with Bombus, these being (I) little queen-worker morphological ... Level ofmean genetic variation (ii) in 20 species ofHymenoptera grouped according to degree.
201
NOTES AND COMMENTS also find no evidence that primitively eusocial species are less genetically variable than are advanced eusocial species. In fact, the trend is actually in the opposite direction, but the best conclusion is that too few data are available for a meaningful comparison. LITERATURE CITED AKRE, R. D. 1982. Social wasps, pp. 1-105. In H. R. Hermann (ed.), Social Insects, v, IV. Academic Press, N.Y. BERKELHAMER, R. C. 1983. Intraspecific genetic variation and haplodiploidy, eusociality, and
polygyny in the Hymenoptera. Evolution 37: 540-545. EVANS, H. E. 1958. The evolution of social life in wasps. Proc. l Oth Int. Congr. EntomoI. 2:449457. EVANS,H.E.,ANDM.J. W.EBERHARD. 1970. The Wasps. Univ. Mich. Press, Ann Arbor. SPRADBERY, J. P. 1973. Wasps. Univ. of Wash. Press, Seattle. WILSON, E. O. 1971. The Insect Societies. Belknap, Cambridge, Mass. Corresponding Editor: P. H. Harvey
Evolution, 39(1).1985, pp. 201-205
DIFFICULTIES WITH THE INTERPRETATION OF PATTERNS OF GENETIC VARIATION IN THE EUSOCIAL HYMENOPTERA ROBIN E. OWEN' Department of Genetics and Human Variation. La Trobe University, Bundoora. Victoria, Australia 3083 Received February 3, 1984. Accepted August 3, 1984 In a recent paper, Berkelhamer (1983) compiled data from the literature on the average genetic variation in 50 species ofHymenoptera (which are haplodiploid) and 51 species of diplodiploid insects. These data were used to test some interesting predictions about the role played by the genetic system (i.e., haplodiploidy vs, diplodiploidy) and the social structure in determining the level of genetic variation. Berkelhamer found the following results: (I) Haplodiploids (Hymenoptera) have a significantly lower level of genetic variation than other insects. Within the Hymenoptera: (2) Polygynous ant species (those which have two or more egg-laying queens coexisting in the same colony) do not have significantly different average heterozygosity values from monogynous species. (3) Eusocial and solitary species do not have significantly different average heterozygosities. (4) Primitively eusocial species show significantly less genetic variation than either advanced eusocial or solitary species. The purpose ofthis paper is to point out that the last conclusion holds only because the four Polistes (social wasp) species are classed by Berkelhamer as advanced eusocial whereas they should be classed as primitively eusociaI. In fact, as I shall show, the significant difference between the primitive and advanced eusocial species arises because Polistes have a particularly high average level of genetic variation
, Present address: Department of Biology, University ofCalgary, Calgary, Alberta T2N IN4, Canada.
while the bumblebees (Bombus spp.) have particularly low levels of genetic variation. I shall also suggest some reasons why this difference between Polistes and Bombus might be expected. Classification o/Polistes.-Wilson (1971 p. 88) listed some important social and demographic characteristics which he used to classify Bombus as primitively eusocial as compared to Apis and the Meliponini. Polistes shares seven out of the ten traits with Bombus, these being (I) little queen-worker morphological differentiation, (2) annual life-cycle, (3) small mature colony size «300 workers), (4) dominance of queen maintained primarily by aggression, (5) no recruitment among workers, (6) lack of chemical alarm communication, and (7) temporal division ofIabor weakly developed. However, Polistes does differ from Bombus in that trophallaxis occurs between adult and larvae, workers construct the eggcells, and larvae are reared in separate cells rather than in groups. At this point it is worth considering Berkelhamer's rationale behind the comparison of genetic variation in primitive and advanced eusocial spcies. A retrospective reason is given in the Discussion after the comparison has been made: ... primitively eusocial species studied are considerably less variable than are either the solitary or advanced eusocial hymenopterans studied. This interesting result is compatible with the idea that eusociality arises in species with relatively inbred populations. Primitively eusocial species are those social species with relatively small nests, little morphological differentiation between queens and workers, and behavioral domination of adult workers
202
NOTES AND COMMENTS
