gy and Ecology 354 (2008) 81–92www.elsevier.com/locate/jembe

Journal of Experimental Marine Biolo

Early life-history processes in benthic octopus: Relationshipsbetween temperature, feeding, food conversion,

and growth in juvenile Octopus pallidus

Jessica André a,⁎, Gretta T. Pecl a, Jayson M. Semmens a, Eric P.M. Grist b

a Tasmanian Aquaculture and Fisheries Institute, Marine Research Laboratories, University of Tasmania,Private Bag 49, Hobart, Tasmania 7001, Australia

b CSIRO Marine and Atmospheric Research, GPO Box 1538, Hobart, Tasmania 7001, Australia

Received 6 July 2007; accepted 21 October 2007

Abstract

Initial growth in cephalopods is exponential, making early life-history critical in determining growth trajectories. Few captivestudies have however examined the early life-history of cephalopods in fluctuating temperatures as would be encountered in thewild. This study investigates the relationship between early growth and the significant factors affecting growth, namely food intake,food conversion and fluctuating environmental temperatures. Pale octopus (Octopus pallidus) hatchlings were reared in captivityunder either a warming or cooling temperature regime. Individual variations and periodicity in feeding rates (Fr), food conversionrates (Cr), growth rates (Gr), and the relationship between these variables and temperature were examined weekly. Food conversionrates were variable between individuals but also within individual octopus, which exhibited large fluctuations in Cr over time,exceeding 100% d−1 in one instance. Although individual Fr, Cr and Gr displayed fluctuations over time, there was no evidence ofperiodicity for any of the variables. Changes in temperature were not significantly correlated to changes in Fr, Cr or Gr. Feedingrate did not appear to influence growth rate or food conversion rate. Food conversion rate was negatively correlated to feeding ratein the same week, and positively correlated to growth rate. Short periods of low or no food consumption were common and thehigh values obtained for food conversion rate for some individuals suggest that octopus can grow substantially with little or no foodintake. Individual variability observed in octopus growth may be dependent on the growth mechanism involved, specifically a finebalance between the continuous hyperplasic and hypertrophic growth found in cephalopods.© 2007 Elsevier B.V. All rights reserved.

Keywords: Cephalopod; Food consumption; Individual-based; Juvenile; Physiology

1. Introduction

Octopus are characterised by rapid non-asymptoticgrowth, which is generally best described by exponentialor power models (Van Heukelem, 1973; Forsythe, 1984;

⁎ Corresponding author. Tel.: +61 3 6227 7265; fax: +61 3 6227 8035.E-mail address: jandre@utas.edu.au (J. André).

0022-0981/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.jembe.2007.10.008

DeRusha et al., 1987; Semmens et al., 2004). Although itis yet to be demonstrated for wild animals, individualsreared in captivity generally exhibit two-phase growthconsisting of an initial exponential growth stage followedby slower growth (Forsythe and Van Heukelem, 1987). Incommon with other cephalopods, individual growth inoctopus is highly variable, even within groups of siblingsreared under identical conditions (Forsythe and Van

82 J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

Heukelem, 1987). Numerous biotic and abiotic factorscan influence growth, including temperature (Forsytheand Hanlon, 1988; Forsythe, 1993; Pecl, 2004), foodquality and quantity, age, size, gender, stage of maturityand level of activity (Forsythe and Van Heukelem, 1987).In the juvenile phase, temperature (Forsythe and VanHeukelem, 1987), food ration (Villanueva et al., 2002),food quality (Villanueva, 1994) and hatching size(Leporati et al., 2007) appear to have the most significantimpact on growth. Higher temperatures result in highergrowth rates for octopus with both benthic (Forsythe andHanlon, 1988; Segawa and Nomoto, 2002) and plank-tonic (Itami et al., 1963; Villanueva, 1995) hatchlings.However, most studies have only used one or severalfixed temperature regimes to investigate the impact oftemperature on growth, with only Leporati et al. (2007)having explored the effect of seasonal temperatures onthe growth of octopus hatchlings. There is in general littleinformation regarding the impact of dynamic tempera-tures, as would be expected in nature, on the physiologyof cephalopods.

Variability in individual food conversion may alsocontribute to differences in individual growth rates. Foodconversion (or gross growth efficiency) is highly variablebetween individuals even for octopus reared on the samediet (Mangold and von Boletzky, 1973). High levels ofactivity, low food intake (Wells and Clarke, 1996) andsexual maturity (Mangold, 1983a; Mangold, 1983b;Klaich et al., 2006) are known to lower food conversionsince less energy is available for growth. Food conver-sion, however, appears independent of sex (Hanlon, 1983;Forsythe, 1984), body size (Mangold, 1983a), and tem-perature (Mangold and von Boletzky, 1973; Mangold,1983a), although a recent study by Klaich et al. (2006)suggests that temperature does affect food conversion.Gross growth efficiency appears variable at the individuallevel, exhibiting apparent periodic fluctuation over time insome species (Mangold and von Boletzky, 1973;Mangold, 1983a; Mangold, 1983b) although the causeof these fluctuations has not been established.

