Latitudinal differences in somatic energy storage: adaptive responses to seasonality in an estuarine fish (Atherinidae: Menidia menidia )

International Association for Ecology

Latitudinal Differences in Somatic Energy Storage: Adaptive Responses to Seasonality in an Estuarine Fish (Atherinidae: Menidia menidia) Author(s): Eric T. Schultz and David O. Conover Reviewed work(s): Source: Oecologia, Vol. 109, No. 4 (1997), pp. 516-529 Published by: Springer in cooperation with International Association for Ecology Stable URL: http://www.jstor.org/stable/4221553 . Accessed: 26/05/2012 10:04 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Springer and International Association for Ecology are collaborating with JSTOR to digitize, preserve and extend access to Oecologia.

http://www.jstor.org

Oecologia (1997) 109:516-529

Eric T. Schultz Latitudinal adaptive (Atherinidae:

? Springer-Verlag 1997

David O. Conover in

diff?rences responses Menidia

to

somatic

seasonally

energy in

an

storage: estuarine

fish

menidia]

Received: 1 November 1995 /Accepted: 13 September 1996

This study focuses on the seasonal accumulaAbstract of somatic tion and depletion energy in the Atlantic an annual estuarine fish. silverside (Menidia menidia), Previous research revealed that northern silversides are winter mortality, while subject to strong size-dependent winter mortality. To southern fish suffer no appreciable in differentiation examine whether there was geographic of we allocation temporal patterns strategies, compared among three populations energy storage and utilization The comparative dealong this gradient in seasonality. sign used monthly or biweekly samples of fish collected in the wild, as well as samples of fish from each popuwhere genetic lation reared in a common environment, can be clarified. Somatic energy stores were differences via gravimetric analysis of neutral storage liquantified revealed that small inpids and lean tissue. Analysis maintained low levels of lipid redividuals relatively for their lower survival in serves, which may account winter. Wild fish in the north rapidly accumulated large somatic reserves, which were depleted over the winter and then increased again during the subsequent spring breeding season. In wild southern fish, relatively small reserves accumulated slowly until breeding commenced of in the spring. The common-environment comparison a genetic basis for revealed somatic patterns storage in reserve accumulation differences among-population rates, but no differences in the amount of reserves stored. of somatic We conclude that the overwinter depletion selective impact on energy acreserves has a significant in seasonal enand allocation cumulation strategies vironments.

E.T. Schultz (El)1 ? D.O. Conover Marine Sciences Research Center, State University of New York, Stony Brook, NY 11794-5000, USA 1Present address: Department of Ecology and Evolutionary Biology, University of Connecticut, 75 N. Eagleville Rd., Storrs, CT 06269-3042, USA fax (860) 486-4320

? ? Allocation Key words Adaptation strategy ? ? variation Allometry Energy reserves Geographic

Introduction in somatic energy reserves is conTemporal variability in sidered to be a widespread to seasonality adaptation animals (Pond 1981; Witter and Cuthill 1993). During of food, seasons in which there is reduced availability reduced energy intake, and/or increased energy output, survival or reproductive success may depend upon somatic energy reserves. Seasonal accumulation and depletion of reserves are evident in many invertebrates 1982; Arts and Evans 1991; (e.g., Tessier and Goulden Dratnal et al. 1993), fishes (Shul'man 1974; Love 1980; and Gill 1987; Larson 1991), reptiles and Weatherley 1976; Wells 1976; Fitzpatrick (Derickson amphibians et al. 1995), birds (King 1972; Blem 1976; Drent and Daan 1980), and mammals (Mrosovsky 1976). variaof intraspecific Demonstration geographical to gradient in the tendency bility along a seasonality store energy would be strong support for the notion that stratseasonal energy storage is an adaptive allocation in rare. Fish low-latievidence is Such surprisingly egy. show less temporal variation in energy tude populations stores than those in high-latitude (Guillepopulations mot 1982; Meffe and Snelson 1993). In some species of birds, the amount of diurnal winter fattening varies di1976). Study of seasonal (Blem rectly with latitude in diet, food resources, and energy reserves fluctuations revealed that of elephants among several populations fattening prior to the dry season was highest in the poof the lowest availability which experienced pulation It food during the dry season (Malpas 1977). usually is are genetically such differences not known whether effects. Genetic variance based or due to environmental in the tendency to store energy has recently been found and Kallman 1993), (Borowsky among fish populations but the selective basis for this variation remains unclear. In this paper, we test the idea that energy reserves are an