TABLE 1. Level of mean genetic variation (ii) in 20 species of Hymenoptera grouped according to degree of sociality. No. of
popu- Mean
Category
Solitary:
Primitively Eusocial: Advanced Eusocial: Bees: Ants:
Species
H
Ageniaspis fuscicollis Diadegma armillata Itoplectis maculator Pimpla turionellae Triclistus yponomeutae Trieces tricarinatus Tetrastichus evonymellae
0.019 0.048 0.056 0.038 0.020 0.000 0.055
Polistes metricus P. variatus
0.065 0.073
Trigona australis T. carbonaria Rhytidoponera impressa R. enigmatica R. confusa R. chalybaea R. purpurea Iridomyrmex purpureus group.! Arthurton-Red form Morgan-Blue form Morgan-Red form Morgan - Black form
0.000 0.000 0.046 0.032 0.034 0.047 0.005
lalions
no. loci
Reference
25 19 20 29 20 14 20
Menken, Menken, Menken, Menken, Menken, Menken, Menken,
9 8
20 20
Metcalf et al., 1984 Metcalf et al., 1984
5 16 2 2 19 10 2
20 20 22 22 22 22 22
Wagner and Briscoe, 1983 Wagner and Briscoe, 1983 Ward,1980b Ward,1980b Ward, 1980b Ward,1980b Ward,1980b
15 15 15 15
Halliday, Halliday, Halliday, Halliday,
0.034 0.063 0.042 0.010
1982 1982 1982 1982 1982 1982 1982
1981 1981 1981 1981
I The sibling species in the I. purpureus group have not yet been named and are identified by collection locality and color. The heterozygosities were calculated using gene frequency data in the first four rows of Table 2 in Halliday (1981).
by the queen. By contrast, advanced eusocial species tend to have larger, perennial colonies, with a greater degree of morphological and behavioral differentiation among castes (Wilson, 1971). Becauseofthis, primitively eusocial species may have population genetic structures more similar to those of species in the early stages of becoming eusocial (Berkelhamer, 1983 p. 543). Thus the reasoning is that the different levels of eusociality reflect different population genetic structures which may have different degrees of genetic variation. Note that Berkelhamer's classification of Polistes is inconsistent with the definition she gives above for primitively eusocial species, and, because of the large number of traits shared with Bombus, there is no justification for including Polistes with the advanced eusocial group. Although Berkelhamer classifies Polistes as advanced eusocial, Vespula vulgaris is classified as primitively eusocial. However, as Wilson (1971 p. 18) points out, the Vespinae are notable for the advanced state of their sociality relative to the Polistinae. Indeed the vespines have marked queenworker size dimorphism, colonies with sometimes thousands of workers, and advanced forms of chemical and auditory communication. It is there-
fore more consistent to place Vespula in the advanced eusocial category. Reanalysis ofthe Data. -Consider first the effect of reclassifying Polistes alone and for the moment keeping V. vulgaris in the primitively eusocial group. From the data on average heterozygosity or genic variation (ii) from Table 2 in Berkelhamer (1983 p. 542) we have the following:
Primitively eusocial Advanced eusocial
Number of Species
fj ± 95% Confidence Limits
13
0.027 ± 0.Ql8
19
0.042 ± 0.012
There is no significant difference in fj (Mann-Whitney U test, one-tailed P> 0.05, two-tailed P> 0.10). Therefore, the significant difference found by Berkelhamer depended on the four Polistes species being classified as advanced eusocial. Since the publication of Berkelhamer's paper, heterozygosity estimates have become available for seven species ofparasitoid Hymenoptera (Menken, 1982), for two species of Trigona, which are ad-
NOTES AND COMMENTS TABLE 2. Summary of mean heterozygosity values for the different categories of Hymenoptera, calculated from the data given in Table I of this paper and Table 2 of Berkelhamer (1983). Number of Category
species
Ii ± SE
Solitary species Eusocial species Primitively eusocial Wasps
25 43 14
0.038 ± 0.005 0.036 ± 0.004 0.035 ± 0.008
6 8 7 29 I 3 25
0.064 ± 0.006 0.013 ± 0.007 0.0 IS ± 0.007 0.036 ± 0.005 0.000 0.005 ± 0.003 0.043 ± 0.005
Polistes
Bees Bombus
Advanced eusocial Wasps Bees Ants
