Nutrition of Breeding Female Ruddy Ducks: the Role of Nutrient Reserves

The condor 96:87s-s97 0 TheCooper omithologicat society 1994

NUTRITION OF BREEDING FEMALE RUDDY DUCKS: THE ROLE OF NUTRIENT RESERVES

RAY T. ~ISAUSKAS* Delta Waterfowl and WetlandsResearch Station, R.R. #I, Portage la Prairie, Manitoba RlN 3A1, Canada

C. DAVLSONANKNEY Department of Zoology, Universityof WesternOntario, London, Ontario N6A 5B7, Canada

Abstract. We determined fat, protein, and mineral content of somatic tissue, and estimated amounts of thesenutrients committed to eggsby Ruddy Ducks (Oxyura jamaicensis) nestingin southernManitoba in 1988. Seasonalnutrient dynamics differed between breeding and nonbreeding females. Nonbreeders had less fat and protein reservesthan did breeders early in the nesting season,but these then increased and subsequentlydeclined, whereas nutrient reservesof breedersdeclined linearly with date from the outset of our study. Onset of breeding was related to a nutrient reserve threshold (early RFG [rapid follicle growth] duckshad 22.7 g more fat and 5.1 g more protein than did non-RFG ducks).Adults showed greaterpropensity to initiate breeding (73% of breeding population) than did yearlings(68% of nonbreeding population), and we relate this to greater fat and protein reservesof adults. Adult breeders had more fat and protein reservesthan did yearlings at every stagein the breeding cycle but no age-related differences in use of fat, protein or mineral reservesfor clutch formation were found. When controlling for date effects, fat reservesdeclined (b = -0.49) as fat was committed to eggs. Early nesters initiated and ended egg laying with significantly more fat compared to late nesters, although rate of fat use was not different. Protein and mineral reserveswere used for clutch formation by early nesters (b = -0.24 and -0.08, respectively), but not by late nesters. Among females nearing end of laying, number of developing follicles was related to mineral, but not fat or protein reserves, indicating that mineral reserveswere a proximate factor constrainingclutch size. Our findings confirm that, for Ruddy Ducks, somatic tissuesare important sourcesof nutrients for clutch formation, and that clutch size is directly related to mineral reserves. Of importance to lifetime reproduction,and thus fitness,is our finding that minimum levels of nutrient reserves appear critical for initiation of breeding by female Ruddy Ducks. Key words: Clutchsize determination;fecundity; nutrient reserves; nutrition; reproduction; Ruddy Ducks; timing of breeding.

INTRODUCTION rossi); he hypothesized that clutch size of indi- Organisms may circumvent resourcebottlenecks vidual Ross’ Geese was proportional to size of by buffering shortfalls in energy or nutrient sup- energy reserves that each has upon arrival at ply through catabolism of nutrient reserves. So- nestingareas. Ankney and MacInnes (1978) found matic tissue can be an important energy source that potential clutch size in Lesser Snow Geese for birds during migration or incubation, or be (Chen caerulescens) was related to nutrient re- a source of nutrients for egg production. Lack serves,thereby supporting Ryder’s ideas for the (1967, 1968) hypothesized that average clutch evolution of clutch size in colonial arctic-nesting size in waterfowl (Anseriformes) evolved in re- geese. lation to average amount of food available to Subsequently, evidence for the role of nutrient laying females. Ryder (1970) modified this hy- reserves in various aspects of reproduction by pothesis to account for greatly reduced feeding waterfowl has increased(see Ankney et al. 199 1). by colonial, arctic-nesting Ross’ Geese (Chen Ankney and Alisauskas ( 199 1a) reviewed use of nutrient reserves in 16 waterfowl species and found that females in all but one species used ’ Received 1 November 1993. Accepted8 June 1994. * Presentaddress: Canadian Wildlife Service, Prairie lipid reservesfor egg production; use of protein and Northern Wildlife ResearchCentre, 115 Perimeter and mineral reservesvaried and appearedrelated Road, Saskatoon,Saskatchewan S7N 0X4, Canada. to differencesin foods consumed during egglay-

