The following is an excerpt from Gwynne, D.T. 2001. Katydids and Bush-crickets: Reproductive Behavior and Evolution of the Tettigoniidae.Cornell University Press, Ithaca New York (To be published in Spring  2001)
 

(Apologies: I have not included a list of references cited.  You are not responsible of these references BUT, if you want any of them please email me.   All the figures that you are responsible for are inserted below)

FOR THE SLIDE SHOW THAT I PRESENTED IN LECTURE GO TO THIS LINK

A Mormon cricket female eats her spermatophore meal (spermatophylax)

Chapter 9. Can Katydids Tell Us Why the Sexes are Different?
Which of the two (sexes) takes the initiative here? Have not the parts been reversed?

J. H. Fabre (1917) The Life of the Grasshopper

Why are males masculine, females feminine and occasionally vice-versa?

George C. Williams (1975) Sex and Evolution
 

Now if we might assume that the males . . . have lost some of that ardour which is usual to their sex, so that they no longer search eagerly for the females; or, if we might assume that the females have become much more numerous than the males . . . then it is not improbable that the females would have been led to court the males, instead of being courted by them.

Charles Darwin (1874) The Descent of Man and Selection in Relation to Sex
 
 

Darwin's (1871) comments on the button-quail (Turnix) and a few other birds have turned out to be an apt description of the "reversed" courtship behavior of certain species in quite an array of different animals including frogs, fishes, and flies, and of course katydids such as Mormon crickets (Chapter 1). In these animals (Table 9.1, Fig. 9.1), males are not very eager to search for mates. Instead females often take this "masculine" competitive role while the "feminine" choosy role is assumed by males (Williams 1975). Research with these exceptional species are central to the study of sexual selection because they allow us to address Williams' question about why typical sexual differences, such as sex-specific flashy traits or weaponry, have evolved. Darwin's quote hints at a possible cause of "role-reversal" - females might court their mates if males are in short supply - and theory since Darwin has pointed to sex ratio as one of several of potential influences on the supply of mateable males and females. This chapter illustrates how Mormon crickets and other tettigoniids have been front and center in tests of sexual-differences theory because of the flexible nature of male and female courtship roles (Chapter 1). It is the experimental work on these katydids that is the main focus of this chapter. Before describing experiments, however, let us examine the natural history of role-reversal in the system that has received most study, the Australian pollen katydid, Kawanaphila nartee, a species in which, as we have already seen, the male provides a large spermatophore meal (Table 6.3) that enhances his mate's fecundity (Table 6.2). The details of both natural history and experimental work with pollen katydids are distilled from a series of studies by biologists working at the University of Western Australia, especially Leigh Simmons, Winston Bailey and me.

As Darwin's (1871) original motivation in proposing sexual selection theory was as an explanation of sexual dimorphism, I will close the chapter (and the book) with a discussion of sex-specific weaponry and other sexually-dimorphic devices in katydids. In keeping with the theme of the chapter, this discussion will highlight some remarkable devices that appear to have evolved by sexual selection among competitive females.

Table 9.1. Groups in which courtship-role reversal has been observed (adapted from Gwynne 1991).
 

