2007 NORTHEASTERN NATURALIST 14(2):269–278 Fecundity, Age and Growth, and Diet of Fundulus diaphanus (Banded Killifish) in Presque Isle Bay, Lake Erie Edward C. Phillips1,*, Yvonne Ewert1, and Patricia A. Speares1 Abstract - Fundulus diaphanus (banded killifish) were the most abundant fish collected in the shallow water areas of Presque Isle Bay, making up 86% of fish collected. Peak spawning dates based on gonadosomatic index and ovary observations were July 9, 2003 and July 18, 2004. Females became sexually mature at age 1. During 2004, the majority of spawning occurred over a 20-day period between July 9 and July 29. Total fecundity increased with both size and age, and the mean total number of eggs produced was 526 ± 37 at age 1 (67 mm ± 1.8), 744 ± 38 at age 2 (79 mm ± 2.4), and 1062 ± 43 at age 3 (93 mm ± 4.1). Based on comparison of the total number of eggs and number of eggs with yolk contained within ovaries, three clutches of eggs may be released during each spawning season. Diets were similar between summer and fall; banded killifish fed almost exclusively on cladocerans and benthic macroinvertebrates. Introduction Fundulus diaphanus Le Sueur (banded killifish) is an abundant fish in the shallow water areas of Presque Isle Bay, Lake Erie. It is an important forage fish for gamefish like Micropterus salmoides Lacépède (largemouth bass), Micropterus dolomieu Lacépède (smallmouth bass), and Esox lucius Linnaeus (northern pike) in Presque Isle Bay (Cooper 1983). Many successful spawning strategies have evolved in various groups of fish. Some species produce eggs continuously during the spawning season, while others produce eggs only once (Burt et al. 1988, Heins 1990, Heins et al. 1992). Differences also occur on the day of spawning, with some species using many spawning acts to release their eggs and others releasing all of their eggs at once (Burt et al. 1988, Heins 1990, Heins et al. 1992). Based on previous observations of spawning behavior (Fournier and Magnin 1975, Richardson 1939, Witt 1979), banded killifish are oviparous fish with groupsynchronous ovum development that release multiple clutches of eggs each reproductive season (Heins 1990, Heins et al. 1992, Wallace and Selman 1981). In other species with similar spawning habits, ovaries cycle through a series of reproductive stages as clutches of synchronously developing eggs are produced from a group of heterogeneous vitellogenic eggs by yolk uptake of eggs in the clutch (Heins et al. 1992, Selman and Wallace 1983, Wallace and Selman 1981). One reproductive measurement that effectively predicts 1Biology Department, Gannon University, 109 University Square, Erie, PA 16541. *Corresponding author - phillips010@gannon.edu. 270 Northeastern Naturalist Vol. 14, No. 2 the spawning peak is the gonadosomatic index (GSI, ovary weight/total body weight x 100) (Barron and Albin 2004, Shein et al. 2004). Banded killifish spawn when water temperatures are from 21 oC to 23 oC (Fournier and Magnin 1975, Richardson 1939). Spawning behavior has been previously described (Richardson 1939, Witt 1979), and female banded killifish release eggs multiple times when stimulated by a male. Fournier and Magnin (1975) found that eggs developed quickly over a six-week period prior to spawning, and that the majority of females spawned within a three-week period in two lakes in Quebec, Canada. They also concluded that sexual maturity was more likely a function of size than of age. Fecundity in banded killifish has shown great variation. Fritz and Garside (1975) found average fecundity to be 128 eggs in Porters Lake and 88 eggs in Kejimkujik Lake, NS, Canada. In two lakes in Quebec, Canada, Fournier and Magnin (1975) found maximum fecundity to be 426 eggs in a 101-mm Lake Renaud female and 266 eggs in an 82- mm Lake St. Louis female. Banded killifish are an important forage species in Presque Isle Bay, but very little is known about their biology in the Great Lakes region. The objectives of this study were to determine the (1) relative abundance, (2) age and growth, (3) reproductive potential, and (4) diet of banded killifish in Presque Isle Bay. Methods Banded killifish were collected during the summer (June, July, August), and fall (September, October) of 2003 and 2004 by manually pulling a 1.2- x 9.1-m (4- x 