The objective of this study was to examine therelationship between growth in juvenile octopus andother significant factors affecting growth, namely foodintake, food conversion and temperature. Laboratoryinvestigations were undertaken using the medium-sized(maximum 1 kg) pale octopus (Octopus pallidus) as amodel species, as it is commercially exploited and eggsare easily obtained. The eggs ofO. pallidus hatch out intofully functional hatchlings (as opposed to paralarvae),which immediately settle on the benthos. Individualfeeding rate, food conversion rate, and growth rate weredetermined for pale octopus hatchlings reared under

identical nutritional conditions with two differentdynamic temperature regimes. As octopus grow expo-nentially, the early life-history is critical in determiningtheir growth trajectory and there has been very little focuson this aspect of their life cycle, except in a fewcommercial species (Segawa and Nomoto, 2002; Iglesiaset al., 2004). This study aims to improve the currentunderstanding of the growth processes in juvenileoctopus.

2. Materials and methods

2.1. Study animals and experimental design

Two brooding females (designated as females A andB) were collected in March 2005 from the commercialpot fishery in north-west Tasmania, Australia (40°49.268; 145° 39.774 west and 40° 50.240; 145° 42.091east, 45 meters depth). Animals were maintained inseparate 250 l tanks in an indoor system at ambient seatemperature (22–11 °C from March to July) until theeggs hatched. The first hatching occurred 123 days aftercollection.

After an acclimatisation period of 24 h, hatchlingswere placed in individual 2 l containers fitted with meshsides to allow water flow and a scallop shell for shelter.The containers were kept in 250 l stock tanks under afluorescent lighting regime which replicated naturaldaylight variation (06.00–18.00 h light, 18.00–06.00 hdark). Two temperature regimes were established: awarmer temperature regime (increasing from 16 °C to18 °C over a period of 36 days then stable at 18 °C for107 days) and a cooler temperature regime (decreasingfrom 16 °C to 14 °C over a period of 36 days then stableat 14 °C for 107 days). Temperature was altered by 1 °Con day 18 and day 36 of the experiment (Fig. 1), andtemperature in the tanks was recorded every 15 min bytwo data-loggers (StowAway Tidbit, USA, http://www.onsetcomp.com).

Four randomly selected day-old hatchlings (two fromfemale A and two from female B all born on 19/07/05)were held under each temperature regime. Individualswere designated as octopus Cool1 to Cool4 andWarm1 toWarm4 according to their temperature regime (Table 1).Growth and food intake was individually monitored for aperiod of 143 days.

2.2. Growth, food consumption, and food conversionrate

Individuals were weighed every 5 to 10 days and theinstantaneous growth rate (Gr), expressed as the percent

Fig. 1. Intended (- - -) and actual (—) temperature regimes recorded during the experiment. Standardised (std) temperature used in the time seriesanalysis are represented by circles for the cool regime and squares for the warm regime.

83J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

increase in body mass per day, was calculated using thestandard exponential growth formula:

Gr ¼ ln B2 � ln B1ð Þ= t2 � t1ð Þð Þ � 100 ð1Þ

where B1 and B2 are body masses (g) at time t1 and t2(Forsythe and Van Heukelem, 1987).

To describe growth as a function of age, linear (Y=a+mX), exponential (Y=aemx) and power (Y=aXm) functionswere fitted to the data sets. The best fit model was chosenon the basis of the highest adjusted r2 values (DeRushaet al., 1987; Hatfield et al., 2001). Adjusted r2 values werecalculated by linear regression to B versus t for the linearmodel, log(B) versus t for the exponential model, and log(B) versus log(t) for the power model. Average growth foreach temperature treatment was calculated as the mean ofindividual growth equation parameter values.

All individuals were fed porcelain crabs (Petrolistheselongatus) collected from around the Hobart area(Tasmania). Two live crabs, whose relative body weighttotalled between 4% and 12% of the octopus bodyweight, were supplied daily to each animal. Beforebeing fed to each individual octopus, crabs were dried

Table 1Individual IDs, lineage (mother A or B), hatch weight, survival time and tempexperiment for 143 days

ID Lineage

Cool (16–14 °C) Cool1 BCool2 BCool3 ACool3 A

Warm(16–18 °C) Warm1 BWarm2 BWarm3 AWarm4 A

Warm4 was kept another 45 days beyond the experiment.

with absorbing paper and weighed on a digital scale to0.001 g accuracy. Uneaten crab remains from theprevious day were removed from each container andfrozen immediately for later analysis. Any live uneatencrab was removed and weighed.

