!" #! $    % #""$#! &'#  ()*)+)) , #$&$-                                       !  "# $%%# "%&%% '    '    ' (  ) *    +       )

  ,     -) $%%#)       .   ) /       .   '   (  ) !       )                    $0) 12 )    ) /34 5"6##162"786$)

        +  '  '             + ) 9 +         : +   +          ) *   '     ))                                  ) *  +    '         '   ;            '   '    ;     <              <          '  ::  ' '         ) *              '         '       '      +       ) * +    '             =          '   +   '  '     '         ) .               + >  '      + >     ) 9 +                                  ' ) *   '   '        6      ) .       +        '    '           +        '    +     ) *         ? ' :           '    '       ' + ?      ) /              =     ;                6    =         ) /    /  + '   '         ''   '               +       )

      +             :: 

!"# $ %  &     '   '  & ( ) *+ &    & ',-./01   & 

@ >A -) ,   $%%#

/334 "2#"62$"1 /34 5"6##162"786$)  &  &&& 6171% B &== ):)= C D &  &&& 6171%E   

                             

   !                      " #

   ! $      %  &  %  & &       '    " #

 (  )   ! *  +        ,, '   " #

-   ! .%& / (  ) '               +    0       ,, 1     "  #

-   !    (  ) 2       ,, '     

            

Contents

Introduction...... 7 The three levels of mate choice...... 7 The meaning of pheromones ...... 9 Study species and Methods...... 14 Natural history...... 14 Culturing ...... 15 Assessment of pheromone streaks...... 16 Results and discussion ...... 18 Individual variation ...... 18 Heritability ...... 23 Cost ...... 24 Conclusions...... 30 Sammanfattning ...... 33 Partnervalets tre nivåer...... 33 Feromoners innebörd...... 35 Studieart och metoder...... 36 Resultat och diskussion ...... 36 Acknowledgments...... 39 References...... 41

Introduction

Chemical signaling is generally regarded as the most ancient and widespread form of communication (Bradbury & Vehrencamp, 1998; Wyatt, 2003). Throughout the living world, chemicals are exchanged between mating part- ners at each step in the complex suite of behaviours leading to reproduction. This has been evident almost since the word ‘pheromone’ was coined in 1959 (Karlson & Lüscher, 1959), and indeed the first identified pheromone was the chemical released by female silk moths, Bombyx mori, to attract males (Butenandt et al., 1959). Since then, the use of pheromones during reproduction has been reported from taxa ranging from bacteria to mammals (see paper I for a review).

The three levels of mate choice Pheromones are ubiquitous in sexual communication, but in general there is little knowledge about what these signals actually communicate. Broadly speaking, sexual communication aims at influencing mate choice, defined as ‘any behaviour that restricts the set of potential mates’ (Wiley & Poston, 1996). Such behaviours should be beneficial whenever they increase the number and/or quality of offspring for choosy individuals as compared to those mating randomly. However, this wide definition of mate choice neces- sitates some sub-categories. First, we might distinguish between ‘direct’ and ‘indirect’ mate choice. Direct mate choice, in the words of Wiley and Poston, ‘requires discrimination between attributes of individuals of the op- posite sex’, whereas indirect mate choice is the result of all other behaviours that restrict the set of potential mates. Second, the set of potential mates might be restricted to the right species, the right mate recognition system (sensu Paterson, 1985) or the right indi- vidual qualities. These three levels of mate choice are defined both by the different cognitive abilities needed to perform them and by different evolu- tionary processes. Thus, species recognition evolves to avoid the costs of mating with the wrong species, mate recognition to coordinate the sexual behaviour between the sexes, and individual mate assessment to increase success in intraspecific competition for mates. From an evolutionary point of view, the results of selection for species and mate recognition are much the same. Both are driven by non-intraspecific processes, i.e. interspecific com-

