower reach fish entered waters off Maine in the summers of 2016 and 2017, otherwise the pattern of shelf migrations in late summer, fall, winter, and early spring were quite similar between spawning reach contingents (Fig 7). Fish that left the Hudson River moved along southern Long Island into Massachusetts waters from May through June with an average transit rate of 10 km d-1, where they remained until late September. After leaving coastal Massachusetts, tagged striped bass moved south at an average rate of 6 km d-1, entering the coastal waters of Virginia off the mouth of the Chesapeake Bay in early-to-mid January. Striped bass remained in this area until mid-March or early April, when they rapidly moved back to the Hudson River at an average rate of 11 km d-1. This pattern repeated each year with relatively little geographic or temporal variation.

Fig 7

Atlantic shelf water telemetry detections of Hudson River striped bass spawning reach contingents.

Top: Detections through time. Upper and Lower reach contingents are defined by clustering of 2017 spawning phenologies. The “Other” category are those not included in the 2017 classification owing to tag loss, skipped spawning, or straying. Detections are ordered by latitude of telemetry receiver array (Maine–Coastal Virginia/Chesapeake Bay). ME = Maine estuary and shelf waters, MA = Massachusetts and Cape Cod Bays, LIS = Long Island Sound, NYB (NY Bight) = New York and New Jersey shelf waters, Hudson = Hudson River Estuary; DE = Delaware River and shelf waters, MD = Maryland, VA = Virginia, CH = Chesapeake Bay. Bottom: Distribution of each year’s detections with regional identification of receiver arrays (left bottom panel).

The lower reach contingent experienced a high level of loss during the May-June period in 2017 and 2018 (Z = 4.9 10−3 d-1; Z = 7.4 10−3 d-1), suggesting periods of intense vulnerability (Fig 8). In comparison, the upper reach contingent experienced lower mortality during this period in 2017 (Z = 1.0 10−3 d-1), but similar losses in 2018 (Z = 4.5 10−3 d-1). Annual post-run Kaplan-Meier survival estimates differed between spawning run contingents only for 2017 (p = 0.02; 2018: p = 0.7). In 2017, annual post-run survival was estimated at 57.1 ± 10.8% and 82.2 ± 5.7%, for the lower and upper reach contingents, respectively. In 2018, annual post-run survival was similar between contingents (lower: 54.5 ± 15.0%, upper: 51.6 ± 9.0%). Note that the median exit date for the lower reach contingent was 9 days earlier than the upper reach contingent in 2017, and that the two contingents exhibited similar exit dates in 2018 (Table 1).

Fig 8

Kaplan-Meier curves of post-run fish by 2017 spawning run contingents classification in 2017 and 2018.

Lines are coded red for the classified upper reach and blue for the lower reach contingents. 95% confidence intervals are shown as envelopes.

Discussion

The three-year biotelemetry study uncovered discrete contingents within spawning runs, a behavior not yet documented for striped bass. Seasonal and sub-seasonal spawning runs are well described where they are overt, such as salmon streaming past the viewing window of a fish lift, or through seasonal fisheries that target spring, fall, or winter spawning aggregations of herring and cod. Here, biotelemetry coupled with the dynamic time warping cluster analysis exposed more-cryptic spawning run contingents that broadly overlapped within the same spring season, but used different reaches of the Hudson River estuary, and exhibited characteristic phenologies. We discovered that individuals largely participate in the same spawning run contingent year after year and that contingent membership had ecological consequences. In one of the two study years, the early run contingent was exposed to greater mortality than the later contingent -associated with early summer shelf fisheries. These differential demographic fates between run contingents can alter population outcomes, particularly if selective mortality occurs for one or the other contingent.

