Environmental Biology of Fishes 65: 289–310, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.
Reproductive biology of the coral-reef goby, Asterropteryx semipunctata, in Kaneohe Bay, Hawaii Lisa A. Privitera Department of Zoology, University of Hawaii, 2538 The Mall, Honolulu, HI 96822, U.S.A. (e-mail:
[email protected]) Received 10 July 2001
Accepted 30 November 2001
Key words: Gobiidae, tropical marine, spawning season, spawning frequency, fecundity Synopsis A population of the Hawaiian coral-reef goby, Asterropteryx semipunctata, was sampled over a 12-month period to determine basic demographics and reproductive parameters. The sexes did not differ in median length or weight, although the largest males were considerably longer and heavier than the largest females. Overall adult sex ratio was 1 : 1; monthly sex ratios did not differ from unity except in June, when there was a significant female bias. Minimum age at maturity (17.5–19 mm SL) was estimated to be 4.5–5 months after hatching. Nearly all fish over 22 mm SL were mature. Mature females were found in all months of the year, and females that showed evidence of recent or imminent spawning were collected in every month except December. Gonadal analyses indicated a peak in breeding during the summer (May–July) and minimal spawning during the winter (January–February). Between 20% and 30% of females showed evidence of having spawned within the 24-h period prior to collection; therefore, it was estimated that females spawned, on average, at least once every five days and perhaps as frequently as every three days. Mean batch fecundity was 708 eggs (±418), and was not well predicted by standard length, body weight, or somatic condition. Relative batch fecundity (mean = 1.44 oocytes mg−1 somatic wet weight) varied seasonally, with higher values in spring and summer (April–July) than in fall and winter (September–January). Reproductive parameters are compared and contrasted with those of other gobiid fishes to elucidate general differences between temperate and tropical species. Introduction Although the reproductive biology of temperate marine gobies has been well studied, little is known about reproduction in tropical and subtropical marine gobiids. This paucity of information is surprising, given that the vast majority of the nearly 2000 species in the family Gobiidae (Nelson 1994) inhabit tropical and subtropical coastal regions. On coral reefs in particular, the diversity and relative abundance of gobies are usually unsurpassed. In Hawaii, information on reproductive biology is available for two of the five amphidromous freshwater gobies (Kinzie 1993, Ha & Kinzie 1996, Way et al. 1998), but few data exist on reproduction for any of the 31 marine gobiid species in Hawaii (Privitera 2001). The coral-reef goby, Asterropteryx semipunctata R¨uppell, 1830, occurs throughout the tropical and
subtropical Indo-Pacific, from Hawaii to the Red Sea (Randall et al. 1990). The typical habitat is algal-coated reef rock and coral rubble of shallow reef flats and lagoons (usually 0.10–0.15 mm in diameter. Females without vitellogenic oocytes were classed as immature. Females with vitellogenic oocytes were provisionally classed as mature, and were further examined to determine the stage of ovarian maturation. The diameters of about 100 oocytes >0.10 mm were measured to the nearest 0.01 mm with an ocular micrometer to determine the size-frequency distribution of vitellogenic oocytes within the ovary. Oocytes were not perfectly spherical, so the maximum diameter was measured. The sizefrequency distribution of oocytes determined whether the ovary was suitable for the measurement of fecundity (see below). The size of the largest oocyte in the sub-sample (Omax ) was used for comparison of ovarian maturation and oocyte development among females. In some cases, the ovaries were also examined histologically (see below). If the size-frequency distribution of oocytes was bimodal, and the advanced group of oocytes was separated distinctly from the smaller oocytes by a gap in size, the potential batch fecundity (number of eggs laid in a single spawning event) could be determined by counting all of the oocytes in the advanced group for the entire ovary. The ovaries from 71 females met this criterion, and batch fecundity was determined for 41 of these. For all bimodal size-frequency distributions, the median size of oocytes in the advanced group (Omed ) was also determined. Relative batch fecundity was calculated as number of oocytes per 1 mg somatic body weight. When only a few (typically 31 mm SL were males, and the largest male was nearly 14% longer and 62% heavier than the largest female (Table 1). As is typical for the Gobiidae, a sexual dimorphism in the urogenital papilla was obvious (Miller 1984). In males, it is long, thin and pointed, with the tip extending to the base of the first or second anal ray. In females it is short, broad, and rounded, with the tip extending no further than half the distance from the base of the papilla to the first anal ray. Fish ≥15 mm SL could be accurately sexed externally by examination of the genital papilla under a dissecting microscope at 10× magnification. The genital papilla was usually undeveloped in specimens 32 mm SL, as well as all fish 17.5 mm SL, possession of yellow spots was unrelated to maturity status (Figure 2), but frequency (%) of spot occurrence decreased significantly with increasing SL (Pearson r = −0.86, p < 0.0001).
