Capillary-Tube Depth Gauges for Diving Animals: an Assessment of Their Accuracy and Applicability

j. Field Ornithol., 59(4):345-354

CAPILLARY-TUBE DEPTH GAUGES FOR DIVING ANIMALS: AN ASSESSMENT OF THEIR ACCURACY AND APPLICABILITY

ALAN E. BURGER l AND RORY P. WILSON 2 BarnfieldMarine Station Barnfield,British Columbia, VOR IBO, Canada

Abstract.--Gaugesto measurethe maximum depthsattained by diving animals were con- structedfrom plastictubing lined with solubleindicator powder, at a costof $0.10 each. In our tests,the differencesbetween real and estimateddepths averaged <3% with single immersionsof gaugesto any depth up to 140 m. With multiple immersionserrors were usually <10% and always <25%. The accuracyof the gaugeswas not affectedby depthor duration of dive, rate of descent,or underwater movementssimulating a bird's swimming. Errors resultedfrom severejarring of the devicesunderwater, plunge-diving,accumulation of moisture within the tubes, and use of excessivehydrophilic indicator. Potential errors associatedwith high air-sea temperature gradientswere not realized in our tests.With careful constructionand deployment the gaugesshould provide accurate depth estimates without adverselyaffecting free-living animals. Methods for attachingthe gaugesto birds are reviewed.

INDICADORES DE PROFUNDIDAD DE TUBOS CAPILARES PARA ANIMALSQUE SESUMERGEN: UNA EVALUACIONDE SU EXACTITUD Y APLICABILIDAD Resumen.--Indicadorespara medir la profundidadm•txima 1ograda pot animalesque se sumergenfue construidade tuberla pl•tsticarevestida con un indicadorsoluble en polvo,a un costode $0.10 cadauna. En nuestraspruebas, las diferenciasentre profundidadesreales y estimadaspromedi6 < 3% despuesde sumergitlos indicadores una solavez a profundidades de hasta 140 m. Con multiples inmersioneslos erroreseran usualmente<10% y siempre <25%. La exactitudde los indicadoresno fue afectadapot la profundidado pot la duraci6n de la inmersi6n,raz6n de descensoo movimientosdebajo del agua simulandolos de un ave nadando.Los erroressurgieron pot sacudidasseveras de losindicadores una vez sumergidos, clavados,acumulaci6n de humedad dentro de los tubos, y uso excesivodel indicador hi- drofilico.Los errorespotenciales asociados con los gradientesde alta temperatufamar-aire no se consideraronen nuestraspruebas. Con la construcci6ny usocauteloso, los indicadores deben proveer estimadosde profundidad exactossin afectar adversamentelos animales de prueba.Los m6todospara colocarlos indicadoresalas avesrueton revisados. The underwaterforaging activities of diving animalsare poorlyknown and difficultto study,but are obviouslyof greatimportance in their lives. We report testson a simplegauge that measuresmaximum divingdepths and is a usefultool in ecological,behavioral, and physiologicalstudies of theseanimals. Capillary-tube depth gaugeshave been reliably usedfor many yearsby SCUBA divers (Miller 1979). The gaugeis a calibrated plastictube with a narrow, uniform bore open to the water at one end. As the depth increases,water is forced into the tube and the depth can

• Currentaddress: Biology Department, University of Victoria,Victoria, British Columbia,V8W 2Y2, Canada. 2 Currentaddress: Institut f•'r Meereskunde,Universit•t Kiel, D•isternbrookerWeg 20, D 2300, Kiel, WestGermany.

345 346] A. E. Burgerand R. P. Wilson j. FieldOrnithol. Autumn 1988

be read off the calibrations at the air-water interface. The device has beenmodified, by lining the lumen with water-solubleindicator dye or powderto functionas a maximum depthgauge. The amountof indicator remaining in the tube upon its recoveryshows the minimum volume of the air spaceand hencethe maximumdepth attained. These gaugeshave beenused to demonstrateremarkable diving abilitiesof free-livingpen- guins (Adamsand Brown 1983, Montague 1985, Kooymanet al. 1971) and auks (Burger and Simpson1986). Somefeatures of the gaugeswere analyzed in these reports, but the present paper is the first rigorous assessmentof their accuracyand applicability.