After defrosting, remains were washed three timeswith ammonium formate (0.5 M) to remove any saltsbefore being dried in an oven for 24 h at 60 °C. Thesamples were then placed in a desiccator for 1 h toremove any moisture and weighed on a digital scale to0.0001 g accuracy. Wet weight and dry weight in gramswere obtained for 30 randomly selected samples ofremains in order to calculate the wet weight to dryweight conversion factor.

The quantity of food consumed in a given day (D)was:

D ¼ wf � wu � wr ð2Þ

where wf denotes the wet weight (g) of live crabs fed tothe octopus, wu denotes the wet weight (g) of any livecrabs uneaten the next day and wr denotes the wetweight (g) of remains from eaten crabs.

erature treatment for the eight Octopus pallidus hatchlings reared in the

Hatch weight (g) Survival (days)

0.281 1430.251 1430.224 1430.258 1430.263 570.303 1430.235 940.206 188

Fig. 2. Schematic diagram of the relationships between temperature (T), feeding rate (Fr), growth rate (Gr) and food conversion (Cr) investigated inthis study.

Fig. 3. Standardised feeding rate Fr and growth rate Gr (in % body weight per day) experienced by individual octopus. Note the change of scale forWarm4.

84 J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

Fig. 4. Autocorrelation plots of the standardised feeding rate Fr (in %body weight per day) under a) the cool temperature regime (n=4), andb) the warm temperature regime (n=2). Bars that protrude beyond thedotted lines indicate significant correlations. Significance level=5%.

85J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

The weight specific feeding rate (Fr) expressed as apercentage of bodymass per daywas calculated accordingto Houlihan et al. (1998):

Fr ¼ Dt= Bm � tð Þð Þ � 100 ð3Þ

where t is the number of days between two weighings,Dt

is the amount of food (g wet weight) consumed over thetime interval t, and Bm is the mean body weight (g wetweight) of the octopus over the time interval t.

Food conversion (%) expresses the amount of foodintake required to fuel a unit amount of growth. To explorethe relationship with Fr and Gr, we have expressed foodconversion as a rate in percentage per day (Cr), which wecalculated according to amodification of theMangold andvon Boletzky (1973) formulae:

Cr ¼ DBð Þ= Dt � tð Þð Þ � 100 ð4Þ

where ΔB is the difference in body mass (g) between twoweighings, t is the number of days between twoweighings, and Dt is the amount of food (g wet weight)consumed over the time interval t.

2.3. Data analysis

For all averages given, a corrected standard errorSE⁎= t×SE, where t is the t-score, was used wheresample size was smaller than 20 (Fowler et al., 1998).Significance level throughout the analysis was set atpb0.05. Time series analysis was performed onindividual feeding rate (Fr), growth rate (Gr), foodconversion rate (Cr), and temperature (T) data. Weightswere obtained every 5–10 days and the resulting timeseries were standardised in terms of 7 day time steps bythe use of linear interpolation. Accordingly, Fr, Gr, Cr

and T were calculated over the same time step. Timeseries exhibiting linear trends were detrended linearly.Time series (Fr,Gr and Cr) were first analysed separatelyusing autocorrelation plots to assist with identifying anycycles. Since data were in weekly time steps, lag 1corresponded to a time shift of 7 days, lag 2 a time shiftof 14 days and so on. Where a significant autocorrela-tion was established, the time series was furtherinvestigated through partial autocorrelation plots. Partialautocorrelation measures the strength of the correlationat specific lag (e.g. lag 4) while removing the effects ofall autocorrelations below that lag (i.e. autocorrelationoccurring at lags 0, 1, 2 and 3). Only autocorrelationanalyses that were supported by partial autocorrelationare presented. The relationships between time series T,Fr, Gr and Cr (Fig. 2) were examined separately using

cross-correlation and partial cross-correlation analyses.Partial correlation coefficients (r) were calculated whensignificant correlations were established. To assess thedirection of influence (e.g. if changes in x causedchanges in y, or the opposite), time series x were time-shifted against y both forward (showing the effect ofx on y, referred to as plots of x on y) and backward(showing the effect of y on x, referred to as plots ofy on x). Given the relatively short lengths of all timeseries (20 data points), cross-correlations and partialcross-correlation analysis beyond a lag 6 (42 days) wereconsidered to be redundant in view of the likelihood of

86 J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

type I and type II errors occurring at higher lags. Themean Fr, Gr and Cr time series were also calculated foreach temperature regime and analysed as previously forindividual time series. The software R version 2.2.1 wasused to carry out all the analyses.

3. Results

The mean hatch weight (g±SE⁎) was 0.231±0.034for hatchlings from female A (range=0.206–0.258 g,n=4) and 0.275 g±0.036 for hatchlings from female B(range=0.251–0.303 g, n=4). Octopus did not appearstressed (inking and jet escape movement were rarelyseen) and spent most of their time sheltered under thescallop shell. Little movement was observed and so thelevel of activity was considered minimal. Two octopusreared under the warm treatment (Warm1 and Warm3)died at 57 and 94 days respectively after hatching, for

Fig. 5. Standardised food conversion rate Cr (in % per day) and temperaturCr between day 136 and day 143 for Cool4.

unknown reasons. Data for these individuals were dis-carded because there were too few data points to es-tablish valid feeding and growth curves.

