Wilson Bull, 106(4), 1994, pp. 679-688 DYNAMICS OF OVARIAN FOLLICLES IN BREEDING DUCKS Daniel Esler’ — Abstract. quantified ovarian rapid follicle growth (RFG) and regression of postovu- I latory follicles of Northern Pintails {Anas acuta), American Wigeon {A. americana), and LesserScaup {Aythya affinis) by a method that accounted forwithin-day variation in follicle size. Objective methods for identifying onset of RFG also are presented; this is crucial for accurate classification of breeding status. Duration of RFG was estimated as 4.2, 5.1, and 5.0 days for pintails, wigeon, and scaup, respectively; these are shorter than previously reported. Diameters offollicles at the beginning ofRFG were estimated to be 8.2, 6.9, and 7.9 mm for pintails, wigeon, and scaup, respectively. For all species, RFG was linear, using follicle diameters, and exponential, using dry masses. Models of RFG and postovulatory follicle regression have practical value for calculating nest initiation dates, number of de- velopingfollicles,clutch size, renesting intervals,and dailyenergy andnutrientcommitment to reproduction of collected breeding females. Received 12 November 1993, accepted 20 April 1994. Rapid follicle growth (RFG) is the period from the time an ovarian follicle begins rapidly accumulating yolk until ovulation (see Lofts and Murton [1973] for descriptions of ovary structure and control). In ovaries of breeding birds, initiation of RFG of successive follicles is staggered in accordance with egg-laying interval. As a result, developing follicles have a distinct size hierarchy that corresponds to the order in which they will be ovulated. Postovulatory follicles are the follicle structures re- maining after ovulation (Lofts and Murton 1973); they regress over time, resulting, similarly, in a size heirarchy within an ovary. Based on this information and assumptions about rates of egg laying, models of RFG and postovulatory regression through time can be developed. Previous studies have described ovarian follicle growth based on changes in mean follicle size by day (e.g., Calverley and Boag 1977, Astheimer and Grau 1990, Alisauskas and Ankney 1992) but did not present model equations. No previous investigators presented continuous models of RFG or postovulatory follicle regression with predictive ca- pabilities that could be used in subsequent studies. My objective was to quantify ovarian follicle dynamics of Northern Pintails (Anas acutcr, hereafter pintails), American Wigeon (A. aiticricana: hereafter wigeon), and Lesser Scaup {Aythya affinis: hereafter scaup) by methods that accounted for within-day variation and objectively identilied ' AlaskaF'ishand Wildlife ResearchCenter. National Biological Survey. 1011 FT F'udorRoad. Anchorage. Alaska 99.S03. 679 680 THE WILSON BULLETIN • Vol. 106, No. 4, December 1994 onset of RFG. I also present models that can be used to discern aspects of breeding biology from ovaries of collected females. METHODS Female pintails were collected in 1990 and 1991 at study sites on Yukon Delta National Wildlife Refuge (NWR) (61°26'N, 165°27'W) and Yukon Flats NWR (66°25'N, 149°59'W), Alaska. In 1991, female wigeon and scaup were collected on Yukon Flats NWR. Ovaries were removed and preserved in 10% formalin. In the laboratory, largest diameters in the plane ofthe stigma ofpreovulatory and postovulatory follicles were measured. Dry masses of preovulatory follicles were recorded. Because follicles preserved in formalin may be fixed in deformed shapes, dry masses of preovulatory follicles may be more accurate than diameters. However, analyses using diameters are advantageous