Size at maturity Minimum size at maturity was similar for males and females. Based on both GSI (Figure 3a) and Omax values, the smallest mature females were 19 mm SL. Based on visual examination of the testicular lobes and sperm duct glands, as well as RI (Figure 3b), the smallest mature males were 17.5 mm SL. The proportion of mature males and females increased with body size until, at a length of about 23 mm SL, virtually all fish were sexually mature (Figure 4); however, the proportion of mature fish 17 mm, n = 194).
294 (a)
(b)
Figure 3. (a) Gonadosomatic index (GSI, as %) vs. standard length for females (n = 279) and (b) reproductive index (RI, as %) vs. standard length for males (n = 216).
17% of females 19–22 mm SL were sexually mature (n = 35), whereas 61% of females in this size range were mature during the remainder of the year (n = 36). For males 18–21 mm SL, 24% (n = 27) were mature from October through January, whereas 91% (n = 41) were mature during the remainder of the year.
Spawning season Data for female reproductive effort and ovarian development (Figures 5, 6, 7a and 8) indicate that this species probably breeds year-round, although spawning may briefly cease or be very limited during part of
295
Figure 4. Frequency vs. standard length for females (n = 279) and males (n = 211). Solid bars are for mature fish and dotted bars are for immature fish.
296
Figure 5. Proportion of mature-size females with oocytes > 0.39 mm vs. month. Actual values are shown within bars; monthly sample sizes of mature-size females (SL >18 mm) are shown above bars.
Figure 6. Percent occurrence of stages of ovarian maturation versus calendar month for all females (n = 277). Immature: pre-vitellogenic oocytes, Omax ≤ 0.10 mm. Developing virgin or ‘possibly spent’ (see Discussion): Omax from 0.11 to 0.21 mm. Ripening: Omax > 0.30, no hydrated oocytes. Ripe: Omax > 0.39 mm, with hydrated or ovulated oocytes. Spent or ‘probably spent’: containing POFs or Omax from 0.22 to 0.30 mm (see Discussion). Data exclude two females with potentially atretic ovaries. Sample sizes vary from 16 to 33 females per month.
297
(a)
(b)
Figure 7. (a) Gonadosomatic index (GSI, as %) vs. calendar month for mature females (n = 200) and (b) reproductive index (RI = testes and sperm duct glands as % body weight) vs. calendar month for mature males (n = 169).
298
Figure 8. Monthly changes in mean reproductive condition for mature-size females (FRC, n = 241) and mature-size males (MRC, n = 194).