517 adaptive response to seasonality by comparing temporal among wild populapatterns of storage and utilization tions of the Atlantic silverside Menidia menidia (L.), and reared in a common environment, among populations where genetic differences can be clarified. A seasonal environment is likely to impose selection in fishes. Wintertime on energy storage capacity is tyof energy pically a period of reduced growth, utilization risk of mortality (Hunt 1969; reserves, and a heightened Shul'man 1974; Cunjak 1988; Johnson and Evans 1991; Griffiths and Kirkwood 1995). Energy storage in the fall function may primarily to sustain individuals through the winter, or to enhance reproductive in performance the subsequent reserves are drawn spring, if internal or upon for spawning migration, gamete production, other forms of parental investment (Reznick and Braun and Gill 1987; Larson 1991; Schultz 1987; Weatherley et al. 1991). We address the question of whether energy storage functions primarily to enhance winter survival or in M. menidia by determining when spring reproduction reserves are depleted relative to the winter season and the initiation of breeding. The storage and utilization of energy are often closely linked to body size. Here, we examine whether large M. menidia have disproportionally relarger energy serves. Larger and/or older individuals in other fish levels of enspecies tend to have higher weight-specific et al. 1988; ergy storage (e.g., Shul'man 1974; Henderson Larson 1991). Depletion rates of energy stores can be because of the expected to decrease in larger individuals, allometric rates (Paloheimo and scaling of metabolic Dickie 1966). The joint effect of these two allometries is that the capacity to endure food deprivation should increase with body size (Downhower 1976; Shuter and Post 1990). The scaling of storage and depletion may often be the reason that larger young-of-the-year individuals have a higher probability of surviving winter than smaller individuals in various fish taxa (review in and Coble 1979; see also Oliver et al. 1979; Toneys Henderson et al. 1988; Post and Evans 1989a; Griffiths and Kirkwood 1995). The physiological size and energy linkage between drive selection for size storage patterns may therefore and growth rate in seasonal environments (Conover evidence has recently that 1992). Important emerged as measured seasonality, by the length of the growing has a potent effect on growth season, evolutionary In M. menidia rates. and Present (Conover 1990), saxatilis D.O. Constriped bass (Morone (Walbaum) in press), and mummiover, J.J. Brown, A. Ehtisham, chogs (Fundulus heteroclitus (L.): Schultz et al. 1996), larval and juvenile fish from northern populations have a higher inherited capacity for growth than fish from southern populations (see also Isely et al. 1987; Delabbio et al. 1990; Torrissen et al. 1993; Nicieza et al. 1994). Hence, where the growing season is short, there has been an evolutionary response for increased growth variation (Conover and rates, a case of countergradient Schultz 1995). The research on latitudinal differences in

energy storage effort to clarify

patterns presented the selective context

here is part of our of this evolutionary

change.

Study

organism

The Atlantic is an abundant silverside resident of salt marshes and bays along the east coast of North AmerUSA (~30? N) to ica, ranging from northern Florida, Newfoundland and the Gulf of St. Lawrence, Canada The life cycle of this species is essentially an(~46?N). nual: less than 1% of breeding adults are 2 + years old and Ross 1982). The spawning season begins (Conover in March in southern populations, and in early June in northern populations (Jessop 1983; Conover and Present The season for juveniles in northern 1990). growing is less than half that for the southernmost populations both because the offspring are spawned populations, later, and because the season ends earlier as waters cool. of temperature-dependent studies Laboratory growth Conover, (D.O. unpublished data) and field surveys and Ross 1982; Ogburn et al. (Bayliff 1950; Conover that growth of juveniles ceases when 1988) indicate water temperatures cool to 10?-12?C. Fish in northern shelf waters at populations migrate out to continental the end of the growing season (Conover and Murawski in northern populations is 1982). Overwinter mortality high and size-selective 1984; see also Marke[Conover vich (1978) and Henderson et al. (1988) for similar results in closely related species]. In southern populations, no offshore migration occurs, winter mortality rates are is not size-selective and low, and mortality (Conover Present 1990; Sosebee 1991).