30-ft) beach seine in several shallow water areas of Presque Isle Bay, Lake Erie. Four locations were sampled: Site 1, parking lot 2 on Presque Isle State Park (N42o07.367', W080o08.843'); Site 2, Niagara boat launch (N42o08.728', W080o07.844'); Site 3, entrance to Marina Bay (N42o09.164', W080o06.770'); and Site 4, Thompson Bay (N42o09.809', W080o04.753'). A total of three seine hauls was taken at each location during each sampling period. Fish collected in each seine haul were sorted as either banded killifish or other species. The banded killifish were anesthetized in 0.1% tricaine methanesulfonate (MS-222) (Sigma Chemicals, St. Louis, MO) and then fixed in 10% formalin. Other fish species were identified in the field, if possible, counted, and released. Identifying other fish to species was not critical to this study because the main purpose of counting other species was to determine the relative abundance of banded killifish in Presque Isle Bay. Banded killifish were returned to the laboratory, measured total length to the nearest mm, and weighed to the nearest 0.01 g. Age and growth data analysis was separated between summer and fall data because summer fish would be even year ages beginning at one year old, and fall fish would begin 2007 E.C. Phillips, Y. Ewert, and P.A. Speares 271 at 0+ years old. The length–weight relationship for banded killifish was calculated using the equation: ln W = a´ + b (ln L). Scales were removed from the area at the tip of the pectoral fin and above the lateral line of each banded killifish and mounted between two microscope slides. Annuli were counted to determine age. Fecundity was determined by first removing the ovaries from gravid females. Ovaries were then weighed so that the gonadosomatic index (GSI, ovary weight/total body weight x 100) could be calculated for each female (Barron and Albin 2004, Shein et al. 2004). Ovaries were then classified into one of five stages based on the criteria used by Heins and Rabito (1986): latent (LA), early maturing (EM), late maturing (LM), mature (MA), and ripe (RE). Specimens classified as MA and RE were considered to be reproductive (Heins and Baker 1989, Heins and Rabito 1986). All females collected on July 9, 2003 and July 18, 2004 were either MA or RE, and mean GSI calculations for these dates were at their highest; therefore, females collected on those dates were used in fecundity analysis because these were the two sampling periods nearest the spawning peak. After ovary classification and GSI calculation, eggs were removed from each ovary and sorted into one of two categories: eggs with yolk or eggs without yolk. These two general categories of eggs used in the analysis were condensed from four more specific types of eggs observed in all ovaries. The four egg types observed were: (1) small white opaque, (2) small yellow (slightly larger than small white and containing some yolk), (3) large opaque yellow (eggs were as large as the next category but were more round with regular edges and were not as translucent), and (4) large golden-yellow translucent with irregular edges. Eggs in categories 1 and 2 were considered to be eggs without yolk because they didn’t appear to be eggs that would be immediately ovulated. Eggs in categories 3 and 4 were considered to be eggs with yolk because they appeared to be eggs that would be ovulated in the near future. These two categories were included in the category of eggs with yolk because, according to the findings of Heins et al. (1992), they could easily be ovulated to produce the currently available clutch of eggs. The relationship between fish length and number of eggs was analyzed using simple linear regression: ln number of eggs = a´ + b (ln L). Simple linear regression analysis was done for total eggs, eggs with yolk, and eggs without yolk. Diets were assessed by first removing the stomachs of preserved fish. Organisms contained in stomachs were identified and counted. Food items of each taxonomic group were then measured by alcohol displacement in either a 5-ml graduated cylinder or a 2-ml pipet to determine the volume of each type of food item. Relative numbers (%N), relative volume (%V), and the relative frequency (%F) (number of fish containing that food item) of 272 Northeastern Naturalist Vol. 14, No. 2 food items in