18781 RUDDY DUCK NUTRIENT RESERVES 879 ing versus normal diets during nonbreeding sea- of May to September, 1985-l 99 1, average water sons. Regardless,the importance of nutrient re- depth of 12 wetlands during our study in May servesfor proximate determination of clutch size and June 1988 were comparatively high (seeFig. in waterfowl remains controversial (Ankney et 1 of Arnold 1994); compared to long-term water al. 1991, Drobney 1991, Arnold and Rohwer fluctuations, the last half of the decade was drier 1991). than average (1961-1991). Also, nesting effort Stiff-tailed ducks (Tribe Oxyurini) have the by Ruddy Ducks during the year of our study heaviest eggsrelative to body mass of all Anat- was the highest recorded in five years (T. W. idae (Lack 1968), and a completed clutch may Arnold, pers. comm.): 1987 (n = 11 nests found weigh more than the female that laid it. Addi- on study area of Arnold [ 19941) 1988 (n = 48), tionally, Ruddy Ducks (Oxyura jamaicensis) lay 1989 (n = 0), 1990 (n = ll), and 1991 (n = 9); one egg/day (Siegfried 1976a, Alisauskas and searcheffort was greaterin yearswith fewer nests. Ankney 1994), which places very high daily de- In general,conditions for nestingby Ruddy Ducks mands for nutrients on females producing large were very good in 1988, and probably about av- clutches. Tome (1984) did not detect significant erage when considered over a longer term. changesin fat or protein reservesof Ruddy Ducks laying eggs,but his small sample sizes reduced FIELD COLLECTIONS the statistical power of his tests and thus may Ninety-seven female Ruddy Ducks were shot have led to incorrect conclusions about use of (under permit from Canadian Wildlife Service) reservesby these ducks. Given the high rate of on wetlands within 50 km of Minnedosa, Man- nutrient demand for egg laying in this species, itoba in 1988. We were especially interested in we were unconvinced specifically about Tome’s samplingbirds in the phaseof rapid follicle growth conclusion that Ruddy Ducks did not show sig- (RFG); therefore, our sampling effort reflected nificant reliance on all nutrient reserves for egg our successat acquiring birds in RFG. After reproduction. trieval, we weighed birds ( 10 g) and immediately Thus, we shot female Ruddy Ducks during the injected 70% ethanol into esophagi to prevent breeding seasonto more fully understand the role post-mortem digestion (Swanson and Bartonek of nutrient reservesin reproductionby thesebirds. 1970). At day’s end, we removed esophagealand Moreover, we wished to: (1) test our ideas (Ali- proventricular contents and placed them in la- sauskasand Ankney 1992) about thresholds of belled containers with additional 70% ethanol. nutrient reserves for initiation of rapid growth Also, after cutting through the ribcage dorsally, of ovarian follicles (see Ankney and Alisauskas we removed the ovary, oviduct and oviducal egg, 199 1b), (2) compare nutrient reservesin breeders if present. After counting postovulatory follicles and non-breeders (seeAlisauskas et al. 1990) (3) (NPoF)we placed reproductive tissue in 10% for- examine seasonal changes in nutrient reserve malin. Carcasseswere then labelled, sealed in thresholds (see Reynolds 1972), and (4) analyze plastic bags, and frozen. the effectsof nutrient reserveson terminal clutch size (see Ankney and Afton 1988, Ankney and CARCASS ANALYSIS Alisauskas 199 1b). Such an approach is more Dissections.After thawing specimens, we took important than simply considering breeding seven measurementsusing dial calipers:bill width birds, becausefactors that affect whether or not (0.1 mm)-widest dimension of ramphotheca, a bird breeds in a given year are more important bill height (0.1 mm)-greatest dorsoventral di- to its fitness than are relatively small-scale vari- mension of ramphotheca, culmen (0.1 mm)- ations in clutch size (see Ankney et al. 1991, distance from anterior tip of bill to feather line, Arnold and Rohwer 199 1). tarsus(0.1 mm)-from distal point of foot bent toward body, to notch at ankle, head length (0.1 METHODS mm)-greatest anterior-posterior dimension of head, head width (0.1 mm)-widest lateral di- STUDY AREA AND HABITAT CONDITIONS mension of head, and middle toe (0.1 mm)- The study area has been describedin some detail distance from base of nail to junction of toe with by Stoudt (1982). Water levels on wetlands are tarsus. We took three measurements with a flat extremely variable within and among years.From ruler: wing chord (1 mm)-distance from wrist measurements made the first day of each month of bent wing to most distal point of primary 880 RAY T. ALISAUSKAS AND C. DAVISON ANKNEY feathers, wing length (1 mm)-distance from and total somatic protein (S,,l,,,): junction of wing and body to most distal point S of primary feathers(with wing flattened and per- pml?,”= L,,, - S&l + L, + Lb,eart+ LIZ”,. pendicular to the vertebral column), and total where L is lean dry mass and F is fat in each length(1 mm)-distance from anterior tip of bill body component. to most posterior point of rectriceswith bird laid CLUTCH NUTRIENTS flat. We took two internal measurements with calipers after removal of skin and feathers: keel Egg composition data for Ruddy Ducks and de- (0.1 mm)-distance along median line from an- tails about laboratory analysis necessaryfor es- terior notch to posterior point of sternum, and timating clutch nutrients committed by each fe- body length(0.1 mm) from joint behind last cer- male in our sample were given by Alisauskas and vical vertebra to joint anterior to first mobile Ankney (1994). Fat invested in clutch formation caudal vertebra. by each duck (R,,,) was estimated following One half of the breast (pectoralisand supra- corucoideus)muscle and musclesof one leg were removed and weighed (0.1 g) separately. The al- where F,, was fat in each enlarged follicle, p,,,,lk imentary tract from junction of proventriculus was a constant (9.18 g) equal to average fat/egg and gizzard to cloaca was excised. Fat depots (n = 12 eggs),NpoF was number of ovulated fol- around neck and abdomen were removed, licles, and FOP_,was fat in remaining ovary tissue. weighed (0.1 g) separately and discarded. Intes- If a duck had an eggin the oviduct, we used the tines were uncoiled and pancreas was removed fat mass of that egg for pYolk(only 13 ducks had and weighed (0.1 g). Lengths of small intestine oviducal eggsfor which egg fat could be deter- (1 mm), large intestine (1 mm) and both caeca mined). together were each measured with a flat steel Protein production associatedwith clutch for- ruler; the bursa of Fabricius was measured (1 mation (R,,O,iJ was estimated as mm), if present.Ruddy Ducks were called “adult” R,,,, = XL,,, + Lovary+ Lopid if no bursa was present (Hochbaum 1942), and + N&&/k + la,bumen) “yearling” if one was present (X = 14.0 mm f 4.5 SD, range 6-21). Gizzard, small intestine, where Lfib LoVo,,,,and LoPidare the lean dry mass large intestine and caecawere separatedfrom one of each follicle, remainder of ovary, and oviduct another with a scalpel, stripped of visible fat de- in each female, and Lyolkand Eaalbumenare the av- pots, and weighed (0.1 g); each was then opened, erage lean dry fractions of yolk (4.77 g) and al- contents were discarded, and, after rinsing with bumen (4.59 g) per egg(n = 12 eggs).When avail- water and patting dry with paper towels, re- able from oviducal eggs(n = 13), we used lean weighed (0.1 g). Heart and liver were weighed dry fractions of egg yolk plus dry albumen from (0.1 g) separately. Breast muscles, leg muscles, such eggsfor &,k and &bumemrespectively. and liver were each dried to constant mass at Minerals in the clutch, RaFh,were estimated by 80°C. All other organs (except fat depots) were R, = N&&~ combined with the carcass(including feathers, except remiges and rectrices which are difficult where &fem= mean mass of eggshell,6.80 g. to grind because of large keratinous shafts); the We assigned each bird a reproductive status carcasswas then homogenizedby passingthrough accordingto state of reproductive organsor pres- a food processorat least three times and mixing ence of a brood patch. thoroughly after each grind. (A) BREEDERS included RFG Ruddy Ducks Somaticnutrients. Determinations of fat, pro- (those with rapid follicle growth) which could be tein, and ash were made using techniques de- either prelayers(n = 32; contained 11 yolky and scribedby Ankney and Afton (1988). From these rapidly developing ovarian follicles, and no atretic determinations, and using S,, = total somatic or ovulated follicles), or layers(n = 26; contained ash, we calculated, for each bird, total somatic L 1 rapidly developing follicle and 2 1 ovulated fat (S& follicle); earlypreluyers (n = 15) were a subcategory of prelayersand were thosewith 5 2 g ovarian fat (Fig. 1); fatelayers (n = 28), a subcategory of layersand earlyincubation, contained I 5 rap- RUDDY DUCK NUTRIENT RESERVES 881

f0 NONRFG I

loo 10’ 102

CLUTCH FAT, Log,o(~Ar+l), (g)

FIGURE 1. Relationbetween dry oviductmass and clutchfat for Ruddy Ducksin variousstages of breeding and nonbreedingas definedin METHODS. Early prelayersare prelayerswith lessthan 2 g of clutchfat (rep- resentedby verticaldotted line). idly developing follicles and 2 1 ovulated follicle. STATISTICAL ANALYSES Ruddy Ducks were in early incubation (n = 7) if they had no developing follicles, but had a brood Structural size correction. We used only some patch and enlarged (I 10 g wet mass) oviduct; structural measurementsto scalevariation in S,,, those in later incubation (n = 5) had no devel- S,,,, and S,, due to body size (Alisauskas and oping follicles, but had a brood patch and a re- Ankney 1987). First, using MANOVA, we con- gressed(< 10 g wet mass) oviduct (Fig. 1). firmed that there were no overall morphometric We also classifieda single female as a renester differencesbetween breedersand nonbreeders (F because it had developing follicles, very small = 0.65; 12, 48 df; P = 0.79), nor between early POFs (similar to those of a duck in later incu- prelayers and NONRFG birds (F = 0.42; 12, 9 bation) and a brood patch. This bird was omitted dc P = 0.92). We conducted principal compo- from any further analyses. nents analysis (PCA) of morphological measure- (B) NONBREEDERS (n = 26) included all ments using both correlation and covariance ma- other birds, i.e., thosewithout enlarged oviducts, trices. Becausemorphological structuresof some or ovulated follicles, or rapidly developing fol- specimenswere damaged during collection, this licles, or brood patches, and seven ducks that resulted in substantial missing data when all 12 had somewhat enlarged oviducts but contained measurements were used in PCA. Therefore, we only atretic and no developing follicles and no repeated PCA after systematically deleting mor- POFs (Fig. 1). NONBREEDER is synonymous, phological variables with the greatest missing in this paper, with NONRFG birds, the appro- data. Correlations between scoresfrom each first priate usagedepending on specific comparisons principal component, and somatic fat, protein and analyses reported. and mineral fractions were compared to deter- 882 RAY T. ALISAUSKAS ANDC. DAVISON ANKNEY mine the best correction for structural size while licles) and all interactions. We used date of first retaining the largest sample size of birds. We ovulation to separate any date effects from any selected a principal component derived from a effectsof commitment to clutch nutrients on nu- correlation matrix of wing chord and body length trient reserves. (loadings 0.7 1,0.7 1, respectively, accounting for For each of the two preceding analyses, we 59% of total variation) because 92 of 97 ducks followed the protocol used by Ankney and Ali- had both measurements recorded, and because sauskas(199 1b). If overall models were signifi- it was correlated with each somatic nutrient, and cant (P < 0.05), we removed non-significant was generally better correlated with these than three-way interactions from the model (P > 0. lo), were univariate measurements. We used least- and analyses were redone. If following this sec- squaresregression of S,,, &_.,, and S,, on this ond analysis, two-way interactions were non-sig- size index (PC,,,,) and found: nificant (P > O.lO), they were deleted from the model and the analysis was repeated using only S,,, = 84.39 + 7.233PC,,,, main effects as predictors in the model. Unlike df = 90, r2 = 0 *04 , P < 0.05, Ankney and Alisauskas (199 1b), we carried these analyses a step further and any non-significant S,,,, = 124.4 + 3.979PC2,,, (P > 0.10) main effectswere removed from the model before a final analysis was done. Type III df = 90, r2=019* 9 P < 0.0001, sums of squares were used for determining P S,, = 19.66 + 0.958PC2,,, values in all analyses. To illustrate graphically any influence of first df = 90, r2 = 0 14 P < 0.0001. . , ovulation date on effect of clutch nutrients on We used the residual (observed - predicted) val- respectivesomatic nutrients, we codedRFG birds ues from these regressionsto calculate new val- as “EARLY” if they were collected on or before ues (J+)for each observation corrected for struc- 12 June (n = 26), and “LATE” if after 12 June tural size: (n = 35). This was near the mean nest initiation date (11 June, see below); this also was the date Yi = Yobs- [a + b(PC,,,,)I + Fobs. before and after which the most similar sample These size-correctedvalues of S,,, S,,,, and S,, sizes could be assigned to each group, while were used in subsequent analyses. Where this maintaining at least a 10 day range in each. We resulted in missing data (n = 5 females) due to used linear regression (PROC REG, SAS 1990) missing morphological measurements, data un- between somatic nutrients and clutch nutrients corrected for size were used. for EARLY and LATE layers separately to fur- We tested for changesin mean length of bursa ther examine details of changesin slopes or in- of Fabricius among yearlings using one-way tercepts with season. analyses of variance (ANOVA) acrossreproduc- To determine whether number of developing tive categories(PROC GLM, SAS 1990). follicles was correlated with nutrient reservesin We tested the hypothesis that initiation of late layers with iV, ~6 (Ankney and Afton clutch formation was related to nutrient reserve 1988), we hrst regressed(PROC GLM, SAS 1990) statusof Ruddy Duck females by comparing nu- each nutrient reserve against NpoF We tested sig- trient reservesof NONRFG females with those nificance of correlations between each set of re- of early prelayers;we used early prelayers to con- siduals representing fat, protein and mineral re- trol for variation in depletion of reservesfollow- serveswith number of developing follicles using ing initiation of follicle growth. Collection date a priori, one-tailed tests. and age were included in a saturated model (PROC GLM, SAS 1990). RESULTS To analyze relations between nutrient invest- ment in eggsand respective somatic nutrients, GENERAL BREEDING CHRONOLOGY we used a general linear model (PROC GLM, In 1988, the first Ruddy Duck was observed on SAS 1990), and analysis of covariance (AN- the study area on 25 April (Fig. 2) during a stan- COVA) of somatic nutrient with respectto clutch dardized roadside censusof 68 wetlands (T. W. nutrient, female age, date of first ovulation (i.e., Arnold, pers. comm.). The earliest known ini- date of collection minus number of ovulated fol- tiation date for 26 nests found on the study area RUDDY DUCK NUTRIENT RESERVES 883