7 petition for communication channels (Cardé & Baker, 1984) and environ- mental change (Paterson, 1985) respectively; both are essentially adaptive (Löfstedt, 1993) and both could be subject to stabilising selection (Cardé & Baker, 1984; Paterson, 1985). In contrast, individual mate assessment evolves by the benefits of choosing a mate that is in some way superior to its conspecifics, insofar as this superiority translates into more or better quality offspring. Moreover, a failure to recognise the right species or mate recogni- tion system might result in no offspring at all, whereas misjudgement of mate quality rarely has such dire consequences. Sexual signals have evolved to influence mate choice at each of these three levels and we can consequently distinguish between species recogni- tion signals, mate recognition signals and mate assessment signals. Each of these three types of signals has its own characteristic design, evolved through different processes and advertising different kinds of information. Thus, species and mate recognition signals could be expected to be relatively uniform within each species and sex respectively. In contrast, mate assess- ment signals must advertise the individual identity of the sender and its qual- ity as a mate, and moreover do this in a way that cannot be faked. Mate as- sessment signals are thus likely to be costly and to vary qualitatively with the condition of the sender in order to provide honest information in a stable signalling system (Grafen, 1990; Rowe & Houle, 1996); they should fur- thermore be highly variable between individuals and are prone to a kind of exaggeration that often seems quite detrimental to the survival of the sender. Although species recognition, mate recognition and mate assessment are not mutually exclusive functions of a signal, it might be helpful to keep these distinctions in mind when investigating the role of a specific sexual signal. In the study of chemical signals in particular, where signal function is often summarily described by the word ‘sex pheromone’, additional insights may be gained by investigating at which level a specific pheromone acts. Studies on the roles of pheromones in sexual behaviour have tended to focus attention on long-range mate attraction (Svensson, 1996) or the impor- tance of chemical signals in species and mate recognition (Löfstedt, 1993; Ptacek, 2000). Indeed, in his seminal review of sexual selection Malte Andersson (1994) cites just nine studies that had explored pheromone-based mating preferences; in striking contrast, several hundred studies are cited that had explored the consequences of visual or acoustic signals. Moreover, while a vast body of research has investigated the molecular composition of pheromones (for reviews, see e.g. Stevens, 1998; Arn et al., 1986), com- paratively little work has been carried out on their behavioural significance at the level of individual mate assessment. However, several lines of evi- dence suggest that the role pheromones may play in mate choice is far

Note that the term ‘mate recognition signal’ used here is not identical to the term ‘sex recog- nition signal’ used in paper I, although the former includes the latter.

8 greater than previously expected. First, there are a number of recent studies indicating individual mate choice by chemical cues (see table 1). Further, other studies have revealed considerable individual variation in pheromone production (e.g. Schlyter & Birgersson, 1989; Buesching et al., 2002), and researchers have also begun exploring the cost of pheromone production (e.g. Bushmann & Atema, 2000; Nahon et al., 1995).

The meaning of pheromones In paper I, we review the evidence for pheromones as indicators of mate quality and their use in individual mate assessment. For a sexual signal to be adaptive, it must convey a benefit to the receiver (Andersson, 1994). In mat- ing systems where individuals obtain immediate benefits from their choice of mate (for example nuptial gifts, territories or oviposition sites), the benefit of choice is relatively intuitive. However, in species where choosy individu- als gain nothing more than the genetic material for the production of off- spring, such as in extra pair copulations or in lek-mating systems, the rea- sons for choice are less obvious (Höglund & Alatalo, 1995). Traditionally, models proposed to explain the evolution of mate choice explore the conse- quences of mate preferences through the direct or indirect gains that an indi- vidual accrues from their choice (Andersson, 1994). These gains, communi- cated through a diverse array of sensory modalities, constitute the ‘meaning’ of the sexual signal. A unique feature of pheromones, as compared to signals communicated through other sensory modalities, is that they are composed of matter that in itself can become a resource. Thus, many insects that rely on plant-derived substances as a source of protection against predators, use these chemicals, or their derivatives, as pheromones in mate attraction (Gullan & Cranston, 1994; Landolt & Phillips, 1997). For example, pyrrolizidine alkaloids, an intensely bitter-tasting group of chemicals found in plant families such as Asteracae and Fabacae, are utilised by a wide range of lepidopterans such as danaid butterflies (Dussourd et al., 1989), arctiid moths (Dussourd et al., 1991; Vonnickischrosenegk & Wink, 1994) and potentially ithomiine butter- flies (Boppre, 1990). In the above species, males incorporate sequestered chemicals into their spermatophores, but also use a proportion of the chemi- cals as pheromones for mate attraction. Females discriminate between males based on the pheromones advertised and preferentially mate with males emitting the strongest chemical signal. Females gain a direct benefit from their choice as the protective chemicals are transmitted from the male to the female via the spermatophore during mating. These chemicals are then in- corporated into the fertilised eggs, reducing their risk of predation. In the above examples, the meaning of the signal to the choosing individ- ual is quite obvious. In the words of Marshall McLuhan, the medium is the

