UBC Faculty Research and Publications

Fine-scale foraging movements by fish-eating killer whales (Orcinus orca) relate to the vertical distributions… Wright, Brianna M; Ford, John K B; Ellis, Graeme M; Deecke, Volker B; Shapiro, Ari D; Battaile, Brian C; Trites, Andrew W Feb 20, 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-40462_2017_Article_94.pdf [ 1.28MB ]
JSON: 52383-1.0343036.json
JSON-LD: 52383-1.0343036-ld.json
RDF/XML (Pretty): 52383-1.0343036-rdf.xml
RDF/JSON: 52383-1.0343036-rdf.json
Turtle: 52383-1.0343036-turtle.txt
N-Triples: 52383-1.0343036-rdf-ntriples.txt
Original Record: 52383-1.0343036-source.json
Full Text

Full Text

RESEARCH Open AccessFine-scale foraging movements byfish-eating killer whales (Orcinus orca)relate to the vertical distributions andescape responses of salmonid prey(Oncorhynchus spp.)Brianna M. Wright1,2,3*, John K. B. Ford2,3, Graeme M. Ellis3, Volker B. Deecke4, Ari Daniel Shapiro5,Brian C. Battaile1,2 and Andrew W. Trites1,2AbstractBackground: We sought to quantitatively describe the fine-scale foraging behavior of northern resident killerwhales (Orcinus orca), a population of fish-eating killer whales that feeds almost exclusively on Pacific salmon(Oncorhynchus spp.). To reconstruct the underwater movements of these specialist predators, we deployed 34biologging Dtags on 32 individuals and collected high-resolution, three-dimensional accelerometry and acousticdata. We used the resulting dive paths to compare killer whale foraging behavior to the distributions of differentsalmonid prey species. Understanding the foraging movements of these threatened predators is important from aconservation standpoint, since prey availability has been identified as a limiting factor in their population dynamicsand recovery.Results: Three-dimensional dive tracks indicated that foraging (N = 701) and non-foraging dives (N = 10,618) werekinematically distinct (Wilks’ lambda: λ16 = 0.321, P < 0.001). While foraging, killer whales dove deeper, remainedsubmerged longer, swam faster, increased their dive path tortuosity, and rolled their bodies to a greater extentthan during other activities. Maximum foraging dive depths reflected the deeper vertical distribution of Chinook(compared to other salmonids) and the tendency of Pacific salmon to evade predators by diving steeply.Kinematic characteristics of prey pursuit by resident killer whales also revealed several other escape strategiesemployed by salmon attempting to avoid predation, including increased swimming speeds and evasivemaneuvering.Conclusions: High-resolution dive tracks reconstructed using data collected by multi-sensor accelerometer tagsfound that movements by resident killer whales relate significantly to the vertical distributions and escaperesponses of their primary prey, Pacific salmon.Keywords: Foraging, Movement, Diving behavior, Biologging, Dtag, Accelerometry, Killer whale, Orcinus orca,Pacific salmon* Correspondence: brianna.wright@dfo-mpo.gc.ca1Marine Mammal Research Unit, Institute for the Oceans and Fisheries,University of British Columbia, AERL Building, Room 247 - 2202 Main Mall,Vancouver, BC V6T 1Z4, Canada2Department of Zoology, University of British Columbia, #4200 - 6270University Blvd., Vancouver, BC V6T 1Z4, CanadaFull list of author information is available at the end of the article© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Wright et al. Movement Ecology  (2017) 5:3 DOI 10.1186/s40462-017-0094-0BackgroundEffective movement patterns during prey searching andcapture are critical to the successful acquisition of re-sources, and are thus a vital component of the foragingbehavior of predators. The efficiency of such move-ments affects an individual’s ability to meet its dailyenergetic requirements, which in turn has a direct im-pact on survival and reproduction, ultimately leading topopulation-level consequences [1, 2]. The ability to ac-curately describe and quantify the kinematic character-istics of foraging behavior is therefore of great interestto ecologists. Analysis of movement patterns by preda-tors during the pursuit phase of hunting can also shedlight on the escape behaviors and predation avoidancestrategies employed by prey. However, detailed behav-ioral studies of movement can be particularly challen-ging to conduct on large marine predators, such askiller whales and other cetaceans, as these species aretypically far-ranging, are only periodically visible at thesurface and move within a complex, three-dimensionalenvironment [1, 3, 4].Most studies of the foraging behavior of fish-eating,or ‘resident’, killer whales in the northeastern PacificOcean have been limited to observations of activity vis-ible at the surface [5–7]. Past studies have shown thatgroups of resident killer whales tend to separate intosmaller subgroups that spread out over several squarekilometers while hunting, but travel in the same generaldirection [5]. Dives by individuals in these subgroupsare typically asynchronous, and are often characterizedby sudden changes of direction, lunges or milling be-havior [5]. Surface observations from previous studiesnoted that foraging whales usually perform sequencesof several short dives followed by a longer dive [5].Capture success during these longer dives can often bedetermined from the presence of fish scales and fleshin the upper water column after the whale has sur-faced [6, 8]. Such physical remains from kills are espe-cially evident when fish are broken up and shared, abehavior that occurs frequently between maternally re-lated individuals [6, 9].In addition to surface observations, a few foragingstudies have deployed time-depth recorders (TDRs)with paddle-wheel swim speed sensors to quantify thediving behavior of resident killer whales [10, 11]. Theyhave shown that dive rate and swim speeds are greaterduring the day than at night [11]. TDR data have alsorevealed that resident killer whales spend very littletime (2.4%) at depths >30 m, but that these deeper di-ves are frequently associated with velocity spikes thatmay indicate fish chases [10]. The utility of TDR tags islimited, however, as they only collect one-dimensionaldepth profiles and thus cannot address questions ofhorizontal or three-dimensional movement and spaceuse. TDR data have not been able to adequately de-scribe how and where resident killer whales capturetheir prey—information that is needed to fully under-stand their foraging ecology and behavior.Resident killer whales feed almost exclusively onPacific salmon (Oncorhynchus spp.) for at least halfof the year (May to October) and preferentially consumeChinook salmon (O. tshawytscha) over other species[6, 8]. Although Chinook is the least abundant sal-monid in the whales’ range [12, 13], it accounted for71.5% of all identified salmon kills (May to December)in a 28-year study of resident killer whale foraging [6].Resident preference for consuming this prey speciesdoes not appear to be influenced by fluctuations inrelative Chinook availability [14]. Annual Chinook sal-mon abundance has been correlated with residentkiller whale survival and birth rates [15], and has alsobeen linked to changes in their social connectivity [16,17]. The ability of resident killer whales to obtain suf-ficient quantities of Chinook therefore has importantconsequences for their population growth and socialorganization. Residents probably target Chinook be-cause their large size and high lipid content makethem the most energetically profitable of all Pacificsalmon species [18, 19], and because Chinook areavailable year-round in the coastal waters of NorthAmerica [6, 12, 20]. Chum salmon (O. keta) is the sec-ond largest Pacific salmonid and the next most com-monly consumed prey species (22.7%) of residentkiller whales, and becomes an important food sourcein September and October [6]. Smaller salmonids,such as coho (O. kisutch) and pink (O. gorbuscha) sal-mon, and various groundfish species are occasionallyconsumed, but do not appear to contribute signifi-cantly to the overall diet of these whales [8].We sought to produce the first quantitative descrip-tion of fine-scale foraging behavior by fish-eating resi-dent killer whales. We used data from multi-sensorarchival tags to reconstruct the three-dimensionalmovements of individual killer whales during foragingdives and other underwater behaviors that are other-wise impossible to visualize in the wild. We catego-rized dives based on their kinematic similarities usinga multivariate classification technique, with the par-ticular goal of identifying foraging dives. By closelyexamining the structure of these foraging dives, wecould compare killer whale hunting behavior to thevertical distributions of various Pacific salmonids tosee if whales targeted the depth ranges typically usedby preferred prey. Reconstructing foraging movementsalso allowed us to identify common escape strategiesemployed by salmon in response to pursuit by residentkiller whales. Our study lays valuable groundwork forfuture research, as reconstructed dive paths could beWright et al. Movement Ecology  (2017) 5:3 Page 2 of 18used to identify foraging habitat, assess space use, andestimate energy expenditure by individuals from thisthreatened population [21], the dynamics of which arelimited by prey availability [15].MethodsStudy area and tagging methodologyWe used archival Dtags [22] to record the diving behaviorof individuals belonging to the northern resident killerwhale community, a population of 290 animals [23] thatranges throughout the coastal waters of the eastern NorthPacific, from central Vancouver Island, British Columbia,Canada to southeastern Alaska, USA [24]. Dtags were de-ployed during August and September (2009–2012) in thecoastal waters of northeastern Vancouver Island and thecentral coast of British Columbia (Fig. 1). The researchplatform was a 10-m command-bridge vessel poweredby a surface-drive propulsion system, which reducedunderwater engine noise that could affect the whales’behavior. When encountered, individual resident killerwhales were identified with an existing photo-identification catalogue [23, 25] using a technique de-veloped by Bigg [26]. We then approached an individ-ual by matching its speed and direction of travel anddeployed a suction-cup attached Dtag from the bowof the vessel using a 7-m hand-held, carbon-fiberpole. Preferred tag placement was just below the baseof the dorsal fin, where the tag’s VHF antenna wouldclear the water when the whale surfaced, to facilitatetracking of the individual. To minimize potential im-pacts of tagging, whales were never tagged twice dur-ing the same study year (and repeat tagging wasavoided whenever possible across study years); we didnot deploy tags on juveniles under 3 years of age.Dtags recorded depth and three-dimensional body orien-tation (using tri-axial accelerometers and