the winter. Females with large (Omax > 0.39), welldeveloped, yolked oocytes were taken from all months (Figure 5), but the proportions of these females were highest during May and June (approximately 60%), and lowest in October–November (18%) and from January through March (range: 9–17% per month). Ripe females (containing hydrated or ovulated ova) were absent from December through February (Figure 6). Females that were spent (containing POFs) or probably spent (with Omax between 0.22 and 0.30 mm, see below) were found in all months except December and May. Thus, females that showed evidence of recent or imminent spawning were collected in every month except December; however, December had a much higher proportion of females with advanced oocytes (Omax > 0.39) than either of the months immediately preceding and following. Some females appeared to cease spawning during the winter. Two mature-size females (22–24 mm SL) collected in December and January had ovaries that appeared to have fully regressed. Each contained two or three large, irregularly-shaped, yolked oocytes >0.32 mm in diameter, but the remainder of oocytes were unyolked and 2.5% were found only in May, July and September. Unlike the GSI and RC values for females, neither male RI nor mean monthly RC (Figure 8) indicated a distinct spawning
299
Figure 9. Gonadosomatic index (GSI) vs. diameter of largest oocyte (Omax ) for 183 females with oocytes > 0.20 mm in diameter. Open circles indicate the 101 females with a unimodal size frequency distribution of vitellogenic oocytes. Solid squares are for the 12 specimens with two slightly overlapping size-frequency distributions of vitellogenic oocytes. Triangles indicate the 70 females with a distinctly bimodal size-frequency distribution of vitellogenic oocytes.
peak in June. In contrast, maximum RI in June was only 1.4%, lower than in any other month from March through October. Oocyte size-frequency distributions For females with Omax > 0.35 mm, both GSI and the development of two discrete size classes of oocytes were well correlated with Omax , but there was considerable individual variability (Figure 9). Oocyte sizefrequency distributions were unimodal in all fish with Omax < 0.36 mm, bimodal in most fish with Omax > 0.39 mm, and bimodal in all fish with Omax > 0.43 mm. In some females with Omax between 0.38 and 0.45 mm, the advanced modal group was not completely separated from the smaller modal group. In these samples, there was a 0.01–0.06 mm long ‘tail’ of heavily yolked opaque oocytes belonging to the advanced modal group that overlapped in size with the more nearly translucent oocytes from the smaller modal group (Figure 10a). All 12 of these samples with overlapping modal groups were taken from October through February. For all females with Omax > 0.45 mm, the advanced modal
group was distinctly separated from the smaller oocytes by a 0.06–0.24 mm wide gap (Figure 10b). For all 71 females with two discrete size classes of oocytes, mean Omax was 0.48 mm (range: 0.38–0.62) and mean Omed was 0.42 mm (range: 0.32–0.51). Mean values were smaller, but within the above ranges, for the 12 females with overlapping modal groups; mean Omax was 0.42 mm (range: 0.39–0.45), and mean estimated Omed was 0.35 (range: 0.32–0.37). When these 83 females were lumped together, Omax declined over the course of the breeding season (Figure 11), but it is not known whether this decline reflects seasonal variation in the size of spawned eggs. The largest unhydrated oocyte was 0.51 mm. Hydrated ova 0.46–0.62 mm in diameter were found in 45 of the 200 mature females examined. Nine of these had a unimodal size-frequency distribution of oocytes. Of these nine, four contained POFs and the largest unhydrated oocytes ranged from 0.21 to 0.32 mm. Unhydrated Omax was larger (0.34–0.39 mm) in the five without POFs. All 36 of the remaining females with hydrated ova had a bimodal size-distribution of oocytes, with Omax > 0.40. In nearly half of these (n = 17), the female had ovulated: the advanced
300
(a)
(b)
Figure 10. Oocyte size-frequency plots from two females representative of (a) incompletely separated and (b) distinctly bimodal oocyte distributions.
301
Figure 11. Diameter of the largest oocyte (Omax, ) vs. calendar month for 83 females with a bimodal size-frequency distribution of vitellogenic oocytes. The regression equation is: y = −0.0004x+12.334 (r2 = 0.41, p < 0.001).