I I I 13 April 3 May 23 May 12 June 2 July DATE FIGURE 2. Arrivalphenology of RuddyDucks as determined from standardizedcensuses of 68 wetlandsnear Minnedosa,Manitoba, 1988 (T. W. Arnold, pers.comm.).

was 26 May, and the latest was 21 June (X = 11 study, then yearlings would have been under- June + 9 days [SD]). Based on 22 nests with sampled. However, becausenest initiation dates available data, clutch size (CS, K = 6.0 f 1.1 ranged from 26 May until 2 1 June for nestsfound [SD]) declined with date (d, May 25 = day 1) as on census routes (T. W. Arnold, pers. comm.), CS = 8.85 - 0.13 d (F = 8.00, r2 = 0.25, P = it is likely that most nest initiations had already 0.0 1). taken place before our last collection date. A second possibility is that yearlings became under- VARIATION IN NUTRIENT RESERVESBY representedin sequential stagesof reproduction STATUS AND AGE becausetheir successat completing each stageis Adult Ruddy Ducks had more body fat and pro- proportionally lower than that of adults. There tein, but not more mineral, than did yearlings was no differencein collection datesbetween ages (Table 1). These differenceswere consistentacross (2-way ANOVA: F = 1.02, df = 1,64, P = 0.32) all stagesof breeding as shown by lack of inter- when controlling for breeding status. Among action between ageand statusfor each dependent ducks showing RFG, mean initiation date for variable. We repeated the analyses after remov- adults was 15 June compared to 16 June for year- ing variation due to Julian date of collection and lings (F = 0.83, df = 1, 69, P = 0.37) thereby found the same patterns (Table 1). ruling out the first explanation while supporting Yearlings comprised a decreasing proportion the second. A final explanation for decreasing in samples organized chronologically by repro- representation by yearlings in progressivelylater ductive status: 68% for NONRFG, 34% for prenesting categoriesis that the bursa of Fabricius, layers, 25% for layers, 14% of early incubators which we used to age Ruddy Ducks, regressesin and 0% of later incubators. This decline (rs = yearlingsas they attempt reproduction. We found - 1.O, n = 5, P = 0.05) may have been an artifact no evidence of decline in bursa length (F = 0.25, of later nesting by yearlings vs. adults. If a large df = 3, 34, P = 0.86) among nonbreeding, pre- proportion of yearlings initiated reproduction af- laying, laying or incubating yearling Ruddy ter 28 June, the last date of collection in our Ducks.

886 RAY T. ALISAUSKAS ANDC. DAVISON ANKNEY

TABLE 2. Comparisonof linear and quadraticmodels fitted to somaticfat, protein and mineral againstcalendar date (1 = 25 May 1988).

somatic N0ttbreediU.g Breeding tissue Lineal Quadratic Linear scaled to Quadratic body size r- ’ P RJ P r ’ P RI P Fat 0.16 0.024 0.22 0.023 0.29 0.0001 0.29 0.0001 Protein 0.00 0.460 0.00 0.427 0.17 0.0003 0.16 0.0011 Mineral 0.02 0.230 0.17 0.045 0.08 0.011 0.06 0.041

VARIATION WITH DATE breeding (17) yearlings (49%; likelihood ratio G Dynamics of nutrient reservesof breeding Rud- = 13.6, df = 1, P < 0.001). dy Ducks differed from those that had not as yet The propensity for female Ruddy Ducks to shown signs of breeding (Fig. 3). Linear models attempt to breed was related not only to age, but of nutrient reserveswith date fit the data better also to size of nutrient reserves.Adults had high- for breeders, whereas quadratic models fit the er levels of somatic fat, protein and mineral than data better for nonbreeders,judging from higher did yearlings (Table 3). Controlling for age and coefficientsof determination (Table 2). Data for date ofcollection, early prelayershad 22.7 g more nonbreeders suggestedstorage of somatic fat, fat and 5.14 g more protein than did NONRFG protein and mineral reserves shortly after our birds (Fig. 4, Table 3), but no differences were collections began. Breeders, on the other hand, evident in mineral reserves(Table 3). No inter- did not store nutrients after collections began, action existed between size of reservesand pro- but had more nutrient reserves than did non- clivity to breed versus age (Table 3). breeders (Fig. 3, see below). It is not known whether some Ruddy Ducks classified as nonbreeders would have become breeders had they ROLE OF NUTRIENT RESERVESIN CLUTCH FORMATION been collected some days later. We used regression analyses to determine rates THRESHOLD FOR INITIATION OF RFG of decline in nutrient reservesrelative to amount The ratio of yearling (n = 17) to adult (n = 9) of respective nutrients invested in eggs(Table 4, NONRFG birds was higher than the ratio of Figs. 5-7). Consistent with data in Figure 3, fat yearlings (n = 4) to adults (n = 11) in early RFG reservesdeclined with calendar date (b = - 2.14 ducks (likelihood ratio G = 5.7, df = 1, P = g/d) in female Ruddy Ducks during RFG. In- 0.017). Similarly, the ratio of breeding (5 1) to dependent and additive to this was a significant nonbreeding (8) adults (86%; includes all cate- effect of egg fat commitment with -0.49 g de- gories from NONRFG through incubation) was cline in fat reserve for every gram of eggfat com- greater than the ratio of breeding (18) to non- mitted (Table 4). Thus taken together, these two

TABLE 3. Equations from general linear model (SAS 1990) based on least-squaresregression with Type III sums of squaresrelating somatic nutrients of 4 1 female Ruddy Ducks to reproductivestatus (nonbreeder = 0, early-RFG = l), age (yearling = 0, adult = 1) and collectiondate. Values in parenthesesare 1 SE.

