Breathing frequencies of northern elephant seals at sea and on …mirounga.ucsc.edu/leboeuf/pdfs/Andrews.2000.pdf · Breathing frequencies of northern elephant seals at sea and on

Respiration Physiology 123 (2000) 71–85

Breathing frequencies of northern elephant seals at sea andon land revealed by heart rate spectral analysis

Russel D. Andrews a,*, Daniel P. Costa b, Burney J. Le Boeuf b,David R. Jones a

a Department of Zoology, Uni6ersity of British Columbia, 6270 Uni6ersity Bl6d., Vancou6er, BC, Canada V6T 1Z4b Department of Biology and Institute of Marine Sciences, Uni6ersity of California, Santa Cruz, CA 95064, USA

Accepted 5 June 2000

Abstract

Elephant seals breathe episodically at sea and on land and surprisingly long apnoeas occur in both situations. Animportant difference is that recovery from apnoeic periods is much quicker at sea, which might be due, in part, todifferences in the ventilatory response. Respiratory frequencies of juvenile northern elephant seals diving at sea andresting on land were estimated from time-frequency maps of the Wigner distribution of heart rate variability.Simultaneous direct measurement of respiration and estimation of respiratory frequency (fR) in the laboratorydemonstrated that the error of estimation was small (mean9S.D.=1.0591.23%) and was independent of themagnitude of fR. Eupnoeic fR at sea was 2.4 times higher than on land (22.092.0 vs. 9.291.3 breaths min−1,respectively), facilitating quick recovery from the preceding dive and allowing a 34% increase in time spent apnoeicat sea versus on land. The overall fR (no. of breaths in a eupnoea divided by the total time of the apnoea+eupnoeacycle) of 2.390.6 breaths min−1 at sea was no different from the rate on land and was inversely related to thepreceding dive duration, suggesting that metabolism on longer dives may be reduced. © 2000 Elsevier Science B.V.All rights reserved.

Keywords: Apnea, elephant seal; Heart, rate, variability, respiratory arrhythmia; Mammals, elephant seal; Pattern of breathing,variability

www.elsevier.com/locate/resphysiol

1. Introduction

It has been suggested that the ability to max-imise the percentage of time spent submerged isjust as important to underwater foragers as their

capacity for making long dives (Fedak et al.,1988). The northern elephant seal (Mirounga an-gustirostris) is a species that possesses extremecapabilities in both categories. Mean dive dura-tions of juveniles and adults range from 15 to 20min, with the maximum observed duration being119 min and yet dives are almost always followedby short surface intervals of :2 min (Le Boeuf etal., 1988, 1996; Stewart and DeLong, 1995). Ele-

* Corresponding author. Tel.: +1-604-8225267; fax: +1-604-8222416.

E-mail address: [email protected] (R.D. Andrews).

0034-5687/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.

PII: S 0034 -5687 (00 )00168 -7

R.D. Andrews et al. / Respiration Physiology 123 (2000) 71–8572

phant seals are, therefore, very efficient divers,spending 80–95% of their time at sea underwater.

This exceptional ability must be related to highrates of gas exchange at the surface, but little isknown about the physiology of recovery fromdiving in this species. Given that elephant sealsare more accurately described as ‘surfacers’ thandivers, their ‘surface physiology’ deserves as muchattention as the more commonly studied ‘divingphysiology’. Heart rate between dives is 174%greater than during dives and 64% higher than therate during eupnoeic periods on land (Andrews etal., 1997). Such high heart rates must facilitaterapid gas transport between tissue and lung andare presumably accompanied by high breathingrates to ensure similarly rapid exchange betweenlung and environment. Ventilation is positivelyrelated to the duration of the preceding dive inWeddell seals (Leptonychotes weddellii ), plateau-ing at about six times the resting minute volumeafter dives of 15 min or longer. This hyperventila-tion is due to a 1.5–2 times increase in tidalvolume and a 3-fold increase in respiratory fre-quency (Kooyman et al., 1971). Unfortunately,Weddell seals are the only pinniped species forwhich it has been possible to quantify the ventila-tory response to voluntary diving in nature.

In preliminary experiments conducted as partof an earlier study (Andrews et al., 1997), at-tempts to record respiration from freely divingnorthern elephant seals using nasal thermistorswere unsuccessful. The seals always managed toremove the thermistors before being released atsea. We were able, however, to obtain R-R inter-val time series at sea and on land for four sealsequipped with electrocardiogram (ECG) record-ing Holter monitors. Northern elephant seals dis-play a pronounced respiratory sinus arrhythmia(RSA) when on land, with heart rate increasingon inspiration and falling on expiration(Bartholomew, 1954; Castellini et al., 1994a,b).RSA also occurs during the eupnoeic periodsbetween short dives in laboratory tanks, albeitwith a reduced amplitude (personal observations)and rhythmic heart rate oscillations that appearto be due to RSA were also seen in sea-goingnorthern elephant seals when at the surface (An-drews et al., 1997). We therefore sought to esti-

mate the respiratory frequency (fR) of freelydiving elephant seals through the analysis of theserhythmic fluctuations in heart rate.