modal group consisted entirely of hydrated eggs with attachment fibers, some of which were in the oviducts and easily extruded by pressure on the abdomen. All of the females with ovulated eggs were taken during a single collection in June, at 13:00 h; over 70% of the 30 mature females in this collection contained advanced modal groups consisting entirely of fully ripened (hydrated or ovulated) eggs. Among ovulated females, the largest unhydrated oocytes were between 0.18 and 0.27 mm, which is similar to the Omax of recently spawned females (Privitera 2001). Postovulatory follicles were found in 17 of the 25 ovaries that were examined histologically. They were present in all 7 females with Omax from 0.22 to 0.30 mm, and in 10 of the remaining females with Omax from 0.31 to 0.34 mm. Fecundity Mean batch fecundity determined from ovarian analyses was 708 ± 418 eggs (range: 134–2210, n = 41). Mean batch fecundity of females with hydrated eggs (n = 20) did not differ significantly from that of females without hydrated eggs (n = 21), therefore all 41 females were included in fecundity analyses (twotailed t-test: t = 0.053, df = 39, p = NS). Clutch
size of aquarium nests (mean = 886 ± 309, n = 18, Privitera 2001) was not significantly different than the mean for batch fecundity (two-tailed t-test: t = 1.62, df = 57, p = NS). Fecundity was not well-predicted by either standard length, somatic body weight, or somatic condition (K), based on either linear or log–log regressions (Table 2); the coefficients of determination (r 2 ) indicated that only about 10% of the variance in fecundity was accounted for by body size, and only 18% of the variance was accounted for by somatic condition. Variation in batch fecundity was best explained by differences in gonad weight. Mean relative batch fecundity was 1.44 oocytes mg−1 somatic wet weight (range: 0.160–4.11), and was not different from the relative clutch size of females that spawned in aquaria (mean = 1.65 eggs mg−1 , Privitera 2001; two-tailed t-test: t = 0.94, df = 56, p = NS). Variation in relative batch fecundity was independent of female length and weight, but there were significant differences between the relative fecundities of ‘early season’ and ‘late season’ samples (Figure 12); females from spring and summer (April–July) had higher relative fecundities than fish from fall and winter (September–January) (Mann–Whitney rank sum: T = 88, n1 = 34, n2 = 7, p < 0.05). Neither absolute batch fecundity nor somatic condition (K = W/SL3 ) varied seasonally.
302 Age determination Fish between 12 and 24 mm SL ranged in age from 107 to 175 days post-settlement. A least-squares regression was fit to the relationship between daily increment Table 2. Linear least squares regressions of size and batch fecundity relationships. Power functions are the antilog forms of linear regressions using logarithms. The regression between standard length (SL) and somatic weight (WS ) includes data from both sexes, whereas the regressions between gonad weight (G) and SL or WS are based only on data from females. Fecundity (F) relationships are based on females with a bimodal sizefrequency distribution of vitellogenic oocytes. Condition factor (K) is defined by the relationship: K = WS /SL3 . Regression
N
r2
WS = 1.20E–02SL3.29 G = 5.01E–12SL8.63 G = 1.17E–06 W2.49 S F = 44.9SL–441 1.43 F = 5.83SL F = 0.61S + 379 F = 28.0 W0.49 S F = 5.45E + 4K–1038 0.74 F = 54.3G F = 13.0G + 304.6
533 276 276 41 41 41 41 41 41 41
0.98 0.66 0.65 0.10 0.06 ns 0.09 0.08 0.18 0.60 0.70
number (y) and SL, y = 6.11SL+34.9 (r2 = 0.93), and was used to predict age at first maturity. The youngest juveniles in the population (7.0 mm SL) were estimated to be approximately 2.5 months old, and minimum age at first maturity (17.5–19 mm SL) was estimated to be 4.5–5 months after hatching. Because growth rates in fishes commonly slow with age after sexual maturity (von Bertalanffy 1957), and because the regression was based on a small sample size, ages for larger individuals could not be reliably estimated from the growth model. Discussion Sexual dimorphism Sexual dimorphism has been reported for most gobies, and may involve body size, body proportions, finray length, pigmentation, dermal denticles, dentition, scent and sound (Miller 1984). Dotu & Mito (1963) reported a sexual dimorphism in the first dorsal fin of A. semipunctata, but did not elaborate further. One might assume that they were referring to length of the third spine filament on the first dorsal fin (D1 III) because it is conspicuously elongated, particularly in large males. The present study demonstrated the sexual
Figure 12. Relative fecundity (eggs mg−1 somatic body weight) vs. calendar month for 41 females with a biomodal size-frequency distribution of vitellogenic oocytes.