FUNCTIONAL PRINCIPLES, SENSITIVITY, AND POTENTIAL ERRORS Althoughcommonly known as capillarydepth gauges, maximum depth gaugesdo not functionby capillarity. The compressionof the air space within the tube follows the general gas law: P•V1 P2V2 -- - (1) T• T 2 where P, V, and T representpressure (in atmospheres),volume (cm3) and temperature(øK), respectively.The effectsof temperaturedifferen- tials are consideredlater in this paper and can be omittedat this stage. Sincethe diameterof the deviceremains constant, the equationcan be modifiedto determinechanges in the length (L) of the air columnin the tube, as follows: PsLs= PdLd (2) where Ps is the pressureat the water surface(1 Arm), Ls is the original lengthof the air columnat the surface,and Pd and Ld are the pressure and lengthof air columnat depthd, respectively.The lengthof air space remaining (Ld) at depth d can be determinedfrom the length of undissolvedindicator (Fig. 1). In seawater,pressure increases by I Arm for every 10.08 m increase in depth,in additionto the I Arm at the surface.The pressureat depth d is thus I + d/10.08. Equation 2 can be modifiedto predictthe depth attained: d=10.08(L • - 1) (3) Given constanttemperatures, the behavior of the air and water in the tube shouldbe highly predictable,producing a curvilinearrelationship betweend and Ld (Fig. 1). The gaugesare clearlymore sensitive to depth changesat shallowdepths. For example,a I m increasein depthproduces a 2.4 mm changein Cd in a 100 mm gaugeat 10 m, but only 0.3 mm at 50 m. The successof the gaugedepends on factorsaffecting the air spacein the tube and errorsmay result from temperaturechanges, accumulation of water dropletsin the lumen, and use of excessiveindicator. We read Vol.59, No. 4 MaximumDepth Gauges for Diving Birds [347

150

lOO

o o 50 lOO 15o DEPTH (m) FIGURE1. Changesin the lengthof undissolvedindicator (Ld, seetext) of maximum depth gaugesat depths0-140 m. Solidand dashedlines indicate the Ld predictedfrom equation 3 for gauges145 and 100 mm long, respectively.Dots show actual means (_+SD) of readingsfrom 10 gaugesof each type lowered to the depths indicated.The sketches showthe penetrationof water into the gaugesat 10 and 140 m, and a gaugetaped to the dorsal plumage of a RhinocerosAuklet Cerorhincamonocerata.

our gaugeswith a maximum precisionof 0.5 mm. The potential errors producedat this scalewere acceptablewithin the depth range of most birds and smaller marine mammals.For example,with a 100 mm long gauge,an estimateddepth of 100 m couldresult from real depthsbetween 94-106 m (Fig. 2).

METHODS FOR TESTING DEPTH GAUGES Depth gaugeswere made from lengthsof flexible plastic Tygon (R) tubing, coatedinternally with a thin layer of icing sugar.The indicator was applied by blowing gently throughthe open tubing to moistenthe lumenwalls and then suckingvery smallamounts of dry icingsugar up the tube. The tube was knottedtightly at one end, without stretchingit, and cut to size. Except where noted,the tubing had an internal diameter of 1.6 mm and the lumen length was either 100 or 145 mm. Sample gaugeswere lowered into seawater at Bamfield, British Columbia. The lengthof undissolvedindicator (Ld) wasread to the nearest0.5 mm using a ruler, without stretchingthe tubing. To simulatethe activitiesof a bird makingmultiple diveswithin a day, a sampleof 10 devices(100 mm long)was lowered to 10 m in fiveepisodes, each separatedby 2-4 h. Each episodeinvolved 50 submersionswhere 348] A. E. Burgerand R. P. Wilson j. FieldOrnithol. Autumn 1988

FIGURE2. Relationship between real depth and estimateddepth, showing the potential error resultingfrom readingprecision of 0.5 mm in a 100 mm gauge(shaded portion). The dottedline showsthat with this potential error, an estimateddepth of 100 m could result from real depthsof 94-106 m.