Taxon	Nature of role-reversal	Male-male competition?	Number of males and females available for mating	What limits female reproduction?
Birds
Phalaropus spp.  (phalarope) (Reynolds 1985)	Fights among females for pre-nesting males	No	Female-biased due to polyandry	Egg incubation exclusively by males
Actitis macularia (spotted sandpiper) (Oring and Lank 1982)	Females compete in areas in which males nest; females court males	Some	Female-biased due to polyandry	Egg incubation exclusively by males
Gallinula chloropus (moorhen) (Petrie 1983)	Fights among females for pre-nesting males	Male fights	Male-biased	Phenotype (quality) of incubating male
Frog
Dendrobates auratus (poison-arrow frog) (Summers 1989)	Occasional fights between females; female takes the active role in courtship	Occasional fights	Possibly female-biased	Male tending of eggs
Fishes
Nerophis ophidion (pipefish)	Male choice (for larger females) and dominance among females (Berglund et al. 1986a; Rosenqvist 1990)	No	Female-biased due to brood-size limitation in males (Berglund et al. 1986b)	Time taken by male to brood eggs on body
Crustacean
Pseudosquilla ciliata (stomatopod) (Hatziolos and Caldwell 1983)	Female initiation of courtship and male choice (for larger females)	Some aggression	Unknown	Possibly male-derived nutrients in ejaculate (known in Squilla)
Arachnids
Zygopachylus albomarginatus (harvestman: Opiliones) (Mora 1990)	Females initiate courtship and engage in aggressive interactions. Males reject certain females.	Males take over nests of rivals	No obvious bias. In fact not all males had nests.	Males tend eggs in mud nests
Empis borealis and Rhamphomyia longicauda (dance flies: Diptera) (Svensson and Petersson 1988; Svensson et al. 1989; Funk and Tallamy 2000)	Females compete in a swarm; males enter the swarm and choose large females	No	Apparently female-biased	Prey-item provided by male (females obtain prey only from males)
Certain populations of Anabrus simplex, the Mormon cricket, Metaballus litus and Kawanaphila nartee (pollen katydids)	In food-limited populations, females fight for access to signalling males; male choice (for larger females) (see text).	Not in these populations	Female-bias caused by food-limitation	Large spermatophore eaten by female which increases fecundity
Belostomatids (giant water bugs: Heteroptera)	Abedus herberti not completely role-reversed: female initiates mating and male coyness (Smith 1979). Diplonychus major females fight to lay eggs (in lab) (Ichikawa 1989)	Male A.herberti display to females (Smith 1979)	Female-biased but only at certain times of the season in D.major (Ichikawa 1989), Belostoma flumineum (Kraus 1989) and A.indentatus (Kruse 1990)	Male aeration of eggs laid on his wing covers

 

A pollen katydid female eats her spermatophore meal

Unlike the neotropical bats listening for the cautious calls of false-katydids (Chapter 8), gleaning long-eared bats (Nyctophilus) of southwestern Australia's jarrah-banksia woodland hear plenty of katydid songs. This is particularly true in early spring when any bat cruising over low vegetation is deluged by a cacophony of ultrasound from dense clusters of the season's first tettigoniid songster (Gwynne et al. 1988), Kawanaphila, literally the "flower lover" but commonly known as a pollen katydid (Zaprochilinae: Chapters 2 and 3). Although the 40 to 60 kHz calls of these insects go undetected by human ears (Gwynne and Bailey 1988; Mason and Bailey 1998), the sounds are pitched right in the frequency ranges used by the local bats to hear their own echolocation clicks (Fullard et al. 1991). Coincidentally, the electronic devices that were originally developed to detect the ultrasonic cries of bats have turned out to be important aids in finding the high-pitched songs of pollen katydids.
 

Our bat-detectors revealed dense populations of one pollen katydid, Kawanaphilanartee (Fig. 9.2a), only in certain locations in the natural bushland of Kings Park, Perth. The park has been our main study site for this small, stick-like flightless species (Chapter 2; Fig. 4.8), by far the most common katydid in the area (Gwynne et al. 1988). In other places within the park, K. nartee appears to be just as common but far fewer males sing. The striking variation in the numbers of callers in different areas of the bushland are reminiscent of the differences between the Greystone and Indian Meadows populations of Mormon crickets described in Chapter 1 except that in pollen katydids this variation occurs over a much smaller spatial scale; in K. nartee huge differences in calling activity can occur over just a few dozen meters (Gwynne et al. 1998).

The most vocal populations of pollen katydids occur in September in places where the katydid's food flowers (Chapter 2) are abundant. Dense choruses of callers can last for the entire K. nartee breeding season near plants with season-long flowering such as Daviesia and Jacksonia bushes (Gwynne et al. 1998). Singing males are also associated with a plant that shows a short spring blooming period, kangaroo paws (Anigozanthos manglesii), the "state-flower" of Western Australia (Simmons and Bailey 1990). Kangaroo paws appear to be a poorer food source for the katydids than flowering bushes, probably because it is bird-pollinated (Hopper 1993) and thus tends to produce less nutritious pollen than flowers visited by pollen-feeding insects. Finally, in areas with few spring flowers of any kind only the occasional K. nartee singer can be heard even though large numbers of katydids may be present in these areas. These populations of katydids are doomed to a life of low rations ⁽¹⁾ except in the rare spot where the pollen famine is dramatically broken by the flowering of a grass tree, Xanthorrhoea preissei (Fig. 9.3), a plant whose floret-crammed stalks can supply abundant food to katydids living close by over the final weeks of the three-month pollen katydid season. A sudden blooming of grass trees occurred during our surveys of pollen katydids in 1993, when two flower stalks pushed up from a large skirt of grass tree leaves in an area with very few other flowers of any kind. We had detected no callers in the immediate area around the stalks but two days after the first florets erupted, and concomitant with an increase in the amount of food in the guts of males (Fig. 9.4), was the nightly appearance of several consistent callers within or close to the grass tree leaves (Gwynne et al. 1998). 