Materials

and methods

Field collection Field sampling was designed to clarify the annual pattern of energy accumulation and depletion over the species range, from a lowlatitude population near the southern limit (South Carolina: SC) to a high-latitude population near the northern limit (Nova Scotia: NS), with one population near the middle of the range (New York: NY). Fishes were collected on a monthly (SC, NY) or biweekly (NS) schedule (Table 1). Water temperature and salinity were determined on each sample date. All populations were sampled in shallow estuarine waters, using a beach seine, the dimensions of which varied with location (SC: 7 m ? 1 m, +-inch mesh; NY: 20 m ? 1 m, 3-mm mesh; NS: 10 m ? 2 m, 3-mm mesh, or 25 m ? 4.5 m, 3-mm mesh). SC fish were collected from a tidal creek in an extensive salt marsh near the Belle Baruch Laboratory, Georgetown, S.C. (33?20'N, 79?10'W). In NY, fish were collected in Great South Bay, Long Island (40?45'N, 72?55'W), over a mixed sand/silt/rock bottom with scattered seagrass (Zostera). In NS, two locations were sampled alternately during the summer and fall of 1993: Porters Point (44?50'N, 63?20') and the Annapolis River (44?40'N, 65?50/W). Specimens were immediately placed on ice after capture. Within 4 h, the total length (TL, to 0.1 mm) of each fish was measured, individuals were bagged separately and labelled, and then frozen. Specimens were initially frozen at -25?C; however, the temperature used for long-term storage was -70?C

518 Table 1 Field collection sample dates, locations, and number of fish analyzed, by population and year-class. Locations: NS Nova Scotia, NY New York, SC South Carolina, PP Porter's Point, AR NS Date lOAug 27Aug 15 Sep 30 Sep 17 Oct 26 Oct 9 May 20 May 3 June 19 June

Site 1993 1993 1993 1993 1993 1993 1994 1994 1994 1994

PP AR PP AR PP AR PP PP PP PP

19 21 19 20 34 372 22 22 22 22

Annapolis River, GSB Great South Bay, FC Fool's Creek, OL Oyster Landing, DD Debidoux Island

NY (1992 year-class)

NY (1993 year-class)