the deviceswere submergedfor 45-50 s and at the surfacefor 10-15 s. This schedulesimulated the activitiesof diving seabirds,such as Common Murres (Uria aalge),which spendabout 7% of the day diving (Cairns et al. 1987) and have a dive: pauseratio of 3.6:1 (Dewar 1924). The effectsof locomotionof an animal were simulatedby pulling and jarring ten 100 mm gaugesfor 30 s at the end of a hand line at 10 m depth. This was done both mildly, to simulate normal locomotion in smallermarine mammalsor birds, and violently,to the maximum degree possibleby hand, to simulatethrashing movements in a large mammal. Controlswere simply held at 10 m for 30 s. To test the effectsof plunge- diving by birds, we droppedsamples of 10 gaugesfrom 8-10 m to the seasurface and allowedthem to submergeto 5 and 10 m deep,in separate trials. Controls were lowered from the sea surface. To testthe effectsof temperature,gauges were placedon blackplumage of a stuffedCommon Murre in full sun for 7 h, followedby submersion in cool seawater.Controls were placed in shadedair. Temperatures in gaugeswere measuredwith a Bailey Thermalert TH-6D thermometer, using a thin wire thermistor inserted2-3 cm down the lumen. Air and sea surfacetemperatures were made with the samethermistor and with a mercury thermometer. Vol.59, No. 4 MaximumDepth Gauges for DivingBirds [349

RESULTS

Accuracyof gauges in a singletest dive.--When submergedto eachdepth once,the gaugeswere highly accurate,with virtually no differencesbetween the readingsobserved and thosepredicted from equation 3 (Fig. 1). The co-efficientof variation amonggroups of ten sampleswas always <2%. The differencesbetween the estimated and real depths averaged 2.4% (range 0.8-5.0%) and 2.6% (0.0-7.2%) for 100 and 145 mm tubes, respectively.Gauges with internal diametersof 1.6 and 0.8 mm were equally accurateat all depths. The effectsof multiplesubmersions and simulatedlocomotion.--The differencesbetween the real and estimateddepths increasedprogressively with increasingfrequency of dives (Fig. 3). These deviationswere the result of small droplets of water remaining on the lumen wall after surfacing,progressively reducing the available air space.The errors in depth estimatesappeared to reach an asymptoteafter 120 dives and remainedwithin 24%. When subsequentlysubmerged for singledives to deeper depths (20, 30, and 35 m), these errors decreasedsubstantially and were usually < 10% (Fig. 3). The mean depth (10.8 _+0.8 m [SD]) predictedby 10 gaugessubjected to mild pulling and jarring, to simulatelocomotion in an animal at 10 m depth,did not differ significantly,after 50 dives,from the depthpredicted from controls(10.9 _ 0.7 m, t = 0.297, ? > 0.05). Similar gauges(internal diameter1.6 mm) subjectedto violentpulling and jarring, however,showed significantlydifferent depths (11.6 _ 0.5 m, n = 10) to controls(10.5 _ 0.2 m, t = 6.46, df = 18, ? < 0.001) after a singledive, due to disruption of the air and water columnsin the tubes.Narrow tubes(internal diameter 0.8 mm) were lessaffected than wider onesby this treatment(mean depth shownafter jarring 10.8 + 0.4 m; control 10.3 + 0.2 m; t = 3.54, df = 18, P < 0.01). Gauges subjectedto simulated plunge-divingshowed significantdeviations from controls,but errors did not increaseafter five plunges.On average, gaugesoverestimated depths by 39% and 9% in submersionsto 5 and 10 m, respectively,after 10 plunges. Effectsof duration of divesand rate of descent.--Readingsof L d and resultantdepth estimatesdid not differ significantlyin gaugesheld at 50 m for periodsof 0.5-8.0 min (Table 1), indicatingthat thesedevices were not significantlyaffected by the dive durationslikely to occurin smaller marinevertebrates. The % error was similarto that expectedfrom reading imprecision(see Fig. 2). Gauges were lowered to 50 and 70 m both rapidly (1.1-1.3 m s-1) and slowly (0.4 m s-l) with no significantdif- ferencesin the resultantreadings, or predicteddepths (Table 2). Rates of ascent,during which water containedin the tubesis forcedout, should not affect the positionof the deepestminiscus recorded by the indicator. Effectsof temperatureand condensation.--Ghangesin temperature(AT) within the gaugescan induceerrors, either by affectingthe volumeof air in the tube, or by causing condensationof water in the lumen. The potentialeffects of AT on the internal volumeof the gauges,and hence •o] A. E. Burgerand R. P. Wilson j. Field Ornithol. Autumn 1988

30-

n-20 - - - 30M

•o 20M • 35M

I 30 60 90 120 NO. OF DIVES TO 10M

FIGUgE3. Effectsof multiple submersionsto 10 m on a 100 mm gauge,followed by single submersionsto 20, 30, and 35 m. Open circlesindicate % differencesbetween observed and expectedreadings of Ld. Dots, bars and vertical lines indicate the mean, SD, and range of % differencesbetween real and predicteddepths. The horizontal dashedline showsthe minimum error resultingfrom reading Ld with a precisionof 0.5 mm.