Mechanisms contributing to individual fidelity to either spawning reach contingent were not evaluated but could relate to learned behaviors and tradition. Given the level of infidelity between run contingents (11%), particularly for the lower reach contingent (25%), and high dispersal and mixing of embryos and larvae [25], persistence in contingent behavior is unlikely related to genetic lineage. For first maturing individuals, associative schooling behaviors with larger and older individuals within the Hudson River estuary could promote initial adoption of either spawning run behavior [29,58,59]. Thereafter spawning reach fidelity may relate to key environmental differences in conductivity, tidal flow, bathymetry and channel dimensions that allow piloting and then navigation to particular spawning reach features. In past telemetry studies, striped bass have shown remarkable fidelity to specific locations summer after summer, indicating precise navigation is within their behavioral repertoire [21,42,60]. Such precise homing behaviors are of course well known for salmon species [61].

Spawning run classifications

The identification of modal migration behaviors within spawning runs and individual fidelity across years required a more nuanced classification approach owing to the regional and temporal overlap between the contingents. We initially considered a more traditional time series hierarchical clustering analysis [62], but found that the approach was too sensitive to unequal phenology durations and small misalignments between daily positions (Fig 2). Dynamic time warping evaluates the degree with which time series must be warped or bent to make them equivalent, and does not have the requirement of equal-length time series [48]. Developed as a machine learning tool in speech recognition applications, it has received scant attention by movement ecologists [51,52]. In its application to inverted U-shaped spawning run behavior, the algorithm performed well, summarizing time series that ranged between 16 and 70 days in duration, and time series that deviated from symmetry and often showed multiple up-estuary excursions (Figs 2 and 4). The approach captured and characterized the two hypothesized reach-specific behaviors despite high variance in individual movements.

Implementing the dynamic time warping approach required complete and overlapping time series for each fish entering the Hudson River. The high longitudinal coverage of the Hudson River Estuary by fixed telemetry receivers aided here, but times series catenation to a start and end point, the Tappan Zee Bridge (river km 40), and interpolation of missing daily records were still required of the approach. Our analysis was informed by the expectation of only two dominant behaviors, and a more refined cluster analysis (k>2), would likely support minority behaviors within these clusters for each year [21]. Still without expectation for additional spawning group behavior, we curtailed our comparison to upper and lower reach groups. This prediction was upheld by the membership persistence within identified groups.

Environmental drivers

Striped bass spawning runs occurred during springtime warming, which ranged 4°C to 23°C across individual spawning phenologies. A slower rate of warming in 2018, was associated with a 5 day lag of the upper reach contingent, but the timing of the lower reach contingent was similar between years. This suggests that spawning run phenologies are influenced, but not tightly cued by temperature. In both years, a minority (<10%) entered early as temperatures ranged 4–8 °C. Entry during winter regime temperatures aligns with historical Hudson River fisheries which targeted fish under ice-cover [22]. Spawning migrations under ice cover have been noted for Canadian systems as well [37]. In both years, the number of fish participating in spawning runs increased rapidly between 10 and 12 °C. Then, most fish departed by the beginning of June as temperatures approached 20 °C. The lower reach contingent likely spawned at lower temperatures than the upper reach contingent with the difference larger in 2018. A recent historical analysis of spawning temperatures in the Hudson River Estuary recorded a range of 11.9 to 23.0 °C [33]. In more intensively studied Chesapeake Bay tributaries, spawning is associated with several-day surges in water temperature within this range [31,63]. In other literature, 12–20°C bracket viable temperatures for embryo and larval growth and development in the Hudson River [32,64] and elsewhere [31,65-67] and in general the most favorable temperatures for larval survival occur between 15 and 19 °C [63], a period of maximum presence by spawning runs in both study years.

Despite general alignment between spawning runs and temperatures conducive for early survival and growth, the 6–9 day difference in contingent timing observed in 2017–2018 could result in differential outcomes on larval growth and mortality. Striped bass are capital spawners, with time of spawning sometimes mismatched to thermal or flow conditions favorable to larval survival [63]. Secor [68] suggested that a protracted spawning season maintained by a size-specific spawning phenology (larger fish spawning earlier [26]) could hedge against this mismatch. Further, warming in the Hudson River Estuary during the recent period (1976–2012) has resulted in a 7-d shift towards earlier spawning [33], increasing the opportunity for spawning-larval survival mismatches. Here, we observed differing spawning run phenologies that could mitigate against “mistimed” spawning. Because larval vital rates are quite sensitive to temperature, the 2–4 °C difference in temperatures observed between spawning reaches in 2017 could have resulted in an order of magnitude difference in larval and juvenile production [63,69,70].