because these measures are obtainable in the field or lab without additional processing. Only laying females (i.e., those that had ovulated at least one follicle) were used as samples for modeling, because only those could have a general hierarchy assigned by day. Ovaries with follicles broken during collection or dissection were included only if the po- sition in the hierarchy ofthe broken follicle was known with certainty. For late-layers with a gap in the follicle hierarchy, only large, developing follicles were used. Within each ovary, follicles were assigned to a DAY, which was a rough estimate ofthe time before ovulation. Forexample, the largest follicle from each ovary was assigned DAY = 1, and it was assumed that it would have ovulated with 24 h. The second largest follicle was assigned DAY = 2, and so forth. Sample sizes by species and DAY are presented in Table 1. For these analyses, I assumed constant laying intervals of 24 h for all species (Alisauskas and Ankney 1992). Rather than describe a rough growth curve based on mean follicle size for each DAY, I incorporated within-day variation in follicle sizes into continuous models. I corrected DAY (CORRDAY) for individual birds, using an adjustment based on that bird’s largest follicle dry mass (DRY) relative to the range in mass between the smallest DAY 1 follicle (SMLFOLL) of the species (Table 1) and an estimate of the individual’s follicle mass at ovulation (LRGFOLL). LRGFOLL was either(1) dry mass ofthe individual’soviductal egg yolk or (2) average yolk dry mass from a sample of oviductal and laid eggs. The former was used, when possible, to account for variation in egg composition among individuals, which is greater than variation within clutches (Duncan 1987). Thus, CORRDAY estimated time before ovulation for the largest follicle of each individual as: CORRDAY = (LRGFOLL - DRY)/(LRGFOLL - SMLFOLL). For other follicles of each individual, CORRDAYwascalculatedbyaddingDAYforeachfollicleandCORRDAYfromthe largest follicle. CORRDAY forthe largest postovulatory follicle ofeach ovary (i.e., days afterovulation) was estimated using the correction fordeveloping follicles: CORRDAY = 1 — (LRGFOLL - DRY)/(LRGFOLL - SMLFOLL). Because postovulatory follicle diameters were subject to more measurement error, CORRDAY based on preovulatory follicles likely was more accurate than deriving a correction factor based on postovulatory follicle sizes. I used an iterative approach to quantify beginning ofRFG for each species. First, I used linear regressions to describe relationships between CORRDAY and follicle diameter for data sets consisting of (1) follicles clearly before RFG (i.e., CORRDAY > 6.0) and (2) follicles definitely in RFG (i.e., CORRDAY < 3.5). Exclusion ofthat range ofpoints avoid- ed using data near the beginning of RFG for all species (Fig. 1). In the second iteration, separate linear regressions were used to describe data less and greater than CORRDAY at the intersectionofmodelsfromthe first iteration.The intersectionofmodelsfromthe second 1 Esler• DYNAMICS OF OVARIAN FOLLICLES 681 Table 1 Ovarian Follicle Sizes of Breeding Ducks NorthernPintail AmericanWigeon LesserScaup Day“ Diameter(mm) Dry mass(g) Diameter(mm) Dry mass(g) Diameter(mm) Dry mass(g) 1 Range 26.3-33.4 4.56-7.80 28.3-34.5 5.63-8.18 30.8-35.6 7.27-10.63 Mean 29.6 6.32 31.4 6.94 33.4 8.71 N 42 40 11 11 15 15 2 Range 20.1-28.4 1.94-5.37 22.7-28.5 2.95-5.66 26.7-30.8 4.12-6.19 Mean 24.1 3.43 26.2 4.27 28.8 5.27 N 42 42 10 10 14 14 3 Range 14.4-23.2 0.65-2.62 17.8-23.3 1.30-2.75 18.6-24.7 1.50-3.93 Mean 17.8 1.38 20.5 2.03 22.0 2.55 N 41 41 8 8 14 14 4 Range 9.3-16.3 0.13-1.06 11.0-18.9 0.27-1.43 11.4-18.5 0.25-1.24 Mean 12.0 0.42 14.6 0.78 15.5 0.81 N 34 31 9 8 15 15 5 Range 6.0-11.8 0.01-0.34 7.6-14.4 0.07-0.57 8.2-13.2 0.06-0.40 Mean 8.1 0.10 10.0 0.24 10.5 0.20 N 26 21 7 6 13 12 6 Range 5.3-8.2 0.01-0.09 5.9-10.9 0.02-0.22 6.8-8.9 0.02-0.07 Mean 6.8 0.04 7.3 0.08 7.7 0.05 N 26 18 9 6 11 8 7 Range 4.7-7.5 0.01-0.05 6.1-7.2 6.4-8. 