Date

Date

5 4 31 5 8 31 4 3

Nov Dec Dec Feb Mar Mar May June

Site 1992 1992 1992 1993 1993 1993 1993 1993

GSB GSB GSB GSB GSB GSB GSB GSB

31 31 37 52 41 46 61 24

SC

Site

1 Oct 2 Nov 3 Dec 5 Jan 14 Mar 23 Mar 12 Apr 11 May 7 June

1993 1993 1993 1994 1994 1994 1994 1994 1994

Date

GSB GSB GSB GSB GSB GSB GSB GSB GSB

21 40 44 18 19 23 22 22 3

16 Aug 13 Sep 12 Oct 12 Nov 13 Dec 27 Jan 24 Feb 8 Mar 7 Apr 6 May 24 May

Site 1993 1993 1993 1993 1993 1994 1994 1994 1994 1994 1994

FC FC FC FC FC OL FC FC FC FC DD

17 39 25 20 4 37 22 22 22 12 14

2 Only 17 values for lean tissue Establishment and maintenance of captive populations

Analysis of composition

Captive stocks of fish from each of the three populations were established in the spring of 1993. Fertilized eggs were collected haphazardly from Bear's Bluff Beach on the Edisto River in South Carolina (33?20'N, 80?20^) on 5 April, from Fireplace Neck Beach in Great South Bay, N.Y. on 11 May, and from Hebb's Landing on the Annapolis River in Nova Scotia on 8 and 29 June. The eggs were transported to the Flax Pond Marine Laboratory of the Marine Sciences Research Center at Old Field, N.Y., and hatchlings were reared following established methods (Conover and Fleisher 1986). Until 23 September, the fish were held on a fixed photoperiod of 14h light: 1Oh dark, in flow-through seawater at 19-27?C. On that date, fish were transferred to 1300- or 250-1 tanks in a greenhouse receiving flow-through seawater. In these tanks, the fish experienced natural seasonal cues: the tanks received only natural illumination, and the water circulating in the tanks was ambient with Flax Pond; however, when tank temperatures fell below 8?C, heaters were activated to prevent cold-induced mortality. Samples were taken for analysis on the date of transfer to the greenhouse, again in late December once the fish were expected to be acclimatized to ambient conditions, and about monthly thereafter (Table 2). Fish samples were handled in the manner described above for the field-collected specimens. The two cohorts of NS fish were reared separately until the late December collection date, and were then pooled. Handling of captive fish followed animal care protocols filed with the State University of New York Animal Care Council, and NIH guidelines.

Dissection and drying

Table 2 Laboratory population sample dates and number of fish analyzed, by source population

The number of fish collected on some dates exceeded the number needed for analysis. In these cases, subsamples were carefully chosen to be representative of the length distribution in the sample. Analysis was restricted to young-of-the-year individuals; the largest individuals (> 125 mm) were aged using scale annuii, and if identified as older fish, they were set aside. Individuals selected for analysis were thawed briefly, remeasured for length, weighed (WETWEIGHT) and dissected to remove any gut contents, to determine sex, and to remove and weigh gonads (GONADWEIGHT). Having taken the dissected wet somatic weight (DISSECT WEIGHT), the sample was diced and refrozen. Samples were dehydrated in a freeze-dryer. In preliminary trials, we determined that 72 h was ample for thorough drying (< 1% of additional weight was lost from specimens in the second 24 h of drying and < 0.5% of weight was further lost in the third 24 h of drying). If the sample dry weight was not determined immediately following dehydration, the vial was capped and stored in a dessicator. Extraction After measuring the dry weight of the whole sample (DRYWEIGHT), the dry material was transferred to an extraction thimble (medium-porosity Alundum) which had been preweighed (THIMBLEWEIGHT). Preliminary trials determined that further grinding of the sample had no effect on extraction efficiency. The weight of thimble + sample (THIMBLEDRY) was taken after 24 h