vanced eusocial Meliponid bees (Wagner and Briscoe, 1983) and for two more species of Polistes (Metcalfet al., 1984). Also Rhytidoponera impressa has been split into five separate species (Ward, 1980a. 1980b) and Halliday (1981) has shown that Iridomyrmex purpureus consists of four reproductively isolated sibling species. These additions and changes to the data listed in Berkelhamer's Table 2 are given in Table I. The mean heterozygosity values for the different categories of Hymenoptera recalculated using the revised and expanded data set are given in Table 2. The primitively eusocial species do not have a significantly different level of genetic variation compared with either the advanced eusocial species (Mann- Whitney U test, one-tailed P > 0.10, twotailed P > 0.20) or the solitary Hymenoptera (MannWhitney U test, one-tailed P > 0.10, two-tailed P > 0.20). Therefore, there is no evidence that the degree of sociality (primitive vs. advanced) per se affects the mean level of genetic variation. However the situation is, in fact, rather complicated because the population genetic structure may vary between species which are all at the same level of sociality. Berkelhamer suggests that this may be the case in ants; polygynous species should have larger effective population sizes than monogynous ones. Although Polistes and Bombus are very similar for most demographic and social traits and are quite distinct from the advanced eusocial species, there are a number of reasons why they may differ in population genetic structure and hence genetic variation. Differences between Polistes and Bombus.-Inspection of Table 2 shows that the six Polistes species have a rather high If while the seven Bombus species have a much lower If. This difference is significant (Mann-Whitney U test, one-tailed P = 0.01, two-
203
tailed P = 0.02). Note that this conclusion can only be regarded as tentative, because the data are from only a small number ofspecies so the absolute magnitude of the means may not be representative of each group and the difference may disappear when more species are examined. Nevertheless supposing for the moment that the differences are real, then is this expected on the basis of population genetics theory? The estimates of genetic variation in the Hymenoptera are based on the analysis of electrophoretically detectable allozymes. The neutralist hypothesis predicts that many if not most allozymes are selectively neutral. If this is correct, then one explanation for the high average heterozygosity in Polistes compared to Bombus is that effective population size N, in Polistes is larger, giving a reduced chance ofloss ofalleles due to drift. Polistes colonies commonly have more than one queen, and Lester and Selander (1981) found that P. exclamans, P. apachus, and P. bellicosus had on average 2.3 functional queens per colony. Bumblebees are always monogynous. Therefore, assuming that the number of colonies which constitutes a population is approximately equal in the two taxa, Polistes species would have 2-3 times as many reproductive females, thus a larger N e- This may not be a valid assumption, but in the absence of definite information it seems reasonable given that both Polistes and Bombus are common insects of comparable abundance, at least in the northern temperate regions. Another mechanism which increases effective population size is multiple mating (Wilson, 1963; Hamilton, 1972; Cobbs, 1977; Griffiths et aI., 1982). In Polistes most queens mate more than once (Page and Metcalf, 1982), usually two to three times. Single mating seems to be the rule in Bombus (Roseler, 1973), though multiple copulations are sometimes observed (Hobbs, 1967). Effective population size in haplodiploids is very sensitive to changes in sex ratio (Crozier, 1976). Bumblebee sex ratios are male-biased (Free and Butler, 1959; Owen et aI., 1980; Owen and Plowright, 1982), but Polistes have sex ratios near I: I (Noonan, 1978; Metcalf, 1980). How the sex-ratio affects N, depends on whether or not the number of reproductive females (queens) is the same in the two taxa. If, in bumblebees, the male biased sexratio is due to an increase in the number of males with no corresponding decrease in the absolute number of queens compared to Polistes, then N, will be larger in Bombus. This approaches, in the limit, a maximum increase of 1.5 times (I thank C. Moran for pointing this out). On the other hand, if the total number of males and queens were fixed in the two taxa, then a male biased sex-ratio would result in a decrease in the number of queens and a relative decrease in N, in Bombus. Clearly on the basis of differences in sex-ratio alone it cannot be assumed that Polistes will necessarily have larger effective population sizes than Bombus. The conclusion is that N, in Polistes could be larger than in Bombus if polygyny and multiple
204
NOTES AND COMMENTS