A closer look at the behavior of calling male pollen katydids in the areas with few flowers revealed that, like Greystone Mormon crickets (Chapter 1) and Darwin's button quails, responsive females did appear to be "much more numerous than the males" and that roles were reversed: "females . . . court the males, instead of being courted by them". (Darwin 1874). We saw an excess of females in areas with few flowers where the occasional caller only had to sing for about a minute before attracting one or more females. We confirmed this experimentally by noting the response of females to caged singers placed in the few-flowers site (Fig. 9.5). These confined callers usually attracted a female after a minute of singing and, if they were allowed to continue singing for a full 10 minutes, could end up with a dozen or more highly responsive females scrambling around on the surface of the cage (Fig. 9.5). By contrast, females at sites with lots of flowers were for the most part uninterested in singers; caged callers placed in these sites often failed to attract any females in a 10 minute singing trial. This was true for both the flowering grass tree and flowering bush sites (Gwynne et al. 1998) (for additional details see the legend to Fig. 9.5).

The reversal in courtship was evident from the behavior of callers and responsive females in sites with poor supplies of food (Simmons and Bailey 1990). First, there was clear evidence of female-female competition when rivals jostled with each other in their race to the caller. Females grappled with front legs and pushed or kicked as each attempted to mount the male (Simmons and Bailey 1990) (Fig. 9.6a). When we allowed females to interact with an experimental calling male, he stopped his continuous stridulation (Fig. 9.2a) as soon as his first suitor appeared. However, if he became separated from the struggling females, as was frequently the case, he would almost always sing again. This time his song was not continuous (Fig. 5.4) but consisted (to our ears, at least) of a single brief "tick". Even this brief burst of sound was sufficient for females to re-orient and again hurry toward the male. In nature, separation of the sexes appeared to be a mainly caused by the male withdrawing from the skirmishing females. This and his coy clicking delayed mating by up to 42 minutes (mean = 6.5 minutes) and may represent behavior that incites competition among his potential partners (see Chapter 7) (Gwynne and Bailey 1999). 

Eventually one of the females mounted the male and engaged genitalia by grasping him in a "genital hold" in which she lowered her ovipositor and pinched the tip of his abdomen in the hinge between the base of her abdomen and ovipositor (Simmons and Bailey 1990) (Fig. 9.2b). But holding the male did not necessarily prevent additional competition. Often a female, occasionally even two females (Fig. 9.6b and c), clambered onto a coupled pair. Movements (up and down, side to side or back and forth) of the interfering rival usually caused the original pair to break up (Gwynne and Bailey 1999) and in about half of the interactions the interloping female usurped the mounted position and went on to mate with the male (Simmons and Bailey 1990; Gwynne and Bailey 1999). Females can also fight over pollen and, although heavier females won these food fights (Simmons and Bailey 1990), we could find no body - size advantage in fights over males either before or after the mounting stage of mating (Gwynne and Bailey 1999).

When it came to male choice, however, there was an advantage for larger females. Males favored heavyweights by providing them with a spermatophore (Simmons and Bailey 1990, details in Fig. 9.2 legend). Males rejected smaller females by pulling out from under them in a similar manner to the coy male Mormon crickets at Greystone (Fig. 1.7). Male pollen katydids rejected mounted females in 38% of cases in poor pollen sites (kangaroo paws), pulling away an average 50 minutes after mounting began. Even if the spurned females persisted and re-gained their genital hold on their mates they were never successful in obtaining a spermatophylax meal (Simmons and Bailey 1990). 

Males occasionally spurned females at pollen-rich (grass tree) sites but they did not target smaller females for rejection. Furthermore, most courtship at grass tree sites showed the roles to be typical; most mate rejection was by females with 14 % of females attracted to males turning away before mounting, and, of the females that did mount, 25% did not grasp their mate in a genital hold. Such reticence by females to link genitalia was never seen in pollen poor sites (Simmons and Bailey 1990). Choosy females at pollen-rich sites may be going after large males because in lab trials with well-fed individuals, there was a significant bias toward the larger of three experimental males available to the female (Gwynne and Bailey 1988). The lack of a male size advantage in mating pairs of pollen katydids collected from the field (Chapter 7) may have been because many of the pairs were sampled from kangaroo paws sites where female choice was not expected to occur (Simmons and Bailey 1990).