9 message. However, in the majority of species the link between the phero- mone and it’s informational content is far less apparent. The most commonly proposed mechanism to ensure the honesty of a mate quality signal is costli- ness (early treatments of this idea are e.g. Williams, 1966; Zahavi, 1975). If a signal is costly to produce, it’s strength or quality might reflect the pheno- typic and possibly genotypic quality of the sender of the signal. Indirect benefit models are mostly founded on the assumption of costli- ness, although the assumed costs are rarely measured (see section ‘Costs’, page 24, for a discussion of this). These models propose that discriminating individuals benefit from their choice of mate through the increased fitness of their offspring. The models require that the preference of the choosing indi- vidual and the preferred trait of the chosen are heritable, and that they covary across generations (Andersson, 1994; Lande, 1981). There are a number of studies indicating good-genes benefits accrued through pheromone-based mate assessment (Table 1). In the cockroach, Nauphoeta cinerea, females use chemical signals to select their mates (Moore & Breed, 1986). Females gain indirect benefits by mating with the male of their choice through the production of offspring that reach sexual maturity more rapidly (Moore, 1994). In addition, discriminating females gain direct benefits from their choice as the time between clutches is also reduced. A good-genes mecha- nism is also proposed for the evolution of pheromone expression in the uni- cellar brewers yeast, Saccharomyces cerevisiae (Nahon et al., 1995). Mate location in the two mating types of S. cerevisiae is mediated by a combina- tion of short peptides (the pheromone), which are relatively cheap to pro- duce, and large proteins that have a post-translational function. The large post-translational proteins are thought to act as honest indicators of quality because they provide information about a cells’ ability to invest in resources and utilise different biochemical pathways (Nahon et al., 1995). An alterna- tive view is that mate choice in S. cerevisiae is based on the relative symme- try between the quantities of chemicals released by the two mates (Pagel, 1993). The ability to withstand infection from parasites might reflect geneti- cally inherited immunological capabilities, and females of the grain beetle, Tenebrio molitor (Rantala et al., 2002), and the house mice, Mus musculus (Ehman & Scott, 2001), has been shown to prefer the scent of non-infected males. However, neither of these studies has explored whether females also gained direct benefits from discriminating against parasitised males, al- though this is a possibility.

10 Table 1. Experimental studies reporting pheromone-based mate assessment. Sex is the emitting sex. Message is the information that the receiver extracts from the sig- nal according to the respective authors; FA = fluctuating asymmetry; Imm. = Im- munocompetence; MHC = major histocompatibility complex. Benefit is the benefit of choosing for the receiver according to the respective authors. Species Sex Message Benefit Reference Insects Drosophila grimshawi Male Male quality Direct/Indir. Droney & Hock 1998 Ips pini Male Predator avoidance Direct Teale et al. 1994 Lutzomyia longipalpis Male Sexiness Fisherian Jones et al. 1998 Nauphoeta cinerea Male Male aggressiveness Direct Moore & Moore 1999 Neopyrochroa flabellata Male Resource Direct Eisner et al. 1996 Nicrophorus orbicollis Male Male size Direct Beeler et al. 2002 Osmoderma eremitica Male Resource Direct Larsson et al. 2003 Panorpa japonica Male Male quality (FA) Indirect Thornhill 1992 Tenebrio molitor Male Male quality (Imm.) Indirect Rantala et al. 2002 Tribolium castaneum Male Male quality Indirect Lewis & Austad 1994 Utetheisa ornatrix Male Resource Direct Dussourd et al. 1991 Crustaceans Homarus americanus Male Dominant vs subord. Indirect Bushman & Atema 2000 Rhynchocinetes typus Male Dominant vs subord. Indirect Diaz & Thiel 2004 Fishes Gasterosteus aculeatus Male Male quality (MHC) Indirect Reusch et al. 2001 Poecilia reticulata Male Male quality Direct/Indir. Shohet & Watt 2004 Amphibian Plethodon vehiculum Both Mate size Direct/Indir. Marco et al. 1998 Reptiles Lacerta monticola Male Male quality (FA) Indirect Martín & Lopez 2000 Thamnophis sirtalis Female Female size Direct Lemaster & Mason 2002 Mammals Homo sapiens Male Male quality (FA) Indirect Thornhill et al. 1999 H. sapiens Male Male quality (MHC) Indirect Wedekind et al. 1995 Mus musculus Male Dominant vs subord. Direct/Indir. Drickamer 1992 M. musculus Male Male quality (MHC) Indirect Penn & Potts 1998 M. musculus Male Good genes & MHC Indirect Roberts & Gosling 2003 M. musculus Male Infection status Indirect Zala et al. 2004 Nycticebus pygmaeus Male Competitive ability Direct/Indir. Fisher et al. 2003

Pheromones may also reflect an individual’s symmetry, a suggested indi- cator of quality. This idea is based on the premise that symmetry is a reflec- tion of an individual’s ability to withstand environmental stress and pertur- bation during development and is thus an indicator of genetic quality (Jennions & Oakes, 1994; Palmer & Strobeck, 1986). However, since sym- metry is a phenotypical trait that probably is dependent on both genes and the environment during development, this idea is controversial. Several stud- ies report a correlation between the symmetry of a particular trait and mate