mating by queens are sufficiently common. However, if this is not the case and bumblebees do have the same or greater numbers of reproductive females per population, then explanations other than differences in N, will be required to account for the difference in genetic variation. Explanations based on selectionist arguments are also not clear-cut. One possibility concerns the production of males by workers which is quite frequent in bumblebees (Owen and Plowright, 1982) but is relatively rare in Polistes (Metcalfand Whitt, 1977; Metcalf, 1980; Lester and Selander, 1981). Workerproduced males affect the probability ofa balanced polymorphism and the average genic variation at polymorphic loci (Owen, 1980). Consider a single locus with two alleles, and let the female genotypes A,A" A,A" A,A, have fitnesses Wll, wJ2 ' W22 respectively and the male genotypes A" A, fitnesses V J and V2 respectively. Then it can be shown (Owen, unpubl.) that the sufficient conditions for there to be at least one stable equilibrium are WllVJ
-
'h~vlwll
>
- wo) < 'hw J2(v J
W 22V2 -
'h~VJ(W22 -
+ v2 ) wo)
where ~ is the proportion (0 ::s ~ ::s I) ofthe males produced by the workers (Owen, 1980). Clearly with overdominance (w J2 > Wll, W22) or dominance (w ll = WJ2 > W 22) in the females, the conditions become more stringent when there are worker-produced males and become increasingly so as the value of ~ increases (i.e., the probability ofa polymorphism occurring is reduced). The mean genic variation H, is also reduced (Owen, unpubl.). With codominance, the situation is different. Using the parameterization W ll = I, WJ2 = I - t/2, W 22 = I - t, VJ = I - sand V2 = I (Lester, 1975; Pamilo, 1979), Owen (unpubl.) found that when sand t were allowed to vary between 0 and 1.0 (i.e., large range of selection coefficients), the probability of a polymorphism and fl increased with increasing values of~. However, when selection differentials were 00.125 for ~ ::s 0.5, the probability of a polymorphism and fl were slightly increased, and, for ~ > 0.5, both were decreased. At selection differentials of 0-0.01, both were again decreased for all values of~. Therefore, except for codominant loci maintained by large selective differences, the occurrence of worker-produced males will lead to a decrease in the average genetic variation at loci maintained by balancing selection. Conclusions (I) Species in the social wasp genus Polistes should be classified as primitively eusocial, not as advanced eusocial as done by Berkelhamer (1983). (2) Primitively eusocial Hymenoptera species do not have significantly different levels ofgenetic variation than advanced eusocial species or solitary Hymenoptera. (3) Six Polistes species have a significantly higher level of genetic variation than seven Bombus species. (4) Polistes and Bombus can be considered to be
on the same level of sociality. However, in general, Polistes species have more queens per colony, more matings by queens, and a smaller proportion of males produced by workers than Bombus species. These differences may cause a sufficient difference in population genetic structure to account for the different levels of genetic variation observed. (5) Genetic variation has been measured in only 14 primitively eusocial species, so any results from comparisons involving these should be interpreted with caution. ACKNOWLEDGMENTS I thank Drs. C. Moran, P. Pamilo, and P. S. Ward for their comments and criticisms of this paper, which was written while on a Postdoctoral Fellowship awarded by the Natural Sciences and Engineering Research Council of Canada. LITERATURE CITED BERKELHAMER, R. C. 1983. Intraspecific genetic variation and haplodiploidy, eusociality and polygyny in the Hymenoptera. Evolution 37: 540-545. COBBS, G. 1977. Multiple insemination and male sexual selection in natural populations of Drosophila pseudoobscura. Amer. Natur. 111:641656. CROZIER, R. H. 1976. Counter-intuitive property of effective population size. Nature 262:384. FREE,J.B.,ANDC.G.BuTLER. 1959. Bumblebees. Collins, London. GRIFFITHS, R. c., S. W. MCKECHNIE, AND J. A. MCKENZIE. 1982. Multiple mating and sperm displacement in a natural population of Drosophila melanogaster. Theoret. Appl. Genet. 62: 89-96. HALLIDAY, R. B. 1981. Heterozygosity and genetic distance in sibling species of meat ants (lridomyrmex purpureus group). Evolution 35: 234-242. HAMILTON, W. D. 1972. Altruism and related phenomena, mainly in social insects. Ann. Rev. Ecol. Syst. 3:193-232. HOBBS, S. A. 1967. Ecology of species of Bombus (Hymenoptera: Apidae) in southern Alberta VI. Subgenus Pyrombombus. Can. Entomol. 99: 1271-1292. LESTER, L. J. 1975. Population genetics of the Hymenoptera. Ph.D. Diss. Univ. Texas, Austin. LESTER, L. J., AND R. K. SELANDER. 1981. Genetic relatedness and the social organization of Polistes. Amer. Natur. 117:147-166. MENKEN, S. B. J. 1982. Enzymatic characterization ofnine endoparasite species ofsmall ermine moths (Yponomeutidae). Experientia 38:14611462. METCALF, R. A. 1980. Sex-ratios, parent-offspring conflict and local competition for mates in the social wasps Polistes metricus and Polistes variatus. Amer. Natur. 116:642-654. METCALF, R. A., AND G. S. WHITT. 1977. Intranest relatedness in the social wasp Polistes me-