The other component of typical courtship roles, the direct fights between males as seen in the Mormon crickets in Indian Meadows (Chapter 1), has never been seen in pollen katydids. Simmons and Bailey (1990) observed no fights among singing male pollen katydids in and around the basal leaves of flowering grass trees. However, males did appear to interact acoustically by increasing the rate at which they produced call ticks (Figs. 5.4., 9.2a) and by moving away when they detected substrate vibrations from a rival's call at increasing intensity (Simmons and Bailey 1992) (see Chapter 7). There is even some evidence that aggression between males can escalate to physical encounters in which males mounted neighboring rivals, thereby preventing calling (Gwynne and Bailey 1988; Bailey and Simmons 1991). 

A male Mormon cricket (left) rejects a female who had mounted him (right)

As we have seen for several katydids, sexually competitive males and choosy females appear when food is plentiful. When food was in short supply there is a reversal in these courtship roles. Recall for example that in gregarious Mormon crickets both males and females fought for nutritious food available from the carcasses of dead band-mates (Chapter 1) (Fig. 9.9). With these observations in mind I proposed a food & sexual differences hypothesis (Gwynne 1984c) arguing that when nutritious food is scarce for katydids, the supply of mateable males is low both because of the low rate at which males can turn nutrients into spermatophylax meals and because hungry females become more promiscuous because they forage for these meals. Male katydids might also respond to food stress by reducing the size of each spermatophylax meal but only to a point because with too small a meal, a male runs the risk of compromising his fitness, either in terms of his ability to inseminate or to be a parent of high quality offspring (see Chapter 6).

But how does the food and sexual-difference hypothesis fit into a more general theory of factors controlling sexual differences in animals (Fig. 9.10)? After all, courtship role-reversals in animals such as Darwin's button quails appear to have little to do with food. However, a factor that is common to all role-reversed species (Table 9.1) is a sex-bias in the numbers of sexually-active individuals. As Darwin (1871) may have recognized (see the chapter opening), a relative increase in the number of sexually-available females, the operational sex ratio (OSR) of Emlen (1976; Emlen and Oring 1977), should cause females to become more sexually competitive. When the numbers of sexually active individuals from the sexes are similar, both may exhibit choosiness and mate competition. Finally, a large increase in the number of sexually-available females should lead to role-reversal: when there are more available females than males, males are in a position to choose and females are forced to compete for access to these males. Therefore sexual selection is more intense on females increases and less intense on males (Fig. 9.10). 

But what are the factors that, in turn, determine the numbers of males and females available for mating? One obvious factor is the adult sex ratio (Parker and Simmons 1996). Sex-biased predation, for example (Chapter 8), could change the operational sex ratio. Although there is some evidence that sex ratio biases in local populations occur and can affect courtship roles (for example, a study of beetles by Lawrence 1986), sex ratios appear typically to be so close to unity they have little influence on the availability of the sexes for mating. 

The supply of mateable males and females, however, could be affected by sex-related activities that take one or the other sex away from mating. In most animals this "time out" from mating (Clutton-Brock and Parker 1992; Parker and Simmons 1996) is greater for females because they have to collect and provide resources to offspring. Females expend effort both in producing large gametes, the eggs, (Bateman 1948) and, in some species, the additional energies (and associated risks) needed for gestation and maternal care (Williams 1966; Trivers 1972). Thus, the rate at which females can produce offspring is much lower than that of males, so males compete to sire these offspring (Thornhill 1986; Clutton-Brock and Parker 1992).

To summarize the theory (Fig. 9.10), in most species, the typically greater parental investment (PI), and thus time out from mating, of females relative to males yields a lower relative potential rate at which females can reproduce. This, in turn, leads to a male-bias in the numbers of individuals available for mating and thus to greater sexual selection on males. 