303 dimorphism of this character, as well as a significant difference between the sexes in the frequency of yellow pigmentation on the caudal peduncle. The functional significance of these two traits is unknown, but it is possible that they are used for signaling purposes during aggressive encounters or to signify dominance status. R.J.F. Smith (unpublished data) noted that when a male A. semipunctata was substantially larger than its tankmates, it possessed yellow spots in only one case out of 35. My own unpublished data also show that high-ranking, territorial males always lack yellow spots, whereas non-territorial males nearly always possess them. Size composition and sex ratio The size range of fish in my sample is slightly narrower than, but within the size range reported by Dotu & Mito (1963), although the newest recruits might have been small enough to elude capture by hand nets. A. semipunctata is smaller in body size than most of the temperate gobiids that have been studied, but is not especially small for a coral-reef goby (see Table 3). This agrees with the general trend within teleosts of smaller average body size in the tropics than in cooler temperate regions (Ebeling & Hixon 1991). Choat (1982) noted that subtropical wrasses (Labridae) and damselfishes (Pomacentridae) tend to be larger than their tropical relatives. Furthermore, body size may increase with latitude within a single species, as DeMartini & Anderson (1978) found in the hexagrammid Oxylebius pictus. Dotu & Mito (1963) reported that A. semipunctata reaches a maximum size of 59 mm TL in southern Japan. This corresponds to a 7% greater length than in Hawaii; however, it is difficult to ascertain whether this size differential reflects a significant geographic difference, as it may stem from only one (or perhaps a few) unusually large specimens. Differences in average body size and population size structure between the two localities are unknown. With regard to the overall 1 : 1 sex ratio, A. semipunctata is typical of gobiids (Miller 1984). Cole (1990) reported an overall sex ratio of 1.8 females per male (n = 186) for this species from several collections in Kaneohe Bay taken from mid-June to early August 1988. This is consistent with the strong female bias that I found during the month of June 1990. Sex differences in susceptibility to capture during breeding can often account for seasonal biases in sex ratio (Gibson & Ezzi 1978, Miller 1984), but this seems an unlikely
explanation for A. semipunctata. Males guarding eggs under pieces of coral rubble were usually easier to catch than the less site-attached females. If sex differences in capture rates did exist, they would more likely have resulted in a male-biased sex ratio when breeding activity was high. Alternatively, differential survival between the sexes during breeding (Gibson & Ezzi 1981) might better explain the marked reduction in proportion of males during the summer, but this hypothesis remains untested. Spawning season Gonadal evidence for a year-round spawning season is supported by Watson & Leis1 , who found A. semipunctata larvae in the plankton during all months of the year in Kaneohe Bay. Protracted breeding seasons are common for fishes in the tropics, whereas most temperate species spawn only during the spring and summer, when planktonic productivity providing food for fish larvae is greatest (LoweMcConnell 1979, Ebeling & Hixon 1991). Spawning seasons of most gobies appear to follow this general trend of an extended breeding season with decreasing latitude. Published estimates of spawning season durations range from 2 to 8 months (mean = 4.4) for 17 temperate species listed in Table 3; with one exception (see below), spawning periods >6 mo are reported only for species breeding at latitudes below 35◦ N. Spawning seasons of tropical and subtropical gobies tend to be longer than those of temperate species; they range in duration from 5 to 12 months (mean = 8.3) for nine species, including A. semipunctata (Table 3). Only one temperate species, the euryhaline Eucyclogobius newberryi, is known to be reproductively active throughout the entire year (Swenson 1999). Grossman (1979) suggested that euryhaline gobies living in harsh, irregularly fluctuating environments might have protracted spawning seasons to ensure that an abrupt physical or biological disturbance does not eliminate a year class. To avoid potential biases arising from phylogenetic constraints, trends relating to latitudinal variation in spawning season are perhaps best exemplified by intraspecific comparisons. Dahlberg & Conyers (1973) noted that the spawning season of Gobiosoma bosc ranges from a 3-month period in New York waters, to 1 Watson, W. & J.M. Leis. 1974. Ichthyoplankton of Kaneohe Bay, Hawaii: a one-year study of fish eggs and larvae. Sea Grant Tech. Report, UNIHI-SEAGRANT-TR-75-01. 178 pp.