Hudson River striped bass exhibited distinct spawning run contingents despite broadly overlapping coastal distributions. The term contingent applies here as population sub-components defined by persistent migration differences [71]. Contingent behaviors can shape overall population outcomes despite periods of high overlap in incidence. In 2017 the departing lower reach contingent entered shelf waters before the upper reach contingent and encountered recreational fishing that selected all sizes of the run (TL>71 cm). In contrast, the in-river fishery was restricted to fish both smaller, 41–71 cm TL, and larger >102 cm TL, protecting the bulk of spawners (Fig 3). We speculate that by departing the Hudson River and entering coastal fisheries 6–9 d earlier in 2016 and 2017, the lower reach contingent experienced greater fishing mortality than the upper reach contingent. Higher vulnerability would persist as this contingent successively entered New England state fisheries as it migrated northward. In 2018, when little difference occurred in median exit dates, mortality was similar between contingents suggesting similar exposure to these fisheries. Directed fishing in the river itself is largely focused in the upper reach [72], such that any latent mortality associated with catch and release of larger spawners would likely bias mortality opposite to observed pattern of contingent mortality. As further circumstantial evidence of higher selectivity on early departing striped bass, consider the “other” (undefined) category, which principally comprised individuals that did not survive until the following spring and therefore could not be included in the clustering procedure. In both 2016 and 2017, other category individuals departed to shelf fisheries at substantially earlier dates and may have been exposed to higher effort resulting in their loss (Fig 7). As Hudson River striped bass move into the mixed stock fishery, fishing mortality is predicted to predominate over “natural mortality” sources for > 7 year old cohorts [36].

During late summer, fall, winter, and spring—the two defined spawning run contingents showed similar north-south shelf migration patterns, matching those reported for migratory striped bass tagged in Southern New England shelf waters and the Potomac River [20,73]. These similar coastal distributions again suggest that differential survival in 2017 was likely related to distinct timing of Hudson River emigration.

The two spawning run contingents could serve to stabilize the overall Hudson River population owing to the portfolio effect. Here, contingents encounter varying mortality and reproductive success but jointly, their outcomes buffer the population under nonstationary mortality regimes. In instances where one contingent is substantially smaller in number, such as indicated here for the lower spawning run contingent, the portfolio effect is diminished, although resilience (persistence) is still enhanced [7,74]. Still, our sampling frame was curtailed to a single year’s tagging effort and egg densities suggest a more equitable spread between spawning reach contingents during the past several decades (Fig 1). Early or later run contingents will interact with climate and anthropogenic forces with outcomes that vary year to year. Early survival and recruitment are sensitive to spawning phenology when thermal and flow conditions are highly variable year to year. We observed that early departure likely caused higher fishing mortality in the lower reach contingent in 2017, but in a scenario of future climate change and warmer springs [33], reproduction and early survival could be favored in the lower reach owing to reducing the current risk of spawning in cold sub-lethal temperatures [31,63]. Facilitating the portfolio effect by promoting contingent conservation is a novel construct in fisheries conservation [5,8,74]. The periodic aggregation behaviors by post-spawning striped bass likely exposes them to a period of intense exploitation, which could select and diminish one or the other contingents. Size and season limitations could conserve “contingents for contingencies,” [29] favoring population stability in the face of future and uncertain exploitation and environmental regimes.

2 Sep 2020

PONE-D-20-18274

Multiple spawning run behavior and population consequences in migratory striped bass Morone saxatilis

PLOS ONE

Dear Dr. Secor,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

==============================

Dear Dave,

We now have two reviews of your manuscript. Both the reviewers and I see potential for a good paper that should be published in PLOS ONE. However, the reviewers have some concerns and criticisms (see below). These need to be addressed in detail befor we can accept the ms. As far as I can see the reviewers' requirements do not demand new analysis, but as ref 1 writes: "My principal criticisms rest on the presentation and interpretation of particular aspects of the study".