We required an objective method that could beautomated for analysis of large amounts of dataand initially attempted to use the Womack (1971)method for the estimation of the time of occur-rence of individual breaths. In this method, alow-pass filter is applied to the heart rate timeseries, deriving a smoothed function which is thencompared with the unfiltered time series in orderto detect inspiration-induced peaks that rise abovethe local average. Despite trying many differentlow-cut-off frequencies and various other filteringalgorithms, we were not able to derive a satisfac-tory method for application to the at sea databecause the RSA amplitude was very low. Insteadof attempting to identify the exact time of eachindividual breath, we then decided to estimate themean frequency of heart rate oscillations over abreathing period by using spectral analysis. TheFourier transform is often used for heart ratespectral analysis, but the breathing periods at seaare usually very short, while on land the breath tobreath intervals can vary considerably over ashort time period. A more appropriate method forshort, nonstationary data is the discrete Wignerdistribution (DWD), a form of time frequencymapping which provides a two-dimensional func-tion of time and frequency (Wigner, 1932; Claasenand Mecklenbrauker, 1980). The DWD has beenshown to correctly map the instantaneous changesin R-R interval spectral content even during con-tinuous slowing of respiration in humans (Novaket al., 1993). Novak et al. (1993) showed that thedominant frequency of the R-R interval fluctua-tions closely followed the changing respiratoryfrequency over a large range, from 28 to 3 breathsmin−1. In this paper we present data to show thatthe DWD of heart rate variability provides agood estimate of fR in northern elephant seals.We then compare our estimates of fR at sea andon land to determine whether an increase in fR

during at sea breathing intervals is an importantelement of the elephant seals ability to minimisethe time spent at the surface.

R.D. Andrews et al. / Respiration Physiology 123 (2000) 71–85 73

2. Materials and methods

2.1. Laboratory 6alidation of fR estimationmethod

Four juvenile northern elephant seals (Table 1)were captured at Ano Nuevo State Reserve, CAand held in captivity at the Long Marine Labora-tory, Santa Cruz, CA for up to 5 days. Details ofthe immobilisation, capture and transport meth-ods are presented in Andrews et al. (1997). Thesurface ECG was recorded from two of these seals(L1 and L2) onto channel 1 of a 2-channel ana-logue Holter monitor (Model 90205, Space LabsInc., Redmond, WA) attached to the fur in themid-dorsal region (Andrews et al., 1997). Respira-tion was monitored in these two seals by record-ing, on channel 2 of the Holter monitor, the signalfrom a thermistor glued to the fur just above oneof the nares. Respiratory and electrocardiographicsignals were recorded from seals L1 and L2 for atleast 4 h during daylight hours, while they restedin a fenced, outdoor pen. The Holter monitortapes were scanned and digitised at 200 Hz usingan FT2000A Medical Workstation computer sys-tem (Space Labs, Inc.).

The ECG of seals L3 and L4 was monitored byusing needle electrodes attached subdermallyalong the dorsal midline. Respiratory movementsof these seals were monitored using two addi-tional needle electrodes placed subdermally oneach side of the thorax at the level of the posteriorend of the foreflippers. These electrodes were at-tached via long leads to an impedance respirome-ter. The ECG and respiration signals wereacquired and stored on a personal computerequipped with an analogue-to-digital convertercard. The ECG was sampled at 200 Hz, whilerespiration was sampled at 2 Hz. Seals L3 and L4were held in a foam-padded metal cage for :4 hof daytime monitoring, after which their elec-trodes were removed and they were returned to anoutdoor pen.

Sections of the records from all four seals werechosen for analysis by selecting time periods dur-ing which the seals were apparently sleeping orresting quietly. Only complete apnoea cycles (anapnoea and subsequent eupnoea) that were fol-

lowed by another apnoea were analysed. To ob-tain R-R interval time series, the QRS complexesof the ECG were detected and the R-R intervalswere calculated. Data segments from at least fiveapnoea cycles were chosen for each seal. Eachsegment consisted of the data 30 sec before thestart of an eupnoea and continued until the end ofthe eupnoea. The time series were linearly interpo-lated at 4 Hz and the low-frequency baseline trendwas removed by applying a moving fourth-orderpolynomial function. Time-frequency maps of theR-R interval variability for each data segmentwere constructed using the discrete Wigner distri-bution, with the parameter settings suggested byNovak et al. (1993). Each map consisted of spec-tra calculated at 0.5-sec intervals (Fig. 1). Foreach of these spectra, the frequency at which themaximum power occurred (peak power fre-quency) was determined. The frequency resolutionof these individual spectra was 0.0156 Hz, or 0.94cycles min−1. Although the peak power frequencyis a function of the time and frequency smoothingof the Wigner distribution, we used it to providean estimation of the fR for that time period. Theindividual spectra of the time-frequency mapswere only used to estimate fR if the spectrum wasdominated by a single peak. Spectra that con-tained multiple peaks at similar power levels werediscarded. This sometimes occurred at late stagesin eupnoeas when the breathing intervals couldbecome especially irregular. If \10% of the spec-tra from a time-frequency map for an eupnoeicsegment were discarded, then that segment wasnot included in the analysis.

In order to calculate the error involved in esti-mating fR from the spectral analysis of R-R inter-val variability, fR was directly measured bycounting the number of breaths (including frac-tions of a breath cycle) that occurred in the period15–60 sec after the start of an eupnoea. Thiscount was then divided by the duration of theperiod (45 sec) to obtain the mean measured fR

for individual eupnoeas. Over the same period,the mean estimated fR was determined by calcu-lating the mean of the peak power frequenciesfrom each of the time-frequency map spectra forthat period. These 45 sec periods were chosen foranalysis to be consistent with the analyses of


Tab

le1

Sum

mar

yof

the

seal

sin

the

labo

rato

ry,

thei

rap

noea

/eup

noea

dura

tion

s,an

dth

eer

ror

inth

ees

tim

atio

nof

eupn

oeic

resp

irat

ory

freq

uenc

yfr

omth

esp

ectr

alan

alys

isof

hear

tra

teva

riab

ility

a

Mea

neu

pnoe

an

Mea

nm

easu

red

Mas

s(k

g)A

ge(m

onth

s)Se

alno

.M

ean

esti

mat

edM

ean

apno

eaM

ean

abso

lute

Mea

n%

DfR

(bre

aths

dura

tion

(min

)du

rati

on(m

in)

valu

eof

%D

fR(b

reat

hsm

in−

1)

min

−1)

L1

14.19

0.8

1614

.29

0.8

1.129

0.8

2.299

2.04

145

115.