11 choice based on chemical signals, and propose that pheromones may act as honest signals of phenotypic and/or genetic quality. In the Japanese scor- pionfly Panorpa japonica, females prefer males with symmetrical fore- wings, a choice that apparently is guided by pheromones alone (Thornhill, 1992). Similarly, female Iberian rock lizards, Lacerta monticola prefer the scent of males with more symmetrical femoral pores (Martin & Lopez, 2000). In humans, Homo sapiens, females are reported to prefer the scent of males with the highest degree of facial symmetry, and show greatest prefer- ence when they are in the most fertile phase of their menstrual cycle (Thornhill & Gangestad, 1999). These ‘good genes’ models assume that a particular trait reflects an in- nate, genetic, quality of the male and on average all females should prefer the same male. However, females may mate for reasons other than to obtain ‘good genes’ for their offspring; they may mate to obtain the most compati- ble genes (Trivers, 1972), or to increase the genetic diversity of their off- spring (Tregenza & Wedell, 2000). Compatibility models assume that to increase the fitness of their offspring, females choose mates that are opti- mally genetically dissimilar to them rather than males that bear the largest or most elaborate ornament (Mays Jr & Hill, 2004; Tregenza & Wedell, 2000). Alternatively, they may mate with the most heterozygous males because these increase the genetic diversity of their offspring. However, as males are unable to pass their heterozygosity per se onto their offspring, the male trait is non-heritable (Mays Jr & Hill, 2004). Empirical support for these models is steadily increasing, and evidence suggests that pheromones may provide a more reliable assessment of genotype prior to mating than either visual or acoustic signals. Thus, female preference for heterozygosity based on chemical assessment has been reported in mice, M. musculus (Penn & Potts, 1998), humans, H. sapiens (Thornhill et al., 2003; Wedekind & Furi, 1997), sticklebacks, Gasterosteus aculeatus, (Reusch et al., 2001; Milinski, 2003), guppies, Poecilia reticulata (Shohet & Watt, 2004) and alpine newts, Tritu- rus alpestris (Garner & Schmidt, 2003). Females may also increase the genetic diversity of their offspring by mat- ing multiply, thus ensuring many sires. There are a few examples of avoid- ance of previous mating partners where the proximate cue used in recogni- tion is assumed to be odour. Given the opportunity, females of the hide bee- tles, Dermestes maculatus (Archer & Elgar, 1999) and the pseudoscorpion Cordylochernes scorpioides (Zeh et al., 1998) prefer to mate with novel males rather than previous mating partners. Similarly, in the adder, Vipera berus, females mate up to eight times during a breeding season, but avoid re- mating with a previous partner (Madsen et al., 1992). While these three spe- cies are known to use pheromone-based mate assessment, none of these studies has assessed explicitly whether polyandrous females used chemical cues to increase the heterozygosity of their offspring.

12 The ‘good genes’ model was once proposed as an alternative to the then more discussed Fisherian model of sexual selection (Zahavi, 1975). Fishe- rian models propose that a female preference for a particular male trait yields indirect benefits through the production of sons with higher reproductive success (Fisher, 1958). From the perspective of chemical communication, indirect Fisherian benefits means that the capacity to produce the pheromone is heritable and that the pheromone does not act as an indicator of viability. To date there is little empirical support for Fisherian models of sexual selec- tion whether or not pheromones are the traits used in mate assessment (Table 1). In addition, much of the data is correlational. In insects, females of the sandfly Lutzomyia longipalpis prefer to with mate males that invest most in wing fanning and pheromone production (Jones & Hamilton, 1998) and fe- males that mate with more attractive males produce sons that are themselves more attractive (Jones et al., 1998). Similar patterns are observed in female tiger moths, Utetheisa ornatrix (Iyengar et al., 2001) and house mice, Mus domesticus (Drickamer, 1992). Even though evidence for a role for pheromones in mate choice is rapidly increasing, the mechanisms are still to a large degree unknown. This is partly due to a mismatch in chemical and behavioural knowledge. As an example, the chemical structure of the pheromone and the modulation of the message are known in only five of the studies in table 1 (i.e. Teale et al., 1994; Moore & Moore, 1999; Eisner et al., 1996; Dussourd et al., 1991; Lemaster & Ma- son 2002). Apart from chemistry, knowledge about the individual variation, the heritability and the cost of pheromones are instrumental in a deeper un- derstanding of pheromones as mate assessment signals. The rest of this the- sis is an attempt to investigate these three factors, using a pheromone of the fruit fly Drosophila grimshawi as a model. However, during the course of this investigation I frequently encountered both answers and new questions that I had not explicitly been looking for. Thus, while individual variation, heritability and cost provides the conceptual framework for this thesis, there will be frequent digressions into other areas of interest for the understanding of the meaning of the male pheromone of D. grimshawi.

13 Study species and Methods

Natural history Drosophila grimshawi Oldenberg belongs to the picture-winged group of Hawaiian Drosophilae. Adults in this species have delayed sexual maturity (females: 25 - 30 days; males 10 - 15 days) and may live for several months in the laboratory (Jones & Widemo, unpublished data). D. grimshawi has a lek-based mating system, where males aggregate on leaves or tree trunks where they engage in competitive and courting behaviour. Females visit the leks only to mate since feeding and egg-laying sites are located elsewhere (Spieth, 1986; Droney, 1990). Males have a wide repertoire of tactile, visual, chemical and probably acoustic signals that are used in interactions with females and other males (Ringo & Hodosh, 1978; Spieth, 1986; Hoikkala et al., 1989). Aggregation is not obligate since solitary males seem to perform all the behaviours associated with lekking, but in the laboratory environment leks typically vary in size from two to 10 males (Droney, 1994). While lekking, males repeatedly drag the tip of their abdomen over the substrate, thus depositing a sticky substance that forms distinctive streaks a couple of millimetres long. The streaks contain a pheromone produced by a

Figure 1. a) Male D. grimshawi. The arrow points to the anal lobes. b) Dor- sal view of the posterior portion of the abdomen with fully extruded anal lobes. From Hodosh et al. 1979, J. Morph., Copyright © 1979 Wiley- Liss, Inc., a subsidiary of John Wiley & Sons, Inc.; Reprinted with permis- sion of John Wiley & Sons, Inc.