Best regards

Geir

==============================

Please submit your revised manuscript by Oct 17 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Geir Ottersen

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. In your Methods section, please provide additional information regarding the permits you obtained for the work. Please ensure you have included the full name of the authority that approved the field site access and, if no permits were required, a brief statement explaining why.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Review of PONE-D-20-18294: Secor et al. 2020 – Multiple spawning run behaviour and population consequences in migratory striped bass

The submitted manuscript presents evidence of distinct spawning runs of Hudson River Estuary striped bass. The authors use two years of biotelemetry data and a dynamic time warping (DTW) analysis to identify distinct migration behaviours associated with population contingents that spawn in the upper and lower portions of the estuary. Evidence for differences in spatial distribution occurred in both years, while temporal differences were more ambiguous (the lower run migrated earlier in 2017, but not in 2018). Additionally, the authors attempt to link differences in phenology with abiotic environmental drivers and present evidence of differential survival among contingents despite similar coastal distributions. Overall, I found the manuscript to be informative, well written (particularly the introduction), and timely given continued interest in the potential for life history diversity to stabilize exploited populations via portfolio effects. I believe the analytical techniques are appropriate given the research questions and commend the authors on exploring novel modeling approaches that have the potential to be relevant to the broader ecological literature. My principal criticisms rest on the presentation and interpretation of particular aspects of the study. If these can be addressed, I believe the manuscript would be suitable for publication. Below I describe my major criticisms, followed by a summary of minor issues.

Major

1) Spawning runs vs contingents: Throughout the manuscript (e.g. line 24, 30, 61-63), the authors move back and forth between using the terms spawning run, spawning run contingent, and contingent to describe the different migration phenologies they observed. Given the first author’s original description of contingents in the 1999 paper, I believe that term is most appropriate, but do not believe it should be used synonymously with spawning run. At least within Pacific salmon, spawning runs are commonly used to describe genetically isolated groups, indeed run timing is one dimension used to define ESUs in the US and CUs in Canada, while the authors note that the upper and lower reaches are likely to have substantial genetic exchange. That being said I recognize that the definition of life history terms can vary region to region and taxa to taxa. Thus, my main recommendation would be to clearly define early in the introduction how each term relates to the other. Note that this may impact the manuscript’s title.

Also in the introduction, it would be beneficial to clarify how spawning location, phenology, or both contribute to classifying contingents in this context since temporal or spatial segregation could be used to classify a run, but evidence for each is somewhat ambiguous here (e.g. straying between groups, downstream advection of eggs, similar migration phenologies in one year).

2) Dynamic time warping – As the authors note, DTW is not commonly encountered in ecological analyses. Indeed, I had to spend a decent amount of time reviewing the relevant literature to understand what was happening under the hood. The authors briefly introduce the technique in the methods and provide some additional information in the discussion, but I believe an additional short paragraph summarizing the primary references would be beneficial for readers who are curious about the technique.

On a related note, I initially struggled to interpret Figure 2, but if adapted, I believe it offers an opportunity to introduce and clarify DTW to readers. First, I would recommend showing the Euclidean plot in the top panel since readers are more likely to be familiar with that distance metric. Second, it would be beneficial to add text insets (e.g., a, b, and c at the beginning, middle, and end of the time series), which are referenced and interpreted in the caption since readers are not fully familiar with the study’s data until the end of the methods, but Fig. 2 is referenced partway through. Finally, I would suggest summarizing what the implications of the Euclidean vs. DTW distance measures are in this simple example (either in the caption or when Fig. 2 is referenced in the text). For completeness, it may be helpful to add a second panel, which shows two time series where Euclidean and DTW estimates provide relatively similar conclusions to contrast visually with the example where they clearly are different. As an aside, in my version the second figure does not have any red or blue, just two shades of brown (I’m using a Mac so perhaps this is a PDF conversion issue?).