89

1.2

5.69

2.0

L2

8.29

1.7

168.

49

1.8

1.889

4.38

3.399

3.22

198

107.

39

1.2

5.39

1.6

7.19

1.2

7.09

1.3

−0.

719

2.92

2.509

1.44

4.99

1.9

154

8.39

3.9

9L

321

3.89

0.3

217.

19

1.1

7.29

1.3

1.919

4.49

4.159

1.80

153

58.

49

1.8

L4

7.49

1.2

4.99

0.8

9.19

3.4

9.29

3.4

1.059

1.23

3.089

0.85

Gra

ndm

ean

an

isth

enu

mbe

rof

apno

ea/e

upno

eacy

cles

.M

ean

valu

esar

epr

esen

ted

asm

ean9

S.D

.;a

gran

dm

ean

isth

em

ean

ofth

ein

divi

dual

seal

mea

ns.

Mea

sure

dfR

,re

spir

ator

yfr

eque

ncy

mea

sure

dby

coun

ting

brea

ths

inth

epe

riod

s15

–60

sec

afte

rth

est

art

ofa

eupn

oea.

Est

imat

edfR

,re

spir

ator

yfr

eque

ncy

esti

mat

ion

base

don

the

mea

npe

akpo

wer

freq

uenc

yfr

omth

eti

me

freq

uenc

ym

apof

the

R-R

vari

abili

tyfo

rth

epe

riod

15–6

0se

caf

ter

the

star

tof

aeu

pnoe

a.%

D,

the

diff

eren

cebe

twee

nth

em

easu

red

and

esti

mat

edf R

expr

esse

das

ape

rcen

tage

.


Fig. 1. Comparison of directly measured respiratory frequency (fR) and estimation of fR from the R-R interval time-frequency map.(A) Time course of the respiration signal, instantaneous heart rate (reciprocal of the R-R interval), and fR, both the measuredinstantaneous fR (reciprocal of the interbreath interval) and the estimated fR. (B) Time-frequency map of the R-R interval time seriesfrom the eupnoeic period in A. (C,D) The spectra from the time-frequency map at 45 sec (D) and 240 sec (C). The peak powerfrequency of each spectrum is indicated, which also provides an estimate of the fR at that time instant. Over the period from 15 to263 sec, the mean measured fR=6.64 breaths min−1, while the mean estimated fR=6.66 breaths min−1.


at-sea eupnoeas. At-sea eupnoeas lasted on aver-age for only 74 sec and the first 15 sec wereignored because of the imprecision in determiningthe start time of eupnoeas at sea.

2.2. Estimation of fR for seals at sea and on thebeach

The dive behaviour and ECG were recordedfrom four translocated seals (three males and onefemale; age: 16–34 months; mass: 150–245 kg,mean mass: 201 kg) as reported in Andrews et al.(1997). These seals were captured at Ano NuevoState Reserve, instrumented at the Long MarineLaboratory, transported either out to sea or tothe opposite side of Monterey Bay and then re-leased. The ECG was recorded as the seals dovecontinuously on their return to Ano Nuevo andfor 12–18 h after they reached land. Only eup-noeic periods longer than 32 sec, in which theECG was completely noise-free and did not con-tain any ectopic or missing beats, were chosen foranalysis. The start and end of eupnoeas at seacould be determined with an accuracy of :95sec, because of the 5 sec sampling interval of thetime-depth recorder (TDR). On land, the start ofeupnoeas and apnoeas was determined from themarked changes in heart rate that accompanythese transitions (Andrews et al., 1997).

The R-R interval time series were treated asdescribed above for the seals in the laboratoryand time-frequency maps were created for eacheupnoea. For eupnoeas at sea, an ‘estimated eup-noeic fR1st min, was calculated for the period from15–60 sec (for the few eupnoeas B60 sec, fR wasestimated for the period from 15 sec to the end ofeupnoea). The mean eupnoea duration at sea wasonly 74 sec, so the estimated eupnoeic fR1st min wasmultiplied by the total duration of the eupnoea inorder to estimate the number of breaths in eacheupnoea. The ‘estimated overall fR’ was defined asthe quotient of the number of breaths in anepisode of eupnoea and the total duration of theapnoea cycle (apnoea duration plus eupnoeaduration).

For eupnoeas on land, an ‘estimated eupnoeicfR1st min’ was calculated as above for the firstminute, and for each subsequent minute of the

eupnoea an estimate was derived for the entireminute (e.g. ‘estimated eupnoeic fR2nd min’ coversthe range from 60–120 sec). For the last fractionof a minute in a eupnoea, an estimate was derivedfor the total remainder of time (e.g. for a 3.6 mineupnoea, the ‘estimated eupnoeic fR4th min, coversthe range from 180–216 sec). The total number ofbreaths in a eupnoea on land were calculatediteratively, by multiplying the fraction of timespent in each 1 min bin by the estimated fR forthat bin and then calculating the sum. In some ofthe data segments for the later periods of eup-noeas (e.g. in the fourth and fifth minute) heartrate became quite arrhythmic and \10% of thespectra did not contain a single dominant peak. Inthose cases, the fraction of time spent in a 1 minbin was multiplied by that seal’s mean value forestimated fR for that bin. For example, if theperiod from 240–300 sec of a 5.0 min eupnoeahad to be discarded, then the number of breathsfor that 1 min period were estimated by multiply-ing the mean estimated eupnoeic fR5th min valuefrom that seal’s other eupnoeas by 1 (because theeupnoea lasted 5 min the fraction of time spent inthe 5th minute bin was 1).