304 Table 3. Summary of spawning data for the family Gobiidae. Given for each are the habitat, maximum size in mm SL (or TL), duration of spawning season in months, range of spawning season, mean fecundity (and range), and method(s) used to determine fecundity (BR = brackish, FW = freshwater, ? = data not available or insufficient sample size). Species
Sp. season range
Fecundity
Method(s)
Source(s)
2
Apr–Jun
2 2 3
Apr–May May–Jun May–Jul
ovarian batch; field clutch ovarian* ovarian* field clutch
1
126–140TL? 230–250TL 75TL
(1054–8978); (1650–6750) (700–2000) (2000–12000) (170–410)
2 3 4, 5
FW
?
3
Apr–Jun
522
lab clutch
6, 7
FW
?
3
May–Jul
?
—
8
BR
195TL?
3
Apr–Jul
(213–3888)
ovarian*
9
BR/FW ?
4
Mar/Apr–Jul
74
4
May–Aug
ovarian*; lab clutch ovarian*
10, 11
marine
(527–863); (80–100) (∼2600–11000)
12, 13
marine
60
5
Mar–Jul
(1000–4000)
ovarian batch
14
marine marine
? 54
6 3–6
603 (293–1300) 500–4000
ovarian* ovarian batch
15 16–18
marine
72
3–6
92
7
2652 (1430–4070); 2674 3274–4788
ovarian batch; lab clutch ovarian batch
19–22
Coyphopterus marine nicholsii BR Lepidogobius lepidus Lythrypnus zebra marine L. dalli marine Eucyclogobius BR newberryi Temperate-subtropical: Gobiosoma bosc BR
Nov–Apr Apr(May)– Jul(Sep) May–Jul/ Jan–Jun Feb–Aug
23
87
7
Sep–Mar
?
—
24
7 8 11–12
Apr–Oct Mar–Oct Jan–Nov/Dec
627 (293–1235) 1058 (745–1593) 607 (362–1010); 407 (100–1000)
ovarian* ovarian* ovarian batch; field clutch
25 25 26, 27
44.5–50
3–7
Jun–Aug; Apr–Sep/Oct
466 (116–1030)
ovarian*
28, 29
41
5
Jan–May
?
—
30–32
71TL
7
Mar–Sep
305 (max 567)
ovarian*
33
340
5
Aug–Dec
9–12
ovarian batch; field clutch lab clutch (ovarian*) —
34
134TL
(55716–343963); (117600–689500) 13800 (184–68300) ?
38
Temperate: Gobius paganellus G. niger G. cobitis Padogobius martensi Odontobutis obscurus Rhinogobius brunneus Neogobius fluviatilis Knipowitschia caucasica Lesueurigobius friesii Pomatoschistus norvegicus P. marmoratus P. microps P. minutus
Habitat
Max. length
marine
112.5–120TL
marine marine FW
Tropical and subtropical: Gobiosoma marine oceanops Microgobius BR gulosus Awaous FW guamensis Lentipes FW concolor Sicyopterus FW extraneus Paragobiodon marine lacunicola P. xanthosoma marine P. modestus** marine P. echinocephalus** marine Eviota lacrimae marine
44 45 52
Sp. season duration
140TL?
12
Oct–Jun/ Jan–Dec Jan–Dec
35–37