3) Environmental linkages – I understand the authors interest in identifying drivers of interannual variability in run timing and agree that discharge and temperature are two plausible candidates for consideration. However, I was surprised that the authors describe strong linkages between temperature and phenology in the discussion given their results. First, since the observational unit in this part of the analysis is an entire run (sampling population level response to a shared driver), it is difficult to draw robust conclusions about the relationship between drivers and phenology with only two years of data. Second, only the upper run delayed in 2018 and the 50% migration point was remarkably stable between years despite a 3-4 C difference in temperature (depending on the run). There was also considerable variation within a run, which suggests that even if temperature does cue the run, it likely only explains a relatively small component of the total variability. In other words, evidence of temperature cues seems to be weak and ambiguous and I actually was under the impression that they would discuss the stability of run timing despite interannual differences in temperature

Similarly, it seemed odd to me that temperature at the mid-point of the run was used as a proxy for thermal conditions that may cue both river entry and egress. Since the data suggest considerable variation in the rate at which warming occurs, it seems as though temperatures at the beginning and end of the summer should be correlated with entry and egress, respectively. Finally, I didn’t understand why discharge data were presented in Fig. 6, but not incorporated into the analysis since they appear to show substantial variation between years as well. If I am misinterpreting the authors analyses (which is possible) I would recommend they clarify their hypotheses and methods, and perhaps introduce additional language that emphasizes the limitations of their approach given the data available.

4) Differential mortality – I found the survival data very interesting and agree with the authors that it is consistent with differential mortality among the contingents, which is clearly of management interest. That being said, it’s not clear to me how a moderately earlier ocean entry (~seven days) led to sustained higher rates of mortality over ~two months for the lower reach group. If differential mortality was driven solely by built up recreational harvest then why wouldn’t the difference in survival be restricted to May/June, rather than continue through July (based on Fig 7)? Is it possible these are delayed mortalities from catch-release fisheries, particularly if lower estuary fish are exposed to higher in river harvest effort? Otherwise there must be some difference in exposure to fisheries, or natural mortality mechanisms, through the summer to drive a consistently divergent pattern; however, that doesn’t seem to be the case based on the coastal detections that are presented. More generally, I would caution the authors against emphasizing harvest as the driver of differential mortality unless they can present data distinguishing natural mortality from fisheries mortality within the tag groups (e.g. a larger proportion of lower reach tags returned by harvesters) or it is established that natural mortality for these age classes is truly negligible. If the latter, appropriate references should be added.

Minor

Line 41: Replace with “enter” singular.

Line 43: Suggest replacing “contributing to” to “potentially stabilizing” given the focus on hypothetical portfolio effects.

Lines 71-73: The phrase “centers” implies to me a multimodal distribution of spawning density; however, the egg data in Fig. 1 appear approximately normally distributed suggesting a continuum of spawning habitat rather than distinct spawning areas. That being said, the rationale in lines 116-118 seems reasonable, I would suggest adding a brief version to the figure caption to help confused readers.

Line 79-80: Suggest re-phrasing to “Points show yearly values of egg density and box plots represent interannual variation”.

Line 88-89: I find the phrase “characteristic migration modes” a little vague. Perhaps specify whether the interest is primarily variation in spatial distribution, temporal distribution, or both? Lines 71-73 suggest the interest is primarily spatial, but the first paragraph, and the bulk of the manuscript, include reference to timing as well.

Line 104: Strictly speaking where do lower estuary individuals fall on the resident/migratory continuum? Also 126 indicates all individuals were migratory ocean contingent. Since fish were tagged in the upper estuary, where individuals are defined as resident, how was this determined?

Lines 128-130: Any concerns of differential effects of seining vs. electrofishing?

Line 142: Duration of holding?

Line 146: Adding receivers to Fig 1 would be helpful.

Lines 152-161: I know that acquiring receiver metadata from tracking networks can be difficult, but if possible, a map showing the distribution of the marine receivers (similar to Secor et al. 2020) would be helpful.

Line 175: It might be useful to insert summary statistics on the extent to which individual time series had to be imputed (i.e. mean + SD of number of imputed days vs. observed).