magnetometers)Fig. 1 Georeferenced tracks (black lines) obtained by dead-reckoning for 31 deployments of archival accelerometry tags (DTags) on northernresident killer whales in British Columbia, Canada during August and September, 2009–2012Wright et al. Movement Ecology  (2017) 5:3 Page 3 of 18at sampling rates of 50 (2009–2011) or 250 Hz (2012)[22]. They also recorded underwater sound, whichhelped to identify surfacing events between dives andthe timing of prey captures. Surfacing events were char-acterized by the sound of the tag impacting the air andthen the water again as the whale re-submerged, whileprey captures coincided with increased flow noise dueto body acceleration. Tags detached automatically [22]and were retrieved for downloading of the data. Priorto analysis, sensor data were downsampled to 5 Hz aspart of the tag calibration process [22].Behavioral observations & prey samplingWe conducted focal follows [27] of tagged individualsand noted surface observations of foraging activity usinga digital voice recorder that was time-synchronized withthe tag clock. We obtained periodic (mean interval =21.7 min) GPS surfacing locations throughout each focalfollow to apply as positional corrections during tag trackreconstruction. GPS fixes were collected with minimaldisturbance to the tagged whale by positioning the boatover the ‘flukeprint’ produced after the whale had re-submerged, and matching this location to the associatedprior surfacing time (as indicated by a beep from theVHF receiver, recorded on the time-synchronized digitalvoice-notes). Fluke prints are circular areas of smoothwater created from displacement by the whale’s bodyand turbulence from its tail stroke as it dives, and re-main visible on the surface for several minutes after thewhale has moved on [28]. The need for concurrent sur-face observations limited the tag deployments to daylighthours. Following the methodology of Ford and Ellis [6],we collected fish scales and tissue fragments using afine-meshed dip net when whales surfaced from success-ful foraging dives. These samples were used to confirmsuccessful predation events and to identify the speciesand age of the captured fish. Fish species were identifiedusing scale morphology or genetics [29] and schlero-chronology was used to establish fish age [30].Dtag calibration and identification of divesSensor data were calibrated to correct for the orientationof the tag relative to the body axes of each trackedwhale, and the raw accelerometer and magnetometerdata were converted into pitch, roll, and heading mea-surements [22]. For some deployments, changes in theposition of the Dtag on the animal due to tag slippagerequired performing new calibrations for every neworientation of the tag. Tag slippage was diagnosed duringcalibration by looking for abrupt shifts in the centraltendencies of the raw accelerometer data, plotted againstdeployment time. To discount possible reactions to be-ing tagged, we excluded the first 10 min of data for eachdeployment from further analysis. Most whales displayedmild behavioral responses to tagging (rolling or a slightflinch as the tag was applied) and resumed their pre-tagging swimming patterns within several surfacings(typically <1–2 min).We identified dives from the calibrated tag data usingan automated filter in MATLAB [31] that defined adive as any event with depth ≥1 m that was bounded bysurfacing events of <1 m depth. The shallow depththreshold ensured that all submersions and surfacingswere detected. Each surfacing represented a singlebreath (identified from the acoustic record) and imme-diate submersion by the tagged animal, although mul-tiple breaths per surfacing (i.e., ‘logging’ behavior,during which the whale remained stationary at the sur-face) was infrequently noted but discounted from theanalysis. We were confident that the MATLAB detec-tion filter estimated the start and end times (relative totime of tag activation) and maximum depth for eachdive with high accuracy because we visually compared arandom sample of 50 dives against correspondingthree-dimensional time-series (or ‘pseudotracks’) ofdive behavior that were independently generated usingTrackPlot 2.3 software [32]. For 96% of these randomlysampled dives, the times (rounded to the nearest sec-ond) and depths (rounded to the nearest 0.1 m) calcu-lated by the MATLAB filter were in agreement withthose generated by TrackPlot. Mismatches (>1 s differ-ences) between the MATLAB- and TrackPlot-generateddive times only arose for two dives, which were bothvery shallow (<2 m) and were bounded by indistinctsurfacing events that likely made them difficult for thefilter to resolve. We retained these two dives in the ana-lysis because the mismatch in both end times was rela-tively minor (<3 s).GPS-corrected dead-reckoning of tag tracksWe generated a time-series of two-dimensional locationdata (x, y) for each whale using dead-reckoning and aMATLAB program (‘ptrack’, developed by Woods HoleOceanographic Institution) that applied a Kalman filterto estimate swim speed from an animal’s pitch and rateof change in depth [22]. These speed estimates werecombined with heading measurements to determine theposition of each whale relative to its starting locationover the length of the deployment. Because dead-reckoning uses estimated prior positions to derive loca-tions farther along the track, absolute position estimateswere subject to compounding spatial error over time. Tominimize this error, we georeferenced the dead-reckonedtag tracks by constraining them through periodic GPS sur-facing (flukeprint) locations that we recorded during thefocal follows [33, 34]. GPS ground-truthing of the dead-reckoned tracks reduced the overall error in the time-series of position estimates, although georeferenced tracksWright et al. Movement Ecology  (2017) 5:3 Page 4 of 18with longer time intervals between recorded GPS sur-facing locations likely contained greater error thantracks with more frequent fixes [34, 35]. Our GPS-corrected dead-reckoning method also could not en-tirely account for positional drift of the whale resultingfrom ocean currents or the influence of forces such asinertia, hydrodynamic lift and buoyancy [36–38]. How-ever, it is important to note that dead-reckoning errorsdue to either environmental factors or time-dependentcumulative error in estimated speed, pitch or compassheading primarily lead to inaccuracies in the absoluteposition of tracks [39]. Here, we present comparisonsof relative movement over small temporal scales (at thelevel of the dive) and we employ kinematic variablessuch as tortuosity that are not impacted by systematicover- or under-estimation of swimming speeds [34].Dead-reckoning combined with GPS fixes thereforeprovided a reliable means of producing high-resolution,continuous tracks of underwater movements by taggedwhales [33–35, 39]. Georeferenced tag tracks were plot-ted using ArcGIS software [40] (Fig. 1).Calculation of kinematic dive variablesTo quantify and compare whale movement patterns, wecalculated a set of kinematic variables for each diveusing both the raw sensor data and the dead-reckonedwhale tracks. These variables included dive duration (s),maximum dive depth (m), two-dimensional dive pathtortuosity (i.e., the degree of convolution in the tagtrack, measured using a straightness index), mean vec-torized Dynamic Body Acceleration (VeDBA), maximumabsolute roll (degrees), mean absolute roll (degrees),estimated overall dive speed (m s−1), and the ratio ofdescent duration to ascent duration. Additional variableswere calculated separately for the descent and ascentphases of each dive: three-dimensional dive path tortu-osity, vertical velocity (m s−1), mean rate of change inroll (degrees s−1), and mean rate of change in pointingangle (degrees s−1). We selected the kinematic variablesbased on their expected ability to distinguish foragingdives from other behaviors. Details concerning the cal-culation of these kinematic dive variables are presentedas Additional file 1 (Appendix A1).Fig. 2 Three-dimensional reconstructions of three foraging dives by northern resident killer whales. Panels (a) (V-shaped dive profile) and (c) (U-shapeddive profile with maintained ~90° off-axis roll position at the bottom of the dive) are side views of Chinook salmon captures at depth, while (b) is anaerial view of a surface chase resulting in a chum salmon capture. Red dots represent the probable locations and times of fish captures. Yellow portionsof the track indicate when the whale rolled sideways >40° in either direction, while blue portions indicate roll <40°Wright et al. Movement Ecology  (2017) 5:3 Page 5 of 18Multivariate statistical analysis of kinematic dive variablesWe used the values of the 16 kinematic dive variablesmeasured during successful foraging dives (those fromwhich we obtained fish scale and/or tissue samples, N =17) as the training set in an iterative linear discriminantanalysis (LDA) to identify other dives that likely alsorepresented foraging behavior. Two of the confirmedforaging dives were discounted from the LDA trainingset (leaving N = 15 dives), as both of these predationevents occurred at the surface, rather than during a dive.Because surface chases were made up of multiple brief,extremely shallow dives (Fig. 2b), the dive-by-dive LDAcould only consider very small portions of a surfacechase at once, and could not treat all of the dives withinthe chase as a single capture event.Prior to performing the LDA, we transformed thekinematic dive variables (except the three measures oftortuosity/straightness) by adding 0.01 to eliminatezeros and then taking the natural logarithm. Sincestraightness is a proportional measure, the logit trans-formation was applied to the three tortuosity variables.We added a small value (ε =minimum non-zero valueof 1-y; where y represented the range of values of thetortuosity variable being transformed) to both the nu-merator and denominator of the logit function to pre-vent proportions equal to 0 or 1 from transforminginto undefined values [41]. We assessed whether thedata transformations had achieved multivariate normal-ity (an assumption of LDA) by comparing Q-Q plotsand histograms of the untransformed versus trans-formed kinematic variables. We standardized the trans-formed dive variables by group membership (i.e., theforaging dive training set versus all other unclassifieddives) prior to running each iteration of the LDA. Mul-tiple iterations were run in succession, with reassign-ment of misclassified dives prior to each iteration, untilno more dives were detected as misclassified in eithercategory (‘foraging’ or ‘non-foraging’). In every iter-ation, the 15 confirmed foraging dives with prey sam-ples were always allocated to the ‘foraging’ training set.Due to the small size of the first training set (N = 15)and the small number