0.03-0.05 Mean 6.2 0.02 6.7 7.2 0.04 N 24 12 7 10 5 “Wherethelargestinaseriesofdevelopingfolliclesfromlayingfemales(whichwouldhaveovulatedwithin24hours) = Day 1,the next largest = Day 2,etc. iteration estimated CORRDAY and follicle diameter at the onset of RFC. I calculated 95% confidence limits around the CORRDAY estimate (Sokal and Rohlf 1981:498). Polynomial models of RFC (i.e., for data with CORRDAY less than the estimate of beginning of RFC) and postovulatory follicle regression were created to de.scribe relation- ships between follicle sizes and CORRDAY, with CORRDAY up to the third order. Higher- order variables were removed if nonsignificant iP > 0.01). Polynomial models afso were derived with CORRDAY as the dependent variable, so that CORRDAY could be predicted from follicles ofcollected birds. RESULT.S CORRDAY (and 95% confidence limits) at the beginning of RFG (i.e., duration of RFG) were estimated to be 4.2 (3.8^,6), 5.1 (4.7-5.6), and 5.0 (4.5-5.4) days for pintails, wigeon, and scaup, respectively; follicle diameters were estimated as 8.2, 6.9, and 7.9 mm, respectively. CORRDAY Follicle diameters were linearly related to during RFCj for all species (Fig. I). Model intercepts estimated diameters at o\illation as mm 32.9, 33.8, and 37.1 for pintails, wigeon, and scaup, respectively. 682 THE WILSON BULLETIN • Vol. 106, No. 4, December 1994 DAYS BEFORE OVULATION Fig. 1. Rapid follicle growth ofthree duck species based on ovarian follicle diameters. Vertical dashed lines represent estimates ofbeginning ofrapid follicle growth. Growth curves of follicle dry masses were best fit with second-order polynomial expressions (Fig. 2). Follicle dry masses at ovulation were estimated to be 8.2, 8.3, and 1 1.0 g for pintails, wigeon, and scaup, re- spectively. Predictive models of RFG (Table 2) estimated CORRDAY Esler• DYNAMICS OF OVARIAN FOLLICLES 683 DAYS BEFORE OVULATION FiCi. 2. Rapid follicle growth ofthree duck species based on ovarian follicle dry masses. Vertical dashed lines represent estimates of beginning of rapid follicle growth. with linear models of diameter and third-order polynomials of dry mass for all species. Postovulatory follicle regression was described by second-order poly- nomials for all species (Fig. 3). Intercepts of these models estimated post- mm ovulatory follicle diameter immediately after ovulation as 13.8 for 684 THE WILSON BULLETIN • Vol. 106, No. 4, December 1994 Table 2 Predictive Models'* Estimating CORRDAY*^ erom Ovarian Eollicle Diameters (DIA) AND Dry Masses (DRY) Group Equation Rapidly growing follicles Northern Pintail Diameter CORRDAY = 5.451 - 0.164DIA Dry mass CORRDAY = 3.894 - 1.162DRY + 0.168DRY2 - 0.011DRY^ American Wigeon Diameter CORRDAY - 6.377 - 0.187DIA Dry mass CORRDAY = 4.800 - 1.607DRY + 0.263DRY- - O.OHDRY^ Lesser Scaup Diameter CORRDAY = 6.217 - 0.166DIA Dry mass CORRDAY = 4.524 - 1.088DRY + 0.134DRY2 - 0.007DRY3 'ostovulatory follicles Northern Pintail CORRDAY = 6.443 - 0.748DIA + 0.023DIA2 American Wigeon CORRDAY = 8.714 - 1.105DIA + 0.038DIA2 Lesser Scaup CORRDAY = 7.744 - 0.823DIA + 0.023DIA2 “Allmodelsn >0.96,P<0.001 forrapidlygrowingfollicles, >0.87,P< 0.001 forpostovulatoryfollicles. '’Thenumberofdaysuntilovulationforrapidlygrowingfolliclesanddayssinceovulationforpostovulatoryfollicles. mm