Lines 183-185: As noted above, given the novelty of dynamic time warping in ecological studies (e.g. line 384) I think a paragraph or so explaining DTW (assumptions and basically what’s going on under the hood) is warranted.

Line 199: The language here is a little unclear to me because the authors do attempt to identify environmental drivers in 2018. I think some text is needed to clarify that the 2017 centroids were assumed to be correct. If this is the case though, it’s not clear to me how individuals that strayed (i.e. identified to one reach in 2017 and the other in 2018) were accounted for in this analysis. Additional clarification on this would be helpful.

Lines 248: I find the use of the term “individual spawning runs” a bit misleading because I most generally think of runs as describing behavior of multiple fish. Hence an individual run suggests, to me, a single categorical classification within a group (e.g. fall rather than spring or summer run), rather than an individual’s behavior. I would suggest replacing with individual migration phenologies, individual migrations, or a similar term as appropriate. Similarly, I think it may be more appropriate to phrase the following sentence as: “The median period of the first run, as defined by centroid 1, is…”

Line 261: Any reason not to include river km, instead of latitude, as the y-axis? It seems to be more commonly referenced in the text than latitude (e.g. line 297), is a bit more intuitive to readers, and presumably it would be an easy swap given the data necessary to make Figure 1. Also spawning is mis-spelled.

Lines 268-269: To clarify, the median periods described here are defined by the median centroids identified for 2018; however these are not shown in in Figure 4 or anywhere else, correct? Personally, I think it might be better to show the 2018 centroids in the lower panels of Figure 4, since these seem to be of more interest than the misclassification patterns (which can be easily described in the text). If this change is made then I believe Table 1 could be removed.

Line 298: Again, it may be nitpicky but I would recommend defining run as a phenotype or suite of migration behaviors, rather than an observation. I.e. the authors studied two runs in two years, rather than four runs. As the authors note in their introduction, interannual stability is a defining characteristic of migratory contingents so it seems inappropriate to consider these as independent replicates.

Lines 315: The phrase 2017-categorized is a little confusing initially. In the case of 2016 individuals, I take it to mean fish that were tagged in the lower portion of the river and then had a centroid 2 type behavior?

Line 359: The authors note that the midpoint of both runs in 2017 occurred at a similar temperature (1.7C difference). In 2018 it was a 1.8C difference. So I’m not sure the authors show strong evidence of a differential response to temperature given that interannual variability in temperature seems to overwhelm differences in the temperatures each run experienced.

Lines 360: Use of persist in this context seems a little out of place.

Lines 361-363: The authors don’t present any data that allows for natural mortality to be distinguished from fishing mortality (e.g. tags returned by harvesters). I would recommend either removing the emphasis on harvest or that the authors add citations demonstrating that the majority of adult mortality is associated with harvest. If the latter, I would still suggest adding qualifying language given our generally poor understanding of natural mortality rates.

Lines 368-369: I’m not sure I follow. I assume the authors mean that first-spawners adopt the behaviors of older individuals that are proximate when migrations begin. However, the two runs do not appear to be segregated spatially or temporally during coastal residence and upstream migrants necessarily bypass downstream spawning habitats so it’s unclear to me where proximity and abundance come into play. I recommend the authors refine this hypothesis a bit more.

Line 370: Strictly speaking, this doesn’t seem to be an alternative hypothesis, but rather an explanation for how distinct behaviors are maintained (while the former seems to be more focused on how they arise).

Line 373: Insert “behavioral” (or synonym) before repertoire.

Line 390: What’s meant by comparable here?

Lines 429-430: Again nitpicky, but it seems that by definition spawning runs will exhibit differential use (temporal or spatial) of a habitat. Suggest rephrasing to something like: “Here we show that Hudson River striped bass exhibit evidence of distinct spawning runs despite broadly overlapping coastal distributions”.

Line 436: I believe this should be Fig 3, however I also thought that Fig. 3 only provided an example comparison of two individuals. If so, then it seems like it would be better to reference Figure 4 and Table 1 (if retained). More generally I find the wording of this sentence confusing. It seems to imply that the lower run in 2017 is the only group to encounter the coastal fishery, which is clearly not the case. I would recommend rephrasing and moving the component on in-river vs. coastal slot limits to a separate sentence.