205
NOTES AND COMMENTS
tricus:a genetic analysis. Behav, Ecol. Sociobiol. 2:339-35 I. METCALF, R. A., J. C. MARLIN, AND G. S. WHITT. 1984. Genetics ofspeciation within the Polistes fuscatus species complex. J. Hered. 75: 117-120. NOONAN, K. M. 1978. Sex ratio of parental investment in colonies ofthe social wasp Polistes fuscatus. Science 199:1354-1356. OWEN, R. E. 1980. Population genetics of social Hymenoptera with worker produced males. Heredity 45:31-46. OWEN, R. E., AND R. C. PLOWRIGHT. 1982. Worker-queen conflict and male parentage in bumble bees. Behav. Ecol. Sociobiol. 11:91-99. OWEN, R. E., F. H. ROOD, AND R. C. PLOWRIGHT. 1980. Sex ratios in bumble bee colonies: complications due to orphaning? Behav, Ecol. SociobioI. 7:287-291. PAGE, R. E., AND R. A. METCALF. 1982. Multiple mating, sperm utilization, and social evolution. Amer. Natur. 119:263-28 I. PAMILO, P. 1979. Genic variation at sex-linked loci: Quantification of regular selection model. Hereditas 91:129-133. ROSELER, P. F. 1973. Die Anzahl der Spermien in Receptaculum Seminis von Hummelkonigin-
nen (Hym., Apidae, Bombinae). Apidologie 4: 267-274. WAGNER, A. E., AND D. A. BRISCOE. 1983. An absence ofenzyme variability within two species of Trigona (Hymenoptera). Heredity 50:97-103. WARD, P. S. 1980a. A systematic revision of the Rhytidoponera impressa group (Hymenoptera: Formicidae) in Australia and New Guinea. Aust. J. Zool. 28:475-498. - - - . 1980b. Genetic variation and population differentiation in the Rhytidoponera impressa group, a species complex of'ponerine ants (Hymenoptera: Formicidae). Evolution 34:10601076. WILSON, E. O. 1963. Social modifications related to rareness in ant species. Evolution 17:249253. - - . 1971. The Insect Societies. Harvard Univ. Press, Cambridge, MA. Corresponding Editor: R. H. Crozier Editor's note: Dr. Berkelhamer has informed me that she has read with interest the papers ofGraur, of Reeve et aI., and of Owen, and that she does not wish to make any formal response.
Evolution, 39( 1), 1985, pp. 205-210
DEMOGRAPHIC GENETICS OF THE GRASS DANTHONIA SPICATA: SUCCESS OF PROGENY FROM CHASMOGAMOUS AND CLEISTOGAMOUS FLOWERS KEITH CLAY' AND JANIS ANTONOVICS Department of Botany, Duke University, Durham, NC 27706 Received September 29,1982.
Darwin (1877) first contrasted vigor of progeny from chasmogamous (CH) and cleistogamous (CL) flowers, and others (Donnelly, 1955, 1979; Wilken, 1982; Cheplick and Quinn, 1982, 1983; Waller, 1984) have subsequently published accounts of this phenomenon. Considerably more data have been published on vigor of progeny from artificially crossand self-pollinated flowers (Griffing and Langridge, 1963; Allard, 1965; Busbice and Wilsie, 1966; Dewey, 1966; Antonovics, 1968; van Wyk, 1981; Schemske, 1983; Schoen, 1983). Diverse results were obtained in both groups of studies, but cross- and self-pollination are not equivalent to chasmogamy and cleistogamy. Differences in fitness between cross- and self-fertilized progeny accrue solely from , Present address: Department of Botany, Louisiana State University, Baton Rouge, LA 70803.
Accepted September 7, 1984
genetic differences between parental plants, but differences between CH and CL progeny may accrue from differences in seed size, nutrient concentration, dispersibility, and germination requirements as well as from the aforementioned genetic differences. Since CH flowers are self-compatible, it is not impossible for a plant with CH flowers to be 100% self-fertilized. There have been no published studies of outcrossing rates of CH flowers in any species, although certain species have mechanisms to promote high degrees of outcrossing (Waller, 1980). As a mode of reproduction, self-fertilization has a number of inherent advantages when compared to cross-fertilization (see Jain, 1976; Charlesworth and Charlesworth, 1979; Lloyd, 1979, for reviews), yet most plant species have reproductive systems other than complete self-fertilization. The most frequently posited "cost" of self-fertilization is in-