This general theory of sexual differences leads to a key prediction, namely that a courtship role reversal should occur when (1) the male parental contribution is greater than the female's, resulting in: (2) a relatively greater time out from mating activity for males, (3) a relatively lower male reproductive rate, (4) a female-bias in numbers of individuals available for mating and (5) a reversal in the strength of sexual selection on the sexes. However, the precise nature of the role reversal can be influenced by additional factors. For example, although sexual competition among females may arise when sexually-active males are in short supply, active choice of mates by males may be absent if males passively accept winners of competitive bouts among females (Table 7.1). Another factor (ignored in the "operational sex-ratio theory" of role reversal outlined above) influencing the degree of mate choice shown by males is variation in the quality of mates. Variation in the quality of the sex being chosen , in this case females, should influence the degree of choosiness in the opposite sex. The greater the variation in fecundity, or ability to provide parental care, the greater the expected level of choosiness among males (Parker 1983; Owens and Thompson 1994; Johnstone et al. 1996) (Fig. 9.10).

Some of these predictions have been tested empirically, for example by showing that courtship role-reversal (in vertebrate species with exclusive male parental care) is associated with a relative lower male reproductive rates (Clutton-Brock and Vincent 1991). However, work with katydids has provided full experimental tests of the predictions, mainly by manipulation of diet. Let us now turn to how this work addresses specific predictions from the food and sexual differences hypothesis and thus provides a general test of sexual differences theory (see predictions 1 to 8 in Fig. 9.10). 

The key proposal of the food and sexual differences hypothesis (1) is that a decrease in high quality food available should result in a greater number of sexually-active females than males.

Our initial observations of role-reversed Mormon crickets and pollen katydids provided some support because mateable males appeared to be in short supply; recall the extreme scarcity of singing males even though many adult males were present. Furthermore, for pollen katydids, an abundance of ready-to-mate females was revealed when responsive females swarmed onto the cages of our experimental calling males (Gwynne et al. 1998). However, to demonstrate that the apparent female-bias in the operational sex ratio is caused by food-restriction it was necessary to manipulate the availability of food. 

To accomplish this we first set up field enclosures (cages) (Fig. 9.11) into which we released individually paint-marked insects. Pollen katydids were the subjects in one experiment (Gwynne and Simmons 1990) and Mormon crickets, the other (Gwynne 1993). In both studies, four of the cages were food-restricted and four were supplied with plenty of food. I will refer to these treatments as "low" and "high" diets, respectively. Importantly, these methods allowed us to control other variables that might influence sexual selection (Sutherland 1987) by keeping sex ratio, density and ages of individuals similar in all cages (see legend to Fig. 9.11). 

Results with both katydids were similar in showing that a low diet increased the number of sexually responsive females as shown in the decreased time it took for a calling male to attract a female. Also, virtually all of these males mated (Fig. 9.12 shows the data for Mormon crickets). We found, as predicted, the opposite effect on the numbers of sexually-available males: in both species there were significantly fewer callers in low-diet cages than in high-diet cages (Fig. 9.12).
 
 
 

These sorts of field assessments of available males and females may not provide the precise estimates of the numbers of responsive males and females necessary to test our first prediction that more sexually-active males than females leads to role-reversal. For example, if any males acquire mates without singing (see Chapter 7), counts of callers will underestimate the number of sexually-active males. Moreover, we would have underestimated the number of sexually-responsive females if less successful females avoid physical combat by being less responsive to singers. 

Maximum mating rates that each sex can achieve can, however, be obtained by giving individual experimental males and females constant access to receptive members of the opposite sex. These experiments were conducted with Requena verticalis (Gwynne 1990b) and pollen katydids (Simmons 1995c). Both studies showed that the extremes of diet caused large differences in the relative remating rates of the sexes. For Requena a high diet increased the mating rate of males to more than twice that of females (operational sex ratio = 0.45). In contrast, the mating rate of females on the low diet slightly exceeded that of the males (operational sex ratio = 1.05) ⁽²⁾. 

Simmons (1995c) examined relative mating rates in pollen katydids on various levels of pollen in the diet as part of a study of relationships between several factors hypothesized to control sexual differences (more on this work later in the chapter). As predicted, male mating rates of pollen katydids increased with increasing food intake whereas female mating rates decreased (Figs 9.14C, D). Interestingly, as Simmons points out, there is a rapid, rather than a gradual, switch in mating rates when food intake falls below 20-30 joules of energy per day. The rapid switch in mating rates correlates with the apparently rapid switch in courtship roles with decreasing food-availability seen in field populations (Fig. 9.13E) (Simmons and Bailey 1990; Gwynne et al. 1998). 