Sfir 9.93 0.000 1 0.45 98.87 (11.51) Rep. Status 22.7 (9.21) 6.06 0.0186 Age 18.01 (8.95) 4.05 0.0514 Dated - 1.50 (0.50) 9.07 0.0047 S.0,0,=1” 10.67 0.0002 0.36 118.69 (2.13) Rep. Status 5.14 (2.72) 3.57 0.0665 Age 8.41 (2.62) 10.29 0.0027 Sarh 9.25 0.0042 0.19 18.55 (0.54) Age 2.36 (0.78) 9.25 0.0042 a Correctedfor variationdue to body size(see Methods). bNonsignifmnt (P > 0. IO) 3-way and Z-way interactionsand main effectsexcluded from model(see Methods) cProbability that the slope = 0. 4Date = collectiondate where 1 = 25 May. RUDDY DUCK NUTRIENT RESERVES 881

YEARLINGS ADULTS 180 I i I 1 7 n CT 180 w 0 0 140 0 a 120 ci,

100 0 l @J O 80 0 .” s 80 I 0 40 VI 20 100 120 140 160 120 140 160 SOMATIC PROTEIN (g)

FIGURE 4. Relation between fat and protein reservesof nonbreeding(NONRFG) female Ruddy Ducks with early prelayers(see METHODS). findings show that, although early nesters and g/g) in the overall model (Table 4), but a signif- late nestershad similar slopes(-0.50 vs. -0.54, icant (P = 0.0657) interaction between these two respectively, Fig. 5) early nesters began laying main effects rendered these estimates unstable. cycles with significantly more fat reserve (114.4 The relation between somatic protein and clutch g) than did late nesters (97.3 g). Age did not protein was significant (b = -0.24 g/g) in EAR- influence size of initial fat reserves, or fat loss LY ducks, but not (b = -0.004 g/g) in LATE with date or with egg fat commitment. ducks (Fig. 6). Similar to fat relationships, how- Somatic protein was significantly negatively ever, the intercept (135.5 f 5.2, 95% CI) for related to date of first ovulation (b = -0.68 g/d) EARLY birds was significantly higher than for and to protein commitments to eggs(b = -0.41 LATE birds (125.0 f 3.8, 95% CI). Age did not

TABLE 4. Equations from general linear model (SAS 1990) based on least-squaresregression with Type III sums of squaresrelating somatic nutrients (Y) of 6 1 female Ruddy Ducks versustheir commitment to clutch nutrients, date of first ovulation and age (yearling = 0, adult = 1). Values in parenthesesare 1 SE.

s/or 13.08 0.0001 0.31 146.07(12.10) R/, -0.49 (0.15) 11.63 0.0012 Dated -2.14 (0.56) 14.52 0.0003 s Pro&w 4.97 0.0039 0.21 142.34(4.44) Rpme,. -0.41 (0.18) 5.25 0.0257 Date -0.68 (0.21) 10.19 0.0023 R prorrmx Date 0.02 (0.01) 3.52 0.0657 s arh 3.01 0.0181 0.22 18.93 (1.99) R,,,, -0.16 (0.08) 4.45 0.0394 Age -4.80 (2.27) 4.48 0.0389 Date (Adults) -0.17 (0.06) 9.12 0.0038 Date (Yearlings) 0.07 (0.09) 0.62 0.4329 R mmmrdx Date 0.007 (0.004) 2.80 0.0998

* Co- for variationdue to bodysize (see Methods). bNonsignificant (P > 0.10) 3-way and 2-way interactions and main effectsexcluded from model(see Methods). ( Probabilitythat the slope= 0. * Date of Grstovldation = collectiondate - nPU*where on 25 May, collection date = 1. 888 RAY T. ALISAUSKAS AND C. DAVISON ANKNEY

0 I I I I 0 20 40 60 80 100 CLUTCH FAT (g)

FIGURE 5. Relation of fat reserves (&) in female Ruddy Ducks to their commitment to clutch fat (I&). EARLY ducks (S,*, = 114.44 - 0.50 R ,*,, r* = 0.11, P = 0.0537) had dates of first ovulation from 24 May to 12 June 1988, and LATE ducks ($, = 97.33 - 0.54 R,,,, r2 = 0.16, P = 0.0102) from 13 June to 22 June 1988.

influence initial protein level, or protein losswith pendent of any effect of eggshell production, date or with eggprotein commitment (Table 4). somatic mineral declined significantly with date Mineral reserves declined with commitment in adults (b = -0.17 g/d), but not in yearlings to eggshellformation (b = -0.16 g/g); however, (Table 4). as with clutch protein, clutch eggshellinteracted Generally, relations between somatic nutrients significantly with date of first ovulation (Table and respective clutch nutrients were not strong 4) in affecting somatic mineral levels; ignoring in significant regressions done separately for age effects, mineral reserves in EARLY ducks EARLY and LATE ducks; coefficients of varia- declined 0.08 g for every gram of eggshellpro- tion calculated for fat, protein, and mineral reduced, whereas no significant relationship exist- serves were only 16%, 31% and 27%, respec- ed among LATE Ruddy Ducks (Fig. 7). Inde- tively, in each model (Figs. 5-7). RUDDY DUCK NUTRIENT RESERVES 889

CLUTCH PROTEIN (g)

FIGURE 6. Relation of somatic protein (SP,Mc,.) in female Ruddy Ducks to their commitment of clutch proteins (R,,,,). EARLY ducks (SP,O,<,.= 135.52 - 0.24 R,,,,, rz = 0.3 1, P = 0.0020) had dates of first ovulation from 24 May to 12 June 1988, and LATE ducks (&,,,, = 125.01 - 0.004 RD,M,,, r* = 0.03, P = 0.9347) from 13 June to 22 June 1988.

THRESHOLD FOR TERMINATION OF RFG DISCUSSION Number of developing follicles was not corre- Maximum daily costsof eggproduction by Rud- lated with either somatic fat (r = -0.2 1, P = dy Ducks, as a percentage of BMR (279%), are 0.288) or protein (r = 0.05, P = 0.813) but was among the highest of all waterfowl (Alisauskas correlated (r = 0.34, P = 0.074, one-tailed test) and Ankney 1992,1994). How do Ruddy Ducks with amount of somatic mineral (Fig. 8) after supply nutrients at a rate sufficient to meet these accounting for variation in nutrient reservesdue very high costs?Why do Ruddy Ducks not pro- to number of ovulated follicles. rate costsover a longer period, e.g., by lowering

@ 20 ____sz___~____. v v l 15

10 1 I I I 1 I I 0 10 20 30 40 50 60 CLUTCH EGGSHELL (g)

FIGURE 7. Relation of somatic ash content (S,,) of female Ruddy Ducks to their eggshellproduction (Rm,_,). EARLY ducks (S,, = 21.28 - 0.08 Rmmn_,,r2 = 0.27, P = 0.0039) had dates of first ovulation from 24 May to 12 June 1988, and LATE ducks (S,, = 19.87 - 0.0007 Rm,ne,a,,r* = 0.03, P = 0.9779) from 13 June to 22 June 1988. 890 RAY T. ALISAUSKAS ANDC. DAVISON ANKNEY

MINERAL (g)

FIGURE 8. Relationbetween somatic ash and numberof developingfollicles in RuddyDucks with > 0 ovulated folliclesand < 6 developingfollicles. laying rates, to offsetdaily costs?We believe that between use of somatic tissuefor clutch nutrients answersto these questions lay in a critical inter- and date. About 49% of fat, about 24% of protein play between use of nutrient reservesand selec- and 8% of minerals in eggs were supplied by tion of wetlands with abundant food that is pre- nutrient reservesin early nesters,but late nesters requisite to a strategyof extreme rates of nutrient relied completely on exogenousprotein and min- supply to eggs. eral while still supplying about half of their clutch Tome (1984) compared mean nutrient re- fat with fat reserves(Table 4). The contribution servesof Ruddy Ducks in groups similar to ours of somatic mineral is not trivial, given the high (Table l), and found that endogenous fat and daily rates of nutrient demand for egg produc- protein were not used significantly for egg pro- tion, and these reserves may be critical in cov- duction, but that endogenousminerals were used ering daily mineral deficits of laying Ruddy to form eggshells.Tome (1984:830) concluded Ducks. that Ruddy Ducks “depend almost exclusively Coefficients of determination for regressions upon dietary intake to meet the energy and nu- of fat, protein and mineral reserveson their re- trient requirements of reproduction”. We are un- spectiveclutch nutrients were comparatively low sure if his failure to document use of fat and (r2 5 0.3 1, Figs. 5-7) even when significant. We protein reservesresulted from small sample size think that both methodological and ecological of Ruddy Ducks in RFG (n = 15), or from his processesmay have confounded these relations. statistical approach (see Alisauskas and Ankney Alisauskasand Ankney (1992) cautioned that use 1992 for a comparison and discussion), but we of species-specificaverages for egg composition suspect that 15 RFG birds were inadequate to entails measurement error in the independent test the hypothesis of no relation between nutri- variable, i.e., egg nutrient commitment. For ex- ent reserves and nutrient commitment to eggs. ample, mean yolk lipid determined from 12 Some of this discrepancy may arise from differ- Ruddy Ducks was 9.18 g and ranged from 7.86 ent years of study. Regardless,with a sample of g to 10.5 1 g (Alisauskas and Ankney 1994). For 6 1 RFG birds, we found a complex relationship a female that laid seven eggs,we estimated that RUDDY DUCK NUTRIENT RESERVES 891

