Resource partitioning facilitates coexistence in sympatric cetaceans in the California Current

Abstract Resource partitioning is an important process driving habitat use and foraging strategies in sympatric species that potentially compete. Differences in foraging behavior are hypothesized to contribute to species coexistence by facilitating resource partitioning, but little is known on the multiple mechanisms for partitioning that may occur simultaneously. Studies are further limited in the marine environment, where the spatial and temporal distribution of resources is highly dynamic and subsequently difficult to quantify. We investigated potential pathways by which foraging behavior may facilitate resource partitioning in two of the largest co‐occurring and closely related species on Earth, blue (Balaenoptera musculus) and humpback (Megaptera novaeangliae) whales. We integrated multiple long‐term datasets (line‐transect surveys, whale‐watching records, net sampling, stable isotope analysis, and remote‐sensing of oceanographic parameters) to compare the diet, phenology, and distribution of the two species during their foraging periods in the highly productive waters of Monterey Bay, California, USA within the California Current Ecosystem. Our long‐term study reveals that blue and humpback whales likely facilitate sympatry by partitioning their foraging along three axes: trophic, temporal, and spatial. Blue whales were specialists foraging on krill, predictably targeting a seasonal peak in krill abundance, were present in the bay for an average of 4.7 months, and were spatially restricted at the continental shelf break. In contrast, humpback whales were generalists apparently feeding on a mixed diet of krill and fishes depending on relative abundances, were present in the bay for a more extended period (average of 6.6 months), and had a broader spatial distribution at the shelf break and inshore. Ultimately, competition for common resources can lead to behavioral, morphological, and physiological character displacement between sympatric species. Understanding the mechanisms for species coexistence is both fundamental to maintaining biodiverse ecosystems, and provides insight into the evolutionary drivers of morphological differences in closely related species.

tioning, but little is known on the multiple mechanisms for partitioning that may occur simultaneously. Studies are further limited in the marine environment, where the spatial and temporal distribution of resources is highly dynamic and subsequently difficult to quantify. We investigated potential pathways by which foraging behavior may facilitate resource partitioning in two of the largest co-occurring and closely related species on Earth, blue (Balaenoptera musculus) and humpback (Megaptera novaeangliae) whales. We integrated multiple long-term datasets (line-transect surveys, whale-watching records, net sampling, stable isotope analysis, and remote-sensing of oceanographic parameters) to compare the diet, phenology, and distribution of the two species during their foraging periods in the highly productive waters of Monterey Bay, California, USA within the California Current Ecosystem. Our long-term study reveals that blue and humpback whales likely facilitate sympatry by partitioning their foraging along three axes: trophic, temporal, and spatial. Blue whales were specialists foraging on krill, predictably targeting a seasonal peak in krill abundance, were present in the bay for an average of 4.7 months, and were spatially restricted at the continental shelf break. In contrast, humpback whales were generalists apparently feeding on a mixed diet of krill and fishes depending on relative abundances, were present in the bay for a more extended period (average of 6.6 months), and had a broader spatial distribution at the shelf break and inshore. Ultimately, competition for common resources can lead to behavioral, morphological, and physiological character displacement between sympatric species. Understanding the mechanisms for species coexistence is both fundamental to maintaining biodiverse ecosystems, and provides insight into the evolutionary drivers of morphological differences in closely related species.