Another interesting result from Simmons' study is that, as in the Requena experiments, mating rates did reverse but only at an extremely low diet (left of the vertical line depicting equal mating rates in Fig. 9.13D). Simmons (1995c) suggests that a more distinct reversal of mating rates might be prevented by male suppression, probably via ejaculatory chemicals, of the female's refractory period and thus preventing her from achieving her preferred (higher) remating rate. There was also a reversal in relative mating rates in data collected from our field cages (Gwynne and Simmons 1990): male remating rate exceeded female rate for the high-diet cages (where the insects exhibited typical courtship roles, see next section) but rates were reversed in low-diet cages (see points marked on the horizontal axis of Fig. 9.13D). This match of field mating rates with courtship roles indicates that estimates of mating rates in nature may after all be reasonable estimates of maximum mating rates.

Another result of these experiments is that the parental spermatophylax meal of pollen katydids, K. nartee, is not reduced in its energy content when males are food stressed (Simmons 1994b) . Similarly, in Requena experimental food stress did not influence spermatophore size until males had mated three times (Fig. 9.13), although well-fed male Requena decrease spermatophylax size when their remating interval is decreased (Chapter 6 and Simmons 1995b). Maintaining a full-sized mating meal may be adaptive for males in food-stressed environments because their spermatophylax nutrients are likely to be particularly important to the fitness of their offspring. Katydid meals that mainly function to increase fertilization success rather than augmenting offspring fitness (see Chapter 6) might be reduced in size when food is limiting. This may be the case in Gampsocleis gratiosa where males reduce the size of their spermatophylaxes when food-stressed (Zhiyun et al. 1998). The reduction in meal size in this species, however, still did not allow males to reach the high remating rate achieved by males high diet males.

If food stress increases the ratio of sexually-available females to males, does this diet-induced change cause the excess of females to compete for mates and for males to choose? This leads to prediction 2: food stress causes a role reversal in courtship behavior.

This prediction is supported by the behavior observed in our caged populations of pollen katydids (Gwynne 1990b) and Mormon crickets (Gwynne 1993). For low-diet cages we found significantly more reproductive interactions in which males rejected mates while competition occurred between females (Figs. 9.14A and 9.15E). Conversely, the incidence of female rejection of mates was significantly higher in high-diet than in low-diet cages. Moreover, obvious cases of male-male aggression (fights) were seen in the high-diet Mormon crickets (Fig. 9.14) although not in pollen katydids (Fig. 9.15). As suggested above, inter-male competition may be more subtle in pollen katydids. 

FIGURE 9.14A. Food-stress (open boxes) causes a reversal in the courtship roles of Mormon crickets whereas ad-lib food (filled boxes) results in the typical roles of male-male competition and female mate choice. Data are represented in “box-plots” with each box encompassing interquartile ranges. Cross bars indicate medians and the top and bottom bars, the full range of data.

An experimental reduction in food also caused role reversal in laboratory populations of Ephippiger ephippiger, from the tribe Ephippigerini, a group that includes species with very large spermatophylax meals (Chapter 6 and Fig. 6.15). Ritchie et al. (1998) found that a low diet caused males to reject females more often than females rejected males, and also caused a complete switch in the direction of sexual competition in that virtually all contests were between females.

Finally, the effects of food stress on male choice and female competition can occur indirectly through parasitism. For example, heavy infections with gregarine protozoans (see Chapter 4) can starve females of the nutrients necessary to achieve high fecundity. In matings with experimentally infected Requena verticalis, the typical courtship, in which there are more female rejections of males than male rejections of females, reverses; males become the more choosy sex (Simmons 1994a). Possible alternative explanations for increased male choosiness is that males are rejecting sick, and potentially infective females, or the males perceive these females as low-quality mates.

When females are forced to compete for limited food, some will be more successful than others. The result is increased variation in female condition, fecundity and, therefore, their quality as mates. The increased variation in mate quality could increase male choosiness (Owens and Thompson 1994). This raises an alternative hypothesis for the male-choice component of role-reversal seen in our food-deprivation experiments; the increase in male choosiness in low-food cages may have been, at least in part, due to increased variation in the quality of females rather than an increase in the relative number of responsive females. After all, the low diet did increase variation in female fecundity and body mass in our field-cage experiments (Simmons and Bailey 1990; Gwynne 1993). 