the stomach of each fish were calculated. From these data, the index of relative importance (IRI) (Pinkas et al. 1971) for each food item was calculated for each age class of fish using the formula IRI = (%N + %V) x %F. Values of IRI can range from 0 to 20,000, with higher values representing food items of greater importance. Fish were divided by age to determine if diet changed with fish age. Age classes for fish collected during the fall were age 0+, age 1+, and age 2+, and age classes for fish collected during the summer were age 1, age 2, and age 3. Fish with empty stomachs were excluded from the analysis. Results Banded killifish had the highest relative abundance of fish in the study, making up 86% of fish captured at all sites combined. The mean relative abundance of banded killifish at site 1 was 86%, at site 2 was 92%, at site 3 was 93%, and at site 4 was 74%. Other species collected were Neogobius melanostomus Pallas (round gobies), largemouth bass, smallmouth bass, Labidesthes sicculus Cope (brook silversides), Notropis atherinoides Rafinesque (emerald shiners), and various other minnow species. Based on GSI and ovary classification, the peak spawning date for 2003 was July 9 and the peak spawning date for 2004 was July 18 (Table 1). The mean GSI for fish collected on July 9, 2003 and July 18, 2004 combined was15.4%, and the range was 6.7 to 22.1%. On the July 9, 2003 sampling date, the mean GSI was 14.9%, and the range was 6.7 to 20.8% and, on the July 18, 2004 sampling date, the mean GSI was 15.8%, and the range was 8.3 to 22.1%. The majority of spawning during 2004 occurred over a 20-day period between July 9 and July 29 (Table 1). GSI was already beginning to increase on the first sampling date of 2004 (June 3). GSI decreased dramatically from the July 29 to the August 8 sampling date, indicating that the majority of spawning was over. Age and growth data were separated for summer and fall periods. During the summer, 243 (117 female, 126 male) banded killifish were weighed, Table 1. Changes in gonadosomatic index by date for adult female banded killifish from Presque Isle Bay in 2004. Date Mean GSI GSI range June 3 3.7 1.8–6.3 June 19 8.8 4.2–16.7 July 9 11.3 5.1–19.0 July 18 15.8 8.3–22.1 July 29 12.7 4.9–21.3 August 8 4.2 2.2–12.5 August 15 0.9 0.3–4.4 September 8 1.1 0.2–1.6 September 22 1.0 0.4–1.7 October 6 1.5 0.7–2.1 2007 E.C. Phillips, Y. Ewert, and P.A. Speares 273 measured, and aged. Fish used in the analysis were collected on the July 9, 2003 and July 18, 2004 peak spawning dates. Total length for males and females combined ranged from 59 to 96 mm, and weight ranged from 1.99 to 8.20 g. Based on scale annuli, three age groups were collected: 1, 2, and 3 years old. Of the 243 fish that were aged, 139 (57.2%) were age 1, 95 (39.1%) were age-2, and nine (3.7%) were age-3. All nine age-3 fish were females. Fish in the other size groups were comprised of both males and females. Length and weight data were also analyzed separately for males and females. Female total length ranged from 64 to 96 mm and weight ranged from 2.96 to 8.20 g. The length–weight relationship (r2 = 0.895, P < 0.001) we observed is described by the following equation: lnW = -11.638 + 3.040 (lnL) (where L = total length in mm). Male total length ranged from 59 to 83 mm, and weight ranged from 1.99 to 5.68 g. The length–weight relationship (r2 = 0.838, P < 0.001) we observed is described by the following equation: lnW = -10.089 + 2.671 (lnL). Summer age groups had some length overlap, but the mean length (± SE) of each group was as follows: age 1, 67 mm (± 1.8); age 2, 79 mm (± 2.4); and age 3, 93 mm (± 4.1). During the fall, 132 banded killifish were weighed, measured, and aged. Total length ranged from 22 to 69 mm, and weight ranged from 0.14 to 3.62 g. Based on scale annuli, three age groups were collected: 0+, 1+, and 2+ years old. Of the 132 fish that were aged, 47 (35.6%) were age 0+, 63 (47.7%) were age 1+, and 22 (16.7%) were age 2+. No age-3+ fish were present during the fall, indicating that no age-3 fish survived spawning. Fall age groups had some length overlap, but the mean length (± SE) of each group was: 0+, 32 mm (± 1.9); age 1+, 52 mm (± 2.0); and age 2+, 63 mm (± 3.2). The length–weight relationship (r2 = 0.980, P < 0.0001) we observed is