64.3 g of egg lipid was produced using average which dive interval increasesuntil another suit- yolk lipid, but using minimum or maximum val- able site is found and intensive foraging resumes. ues, the estimate becomes 55.0 g or 73.6 g, re- Variation in rate of exogenous nutrient acqui- spectively, a difference of 34%. A useful test of sition may mean that individual females show the magnitude of error could be done with species variable rates of decline in nutrient reservesdur- (such as non-parasitic hole-nesters whose nests ing RFG. Nutrient reserves may be critical in are easier to find and monitor) for which nutrient ensuring an uninterrupted sourceof eggnutrients composition of entire clutchescould be assigned when individual females face shortfalls in ex- to females that produced them. Coefficients of ogenousnutrients over a short time interval. Our determination for regressions of nutrient re- estimates of rate of decline in nutrient reserves serveson eggnutrients using species-specificav- (Table 4) no doubt are confounded by theseother eragesfor eggcomposition vs. female-specificegg factors, but the estimates may nevertheless be composition then could be compared. Only 13 thought of as the population average for the rate ducks in our sample had undamaged oviducal of nutrient reserve decline in responseto eggnu- eggsso we could not properly evaluate the effect trient demand. Inter-female variability in avail- of using a species average of egg composition ability of exogenous nutrients is probably the rather than female-specific values of egg nutri- most important factor confounding the relation ents. between nutrient supply to eggsand decline in nutrient reserves,resulting in a low coefficient of determination for these regressions (Table 4). ROLE OF EXOGENOUS NUTRIENTS IN These considerations probably apply to most EGG FORMATION birds where a mixed strategyof endogenousand High unexplained variation in nutrient reserves exogenousnutrients are used for eggs,but likely of laying Ruddy Ducks also may stem from a are more pronounced in Ruddy Ducks than in combination of ecological factors: inter-female other waterfowl; this is becausea female Ruddy variation in nutrient reservesat the start of laying Duck rarely leaves the wetland in which its nest and/or inter-female variation in assimilation of is located (Tome 1984), and an important char- exogenousnutrients, exacerbatedby variation in acteristic of habitat quality is the invertebrate food availability and efficiency of prey intake. prey assemblageassociated with that wetland. The importance of feeding during egg-laying Another source of among-female variation in by Ruddy Ducks is clear (Tome 1991). Their useof reservesfor eggproduction in Ruddy Ducks main foods are benthic invertebrates, primarily could be related to their proclivity to lay eggsin Chironomidae (Siegfried 1973; Woodin and nestsof conspecificsor of other species(Siegfried Swanson 1989; Alisauskas and Ankney, unpubl. 1976a, Joyner 1983). If, for example, in a given data) which may be difficult to locate and more year some females are nestersand others are par- costly to obtain compared to prey exploited by asites, then these groups could differ in their use other duck species.First, to obtain them, birds of reserves. Some female Ruddy Ducks appar- must dive and strain them from the soft bottom ently lay eggseven though unpaired (Gray 1980) of a wetland basin. The difficulty is compounded and thesemay be parasites.Female Ruddy Ducks for Ruddy Ducks becausethey forage using tac- in poor body condition may lay eggsparasitically tile rather than visual cues(Tome and Wrubleski in nests of other Ruddy Ducks, as has been sug- 1988). Second, chironomid larvae are patchily gestedfor other waterfowl (Laughlin 1975, Pien- distributed and there is great variation in larval kowski and Evans 1982). If so, they then could density among patches(Gray 1980, Tome 199 1). use up all their reserves to produce eggsas no Third, patches are temporally variable because reserveswould be required for incubation. The of asynchronous maturation of larvae among preponderance of parasitic egg-laying among patches (Tome 1988). Disappearance of a chi- yearlings, as has been described in Barrow’s and ronomid patch, through either depletion by Rud- Common Goldeneyes (Bucephala islandica and dy Ducks or development of chironomid larvae, B. clangula, Eadie 1989) and Canvasbacks (Ay- motivates patch switching (Tome 1988, 1989). thya valisneria, Sorenson 1993) if also true for Tome (198 1) observed that foraging Ruddy Ruddy Ducks, may also explain our finding that Ducks dive repeatedly in the same portion of a yearlings become systematically under-repre- wetland until the location is abandoned, after sentedin our categoriesof chronologicallyranked 892 RAY T. ALISAUSKAS AND C. DAVISON ANKNEY reproductive states,i.e., from NONRFG to later plain why most Aythyini (diving ducks) show a incubation (Table 1). prolonged interval between arrival and nesting We could not evaluate annual variation in nu- (Hochbaum 1944, Alisauskas et al. 1990, Barzen trient reserve use with our data. The year of our and Serie 1990, Afton and Ankney 1991) com- study was about average in wetland conditions pared to some dabbling ducks (e.g., Mallard and at Minnedosa, Manitoba, but exceptional in Northern Pintail), which arrive on nesting areas nesting effort by Ruddy Ducks (seeMETHODS). with relatively large nutrient reserves and nest Patterns uncovered among Ruddy Ducks breed- shortly thereafter (Krapu 1974, 198 1). ing in 1988 probably typify those to be found in Ruddy Ducks are unusual among North a variety of years. Annual variation in suitable American anatids, becausepairing occurson the habitat likely affects the proportion of females breeding grounds, and, apparently, males often that surpassthe nutrient reserve threshold (see provide females with little more than gametesas below), and that subsequently breed. We suggest there is great variation in duration and existence that this has greater consequences for nesting of identifiable pair bonds (Hochbaum 1944, population size and recruitment of young than Siegfried 1976b, Gray 1980). Rohwer and An- do subtle annual changesin rate of nutrient re- derson (1988) proposed that evolution of pair serve decline for egg production. formation schedulesin waterfowl involved a bal- ance of (a) benefits to females (nutrient acqui- PRENESTING STORAGE OF sition) and to males (mate access),against (b) NUTRIENT RESERVES coststo paired males of mate defense and vigi- Our findings suggestthat sufficient nutrient re- lance (see also Afton and Sayler 1982). Thus, serves are a prerequisite for start of RFG in fe- Ruddy Ducks fit this pattern in an extreme way male Ruddy Ducks. In 1988, there was a one by delaying courtship and pairing until after ar- month interval between first arrival (25 April) rival onto nesting grounds, where some nutrient and first nesting (26 May) and the highest rate storage occurs (Gray 1980; Tome 1984, this of nest initiation occurring in mid-June. In Cal- study). The spatio-temporal unpredictability of ifornia, Ruddy Ducks stored body fat after ar- prey important for nutrient storageby females, rival on nesting areas (Gray 1980). Tome (1984) and as a direct source of exogenousnutrient for reported lower lipid and mineral reservesin “ar- eggs,may make territorial defense by males un- riving” than in “prelaying” females and inter- economical. Instead, when Ruddy Ducks form preted this as storage. He gave no information, identifiable pair bonds, males only defend the however, about timing of collections of birds in female and an area immediately around her each of thesecategories, potentially confounding (Siegfried 1976a, 1976b). conclusionsabout storagewith those about lower Tome (1984) hypothesized that nutrient stor- nutrient reserves of non-breeders as compared age in Ruddy Ducks is deferred until after spring to breeders. We did not sample Ruddy Ducks migration because high wingloading (Raikow upon first arrival, so we have incomplete infor- 1973) may hinder efficient migratory flight if nu- mation about amounts of nutrients stored on trients were stored before or during migration. nesting areas before nesting versus amounts However, Ruddy Ducks shot during fall migra- stored during spring migration. Regardless,our tion at Long Point (n > 100) Ontario, an im- data showthat, early in the nesting season,breed- portant stagingarea, appeared as fat as prelaying ers were fatter and more muscular than were ducks collected in Manitoba (Alisauskas and non-breeders on comparable dates,and that non- Ankney, pers. observ.). It is unclear to us why breeders stored nutrients after arrival on nesting excessive wingloading would be a factor in re- areas(Fig. 3). Thus, averageschedules ofnutrient serve storage during spring, but not fall migra- storage were confounded by individual variation. An alternative hypothesis is that food is tion, with some individuals attaining adequate more limiting during spring migration. Outside reserves for nesting well before others. We sus- the breeding season,Ruddy Ducks consume an- pect that additional variation in prenesting body imal matter primarily (Euliss et al. 199 1, Hoppe composition results from differential feeding ef- et al. 1986) but can consume a variety of seeds ficiency of individual females because of expe- and submergedaquatic vegetation (Bellrose 1976 rience, age (see below), or accessto abundant and referencestherein). Aquatic vegetation is not food. Prenesting nutrient storage may also ex- very abundant in spring due to overwinter mor- RUDDY DUCK NUTRIENT RESERVES 893