Kvarnemo and Simmons (1999) addressed this issue with pollen katydids. First, in an experiment in which each male had encountered a single low, high, or medium quality (fecundity) female, males were, on average, more likely to reject a subsequent female compared to a treatment in which each male encountered a medium quality female. However, only in a second experiment could each male assess variance per se by encountering nightly sets of three females varying in quality. This time the results showed no significant difference between the mate rejection rates of these males and a second treatment in which males encountered trios of females similar in quality. 

More work is needed on the influence of mate quality on male choosiness (see also Kvarnemo and Simmons 1998). This is an important factor influencing male choosineess. After all, there has to be some level of variation in the quality of mates available to males or there would be no benefits to male choosiness in the first place!

A courtship role-reversal should be associated with a reversal in sexual selection; fights by females and choice by males is expected to impose greater sexual selection on females than males. Sexual selection is somewhat of a more difficult process to understand than competition and mate choice. Indeed, whole books have been devoted to defining and measuring sexual selection (e.g., Bradbury and Andersson 1987; Clutton-Brock 1988). There seems to be general agreement that sexual selection occurs on a sex when members of that sex compete for access to mates (in the case of males, fertilizations), or to the best mates (Chapter 1). Therefore, any trait that conveys an advantage in direct sexual competition, or is preferred by the opposite sex, is a sexually selected trait. "Advantage" refers to fitness, which can be measured in success in acquiring mates. Therefore, one way of estimating the "opportunity" for sexual selection in a sex is to measure variation in mating success (Bradbury and Andersson 1987). 

Success at mating results in success in producing offspring. Therefore positive relationships between number of offspring and mating frequency, is a convenient way of expressing male-female differences in sexual selection (Bateman 1948; Arnold and Duvall 1994) (see the plot at bottom of Fig. 9.10). A greater slope of the relationship for one sex, typically males, than that of the other sex indicates stronger sexual selection on the former sex. This is because a greater slope can indicate large fitness returns for each mate won in the competition for copulations. Note, however, that a positive correlation between mating frequency and number of offspring may not necessarily be a result of mate competition. For example, even when female katydids are not competing for mates, each additional mating (spermatophore meal) enhances their fecundity (Table 6.2). Therefore other sex differences, such as in variation in numbers of mates (Arnold and Wade 1984) , may be necessary. 

If a positive relationship between mating frequency and fecundity indicates sexual selection, a reversal in sexual selection should be the case when the female slope exceeds the male slope (i.e., the slope of the female line in the plot at the bottom of Fig. 9.10 becomes steeper than that of the male line: See Figure below)) (Arnold and Duvall 1994; Lorch 1999). However, no study has yet examined this prediction, particularly in field populations. Although I lack data on male mating frequencies in nature, there are some for female mating frequencies in role-reversed populations which can be compared with data from populations with typical roles. The prediction is that food stress increases sexual selection on females as revealed in (i) increased variation in mating success and (ii) an increase in the slope of the fecundity x mating-frequency plot (Bateman curve) for females (Prediction 3: . 

The first part of this prediction was supported in cage experiments with pollen katydids: variance in female mating success was greater in low-food than high-food cages (Gwynne and Simmons 1990).

The second part of the prediction was not supported in a study of role-reversed and typical-role populations of Mormon crickets observed in nature: although the slopes did not differ, both populations showed positive slopes between mating frequency (determined by counts of spermatodoses, see Chapter 6 and Fig. 2.31e) and number of mature eggs in the ovary (Gwynne 1984c). One problem with these field data, however, is that fecundity estimates using only ovarian eggs were poor because females may have laid many of their eggs. However, this was not a problem in cage experiments with Mormon crickets because females laid no eggs. Moreover, the fecundity measure I used in the cage studies was total mass of mature ovarian eggs (egg number x egg mass, the latter known to be influenced by spermatophylax feeding: Chapter 6). Finally, in the second replicate of these experiments (in 1990) I estimated female mating frequency directly from observed number of spermatophores received by marked individuals. The results (from the data collected in Gwynne 1993) were equivocal: data from 1989 supported the hypothesis with a highly significant relationship in low-diet females (where behavioral role-reversal was observed) (Spearman rank correlation R_s = 0.38, P < 0.003) but no relationship in high-diet females (R_s = -0.1, P > 0.5); however, in 1990 both low diet (R_s = 0.35, P < 0.03) and high diet females (R_s = 0.34, P < 0.003) showed a significant positive relationship between mating (spermatodose) number and total fecundity.