described by the following equation: lnW = -11.032 + 2.919 (lnL). Females collected on the July 9, 2003 and July 18, 2004 peak spawning dates were used to analyze fecundity. The 117 females used in age and growth analysis were used to determine fecundity. Female banded killifish became mature at 1 year old. The mean length of gravid females was 78 mm (range 64–96 mm). The mean number of total eggs per female was 694 ± 39. The relationship (r2 = 0.74, P < 0.001) between length and total number of eggs per female is explained by the following equation: ln number of eggs = -6.00 + 2.85 (lnL). The mean number of eggs with yolk per female was 201 ± 16. The relationship (r2 = 0.48, P < 0.001) between length and number of eggs with yolk per female is explained by the following equation: ln number of eggs with yolk = -8.470 + 3.15 (lnL). The mean number of eggs without yolk per female was 496 ± 32. The relationship (r2 = 0.69, P < 0.001) between length and number of eggs without yolk per female is explained by the following equation: ln number of without yolk eggs = -6.28 + 2.94 (lnL). The mean ovary weight per female was 0.565 ± 0.046. The relationship (r2 = 0.55, P < 0.001) between 274 Northeastern Naturalist Vol. 14, No. 2 length and ovary weight per female is explained by the equation: ln ovary weight = -14.53 + 3.41 (lnL). Mean egg numbers per female also differed as a factor of age. Total number of eggs per female was 526 ± 37 at age one, 744 ± 38 at age two, and 1062 ± 43 at age three. Total number of eggs with yolk per female was 166 ± 17 at age one, 215 ± 20 at age two, and 337 ± 23 at age three. Total number of eggs without yolk per female was 360 ± 27 at age one, 528 ± 24 at age two, and 795 ± 39 at age three. In age 1 females, 31.6% of total eggs were eggs with yolk. In age-2 females, 28.9% of total eggs were eggs with yolk, and in age-3 females, 31.7% of total eggs were eggs with yolk. Banded killifish diets were analyzed based on three age classes and were separated into summer diets (N = 243 fish, ages 1, 2, and 3) and fall diets (N = 132 fish, ages 0+, 1+, 2+). Fish of all age classes fed almost exclusively on cladocerans and benthic macroinvertebrates (Tables 2 and 3). Diets were very similar between summer and fall, but changed in composition with age. Age-0+ and age-1 fish relied very heavily on cladocerans (IRI 18,346 for 0+ and 14,208 for age 1). Those fish also fed on chironomids and hydroptilids, but chironomids were a more important part of the diet of age-1 fish than they were of age 0+ fish (Tables 2 and 3). Age-1+ and age-2 fish had a much broader diet breadth than the previous group and fed on seven different prey items. Cladocerans were still the most important prey item (IRI 8363 for age 1+ and 9080 for age– 2), but chironomids became more important in the diets of this group of fish (Tables 2 and 3). Age-2+ and age-3 fish had the same diet breadth as the previous group, feeding on the same prey items, but the proportion of prey items changed with these age groups. This age group relied more heavily on chironomids and amphipods, and overall fed on fewer and Table 2. Diets of banded killifish collected from Presque Isle Bay during summer 2003 and 2004. Diet is shown for three age classes. %F = the percentage of fish containing that food item, %N = % composition by number, %V = % composition by volume, IRI = index of relative importance = (%N + %V) x %F. Age 1 (N = 139) Age 2 (N = 95) Age 3 (N = 9) Food Item %F %N%V IRI %F %N %V IRI %F %N %V IRI Crustacea Cladocera 96 93.0 55.0 14,208 94 81.0 15.6 9080 58 51.6 8.6 3492 Amphipoda 14 0.8 4.8 78 42 13.0 47.1 2524 Diptera Chironomidae 78 6.2 37.2 3385 89 12.1 57.0 6150 74 6.0 14.0 1480 Adult Diptera 7 1.7 5.1 48 9 5.0 12.1 154 Trichoptera Hydroptilidae 9 0.8 7.8 77 9 2.2 8.8 99 21 11.4 7.3 393 Ephemeroptera Baetidae 6 1.7 5.1 41 12 8.7 6.2 179 Hexagenia 6 0.6 3.6 25 7 4.3 4.7 63 2007 E.C. Phillips, Y. Ewert, and P.A. Speares 275 larger prey items (Tables 2 and 3). One difference between summer and fall diets of the two older age groups is the presence of plant material during the fall but not the summer. The plant material that was found in the stomachs during the fall was a seed of an unidentified plant. Discussion Banded killifish had the highest relative abundance of any fish collected