tality of shoots.During winter, many prairie wet- ent reservesare unimportant in control of clutch lands freeze to the bottom and, during spring size is seriously flawed becausethey ignored non- thaw, bottoms of wetlands which are open at the breeding (functionally a clutch size of zero). Eco- surface are frozen, thereby restricting accessto logical costs and benefits involved in evolution submergent macrophytes, root stalks and tubers. of nutritionally basedthresholds for breeding are Under such conditions, Ruddy Ducks, which are unknown, but likely entail low inclusive fitness benthic invertebrate specialistsduring breeding, associatedwith small clutches,reinforced by pre- may face severe food shortages.As benthic in- dation risk for both clutches and females, a gen- vertebratesare the last foods to become available eral and important feature of waterfowl nesting in spring, food inaccessibility on spring staging ecology (Arnold et al. 1987, Sargeant and Rav- areas may explain why Ruddy Ducks are among eling 1992). the latest migrants in spring (Hochbaum 1944). Reynolds (1972) first hypothesized that there Similarly, but due instead to the low quality of is a nutrient threshold for reproduction by wa- an alternate food base, Lesser Scaup (Aythya uf- terfowl, and his model incorporated covariation finis), which show a similar seasonaldiscrepancy in clutch size and laying date. Consequently, nu- in fatness (A. D. Afton, pers. comm.), may also trient reserve thresholdsfor initiation of egglay- face chronic food shortagesduring spring migra- ing have been reported for Ring-necked Ducks tion preventing proper conditioning for nesting (Aythyu collaris, Alisauskas et al. 1990), Gad- until after arrival on nesting areas (Afton and walls (Anus strepera, Ankney and Alisauskas Ankney 199 1). 199 lb) and Ruddy Ducks (this study). As these also were the only studies that have tested the CONTROL OF REPRODUCTION hypothesis, we predict that such thresholds are By comparing number of visible females (n = a general feature of reproduction by temperate 50) with number of nests found (n = 40) on his and arctic nestingwaterfowl. Moreover, we clearly study area, Siegfried (1976a) concluded that 20% documented a seasonal decline in nutrient re- of Ruddy Ducks did not nest. Based on collec- serve thresholdsfor initiation of breeding. Adap- tions of a few birds, he suggestedthat food short- tive reasonsfor sucha decline are lessclear, how- agesdid not limit onset of nesting in Manitoba. ever, but may be related to an interplay between Gray (1980) found that nesting activity of female seasonalchanges in food abundance and seasonal Ruddy Ducks in California occurred about two changes in recruitment rates. We suspect that weeksafter numbers of chironomid larvae peaked postfledging survival of young probably is di- in nesting areas; she offered this as evidence that minished in Ruddy Ducks that nest late, as has availability of exogenousnutrients plays an im- been demonstrated in other waterfowl (see Roh- portant role in motivating Ruddy Duck females wer 1992 for a review), and this should result in to nest. Our data indicate that non-breeding is selection for early nesting. We believe that Rud- related to size of nutrient reserves,but we suspect dy Ducks have evolved the capacity to nest be- that nesting attempts are probably governed by fore exogenousnutrients are abundant if they can both appropriate levels of nutrient reserves,and succeed in storing sufficient nutrient reserves. food abundance on nesting wetlands relative to Birds able to store enough fat to surpassthe early nutrient demands for egg formation. Siegfried seasonthreshold are the first to nest; birds that (1976a) suggestedthat availability of green veg- store fat at a slower rate after arrival nest later, etation for nesting habitat limited nesting activ- but, in this case, less cumulative fat will be nec- ity of Ruddy Ducks, but we think that nutrient essaryto initiate breedingbecause exogenous food availability, either from nutrient reserves(Table is more abundant (see Alisauskas and Ankney 4) or directly from food (particularly chironomid 1992). As the nesting season progresses,initia- larvae), is far more important in this regard. tion thresholds,perhaps governed endocrinolog- To forgo breeding has greaterconsequences for ically, decline so that a breeding effort becomes lifetime reproductive successthan doesvariation more likely even with fewer fat reserves. Our in clutch size, whether it is argued that nutrient analyses (Fig. 3, Tables 3, 4) suggestthat fat re- reservesare important in controlling such vari- servesdecline seasonally,but given that nutrient ation (Ankney et al. 1991) or irrelevant to such reserves are stored during the prelaying period, variation (Arnold and Rohwer 199 1). Thus, Ar- we suspectthat birds merely begin to nest after nold and Rohwer’s (199 1) argument that nutri- having stored less reserve when nesting late as 894 RAY T. ALISAUSKAS ANDC. DAVISON ANKNEY