NS

NY

SC

Date

Date

Date

26 Oct 30 Oct 20 Dec 21 Dec 18 Feb 25 Mar 3 May

1993 1993 1993 1993 1994 1994 1994

20 15 40 20 31 10 3

27 Sep 16 21 18 25 3

Dec Dec Feb Mar May

1993

25

27 Sep

1993

20

1993 1993 1994 1994 1994

25 7 25 12 12

20 Dec

1993

20

18 Feb 25 Mar 3 May

1994 1994 1994

20 14 5

519 at 50?C, and then the lipid in the sample was extracted in petroleum ether, using a Soxhlet extractor custom-designed for processing multiple samples. The extractor was run for 5 h, for about 20 cycles of exposure to fresh extracting solvent. Previous comparisons of different techniques have indicated that Soxhlet extraction under petroleum ether, for 3-6 h, efficiently maximizes extractions of nonpolar lipids while removing virtually no nonlipid material (Dobush et al. 1985). The extracted sample was then redried at 50?C for 24 h, and weighed (THIMBLELEAN). Finally, the lean sample was ashed for 16 h in a muffle furnace at 600?C, then cooled and weighed when the sample temperature had stabilized at 50?C (THIMBLEASH). At high ashing temperatures, there is likely to be some dissociation of CaC03 (Busacker et al. 1990). We determined that this had a negligible effect on our results by ashing a set of samples for 24 h at 450?C, and reashing at 600?C for 16 h. The higher temperature resulted in an additional loss of only 4.9% of the ash weight, or 1% of the initial dry weight of the sample. All weights were read to the nearest milligram. The number of specimens analyzed is listed by collection date in Table 1 (for the field-collected fish) and Table 2 (for specimens taken from the laboratory population). Up to 24 samples could be processed in our extractor at any one time. To determine whether differences in conditions or minor changes in procedure might affect our results, we included aliquots of a standard homogenate in a portion of our runs. Analysis of these aliquots indicated that run effects were small (run-to-run standard deviations, lipid and protein expressed as percentage of dry weight: lipid SD 0.6%, CV 14%, lean SD 0.012%, CV 1.5%). The factor contributing most to differences in readings from run to run was atmospheric humidity (personal observation). Repeated readings of a sample on different days typically differed by no more than 5 mg. Compositional statistics were calculated as follows. Gonosomatic Index (GSI) = GONADWEIGHT/ (DISSECTWEIGHT + GONADWEIGHT) Lipid = (THIMBLEDRY - THIMBLELEAN) ? DRYWEIGHT/ (THIMBLEDRY - THIMBLEWEIGHT) Lean = (THIMBLELEAN - THIMBLEASH) ? DRYWEIGHT/ (THIMBLEDRY - THIMBLEWEIGHT) Total energy (kcal; Brett and Groves 1979) = 4.8 ? Lean + 9.45 ? Lipid The weight of lipid and lean tissue for the entire specimen was corrected for any material lost during transfer to the thimble. In a few cases, the sample was too large for a thimble and had to be ground and subsampled. The formula for total energy assumes that all lean tissue is protein, disregarding phospholipids which were not extracted by the solvent, and carbohydrates, which are a minor constituent offish tissues (Love 1980). Composition of gonads To estimate the changes in gonad lipid and lean weights with spring maturation, gonad energy content was analyzed for the period of gonad recrudescence and spawning (NS: 9 May-19 June; NY: 14 March-11 May; SC: 24 February-7 April, gonads inadvertently discarded from last collections). The gonads were too small in all but the last NS and last NY collections for analysis of individual samples, so we pooled gonads into lots with a wet weight of at least 500 mg (up to 12 fish per lot) to furnish three samples for each collection and sex. Female gonads were more energy-dense than male gonads (1.4 kcal/g wet weight vs 0.9 kcal/g wet weight). Gonad energy density also varied among populations, and collections within a population within each sex (nested analysis of variance, ? for main and nested effects 0.0001-0.0005). However, there was no clear temporal or latitudinal pattern in these differences. We therefore used the average gonad energy density of each collection to estimate individual gonad energy totals (energy density ? wet gonad weight) for spring collections in which gonads were too small for individual energy analysis.