Temperatures in the two treatments fluctuated morethan originally intended due to mechanical failures of theheater/chiller units (Fig. 1). The treatments neverthelessfollowed the intended main trend of increasing (warm)and decreasing (cool) temperature with time. Theseshorter term temperature fluctuations were an integralpart of the subsequent time series analyses which wereperformed.

3.1. Food consumption rate, food conversion rate, andgrowth rate

All octopus started feeding within hours afterhatching, preying on the smallest size crabs (2 mmcarapace length). Five out of the six individuals exhibited

e experienced by individual octopus. It was not possible to calculate

Fig. 6. a) Autocorrelation plot of food conversion rate for Warm2;b) autocorrelation plot of growth rate for Cool4. Bars that protrude beyondthe dotted lines indicate significant correlations. Significance level = 5%.

87J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

at least one subsequent period of “starvation” (rangingfrom 1 to 7 days) in which they did not eat and lostweight. In general, daily food consumption D increasedwith body weight. The mean D (g wet weight±SE) forthe warm regime was 0.024±0.002 and 0.020±0.001 forthe cool treatment.

The mean feeding rate (% body weight per day±SE)was 2.87±0.16 for octopus reared under the warmtreatment (range=0.79–5.69% body weight day−1,n=40), and 2.45±0.12 for octopus reared under thecool treatment (range=0.10–5.93% body weight day−1,n=80). Although fluctuations showed some regularityin these standardised time series (Fig. 3), autocorrelationplots revealed no significant periodicity (Fig. 4).

Food conversion rates Cr were variable from one weekto the next (Fig. 5), ranging from −8.10 to 70.40% d−1

under thewarm regime (mean%per day±SE=9.57±1.93,n=40) and from −6.30 to 23.90% d−1 for octopus rearedunder the cool treatment (mean=8.61±0.67, n=79). Cr

could not be calculated for Cool4 between day 136 and143 since this individual consumed no food during thisperiod. Food conversion rate however is expected toexceed 100% d−1 for that period since substantial growthwas observed despite the lack of food intake. Onlyindividual Warm2 displayed evidence of periodicity infood conversion rate with a highly significant negativeautocorrelation at lag 1 (Fig. 6a).

Individual growth under both temperature regimeswasbest described by an exponential curve (Fig. 7) withadjusted r2 values for individual growth trajectoriesranging between 0.91 and 0.99. The mean best fit growthcurve was B=0.227e0.015t under the warm regime andB=0.264e0.013t under the cool regime. Octopus under thewarm regime (growth ratem=0.015) therefore on averagegrew faster than those under the cool regime (m=0.013).Warm4 exhibited a two-phase growth pattern in which astriking change in growth rate occurred at around133 days. The resulting two-phase growth curve is bestdescribed by the exponential equations: B=0.208e0.014t

(r2=0.987) from 0 to 133 days and B=0.3142e0.017t

(r2=1) from 133 to 143 days.In close agreement with the exponential growth rate

coefficients m obtained from the regression modelsdescribed above, the mean growth rate Gr (% bodyweight per day±SE) was 1.67±0.31 under the warmregime (range=−1.29–11.89% body weight day−1,n=40) compared to 1.38±0.10 under the cool regime(range=−0.57–3.68% body weight day−1, n=80).Although there appeared to be some regular fluctuationsin the growth rate time series of all individuals (Fig. 3),only Cool4 displayed a statistically significant periodicityin Gr, with a negative cross-correlation at lag 1 (Fig. 6b).

3.2. Effect of temperature on Fr, Gr and Cr

Overall, cross-correlation analyses suggested fluctua-tions in temperature did not drive changes in Fr, or Gr orCr. Analysis of the meanCr time series for the cool regimeshowed a significant positive cross-correlation between TandCr at lag 0 (r=0.64) (Fig. 8a), suggesting that changesin food conversion rates in the current week werecorrelated with fluctuations in temperature. This signifi-cant correlation was most likely influenced by individualCool4, which displayed a similar cross-correlation patternwith temperature (r=0.68 for lag 0) (Fig. 8b). The absenceof significant cross-correlations for the other five indivi-duals suggests that temperature had aminimal effect onCr.

Fig. 7. Raw data of individual (—) and average (- - -) growth under the a) cool and b) warm temperature regime.

88 J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

No significant cross-correlations were obtained betweenthe mean Fr time series and temperature, or betweenindividual Fr time series and temperature (Fig. 8c and d),suggesting that fluctuations in temperature had no effect onfeeding rate. Similarly, temperature fluctuations appearedto have little influence on growth rate fluctuations, asrevealed by the absence of significant cross-correlationsbetween mean Gr time series, or individualGr time series,and temperature (Fig. 8e and f).