| INTRODUCTION
Understanding the role of resource competition and partitioning between sympatric species in driving differential habitat use and foraging strategies has long been a fundamental question in ecology (Brown & Wilson, 1956;Fenchel, 1975;Grant, 1972). Resource partitioning promotes the coexistence of species that compete for shared limited resources (Schoener, 1974;Toft, 1985;Walter, 1991) via qualitative, temporal, or spatial differences in how these resources are exploited (Brown & Wilson, 1956;Murray & Brown, 1993;Pfennig, Rice, & Martin, 2006).
These Balaenopterid whales are characterized by a unique combination of morphological traits (ventral pleats and baleens) allowing them to use lunge feeding to capture and swallow large quantities of small individual prey in a single batch feeding event (Goldbogen et al., 2012;Kawamura, 1980). The availability of large, dense prey aggregations are therefore requisite for successful feeding (Croll et al., 2001;Goldbogen et al., 2011Goldbogen et al., , 2012Santora, Reiss, Loeb, & Veit, 2010). Such prey aggregations occur in discrete regions of exceptionally high productivity, often associated with fronts, upwelling centers, and steep topography leading to strong spatial and temporal patchiness (Brentnall, Richards, Brindley, & Murphy, 2003;Croll et al., 2005;Santora, Sydeman, Schroeder, Wells, & Field, 2011). This often results in seasonal sympatry of multiple filter-feeding whale species in spatially-restricted productive areas. In these instances, resource partitioning facilitates reduced interspecific competition and increased energy gain across species. For example, differences in resource utilization have been described in sympatric humpback (Megaptera novaeangliae) and Antarctic minke (Baleanoptera bonaerensis) whales around the Western Antarctic Peninsula, where minke whales appear to target deeper krill aggregations than larger humpback whales, as well as exhibit differences in their horizontal spatial distributions (Friedlaender, Johnston, Fraser, Burns, & Costa, 2011;Friedlaender, Lawson, & Halpin, 2009). Little is known, however, on the multiple mechanisms for resource partitioning that may occur simultaneously, such as differences in target prey species, temporal distribution, and spatial distribution in foraging.
Here, we evaluate the hypothesis that multiple differences in foraging behavior facilitate resource partitioning by examining the foraging ecology of two closely-related sympatric whale species. We integrate complementary datasets, including field surveys, stable isotope analysis, and remote-sensing of oceanographic variables, to simultaneously evaluate the qualitative (trophic), temporal and spatial differences in foraging of blue and humpback whales. These species co-occur seasonally in the highly productive waters of Monterey Bay in central California, USA. Understanding the mechanisms for species coexistence is both fundamental to maintaining biodiverse ecosystems and provides insight into the evolutionary drivers of behavioral, morphological and physiological differences in closely related species.

| Study area
Monterey Bay (36.80°N, 121.90°W) is a large (~1200 km 2 ) open bay located off the central California coast (Benson, Croll, Marinovic, Chavez, & Harvey, 2002;Croll et al., 2005; Figure 1) and is divided by the Monterey Submarine Canyon; one of the largest in the world (Shepard & Marshall, 1973). Two nearly equal shallower shelves (up to 140 m deep and 10 to 15 km wide) surround the deeper waters of the canyon located in the center of the bay (Greene, Maher, & Paull, 2002). This highly productive coastal area is strongly influenced by two seasonal upwelling modes: a spring/summer wind-driven upwelling period and a winter relaxation or downwelling period (Black, Schroeder, Sydeman, Bograd, & Lawson, 2010;Black et al., 2011;Bograd et al., 2009;Schroeder et al., 2009).

K E Y W O R D S
Balaenoptera, foraging ecology, interspecific competition, resource partitioning, species coexistence, trophic position

| Oceanographic data
Satellite oceanographic data hypothesized to influence whale habitat use and previously shown to affect other marine predators were obtained for a 3 × 3 pixel (i.e. 9 km 2 ) region centred in Monterey Bay (36.80°N, 121.90°W) for the study period 1997-2006 (Black et al., 2010(Black et al., , 2011Schroeder et al., 2013). A monthly upwelling index (UI; m 3 s −1 100 m −1 ) was derived by the NOAA Fisheries, Environmental Research Division (http://www.pfeg.noaa.gov). The index is based on estimates of offshore Ekman transport driven by geostrophic wind stress derived from 6-hourly, synoptic, surface atmospheric pressure fields (Bakun & Nelson, 1991). Ocean color (Sea-Viewing Wide Field Sensor, SeaWiFS, and Moderate Resolution Imaging Spectrometer, MODIS AQUA) data were obtained from the NASA Ocean Biology Processing Group (OBPG). Monthly averages of Chlorophyll-a (mg m −3 ) and depth-integrated primary production (mg C m −2 day −1 ) were derived using data processed with the standard OBPG methods for chlorophyll-a, and the standard Vertically Generalized Production Model (VGPM) for primary production as described by Behrenfeld and Falkowski (1997), modified to use a non-varying chlorophyllspecific primary productivity term (P B opt ) as described by Kudela, Cochlan, Peterson, and Trick (2006). A cumulative upwelling index (CUI) at 36°N, 122°W and the North Pacific High (NPH) position and strength indices (NPH's monthly areal extent (A) and maximum pressure (Pmax)) were included to examine broader scale forcing of the Monterey Bay Ecosystem (Schroeder et al., 2013). F I G U R E 1 Monterey Bay study area. Monterey Submarine Canyon is characterized by waters deeper than 500 m (light grey lines). Black lines indicate transect lines (10-22 km in length; 5.5 km apart) followed during shipboard surveys to sample krill and forage fish distributions and densities using hydroacoustics, as well as whale distribution and density. Black circles indicate net tow sites to sample zooplankton abundance.

