SEX RATIO OF ODONATAAT EMERGENCE

Abstract


Introduction
The sex ratio of Odonata has attracted attention from several authors since Tiimpel (1899) and Tillyard (1905Tillyard ( , 1917)).Most observations reviewed by Corbet (1962) and Lawton (1972) have been broadly consistent with Tillyard's tentative conclusion (1917: p. 326) that the number of males and females is approximately equal and that females are slightly in excess.
Because the behaviour of adult males and females differs greatly, it can be nearly impossible to obtain precise measures of sex ratio during the adult stage.Indeed the only time in the life history when sex ratio can be measured without being subject to bias of unknown extent is emergence, when every individual leaves a cast skin on a support close to the water's edge.It is worth emphasising that, by virtue of their size and often high population density in small bodies of water, Odonata offer uniquely favourable opportunities for measuring sex ratio at emergence: -the emergence period of many species, especially in temperate latitudes, is relatively brief (a few weeks); cast skins (exuviae) normally remain on emergence supports for several days and, especially in Anisoptera, are conspicuous and readily identifiable to species; -exuviae, if intact, can be sexed unequivocally; -numbers emerging during one emergence period from a small or medium-sized water body are often manageable for a person making daily collections of exuviae; and in many water bodies all emergence supports are, or can be rendered, acces-sible, so that counts can be made virtually exhaustive.
Thus it is often possible to measure sex ratio precisely in the whole population of one species emerging from one habitat.So, if due allowance is made for sources of potential bias, collections of exuviae can provide sex-ratio data of unrivalled quality.
Since the reviews cited above, many accounts of sex ratio at emergence have been published, and we now see merit in critically appraising the data available.Accordingly, our aims in this paper are: -to present and review records available to date; -to identify sources of possible bias in field-derived data and to recommend minimum acceptable criteria for ensuring that such data are accurate and comparable; -to suggest ways of analysing and presenting data; -to record the range and variability of sex ratio measurements in different taxa; and -to draw tentative inferences regarding the causes of sex ratio imbalance.

Methods
In this review we include only data that, as far as we have been able to ascertain, satisfy criteria designed to avoid or minimise bias and which accordingly can constitute useful guidelines for field workers.Inspection of published records convinces us that it is necessary to specify all of these criteria.The criteria, which extend and supersede those listed by Corbet (1962: p. 117), follow, accompanied by explanatory notes in square brackets.
Criteria for measuring sex ratio of a given species at emergence by collecting exuviae 1. Species identification is secure, and only one species is included.2. Counts include only exuviae and not teneral adults.3. Collections are all from the same emergence period (within one season).[Where information exists, allowance is made for possible heterogeneity due to the existence within the emergence period of time-segregated moieties derived from cohorts with different voltinism, e.g. the second peak of emergence of Anax imperator which, when it contains an excess of females, reduces the sex ratio of the annual emerging population (Corbet, 1962:p.117).It will not always be possible to allow for this variable, partly because voltinism within an emergence period is not necessarily uniform, and partly because, even if heterogeneity within an emergence period is confirmed, as inA imperator, the two moieties probably show temporal overlap.] 4. Collections are all from the same body of water. 5. Collections are made in a standardised way.[The same effort is devoted on each occasion, preferably to the whole water body, resulting in exhaustive collections, or to a constant part of the water body, resulting in standardised samples.]6.Only exuviae that can be sexed are included in totals.[If possible, the number of exuviae that cannot be sexed is recorded.It is confirmed that such exuviae are not localised in time or place in a way that might distort the record of overall sex ratio.]7. Collections are made regularly and frequently and at about the same time each day.
[Collections are made preferably daily and no less often than weekly.Weather (e.g.strong or gusty wind or heavy rain) likely to dislodge exuviae is noted in case it jeopardises reliability or continuity of the record.] 8. Collections include the whole emergence period in any year.[Collections begin a few days before, and cease a few days after, the emergence period.For bi-or trivoltine species each emergence period is treated separately.Non-seasonal species that lack a restricted emergence period are not eligible for analysis.]9.The size of the collection used to compute the sex ratio exceeds 99 and preferably exceeds 299.10.Collecting procedure is described in the report.11.The whereabouts of the collected material is stated.
We have satisfied ourselves, either from scrutiny of the published account or from correspondence with the author, that each record included in Tables 1-3 meets the criteria listed above closely enough to provide data that are comparable inter se.When scrutinising records we were obliged to omit records, in whole or in part, from 28 sources because they failed to meet the criteria in one or more respects.Abbreviations as listed in Table I.
Before including records from 11 other sources we had to recalculate sex ratios to allow for mathematical error, inclusion of unsexed exuviae or combination of two or more emergence periods.Notwithstanding the approach adopted by Corbet (1962: p. 117) in the first review of sex ratio in Anisoptera, we recommend against inclusion of records based on parts of a year, and on several habitats combined, for non-seasonal tropical species.Although sex ratios obtained by Corbet (1962: table VII) for tropical species, based on collections made on several occasions and at several places on Lakes Albert and Victoria in East Africa (Table 5), conform well with expectation based on values for temperate Anisoptera that meet the stated criteria, we consider it premature to assume that sex ratio in non-seasonal species is independent of time of year or place.We therefore recommend that such records are not treated as comparable with those from sites, be they tropical or temperate, where emergence is seasonally restricted and where the criteria listed above can be applied.In contrast, the results of Mathavan & Pandian (1977) for five species of Libellulidae are acceptable for purposes of the present review, having been obtained where emergence was seasonal (following the north-east monsoon in Tamil Nadu, India) and confined to about eight months each year.The possibility cannot be excluded, however, that one species studied by Mathavan & Pandian (1977), namely Pantala flavescens, had an emergence curve which, being somewhat bimodal, might have reflected the emergence of some individuals which were the progeny of those that emerged earlier in the same emergence season.
Data that meet our criteria have been segregated among Tables 1-3 according to size of collection (the minimum acceptable total being 100) on the assumption that larger collections provide results in which greater confidence can be placed.Because for statistical reasons ratios cannot be averaged, we present data graphically in Figs.1-16 in a form that we believe is likely to reveal central tendency, range and variability in records of this kind.Sex ratios are expressed as percentage males of the total.
For each record where the ratio differs "significantly" from 50% (as indicated by a value of P of 0.05 or less, using conventional chi-squared analysis) we record the value of P, although we do not thereby imply that a ratio of 50% constitutes expectation.Indeed there is now ample justification for assuming that, in many species, the usual sex ratio is not 50%.Table 4. Frequency distributions of sex ratios according to family (Cordulegastridae omitted).This table is based on all records in Tables 1-3, and therefore features multiple records for some species.In Table 5 we include data that, though not complying with our criteria, may later prove to be informative.
Authorities for scientific names of species included in Tables are given on the first occasion of mention there, and otherwise on first mention in the text.
Table 5. Sex ratios of non-seasonal tropical species.Data are from table VII in Corbet (1962); entries for each species combine records made on several occasions and at several place on Lakes Albert and Victoria in East Africa.