3.3. Feeding rate and growth rate

Interestingly, fluctuations in feeding rate did notappear to drive fluctuations in growth rate as evidencedby the absence of significant cross-correlations of Fr

on Gr (Fig. 9a and b). Fluctuations in Gr however,appeared to influence fluctuations in Fr under the warmregime, with a significant positive cross-correlation ofGr on Fr at lag 1 (Fig. 9c). This significant correlationwas most likely influenced by individual warm2, whichdisplayed a similar cross-correlation pattern (Fig. 9d).The absence of significant cross-correlations for theother five individuals suggests that the overall effect ofGr fluctuations on changes in Fr were minimal.

3.4. Feeding rate and food conversion rate

Analysis of the mean times series showed asignificant negative cross-correlation at lag 0 for Cr onFr under the cool regime (r=−0.68) (Fig. 10a). Thissignificant relationship was mainly due to consistentnegative cross-correlations exhibited by all individualsfor Cr on Fr at lag 0, even though only that of Cool4 wasstatistically significant at this lag (r=−0.63) (Fig. 10c).The analysis for individual Warm4 also revealed asignificant negative cross-correlation at lag 0 (r=−0.79)(Fig. 10e), although no such pattern was evident fromthe analysis of the mean time series for the warm regime(Fig. 10d). Overall, individuals that demonstrated a highfood conversion rate tended to have a low feeding rate inthe same week, and vice versa. Fluctuations in foodconversion rate also appeared to drive fluctuations infeeding rate in subsequent weeks for some individuals.Analysis of the mean times series of Cr on Fr for thewarm regime revealed a significant positive cross-correlation at lag 3 (Fig. 10d). This was mainly due topositive cross-correlations at this lag exhibited byWarm2 and Warm4, even though neither of them wasstatistically significant. Analysis of the mean time series

Fig. 8. Cross-correlation plots showing the influence of temperature (T) onfeeding rate (Fr), growth rate (Gr) and food conversion rate (Cr): a)Ton themean Cr for the cool regime; b) T on Cr for individual Cool4; c) T on themean Fr for the cool regime; d) T on the mean Fr for the warm regime;e) T on the meanGr for the cool regime; f) T on the meanGr for the warmregime. Bars that protrude beyond the dotted lines indicate significantcorrelations. Significance level = 5%.

Fig. 9. Cross-correlation plots showing the relationship between feedingrate Fr and growth rate Gr: a) mean Fr on mean Gr for the cool regime;b) mean Fr on mean Gr for the warm regime; c) mean Gr on mean Fr forthe warm regime; d) Gr on Fr for Warm2. Bars that protrude beyond thedotted lines indicate significant correlations. Significance level = 5%.

89J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

of Cr on Fr for the cool regime did not reveal anysignificant cross-correlations although the analysis forindividual Cool3 revealed a significant positive cross-correlation at lag 4 (Fig. 10b).

Conversely however, the cross-correlation analysisof Fr on Cr revealed that there were no significantcorrelations for any individual, which suggests thatfluctuations in Cr were not determined by fluctuationsin Fr.

3.5. Growth rate and food conversion rate

The time series for Gr and Cr supported the existenceof a relationship between Gr on Cr with evidence of apositive cross-correlation occurring at a time lag ofunder a week (lag 0). All individuals but Cool4 hadstatistically significant cross-correlations at lag 0(0.85b rb0.97) (Fig. 11). The lack of a significant cor-

relation for this individual can be accounted for bythe fact that Cool4 exhibited an extremely highCr (238% d−1) and was hence an outlier. Individualcross-correlation plots of Gr on Cr and Cr on Gr did notshow any significant correlations other than at lag 0,suggesting that fluctuations in food convertion rate onlyhad an immediate impact on growth rate or vice versa.

4. Discussion

The relationship between Fr, Gr, Cr and T is highlycomplex with this detailed study revealing unexpectedresults such as the lack of a correlation betweenfluctuations in feeding rate and growth rate, and theobservation that dynamic temperatures have little effecton any of the variables examined. Strong individualvariability was evident and may in part be explained bythe dynamic growth process.

4.1. Food consumption, food conversion rate andgrowth rate

As for other octopuses reared in laboratory (Nixon,1966; Mather, 1980; Boyle and Knobloch, 1982;Houlihan et al., 1998), individuals did not feed everydaypotentially as a consequence of stress in captivity(Houlihan et al., 1998). Feeding rates were comparableto those of sub-adult or adult octopus from other species

Fig. 10. Cross-correlation plots showing the relationship between foodconversion rate Cr and feeding rate Fr: a) mean Cr on mean Fr for thecool regime; b) Cr on Fr for Cool3; c) Cr on Fr for Cool4; d) mean Cr

on mean Fr for the warm regime; e) Cr on Fr for Warm4. Bars thatprotrude beyond the dotted lines indicate significant correlations.Significance level = 5%.