of whales represented by thesedives (N = 7), it was possible that idiosyncratic behaviormight influence how the LDA identified foraging dives.To determine the relative influence of repeated mea-sures (i.e., the factor of ‘individual’) on the LDA results,we cross-validated the algorithm’s ability to correctlyidentify foraging dives regardless of within-individualbehavior patterns by re-running the analysis with theremoval of each whale’s dives in turn from the firsttraining set (‘leave-one-out’ method [42]). This pro-vided a direct test of the LDA’s capacity to correctlyclassify dives that were not used to calculate the ori-ginal discriminant function.Following the iterative LDA, we analyzed the ‘non-foraging’ dives using X-means clustering [43, 44] toidentify further dive types unrelated to feeding behav-ior. X-means clustering does not rely on a priori know-ledge of group membership [43], which made it suitablefor identifying dive types that lacked ‘true positive’examples for constructing a training set. Wilks’ lambdatests were performed to determine if the two pairs ofdive type groupings, as determined by the LDA (foragingversus non-foraging dives) and X-means clustering(various non-foraging dive behaviors), were statisticallydifferent from one another. We summarized the untrans-formed kinematic dive variables by dive type usingmedians (M) and interquartile ranges (IQR), due to thehighly skewed distributions of many of these variables.Meta-analysis of Pacific salmon vertical distributionTo compare whale diving behavior with that of theirprey, we conducted a meta-analysis of the summer andfall vertical distributions of Pacific salmon species. Usingreported mean swimming depths from salmon ultrasonictelemetry and tagging studies (N = 12), we calculated anoverall average swimming depth for each salmon species,which was compared to killer whale foraging divedepths. Where possible, we included mean nocturnaland diurnal swimming depths of tagged salmon as separ-ate values, which allowed the meta-analysis to accountfor diel variation in depth distribution. If separate dayand night values were not available, we used the meanswimming depth for all times of day combined.We also summarized scientific test fishery studies(N = 8) that measured or reported information aboutthe vertical distribution of salmon. We only includedstudies that reported catch depth for at least 10 indi-vidual fish per species. Data from all seasons andtimes of day were included to ensure that seasonal anddiel variations were captured in the analysis. For eachsalmon species, we determined the depth ranges overwhich the majority of fish were caught during eachstudy. These species-specific depth ranges were com-pared to the maximum foraging dive depths of taggedresident killer whales to determine if foraging divescorresponded to the depth range of preferred prey(Chinook salmon).All studies included in the meta-analysis (both tag-ging and test fishery) were generally conducted on ma-turing or adult fish (i.e., those ≥ 2 years old). However,in some cases, fish ages were not specified or studiescombined data from juvenile and adult individuals. Wedid not include studies involving only juvenile salmon(first year at sea) because this age group is not con-sumed by resident killer whales [6]. To obtain a suffi-ciently large data set, studies in both coastal and highseas habitats were considered.Wright et al. Movement Ecology  (2017) 5:3 Page 6 of 18ResultsTag deployments and dive identificationDtags were deployed on 34 occasions on 32 differentnorthern resident killer whales (Table 1, Fig. 1). Thetagged whales included 8 adult females (≥12 y), 14 adultmales (≥12 y), and 10 juveniles (3–11 y; 5 females, 2males and 3 of unknown sex). Two individuals, A66 andA83, were tagged twice, although the second deploymenton A83 was too brief to permit analysis (Table 1). Intotal, data from three deployments were not analyzedbecause they had short durations and lacked divesdeeper than the 10 m required for calibration. The 31calibrated tag deployments ranged from 0.3 to 11.8 h induration, yielding a total of 126.1 h of sensor data(Table 1). The MATLAB dive detection filter identified atotal of 11,319 dives (≥1 m).Table 1 Deployments (N = 34) of digital archival tags (Dtags) on 32 northern resident killer whales in British Columbia (2009–2012)Tag ID Deployment date(dd/mm/yyyy)Deployment location Whale ID Sex Age(y)Deploymentduration (h)# divesanalyzedoo09_231a 19/08/2009 50° 46.500 N 127° 24.066 W G52 F 16 7.41 542oo09_234a 22/08/2009 50° 56.870 N 127° 47.920 W A46 M 27 3.92 342oo09_235a 23/08/2009 50° 49.758 N 127° 43.463 W A72 F 10 5.22 486oo09_236a 24/08/2009 50° 51.032 N 127° 31.560 W I45 M 24 2.37 151oo09_237a 25/08/2009 50° 47.670 N 127° 31.891 W I57? F 20 0.07 0oo09_237b 25/08/2009 50° 48.336 N 127° 36.855 W I71 F 16 0.28 0oo09_237c 25/08/2009 50° 49.336 N 127° 41.669 W I83 F 12 1.15 93oo09_237d 25/08/2009 50° 56.672 N 128° 02.190 W I53 M 23 3.28 314oo09_238a 26/08/2009 50° 51.117 N 127° 49.327 W I111 ? 3 11.64 1123oo09_239a 27/08/2009 50° 49.516 N 127° 42.441 W A66 M 13 2.15 145oo09_240a 28/08/2009 50° 56.073 N 127° 41.825 W A37 M 32 3.63 353oo09_243a 31/08/2009 50° 53.767 N 127° 39.881 W I39 M 29 3.11 233oo09_244a 01/09/2009 51° 00.065 N 127° 49.085 W R25 M 22 4.24 299oo09_245a 02/09/2009 50° 47.268 N 127° 32.671 W I46 M 24 5.89 483oo09_245b 02/09/2009 50° 46.975 N 127° 15.357 W I62 M 21 1.52 109oo09_247a 04/09/2009 50° 30.813 N 126° 23.110 W A62 F 15 1.27 157oo10_256a 13/09/2010 50° 57.047 N 127° 44.552 W G64 F 10 7.59 828oo10_260a 17/09/2010 50° 53.982 N 127° 38.038 W A75 F 8 6.97 604oo10_261a 18/09/2010 50° 54.141 N 127° 38.604 W A38 M 39 3.22 291oo10_264a 21/09/2010 51° 03.696 N 127° 58.168 W G39 M 24 1.60 116oo10_265a 22/09/2010 50° 51.936 N 127° 33.151 W G49 F 20 2.92 299oo11_224a 12/08/2011 51° 51.844 N 128° 15.430 W R40 F 10 2.12 215oo11_224b 12/08/2011 51° 23.548 N 128° 08.301 W G32 M 29 0.34 12oo11_240a 28/08/2011 50° 57.018 N 127° 43.853 W I104 F 9 3.95 361oo11_244a 01/09/2011 50° 55.329 N 127° 42.107 W C14 M 26 2.84 175oo11_244b 01/09/2011 51° 00.448 N 127° 58.949 W C24 M 11 1.15 82oo11_245a 02/09/2011 50° 47.917 N 127° 35.362 W I43 M 28 11.80 856oo11_246a 03/09/2011 50° 48.852 N 127° 39.618 W G31 F 30 3.81 466oo11_248a 05/09/2011 50° 49.609 N 127° 42.700 W A83 ? 6 0.48 21oo11_248b 05/09/2011 50° 50.738 N 127° 46.718 W A80 M 7 2.97 298oo11_267a 24/09/2011 50° 40.754 N 127° 03.117 W A34 F 36 7.19 620oo12_232a 19/08/2012 51° 01.358 N 127° 41.391 W I106 ? 8 5.78 751oo12_235a 22/08/2012 50° 55.672 N 127° 42.149 W A83 ? 7 0.07 0oo12_235b 22/08/2012 50° 49.325 N 127° 28.907 W A66 M 16 4.51 494Tag IDs reflect the year (e.g., 09) and Julian day (e.g., 231) of tag deployment. Whale IDs, ages and sexes are from published photographic identification catalogues ofnorthern resident killer whales [35, 37]Wright et al. Movement Ecology  (2017) 5:3 Page 7 of 18Structure of confirmed foraging divesPrey fragments (fish scales and/or flesh) were collected for17 confirmed kills that were made by seven of the taggedindividuals (Table 2). Scale analysis revealed that nine ofthese kills were Chinook salmon, six were chum, and twowere coho. Salmon caught by the tagged whales ranged inage from 2 to 5 y, with the majority (N = 11, 65%) being4–5 y (Table 2). The pseudotracks for the confirmed for-aging dives (with prey samples) revealed a general patternof convoluted, spiraling and kinematically complex pathsduring descents, with relatively abrupt transitions (usuallyat the point of maximum depth) to directional, linear as-cents (Fig. 2). Analysis of tag acoustic records suggestedthat these sudden behavioral transitions likely occurredimmediately following prey captures, which allowed us toestimate capture times and depths for successful kills(Table 2). Often, the estimated capture time correspondedwith a marked increase in flow noise on the Dtag acousticrecord (due to body acceleration) that was followed bycrunching sounds (likely indicative of prey processing). Afew surface chases were also observed; one chum salmoncapture involved only a surface chase (Fig. 2b), whereasfour other captures (2 chum, 2 coho) involved surface pur-suits followed by a deeper dive that resulted in prey cap-ture. One surface-caught Chinook was taken by a taggedwhale (oo12_235b, Table 2) that made a sudden leap atthe surface, without any evidence of a pursuit prior to thecapture event.In all but three of the captures at depth (N = 15), theprobable capture depth corresponded to the maximumdepth attained by the whale during the dive (Table 2).Regardless of the salmon species caught, the majorityof capture depths (82%) were deeper than 100 m(Table 2). Most of the deeper confirmed foraging diveshad V-shaped time-depth profiles (N = 11, Fig. 2a).However, a few were U-shaped (N = 4) with relativelyflat bottom phases accompanied by a sustained bodyroll of approximately 90° (i.e., individuals swimming ontheir sides; Fig. 2c). The bottom phases of U-shaped di-ves also typically contained many tight loops and thewhales’ swim paths were more convoluted on average(mean 2D whole dive straightness index = 0.83 ± 0.13SD, N = 4).Multivariate statistical analysis of kinematic dive variablesLinear discriminant analysis (LDA) of the 11,319 identi-fied dives detected 701 putative foraging dives over 25iterations, including the confirmed foraging dives withprey samples used as the initial training set (N = 15; twosurface captures discounted). The coefficients of theTable 2 Summary of confirmed foraging dives (N = 17) resulting in fish kills by 7 tagged northern resident killer whales over 4 years(2009–2012) of Dtag deploymentsTag ID Whale ID Sex Age (y) Date of kill(dd/mm/yyyy)Capture time(hh:mm:ss)Capture deptha (m) Fish species Fish ageb(European)Fish age (y)oo09_234a A46 M 27 22/08/2009 18:46:35 101.6 Chinook 1.1 3oo09_240a A37 M 32 28/08/2009 13:02:28 165.7 coho x.1 ≥2oo09_240a A37 M 32 28/08/2009 13:29:29 119.4 coho 1.1 3oo10_256a G64 F 10 13/09/2010 16:26:52 134.5 chum 0.4 5oo10_256a G64 F 10 13/09/2010 16:44:18 123.7 * chum 0.4 5oo10_265a G49 F 20 22/09/2010 17:46:02 130.5 chum 0.4 5oo10_265a G49 F 20 22/09/2010 17:53:26 133.7 chum 0.3 4oo11_246a G31 F 30 03/09/2011 13:24:46 201.9 Chinook 0.3 4oo11_246a G31 F 30 03/09/2011 13:39:04 264.8 Chinook 0.3 4oo11_246a G31 F 30 03/09/2011 14:43:15 131.1 Chinook 0.3 4oo11_246a G31 F 30 03/09/2011 14:50:32 204.5 Chinook 0.3 4oo11_246a G31 F 30 03/09/2011 15:05:47 180.7 Chinook 0.3 4oo12_232a I106 ? 8 19/08/2012 15:43:54 0.7 † chum 0.3 4oo12_232a I106 ? 