14 pair of glands situated at either side of the male anus (Hodosh et al., 1979) (fig. 1). This pheromone attracts both females and males from a distance (Droney, 1994). Pheromone streaks can be quantified by counting the total number laid down by a male over a standard time period (Droney & Hock, 1998). Moreover, the quantity of streaks produced by a male correlates posi- tively to his mating success and can be used as an estimate of male phero- mone production (Droney, 1996; Droney & Hock, 1998). In contrast to many other lekking species, D. grimshawi males do not defend territories (Droney, 1992) and pheromone depositions from a male can be spread all over the lek. Moreover, males may deposit secretions on top of streaks al- ready deposited by themselves or other males. The specific chemical composition of the pheromone is unknown, but the general composition of the anal gland secretes of three species in the adias- tola sub-group of picture-winged Drosophila has been examined, revealing them to be a mixture of 3 to 14 hydrocarbons with chain-lengths varying from 19 to 30 carbons (Tompkins et al., 1993). Since the secretions of D. grimshawi forms distinctive streaks that are visible for several months, it seems probable that these volatile hydrocarbons are contained within a ma- trix of some less volatile substance.

Culturing The two stocks we used in these studies had different origins and histories in the lab. The “wild stock” was originally derived from five wild-caught fe- males captured from the Makawao forest reserve north of the Haleakala cra- ter on the island of Maui in 2000, and bred in our lab since then. The “lab stock” is an isofemale line, also known as the G1-stock, collected from Au- wahi south of the Haleakala crater in 1965 and bred in our lab since 1999. The populations were maintained in a series of 2.0 litre jars (20-30 adults per jar) in the laboratory on a standard Drosophila rearing medium (malt and agar). The base of the jars was filled with moist sand (approximately 2 cm height). Adult flies were given three glass vials (length 100 mm, inner di- ameter 27 mm), lined with damp, double-thickness tissue paper (20 u 30 mm) and containing 5 ml of standard medium at the bottom. Both sexes feed on the malt medium and females use the same malt substrate for oviposition. Vials were removed weekly and emerging larvae were fed a standard larval medium (malt, semolina, agar and dry yeast) ad libitum. At approximately two weeks, vials containing final instar larvae were placed in 2.0 litre jars lined with moist sand for pupation and left until adult eclosion. All individu- als were held at constant temperature (20 r 5 qC) and relative humidity (60 %) on a 14:10 h light: dark photoperiod (0730-2130).

15 Assessment of pheromone streaks To assess pheromone production, we covered the interior surface of a stan- dard food vial with a rolled up, rectangular sheath of transparent plastic film (900 u 850 mm). Males where placed solitary in the vials and left to deposit pheromones on the film for 24 hours. As males can deposit more than 1000 pheromone streaks per day (Droney & Hock 1998; Widemo & Jones, unpub- lished data) the plastic film was changed daily to minimise overlap of pheromone streaks. The films were photographed using a flatbed scanner (Agfa Snapscan 1212) (fig. 2 a). The digitalized images were composed of 255 grey tone levels with 255 as the lightest and had 15.6 pixels/mm. Images were cleaned of obvious dirt (e.g. fly faeces) by editing in Adobe Photoshop LE, and further processing was carried out using the ImageJ 1.30g (National Institutes of Health, USA) image processing software. Grey tone values were approximately normally distributed in the original images but mean values varied. Grey tone value distributions were standardized around the mean to facilitate comparisons between images. The minimum displayed value was set to 40 to enhance contrast, and a median filter was applied. The section of each image containing pheromone depositions was circled. Since the phero- mone streaks were the lightest parts of these sections, I measured the area of all regions with a grey tone value higher than the mean + 20 and an area greater than 10 pixels to obtain the total area of the pheromone depositions (fig. 2 b).

Figure 2. a) Original photo of pheromone streaks deposited on plastic film. b) The same image, processed in ImageJ.

To check the consistency of this method of measurement, I repeated the process at three different days with a sample of 12 images. The repeatability, computed according to the method described in Lessels and Boag (1987), for this sample was 0.98. The amount of pheromone deposited by male D. grim-

16 shawi has previously been estimated by manual counting of pheromone streaks (Droney & Hock, 1998). I repeated this procedure for a small sub- sample (n = 20) of our pheromone films and found a positive correlation between number of streaks and area measured as above (r2 = 0.35, t = 3.12, p = 0.006).