Fig. 11. Cross-correlation plots of food conversion rateCr on growth rateGr for: a)Cool1; b)Cool2; c)Cool3; d) Cool4; e)Warm2; e)Warm2.Barsthat protrude beyond the dotted lines indicate significant correlations.Significance level = 5%.

90 J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

(Boyle and Knobloch, 1982; Mangold, 1983a; Houlihanet al., 1998; Segawa and Nomoto, 2002) but were muchlower (0.10 to 5.93% d−1) then feeding rates of 18.8 to32.3% d−1 found for captive juvenile O. ocellatus(Segawa and Nomoto, 2002), a similar benthic octopusto O. pallidus in size and life cycle. Unlike the presentstudy however, O. ocellatus were fed ad libitum whichcould account for the observed differences in feedingrates. Mean food conversion rates found for O. palliduswere higher than that reported for immatureO. tehuelchus(Klaich et al., 2006) another species with benthichatchlings (4.4% d−1 at 15 °C and 5.2% d−1 at 10 °C—calculated from the original food conversion in %). Foodconversion rates were highly variable on a weekly basis,withCr as high as 70.4% d−1. Similar weekly fluctuationshave been reported in previous studies (Mangold and vonBoletzky, 1973; Mangold, 1983a; Mangold, 1983b). Oneindividual displayed a Cr in excess of 100% d−1 whichsuggests that all the growth observed cannot be explainedby current or very recent food intake.

Growth trajectories were exponential throughout theexperimental period but animals did not display a classictwo-phased growth pattern (exponential growth phasefollowed by a slower growth phase) as observed for othercultured octopus species (Forsythe and Van Heukelem,1987; Semmens et al., 2004). Growth rates of 1.38% and1.67% for the cool andwarm treatment respectively werecomparable to the values of 1.43% (cool treatment) and1.73% (warm treatment) found by Leporati et al. (2007)for O. pallidus hatchlings reared under fluctuating tem-perature regimes similar to those employed in this study.

Food consumption, food conversion and growth ratesall exhibited fluctuations over time. These fluctuationsoverall showed little statistically significant periodicity ona weekly basis but may be linked to some underlyingphysiological growth mechanism, as discussed furtherbelow in Section 4.3.

4.2. Effect of temperature on Fr, Gr and Cr

Temperature is one of the most important abioticfactors affecting the physiology of poikilothermic

91J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

animals. The minimal effect between food conversionrate and temperature in the present study is consistentwith results of previous studies on cephalopods(Mangold and von Boletzky, 1973; Mangold, 1983a)which found that food conversion rate had nodependence on temperature. Temperature-dependenceof growth rate and feeding rate is on the other hand wellestablished for octopus, with higher temperaturesresulting in higher food intake (Mangold and vonBoletzky, 1973; Joll, 1977; Segawa and Nomoto, 2002)and growth (Semmens et al., 2004) within the thermalrange of a species. The general lack of statistical supportfor a relationship between Fr and temperature or Gr andtemperature in this study may be connected with acomplex response to the fluctuating temperature regime.Since maintenance costs are temperature-related (Wellsand Clarke, 1996), short-term fluctuating temperatureswould most likely produce variations in maintenancecosts that must ultimately result in variations in growth.This could obscure the direct dependence of growth ontemperature. The substantial variations observed in foodconversion rates could also have contributed to the lackof a strong relationship between growth and temperatureas suggested by Joll (1977). The pattern observed in thisstudy is however more likely to reflect the situation inthe wild since short-term environmental temperaturefluctuations would be frequently encountered by oc-topus, either from individuals moving between deep andshallow waters or through exposure to different tem-perature water masses.

4.3. Relationship between Fr, Gr and Cr

Fluctuations in feeding rate did not appear to drivefluctuations in growth rate, which suggests that theenergy from food may not be utilised for immediateweight gain. There appeared to be minimal feedbackbetween growth and feeding as growth rates in a givenweek did not generally influence feeding rates insubsequent weeks and vice versa.

Fluctuations in feeding rate did not appear to affectfood conversion rate, suggesting that energy from foodwas not translated into an increase in weight on a regularbasis. Interestingly however, food conversion rateappeared to have an immediate impact on feeding rate.The negative cross-correlations between these twovariables suggest that the physiological mechanismunderlying food conversion may have an immediateantagonist effect on feeding. The observation thatcurrent food conversion rates could have an impact onfeeding rates 3 weeks later in some individualsillustrates the length of time lag which may be involved.