Statistical analysis The weight of tissue components is expected to conform to the scaling power equation W = aLb, where W is the weight of lean or lipid and L is length. Both variables can be transformed to yield: lnW = a + b(lnL), and can be estimated using standard regression techniques. Because both W and L are subject to random errors, and neither are under the control of the investigator, Model II regression is appropriate (Sokal and Rohlf 1981). Data analysis in this study was conducted with PC-SAS (version 6; SAS Institute, Cary, N.C.). Testing our predictions required that we employ two regression techniques in our analysis. One prediction was that tissue components (particularly lipid) scale hyperallometrically, i.e., are disproportionally greater in large individuals. This prediction concerns the functional relationship between W and L, particularly the exponent (or regression slope) b. Analysis of functional relationships should be conducted using reduced major axis or geometric mean regression (RMA; Ricker 1973; Laws and Archie 1981), because least-squares (LS) estimates of the regression slope are biased when both variables are subject to error (see also Sokal and Rohlf 1981; LaBarbera 1989). An exponent greater than 3 is consistent with hyperallometric scaling. We tested the hypothesis in one-tailed tests against the null hypothesis that the slope equalled 3 (isometric scaling), or was less than 3 (hypoallometric scaling). The standard error for RMA slopes was approximated as the SE for the slope in LS regression of the same data (Ricker 1973; Sokal and Rohlf 1981). When multiple slopes were tested individually, we used the sequential Bonferroni procedure (Rice 1989) to adjust the significance level, keeping the analysis-wide significance level to 5%. To test other predictions, regarding temporal and among-population variations in the amount of tissue components, we employed LS techniques such as analysis of covariance (ANCOVA) and LS regression. LS approaches are the most appropriate for this line of testing because the objective of the model is prediction: interest is primarily focused on estimating the expected weight of a tissue component, given individual size. To examine the seasonal storage and depletion of tissue components, we quantified deviations from expected component weights. In some cases, use of LS means (Searle et al. 1980) was appropriate, comparing mean component weights of fish from different collections at a common average length. This is similar to comparing component weights of, for example, a 'standard fish' (MacKinnon 1972). However, LS means were not appropriate when collections differed in the scaling exponent (significant interaction between collection and length) and/or differed in length distributions so much that estimation would have required considerable extrapolation beyond the data. In these cases, rather than comparing component weights at a common length, we compared deviations from expected component weights (given length). Expected component weights were estimated in LS regressions against length, pooling across dates but analyzing each population and year-class separately. Because a single regression model was used for samples taken throughout the year, expected values represented size-dependent, but season-independent, component weights. The deviations from expected values (referred to hereafter as 'storage residuals'), averaged by collection date, represent our best estimate of seasonal storage and utilization. Similar approaches have been taken by others (Ankney and Afton 1988; Ankney and Alisauskas 1991; Fechhelm et al. 1995). The storage residuals are not completely unbiased estimates of tissue component storage. We found significant differences in LS regression slopes among collection dates (ANCOVAs of component weights vs length and date, length-by-date interaction ? < 0.05). This slope heterogeneity means that sampling errors in length would contribute to sampling errors in storage residuals; for

520 instance, if we had over-sampled larger individuals in a collection with an especially steep slope, then the residual would be biased upwards. Biases in the residual estimates should be slight, however, because care was taken to avoid size biases in collecting fish in the field, and in selecting subsamples for composition analysis.

A) e

Results Field

?. 7 ?. &?**#?? co 2 -1-1-1-1 19 July 17 Oct

15 Jan

15 Apr

14 July

19 July

~l 17 Oct

? 15 Jan

G 15 Apr

14 July

19 July

1-1-G 17 Oct

15 Jan

15 Apr

14 July

19 July

p-1-G 17 Oct

15 Jan

15 Apr

14 July

populations

Scaling of somatic tissue components and energy in the wild scaled hyperallometrically Tissue components generally in populations or isometrically of M. menidia. Pooling and yearcollections across dates for each population of lipid mean regression class, we performed geometric and lean tissue components, and total somatic energy, that larger inThese analyses indicate versus length. dividuals have disproportionately large lipid stores: the between of the functional lipid and relationship slope exceeded 3 in all three populations length significantly of lean weight was (Table 3). The scaling relationship generally tighter than that of fat weight (smaller SEs), and more isometric (closer to 3). Lean tissue was sigin the two NY year-classes. nificantly hyperallometric in the NS scaled Somatic energy hyperallometrically but was isoand the two NY year-classes, population metric in the SC population. that there was temindicated analysis Exploratory in rethe of the functional slopes poral heterogeneity for therefore conducted an and we analysis lationships, on lipid. In NS, lipid each date separately, focusing the summer and scaled hyperallometrically throughout fall (Fig. 1A); when the fish returned to spawn in the spring, the slopes were still higher than 3 but the scaling was weaker (larger SEs). There was occarelationship of lipid in the NY 1992 year-class sional hyperallometry slopes were hy(Fig. IB). In the NY 1993 year-class, in the late fall (Fig. 1C), while in SC, midperallometric

Table 3 Allometry exponents for lipid, lean, and somatic energy, for field and laboratory populations of M. menidia. Exponents estimated as slopes in ln-ln RMA regression of response variables against total length, pooling across all dates within each population/year-class combination. One-tailed significance tests were Population