17 Results and discussion

Individual variation Studies of individual variation in chemical signals have been held back by the technical difficulties of analysing the often very minute quantities of these chemicals that each individual produces (Schlyter & Birgersson, 1989), but also by the dominating idea that the main function of sex pheromones are in species and mate recognition. The traditional paradigm (e.g. Roelofs & Cardé, 1974; Cardé & Baker, 1984) as well as the ‘species-recognition con- cept’ proposed by Patterson (1985) and others, maintains that pheromones should be subject to stabilising evolution and individual variation thus unlikely. However, since the beginning of the 1980’s a substantial amount of individual variation in pheromones has been revealed (paper I). This varia- tion may be achieved through quantitative or qualitative differences in the pheromone blend, or via differential release rates. In general, the quantity of pheromone released varies more than the ratios of its components. There is great variation in the amount of pheromone deposited by males of D. grimshawi and the repeatability of the pheromone deposition by indi- vidual males is fairly high, with some males consistently depositing less pheromone than other males (Droney & Hock, 1998). Less is known, how- ever, about qualitative variation in the pheromone. In two experiments (pa- pers II and III), we set out to look for qualitative variation, but since I could not analyse the pheromone chemically, we had to resort to somewhat indi- rect means. In paper II, I wanted to confirm the attractive properties of pheromone depositions at close distance, since this approximates a mate choice situation on a lek. Earlier experiments (Droney, 1994) were more concerned with the pheromone as a long distance “to-the-lek”-signal. I per- formed experiments on both wild and lab stock flies. First, I tested the attrac- tion of males and females to pheromone depositions from males of their own stock. Second, I performed a cross-attraction experiment, where I exposed males and females of either stock to pheromone streaks deposited by males of the other stock. Individual flies in both experiments where released into 0.5 litre plastic boxes containing one food vial marked with pheromone and one without pheromone. After 30 minutes, the whereabouts of the fly was noted. The amount of pheromone deposited in each vial was only recorded for the second experiment.

18 At the close distance and confined space of the bioassays, the phero- mone’s role as a mate recognition signal is probably less important than its role in mate assessment. It seems likely that the amount of pheromone in the marked vials was enough to guide the flies to the lek in most of the bioas- says. The experiments should thus approximate a mate assessment, or in the case of males competitor assessment, situation when a fly has landed on a lek. If assessment takes place, flies should act differently on quantitative or qualitative variation in the pheromone, for example preferring large amounts of pheromone or certain compositions. Wild males deposited on average r SD 115.8 r 92.5 mm2 of pheromone (N = 78) and lab males 99.2 r 79.1 mm2 of pheromone (N = 104), which was not a significant difference in amount of pheromone (t180 = -1.30, P = 0.19). In the first experiment, males and females of both stocks where strongly attracted to their own stock pheromone. In the second experiment however, the only significant attraction was for lab females to lab pheromone and there was no evidence for cross-attraction. This disparity between the ex- periments was probably due to differences in the amount, and possibly freshness, of pheromone. In the first experiment, I let two males deposit in each vial, and these vials where used in the experiment about three hours later, whereas only one male was used as depositor in the second experiment and there was a one day interval allowing for photography before the plastic film was used in the experiment. However, it cannot be excluded that the stronger attraction in the first experiment was due to there being pheromone depositions from two males, thus constituting a lek, instead of only one. A logistic regression on the second experiment revealed no effects on choice of vial from either the stock of the depositing males, the stock of the responding flies or the sex of the responding flies, and there were no interac- tions between these factors. However, there was at least a suggestion of an effect of area of pheromone depositions (LLR = 3.12, df = 6, P = 0.08). Bio- assays where the flies first visited the vial marked with pheromone deposi- tions contained on average r SD 117.22 r 80.58 mm2 of pheromone (N = 66) and bioassays where the flies first visited the unmarked vial contained on average 88.70 r 71.72 mm2 of pheromone (N = 56) which was a significant difference in area (t120 = -1.30, P = 0.04). While the first experiment demonstrates that both males and females are attracted to pheromone of their own stock, the results with respect to quanti- tative versus qualitative variation in pheromone deposition are somewhat ambiguous. However, lab females seem to be able to differentiate between pheromone from their own and wild stock males, indicating a qualitative difference between the two since there was no difference in the amount of pheromone deposited by the two stocks. More surprising is perhaps the weak effect of amount of pheromone on attraction. This implies that other factors than quantity are important for attraction, or that the pheromone acts as a

19 species/mate recognition signal where the amount of pheromone carries no information as long as it exceeds the threshold level necessary for attraction. In paper III, we made a further attempt at looking for qualitative variation in the pheromone by testing whether males can discriminate between self and non-self pheromone depositions and if there is a stronger response to depositions from several competitors as compared to one. On the first day of the experiment, males of both stocks were placed singly in fresh food vials and left to deposit pheromones on plastic films for 24 hours. Films were photographed, and males that had deposited pheromones were randomly assigned to three treatments: treatment A were males received the same plas- tic film as they had deposited on the previous day; treatment B were males received a film with depositions from another male, and treatment C were males received a film without pheromone depositions. The males were left to deposit pheromones on these films for another 24 hours, after which the films where once again photographed and the flies killed. We used the films resulting from treatments A and B further, by exposing novel males to them. A new set of males received fresh food vials with films containing two days of pheromone depositions from either one male (treatment D; films from treatment A) or two males (treatment E, films from treatment B) (fig. 3). The males were left in the vials for 24 hours after which the films were photo- graphed and the males killed.