TABLE 5. Eggnutrient demand relative to somatic et al. 1990). We believe that the propensity to tissue levels in temperate nesting waterfowl. Amount nest, timing of nesting and clutch size for indi- of somatic nutrient was determined as the intercept of vidual Ruddy Ducks is an outcome of a complex regressionsof nutrient reserves on clutch nutrients. When those values were unavailable, we used the av- interaction between ultimate causes(strong se- erage present in prelaying females with rapidly devel- lection to nest early) and proximate prerequisites oping follicles, unlessindicated otherwise in footnotes. (strongreliance on nutrient reserves).We suspect that there is variation in nesting dates because (Eggnutrient/somatic nutrient) x 100 not all birds are able to surpassnutrient reserve Species Mineral Lipid Protein thresholdsearly in the nesting seasondue to vari- Ruddy Duck 33.1 9.7 7.0 able condition at arrival onto nesting areas, and Lesser Scaupb 14.0 10.1 4.2 variable rates of nutrient storageafter arrival due Gadwalh 12.4 5.5 4.6 to variation in wetland quality and individual Shoveleld 14.6 7.7 5.9 Ring-necked Duck 17.2 7.8 4.6 variation in foraging proficiency. A significant Canvasback’ -h 5.7 4.3 portion of this variation is related to age (see Redheadg - 5.0 4.5 below). Mallard 16.6 4.0 2.9 aThis study(Table 6), Alisauskasand Ankney(1994). CONTROL OF CLUTCH SIZE b Afton and Ankney(1991). CAnkney and Alisauskas(1991b801); the averageash content of Notwithstanding Arnold and Rohwer’s (199 1) 26.5 g was usedas the estimateof sanatlc mineralin prelayers. d Ankneyand Afton (I 988). skepticism, our study shows that nutrient re- cAlisauskas et al. (1990, unpubl.data). r Banen and Swie (1990); somatictissue estimated from intercepts serves are dually important for egg production, reportedfor regressionsof nutrientreserves vs. massof largestdeveloping folliclein prelayers. as they influence both nesting propensity and, in * Alisauskasand Ankney(1992), Noyesand Jarvis(1985). the caseof mineral reserves,clutch size in Ruddy hNot available. ’ Young (1993);,somatictissue estimated from mean of interceptsre- Ducks. In the caseof protein and mineral, lower pated for regressmnsof nutrientreserves vs. reproductivenutrients. requisite levels of somatic tissue for clutch formation later in the nesting seasonwas related to opposedto early in the season.Fat reserveswere greater reliance on exogenoussources of protein used at the same rates for eggformation regard- and mineral (Figs. 6,7), sothat clutch size apriori less of when ducks initiated clutches, although, is not likely related to mineral or protein reserves again, ducks nesting later did so with lower fat in late nesters.However, a declining fat threshold reservesthan did those nesting earlier. Declining was not related to extent to which fat reserves fat reserves with lateness of nesting is in turn were used to supply clutch fat (Fig. 5). Instead, related to a seasonaldecline in clutch size. Ruddy Ruddy Ducks used fat reserves to supply about Ducks may continue to nest late, laying smaller half of their clutch nutrients, regardlessof nesting clutches, because some fitness may accrue to date. Also, later-nesting ducksinitiated RFG with them, even though it likely is diminished. lessfat reserve than did early nesting ducks, such We believe that less protein and mineral re- that less fat reserve was available for egg for- serve is used later for eggsbecause of seasonal mation. This seasonal decline in available fat increases in the exogenous availability of these reservesis related to, and may explain, the sea- nutrients. For example, emergence of adult chi- sonal decline in clutch size of Ruddy Ducks. ronomids, preceded by maximum instar size of Number of developing follicles in females larvae, in Manitoba is greatest in June or July, nearing end of laying was correlated with fat re- although there can be marked annual variation serves in Shovelers (Ankney and Afton 1988) in chronology of emergence(Wrubleski and Ro- and with protein reserves in Gadwalls (Ankney senberg 1990). Some Ruddy Ducks consumed and Alisauskas 199 1b). In Ruddy Ducks, how- many leeches (Class Hirudinea, Alisauskas and ever, number of developing follicles was related Ankney, unpubl. data), which may also be larger to somatic mineral levels, which suggestsa major or more abundant as the nesting season pro- differencein nutrient constraintson Ruddy Ducks gresses.Thus, to begin nesting in May requires compared to these other species.Although feed- greater reliance on endogenous protein than in ing ecology (including diet during laying) of all June, when preferred animal food is likely more these speciesis different, demand for minerals abundant. Generally, protein is lesslimiting than during eggshell production by Ruddy Ducks is fat because nesting ducks probably select wet- extraordinary: massof a single eggshellis a much lands with abundant protein sources(Alisauskas greater percentage of somatic minerals in pre- RUDDY DUCK NUTRIENT RESERVES 895 laying Ruddy Ducks than it is in other species cupying wetlands that are highly productive of for which data are available (Table 5). Con- such animals (Alisauskas et al. 1990). However, sumption of gastropods, a rich source of min- becauseinvertebrate protein is spatially and tem- erals, increased from laying to early incubation porally ephemeral, there is a premium on its rap- in female Ruddy Ducks (Alisauskasand Ankney, id exploitation (Alisauskas and Ankney 1992). unpubl. data), presumably to replenish depleted Most temperate-nesting ducks obtain exogenous cortical bone (Taylor and Moore 1954). protein at a rate enabling production of one egg/ day, and, other than Gadwalls (Ankney and Ali- AGE-RELATED PATTERNS sauskas 199 1b) and Ruddy Ducks (this study), Older waterfowl generally nest earlier, have high- they do this without using protein reserves (see er average clutch sizes and greater breeding pro- Ankney and Alisauskas 199 la, Alisauskas and pensity than do younger individuals (see Afton Ankney 1992 for reviews). (Although many large- 1984, Rohwer 1992521 and referencestherein). bodied waterfowl, such as arctic-nesting geese, Age-related variation in nutrient reserves may may rely extensively on nutrient reserves for explain at least some age-related patterns in re- clutch formation [e.g., Ankney and MacInnes productive rates (Krapu and Doty 1979, Ali- 19781,there appearto be rate-limiting effectssuch sauskasand Ankney 1987) but in Ruddy Ducks, that egg-laying intervals are > 1 day [Alisauskas it had its greatest effect on propensity to breed. and Ankney 19921).Some of this ability is related Adults showed greater likelihood of initiating to differences between diet during the breeding breeding than did yearlings and this was corre- seasonand other times of the annual cycle (Ank- lated with greater levels of fat and protein re- ney and Alisauskas 199 1a). Although it is phys- servesof adults. Of females that initiated a laying iologically possible to capture and convert ani- cycle, no differenceswere apparent in reserve use mal protein from wetlands directly into egg between ages. In Gadwalls and Canvasbacks, proteins, it may be ecologically impossible to yearlings began laying cycles with less fat than convert either carbohydrate or protein into egg did adults, and relied less on lipid reserves for lipid at a rate sufficient to meet nutrient demands eggproduction than did adults (Ankney and Ali- of one egg/day (Alisauskas et al. 1990). Thus, we sauskas199 1b, Barzen and Serie 1990). In Rud- think that the tactic to store and use body fat is dy Ducks, those yearlings (a minority of them) a trait that has coevolved with the evolution of that surpassednutrient reservethresholds for ini- rapid reproductive rates by temperate-nesting tiation of breeding were indistinguishable from ducks in highly productive habitats. Prenesting breeding adults in their patterns of nutrient re- storage of nutrient reserves also is probably an serve use for eggformation. Becausesome year- important factor that can advance nesting sea- lings breed, and because their breeding propen- sons before exogenousnutrients reach an abun- sity is related to their ability to acquire sufficient dance that is sufficient to allow reliance on them nutrient reserves, which, in turn, is related to for forming eggs. adeptness at foraging, it is likely that breeding Although Ruddy Ducks use the same land- constraint rather than restraint is operating as scapesas other ducks nesting in the prairie pot- suggestedfor waterfowl by Rohwer (1992). hole region of the northern plains of North America, they are extraordinary in their extreme CONCLUSIONS nutrient demand for eggproduction. The pattern We believe (Alisauskas and Ankney 1992) that of nutrient reserve use for egg production by three factors were responsible for the evolution Ruddy Ducks fits those of waterfowl in general of high rates of nutrient secretion by ducks rel- as use of lipid reservesduring egg production is ative to other species.First, it is a necessity for virtually ubiquitous (Ankney and Alisauskas their large nutrient-rich eggs(Lack 1967, 1968). 199 la). However, Ruddy Ducks also showedsig- Second, we agree with Arnold et al. (1987) that nificant use of both somatic protein (this study) open-nesting waterfowl have undergone strong and mineral (Tome 1984, this study) for pro- selection to reduce the number of days that eggs ducing eggs.We believe that predation pressure and females are exposed to predators. Finally, and ephemeral food sourceshave been respon- we also think that because of their adaptations sible for evolution of rapid nutrient secretion in for exploiting aquatic invertebrates, most species Ruddy Ducks, as argued for other temperate are capable of avoiding protein shortageby oc- ducks. Storageand use of fat, protein and mineral 896 RAY T. ALISAUSKAS AND C. DAVISON ANKNEY reserves, in addition to adaptations for efficient ANKNEY,C. D., AND R. T. ALISAUSKAS. 1991a. The exploitation of benthic invertebrates from highly use of nutrient reserves by breeding waterfowl. Proc. 20th Inter. Omith. Congress,Christchurch, productive wetlands, enable Ruddy Ducks to New Zealand. Vol. I:2 170-2 176. produce exceptionally large eggsat a rapid rate. ANKNEY,C. D., AND R. T. ALISAUSKAS.1991b. Nu- However, proximate limitations on attempted trient-reserve dynamics and diet of breeding fe- breeding and clutch size operate in many cases male Gadwalls.~Condor93:799-810. becauseof individual variation in ability to store ANKNEY.C. D.. A. D. EON. AND R. T. ALISAUSKAS. 1991. The role of nutrient reserves in limiting sufficient nutrient reserves, and to find and ex- waterfowl reproduction. Condor 93:1029-1032.- ploit high quality patches of aquatic inverte- ARNOLD.T. W. 1994. Effects of suuolemental food brates. on egg production in American sots. Auk 110: 337-350. ACKNOWLEDGMENTS ARNOLD,T. W., AND F. C. ROH~ER. 1991. Do egg formation costs limit clutch size in waterfowl? A We thank Bruce Batt, former Director of the Delta skeptical view. Condor 93:1032-1038. Waterfowl and Wetlands ResearchStation, for his sup- ARNOLD, T. W., F. C. ROHWER,AND T. ARMSTRONG. port of this research,Sandi Johnsonfor help with field- 1987. Eggviability, nest predation, and the adap- work, and Jamile Amery for dissectionsand chemical tive significance of clutch size in prairie ducks. analysis.This work was funded bv the Natural Sciences Am. Nat. 130:643-653. and EngineeringResearch Council (Canada) through a BARZEN,J. A., ANDJ. R. SERIE. 1990. Nutrient reserve researchgrant to CDA and a postdoctoral fellowship dynamics of breeding Canvasbacks.Auk 107:75- to RTA. Todd Arnold conducted roadside censuses 85. and graciouslysupplied information on arrival, nest- BELLROSE,F. C. 1976. Ducks, geese and swans of ing, and clutch size chronology. We received valuable North America. StackpoleBooks, Harrisburg,PA. comments from Al Afton, Todd Arnold, Bob Clark, DROBNEY.R. D. 1991. Nutrient limitation of clutch Tony Diamond, JonathanThompson, Mike Tome and size in waterfowl: is there a universal hypothesis? two anonymous reviewers. Condor 93:1026-1028. EADIE,J. M. 1989. Alternative reproductive tactics LITERATURE CITED in a precocial bird: The ecology and evolution of AFTON, A. D. 1984. Influence of age and time on brood parasitism in goldeneyes.Ph.D.diss., Univ. reproductiveperformance of female LesserScaup. of British Columbia, Vancouver. Auk 101:255-265. EULISS,N. H., R. L. JARVIS,AND D. S. GILMER. 1991. AFTON,A. D., ANDR. D. SAYLER. 1982. Social court- Feeding ecology of waterfowl wintering on evap- ship and pairbonding of Common Goldeneyes, oration ponds in California. Condor 93:582-590. Bucephalaclangula, wintering in Minnesota. Can. GRAY, B. J. 1980. Reproduction, energeticsand so- Field Nat. 96:295-300. cial structureofthe Ruddy Duck. Ph.D.diss.,Univ. &TON,‘ A. D., AND C. D. ANKNEY. 1991. Nutrient of California, Davis, CA. reserve dynamics of breeding Lesser Scaup:a test HOCHBAUM,H. A. 1942. Sex and age determination of competing hypotheses.Condor 93:89-97. of waterfowl by cloaca1examination. Trans. N. ALISAUSKAS,R. T., AND C. D. ANKNEY. 1987. Age- Amer. Wildl. Conf. 7~299-307. related variation in the nutrient reservesof breed- HCXHBAUM,H. A. 1944. The Canvasbackon a Prai- ing American Coots (Fulica americana). Can. J. rie Marsh. Univ. Nebraska Press, Lincoln, NE. Zool. 65:2417-2420. HOPPE,R. T., L. M. SMITH,AND D. B.WEBSTER.1986. ALISAUSKAS,R. T., AND C. D. ANKNEY. 1992. The Foods of wintering diving ducks in South Caro- cost of egg laying and its relationship to nutrient lina. J. Field Omith. 57: 126-l 34. reservesin waterfowl, p. 30-6 1. In B.D.J. Batt et JOYNER,D. E. 1983. Parasiticegg laying in Redheads al. [eds.], Ecology and management of breeding and Ruddy Ducks in Utah: incidenceand success. waterfowl. Univ. of Minnesota Press, Minneap- Auk 100:717-725. olis. K~APU. G. L. 1974. Feedine ecoloev of Pintail hens ALISAUSKAS,R. T., AND C. D. ANKNEY. 1994. Costs during reproduction. Auk 91:27x-290. and rates of eggformation in Ruddy Ducks. Con- KRApu, G. L. 1981. The role of nutrient reservesin dor 96:11-18. Mallard reproduction. Auk 98:29-38. ALISAUSKAS, R. T., R. T. EBERI%4RDT, AND C. D. K~APU, G. L., AND H. A. Don. 1979. Age-related ANKNEY. 1990. Nutrient reserves of breeding aspectsof Mallard reproduction. Wildfowl 30:35- Ring-neckedDucks (Aythyacollaris). Can. J. Zool. 39. 68:2524-2530. LACK,D. 1967. The significanceof clutch-sizein wa- ANKNEY, C. D., ANDC. D. MACINNES. 1978. Nutrient terfowl. Wildfowl 18:125-128. reservesand reproductive performance of female LACK, D. 1968. Ecological adaptations for breeding Lesser Snow Geese. Auk 95:459-47 1. in birds. Methuen, London. ANKNEY.C. D.. AND A. D. AFTON. 1988. Bioener- LAUGHLIN,K. F. 1975. The bioenergeticsof the Tuft- get& of breeding Northern Shovelers: diet, nu- ed Duck, Aythya filigula (L.). Ph.D.diss., Univ. trient reserves,clutch size and incubation. Condor of Sterling, Sterling, England. 90~459-472. Novns, J. H., AND R. L. JARVIS. 1985. Diet and nu- RUDDY DUCK NUTRIENT RESERVES 897