8 19/08/2012 16:51:35 87.6 chum 0.2 3oo12_235b A66 M 16 22/08/2012 14:36:49 102.7 * Chinook 0.1 2oo12_235b A66 M 16 22/08/2012 15:43:56 6.6 * Chinook 0.2 3oo12_235b A66 M 16 22/08/2012 15:57:38 0 † Chinook 0.3 4Capture times were determined using a combination of visual (sudden behavioral transitions in the 3-dimensional TrackPlot reconstructions of foraging dives) andacoustic (marked increases in tag hydrophone flow noise due to body acceleration) evidenceaExcluding the two surface captures (†), all but three foraging dives (*, maximum depths = 141.4, 103.9 and 32.0 m, respectively) had estimated capture depthsthat corresponded to the maximum dive depth, as measured by the Dtag pressure sensorbFish ages are displayed according to the European system, which indicates the number of freshwater and marine annuli (rings) found in the fish scales, separatedby a decimal point. Scales for which the number of annuli could not be determined are denoted by an “x” in place of a numberWright et al. Movement Ecology  (2017) 5:3 Page 8 of 18linear discriminant function indicate the weights ap-plied to each kinematic dive variable (Table 3), andvariables with larger discriminant coefficients (abso-lute values) therefore provided the most separationbetween foraging and non-foraging dive types [45]. Inthe final iteration (25th) of the discriminant function,the variables that best distinguished foraging fromnon-foraging dives were dive duration (min), verticaldescent velocity (m s−1), vertical ascent velocity (m s−1),and the ratio of descent to ascent duration (Table 3). Fol-lowing the LDA, X-means clustering split the remaining10,618 non-foraging dives into two additional types,which we designated as ‘respiration’ (N = 7,050) and‘other’ (N = 3,568).Compared to other dive types, foraging dives identifiedby the LDA (N = 701) were typically deeper (M = 34.0 m,IQR = 71.0 m; Fig. 3) and lasted longer (M = 2.9 min, IQR= 2.4 min; Fig. 3). Foraging whales also swam at greater es-timated speeds (M = 2.1 m s−1, IQR = 1.1 m s−1) than theydid during ‘other’ dives, but displayed no difference inspeed compared to respiration dives (Fig. 4, Table 3). For-aging dive rates of descent (M = 0.7 m s−1, IQR = 0.7 m s−1) and ascent (M = 0.6 m s−1, IQR = 0.8 m s−1), measuredas vertical velocities, were considerably faster than theywere for non-foraging dives (Fig. 4, Table 3). Straightnessindices in both two (whole dive) and three dimensions(descent and ascent phases) for putative foraging dives(M = 0.93–0.95) were marginally lower than those ofrespiration dives (M = 0.99–1.00), indicating that whalemovement paths were more convoluted and less direc-tional (i.e., had higher tortuosity) during foraging (Fig. 5,Table 3). Confirmed foraging dives had even lowerstraightness indices, particularly during the descentphase (M = 0.81). However, median straightness values(M = 0.93–0.97) for other dive behaviors were similar tothose displayed during putative foraging dives (Table 3).Whales engaged in foraging dives also rolled to a greaterextent than during non-foraging dives (Fig. 6). Medians ofboth the mean body roll (M = 21.6°, IQR = 41.3°) and max-imum body roll (M = 132.3°, IQR = 128.4°) values recordedwithin each dive were considerably higher during foragingdives (Table 3). The summary statistics for the LDA for-aging training set (N = 15) indicated an even strongerkinematic differentiation from the non-foraging dive cat-egories (Table 3). The confirmed foraging dives in thetraining set had much greater durations (M = 3.7 min,IQR = 2.4 min), depths (M = 133.7 m, IQR = 61.6 m),mean (M = 65.9°, IQR = 29.2°) and maximum (M = 179.8°,IQR = 0.2°) body roll values, overall swim speeds (M =2.7 m s−1, IQR = 0.5 m s−1), and vertical velocities (des-cent: M = 1.0 m s−1, IQR = 0.7 m s−1, ascent: M = 1.9 m s−1, IQR = 1.0 m s−1), as well as lower straightness indices(M = 0.81–0.90; Table 3).Non-foraging ‘respiration’ dives identified by X-meansclustering were extremely shallow (M = 2.8 m, IQR =1.3 m), comparatively brief in duration (M = 0.3 min,Table 3 Median values (M) of untransformed kinematic dive variables (interquartile ranges, IQR, shown in parentheses) by dive type,recorded for 30 northern resident killer whales carrying Dtags (31 deployments)Dive variable Foraging training set Foraging dives Respiration dives Other dives Coefficientsof lineardiscriminantN = 15 N = 701 N = 7050 N = 3568Dive duration (min) 3.68 (2.35) 2.94 (2.36) 0.33 (0.27) 0.34 (0.29) −1.7075Maximum dive depth (m) 133.67 (61.57) 34.00 (71.02) 2.75 (1.34) 3.15 (1.92) 0.25372D dive straightness index 0.86 (0.26) 0.95 (0.12) 1.00 (0.004) 0.97 (0.05) −0.10453D descent straightness index 0.81 (0.16) 0.94 (0.12) 0.99 (0.02) 0.93 (0.10) −0.10663D ascent straightness index 0.90 (0.09) 0.93 (0.11) 0.99 (0.02) 0.93 (0.07) 0.0362Mean VeDBA 0.17 (0.06) 0.10 (0.07) 0.09 (0.05) 0.11 (0.08) 0.1421Maximum absolute roll (deg) 179.84 (0.15) 132.32 (128.43) 14.12 (20.72) 28.75 (32.24) −0.1204Mean absolute roll (deg) 65.92 (29.17) 21.58 (41.32) 5.50 (7.12) 9.71 (10.55) −0.0770Overall swim speed (m s−1) 2.72 (0.45) 2.08 (1.12) 1.91 (0.90) 1.17 (0.75) 0.0801Descent : ascent duration 1.41 (1.31) 0.85 (0.95) 1.05 (0.53) 1.03 (0.61) −5.1212Vertical descent velocity (m s−1) 0.98 (0.72) 0.66 (0.66) 0.29 (0.15) 0.32 (0.17) −5.9038Vertical ascent velocity (m s−1) 1.93 (1.02) 0.57 (0.76) 0.30 (0.16) 0.33 (0.20) 4.9175Descent Δroll/time (deg s−1) 27.08 (13.47) 8.56 (13.57) 4.93 (5.40) 7.34 (6.85) 0.1164Ascent Δroll/time (deg s−1) 43.52 (26.17) 9.70 (12.55) 4.11 (4.18) 6.45 (5.69) 0.1094Descent Δpointing angle/time (deg s−1) 57.36 (22.44) 22.62 (29.98) 21.50 (20.53) 26.35 (34.65) 0.0352Ascent Δpointing angle/time (deg s−1) 51.64 (23.02) 22.05 (21.65) 11.24 (8.52) 15.37 (12.48) −0.2077Coefficients of the linear discriminant indicate weights applied to each dive variable, with larger absolute values indicating variables that provided greaterseparation between the foraging and non-foraging dive typesWright et al. Movement Ecology  (2017) 5:3 Page 9 of 18Fig. 4 Comparative swim speeds between the three identified divetypes made by 30 tagged northern resident killer whales (F = foraging,R = respiration, O = other behaviors; N = 11,319 total dives from 31 tagdeployments). Whole dive velocity was calculated by dividing the3-dimensional dive path length (determined using dead-reckoning) bythe total dive time, and included both descent and ascent phases.Vertical velocities for descent and ascent phases were based solelyon depth sensor dataFig. 5 Comparative kinematic tortuosity variables between thethree identified dive types made by 30 tagged northern residentkiller whales (F = foraging, R = respiration, O = other behaviors;N = 11,319 total dives from 31 tag deployments). The straightnessindex, indicating relative tortuosity, was calculated in two dimensions(x-y plane only) over entire dives and in three dimensions for thedescent and ascent phases. Lower values of the straightness indexindicate more convoluted paths of whale movement, while valuesapproaching 1 indicate directional, straight-line pathsFig. 6 Comparative maximum and mean body roll (absolute values,in degrees) by 30 tagged northern resident killer whales engaged inthree identified dive types (F = foraging, R = respiration, O = otherbehaviors; N = 11,319 total dives from 31 tag deployments)Fig. 3 Maximum dive depths (m) and dive durations (min) of foraging(N = 701) and non-foraging (N = 10,618) dives by 30 tagged northernresident killer whales (number of deployments = 31). Confirmedforaging dives (N = 17) are marked by coloured data points indicatingthe species of salmon killed (Chinook, coho or chum). Non-foragingdives (gray data points) did not exceed 21 m in depthWright et al. Movement Ecology  (2017) 5:3 Page 10 of 18IQR = 0.3 min), and while only slightly slower than for-aging dives in terms of overall speed (M = 1.9 m s−1,IQR = 0.9 m s−1), they had considerably slower medianvertical descent and ascent velocities (M = 0.3 m s−1,IQR = 0.2 m s−1; Fig. 4, Table 3). Movement withinthese dives was highly directional, with almost no tor-tuosity (M ≥ 0.99 for all 3 straightness index measures,Fig. 5) and limited mean (M = 5.5°, IQR = 7.2°) andmaximum (M = 14.1°, IQR = 20.7°) body roll (Fig. 6,Table 3). The kinematics of this dive type correspondedwell with surface observations of whales submersingthemselves for extremely brief periods between singlebreaths, a movement that occurs repeatedly betweendeeper dives and is present during all activity states(e.g., resting, foraging, travelling and socializing). Whilenot really a true ‘dive’, these surface breathing boutsare conducted for the sole purpose of gas exchangeduring forward propulsion [11], and so we refer tothem throughout as ‘respiration dives’, primarily forconvenience.The second type of non-foraging dive was designatedas ‘other’ because the overall kinematic structure wasintermediate between foraging and respiration dives.Like respiration dives, these dives were comparativelyshallow (M = 3.2 m, IQR = 1.9 m) and short in duration(M = 0.3 min, IQR = 0.3 min). Overall dive speed (M =1.2 m s−1, IQR = 0.8 m s−1), as well as descent and as-cent vertical velocities (M = 0.3 m s−1, IQR = 0.2 m s−1),were almost identical to those of respiration dives andwere slower than during foraging dives (Fig. 4, Table 3).However, ‘other’ dive behaviors had straightness indicesthat were more similar to those of foraging dives andindicated slightly higher path tortuosity (M = 0.93–0.97;Fig. 5, Table 3). During ‘other’ dive behaviors, killerwhales also exhibited a higher level of mean (M = 9.7°,IQR = 10.6°) and maximum (M = 28.8°, IQR = 32.2°)body roll than for respiration dives, although not to thesame extent as during foraging dives (Fig. 6, Table 3).Given the large number of dives in this category (N =3,568) and the intermediate values of many of the kine-matic dive variables (Table 3), it probably includes avariety of other previously described behaviors by resi-dent killer whales, such as socializing, travelling, restingand beach rubbing [5]. X-means clustering was likelyunable to further separate the different behaviors withinthis category because the kinematic predictor variableswere chosen specifically for their expected ability toidentify foraging dives.All three dive types (foraging, respiration, and other)were detected in all but one of the 31 tag deployments;one short deployment (0.34 h) on the adult male G32(Table 1) contained no foraging dives. As a percentageof recorded dives (number of dives, not time-budget) foreach individual, foraging dives made up an average of6.7% (SD = 3.5%), while respiration dives comprised 64.7%(SD = 21.0%) and other dive behaviors 28.6% (SD = 20.0%).The considerably higher occurrence of respiration ‘dives’was likely because they represent bouts of surface breath-ing that are present throughout all killer whale activitystates. Kinematic characteristics of foraging and non-foraging dives detected by the LDA were significantlydifferent (Wilks’ lambda: λ16 = 0.321, p < 0.001), as werethe kinematic characteristics of two non-foraging divetypes (respiration, other) detected by