| Line-transect surveys
to and in March 2003. Each survey consisted of seven transect lines ranging from 10 to 22 km in length and spaced 5.5 km apart (Figure 1). A survey totaled ~126 km and was typically completed in two consecutive days at a ship speed of 18.5 km h −1 (10 knots). The entire survey area, ~909 km 2 , included all of Monterey Bay and the waters off the Monterey Peninsula (except nearshore regions) beginning at the 55 m (30 fathom) isobaths and extending WNW to 122.08°W longitude (Figure 1, for more details on the survey design, see Benson et al., 2002;Croll et al., 2005).

| Hydroacoustic survey and net sampling
Krill (i.e., euphausiids) and forage fish densities were estimated along survey transects from 1997 to 2006 and from 2003 to 2006, respectively, using Simrad echosounders. Krill hydroacoustic data were analysed based on the methods described in Croll et al. (1998) and Hewitt and Demer (1993) to provide relative integrated measures of acoustic backscatter for euphausiids for the whole survey grid. Krill schools were identified and scrutinized from other scattering organisms based upon school morphology and frequency-specific differences in backscatter strength. Volume backscattering was integrated vertically from 5 m below the surface down to either 200 m or 5 m above the bottom, and averaged over 1 km horizontal intervals of each transect. Krill schools detected hydroacoustically were additionally confirmed by periodic targeted net tows. The net was towed obliquely to either 10 m above the bottom or to 200 m depth. Krill density (number/1000 m 3 ) and zooplankton biovolume (total zooplankton displacement volume, mL/1000 m 3 ) were calculated based on conventional MOCNESS/BONGO net sampling as described in Marinovic, Croll, Gong, Benson, and Chavez (2002).
Forage fish schools were identified from hydroacoustic records based upon school morphology and differential target strengths, as organisms with swim bladders (e.g., many fishes) scatter greater at lower frequencies than those without (e.g., krill; Simmonds & MacLennan, 2005). Due to difficulties associated with distinguishing among fish species using hydroacoustic data, we included all hydroacoustic records matching our schooling fish criteria under the general term forage fish. Forage fish density was calculated as school encounter rate, where individual schools were identified and the number of schools encountered per kilometer of transect-line was calculated.

| Spatio-temporal patterns
To examine temporal patterns in prey distribution and density, hydroacoustic and net sampling data were averaged monthly across all  Total number of sightings per day per species was computed to estimate relative daily abundance (calculated as mean number of individual whale sightings per day per species). While opportunistic data are not appropriate to obtain reliable absolute density estimates, they can be used to examine phenology, or the timing of when both species used the bay, and seasonal trends. To examine species phenology, we only used years where trips were conducted from at least April to November (n = 6 years, 1996-1998; 2000-2002), encompassing the main season for both species. We used two criteria to define the arrival date: (i) arrival date is the date of first sighting of a species, (ii) this first sighting should be followed by regular sightings over the following 10 days. The departure date was considered as the date of last sighting of a species in the bay.