Figure 3. Experimental de- sign to investigate effects of self and non-self pheromone streaks on pheromone deposi- tion. At day 3 and 5, the treatments got substrates (white parallelograms) with different histories.

Lab males deposited more pheromone in the presence of streaks from a strange male than when exposed to their own streaks (fig. 4) and invested more when encountering streaks from two as compared to one strange male. Males given a blank film on the experimental day invested somewhere in between males that had access to their own streaks and those that received streaks from another male. Wild males showed no response to the manipula- tion in the first part of the experiment, but deposited much less in the pres- ence of streaks from two as compared to one strange male in the second part. This may seem difficult to explain, but a possible proximate explanation is that wild males require larger amounts of pheromone from a single strange

20 male in order to adjust their depositing behaviour. Data from this and the previous experiment suggest that wild males deposit somewhat more phero- mone than lab males, and that wild flies require larger amounts of phero- mone for being attracted over a short distance. Clearly, both stocks of D. grimshawi adjusted their pheromone depositing behaviour in response to experimental manipulation, proving the male ability to distinguish between competitors from qualitative differences in phero- mone streaks alone. Female ability to sample the aggregate pheromone streaks on a lek and determine relative male investment will provide females with an effective way of assessing male investment in lekking over a much longer time period than a female visit. Time spent on the lek is a good pre- dictor of male investment and success in D. grimshawi (Droney, 1992) as in most other lekking species (Höglund & Alatalo, 1995). Thus, males need not signal competitive ability directly using qualitative variation in the phero- mone and females need not prefer specific qualitative pheromone traits; fe- male ability to distinguish between deposits from different individuals in a lekking situation will suffice for assessing male quality. In this system, indi- vidual recognition of pheromone signatures might evolve without any bene- fits to females from mating with males displaying certain chemical qualities in their pheromone depositions. Male ability to determine male identity from pheromonal deposits may be beneficial even if males were unable to determine the relative investment of the competitors. Large leks usually mean increased chances of encountering prospective mates (Höglund & Alatalo, 1995) and may also result in better chances for lower ranking males to obtain matings (Widemo & Owens, 1995; Widemo & Owens, 1999). Thus, it will be beneficial for males sam- pling leks to be able to determine the number of males that have deposited pheromones. Being able to determine not only how many males that have been present but also how much they have invested will, obviously, enable males to assess the investment in lekking activities from competitors. Mat- ings in D. grimshawi often last more than a minute (own observation) and males often attempt to disrupt matings. An additional benefit to high quality males from investing in pheromone deposition could be that lower ranking males may be less likely to disrupt such males. Male interactions in lekking species often have significant impact on mating success, as males are tightly clustered in space. Here, signalling systems for conveying information be- tween males are especially likely to evolve (Widemo, 1997). It seems likely that the pheromone signalling system in D. grimshawi originally evolved for attracting females to leks. The transition to a signalling system for conveying information about individuals may well, however, at least in part have been driven by benefits from male-male signalling.

21 Figure 4. Total area of pheromone deposited over the stimulus day and experimental day in the first part of the experiment. Males in treatment A were exposed to their own pheromone deposits, while males in treatment B were exposed to streaks from a strange male. Males in treatment C were given a blank film on the stimulus and experimental day and the areas were added. Boxes are s.e. and whiskers 95% confi- dence intervals. N = 13 Awild, 13 Bwild, 14 Cwild, 17 Alab, 14 Blab, 18 Clab.

The benefits from distinguishing between the pheromone from different males for males and females that I have discussed here are reminiscent of the scent matching hypothesis (Gosling & Roberts, 2001) being put forward to explain territorial marking in mammals. According to this hypothesis, in- truders match the scent marks on a territory with the owner in order to assess his ownership and competitive abilities. This might help males to avoid fights they are likely to lose (Gosling et al., 1996) and females in mate choice (Rich & Hurst, 1998). Lekking antelope have been shown to use ol- factory cues from the soil on male territories to determine past success of the territory holder (Deutsch & Nefdt, 1992) and the scent-matching hypothesis has been suggested to work also in non-territorial species (Wyatt, 2003). Whether male or female D. grimshawi are able to match individual males to streaks deposited remains to be tested. Both males and females have the opportunity to sample old and freshly deposited streaks on the lek, however, as males do not defend territories.