X-means cluster-ing (Wilks’ lambda: λ16 = 0.323, p < 0.001). The majorityof variance in the kinematic dive variables (~68%) canthus be attributed to the grouping factor, meaning thatboth LDA and X-means clustering distinguished divetypes that differed significantly in their kinematic struc-tures. The non-independence of samples (dives), due tothe temporal autocorrelation inherent in time-seriesdata, means that the level of significance implied by theWilks’ lambda P-values is likely somewhat inflated.However, the LDA separated dive types consistently(even when reduced training sets were used duringleave-one-out validations), suggesting that within-group(dive type) variance is much lower than between-groupvariance, and that the dive types can be consistently dif-ferentiated from one another. The leave-one-out valida-tions also confirmed that the LDA was not undulyinfluenced by idiosyncratic variations in foraging divestructure, since all of the omitted individual’s successfulforaging dives (with prey samples) that had been ex-cluded from the training set were reclassified as ‘for-aging’ by the final iteration of each validation test.Some of the kinematic variables used in the LDA didnot distinguish foraging from non-foraging dives as wellas we had expected. Vectorized Dynamic Body Acceler-ation (VeDBA) was very similar between foraging dives,respiration dives, and other dive behaviors (Table 3).Rates of change in both body roll and pointing angle(descents and ascents; degree s−1) tended to be similarbetween foraging dives and other behaviors, but weregenerally lower for respiration dives (Table 3). The ratioof descent to ascent durations was expected to be higher(>1.0) for foraging dives, on the basis that descents in-volving tortuous chase behavior should take longer thandirectional ascents covering the same depth range. Thisvariable also had a higher absolute value for its lineardiscriminant coefficient (Table 3), which implies that itwas relatively important in predicting group membership(i.e., dive type). Although descent to ascent duration wasgreater for the LDA training set of confirmed foragingdives (M = 1.41, IQR = 1.31), it was actually lower (<1.0)for the putative foraging dives (M = 0.85, IQR = 0.95)than for other behaviors (M = 1.03, IQR = 0.61), and res-piration dives (M = 1.05, IQR = 0.53) (Table 3). Overall,however, the IQRs for this variable across the three diveWright et al. Movement Ecology  (2017) 5:3 Page 11 of 18types indicate that the distribution of values for ascent:-descent duration is basically equivalent regardless of divetype.Meta-analysis of pacific salmon vertical distributionThe meta-analysis of ultrasonic telemetry and archivaltagging studies showed that Chinook salmon swim atan average depth of 43.4 m (SD = 15.4 m, Fig. 7) incoastal and offshore Pacific waters [46–49]. In contrast,chum salmon swim at an average depth of 22.0 m (SD= 19.0 m), while coho (x ± SD = 9.4 ± 2.2 m), sockeye(9.4 ± 6.1 m), pink (9.0 ± 3.7 m) and steelhead (4.6 ±3.2 m) are surface-oriented species found at averagedepths of less than 10 m (Fig. 7) [48–57]. The meta-analysis of test fishery studies indicated similar patternsof vertical distribution, with most Chinook beingcaught below 30 m (range = 15–100 m) [49, 58–62],while all other salmon species tended to be caught atdepths shallower than 30 m (range = 0–45.5 m) (Fig. 8)[58–64]. The maximum foraging dive depths of taggednorthern resident killer whales (M = 34.0 m, IQR =71.0 m, N = 701) overlapped considerably (Figs. 7 & 8)with the average vertical distribution of Chinook, butdid not correspond well to the swimming depths ofother salmon species.DiscussionAnalysis of dives by northern resident killer whales re-vealed that dive depth, path tortuosity, body rotation andestimates of velocity are reliable metrics for distinguishingforaging from non-foraging behavior. Most notably, Dtag-recorded kinematics showed that foraging dives by north-ern residents attained and often exceeded the expecteddepth distribution of Chinook salmon, their preferredprey. Analyzing whale movement patterns during preypursuit also revealed several strategies that salmon mayuse to escape air-breathing predators.Kinematic structure of foraging divesForaging dives by resident killer whales were characterizedby greater maximum depths and dive durations, moreconvoluted dive paths, higher levels of body rotation, andincreased swimming speeds. The median maximum depth(133.7 m, IQR = 61.6 m) of confirmed foraging dives(training set, N = 15) in our study corresponded with theaverage maximum depth (calculated per tag) of 140.8(±61.8 SD) m reported by Baird et al. [11] for southernresident killer whales. The variability in maximum divedepths was also remarkable similar between the two stud-ies. Median durations for both LDA-identified (2.9 min,IQR = 2.4) and confirmed (3.7 min, IQR = 2.4) foragingdives from the Dtag data were marginally greater than themean daytime dive durations measured by TDRs deployedFig. 7 Maximum depths (m) of foraging dives (N = 701) by 30tagged northern resident killer whales (grey box plot) and overallmean ocean swimming depths (white box plots) of six species ofPacific salmon, as reported in tagging and ultrasonic telemetrystudies (N = 12) of maturing or adult fish (≥2 years) in summeror autumnFig. 8 Catch depths (m) of six species of Pacific salmon taken by troll,gillnet or trawl fishing (dark grey boxes), and maximum foraging divedepths (1st–3rd quartiles, shaded band) of 30 tagged northern residentkiller whales. The range of maximum foraging depths shown herespans the interval between the 25th and 75th percentiles (22.2–93.2 m)of all LDA-detected foraging dives (N = 701), and the fishery catchdepths are from salmon vertical distribution and bycatch studies. Foreach species of salmon, individual boxes represent separate studies(some studies appear more than once if conducted on multiplespecies). Dashed lines indicate the total depth interval (m) fished, anddark grey boxes represent the depth interval (m) in which the largestpercentage of fish was caught during each study. Catch data are fromall seasons and times of day, taken in both coastal and high seashabitats (N = 8 studies, minimum of 10 individual fish/species/study)Wright et al. Movement Ecology  (2017) 5:3 Page 12 of 18on southern residents (2.8 ± 0.5 for adult males, 2.1 ± 0.6for adult females) [11]. However, the study by Baird et al.[11] pooled all dives ≥1 min together regardless of activitystate, which may help to explain this difference.The increased roll and greater tortuosity displayedduring the descent phase of foraging dives by residentkiller whales (Fig. 2a) may serve to facilitate acousticsearching. Odontocetes have narrow, conically-shapedsonar beams that allow them to effectively discrimin-ate the size and distance of detected targets [65].However, to initially locate prey, an area much largerthan the whale’s beam width must be scanned. Rollingbehavior may increase the area covered by the sonarbeam and thus improve the likelihood of detectingprey. Akamatsu et al. [65] found that finless porpoises(Neophocaena phocaenoides) rolled extensively duringdives with greater acoustic search effort and DeRuiteret al. [66] found that in captive harbor porpoises(Phocoena phocoena), both click rate and variance inroll angle increased around the time of fish capture.Similar increases in rolling behavior were prevalent inthe foraging dives made by the tagged resident killerwhales in our study (Table 3, Fig. 6), and could servethe same purpose. Increased body rotation may there-fore be a useful metric for identifying foraging behav-ior in future studies.Sustained off-axis body roll positions performed byhunting killer whales during U-shaped dives may alsoimprove maneuverability and swimming performanceduring fish pursuits along the sea floor. Cetaceans gener-ate hydrodynamic thrust for swimming by moving theposterior third of their bodies and tail flukes dorso-ventrally [67]; the average vertical amplitude of fluke tipmovement during swimming by killer whales is greaterthan 20% of their body length [68, 69]. Tail stroke ampli-tude would therefore be restricted when moving closelyalong the sea floor in an upright position. In our study,tagged animals often rotated their bodies approximately90° to the right or left during the level bottom phases offoraging dives with U-shaped profiles (Fig. 2c). If whaleswere chasing salmon along the sea floor, as this diveshape implies, then turning sideways would allow unre-stricted fluke movement and ensure that high swimmingspeeds were achieved. Although whales swimming side-ways along the bottom may lose the additional thrustand propulsive efficiency generated by ground effect[70], the benefits of unimpeded fluke movement likelyoutweigh this cost, because the flukes must be very close(within one span length) to the sea floor for ground ef-fect to be of consequence [71].In addition to depth and tortuosity, estimated swim-ming velocity was also an effective way to identify for-aging dives. Based on dead-reckoned tracks, the medianestimated speed of foraging whales in our study was2.1 m s−1 (N = 701; Table 3, Fig. 4). However, comparedto all foraging dives combined, dives resulting in suc-cessful kills (N = 15) were slightly faster (M = 2.7 m s−1)and several foraging dives had estimated speeds exceed-ing 4.0 m s−1 (max = 6.7 m s−1; Fig. 4). The somewhatslower median speed of LDA-identified foraging dives,relative to those dives resulting in confirmed kills, islikely due to this dive category also containing unsuc-cessful, aborted chases (Table 3). Using theodolite tech-niques, Williams & Noren [72] estimated maximumswimming speeds for adult resident killer whales of2.7 m s−1 (females) and 3.0 m s−1 (males), which issimilar to the median speed of 2.7 m s−1 for our con-firmed foraging dives. Estimated speeds for the fastestforaging dives (>4.0 m s−1) recorded in our study arelikewise comparable to the average maximum velocityof 5.98 m s−1 recorded by Fish [68] for captive killerwhales performing turning maneuvers. They are alsosimilar to the theoretical sustainable aerobic swimmingspeed of 4.7-5.6 m s−1 estimated for adult killer whalesby Guinet et al. [73]. Mean sustained swimming speedfor killer whales chasing bluefin tuna (Thunnus thyn-nus) in the Strait of Gibraltar was 3.7 ± 0.2 m s−1 [73],which unsurprisingly is slightly faster than the medianspeed of northern resident killer whales in our study,which were hunting moderately slower-moving Pacificsalmon. Roos et al. [74] estimated a mean swimmingspeed of 1.89 ± 0.61 m s−1 (range = 0.69–4.05 m s−1) forNorwegian herring-feeding killer whales based on lowfrequency flow noise from Dtag hydrophone recordings.This mean swimming speed for the Norwegian whalesis comparable to our median foraging dive speed fornorthern