| Spatio-temporal patterns
To examine temporal patterns in whale distribution, density, and relative abundance, survey data were averaged monthly across all sampled stations and whale watching data were averaged monthly across all trips. These monthly averages were then combined into a time series (1997-2007 for line-transect data, 1993-2004

| Whale trophic position and inferred diet
To examine trophic position of blue and humpback whales, we conducted stable isotope analysis on skin samples obtained from biopsy darts. Stable isotope analysis is a powerful tool for assessing the diet composition and trophic level of predators (Newsome, del Rio, Bearhop, & Phillips, 2007;Post, 2002). All samples used in this study were obtained in 2005 as part of a larger biopsy sampling program (Fleming, Clark, Calambokidis, & Barlow, 2015). A total of 39 individuals were sampled (25 humpback, 14 blue whale) along the U.S. California Current range of the whales (Washington to California).
Skin samples were collected using a small stainless steel biopsy dart fired from a crossbow (summarized in Ralls and Hoelzel (1992)).
Cetacean skin is a metabolically active tissue, which reflects recent di- Samples were separated by species and early vs. late upwelling/ oceanic time period in Monterey Bay, using July 15th as the approximate transition date (Pennington & Chavez, 2000). Tissue samples were sent to the Colorado Plateau Stable Isotope Library, Northern Arizona University (Flagstaff, Arizona). Samples were oven dried followed by lipid extraction (Soxhlet) and homogenized for determination of δ 15 N. Nitrogren isotope ratios are useful for assessing trophic position, as higher trophic levels have higher values of δ 15 N (Post, 2002). In addition, δ 15 N in humpback whales have been shown to be significantly positively related to forage fish abundance and negatively related to krill abundance in the California Current System, allowing diet inferences (Fleming et al., 2015). Analyses were conducted using a Finnigan Delta Plus isotopic ratio mass spectrometer (Thermo Electron Corporation, Waltham, MA). Detailed description of methods is presented in Fleming et al. (2015).

| Oceanographic conditions and predator/prey temporal distribution
During the peak in krill density in May, fish density was at its lowest (6.3 ± 8.7 schools km −1 Figure 2d).
From whale watch records between 1993 and 2004, mean blue whale arrival and departure dates to Monterey Bay were July 22th (±22 days) and October 31st (±49 days), respectively (Figure 3a).
Relative monthly abundance, measured as the mean number of sightings per day in a given month, of blue whales peaked in August (8.7 ± 1.5 month, Figure 3a), whereas humpback whale relative abundance was bimodal with peaks occurring in late April (4.8 ± 0.4 month) and July (7.6 ± 1.4 month, Figure 3b). Relative blue whale monthly abundance was positively correlated with the monthly cumulative upwelling index (CUI; Table 1). Relative humpback whale monthly abundance was positively correlated with the monthly upwelling index at 36N, monthly krill density, and monthly areal extent (A) and maximum pressure (pmax) of the NPH (Table 1). Annual peak in the relative abundance of blue whales was not correlated with the density of krill in July or August, however annual peak in the relative abundance of humpback whales was positively correlated with the density of krill in June (Spearman's correlation, ρ = 0.88, p = .07, n = 5).

| Predator/prey spatial distribution
Over a ten-year period, sightings of blue whales were more common off the continental shelf than on the continental shelf (84.4 ± 6.8%  Figure 4b).
In addition, there were significantly more sightings of both blue

| DISCUSSION
Long-term studies integrating complementary datasets offer unique opportunities to understand the ecology of wide-ranging and longlived species. Our study highlights the differences in the spatiotemporal distribution and foraging behavior of two seasonally sympatric whale species (blue and humpback), and suggests that these distributional and behavioral differences facilitate reduced competition between these two closely related species.