Line 436: Outmigration data from 2016 are not clearly presented in the results, except in Figure 7 which is difficult to interpret at fine temporal scales. I’d recommend adding somewhere if they’re going to be discussed in detail.

Line 444-447: I’d finish this paragraph off by clearly stating that the similar coastal distributions suggest that differential survival in 2017 is likely driven by early ocean entry phenology

Figure 7: I would recommend increasing the resolution on the final figure version and removing the panel borders that separate different years of observation. I think it would be better to show this as a one-panel figure with different years denoted along the x-axis so it’s immediately clear to readers that these are year-round observations (rather than 3 distinct tag deployments).

Table 1: I found Table 1 a little time consuming to interpret. I would recommend replacing with a figure providing equivalent summary statistics for easier interpretation. Alternatively, Figure 4 could be expanded to clearly show centroids in 2018, which would serve a similar purpose.

Reviewer #2: Summary

In their manuscript, Secor et al. describe their identification of two spawning subgroups within the Hudson River Estuary population of ocean-migrating striped bass that spawn annually. They suggest that these two subgroups are spatially segregated during spawning and follow similar migratory paths but exhibit different phenologies (i.e., staggered within the spawning season) as they enter and leave their respective spawning grounds. The mean daily latitudes of fish within these spawning subgroups were tracked with acoustic telemetry. Movement patterns of subgroups into the spawning reaches were then defined (upper and lower spawning runs) and time series differences analyzed using dynamic time warping. Individuals are proposed to maintain membership within established groups over time, based on 2 years of recapture data (2017-2018), where characterization of the upper versus lower spawning run subgroup was based on the 2017 spawning run. The lower spawning run occurred earlier (entered the study area and left the study area earlier) than the upper spawning run in 2017, but not 2018. They also find that these two subgroups follow similar routes along the Atlantic shelf outside of the spawning season. Moreover, the authors argue that different mortality rates between the subgroups help to stabilize the larger population dynamics.

Major comments

I think this is an interesting study that characterizes patterns of movement behavior for striped bass within the Hudson River Estuary and on the coastal shelf, and provides some support for spatial segregation of an upper and lower run during spawning. The manuscript was written well, and generally easy to understand. Survival analyses and conclusions from the survival analyses seemed appropriate (assuming that lower run fish were adequately detected, which was not clear to me based on Lines 149-150). The authors did a good job relating the increased mortality rate of the lower spawning group with the earlier (relative to the upper spawning group) entrance into shelf fisheries in 2017. This earlier migration behavior was potentially initiated in response to a steady increase in water temperatures throughout April that also triggered their earlier movement up-estuary to spawn.

A main concern is that the membership of individual tagged fish within the lower spawning run centroid was only consistent across 2017-2018 for 8 of 12 fish. Thus, although it is true that only 11% of individuals were misclassified between the two years overall, all of those misclassified were originally thought to be part of the lower spawning group, with few fish in that group overall. If the study was able to recapture fish in 2019, and there was evidence of continued site fidelity by those lower spawning group individuals, the case for spatially segregated groups with consistent membership would have been much stronger.

In addition, this group seems to have much lower membership overall, so I wonder how much it would be able to contribute to an overall stabilizing portfolio effect. If there is evidence of at times higher spawner abundances in the lower reach than in the upper reach (e.g., based on the decades of monitoring spawning adults mentioned in Lines 118-120), please include in Discussion. It is mentioned in Lines 452-454 that future climate change and warmer springs could lead to favoring the lower run. Please expand on how warmer springs could favor the lower run because it is not clear to me. Addressing these issues would help support the authors proposal that the two spawning runs could contribute to a portolio effect that stabilizes population dynamics for this contingent of striped bass.

Minor comments

• Line 116-117: “Spawning adults associated with these reaches are believed to be centered” – what is this belief based on? Observations from the authors?