residents (2.1 m s−1), while the maximumspeed is similar to the maximum speeds of the fastestforaging dives in our Dtag dataset.Estimated swimming speeds were calculated usingdive path lengths that relied on dead-reckoning, andtherefore contained cumulative error that may have ledto the over- or underestimation of distances travelled [34],thus impacting speed calculations. However, we mini-mized such errors by correcting track placement usingperiodic GPS surfacing locations, and by constraining ouranalysis to comparisons of kinematic metrics summarizedover very brief time periods (i.e., dive-by-dive). Swimspeed estimates from our Dtag tracks concurred withswimming speeds from other studies of fish-eating killerwhales that were obtained using different methodologies,and therefore we are reasonably confident of their accur-acy. Unlike estimated overall swimming speed, verticalvelocities likely contained minimal error because theywere calculated directly from depth sensor measurements;however, these values greatly underestimate true swim-ming speeds because movement in the horizontal planewas not considered. Regardless, both overall swimmingWright et al. Movement Ecology  (2017) 5:3 Page 13 of 18speed and median vertical velocity were much higher forforaging dives than for any other dive type (Table 3). Ver-tical velocities were equally high for both the descent andascent phases of foraging dives (Table 3)–due to the pur-suit of rapidly fleeing prey during descents, and the needto return to the surface quickly during ascents to replenishoxygen stores depleted at depth. Hydrophone recordsfrom ascent phases of confirmed foraging dives typicallycontained pulses of flow noise caused by fluke strokes asthe animal ascended, which substantiated depth sensor-detected increases in vertical velocity.Foraging dive depth selectivity and the verticaldistribution of preferred preyWe determined that the maximum depths of foraging di-ves by northern resident killer whales overlapped with theaverage swimming depth of Chinook salmon tracked dur-ing tagging studies (Fig. 7), as well as with test fisherycatch depths (15–100 m) for Chinook (Fig. 8). Conversely,there was almost no correspondence between maximumdive depths of foraging whales and the average swimmingdepths of other salmon species (Fig. 7), except for chum,which is the second most commonly consumed prey spe-cies by northern residents [6]. This overlap between max-imum foraging dive depths (i.e., estimated fish capturedepths) and the vertical distributions of Chinook andchum salmon suggests that resident killer whales areintentionally diving to depths where preferred prey ismore likely to occur. Although the vertical distribution ofsalmon changes on seasonal and diel scales, and is affectedby many physiological and ecological factors [46, 49, 75],tagging and fisheries studies consistently indicated thatChinook salmon are located deeper in the water columnthan other salmonids. This means that when Chinook sal-mon abundance is low, killer whales may continue to diveto the deeper depths used by their preferred prey, butwould experience low encounter rates and poor energeticreturns.Although the foraging dive depths of killer whalesoverlapped with the vertical distribution of Chinook sal-mon, whales also extended their foraging dives to muchgreater depths of up to 379 m (Figs. 3 & 7). Chinookhave been intercepted as bycatch by trawlers at depthsof 325 [49] and 482 m [76]. Ultrasonic tracking has alsoshown that Chinook salmon swim to depths of 300–400 m, and that fish performing deep dives (>200 m) aresignificantly larger (x = 87.2 cm) than those remaining atshallower depths (x = 77.3 cm) [46]. These deep-divingindividuals correspond in length to 4–5 y Chinook [77],which are the size classes most frequently consumed byresident killer whales [6]. This suggests that whales maydive beyond the average swimming depth of most Chi-nook to increase their chance of locating the larger andmore energetically profitable 4–5 y old fish.Predation avoidance strategies of pacific salmonKiller whale foraging dives would also be expected toexceed the typical swimming depth of Chinook if salmonswim towards the sea floor as an escape response. Themaximum depths of successful foraging dives weregreater (Table 3) than the average depth that Chinookare found at (Figs. 7 & 8), implying that whales maychase Chinook to greater depths before catching them.Furthermore, tortuous dive paths that resembled chasesoccurred primarily during the descent phase for 12 ofthe 15 successful foraging dives (Table 2), with the cap-ture point corresponding to the maximum depth of thedive (e.g., Fig. 2a). This occurred regardless of the spe-cies of salmon caught, suggesting that rapid descentsmay be a general predator avoidance response of manyPacific salmon. The other three confirmed foraging diveshad capture points that did not correspond to maximumdive depths (Table 2). Of these, two were U-shaped divesthat had the same high track tortuosity evident in thedescent phases of the other 12 dives (Fig. 2c). The levelbottom phases of these two dives suggest the whaleswere chasing fish along the sea floor and the fish couldnot escape to deeper water, which could explain why theestimated capture points for these dives did not corres-pond to the maximum dive depths.Most of the confirmed kills of chum and coho (N = 7out of 8), which are normally shallow-swimming species(Fig. 7), had estimated capture depths exceeding 80 m(Table 2, Fig. 3). While shallow water chases precededthe captures of coho and chum for five of these eight(63%) successful foraging dives, only one chum captureactually occurred near the surface (Table 2). This also indi-cates that rapid descent is likely a common escape responsein all species of Pacific salmon. Foraging whales may haveopportunistically encountered chum and coho close to thesurface, where they are typically found, and subsequentlypursued them to greater depths before making a successfulcapture. An underwater video (Ellis GM, Towers JR, FordJKB: 'A' pod juveniles with chum salmon, unpublishedmedia) further supports the hypothesis that surface-oriented salmon species will dive when threatened by apredator. This video shows two young whales echolocatingon a chum salmon, which then dives towards the bottomafter one of the whales bites its caudal fin. Tagging studiesof Pacific salmon provide additional evidence that divingsteeply could represent an escape response, as fish oftenperformed very deep dives immediately after post-taggingrelease [46, 52, 55–57, 78–80]. In addition, tagged chum sal-mon dove to the sea floor in 12 of 16 encounters with Dall’sporpoises (Phocoenoides dalli), a potential predator [81].Although fleeing is energetically costly [82], swimmingdownward may be an effective strategy for fish to escapean air-breathing predator, such as a killer whale. Thelikelihood that a pursuing whale would have to return toWright et al. Movement Ecology  (2017) 5:3 Page 14 of 18the surface to breathe before intercepting its preywould increase with greater dive depths (i.e., longerpursuit times). Salmon may also descend to avoid pre-senting the visual target of a dark body silhouettedagainst light coming from the surface [83]. Lastly,swimming to the sea floor could allow fish to use rockycrevices and other bathymetric features as refuges fromlarge predators.Since Pacific salmon appear to descend rapidly as anescape response, determining depths at which chasesare initiated would provide a better estimate than themaximum foraging dive depths we used here (Fig. 7)for identifying the depth ranges targeted by whalessearching for prey. This would involve determiningthe depth of the transition between the search andpursuit phase of each foraging dive, which could beaccomplished by combining kinematic and acousticanalyses. The beginning of a chase is likely indicatedby increases in swimming velocity, dive depth, pathtortuosity, and the production rate of echolocationclicks. Kinematic analysis of foraging dive behaviorwould therefore benefit from knowledge about howclose a killer whale must be to a fish before it is ener-getically worth pursuing [84], as well as the thresholddistance at which Pacific salmon are capable of detect-ing large predators. Echolocating killer whales canprobably detect fish at depth during surface transitsand initiate a foraging dive in response, as they arecapable of sensing Chinook salmon at distances of upto 100 m in quiet conditions [84]. Construction ofechograms from Dtag acoustic records to determinetarget ranges [85] and how these correspond to for-aging movements would therefore also aid in identify-ing the sections of the water column important tokiller whales for the search and initial pursuit phasesof hunting.The tortuous and non-linear swim paths exhibited byforaging killer whales (Fig. 2) support past field obser-vations which indicated that, in addition to performingsteep dives, salmon also attempt to avoid capture byunpredictably altering their swimming trajectories [86].The smaller body size of salmon relative to that of killerwhales allows them to execute tighter turning angles atfaster rates, making them more maneuverable than theirlarger predators [68, 82, 87]. Evasive movements increasethe probability of escape by taking the fish out of the dir-ect pursuit path of the predator [82]. To intercept erratic-ally swimming prey (and maintain knowledge of preylocation using highly-directional echolocation clicks),killer whales must match the convoluted flight path of thefish. In our study, the response of tagged whales to thepresumed evasive maneuvers of salmon resulted in lowerstraightness indices for foraging dives compared to respir-ation dives (Fig. 5, Table 3). However, mean straightnessindices for foraging dives were similar to those of the‘other’ dive category (Fig. 5, Table 3), indicating that mea-sures of tortuosity alone may be insufficient to distinguishforaging from non-foraging behavior.Rate of change in pointing angle was also expected tobe noticeably higher during foraging dives compared tonon-foraging dives (particularly for descents, where themajority of chasing behavior occurred), since the orien-tation of the whale’s longitudinal axis should changemore rapidly in response to prey maneuvers. Althoughthe rate of change in pointing angle for the training setof confirmed foraging dives (N = 15) was noticeablyhigher for both descents and ascents, this kinematicvariable was roughly comparable across the three divecategories identified by the LDA (Table 3). This impliesthat other behaviors (e.g., socializing or beach-rubbing)also involve rapid orientation changes, which is sup-ported by surface observations of resident killer whales.As expected, the change in pointing angle over timewas generally higher for descents (chasing) than