during this study, and there are no previous studies for comparison. Because collections during this study were confined to shallow water areas (maximum depth of 1.2 m), results do not reflect their overall abundance in Presque Isle Bay. Three age classes were found in banded killifish during both summer and fall sampling. Maximum ages in Presque Isle Bay were 2+ years during the fall and 3 years during the summer. These findings are consistent with those of Witt (1979), who found 3 age classes during June and July sampling in Green Falls Reservoir, CT. A difference between our findings and those in the Green Falls Reservoir population was the length distribution of the fish. The mean length of fish in Presque Isle Bay was longer for each age class: 67 compared to 37 mm for age 1, 79 compared to 52 mm for age 2, and 93 compared to 59 mm for age 3. Possible explanations for the size differences are genetic differences between the two populations, or differences in food availability or water temperatures between the two locations. Fritz and Garside (1975) found four age classes in banded killifish in Nova Scotia. Fish from that study were also smaller than fish from Presque Isle Bay at ages 1, 2, and 3; however, age- 4 fish were similar in length to age-3 fish from Presque Isle Bay. The presence of age-4 fish in Nova Scotia could be a result of the higher Table 3. Diets of banded killifish collected from Presque Isle Bay during fall 2003 and 2004. Diet is shown for three age classes. %F = the percentage of fish containing that food item, %N = % composition by number, %V = % composition by volume, IRI = index of relative importance = (%N + %V) x %F. Age 0+ (N = 47) Age 1+ (N = 63) Age 2+ (N = 22) Food Item %F %N %V IRI %F %N %V IRI %F %N %V IRI Crustacea Cladocera 98 96.0 91.2 18,346 91 77.2 14.7 8363 42 36.6 4.4 1722 Amphipoda 22 6.6 11.4 396 57 18.4 45.7 3654 Diptera Chironomidae 12 3.2 7.7 131 82 9.1 51.0 4,928 77 24.1 24.2 3719 Trichoptera Hydroptilidae 5 0.8 1.1 10 22 4.1 9.9 308 26 9.4 10.0 504 Ephemeroptera Baetidae 5 0.5 3.7 21 9 3.0 5.1 73 Hexagenia 18 6.2 6.2 223 Plant Material 5 2.5 9.3 59 9 2.3 4.4 60 276 Northeastern Naturalist Vol. 14, No. 2 latitude producing a shorter growing season resulting in slower growth rates but a longer life expectancy. Actual fecundity is difficult to determine from egg counts in sexually mature females because there is no way to determine how many of the eggs contained within the ovary will be ovulated during the spawning season, or how many have already been released. All eggs present in the ovary have the potential to be ovulated, but some could be reabsorbed instead. Our fecundity predictions are for total possible fecundity based on the total number of eggs present in the ovary. Clutch size is based on the number of large eggs with yolk that could possibly be released in a relatively short time. Smaller eggs with yolk would probably be part of the next clutch, but could potentially be reabsorbed. Small eggs without yolk would also most likely be part of a subsequent clutch; however, if any eggs were going to be reabsorbed, these would be the most likely. Based on these assumptions, it appears as though three clutches can be produced during each spawning season. In all three age classes, approximately 30% of eggs contained within the ovary were eggs with yolk: age one, 166 ± 17 eggs with yolk of 526 ± 37 total eggs; age two, 215 ± 20 eggs with yolk of 744 ± 38 total eggs; and age three, 337 ± 23 eggs with yolk of 1062 ± 43 total eggs. Previous studies have shown fecundity to be less than what we observed. Fournier and Magnin (1975) found the maximum number of eggs to be only 426 in a 101-mm female in Lake Renaud, Quebec. Fritz and Garside (1975) found 128 and 88 to be the average number of ova produced in Porters Lake and Kejimkujik Lake, NS. Because no distinction was made in these studies about which eggs were counted, it is assumed that all eggs contained within the ovary were counted. We found the length of the spawning season to be similar to the 3- week spawning season found in Quebec by Fournier and Magnin (1975). Changes in GSI over time documented in that study were also similar to our findings. Surprisingly, peak spawning occurred earlier in the year in Quebec than in Presque Isle Bay. Because