Date Fig. ??-D Allometric exponents for lipids in field-collected M. menidia. Slope ( ? SE) in ln-ln RMA regression of lipid weight vs dry somatic body weight, against date. Slopes found to be significantly higher than 3 (one-tailed test, ? < 0.05, ? value adjusted by sequential Bonferroni procedure over 38 tests) are marked with an asterisk. A Nova Scotia (NS) Line joins data from Porter's Point collections only; dashed line for overwinter period when fish were offshore. ? New York (NY), 1992 year-class. C NY, 1993 year-class. D South Carolina (SC)

were

winter

slopes perallometric

slopes

conducted on the null hypothesis that the slopes equalled 3; significance levels adjusted using a sequential Bonferroni procedure (adjusting for 12 tests of field-collected and 9 tests of laboratoryreared populations)

Lean

Lipid

(Fig. ID). The hyhyperallometric observed in the NS, NY 1993, and

Energy

Slope

SE

?

Slope

SE

P

Slope

SE

?

5.73 3.72 6.63 4.05

0.21 0.18 0.31 0.16

0.0001 0.0002 0.0001 0.0001

2.91 3.49 4.93 2.91

0.03 0.06 0.04 0.02

1 0.0001 0.0001 1

3.52 3.26 5.04 3.01

0.08 0.07 0.07 0.03

0.0001 0.0002 0.0001 1

4.01 4.61 4.32

0.11 0.26 0.37

0.0001 0.0001 0.002

2.95 2.76 2.73

0.05 0.07 0.09

1.00 1.00 1.00

3.41 3.33 3.11

0.07 0.13 0.18

0.0001 0.03 1.00

Field collected NS NY 1992 NY 1993 SC Laboratory reared NS NY SC

521 tended SC populations higher (see below).

to occur

when

lipid

levels

were

Temporal changes in storage of tissue components and somatic energy in the wild patterns in storage residuals in high-latitude Temporal that both lipid and lean tissues indicate populations in the fall, and are depleted in the winter. In accumulate over the summer NS and NY, lipid residuals increased Peak residuals in the and through the fall (Fig. 2A-C). values in NS fish and fall were 190% above expected 195% above in NY fish. The fall season lean tissue acobserved 4%-8% was smaller (Fig. 2A-C; cumulation In the winter, observed above expected). lipid weight dropped below expected values (58% in May among NS fish, 50% in March among NY fish), and lean tissue was over 10%?16% below expected) also depleted (observed the winter. folpopulation Storage residuals in the low-latitude lowed a different temporal pattern. Peak accumulation occurred in the of lipid (and lean) in the SC population values above exobserved 100% 2D; lipid spring (Fig. A)

0.2 ~

0.0-

0.0

*o*-*- *-

-1.5 19 July

?^r 17 Oct

15 Jan

15 Apr

B)

r 0.2

19 July 9

17 Oct

15 Jan

15 Apr

1.5 0.0

?o- NS, and NY > SC, both ? = 0.0001), but NS and SC did not differ in lean content (P = 0.94). In terms of total somatic energy, NS had of energy than the other two clearly higher amounts = ? 3C; 0.0001), but SC and NY did (Fig. populations not differ (P = 0.44). NS fish realized high levels of somatic storage in a comseason. short growing Interpopulation relatively accumulation of material and of rate the energy parisons over the growing season are possible, given several ashatched at sumptions. (1) We assume that all individuals of the breeding season at each latitude the midpoint and Present (1990): SC, 7 May; NY, 21 [from Conover May; NS, 15 June]. (2) We assume that the dry weight of newly hatched 5-mm larvae is 0.1 mg, from a regression of weight versus length [T. Present (unpublished data): In dry weight = -7.8 + 3.4(ln length); R2 = 0.95, ? = 29; length range 15-28 mm]. (3) We assume that the comof a newly hatched larva is 5% lipid and 80% position is lean (by dry weight). The specific rate of accumulation et al. 1990): estimated as (Busacker 100 ? [In (fin al value)-ln

(initial

value)/age

in days].

522 1? 0A

A)

.?*.?

'

3-1-1 c