ascents(transiting to the surface); however, this was true for alldive types, not just foraging (Table 3). This finding im-plies that whales ascending from dives are likely return-ing to the surface using the most direct routes,probably because the need to replenish oxygen and off-load carbon dioxide take precedence over other activ-ities near the end of a dive.Higher swimming speeds exhibited by killer whalesduring foraging dives likely arose as a response to in-creased swimming speeds of fleeing prey. Unfortu-nately, very few studies have directly measured themaximum or burst swimming speeds of adult Pacificsalmon in saltwater: the measure of swimming per-formance most relevant to avoiding predators [88].Data logger measurements from a wild adult (4 wintersat sea; European age 0.4) chum salmon in the Bering Seameasured a maximum speed of 2.8 m s−1 [53], which iscomparable to the median speed recorded for successfulkiller whale foraging dives in our study (M = 2.7 m s−1,N = 15; Table 3). Randall et al. [89] determined themean burst critical swimming speed (Ucrit [90];) ofChinook salmon in saltwater to be 2.32 body lengths s−1,or about 0.731 m s−1 (sustained for 30–60 s). However,fish in that study were previously fatigued prior to de-termining burst Ucrit and were relatively small (meanfork length = 31.5 cm). It is probable that the larger sizeclasses of Chinook typically consumed by resident killerwhales (mean fork lengths = 80.8–93.4 cm) [6] are cap-able of swimming much faster than this when pursuedby whales. Killer whale maximum swim speeds were ex-pected to approximate or marginally surpass those ofPacific salmon, since they are unlikely to expend add-itional energy by swimming faster than is required forprey capture.Wright et al. Movement Ecology  (2017) 5:3 Page 15 of 18ConclusionsUsing high-resolution accelerometer tags, we providethe first quantitative description of fine-scale foragingmovements by resident killer whales hunting Pacific sal-mon. Increased dive depth, tortuosity, body roll, and es-timated swimming velocity were determined to be themost useful kinematic measures for distinguishing for-aging from other dive behaviors. Reconstructed divepaths indicated that foraging dives targeted the expecteddepth distribution of Chinook salmon, the preferred preyof resident killer whales, and whale movements duringprey pursuit also revealed probable escape strategiesused by salmon to avoid capture (rapid descent, evasivemaneuvering, and increased swimming speeds). Futurestudies could build on our findings by using Dtag re-cords to assess space use and energy expenditure bykiller whales during different activity states. Our recon-structed Dtag tracks and the kinematic characteristics offoraging dives we have identified also provide a com-parative baseline for evaluating the impacts of variousanthropogenic disturbances on resident killer whale for-aging behavior.Additional fileAdditional file 1: Appendix A1. Methods: Calculation of Kinematic DiveVariables. (DOCX 16 kb)AbbreviationsGPS: Global positioning system; IQR: Inter-quartile range; LDA: Lineardiscriminant analysis; M: Median; ODBA: Overall dynamic body acceleration;TDR: Time-depth recorder; VeDBA: Vectorized dynamic body acceleration;VHF: Very high frequencyAcknowledgementsWe thank M. deRoos for his crucial role deploying the Dtags and collectingthe prey samples, and J. Towers for his assistance in the field. We are gratefulfor the logistical support we received from A. Ceschi, B. Weeks, and the staffof God’s Pocket Resort, B. Falconer and the crew of the S/V Achiever, and J.and M. Borrowman of Orcella Expeditions. We appreciate the assistanceprovided by E. Stredulinsky and E. Zwamborn in transcribing the digitallyrecorded audio field notes. Thanks to C. Ware for guidance in the use ofTrackPlot visualization software and R. Joy for advising us on the applicationof multivariate statistics for categorizing dive behavior data. T. Hurst and A.Bocconcelli of Woods Hole Oceanographic Institution provided support forthe calibration and archiving of Dtag data. We also thank S. DeRuiter, A.Edwards, and P. Miller for their suggestions regarding methodologicalapproaches for analysis of accelerometry data. We are grateful to the staff ofthe Fish Ageing Lab and the Molecular Genetics Lab at the Pacific BiologicalStation, Fisheries and Oceans Canada, for analyzing the salmon scales andtissue fragments collected from killer whale foraging dives. Earlier versions ofthis manuscript benefited from helpful comments and editing by J. Watson.FundingThis work was supported by the Species at Risk Program, Fisheries andOceans Canada; the University of Cumbria’s Research and ScholarshipDevelopment Fund; a Marie Curie Intra-European Fellowship (IEF) to VD; aUniversity of British Columbia Zoology Graduate Fellowship to BW; and aNatural Sciences and Engineering Research Council (NSERC) Alexander Gra-ham Bell Canada Graduate Scholarship to BW.Availability of data and materialsThe datasets generated and analyzed during the current study are availablefrom John K.B. Ford (john.k.ford@dfo-mpo.gc.ca) on reasonable request.Authors’ contributionsJF, VD and GE conceived the idea for the study and BW, GE, VD and AScollected the data. Methodology for tag data analysis was developed byBW, BB, AT and JF. Dtag track reconstruction and analysis was conductedby BW with support from BB and AS. BW interpreted the results withcontributions from BB, VD, GE, AT and JF. BW wrote the manuscript andAT and JF provided detailed revisions of its content. All authors read andapproved the final manuscript.Competing interestsThe authors declare that they have no competing interests.Consent for publicationNot applicable.Ethics approval and consent to participateOur study was conducted under University of British Columbia Animal CarePermit no. A11-0140 and Fisheries and Oceans Canada Marine MammalResearch License no. MML-001. All field procedures were approved by theUniversity of British Columbia Animal Care Committee (ACC) and thePacific Region Animal Care Committee of Fisheries and Oceans Canada(Pacific Biological Station), and complied with the standards of animal careset by the Canadian Council on Animal Care.Author details1Marine Mammal Research Unit, Institute for the Oceans and Fisheries,University of British Columbia, AERL Building, Room 247 - 2202 Main Mall,Vancouver, BC V6T 1Z4, Canada. 2Department of Zoology, University ofBritish Columbia, #4200 - 6270 University Blvd., Vancouver, BC V6T 1Z4,Canada. 3Pacific Biological Station, Fisheries and Oceans Canada, 3190Hammond Bay Road, Nanaimo, BC V9T 1K6, Canada. 4Centre for WildlifeConservation, University of Cumbria, Rydal Road, Ambleside, Cumbria L229BB, UK. 5Woods Hole Oceanographic Institution, 266 Woods Hole Road,Woods Hole, MA 02543-1050, USA.Received: 20 October 2016 Accepted: 30 January 2017References1. Nowacek DP, Christiansen FC, Bejder L, Goldbogen JA, Friedlaender AS.Studying cetacean behaviour: new technological approaches andconservation applications. Anim Behav. 2016;210:235–44.2. Pyke GH. Optimal foraging theory: A critical review. Annu Rev Ecol Syst.1984;15:523–75.3. Cooke SJ, Hinch SG, Wikelski M, Andrews RD, Kuchel LJ, Wolcott TG, et al.Biotelemetry–a mechanistic approach to ecology. Trends Ecol Evol.2004;19:334–43.4. Mann J. Behavioural sampling methods for cetaceans: A review and critique.Mar Mamm Sci. 1999;15(1):102–22.5. Ford JKB. Acoustic behaviour of resident killer whales (Orcinus orca) offVancouver Island, British Columbia. Can J Zool. 1989;67:727–45.6. Ford JKB, Ellis GM. Selective foraging by fish-eating killer whales Orcinusorca in British Columbia. Mar Ecol Prog Ser. 2006;316:185–99.7. Heimlich-Boran JR. Behavioral ecology of killer whales (Orcinus orca) in thePacific Northwest. Can J Zool. 1988;66(3):565–78.8. Ford JKB, Ellis GM, Barrett-Lennard LG, Morton AB, Palm RS, Balcomb KC.Dietary specialization in two sympatric populations of killer whales(Orcinus orca) in coastal British Columbia and adjacent waters. Can J Zool.1998;76(8):1456–71.9. Wright BM, Stredulinsky EH, Ellis GM, Ford JKB. Kin-directed food sharingpromotes lifetime natal philopatry of both sexes in a population of fish-eating killer whales, Orcinus orca. Anim Behav. 2016;115:81–95.10. Baird RW, Hanson MB, Ashe EE, Heithaus MR, Marshall GJ. Studies of foraging in“southern resident” killer whales during July 2002: Dive depths, bursts in speed,and the use of “Crittercam” system for examining sub-surface behaviour.Report prepared for the National Marine Fisheries Service, National MarineMammal Laboratory; 2003. p. 17.Wright et al. Movement Ecology  (2017) 5:3 Page 16 of 1811. Baird RW, Hanson MB, Dill LM. Factors influencing the diving behaviour offish-eating killer whales: sex differences and diel and interannual variation indiving rates. Can J Zool. 2005;83:257–67.12. Groot C, Margolis L, editors. Pacific salmon life histories. Vancouver: UBCPress; 1991.13. Riddell B. Pacific salmon resources in central and north coast BritishColumbia. Vancouver: Pacific Fisheries Resource Conservation Council; 2004.14. Ford JKB, Wright BM, Ellis GM, Candy JR. Chinook salmon predation byresident killer whales: seasonal and regional selectivity, stock identity ofprey, and consumption rates. Can Sci Adv Sec Res Doc. 2009;101:iv + 43.15. Ford JKB, Ellis GM, Olesiuk PF, Balcomb KC. Linking killer whale survival andprey abundance: food limitation in the oceans’ apex predator? Biol Lett.2010;6(1):139–42.16. Foster E, Franks DW, Morrell LJ, Balcomb KC, Parsons KM, van Ginneken A,Croft DP. Social network correlates of food availability in an endangeredpopulation of killer whales, Orcinus orca. Anim Behav. 2012;83(3):731–6.17. Parsons KM, Balcomb KC, Ford JKB, Durban JW. The social dynamics ofsouthern resident killer whales and conservation implications for thisendangered population. Anim Behav. 2009;77(4):963–71.18. O’Neill SM, Ylitalo GM, West JE. Energy content of Pacific salmon as preyof northern and southern resident killer whales. Endanger Species Res.2014;25:265–81.19. Stansby ME. Chemical characteristics of fish caught in the northeast PacificOcean. Mar Fish Rev. 1976;38:1–11.20. Healey MC. Coastwide distribution and ocean migration patterns of stream-and ocean-type Chinook salmon, Oncorhynchus tshawytscha. Can Field Nat.1983;97:427–33.21. Fisheries and Oceans Canada. Recovery strategy for the northern andsouthern resident killer whales (Orcinus orca) in Canada. Species at Risk ActRecovery Strategy Series. Ottawa: Fisheries and Oceans Canada; 2011.22. Johnson MP, Tyack PL. A digital acoustic recording tag for measuring theresponse of wild marine mammals to sound. IEEE J Ocean Eng. 2003;28(1):3–12.23. Towers JR, Ellis GM, Ford JKB. Photo-identification catalogue and status ofthe northern resident killer whale population in 2014. Can Tech Rep FishAquat Sci. 2015;3139:iv + 75.24. Ford JKB, Ellis GM, Balcomb KC. Killer whales: the natural history andgenealogy of Orcinus orca in the waters of British Columbia andWashington. 