both pintails and wigeon, and 18.2 for scaup. Predictive models (Ta- CORRDAY ble 2) can be used to estimate based on diameters ofpostovu- latory follicles. DISCUSSION Consistent and objective criteria have not been used for defining be- ginning of RFG for ducks (i.e., defining a measure for distinguishing between developing and nondeveloping follicles), which is essential for determining breeding status. Ovary masses of 3.0 g have been used for pintails (Krapu 1974), Mallards {Anas platyrhynchos; Krapu 1981), and Ring-necked Ducks (Aythya collaris; Hohman 1986). Follicle diameters have been used for Ruddy Ducks {Oxyura jamaicensis; 8.0 mm; Tome 1984), Canvasbacks (A. valisineria; 7.5 mm; Barzen and Serie 1990), and pintails (6.0 mm; Phillips and van Tienhoven 1962, Mann and Sedinger 1993). Follicle dry mass of0.10 g was used for Northern Shovelers {Anas clypeata\ Ankney and Afton 1988) and Ring-necked Ducks (Alisauskas et al. 1990). Conservative estimates were obtained by using dry mass of the second smallest “developing” follicle from samples of hens with complete sets of follicles; criteria by this method have included 0.20 g for scaup (Afton and Ankney 1991), 0.40 g for Gadwall {A. strepera\ Esler• DYNAMICS OF OVARIAN FOLLICLES 685 DAYS AFTER OVULATION Fig. 3. Regres.sion of postovulatory follicles of three duck species. Ankney and Alisauskas 1991), and 0.39 g for Mallards (Young 1993). Clearly, it would be valuable to derive consistent methods for interpre- tation of breeding status from ovaries. Otherwise, there is danger of mis- interpreting breeding status and affecting associated analyses. Initiation of RFC can be determined objectively by the methods pre- sented here. To apply this information to determine waterfowl breeding 686 THE WILSON BULLETIN • Vol. 106, No. 4, December 1994 status, I suggest adding a conservative buffer to follicle size estimates at the beginning of RFG to be certain that follicles are in RFG. For example, mm for the species in this study, 10 is an appropriate distinction between RFG and non-RFG follicles. Only three pintail follicles were >10 mm before the beginning ofRFG (Fig. 1). From polynomial models describing relationships between follicle diameters and dry masses {P < 0.001, > mm 0.98), I found that 10 corresponded to 0.12, 0.15, and 0.10 g dry mass for pintails, wigeon, and scaup, respectively; thus, 0.15 g dry mass also is an appropriate distinction for these species. Follicles before the beginning of RFG were <0.15 g, again with the exception of the three pintail follicles. Duration of RFG can be estimated in several ways. Rough estimates can be obtained by multiplying the maximum number of developing fol- licles by the egg-laying interval (Alisauskas and Ankney 1992). Renest intervals have been used as a maximum estimate (Grau 1984). Duration of RFG also has been estimated by examining rings in cross-sections of yolk that form as yolk material is deposited; each pair of rings was pre- sumed to represent daily growth (e.g., Grau 1976, 1984; Roudybush et al. 1979; Astheimer and Grau 1990). However, Alisauskas and Ankney (1994) suggested that the ring method may not work for laying waterfowl with a diphasic feeding regime that may lay down more than one set of rings each day. This pattern was found in Japanese Quail {Coturnix co- turnix) fed twice daily (Dobbs et al. 1976). The method I presented here has advantages over other methods because the results are more exact and, unlike the ring method, laboratory analyses are not required and assumptions regarding yolk deposition are not necessary. However, col- lection