| A highly productive hotspot
Monterey Bay has been described as a highly productive "hotspot" supporting high concentrations of forage fishes (e.g., anchovy, sardines), krill and top predators (Santora et al., 2012). Our long-term dataset ranging from physical oceanography to top predator ecology showcases this productivity and underscores its seasonal nature. Our study shows that the spring transition to summertime upwelling leads to a peak in phytoplankton standing crop in March-April which results in krill density peaking in May/June (see also Croll et al., 2005;Marinovic et al., 2002). A secondary peak in krill density occurs again in August as individual krill spawned during late upwelling enter the population, and adults, resulting from the primary recruitment pulse in May, are still surviving in the population (see also Marinovic et al., 2002;Croll et al., 2005). Thus, while krill are present year-round in Monterey Bay, their density is strongly seasonal, with two peaks largely driven by physical/biological coupling of upwelling dynamics.
Our study also reveals that forage fishes were likewise encountered throughout the year, but similarly displayed strong seasonality associated with upwelling dynamics (Black et al., 2011). However, in an opposite pattern to krill density, forage fish density was greatest during relaxation periods.
Our results show that krill species in Monterey Bay are associated with the Monterey Submarine Canyon shelf break, and are more abundant in the southern part of the bay and off the continental shelf compared to on the shelf. The relationship between krill density and steep topographic features -particularly channels and shelf edges -has been previously described in Monterey Bay (Croll et al., 1998(Croll et al., , 2005Santora et al., 2011) and elsewhere (Lavoie, Simard, & Saucier, 2000). In contrast, forage fishes were more abundant on the continental shelf and north of the Monterey Bay Canyon. Upwelling dynamics lead to temporal and spatial lags between peak upwelling, peak primary production, and peak secondary production and enhanced grazing (Wilkerson, Lassiter, Dugdale, Marchi, & Hogue, 2006). These temporal and spatial lags, combined with the upwelling shadow mechanism described for Monterey Bay (Graham & Largier, 1997)

| Foraging strategies and resource partitioning
Blue and humpback whales are both filter feeders with high energetic demands, requiring dense prey aggregations to feed successfully (Croll et al., 2001;Goldbogen et al., 2011). Blue whales are known to exclusively feed on krill (Sears & Perrin, 2009), whereas humpback whales have a more diverse diet, including krill as well as small schooling fish (Mann, 2000). Such dense aggregations only occur in discrete regions of exceptionally high productivity and are characterized by strong spatial and temporal patchiness, presenting the potential for resource competition between these species.
Our results suggest that due to their more generalist diet, humpback whales are able to switch prey depending on relative abundances, which may serve as a mechanism to reduce interspecific competition.
Over the study period, humpback relative abundance and density were consistently high throughout the entire upwelling season; even during months when forage fish density was low, i.e., in May and June.
The annual peak in the relative abundance of humpback whales was positively correlated with the density of krill in June, suggesting that humpbacks may target the early season peak in krill density. Later in the season, we suggest that humpbacks switch to feeding upon forage fishes as fish density increases. Our stable isotope analysis for 2005 supports this hypothesis: humpbacks had greater mean δ 15 N values than blue whales, suggesting a diet comprised of higher trophic level prey. While we could not verify δ 15 N values of krill and forage fish species at the time of sampling, stable isotopes processed using the same methods in humpback whales in the California Current accurately reflected changes in prey choice and availability between krill and forage fish over multiple years (Fleming et al., 2015). In addition, a review of the literature for isotopic values of prey (i.e. krill and forage fish species) in the California Current System (Becker, Peery, & Beissinger, 2007;Brodeur, Suchman, Reese, Miller, & Daly, 2008;Miller et al., 2013;Sydeman, Hobson, Pyle, & McLaren, 1997) suggested δ 15 N values for krill are on average 2 to 3% lower than δ 15 N values for forage fish species, which is agreement with our interpretation of the results.
In comparison, blue whales had a narrower peaked abundance distribution than humpback whales. Humpback whales often remain for a long period of time (~6 months) in a single foraging area (Baker et al. 2013), whereas blue whales have been shown to move great distances between regions of seasonal productivity (Calambokidis, Barlow, Ford, Chandler, & Douglas, 2009), feeding in a more localized area for about 3 weeks before migrating to another area (Bailey et al., 2009 (Black et al., 2010(Black et al., , 2011Schroeder et al., 2013). While humpbacks seemed to respond more to large scale forcing, blue whales seemed to be influenced by more localised cues.
The annual peak in blue whale density which occurred on average between mid-July and August, at the same time as the secondary, more predictable, late summer peak in krill, was correlated with the maximum biovolume of zooplankton in the Bay. Our results also reveal an interesting difference between both species: while monthly humpback whale density was linked with the monthly krill density in Monterey Bay, monthly blue whale density was more closely correlated with the monthly density of krill schools. Blue whales have extremely high prey demand, requiring up to two tons of prey per day (Goldbogen et al., 2011;Rice, 1978). This constrains them to feed upon extremely dense euphausiid schools, presumably leading to a relatively high threshold density of both prey and schools for feeding events to be profitable (Acevedo-Gutierrez, Croll, & Tershy, 2002;Goldbogen et al., 2011;Hazen et al., 2015). This may be why blue whales in Monterey Bay were attuned to predictable and low variance peaks in prey compared to periods of higher and more variable prey density and timed their arrival with the late summer peak in krill.
Within the bay, the two species overlapped spatially; both species were preferentially found south of the Monterey Canyon. However, blue whales were mainly observed offshore, where krill abundance was the highest, whereas humpback whales foraged both on and off the continental shelf, most likely related to their target prey. Blue whales therefore appear to target specific spatial and temporal patterns within Monterey Bay to efficiently feed on predictably dense krill schools along the shelf break and off-shelf (Croll et al., 2005).