La, were calculatedfrom equation 1, and for temperaturesof 0-40 C (273-313 øK): ALa = 0.0034 AT (4) The effectsof AT on depthestimates depend on the depthattained, since the relationshipbetween d and La is not linear. For example,a 20 C decreasein a 100 mm gaugecould cause a decreasein La of 7% (equation 4), resultingin potentialoverestimations in d of 15% and 9% at 10 and 50 m depth,respectively (calculated using equation 3). Temperaturevariations within a water columnare generallysmall and unlikely to affect depth estimates,but more seriouserrors are possible due to large temperature gradientsoften found between air and water. The air within transparenttubes heats up in sunshinedue to a "green- house"effect, and this is exacerbatedby contactwith plumage(Table 3). The gaugeshave little thermal massand cool rapidly when placedin water. In our samplesthe averageinternal temperaturesdropped from 31 C to 16 C within 30 s of submersion.Air within gaugesmoving from hot air to coolsea will contract,potentially producing an overestimateof the maximum depth attained. These potential errors were not realized in our tests:gauges lying on blackplumage in the sun for 7 h with internal temperaturesat least 25 C higher than controlsin the shade, did not differ from controlswhen submergedin cool seawater (Table 3, t-tests, P > 0.05). Depths estimatedfrom actual readingswere much closerto the real depth than those predictedwith temperatureinduced errors. Other factors,such as expansionand contractionof the plastic,might havenegated the changesin the air spaceinduced by temperaturechanges. Internal condensationcaused by rapid coolingof hot air, with a high relativehumidity due to proximity of the sea, is anotherproblem likely Vol.59, No. 4 MaximumDepth Gauges for DivingBirds [351

TABLE1. The effectof dive durationat 50 m depth with maximum depthgauges of length 100 mm and 145 mm.

Time at 100 mm gauges 145 mm gauges depth (min) Depth predicted(m) Error • (%) Depth predicted(m) Error (%) 0.5 51.0 + 3.1 2.0 51.2 + 1.9 2.4 1.0 51.4 + 4.0 2.8 51.6 + 2.0 3.2 2.0 51.4 + 4.7 2.8 52.5 + 2.4 5.0 4.0 50.6 + 3.2 1.2 53.2 + 3.1 6.4 8.0 50.2 + 3.8 0.4 53.5 + 3.4 7.0 ANOVA F = 0.188, P > 0.05 F = 1.426, P > 0.05

• Deviation betweenthe real depth and the depth predictedfrom the indicatorreading using eq. 3.

to occurin hot areas.For example,air at 100% relative humidity contains up to 168 g/m 3 of water at 60 C, but only 9 g/m 3 at 10 C (Jorgensen 1979). Thus, a 100 mm long gauge,with an internal diameterof 1.6 mm (volume ca. 200 mm3), at 60 C and 100% R.H. contains3.4 x 10-5 g of water vapor.When immersedin water at 10 C, 95% of the water vapor will condense(i.e., 3.2 x 10-5 g). If all the condensedwater remainsin the tube, approximately 1 mm3 of water (causinga 0.5 mm error in Ld) will condense in the lumen after 30 dives. Under extreme conditions the dropletsmay coalesceand run down the tube dissolvingthe indicator; this was experiencedwhen deployinggauges mounted on dark-plumaged cormorantsin southernAfrica (Wilson,pers. obs.). Condensation problems can be reducedby minimizingthe periodof deploymentand by attaching gaugeson the undersidesof the birds to reduceinsolation.