of birds is required. My estimates of duration of RFG are shorter than the six days previ- ously described for these species (Phillips and van Tienhoven 1962, Ali- sauskas and Ankney 1992). These results are corroborated by examining ranges and means of follicle sizes by DAY (Table 1) for each species; follicles were nondeveloped, on average, on DAY 5 (4-5 days from ovu- lation) for pintails and DAY 6 (5-6 days from ovulation) for wigeon and scaup. Without comparably treated data from mid-continent breeding ar- eas, it is unknown ifthere is geographic variability in RFG duration. Short RFG duration may be advantageous for species that exploit unpredictable food resources or that experience high rates of nest predation (Alisauskas and Ankney 1992). However, shorter RFG results in increased daily costs of egg production (Alisauskas and Ankney 1992, 1994). Although researchers have used ovary characteristics to determine wa- terfowl breeding status, other values of RFG models have not been ex- ploited. When applied to ovaries of individuals, these models can identify Esler• DYNAMICS OF OVARIAN FOLLICLES 687 important aspects of their basic breeding biology. For example, nest ini- tiation dates ofbirds with developing follicles can be estimated accurately CORRDAY by determining and adding a day for the time the follicle is in the oviduct (Alisauskas and Ankney 1992). Time of day of ovulation also can be estimated. Models of RFG allow detection of breaks in the follicle hierarchy of individuals late in their laying sequence, differenti- ating follicles that would be laid from those that would not; in such cases, clutch size is the number of developing follicles plus the number of post- ovulatory follicles. Some analyses of nutrient reserves require accurate distinction of the number of follicles remaining to be laid (e.g., Ankney and Alisauskas 1991, Esler and Grand 1994). Renesting intervals can be determined by estimating days since ovulation of the last follicle of the first nest (using postovulatory follicle models) and days until laying of the first egg of the renest (using models of RFG). Furthermore, for as- sessments of nutrient and energy commitment to clutch formation (e.g., Drobney 1980, Astheimer 1986, Alisauskas and Ankney 1994), RFG models could provide accurate daily changes. Postovulatory follicles have been used as objective measures of clutch size and incidence of brood parasitism (e.g., Kennedy et al. 1989). Per- sistence of postovulatory follicles is variable among taxa (see review in Semel and Sherman 1991). Postovulatory follicles of Wood Ducks {Aix sponsci) were detectable for <30 days after ovulation (Semel and Sherman 1991); I suspect this is true for species in this study also. Models for pintails, wigeon, and scaup described regression ofpostovulatory follicles for only a few days after ovulation; these have value for determining clutch size and, for birds early in incubation, how long they have been incubating. ACKNOWLEDGMENTS I thank D. L. Boyd, P. L. Flint, J. F. Kormendy, S. McDonald, and R. Migoya for assis- tance with bird collections. J. B. Grand and the staffs of the Yukon Delta and Yukon Flats NWRs are thanked for logistical and administrative support. Ovarian examination was as- sisted by P. L. Flint, T. F. Fondell, J. B. Grand, and J. F. Kormendy. Dry follicles were weighed by S. A. Lee. J. Beebee assisted with data analysis. I thank R. T. Ali.sauskas, D. V. Derk.sen, C. R. Ely, P. L. Flint, J. B. Grand, C. R. Grau, M. R. Petersen, and M. W. Tome for review of the manuscript. LITERATURE CITED Arrow, A. D. AWt) C. D. Anknf-y. 1991. Nutrient-reserve dynamics of breeding Lesser Scaup: a test ofcompeting hypotheses. Condor 93:89-97. Alisauskas, R. T. and C. D. Anknly. 