Student’s paired t-tests were used to compareapnoea and eupnoea durations and respiratoryfrequencies at sea and on land, except where itwas appropriate to perform a one-way repeatedmeasures ANOVA. Significance was accepted atthe level of PB0.05 except when it was necessaryto use the sequential Bonferroni method to min-imise type-I errors (Rice, 1989). Relationshipswere examined using least-squares linear regres-sion.

3. Results

3.1. Laboratory 6alidation

During periods of simultaneous monitoring ofrespiration and heart rate, the estimated fR fromthe 0.5 sec spectra of the DWD time-frequencymap matched the measured instantaneous fR

closely, even when the breathing intervals werequite variable over time (Fig. 1). Therefore, themean estimated fR was usually within a few per-


Fig. 2. The algebraic (A) and absolute value (B) of % D, the difference between measured and estimated fR, plotted against themeasured fR for individual eupnoeic segments from all four seals (L1–L4).


Fig. 3. Calculation of ‘estimated eupnoeic fR1st min, from the heart rate for a 1.25 min surface interval marked (*) in (A) and shownin an expanded view in (B). (C) The DWD time-frequency map of the R-R variability. (D) Time series of the estimated fR basedon the peak power frequency of each of the spectra taken at 0.5 sec intervals from the time-frequency map. The estimated eupnoeicfR1st min was 20.6 breaths min−1.


Fig. 4. Estimated eupnoeic fR1st min plotted against the duration of the preceding dive or beach apnoea for each of the translocatedseals.

cent (range −5.6 to +10.0%) of the mean mea-sured fR for individual eupnoeic segments (Fig. 2).The mean algebraic error in the estimation of fR

for all seals was only 1.0591.23% and the meanabsolute value of the error was 3.0890.85%(Table 1). There was no relationship between themeasured fR and the error in its estimation (Fig.2).

3.2. Respiration at sea and on the beach

The DWD time-frequency maps of heart ratevariability for eupnoeic segments at sea were usu-ally dominated by a peak at the presumed respira-tory frequency, although there was oftenconsiderable spectral power at other frequenciesas well (Fig. 3C). The estimated eupnoeic fR1st min

for individual surface intervals tended to increasewith the duration of the preceding dive (Fig. 4).

In the two seals that made many dives in excess of15 min, the estimated eupnoeic fR1st min appearedto plateau at :25 breaths min−1 (Fig. 4). Themean estimated eupnoeic fR1st min for individualseals ranged from 19.091.6 to 23.392.1 breathsmin−1 (Table 2). There was a significantly posi-tive linear relationship for each seal (r2=0.20−0.67) between the estimated number of breathsper episode and the duration of the precedingdive. However, there was a negative relationshipbetween the estimated overall fR and precedingdive duration (significant for three seals, r2=0.23−0.43). Overall fR was even more stronglyrelated in an inverse fashion to the percent timesubmerged (significant for all four seals, r2=0.28−0.96).

Time-frequency maps of heart rate variabilityduring eupnoeas on the beach were very similar tothose for the seals in the laboratory. There was a


Tab

le2

Dur

atio

nsof

apno

eaan

deu

pnoe

aan

dre

spir

ator

yfr

eque

ncie

sfo

rtr

ansl

ocat

edse

als

atse

aan

don

land

a

Seal

tag

no.

Mea

n%

tim

eM

ean

apno

eaM

ean

eupn

oea

Mea

nes

tim

ated

eupn

oeic

nM

ean

esti

mat

edov

eral

lfR

Mea

nes

tim

ated

no.

offR

1st

min

(bre

aths

min

−1)

apno

eic

dura

tion

(min

)br

eath

spe

rep

isod

e(b

reat

hsm

in−

1)

dura

tion

(min

)

At

sea:

GG

571

1.99

0.5

3128

.39

10.4

14.69

6.1

1.219

0.34

91.6

22.99

3.0

1.99

0.8

25.69

7.1

23.29

2.1

3813

.99

5.7

GH

929

91.6

1.109

0.26

19.09

1.6

342.

29

0.5

30.49

7.1

12.59

2.9

1.599

0.34

88.3

GJ3

2586

.222

.99

1.4

3.29

0.6

24.69

8.1

GJ7

1123

6.99

2.0

1.069

0.30

89.49

2.6

22.09

2.0

2.39

0.6

27.29

2.6

Gra

ndm

ean:

12.09

3.5

1.249

0.24

On

land

:2.

89

0.8

38.69

10.3

9.49

3.7

5.19

1.5

24G

G57

162

.98.

29

1.4

2.09

0.6

33.99

11.8

GH

929

1713

.59

5.1

4.49

1.6

74.2

8.29

1.8

3.59

0.7

36.09

15.4

11.09

2.1

58.0

GJ3

254.

39

2.1

6.09

2.0

2371

.69.

69

2.0

2.39

0.4

26.49

9.6

15G

J711

8.49

3.2

3.29

1.3

2.69

0.6

33.79

5.2

9.29

1.3*

4.29

0.78

*66

.79

7.5*

9.39

3.1

Gra

ndm

ean:

aM

ean

valu

esar

epr

esen

ted

asm

ean9

S.D

.;a

gran

dm

ean

isth

em

ean

ofth

ein

divi

dual

seal

mea

ns.

Eup

noei

cfR

1st

min

,th

ees

tim

ated

fRfo

rth

epe

riod

15–6

0se

caf

ter

the

star

tof

aeu

pnoe

a.O

vera

llfR

,th

ees

tim

ated

num

ber

ofbr

eath

sof

aeu

pnoe

adi

vide

dby

the

dura

tion

ofth

eco

mpl

ete

apno

eacy

cle.