DISCUSSION Our testsrevealed several real or potential sourcesof error associated with maximum depthgauges. Accumulation of water dropletswithin the lumen, through condensation,prolonged deployment, or useof excessive hydrophilicindicator, can lead to significanterrors. These, and errors resultingfrom imprecisionin reading Ld, will have progressivelylarger effectsas depth increases.The probability of sucherrors can be reduced throughminimizing the amountof hydrophilicindicator used; minimizing deploymenttimes; reducing condensation by mountingdevices where they will be shadedby the animal's body;using longer tubesfor deep diving animals;and maximizingreading precision. Inspection of recoveredgauges will usuallyreveal evidence of moistureaccumulation and affectedgauges can be discarded. The mixing of the air and water columnswithin the gauge,resulting from severejarring, causedsignificant errors and might precludethe use of the techniqueon largeseals or whales.The useof narrow tubesreduced sucherrors. Underwater locomotionof birds shouldnot affectthe accuracy 352] A. E. Burgerand R. P. Wilson J.Field Ornithol. Autumn 1988

TABLE2. The effectof the rate of descenton depthsestimated by maximum depth gauges of length 100 and 145 mm (n = 10 for each treatment).

Gauge length Depth Rate of descent Mean (_SD) predicted Paired t-test (mm) (m) (m. s-l) depth (m) (t) 100 50 1.3 52.9 + 5.5 1.500, 100 50 0.4 50.4 + 3.4 NS

100 70 1.1 69.6 + 4.3 0.318, 100 70 0.4 69.2 + 6.4 NS 145 50 1.1 49.7 + 1.2 0.658, 145 50 0.4 49.4 + 0.8 NS 145 70 1.2 73.3 + 4.2 0.286, 145 70 0.4 73.9 + 3.1 NS

NS = No significantdifference (P > 0.05). of maximum depthgauges, but gaugesused on plunge-divingbirds should be testedthoroughly for potentialerrors. Repeatedsubmersions to the samedepth led to overestimatesof maximum depths,and this might affectresults obtained from animalsdiving regularly to fixed depths,such as obligatebottom feeders. The size of error dependson the frequencyof diving and the depth attained. Max- imum depth estimateswill be more accuratewith epipelagicand mid- water foragers,which tend to make infrequent forays into deeperdepths. Errors in depth estimatesaveraged <3% with single immersionsto any depth,and with multiple immersionswere usually < 10% and almost always <25%. The accuracyof the gaugeswas not affectedby depths per se,the rate of descentor the duration of submersion,within the ranges likely to be encounteredin birds or smaller marine mammals. Errors related to large air-water temperature gradients,although potentially serious,were not significant in our tests. In general we feel confident that, with proper attention to constructionand deployment, maximum depth gaugesare reliable and accurate. Most of our findings probably apply to gaugesof differentdiameters and lengths,but we urge researchers to do their own tests,ideally with the same air and sea temperatures likely to be encounteredby the animal being studied. Maximum depth gaugeshave been deployedusing harnesses (Adams and Brown 1983), by suturing the devicesto the skin (Kooyman et al. 1971), and attachedto flipper bands (Montague 1985) and leg bands (Burger and Simpson1986). We found waterproof adhesivetape (Su- perstikbrand) ideal for attachinggauges to the contourfeathers of alcids (Fig. 1). Gauges not recoveredfall off as the tape gradually losesits adhesiveness or the bird molts. Other devices have been attached to the feathersof penguinsand the fur of sealsusing small hoseclamps or glue. Harnessesand other bulky attachmentsare unsuitablefor aquaticbirds, sincethey affect the birds' hydrodynamicstreamlining and foragingef- ficiency(Wilson et al. 1986). Very long gaugesneeded for deepdiving Vol.59, No. 4 MaximumDepth Gauges for DivingBirds [353

TABLE3. The effectsof solarradiation on internal temperaturesand subsequentaccuracy of maximumdepth gauges after submersionto 50 m in seawaterat 12 C. The means _ SD are shown, and n = 5 gaugesin each case.

Gauge length (mm) 100 100 145 145 Treatment prior to On black Shaded On black Shaded submersion: plumage in air plumage in air in sun in sun Internal temperatures(øC) 56 _+ 6 23 _ 1 54_+ 3 25 _+ 1 AT gauge-sea(øC) 44 11 42 13 Following dive: Depth predictedif AT had full effect (m) • 60.5 52.4 59.7 52.4 Depth predictedfrom actual Ld reading(m) 49.2 + 0.8 49.5 + 1.0 51.6 + 4.0 49.8 + 2.7

• Using eq. 4 to calculateeffect of AT on Ld and eq. 2 to calculatethis effecton estimate of d (withoutany temperatureeffect, Ld is 16.7 and 24.2 mm, for 100 and 145 mm gauges, respectively,at 50 m).

speciescould be coiledand mountedon a small disc,and tapedor glued flush with the body contour.