trition of breeding female Redhead and Canvas- vasbacks:frequency, success,and individual be- back ducksin Nevada. J. Wildl. Manage. 49:203- havior. Auk 110:57-69. 211. STOVDT.J. H. 1982. Habitat use and nroductivitv of Pm~~owsru, M. W., ANDP. R. EVANS. 1982. Clutch Canvasbacksin southwesternManitoba, 196l-72. parasitism and nesting interference between shel- U.S. Fish Wildl. Serv. Snec. Sci. Ren. Wildl. 248. ducks at Aberlady Bay. Wildfowl 33: 159-163. SWANSON,G. A., AND J. C.-BARTONEK.~1970. Bias RAmow, R. J. 1973. Iocomotor mechanismsin North associatedwith food analysisin gizzards of Blue- American ducks. Wilson Bull. 85:295-307. winged Teal. J. Wildl. Manage. 34:739-746. REYNOLDS,C. M. 1972. Mute swan weights in rela- TAYLOR,T. G., AND J. H. MOORE. 1954. Skeletal tion to breeding. Wildfowl 23: 11l-l 18. depletion in hens laying on a low calcium diet. ROHWER,F. C. 1992. The evolution of reproductive Brit. J. Nutr. 8:112-124. patterns in waterfowl, p. 486-539. In B.D.B. Batt TOME, M. W. 1981. Reproductive bioenergetics of et al. [eds.],Ecology and Management of Breeding female Ruddy Ducks in Manitoba. M.Sc.thesis, Waterfowl. Univ. Minnesota Press. Minneanolis. University of Maine, Orono, Maine. ROHWER, F. C., ANDM. G. ANDERSON. 1988. Female- TOME.M. W. 1984. Chanaesin nutrient reservesand biased philopatry, monogamy, and the timing of organ size of female Ruddy Ducks breeding in pair formation in migratory waterfowl, p. 187- Manitoba. Auk 101:830-837. 22 1. In R. F. Johnson, [ed.], Current ornithology, TOME, M. W. 1988. Optimal foraging: food patch Vol. V. Plenum Publishing Corp., New York. depletion by Ruddy Ducks. Oecologia (Berl.) 76: RYDER,J. P. 1970. A possiblefactor in the evolution 27-36. of clutch size in Ross’ Goose. Wilson Bull. 82:5- TOME,M. W. 1989. Food patch depletion by Ruddy 13. Ducks: foragingby expectationrules. Can. J. Zool. SARGEANT,A. B., AND D. G. RAVELING. 1992. Mor- 6712751-2755. tality during the breeding season,p. 396422. In TOME,M. W. 1991. Diurnal activity budgetof female B.D.J. Batt et al. [eds.], Ecologyand management Ruddy Ducks breeding in Manitoba. Wilson Bull. of breeding waterfowl. Univ. of Minnesota Press, 103:183-189. Minneapolis. TOME,M. W., AND D. A. WRUBLESKL1988. Under- SAS INSTITWIX 1990. SASSTAT user’s guide, Ver- water foraging behavior of Canvasbacks,Lesser sion 6, 4th ed. SAS Institute, Cary, NC. Scaups,and Ruddy Ducks. Condor 90: 168-l 72. SIEGFRIED,W. R. 1973. Summer foods and feeding WOODIN,M. C.. AND G. A. SWANSON.1989. Foods of the Ruddy Duck in Manitoba. Can. J. Zool. 5 1: and dietary strategies of prairie-nesting Ruddy 1293-1297. Ducks and Redheads. Condor 91:280-287. SIEGFRIED,W. R. 1976a. Breeding biology and par- WRUBLESKI,D. A., ANDD. M. ROGENBERG.1990. The asitism in the Ruddy Duck. Wilson Bull. 88:566- Chironomidae (Dintera) of Bone Pile Pond. Delta 574. Marsh, Manitoba, Cana’da.Wetlands l&243-275. SIEGFRIED,W. R. 1976b. Social organization in Rud- YOUNG,A. D. 1993. Intraspecific variation in use of dy and Maccoa Ducks. Auk 93:56CL570. nutrient reserves by breeding female Mallards. SORENSON,M. D. 1993. Parasitic egg laying in Can- Condor 95:45-6.