• Line 117: Should 160-120 be 160-200? However, the lower boundary of the red rectangle in Figure 1 looks like it is greater than 160.

• Line 189: I am not seeing blue and red time series in Figure 2, as described here in the figure caption, just shades of brown.

• Line 213: Can leave out “excluding May”

• Line 293: I see a leveling off in Fig. 6 that starts the beginning of May and extends through mid-May, rather than a plateau only during mid-May

• Line 383-384: The authors included this statement: “Developed as a machine learning tool in speech recognition applications, we believe this is its [dynamic time warping] first application in comparing animal movement patterns.” As I wasn’t familiar with dynamic time warping, I did a quick search to see that it has recently been a suggested approach for analysis of animal movement as demonstrated within Cleasby et al. (2019), and applied to seabirds. I realize this is a recent paper, so understandably missed. I suggest the authors include this reference within the manuscript and adjust their statement accordingly.

References:

Cleasby et al. 2019. Using time-series similarity measures to compare animal movement trajectories in ecology. Behavioral Ecology and Sociobiology 73.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PLOS_One_2018AUG.docx

Click here for additional data file.

2 Oct 2020

Please see attached file Response to Reviewers_Cot2

Submitted filename: Response to Reviewers_Oct2.docx

Click here for additional data file.

26 Oct 2020

PONE-D-20-18274R1

Multiple spawning run contingents and population consequences in migratory striped bass Morone saxatilis

PLOS ONE

Dear Dr. Secor,

Thank you for submitting your revised manuscript to PLOS ONE.

I have had one of the original referees review the new version of your manuscript. The reviewer is generally very pleased, but has a few minor points (s)he wants you to address.

When you have taken care of these I will recommend your manuscript for publication. I see no need for further review given that the remaining issues are minor.

Please submit your revised manuscript by Dec 10 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Geir Ottersen

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Again, I greatly enjoyed the opportunity to read this manuscript and commend the authors on their work. They thoroughly addressed my original comments and I particularly appreciated the additional clarity regarding DTW. I only have a few additional suggestions listed below.

Line 33-35: Some awkward wording in this sentence.

Line 354: Contingents mis-spelled as continents.

Line 357-359: Check edits.

Figure 7. The figure has been updated to include receiver detections in the authors’ reply (bottom), but not in the attached figure preceding the revised manuscript. Editorial office should double-check before proofing.

Line 431: Suggest inserting a but before used to contrast the contingents differences from their similarities.

Line 432-438: Suggest rephrasing to clarify that in one of two years the contingents were exposed to early summer fisheries at different rates, which was associated with greater mortality.

Line 531: Add “with” before catch and release.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Cameron Freshwater

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

28 Oct 2020

All changes have been made according the reviewers helpful comments and edits.

Reviewer #1: Again, I greatly enjoyed the opportunity to read this manuscript and commend the authors on their work. They thoroughly addressed my original comments and I particularly appreciated the additional clarity regarding DTW. I only have a few additional suggestions listed below.

Line 33-35: Some awkward wording in this sentence.

Line 354: Contingents mis-spelled as continents.

Line 357-359: Check edits.

Figure 7. The figure has been updated to include receiver detections in the authors’ reply (bottom), but not in the attached figure preceding the revised manuscript. Editorial office should double-check before proofing.

Line 431: Suggest inserting a but before used to contrast the contingents differences from their similarities.

Line 432-438: Suggest rephrasing to clarify that in one of two years the contingents were exposed to early summer fisheries at different rates, which was associated with greater mortality.

Line 531: Add “with” before catch and release.

Submitted filename: Response to Reviewer_Oct26.docx

Click here for additional data file.

10 Nov 2020

Multiple spawning run contingents and population consequences in migratory striped bass Morone saxatilis

PONE-D-20-18274R2

Dear Dr. Secor,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Geir Ottersen

Academic Editor

PLOS ONE

13 Nov 2020

PONE-D-20-18274R2

Multiple spawning run contingents and population consequences in migratory striped bass Morone saxatilis

Dear Dr. Secor:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Geir Ottersen

Academic Editor

PLOS ONE