2nd ed. Vancouver: UBC Press; 2000.25. Ellis GM, Towers JR, Ford JKB. Northern resident killer whales of BritishColumbia: Photo-identification catalogue and population status to 2010.Can Tech Rep Fish Aquat Sci. 2011;2942:vi + 71.26. Bigg MA. An assessment of killer whale (Orcinus orca) stocks off VancouverIsland, British Columbia. Rep Int Whaling Commn. 1982;32:655–66.27. Altmann J. Observational study of behavior: Sampling methods. Behaviour.1974;49(3):227–67.28. Levy R, Uminsky D, Park A, Calambokidis J. A theory for the hydrodynamicorigin of whale flukeprints. Int J Nonlinear Mech. 2011;46(4):616–26.29. Withler RE, Candy JR, Beacham TD, Miller KM. Forensic DNA analysis ofPacific salmonid samples for species and stock identification. Environ BiolFishes. 2004;69:275–85.30. MacLellan SE. Guide for sampling structures used in age determination ofPacific salmon. Nanaimo: Fisheries and Oceans Canada; 2004.31. The Mathworks, Inc. MATLAB, version 7.8.0 (R2009a). 2009. Natick, MA.32. Ware C, Arsenault R, Plumlee M, Wiley D. Visualizing the underwaterbehavior of humpback whales. IEEE Comput Graph Appl. 2006;26(4):14–8.33. Battaile B. TrackReconstruction: Reconstruct animal tracks frommagnetometer, accelerometer, depth, and optional speed data (R package),version 1.1. 2014. http://cran.r-project.org/web/packages/TrackReconstruction/index.html.34. Wilson RP, Liebsch N, Davies IM, Quintana F, Weimerskirch H, Storch S, et al.All at sea with animal tracks; methodological and analytical solutions for theresolution of movement. Deep-Sea Res II. 2007;54:193–210.35. Bidder OR, Walker JS, Jones MW, Holton MD, Urge P, Scantlebury DM, et al.Step by step: reconstruction of terrestrial animal movement paths by dead-reckoning. Mov Ecol. 2015;3:23.36. Wilson RP, Shepard ELC, Liebsch N. Prying into the intimate details ofanimal lives: use of a daily diary on animals. Endanger Species Res.2008;4:123–37.37. Brown DD, Kays R, Wikelski M, Wilson R, Klimley AP. Observing theunwatchable through acceleration logging of animal behavior. AnimBiotelem. 2013;1:20.38. Johnson M, de Soto NA, Madsen PT. Studying the behaviour and sensoryecology of marine mammals using acoustic recording tags: a review. MarEcol Prog Ser. 2009;395:55–73.39. Bidder OR, Soresina M, Shepard ELC, Halsey LG, Quintana F, Gomez-Laich A,et al. The need for speed: testing acceleration for estimating animal travelrates in terrestrial dead-reckoned systems. Zoology. 2012;115:58–64.40. Environmental Systems Research Institute. ArcGIS Desktop: EducationEdition, version 10.2. Redlands: ESRI; 2013.41. Warton DI, Hui FK. The arcsine is asinine: the analysis of proportions inecology. Ecology. 2011;92(1):3–10.42. Mundry R, Sommer C. Discriminant function analysis with nonindependentdata: consequences and an alternative. Anim Behav. 2007;74:965–76.43. Ishioka T. Extended K-means with an efficient estimation of the number ofclusters. Lect Notes Comput Sci. 2000;1983:17–22.44. Ishioka T. An expansion of X-means for automatically determining theoptimal number of clusters: Progressive iterations of k-means and mergingof clusters. Proc 4th IASTED Int Conf Comp Intell. 2005;91–96.45. Tabachnick BG, Fidell LS. Using multivariate statistics. 6th ed. Boston:Pearson Education; 2013.46. Candy JR, Quinn TP. Behavior of adult chinook salmon (Oncorhynchustshawytscha) in British Columbia coastal waters determined from ultrasonictelemetry. Can J Zool. 1999;77(7):1161–9.47. Hinke JT, Watters GM, Boehlert GW, Zedonis P. Ocean habitat use inautumn by Chinook salmon in coastal waters of Oregon and California. MarEcol Prog Ser. 2005;285:181–92.48. Ogura M. Migratory behavior of Pacific salmon (Oncorhynchus spp.) in theopen sea. Bull Nat Res Inst Far Seas Fish. 1994;31:1–139.49. Walker RV, Sviridov VV, Urawa S, Azumaya T. Spatio-temporal variation invertical distributions of Pacific salmon in the ocean. N Pac Anadr FishComm Bull. 2007;4:193–201.50. Azumaya T, Ishida Y. Mechanism of body cavity temperature regulation ofchum salmon (Oncorhynchus keta) during homing migration in the NorthPacific Ocean. Fish Oceanogr. 2005;14(2):81–96.51. Ishida Y, Yano A, Ban M, Ogura M. Vertical movement of a chum salmonOncorhynchus keta in the western North Pacific Ocean as determined by adepth-recording archival tag. Fish Sci. 2001;67:1030–5.52. Soeda H, Yoza K, Shimamura T, Hasegawa E. On the swimming behavior ofchum salmon in early migratory season off the coast of Hokkaido, OkhotsuSea. Nippon Suisan Gakkaishi. 1987;53(1):1827–33.53. Tanaka H, Naito Y, Davis ND, Urawa S, Ueda H, Fukuwaka M. First record ofthe at-sea swimming speed of a Pacific salmon during its oceanicmigration. Mar Ecol Prog Ser. 2005;291:307–12.54. Tanaka H, Takagi Y, Naito Y. Swimming speeds and buoyancy compensationof migrating adult chum salmon Oncorhynchus keta revealed by speed/depth/acceleration data. J Exp Biol. 2001;204:3895–904.55. Yano K, Nakamura A. Observations on the effect of visual and olfactoryablation on the swimming behavior of migrating adult chum salmon,Oncorhynchus keta. Jpn J Ichthyol. 1992;39(1):67–83.56. Quinn TP, Terhart BA, Groot C. Migratory orientation and verticalmovements of homing adult sockeye salmon, Oncorhynchus nerka, incoastal waters. Anim Behav. 1989;37:587–99.57. Ruggerone GT, Quinn TP, McGregor IA, Wilkinson TD. Horizontal and verticalmovement of adult steelhead trout, Oncorhynchus mykiss, in the Dean andFisher Channels, British Columbia. Can J Fish Aquat Sci. 1990;47:1963–9.58. Argue AW. A study of factors affecting exploitation of Pacific salmon in theCanadian gantlet fishery of Juan de Fuca Strait. Can Fish Serv Pac Reg TechRep. 1970;11:259.59. Godfrey H, Henry KA, Machidori S. Distribution and abundance of cohosalmon in offshore waters of the North Pacific Ocean. Intl N Pac Fish CommBull. 1975;31:80.60. Milne DJ. Selectivity of trolling lures. Fish Res Board Can Prog Rep Pac CoastStations. 1955;103:3–5.61. Parker RR, Black EC, Larkin PA. Fatigue and mortality in troll-caught Pacificsalmon (Oncorhynchus). J Fish Res Board Can. 1959;16(4):429–48.62. Taylor FHC. The British Columbia offshore herring survey, 1968–1969. FishRes Board Can Tech Rep. 1969;140:54.63. Machidori S. Vertical distribution of salmon (genus Oncorhynchus) in thenorthwestern Pacific. Bull Hokkaido Region Fish Res Lab. 1966;31:11–7.64. Manzer JI. Preliminary observations on the vertical distribution of Pacificsalmon (Genus Oncorhynchus) in the Gulf of Alaska. J Fish Res Board Can.1964;21:891–903.Wright et al. Movement Ecology  (2017) 5:3 Page 17 of 1865. Akamatsu T, Wang D, Wang K, Li S, Dong S. Scanning sonar of rollingporpoises during prey capture dives. J Exp Biol. 2010;213(1):146–52.66. DeRuiter SL, Bahr A, Blanchet M, Hansen SF, Kristensen JH, Madsen PT, TyackPL, Wahlberg M. Acoustic behaviour of echolocating porpoises during preycapture. J Exp Biol. 2009;212(19):3100–7.67. Fish FE, Nusbaum MK, Beneski JT, Ketten DR. Passive cambering and flexiblepropulsors: cetacean flukes. Bioinspir Biomim. 2006;1(4):S42–8.68. Fish FE. Balancing requirements for stability and maneuverability incetaceans. Integr Comp Biol. 2002;42(1):85–93.69. Fish FE, Peacock JE, Rohr J. Phase relationships between body componentsof odontocete cetaceans in relation to stability and propulsive mechanisms.Proc 1st Int Symp Aqua Bio-Mech. 2000;57–60.70. Quinn DB, Lauder GV, Smits AJ. Flexible propulsors in ground effect.Bioinspir Biomim. 2014;9:036008.71. Webb PW. Simple physical principles and vertebrate aquatic locomotion.Am Zool. 1988;28:709–25.72. Williams R, Noren DP. Swimming speed, respiration rate and estimated costof transport in adult killer whales. Mar Mamm Sci. 2009;25(2):327–50.73. Guinet C, Domenici P, de Stephanis R, Barrett-Lennard L, Ford JKB, VerborghP. Killer whale predation on bluefin tuna: exploring the hypothesis of theendurance-exhaustion technique. Mar Ecol Prog Ser. 2007;347:111–9.74. Roos MMH, Wu G, Miller PJO. The significance of respiration timing in theenergetics estimates of free-ranging killer whales (Orcinus orca). J Exp Biol.2016;219:2066–77.75. Quinn TP. The behavior and ecology of Pacific salmon and trout. Vancouver:UBC Press; 2005.76. Erickson DL, Pikitch EK. Incidental catch of chinook salmon in commercialbottom trawls off the US west coast. N Am J Fish Manag. 1994;14(3):550–63.77. Kuhn BR. The MRP-reporter program: a data extraction and reporting toolfor the mark recovery program data-base. Can Tech Rep Fish Aquat Sci.1988;1625:145.78. Ogura M, Ishida Y. Swimming behavior of coho salmon, Oncorhynchuskisutch, in the open sea as determined by ultrasonic telemetry. Can J FishAquat Sci. 1992;49:453–7.79. Ogura M, Ishida Y. Homing behavior and vertical movements of fourspecies of Pacific salmon (Oncorhynchus spp.) in the central Bering Sea. CanJ Fish Aquat Sci. 1995;52:532–40.80. Wada K, Ueno Y. Homing behavior of chum salmon determined by anarchival tag. N Pac Anadr Fish Comm Doc. 1999;425:29.81. Yano K, Ichihara T, Nakamura A, Tanaka S. Escape behavior of the chumsalmon Onchorhynchus keta upon encountering Dall’s porpoisePhocoenoides dalli. Bull Jpn Soc Sci Fish. 1984;50(8):1273–7.82. Godin JJ. Evading predators. In: Godin JJ, editor. Behavioural ecology ofteleost fishes. Oxford: Oxford University Press; 1997. p. 190–229.83. Denton E. On the organization of reflecting surfaces in some marineanimals. Philos Trans R Soc Lond B. 1970;258(824):285–313.84. Au WWL, Ford JKB, Horne JK, Allman KAN. Echolocation signals of free-ranging killer whales (Orcinus orca) and modeling of foraging for chinooksalmon (Oncorhynchus tshawytscha). J Acoust Soc Am. 2004;115(2):901–9.85. Madsen PT, Johnson M, de Soto NA, Zimmer WMX, Tyack P. Biosonarperformance of foraging beaked whales (Mesoplodon densirostris). J Exp Biol.2005;208(2):181–94.86. Hoelzel AR. Foraging behavior and social group dynamics in Puget Soundkiller whales. Anim Behav. 1993;45(3):581–91.87. Domenici P. The scaling of locomotor performance in predator-preyencounters: from fish to killer whales. Comp Biochem Physiol A.2001;131:169–82.88. Reidy SP, Kerr SR, Nelson JA. Aerobic and anaerobic swimming performanceof individual Atlantic cod. J Exp Biol. 2000;203:347–57.89. Randall DJ, Mense D, Boutilier RG. The effects of burst swimming on aerobicswimming in chinook salmon (Oncorhynchus tshawytscha). Mar BehavPhysiol. 1987;13:77–88.90. Brett JR. The respiratory metabolism and swimming performance of youngsockeye salmon. J Fish Res Board Can. 1964;21:1183–226.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submitSubmit your next manuscript to BioMed Central and we will help you at every step:Wright et al. Movement Ecology  (2017) 5:3 Page 18 of 18


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items