Different degrees of lunar synchronization of ovary development between two morphs of a Tanganyikan cichlid fish
The degree of lunar synchronization of spawning is thought to be related to a strategy that decreases predation on the brood in Tanganyikan substrate-brooding cichlids. Here, I examined the periodic change of ovary development in two morphs of Telmatochromis temporalis: the normal morph uses burrows under stones as spawning nests, whereas the dwarf morph uses holes within empty snail shells. The normal morph showed a significant lunar synchronization of ovary development, but the dwarf morph did not. In the normal morph, spawning prior to the full moon probably decreases the incidence of approaching brood predators and increases the guarding efficiency of parents. In the dwarf morph, however, lunar cyclic spawning may be dispensable, because the spawning nests within shells are probably highly effective for predator avoidance. These two morphs are closely related, and then will be a good model to clarify the widely observed phenomenon of lunar cyclic spawning.
KeywordsCichlidae Gonado-somatic index Lunar cyclic spawning Spawning nest Predator avoidance strategy
Although biological rhythmicity that is synchronized with lunar or semi-lunar cycles is a widespread phenomenon in marine organisms (Leatherland et al., 1992; Hernández-León, 2008), it is rare among organisms inhabiting freshwater bodies where tidal fluctuation is limited or absent. However, lunar cyclic spawning has been reported in nine substrate-brooding species of cichlid fish belonging to the tribe Lamprlogini in Lake Tanganyika, Africa (Nakai et al., 1990; Rossiter, 1991). In these species, a pair of parents or a female guards the eggs and yolk-sac larvae, which cannot easily escape from predators by themselves. The spawning cycles of these species are thought to improve the survival of the vulnerable brood (eggs and yolk-sac larvae) or dispersing young (Nakai et al., 1990; Rossiter, 1991). Three possible explanations have been proposed: (1) spawning prior to the full moon reduces the predation on the vulnerable brood by nocturnal predators (bagrid catfish) that are not active during the full moon (Rossiter, 1991), (2) spawning prior to the full moon enhances the effectiveness of nocturnal parental guarding of the vulnerable brood under the maximum lunar illumination during the full moon (Nakai et al., 1990; Rossiter, 1991), and (3) dispersal of young during the fourth quarter of the lunar cycle and the new moon improves the survival of young dispersing under the cover of darkness (Nakai et al., 1990). If any of these explanations is true, the degree of lunar synchronization of spawning will vary according to the strategies for decreasing predation on the vulnerable brood and/or dispersing young. Comparison between more closely related populations is better to test this hypothesis because it will minimize the effect of phylogeny.
The algae-feeding Tanganyikan cichlid, Telmatochromis temporalis Boulenger, is an iteroparous substrate brooder. Like other substrate-brooding species that are known to exhibit lunar synchronized spawning, this fish also belongs to the tribe Lamprologini (Takahashi, 2003). T. temporalis is dimorphic for body size. The two morphs lay eggs and guard the brood in different types of nest (Takahashi, 2004; Takahashi et al., 2009). The normal morph possesses a moderate-sized body for a Tanganyikan rock-dwelling cichlid [88 mm in standard length at maximum size (SLmax) in males, 62 mm SLmax in females], and is one of the most common fish on rocky shorelines. This morph uses burrows under stones as spawning nests and shelters (Mboko & Kohda, 1999; Katoh et al., 2005). The dwarf morph is about half of the normal morph in body size (45 mm SLmax in males, 29 mm SLmax in females) and invariably inhabits shell beds, in which the lake bottom is covered by a high density of empty snail shells of the gastropod Neothauma tanganyicense Smith (Takahashi et al., 2009: fig. 1d). This morph uses the empty shells as spawning nests and shelters. A population genetic study showed that the normal and dwarf morphs from Wonzye (08º43′31″ S; 31º07′55″ E; near Mpulungu, Zambia, at the southern end of the lake) were closely related but isolated from each other, although their geographical distributions partly overlap (Takahashi et al., 2009).
In this study, I observed the periodic changes of ovary development in the normal and dwarf morphs from Wonzye, and examined whether the ovary development was synchronized with the lunar cycle. Based on my findings, I discuss the relationship between the degree of lunar synchronization of ovary development and predator avoidance strategies.