| Character displacement hypothesis
We suggest that the differences in habitat use and foraging behavior observed between blue and humpback whales in our study are consistent with interspecific differences in body size and lend support for the character displacement hypothesis. The character displacement hypothesis contends that competition between species drives the evolution of differences in morphology (Brown & Wilson, 1956;Grant, 1972;Grant & Grant, 2006). For example, a study examining the spatial association between baleen whales and their principal prey, Antarctic krill Euphausia superba, near the South Shetland Islands (Antarctic Peninsula), found that humpback, fin and minke whales partition foraging habitat based on the size of their prey (Santora et al., 2010): humpback were associated with small (<35 mm) juvenile krill, fin whales Balaenoptera physalus were associated with large (>45 mm) mature krill located offshore and Antarctic minke whales were associated with intermediate sized krill (35-44 mm). Different size and morphology of baleen plates may have influenced these whales' divergent prey selections (Gaskin, 1982).
Differences in overall body size may also manifest in decreased competition, as these lead directly to differences in cost of transport, energy storage, and prey capture rates, and indirectly to differences in fasting and migratory abilities (Dial et al., 2008;Domenici, 2000;Howland, 1974;Huey & Hertz, 1984;Peters, 1986). The extreme body size of blue whales is associated with both a lower mass-specific metabolic rate and cost of transport, an advantage for long distance travelling (Goldbogen et al., 2011). Large lipid reserves can also serve as a buffer from variability in coastal productivity, which is particularly important for a predator specialised in a patchy and ephemeral resource such as krill. The higher mass-specific engulfment capacity of blue whales has also been suggested to be more efficient for krill feeding, while the humpback's lower capacity values and higher manoeuvrability related to its smaller body size (Domenici, 2000;McGuire & Dudley, 2005) may be better for exploiting more agile prey like fish (Goldbogen et al., 2011;Figure 5b). Evaluating the extent to which differential selection on body size may be occurring between the two species to facilitate resource partitioning would be a ripe area for future research.

| CONCLUSION
Areas of high seasonal productivity, such as Monterey Bay, are critical foraging areas for a wide range of species (Block et al., 2011). In these areas, the probability of overlapping distributions and resource competition among species is high and can therefore present an opportunity to examine, in the field, how competition for resources may drive resource partitioning between seasonally sympatric, closely related species. Ultimately, such partitioning can lead to behavioral, morphological, and physiological character displacement between sympatric species. Our long-term study using complementary approaches reveals several mechanisms operating simultaneously that may facilitate coexistence among some of the largest co-occurring species on Earth.

ACKNOWLEDGMENTS
We would like to thank and memorialize two tireless and enthusiastic boat captains that helped make this

DATA ACCESSIBILITY
We will publicly archive data supporting our findings on Dryad.