Sequ

enti

alB

onfe

rron

ipr

oced

ure

was

used

tom

inim

ise

Typ

e1

erro

rsin

the

mul

tipl

e(6

)pa

ired

t-te

sts.

*Si

gnifi

cant

diff

eren

cebe

twee

nth

e‘a

tse

a’an

d‘o

nla

nd’

valu

e.


trend for the estimated eupnoeic fR1st min for indi-vidual eupnoeas on the beach to be directly re-lated to the duration of the preceding apnoea, atleast for apnoeas up to 10 min long (Fig. 4).Mean estimated eupnoeic fR1st min for all seals was9.291.3 breaths min−1, which was 58% lowerthan the estimated eupnoeic fR1st min at sea (Table2). The mean estimated fR for each subsequentminute of eupnoea tended to decrease slightly.There were adequate sample sizes from all fourseals to allow comparison of fR1st min, fR2nd min,fR3rd min and fR4th min, with a repeated measuresANOVA (multiple pairwise comparisons donewith the Student-Newman–Keuls test). Only fR3rd

min and fR4th min was significantly lower than fR1st

min and the mean fR4th min was 7.791.0 breathsmin−1. The mean estimated number of breathsper episode on land was 3495, which was notsignificantly different from the mean value of2793 breaths per episode at sea (Table 2). Themean estimated overall fR was 2.690.6 breathsmin−1 on land and 2.390.6 breaths min−1 at sea(Table 2). Because the number of breaths forsome long eupnoeas on land had to be calculatedwith mean fR values for individual 1 min binsinstead of with the actual estimated fR for thatmin of the eupnoea, we did not examine relation-ships between the duration of the preceding ap-noea and either the number of breaths per episodeor the overall fR.

4. Discussion

The estimation of respiratory frequency fromthe spectral analysis of heart rate using the dis-crete Wigner distribution appears to be a satisfac-tory method for situations in which it is possibleto record the ECG but not respiration. Althoughwe were not able to validate the method by simul-taneous recording of respiration and estimation offR in seals at sea, we are confident nonethelessthat our estimates of fR for seals at sea arereasonable. In the laboratory we observed nocorrelation between measured fR and the error inestimating fR, even though one of the seals (L1)breathed at an unusually high rate (Table 1, Fig.2). The RSA amplitude was reduced in seal L1

compared with the others, and the magnitude ofthe peak R-R interval spectral power was dimin-ished, but the error in estimating fR was still verylow. RSA amplitude in the seals breathing at seawas even lower, so the level of spectral power atthe peak power frequency (the presumed fR) wasvery low (compare Figs. 1 and 3). Overall spectralpower, however, was also reduced and so therewas still a single, obviously dominant, peak inmost spectra.

A reduction in R-R interval spectral power atthe respiratory frequency and at all other frequen-cies is also seen at high respiratory rates duringexercise in humans (Arai et al., 1989). RSA isdependent upon both fR and tidal volume (VT),and while the phase and amplitude may change,there is usually a 1:1 relationship between therespiratory movements and RSA heart rate fluctu-ations (Angelone and Coulter, 1964; Hirsch andBishop, 1981). The coherence between the R-Rinterval power spectrum and the respiratory signalpower spectrum is usually near one and is inde-pendent of fR in humans (Patwardhan et al.,1995). This would appear to be the case forelephant seals as well, at least for seals on thebeach breathing at a limited range of frequencies.Our estimates of fR for seals at sea are alsosupported by recent, serendipitous measurementsof fR from the recording of breath sounds oftranslocated northern elephant seals carryingacoustic monitoring devices (Fletcher et al., 1996).The mean eupnoeic fR, between dives that aver-aged 14.7 min, ranged from 22.091.0 to 24.691.6 breaths min−1 for three seals. These valuesare very similar to the values of fR that weestimated in this study.

The mean post-dive eupnoeic fR1st min was 2.4times greater than the mean post-apnoea eupnoeicfR1st min on land (Table 2). Eupnoeic fR1st min at seaalso increased in direct relation to the duration ofthe preceding dive, at least for dives up to 15 minin length (Fig. 4). These high breathing ratespermit elephant seals to spend only short periodsat the surface for gas exchange, resulting in a highpercentage of time spent submerged, or apnoeic.The mean duration of apnoea at sea was notsignificantly different from that on land, and thenumber of breaths per episode was also not differ-


ent. The much higher fR at sea, however, resultedin 34% increase in the time spent apnoeic at seacompared with apnoeic time on land.

This difference in fR between diving seals andseals resting on land contrasts with a study ofnorthern elephant seal pups. In the laboratory,pups performed equivalent length breath-holdswhether sleeping underwater in a shallow tank orsleeping in dry conditions, and there was no dif-ference in the eupnoeic fR between wet and dryconditions (Castellini et al., 1994a). It is possiblethat the increased cardiac and respiratory rates ofjuveniles during recovery from dives comparedwith recovery from sleep apnoeas are due to moreextreme changes in blood gas levels and therefore,greater levels of respiratory drive during recoveryfrom diving. When northern elephant seal pupsthat are awake and breathing regularly, withoutextended apnoeas, are exposed to hypercapnicbreathing gas, fR approximately doubles (Milsomet al., 1996). Exposure to hypercapnia duringbouts of sleep apnoea, however, causes only asmall increase in the instantaneous breathing ratefor the first five breaths after an apnoea, andmoderate hypoxia has no effect on fR whether thepups are asleep or awake. It is therefore not clearwhether increased PaCO2

and decreased PaO2

would be expected to cause as large an increase infR as the one we saw in recovery from divescompared with recovery from apnoeas on land inelephant seals.