CONCLUSIONS Maximum depth gaugesare useful toolsfor the study of underwater foragingof birdsand marine mammals. They are unbreakable,very small and light (our 100 mm gaugeshad a mass< 1 g), inexpensiveand easily constructed.Their flexibility and small size ensure that the hindrance associatedwith larger devices(Wilson et al. 1986) will be avoided.Even if the probabilityof recoveryfrom free-living animals is very low, the low cost and neglible impact on the animals makes this a very viable technique.Attaching or detachinga gaugeto a capturedanimal should be as quick as conventionalbanding or taggingprocedures, and no more stressful to the animal. Thesegauges provide only a singlepoint estimateof the deepestdive. Among epipelagicand mid-water foragersthese might representrare exploratorydives beyondthe animals' normal foraging depths.More elaboratedepth gauges, using pressure transducers (Kooyman et al. 1983), autoradiography(Wilson and Bain 1984) or light-emittingdiodes (R. P. Wilson et al., unpubl.), providemore comprehensivedata by estimating the time spentat eachdepth. These gaugesare, however,far more ex- pensive,heavier and bulkier than maximum depth gaugesand demand more time to deployand interpret the data. Maximum depth gauges effectivelygive an initial insightinto an animal'sunderwater capabilities and couldenhance studies of diving physiology,habitat use and foraging ecology.Where very low recoveryrates preclude the useof moreexpensive devices,or where the birds are too small to carry more elaboratedevices, maximum depth gaugesmay representthe only viable techniqueto mea- sure diving depths. 354] A. E. Burgerand R. P. Wilson J.Field Ornithol. Autumn 1988

ACKNOWLEDGMENTS

We thank David Duncan, Suzanne Dionne, Don Garnier, David Reid, and Marie-Pierre Wilson for assistancein testingthe devices,Dr. George Bourne for the loan of equipment and Drs. D. K. Cairns, D.C. Duncan, W. A. Montevecchi, and D. Schneiderfor valuable commentson the manuscript.This work was supportedby grantsto A.E.B. from NSERC Canada and the Western Canadian Universities Marine BiologicalSociety.

LITERATURE CITED

ADAMS,N.J., ANDC. R. BROWN. 1983. Diving depthsof the GentooPenguin (Pygoscelis papua). Condor 85:503-504. BURGER,A. E., ANr•M. SIMPSON.1986. Diving depthsof Atlantic Puffins and Common Murres. Auk 103:828-830. CAIRNS,D. K., K. A. BREDIN,AND W. h. MONTEVECCHI. 1987. Activity budgetsand foragingranges of breedingCommon Murres. Auk 104:218-224. DEWAR,J. M. 1924. The bird as a diver. Witherby, London. JORGENSEN,S. E. (ed.) 1979. Handbookof environmentaldata and ecologicalparameters. PergamonPress, New York. KOOYMAN,G. L., C. M. DRABEK,R. ELSNER,AND W. B. CAMPBELL. 1971. Diving behaviorof the Emperor Penguin (Aptenodytesforsteri).Auk 88:775-795. , J. 0. BILLt•PS,AND W. D. FARWELL. 1983. Two recentlydeveloped recorders for monitoringdiving activity of marine birds and mammals.Pp. 197-214, in A. G. Macdonaldand I. G. Priede,eds. Experimental biology at sea.Academic Press, London and New York. MILLER, J. w. (ed.) 1979. NOAA diving manual: diving for scienceand technology,2nd ed. U.S. Dept. Commerce,National Oceanic and AtmosphericAdministration. MONTAGUE,T. L. 1985. A maximum dive recorderfor Little Penguins.Emu 85:264- 267. WILSON,R. P., ANDC. R. BAIN. 1984. An inexpensivedepth gauge for penguins.J. Wildl. Manage. 48:1077-1084. •, W. S. GRANT,AND D.C. DUFFY. 1986. Recordingdevices on free-rangingmarine animals:does measurement affect foragingperformance? Ecology 67:1091-1093. Received2 Jun. 1987; accepted22 May 1988.