1992. The cost ofegg laying and its relationship to nutrient reserves in waterfowl. Pp. 30-61 in Ecology and management of breeding waterfowl (B. D. J. Batt, A. D. Alton. M. G. Anderson, C. D. Ankney. D. H. .lohnson. J. A. Kadlec, and G. L. Krapu,cds.). Univ.ofMinnesota Press, Minneapolis. Minnesota. 688 THE WILSON BULLETIN • Vol. 106, No. 4, December 1994 AND . 1994. Costs and rates ofegg formation in Ruddy Ducks. Condor 96: 1-18. 1 , R. T. Eberhardt, and C. D. Ankney. 1990. Nutrient reserves of breeding Ring- necked Ducks (Aythyo collaris). Can. J. Zool. 68:2524-2530. Ankney, C. D. and A. D. Afton. 1988. Bioenergetics of breeding Northern Shovelers: diet, nutrient reserves, clutch size, and incubation. Condor 90:459^72. AND R. T. Alisauskas. 1991. Nutrient-reservedynamicsanddietofbreedingfemale Gadwalls. Condor 93:799-810. Astheimer, L. B. 1986. Egg formation in Cassin’s Auklet. Auk 103:682-693. AND C. R. Grau. 1990. A comparison of yolk growth rates in seabird eggs. Ibis 132:380-394. Barzen, J. a. and J. R. Serie. 1990. Nutrient reserve dynamics ofbreeding Canvasbacks. Auk 107:75-85. Calverley, B. K. and D. A. Boag. 1977. Reproductive potential in parkland- and arctic- nesting populations ofMallards and Pintails (Anatidae). Can. J. Zool. 55:1242-1251. Dobbs, J. C., C. R. Grau, T. Roudybush, and J. Wathen. 1976. Yolk ring structure of quail subjected to food deprivation and refeeding. Poultry Sci. 55:2028-2029. Drobney, R. D. 1980. Reproductive bioenergetics ofWood Ducks. Auk 97:480^90. Duncan, D. C. 1987. Variation and heritability in egg size ofthe Northern Pintail. Can. J. Zool. 65:992-996. Esler, D. and j. B. Grand. 1994. The role of nutrient reserves for clutch formation by Northern Pintails in Alaska. Condor 96:422^32. Grau, C. R. 1976. Ring structure ofavian egg yolk. Poultry Sci. 55:1418-1422. . 1984. Egg formation. Pp. 33-57 in Seabird energetics (G. C. Whittow and H. Rahn, eds.). Plenum Press, New York, New York. Hohman, W. L. 1986. Changes in body weight and body composition of breeding Ring- necked Ducks (Aythya collaris). Auk 103:181-188. Kennedy, E. D., P. C. Stouffer, and H. W. Power. 1989. Postovulatory follicles as a measure of clutch size and brood parasitism in European Starlings. Condor 91:471- 473. Krapu, G. L. 1974. Eeeding ecology ofPintail hens during reproduction. Auk 91:278-290. . 1981. The role of nutrient reserves in Mallard reproduction. Auk 98:29-38. Lofts, B. and R. K. Murton. 1973. Reproduction in birds. Pp. 1-107 in Avian biology, Vol. 3 (D. S. Earner, J. R. King, and K. C. Parkes, eds.). Academic Press, New York, New York. Mann, E E. and J. S. Sedinger. 1993. Nutrient-reserve dynamics and control of clutch size in Northern Pintails breeding in Alaska. Auk 110:264-278. Phillips, R. E. and A. van Tienhoven. 1962. Some physiological correlates of Pintail reproductive behavior. Condor 64:291-299. Roudybush, T. E., C. R. Grau, M. R. Petersen, D. G. Ainley, K. V. Hirsch, A. P. Gilman, AND S. M. Patten. 1979. Yolkformation in some charadriiformbirds. Condor 81:293- 298. Semel, B. and P. Sherman. 1991. Ovarian follicles do not reveal laying histories ofpost- incubation Wood Ducks. Wilson Bull. 103:703-705. SoKAL, R. R. AND P. J. Rohlf. 1981. Biometry. W. H. Freeman and Co., New York, New York. Tome, M. W. 1984. Changes in nutrient reserves and organ size of female Ruddy Ducks breeding in Manitoba. Auk 101:830-837. Young, A. D. 1993. Intraspecific variation in the use of nutrient reserves by breeding female Mallards. Condor 95:45-56.