It is not even clear whether blood gas levels willdiffer at the end of dives and beach apnoeas ofsimilar duration. Furthermore, the direction ofany difference may not be what one might expect,despite that during a dive seals are swimming andduring most apnoeas on land seals are simplysleeping. For instance, during one episode of sleepapnoea in a Weddell seal lying on ice, PaO2

fell to25 mmHg and PaCO2

rose to 55 mmHg after only4 min (Kooyman et al., 1980). Such a low PaO2

was not usually seen in freely diving seals untilafter at least 15 min of submergence and PaCO2

never rose above 53 mmHg, even in dives as longas 27 min (Qvist et al., 1986).

Although simultaneous measurements of fR, VT

and alveolar and blood gases in elephant sealsduring both diving and sleep apnoea are needed

before an attempt to resolve the issue can bemade, it seems safe to propose that there areimportant differences in the control of the car-diorespiratory responses to diving and sleep ap-noea. Perhaps during diving there is a change inthe set point and gain of the ventilatory CO2

response. Rather than being an automatic re-sponse to respiratory drive, irrespective of situa-tion, it appears that the high heart rates andbreathing rates after a dive uniquely serve toensure rapid, as opposed to adequate, gas ex-change, thereby decreasing the amount of timespent at the surface.

Although the elephant seal’s mean eupnoeicbreathing rate at sea of 22 breaths min−1 is 2.4times higher than on land, terrestrial mammalsare often capable of increasing fR by four to eighttimes during strenuous exercise. If elephant sealsare under pressure to decrease the fraction of timethey spend at the surface, one might ask why theydo not breathe even faster between dives. Evensedentary calves (mass :180 kg) are able toincrease fR up to 65 breaths min−1 during exer-cise (Kuhlmann et al., 1985), and the fR of ponies(mass :150 kg) exercising near their maximalwork load reaches 95 breaths min−1, an increaseof five times over resting fR (Bisgard et al., 1978).Of course, respiration in elephant seals floatingmotionless at the surface cannot be assisted bycoupling between respiratory and locomotormovements, as may be the case in runningmammals.

Nevertheless, if elephant seals breathe with verylarge tidal volumes after dives, like Weddell andgrey (Halichoerus grypus) seals (Kooyman et al.,1971; Reed et al., 1994), reasonably high levels ofventilation may still be achieved. Based on therelationship between body mass and lung volumederived from other marine mammal species(Kooyman, 1989), the predicted total lung capac-ity of a 201-kg elephant seal is 16.2 L. Themaximal VT during recovery from voluntary divesin Weddell and grey seals is between 46 and 49%of total lung capacity (TLC) (Kooyman et al.,1971; Reed et al., 1994). If we assume that VT inelephant seals is 50% of TLC, or 8.1 L, then witha fR of 22 breaths min−1 they would have anexpired ventilation of 178 L min−1. At a maximal


breathing rate of 27 breaths min−1, ventilationwould reach 219 L min−1. Similarly sized calvesand ponies reach ventilation levels of 251 and 435L min−1, respectively, but a comparison of alve-olar ventilation may be more appropriate becausethe dead space (VD) to VT ratio tends to be verysmall in phocid seals. (Kooyman et al., 1971; Craigand Pasche, 1980). With a VD/VT of 0.15, at themaximal fR of 27 breaths min−1, elephant sealalveolar ventilation might reach 186 L min−1,exceeding the alveolar ventilation reached bycalves (160 L min−1), but still much less than thatobserved in the highly athletic ponies (234 Lmin−1).

A factor that may be limiting total ventilation inelephant seals is the extra work required to breathewhile immersed in water. When juvenile elephantseals reach the surface and begin to breathe theyare still immersed up to the level of the base of theskull, and the midpoint of the lung is subject to ahydrostatic pressure of :40 cm of seawater (4.0kPa). The work of breathing increases by 60%when humans are immersed up to the neck, andthis is partly due to an increase in airway resistancebecause of compression of the extrathoracic air-ways (Agostoni et al., 1966; Hong et al., 1969). Thetrachea, bronchi and terminal airways of pin-nipeds, however, have an unusual amount of mus-cular and cartilaginous reinforcement (Denisonand Kooyman, 1973; Tarasoff and Kooyman,1973). The function of this reinforcement may beto limit nitrogen absorption during deep dives, byallowing orderly collapse of the lung (Scholander,1940), or to allow high expiratory flow at low lungvolumes (Denison and Kooyman, 1973; Kerem etal., 1975). On the other hand, the strengthenedairways may also serve to resist hydrostatic com-pression during immersed breathing. Nonetheless,Kerem et al. (1975) observed that inspiratory flowrates of immersed California sea lions (Zalophuscalifornianus) were much lower than expiratoryrates and although they thought this result waspuzzling, it is likely that while pinnipeds can expirelarge volumes very quickly, inspiration is limitedby the difficulty of breathing against a large posi-tive pressure.

A potential trade-off between oxygen transportand oxygen storage capacity may also limit the

maximum ventilation rate that elephant seals can,or should, achieve. The resting hematocrit (Hct) ofnorthern elephant seals is quite high (range :50–67%), and it can increase both during apnoeas onland and during diving in the laboratory (Castelliniet al., 1986; Hedrick et al., 1986; Wickham et al.,1990; Hedrick and Duffield, 1991; Thorson, 1993;Thorson and Le Boeuf, 1994). Although such ahigh Hct is an important component of the north-ern elephant seal’s exceptional oxygen storagecapacity, the concomitant exponential increase inblood viscosity may severely limit maximal oxygentransport (Hedrick et al. 1986; Hedrick andDuffield, 1991). There is little difference in surfaceinterval heart rate between short and long dives atsea (Andrews et al., 1997), suggesting that a heartrate of :110–120 beats min−1 is the highest ratepossible in juvenile elephant seals. If during recov-ery from dives cardiac output reaches a limit, thenfurther increases in ventilation would contributevery little to gas exchange at the expense ofincreases in the work of breathing. Extremely highHct may increase oxygen storage, enabling longermaximal dive times, but would also decrease oxy-gen transport, extending the time needed for recov-ery. Consequently, there may be a trade-offbetween the ability to make very long dives and theability to reduce the surface interval between dives.With dive durations of :20 min and 90% of timeat sea spent submerged, elephant seals may havereached the limit.