The results obtained in this study may in part beexplained by the dynamic growth process. Cephalopodsgrow through hyperplasia (i.e. generation of new musclefibres) and hypertrophy (i.e. the increase in size ofexisting muscle fibres) throughout their lives (Pecl andMoltschaniwskyj, 1997; Moltschaniwskyj, 2004).These two processes are likely to be different in termsof energy demand and weight gain produced: hyperpla-sia is likely to be energetically costly (due to the creationof new fibres and structural elements) with potentiallylittle weight gain, whereas hypertrophy may be lessenergy-demanding (expansion of muscle fibres can beachieved by increasing intracellular fluid level) butproduce larger weight gains. If growth consists of analteration between phases of predominantly hyperplasicgrowth followed by phases of predominantly hyper-trophic growth, growth rate (and food conversion rate)would also fluctuate accordingly with phases of lowGr corresponding to hyperplasia and phases of highGr corresponding to hypertrophy. This could explainwhy biochemical techniques used to estimate instanta-neous growth rate have had mixed success in octopusand cephalopods in general (Semmens and Jackson,2005). High RNA concentrations, which in octopusshow a linear relationship with rates of protein synthesis(Houlihan et al., 1990), may not reflect a period of highgrowth rate if the individual is undergoing hyperplasia.Such a growth mechanism could also result in Cr valuesin excess of 100% as observed in this study if the amountof food consumed is small in relation to the weight gainedthrough hypertrophy of the muscle cells.

The relative length of the hypertrophic/hyperplasicphases is speculative at this stage but the general lack ofperiodicity in feeding rates, growth rates or conversionrates suggest that these phases may not be physiologi-cally fixed and could be influenced by external factors.

The validity of this hypothesis needs to be formallytested through further experimental work. Precise mea-surements of Fr, Gr, Cr and T in aquaria along withhistological analysis of muscle tissue from various partsof the mantle would be required to identify whether analternation between hyperplasia and hypertrophy doesoccur. Also, the time scale of 1 week for the mea-surement of variables may not be adequate to capture allthe processes in place and choosing shorter time-scalesmay reveal additional information.

4.4. Conclusion

The relationship between food consumption, foodconversion rate, growth and temperature during theearly life-history of octopus is complex. Temperature

92 J. André et al. / Journal of Experimental Marine Biology and Ecology 354 (2008) 81–92

unexpectedly, in general did not appear to have astatistically significant impact on feeding, growth orfood conversion rates. Feeding rate also did not appearto impact food conversion rate or growth rate. Shortperiods of starvation were common and the high valuesobtained for food conversion rate for some individualssuggest that octopus can grow substantially with little orno food intake, which possibly implies a re-evaluationon our understanding of the way octopus growth occursis necessary. Individual variability observed in octopusgrowth may be dependant on the growth mechanisminvolved, more specifically the fine balance between thehyperplasic and hypertrophic growth found in cephalo-pods. This study highlights the extreme plasticity andcomplexity of cephalopod growth and the need forfurther work in this area.

Acknowledgments

We thank Stephen Leporati, Natalie Moltschaniws-kyj, Allan Beech, Ed Smith, and all the people from theMarine Research Laboratory, Tasmanian Aquacultureand Fisheries Institute who helped with data collection,system maintenance and valuable comments. This studyis part of the joined CSIRO/University of TasmaniaQuantitative Marine Science (QMS) program. [SS]

References

Boyle, P.R., Knobloch, D., 1982. On growth of the octopus Eledonecirrhosa. J. Mar. Biol. Assoc. U. K. 62, 277–296.

DeRusha, R.H., Forsythe, J.W., Hanlon, R.T., 1987. Laboratorygrowth, reproduction and life span of the Pacific pygmy octopus,Octopus digueti. Pac. Sci. 41, 104–121.

Forsythe, J.W., 1984. Octopus joubini (Mollusca: Cephalopoda): adetailed study of growth through the whole life cycle in a closedseawater system. J. Zool. (Lond.) 202, 393–417.

Forsythe, J.W., 1993. A working hypothesis of how seasonaltemperature change may impact the field growth of youngcephalopods. In: Okutani, T., O'Dor, R.K., Kubodera, T. (Eds.),Recent Advances in Fisheries Biology. Tokai University Press,Tokyo, pp. 133–143.

Forsythe, J.W., Hanlon, R.T., 1988. Effect of temperature on laboratorygrowth, reproduction and life span of Octopus bimaculoides. Mar.Biol. 98, 369–379.

Forsythe, J.W., Van Heukelem, W.F., 1987. Growth. In: Boyle, P.R.(Ed.), Cephalopod Life Cycles, vol. 2. Academic Press, London,pp. 135–156.

Fowler, J., Cohen, L., Jarvis, P., 1998. Practical Statistics for FieldBiology, 2nd ed. John Wiley & Sons, Chichester.

Hanlon, R.T., 1983.Octopus briareus. In: Boyle, P.R. (Ed.), CephalopodLife Cycles, vol. 1. Academic Press, London, pp. 251–266.