Materials and methods
Sampling of fish
Using SCUBA diving, 6 to 11 females of the normal morph were collected every 3 to 5 days between 27 September and 2 December of 2005 on rocky shorelines at 1.0 to 4.9 m depth (N = 173), and 5 to 12 females of the dwarf morph were collected every 3 or 4 days between 29 September and 1 December of 2005 on shell beds at 9.5 to 10.3 m depth (N = 174). Fish were transported to the laboratory at Mpulungu and killed in a solution of anesthesia FA 100 (Takeda Pharmaceutical Co. Ltd.) within 6 h after collection. The ovaries were extracted from all females. The bodies and ovaries were dabbed on tissue paper to remove excess moisture before weighing, and were weighed to the nearest 1 mg. When the ovary was lighter than the minimum readable weight of the electronic balance (1 mg), the weight was taken to be 0 mg. However, zero values cannot be subjected to log-transformation. To avoid this problem, I added 1 mg to all ovary weights. The gonado-somatic index was calculated as IG = 100 WOWB−1 (%), where WB is the body weight (mg), and WO is the ovary weight + 1 (mg). Note that IG is not a useful descriptor of gonad investment in some organisms (Tomkins & Simmons, 2002). However, this index was used in this study, because the normal and dwarf morphs showed isometric relationships between WB and WO (see “Results”).
Test of lunar synchronization
Body weight and ovary weight
Tomkins & Simons (2002) pointed out that IG was not a useful descriptor of gonad investment in organisms that showed an allometric relationship between the gonad and somatic weights. In this study, however, the isometric function provided a significantly better fit than the allometric function for explaining the relationship between WB and WO in each morph (Fig. 1) (test of difference between allometric and isometric functions: F = 0.483, df = 1 and 158, P = 0.488 in the normal morph; F = 0.034, df = 1 and 172, P = 0.854 in the dwarf morph). This result means that IG is not significantly affected by WB after maturity; therefore, IG was used as the descriptor of gonad investment in the present study. In the normal morph, the variance of log(WO) seemed to increase with log(WB) (Fig. 1), but this tendency was not significant (the correlation coefficient between log(WB) and squared deviates was r = 0.071, P = 0.370).
Lunar synchronization of ovary development
This study revealed that (1) the ovary development of the normal morph was significantly synchronized with the lunar cycle, and (2) the ovary development of the dwarf morph was not significantly synchronized with the lunar cycle (P = 0.096), or at least the cycle was less pronounced compared to the normal morph. The ovary development of the normal morph peaked during the second quarter of the lunar cycle, suggesting high activity of reproduction during this period, in accord with the findings for nine other substrate-brooding species of Tanganyikan cichlids (Nakai et al., 1990; Rossiter, 1991). Three possible explanations have been proposed for the lunar cyclic spawning of the Tanganyikan substrate brooders, and two of these explanations are applicable to the normal morph.
The first applicable explanation is synchronization of the vulnerable stages of the brood (egg and yolk-sac stages) with a period when nocturnal predators are not active (Rossiter, 1991). The bagrid catfish is one of the main predators in the lake (Fryer & Iles, 1972). Small bagrid catfishes (Phyllonemus spp. and young of Chrysichthys spp.) are ubiquitous in rocky shorelines, which are the main habitat of the normal morph. Predation by these catfishes will critically affect the survival of the broods of the normal morph. Young of the normal morph hatch out about 3 days after spawning (Katoh et al., 2005), and therefore the vulnerable stages of the brood occur around the full moon, which accords with a period when bagrid catfishes are not active (McKaye, 1983; Rossiter, 1991).
The second applicable explanation is the synchronization of the vulnerable stages of the brood with a period when the nocturnal parental guarding is effective (Nakai et al., 1990; Rossiter, 1991). The parents of the normal morph attack and repel the brood predators when they approach (Mboko & Kohda, 1999). The ambient light during the full moon may assist the parents in visual detection at night, and in repelling approaching nocturnal brood predators, for example, spiny eels (Ochi et al., 1999).