Elephant seals do, however, sometimes makeexceptionally long dives that are nonetheless fol-lowed by short surface intervals, which suggeststhat metabolism must be reduced in such dives sothat an extended recovery period at the surface isnot necessary. Although metabolic rate reductionsduring diving have been proposed frequently (forreviews see Boyd and Croxall (1996), Butler andJones (1997) and Kooyman and Ponganis (1998)),there has never been a direct measurement ofmetabolic rate during a dive for any marine mam-mal. In translocated juvenile elephant seals, thereis an inverse relationship between diving heart rateand dive duration, suggesting that metabolic ratemay be adjusted in a similar fashion (Andrews etal., 1997). In this study we reported that the overallfR (number of breaths per dive cycle divided by the


dive plus surface interval duration) is also in-versely related to dive duration, providing furthersupport that metabolic rate is reduced on longerdives.

Acknowledgements

We thank Dan Crocker, Pat Morris, GuyOliver, Phil Thorson, Jeannine Williams, YuanYeh, and numerous other students from UCSanta Cruz and the University of British Colum-bia for field and lab assistance, Bob Burr andVera Novak for help with spectral analysis ofheart rate, and Bill Milsom for reviewing a draftof this manuscript. This work was supported inpart by Office of Naval Research Grants N00014-93-J-1680 and N00014-91-J-4107, gifts from theG. Macgowan trust through the courtesy of G.A.Malloch, an NSERCC Research Grant, and anNSERCC International Collaborative Grant.

References

Agostoni, E., Gurtner, G., Torri, G., Rahn, H., 1966. Respira-tory mechanics during submersion and negative-pressurebreathing. J. Appl. Physiol. 21, 251–258.

Andrews, R.D., Jones, D.R., Williams, J.D., Thorson, P.H.,Oliver, G.W., Costa, D.P., Le Boeuf, B.J., 1997. Heartrates of northern elephant seals diving at sea and restingon the beach. J. Exp. Biol. 200, 2083–2095.

Angelone, A., Coulter, N.A. Jr, 1964. Respiratory sinusarrhythmia: a frequency dependent phenomenon. J. Appl.Physiol. 19, 479–482.

Arai, Y., Saul, J.P., Albrecht, P., Hartley, L.H., Lilly, L.S.,Cohen, R.J., Colucci, W.S., 1989. Modulation of cardiacautonomic activity during and immediately after exercise.Am. J. Physiol. Heart Circ. Physiol. 25 256, H132–H141.

Bartholomew, G.A. Jr, 1954. Body temperature and respira-tory and heart rates in the northern elephant seal. J.Mamm. 35, 211–218.

Bisgard, G.E., Forster, H.V., Byrnes, B., Stanek, K., Klein, J.,Manohar, M., 1978. Cerebrospinal fluid acid-base balanceduring muscular exercise. J. Appl. Physiol. 45, 94–101.

Boyd, I.L., Croxall, J.P., 1996. Dive durations in pinnipedsand seabirds. Can. J. Zool. 74, 1696–1705.

Butler, P.J., Jones, D.R., 1997. Physiology of diving of birdsand mammals. Physiol. Rev. 77, 837–899.

Castellini, M.A., Costa, D.P., Huntley, A., 1986. Hematocritvariation during sleep apnea in elephant seal pups. Am. J.Physiol. Regulat. Integr. Comp. Physiol. 20 251, R429–R431.

Castellini, M.A., Milsom, W.K., Berger, R.J., Costa, D.P.,Jones, D.R., Castellini, J.M., Rea, L.D., Bharma, S., Har-ris, M., 1994a. Patterns of respiration and heart rateduring wakefulness and sleep in elephant seal pups. Am. J.Physiol. Regulat. Integr. Comp. Physiol. 35 266, R863–R869.

Castellini, M.A., Rea, L.D., Sanders, J.L., Castellini, J.M.,Zenteno-Savin, T., 1994b. Developmental changes in car-diorespiratory patterns of sleep-associated apnea in north-ern elephant seals. Am. J. Physiol. Regulat. Integr. Comp.Physiol. 36 267, R1294–R1301.

Claasen, T.A.C.M., Mecklenbrauker, W.F.G., 1980. TheWigner distribution — a tool for time frequency signalanalysis. I. Continuous-time signals. Philips J. Res. 35,217–250.

Craig, A.B.J., Pasche, A., 1980. Respiratory physiology offreely diving harbour seals (Phoca 6itulina). Physiol. Zool.53, 419–432.

Denison, D.M., Kooyman, G.L., 1973. The structure andfunction of the small airways in pinniped and sea otterlungs. Respir. Physiol. 17, 1–10.

Fedak, M.A., Pullen, M.R., Kanwisher, J., 1988. Circulatoryresponses of seals to periodic breathing: heart rate andbreathing during exercise and diving in the laboratory andopen sea. Can. J. Zool. 66, 53–60.

Fletcher, S., Le Boeuf, B.J., Costa, D.P., Tyack, P.L., Black-well, S.B., 1996. Onboard acoustic recording from divingnorthern elephant seals. J. Accoust. Soc. Am. 100, 2531–2539.