the timing of spawning is driven by water temperature, however, annual changes based on yearly weather patterns could easily explain that difference. During the 2003 and 2004 sampling seasons, we had an unusually cold and rainy spring and early summer, which probably caused the spawning season to occur later than usual. By comparison, in 2005 the water temperature reached 23 oC by June 13, and although no data were collected, banded killifish were observed showing spawning behaviors. We found age of sexual maturity to be one year old, which is different from the findings of Fournier and Magnin (1975), who found sexual maturity to occur at two years of age. They also stated that sexual maturity is more likely a function of size than age. The difference in sexual 2007 E.C. Phillips, Y. Ewert, and P.A. Speares 277 maturity findings between the two studies may be a function of the latitude at which they were conducted. Quebec is considerably farther north than Presque Isle Bay, and the growing season for killifish in that area is probably considerably shorter. Another finding that confirms that possibility is that we found 64 mm to be the smallest gravid female compared with 42 mm in Lake Saint-Louis, PQ. Previous studies have all found the diets of banded killifish to be similar to what we observed in Presque Isle Bay (Keast 1978, Smith 1947, Witt 1979). All studies found invertebrates to make up the bulk of the diet. Previous studies have been divided on the presence or absence of plant material in the diets of banded killifish (Keast 1978, Smith 1947, Witt 1979). The only plant material we found were plant seeds in the diets of age 1+ and 2+ fish during the fall. Literature Cited Barron, J.N., and H.T. Albin. 2004. Reproduction in Etheostoma zonale across three breeding seasons in Ohio. American Midland Naturalist 151:1–14. Burt, A., D.L. Kramer, K. Nakatsuru, and C. Spry. 1988. The tempo of reproduction in Hyphessobrycon pulchripinnis. Environmental Biology of Fishes 22:15–27. Cooper, E.L. 1983. Fishes of Pennsylvania and the Northeastern United States. The Pennsylvania State University Press, University Park, PA. Fournier, P., and E. Magnin. 1975. Reproduction du petit barré de l’est Fundulus diaphanus diaphanus (Le Sueur). Naturaliste Canada 102:181–188. Fritz, E.S., and E.T. Garside. 1975. Comparison of age composition, growth, and fecundity between two populations each of Fundulus heteroclitus and F. diaphanus (Pisces: Cyprinodontidae). Canadian Journal of Zoology 53:300–311. Heins, D.C. 1990. Field evidence for multiple clutches in the longnose shiner. Copeia 1990:579–582. Heins, D.C., and J.A. Baker. 1989. Growth, population structure, and reproduction of the percid fish, Percina vigil. Copeia 1989:727–736. Heins, D.C., and F.G. Rabito. 1986. Spawning performance in North American minnows: Direct evidence of multiple clutches in the genus Notropis. Journal of Fish Biology 28:343–357. Heins, D.C., J.A. Baker, and W.P. Dunlap. 1992. Yolk loading in oocytes of darters and its consequences for life-history study. Copeia 1992:404–412. Keast, A. 1978. Trophic and spatial interrelationships in the fish species of an Ontario temperate lake. Environmental Biology of Fishes 3:7–31. Pinkas, L., M.S. Oliphant, and I.L.K. Iverson. 1971. Food habits of albacore, bluefin tuna, and bonito in California waters. California Department of Fish and Game, Sacramento, CA. Fish Bulletin 152. Richardson, L.R. 1939. The spawning behaviour of Fundulus diaphanus (Le Sueur). Copeia 1939:165–167. Selman, K., and R.A. Wallace. 1983. Oogenesis in Fundulus heteroclitus. III. Vitellogenesis. Journal of Experimental Zoology 226:441–457. 278 Northeastern Naturalist Vol. 14, No. 2 Shein, N.L., H. Chuda, T. Arakawa, K. Mizunao, and K. Soyano. 2004. Ovarian development and final oocyte maturation in cultured sevenband grouper Epinephelus septemfasciatus. Fisheries Science 70:360–365. Smith, M.W. 1947. Food of killifish and white perch in relation to supply. Journal of the Fisheries Research Board of Canada 7:22–34. Wallace, R.A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in teleosts. American Zoologist 21:325–343. Witt, A.L. 1979. Life history of banded killifish Fundulus diaphanus in Green Falls Reservoir, Connecticut. M.Sc. Thesis. University of Connecticut, Storrs, CT.