The other proposed explanation for lunar cyclic spawning, namely, that spawning prior to the full moon improves the survival of young dispersing under the cover of darkness (Nakai et al., 1990), is unlikely in the normal morph. The young of some substrate-brooding species leave the spawning nest immediately after they complete yolk absorption. The period of yolk absorption is about 2 weeks (Kuwamura, 1997), resulting in the dispersal of the young during dark nights during the fourth quarter and new moon (Nakai et al., 1990). However, the young of the normal morph remain in the spawning nest for more than 1 month after yolk absorption (Mboko & Kohda, 1999). The timing of dispersal of the free-swimming young of the normal morph may be decided by some environmental cue.
As discussed above, the lunar cyclic spawning of the normal morph may improve the survival of the brood by synchronizing the vulnerable stages of the brood with the period when the nocturnal brood predators (bagrid catfish) are not active and the nocturnal parental guarding is effective. On the other hand, the dwarf morph did not show clear lunar synchronization of ovary development. The dwarf morph uses empty snail shells as spawning nests in shell beds. The eggs and yolk-sac larvae were always found with a female close to the end of the hole within a shell (N = 12, observed in November of 2005 and October to November of 2007 by the author), suggesting that females spawn and care for the brood there. The end of the hole is very small and is invisible from the outside. The predators would probably have trouble finding and accessing the brood. Spawning nests within empty shells, therefore, will be very effective for preventing predation on the vulnerable brood of the dwarf morph, and therefore lunar cyclic spawning may be dispensable for predator avoidance. Release from the limitation of the spawning timing may reduce the degree of lunar synchronization of the reproduction of the dwarf morph.
Different degrees of lunar synchronization in spawning have been reported among eight species of Tanganyikan substrate-brooding cichlids, and these variations were suggested to be related to the spawning sites (Nakai et al., 1990). The present statistical test of the lunar synchronization using a cosine-curve function supports this suggestion. However, this explanation for the different degrees of lunar synchronization is based on circumstantial evidence, and other explanations remain possible. In fact, females of Lamprologus callipterus and Altolamprologus compressiceps use shells as spawning sites like the dwarf morph of T. temporalis, but show lunar spawning (Nakai et al., 1990). Other factors, such as spawning position within the shells, may also play a role. This study was conducted from the end of dry season to the beginning of rainy season covering two lunar cycles. More analyses of other life-history traits and replicates in other seasons will be needed to reveal the mechanism and the adaptive significance of the lunar synchronization.
Lunar cyclic spawning is a well-documented feature in marine organisms (Leatherland et al., 1992; Hernández-León, 2008). More detailed analyses of the normal and dwarf morphs of T. temporalis will help to clarify the mechanism and evolution of this phenomenon.
I express my sincere thanks to M. Hori for providing facilities for research; to H. Phiri, D. Sinyinza and other staff of the Lake Tanganyika Research Unit in Mpulungu for their full cooperation; to E. Nakajima and D. Sinyinza for comments on the manuscript; to T. Sota, S. Takeyama and K. Ota for advice on research; and F. Tashiro for assistance. This study was carried out in accordance with Guidelines for Animal Experimentation, Kyoto University. This study was supported by Grants-in-Aid for JSPS Fellows (No. 20188), for Special Purposes (No. 18779002) and for Young Scientists (No. 20770065), and Global COE Program (A06), MEXT, Japan.
- Fryer, G. & T. D. Iles, 1972. The Cichlid Fishes of the Great Lakes of Africa: Their Biology and Evolution. Oliver & Boyd, Edinburgh.Google Scholar
- Kuwamura, T., 1997. The evolution of parental care and mating systems among Tanganyikan cichlids. In Kawanabe, H., M. Hori & M. Nagoshi (eds), Fish Communities in Lake Tanganyika. Kyoto University Press, Kyoto: 57–86.Google Scholar
- Leatherland, J. F., K. J. Farbridge & T. Boujard, 1992. Lunar and semi-lunar rhythms in fishes. In Ali, M. A. (ed.), Rhythms in Fishes. Plenum Press, New York: 83–107.Google Scholar
- Ochi, H., Y. Sato & Y. Yanagisawa, 1999. Obligate feeding of cichlid eggs by Caecomastacembelus zebratus in Lake Tanganyika. Journal of Fish Biology 54: 450–459.Google Scholar