Hedrick, M.S., Duffield, D.A., 1991. Haematological and rhe-ological characteristics in seven marine mammal species:physiological implications for diving behaviour. J. Zool.(London) 225, 273–283.

Hedrick, M.S., Duffield, D.A., Cornell, L.H., 1986. Bloodviscosity and optimal hematocrit in a deep-diving mammal,the northern elephant seal (Mirounga angustirostris). Can.J. Zool. 64, 2081–2085.

Hirsch, J.A., Bishop, B., 1981. Respiratory sinus arrhythmia inhumans: how breathing pattern modulates heart rate. Am.J. Physiol. Heart Circ. Physiol. 10 241, H620–H629.

Hong, S.K., Cerretelli, P., Cruz, J.C., Rahn, R., 1969. Me-chanics of respiration during submersion in water. J. Appl.Physiol. 27, 537–538.

Kerem, D.H., Kylstra, J.A., Saltzman, H.A., 1975. Respira-tory flow rates in the sea lion. Undersea Biomed. Res. 2,20–27.

Kooyman, G.L., 1989. Diverse Divers. Springer-Verlag,Berlin.

Kooyman, G.L., Ponganis, P.J., 1998. The physiological basisof diving to depth: birds and mammals. Ann. Rev. Physiol.60, 19–32.

Kooyman, G.L., Kerem, D.H., Campbell, W.B., Wright, J.J.,1971. Pulmonary function in freely diving Weddell seals,Leptonychotes weddelli. Respir. Physiol. 12, 271–282.

Kooyman, G.L., Wahrenbrock, E.A., Castellini, M.A., Davis,R.W., Sinnett, E.E., 1980. Aerobic and anaerobicmetabolism during voluntary diving in Weddell seals: Evi-


dence of preferred pathways from blood chemistry andbehavior. J. Comp. Physiol. B 138, 335–346.

Kuhlmann, W.D., Hodgson, D.S., Fedde, M.R., 1985. Res-piratory, cardiovascular, and metabolic adjustments to ex-ercise in the Hereford calf. J. Appl. Physiol. 58,1273–1280.

Le Boeuf, B.J., Costa, D.P., Huntley, A.C., Feldkamp, S.D.,1988. Continuous deep diving in female northern elephantseals, Mirounga angustirostris. Can. J. Zool. 66, 446–458.

Le Boeuf, B.J., Morris, P.A., Blackwell, S.B., Crocker, D.E.,Costa, D.P., 1996. Diving behaviour of juvenile northernelephant seals. Can. J. Zool. 74, 1632–1644.

Milsom, W., Castellini, M., Harris, M., Castellini, J., Jones,D., Berger, R., Bahrma, S., Rea, L., Costa, D., 1996.Effects of hypoxia and hypercapnia on patterns of sleep-as-sociated apnea in elephant seal pups. Am. J. Physiol.Regulat. Integr. Comp. Physiol. 40 271, R1017–R1024.

Novak, V., Novak, P., De Champlain, J., Le Blanc, A.R.,Martin, R., Nadeau, R., 1993. Influence of respiration onheart rate and blood pressure fluctuations. J. Appl. Phys-iol. 74, 617–626.

Patwardhan, A.R., Evans, J.M., Bruce, E.N., Eckberg, D.L.,Knapp, C.F., 1995. Voluntary control of breathing doesnot alter vagal modulation of heart rate. J. Appl. Physiol.78, 2087–2094.

Qvist, J., Hill, R.D., Schneider, R.C., Falke, K.J., Liggins,G.C., Guppy, M., Elliott, R.L., Hochachka, P.W., Zapol,W.M., 1986. Hemoglobin concentrations and blood gastensions of free-diving Weddell seals. J. Appl. Physiol. 61,1560–1569.

Reed, J.Z., Chambers, C., Fedak, M.A., Butler, P.J., 1994.Gas exchange of captive freely diving grey seals (Hali-choerus grypus). J. Exp. Biol. 191, 1–18.

Rice, W.R., 1989. Analyzing tables of statistical tests. Evolu-tion 43, 223–225.

Scholander, P.F., 1940. Experimental investigations on therespiratory function in diving mammals and birds.Hvalrad. Skr. 22, 1–131.

Stewart, B.S., DeLong, R.L., 1995. Double migrations of thenorthern elephant seal, Mirounga angustirostris. J. Mamm.76, 196–205.

Tarasoff, F.J., Kooyman, G.L., 1973. Observations of theanatomy of the respiratory system of the river otter, seaotter, and harp seal. II. The trachea and bronchial tree.Can. J. Zool. 51, 171–177.

Thorson, P.H., 1993. Development of Diving in the NorthernElephant Seal, Ph.D. thesis. University of California, SantaCruz, CA.

Thorson, P.H., Le Boeuf, B.J., 1994. Developmental aspects ofdiving in northern elephant seal pups. In: Le Boeuf, B.J.,Laws, R.M. (Eds.), Elephant Seals: Population Ecology,Behavior, and Physiology. University of California Press,Berkeley, CA, pp. 271–289.

Wickham, L.L., Costa, D.P., Elsner, R., 1990. Blood rheologyof captive and free-ranging northern elephant seals and seaotters. Can. J. Zool. 68, 375–380.

Wigner, E.P., 1932. On the quantum correction for thermody-namic equilibrium. Phys. Rev. 40, 749–752.

Womack, B.F., 1971. The analysis of respiratory sinusarrhythmia using spectral analysis and digital filtering.IEEE Trans. Biomed. Eng. BME 18, 399–409.

..

Breathing frequencies of northern elephant seals at sea and on …mirounga.ucsc.edu/leboeuf/pdfs/Andrews.2000.pdf · Breathing frequencies of northern elephant seals at sea and on

Documents