Measuring sexual selection using variation in mating frequency and correlations between this variable and fecundity ignores a crucial aspect of the process of sexual selection because we need to know what traits give successful individuals a competitive edge (Darwin 1871). Increased success in sexual competition (moving up the fecundity x mating frequency curve in Fig. 9.10) should be directly related to increased size or elaboration of the sexually-selected trait ("secondary sexual character:" Hunter 1786) such as a peacock's tail or (in a role-reversed example) bright coloration in female button quails. In Mormon crickets a less eye-catching trait, but one that is nevertheless under sexual selection in females is body size (Chapter 1); in role-reversed Mormon crickets, as well as Metaballus litus (Gwynne 1985) and pollen katydids (Simmons and Bailey 1990), males preferred larger, more fecund females as mates. This brings us to prediction 4: under food stress larger females are expected to mate more frequently as revealed by a body size x mating frequency correlation.

This prediction was supported in the cage experiments with both pollen katydids and Mormon crickets. In pollen katydids, the pooled data from all replicates within each food treatment (see legend to Fig. 9.11) showed that heavyweight females in low-diet cages mated significantly more frequently. On the other hand, there was, as predicted, no body-size mating advantage in high-diet cages (Gwynne and Simmons 1990). 

However, a positive correlation between female weight and mating frequency may be due to weight gain from the more frequent spermatophylax meals acquired by large females rather than a large females attracting more matings as posited by the sexual selection hypothesis (Gwynne and Simmons 1990). This cause and affect problem was dealt with in work with Mormon crickets because our measure of body size was pronotum length, a metric that does not change after the adult molt. There was a positive correlation between pronotum length and number of spermatodoses (mating frequency) only in role-reversed populations (Gwynne 1984c) and experimental manipulation of diet (Gwynne 1993) confirmed this. This correlation was found in the low-diet treatment where we observed role-reversal but not in high diet cages where the roles were typical (Fig. 9.14B) .  These data support the hypothesis that sexual selection on females favors larger body size in role-reversed populations.

FIGURE 9.14B.  There was detectable sexual selection for increased female body size in the food-stressed cages where sexual competition among females had been observed.  This was revealed in a significant positive regression between female size (pronotum length) and mating frequency (open points and dashed line).  No significant regression exists for data from cages with ad lib food (closed points and continuous line).  Because these correlations did not differ between the four replicate cages within each food treatment, they were pooled for analysis within each treatment.  Females that had mated were easily spotted by the large spermatophylax “tag”.

In theory (Fig. 9.10), a reversal in sexual selection on the sexes is caused by the time out from mating activity by males exceeding the time out of females (Parker and Simmons 1996). The greater time out of males, per se, is argued to be due male investment in individual offspring (at the cost of investing in other offspring) exceeding that of the female (Trivers 1972). Is there any evidence for katydids that there is a reversal in parental investment by the sexes when food is limited (prediction 5)? 

For katydids there have been no measures of relative parental investment in terms of the cost in producing other offspring. However, because investment by both sexes appears to be similar in currency ⁽³⁾ - both sexes make a prezygotic nutrient contribution to eggs (Bowen et al. 1984) - it may be possible to estimate relative parental investment, by measuring investment in individual offspring. For pollen katydids in particular male and female parental investment appears to be similar because pollen and spermatophylax nutrition have similar effects on the number and size of eggs that the female subsequently produces (Simmons and Bailey 1990). Thus when food is limited the proportion of materials of male origin (spermatophylax) in individual offspring (eggs) is expected to increase because females should have less material from other sources in their own reserves (Gwynne 1991). This prediction was supported in studies in which male (spermatophore nutrients) and female contributions to eggs were labelled.

The photo at bottom shows a Rhamphyomya longicauda female (dance fly: Diptera: Empididae) inflating her abdomens before entering an all female competitive swarm. The inflated abdomen and leg hairs of a female in the swarm exaggerates her size to males entering the swarm from below.

1. Females but not males have been observed to move away from sites with less food (Shelly and Bailey 1995) and may use male song (Bailey and Simmons 1991) or food odors to find local food patches. However, the high risks of movements probably mean that hungry pollen katydids do not attempt random searches for food over very long distances.

2. Mating rates of the sexes are equivalent to the operational sex ratio when the adult sex ratio is 1:1 and maintenance costs of mating activity are very small (Clutton-Brock and Parker 1992).

3. In contrast to other animals in which parental investment by the sexes can come in different currencies: e.g. investment in gametes versus in parental care (Knapton 1984).