Results
Species represented by records that meet our criteria are predominantly those which have large, readily identifiable exuviae and that exist in conveniently large numbers in water bodies of moderate size.Because of this bias, conclusions applying to Odonata in general can only be tentative until further information reveals if, and to what extent, it is appropriate to combine data for different taxa.In the meantime, aggregate distributions appear to show a level of consistency that deserves comment.The data, presented in the Tables and Figures, encompass 50 species of Anisoptera, from 25 genera and 5 families, and 16 species of Zygoptera, from 8 genera and 2 families.

Discussion
Sex determination in Odonata is genetic, not environmental (see Conover & Heins, 1987), and is the subject of a searching study by Kiauta (1969) whose relevant conclusions are as follows.The original mode of sex determination in the order is of the XO/XX type, the male being the heterogamic sex, and exists in all primary complements, regardless of chromosome number, i.e. the degree of specialisation exhibited by the taxa concerned.In secondary complements a neo-XY sex determination occurs, where the original X has been fused with an autosome.Such an occurrence, which is often irreversible and which can be found in some cells (or stages) but not in others of a single individual, seems unrelated to a taxon's inferred phylogeny.If stabilised, the neo-XY condition tends to evolve further, as in Gomphidae, into a secondary XO type.
Pending evidence to the contrary, one must assume that the sex-determining mechanisms documented for Odonata produce a sex ratio of unity in the newly fertilised zygote.It is therefore plausible to hypothesise that any variability and imbalance in the sex ratio at emergence reflect differential mortality during the aquatic stages.The analysis by Lawton (1972) for Zygoptera showed that, in the species he studied, sex ratio does not change significantly during the last five larval instars (i.e.those during which larvae can be sexed by use of external characters).If this situation applies generally, causes of imbalance could reasonably be sought during the egg or early larval stages.Beyond this, one can only speculate, bearing in mind that predation plays a major role in larval mortality (see Johnson, 1991) and that microhabitats occupied by larvae vary widely and correlate closely with larval form and behaviour, especially among Anisoptera; such microhabitats also change during larval ontogeny (see Corbet, 1999).These considerations suggest one arena in which male and female larvae might sustain differential mortality.In this connexion, it is noteworthy that agonistic ("territorial") interactions occur primarily among male, but not female, larvae of Calopteryx splendens (Harris) (Ryazanova, 1988).
Whether or not the causes of sex-ratio imbalance in Odonata can be discovered, it can be stated securely: that, in species included in Tables 1-3, sex ratio at emergence often exhibits an imbalance; that the extent of the imbalance differs among families; and that sex ratio can vary greatly or very little within a species.It can be stated further that, having regard to the widely different mating frequencies between males and females and often intense sexual selection among males (see Corbet,1999), imbalances of the extent reported here are unlikely to have any significant effect on the reproductive potential of the adult population, although the size of the next generation might be reduced when, as happens occasionally (e.g.Gomphus vulgatissimus, Fig. 16), the number of females is exceptionally low.
If future work reveals consistent differences in sex ratio between species or higher taxa, a reasonable approach might be to seek correlations between sex ratio and larval microhabitat, especially among Anisoptera (Corbet, 1999:table A.5. 7), between sex ratio and habitat (i.e.type of water body) and also between sex ratio and voltinism, according to the prediction that the longer the more susceptible sex is exposed to larval mortality factors, the greater is the imbalance in its sex ratio at emergence.It must be said, however, that this prediction is not supported by the greater number of females in the second (univoltine) peak of emergence of Anax imperator in an otherwise semivoltine emerging population (Corbet, 1957).Because of wide variability in ecological conditions, between habitats and years, we consider it unlikely that any detectable correlations will be clear-cut.So the causes of sex-ratio imbalance in Odonata are likely to remain elusive.
Figure 2. Zygoptera: records of sex ratio according to size of collection (log scale).

Figure 3 .
Figure 3. Odonata: sex ratio according to family and frequency, based on entries in Tables1-3(Cordulegastridae omitted).For frequencies (height of bars) see Table4.The dotted line indicates a sex ratio of unity, i.e. 50% males.

Table 1 .
Sex ratios based on totals between 100 and 199.The collecting locality (coli.loc.)

Table 2 .
Sex ratios based on totals between 200 and 299.Conventions as in Table1.
Abbreviations as listed in Table1.

Table 3 .
Sex ratios based on totals exceeding 299.Conventions as in Table1.