Hatfield, E.M.C., Hanlon, R.T., Forsythe, J.W., Grist, E.P.M., 2001.Laboratory testing of a growth hypothesis for juvenile squid Lo-ligo pealeii (Cephalopoda: Loliginidae). Can. J. Fish. Aquat. Sci.58, 845–857.

Houlihan, D.F., Kelly, K., Boyle, P.R., 1998. Correlates of growth andfeeding in laboratory-maintained Eledone cirrhosa (Cephalopoda:Octopoda). J. Mar. Biol. Assoc. U. K. 78, 919–932.

Houlihan, D.F., MacMillan, D., Agnisola, C., Trara Genoino, I., Foti,L., 1990. Protein synthesis and growth in Octopus vulgaris. Mar.Biol. 106, 251–259.

Iglesias, J., Otero, J.J., Moxica, C., Fuentes, L., Sánchez, F.J., 2004.The completed life cycle of the octopus (Octopus vulgaris, Cuvier)under culture conditions: paralarval rearing using Artemia andzoeae, and first data on juvenile growth up to 8 months of age.Aquac. Int. 12, 481–487.

Itami, K., Izawa, Y., Maeda, S., Nakai, K., 1963. Notes on the laboratoryculture of octopus larvae. Nippon Suisan Gakkaishi 29, 514–520.

Joll, L.M., 1977. Growth and food intake ofOctopus tetricus (Mollusca:Cephalopoda) in aquaria. Aust. J. Mar. Freshw. Res. 28, 45–56.

Klaich, M.J., Ré, M.E., Pedraza, S.N., 2006. Effect of temperature,sexual maturity and sex on growth, food intake and gross growthefficiency in the “pulpito” Octopus tehuelchus (d'Orbigny, 1834).J. Shellfish Res. 25 (3), 979–986.

Leporati, S.C., Pecl, G.T., Semmens, J.M., 2007. Cephalopod hatchlinggrowth: the effects of initial size and seasonal temperatures. Mar.Biol. 151, 1375–1383.

Mangold,K., 1983a.Eledonemoschata. In: Boyle, P.R. (Ed.), CephalopodLife Cycles, vol. 1. Academic Press, London, pp. 387–400.

Mangold, K., 1983b.Octopus vulgaris. In: Boyle, P.R. (Ed.), CephalopodLife Cycles, vol. 1. Academic Press, London, pp. 335–364.

Mangold, K., von Boletzky, S., 1973. New data on reproductivebiology and growth of Octopus vulgaris. Mar. Biol. 19, 7–12.

Mather, J.A., 1980. Some aspects of food intake in Octopus joubiniRobson. Veliger 22, 286–290.

Moltschaniwskyj, N.A., 2004. Understanding the process of growth incephalopods. Mar. Freshw. Res. 55, 379–386.

Nixon, M., 1966. Changes in body weight and intake of food ofOctopus vulgaris. J. Zool. (Lond.) 150, 1–9.

Pecl, G.T., 2004. The in situ relationships between season of hatching,growth and condition in the southern calamary, Sepioteuthisaustralis. Mar. Freshw. Res. 55, 429–438.

Pecl, G.T., Moltschaniwskyj, N.A., 1997. Changes in muscle structureassociated with somatic growth in Idiosepius pygmaeus, a smalltropical cephalopod. J. Zool. (Lond.) 242.

Segawa, S., Nomoto, A., 2002. Laboratory growth, feeding, oxygenconsumption and ammonia excretion of Octopus ocellatus. Bull.Mar. Sci. 71 (2), 801–813.

Semmens, J.M., Jackson, G.D., 2005. Evaluation of biochem-ical indices for assessing growth and condition of the deepwatersquid Moroteuthis ingens. Mar. Ecol. Prog. Ser. 289, 215–223.

Semmens, J.M., Pecl, G.T., Villanueva, R., Jouffre, D., Sobrino, I.,Wood, J.B., Rigby, P.R., 2004. Understanding octopus growth:patterns, variability and physiology. Mar. Freshw. Res. 55, 367–377.

Van Heukelem, W.F., 1973. Growth and life-span of Octopus cyanea(Mollusca:Cephalopoda). J. Zool. (Lond.) 169, 299–315.

Villanueva, R., 1994. Decapod crab zoeae as food for rearingcephalopod paralarvae. Aquaculture 128, 143–152.

Villanueva, R., 1995. Experimental rearing and growth of planktonicOctopus vulgaris paralarvae. Can. J. Fish.Aquat. Sci. 52, 2639–2650.

Villanueva, R., Koueta, N., Riba, J., Boucaud-Camou, E., 2002.Growth and proteolytic activity of Octopus vulgaris paralarvaewith different food rations during first-feeding, using Artemianauplii and compound diets. Aquaculture 205, 269–286.

Wells, M.J., Clarke, A., 1996. Energetics: the costs of living andreproducing for an individual cephalopod. Philos. Trans. R. Soc.Lond. B Biol. Sci. 351, 1083–1104.