22 Heritability The evolutionary potential of traits is determined to a large extent by their level of additive genetic variation, the heritability (i.e. the ratio between the additive genetic variance of the character to its total phenotypic variance, see Falconer and Mackay, 1996). Low levels of heritability are generally ex- pected for characters closely linked with fitness, since genes coding for such traits are predicted to reach fixation in populations (e.g., Falconer & Mac- kay, 1996). In the context of sexual signalling, pheromonal traits should be subject to strong stabilising selection if they function simply in terms of mate-/species recognition. If so, the heritability of pheromonal traits would be predicted to be fairly low. In contrast, if they are used as indicators of mate quality, heritability might be high since such characters are subject to trade offs with characters more important to immediate survival. In paper IV, we let males of both stocks, with known pheromone produc- tion and thorax length, mate with randomly selected females. Each male was paired to one female, and time to mating was recorded. We then recorded the number of offspring resulting from these matings, and measured pheromone production and thorax length for a sample of the sons. While large males turned out to have a mating advantage in this experiment, males that depos- ited large amounts of pheromone did not have greater mating success. In earlier experiments, Droney (1992) showed that large males had greater mat- ing success than small at a lek, but that this was due to a strong correlation between body size and lekking duration. Thus, in a lekking context with both male-male and male-female interactions, large males can invest more time in lekking but body size per se does not affect mating success (Droney, 1992). An analysis of covariance revealed that the relationship between phero- mone production and body size differed in the two stocks. The phenotypic correlation between these two traits was negative in the wild but not in the laboratory stock. Taken together with the results of the behavioural trials, this suggests the following interpretations of the results. First, given that females exercise considerable behavioural “control” over mating in D. grim- shawi (Droney & Hock, 1998), our results indicate that females do prefer large males but seem to ignore the amount of pheromone that males deposit. Second, if there is indeed an evolutionary trade-off between pheromone pro- duction and body size, as our phenotypic data suggests, then variation in pheromone production may be maintained because of frequency dependent sexual selection among males. Large males are obviously more successful in acquiring matings through female mate choice while smaller males deposit- ing more pheromone may have an advantage in male-male interactions. In spite of the often negative correlation between male body size and pheromone deposition in D. grimshawi, there is some evidence suggesting that both characters are condition dependent in their phenotypic expression. For example, long-lived males are on average larger and spend proportion-

23 ally more days on pheromone deposition than short-lived (Droney, 1990; paper V). Males reared on low-protein medium deposit less pheromone than do males reared on a high-protein medium (Droney & Hock, 1998). Interest- ingly enough, females in the present study produced more offspring when mated with males that where large and deposited large amounts of phero- mone. This suggests that the most parsimonious explanation for this result is a direct effect in females: males in better phenotypic condition might have been able to transfer larger ejaculates containing more sperm, leading to a higher offspring production in females. Another line of evidence suggesting that pheromone production is condi- tion dependent is the fact that pheromone production in male offspring was correlated with time to mating in the parental generation, in spite of the lack of additive genetic variation (see below). Male D. grimshawi are quite per- sistent in their courtship, but since females are reluctant to mate we suggest that the time to mating may to a large extent reflect female condition. Thus, females in good condition might take longer to accept courting males as mates. Variation in pheromone deposition among offspring might then be related to maternal phenotypic effects rather than any paternal genetic ef- fects. We failed to detect any additive genetic variation for either size or pheromone production in either of our stocks, despite extensive phenotypic variation. Similar results were found by Jones and Widemo (In press), who found no correlation between the number of pheromone streaks deposited by fathers and their sons. While this result is unsurprising for the lab stock, since it is essentially an isofemale line and thus should exhibit very low lev- els of genetic variation, the additive genetic architecture of the wild stock did not differ significantly from the lab stock. The extent to which these low levels of genetic variation reflect the situation in natural populations is cur- rently difficult to evaluate, since the relatively low numbers of founders in the wild stock might have significantly depressed genetic variation to levels which are below those found in natural populations. In any case, the fact that body size and pheromone production was negatively phenotypically corre- lated in the absence of additive genetic variation for either trait strongly sup- ports the conclusion of an allocation trade-off between the two traits. This is because genetic correlations can confound the interpretation of phenotypic correlations (Roff, 1996). Without genetic variation, such as in clones, this problem is diminished.

Cost Indirect benefits models of sexual selection are based on three assumptions that centre on the costs of signalling (Grafen, 1990). Firstly, the cost must reduce the fitness of the signaller. Thus, it is not enough to show that a signal

24 requires high levels of time or energy expenditure. Ideally, it should reduce the number of offspring produced in future generations, but a more feasible measure of fitness reduction would be higher mortality or reduced reproduc- tion for the signaller (Kotiaho, 2001). Secondly, the signal must be depend- ent on the phenotypic and genotypic condition of the signaller. Condition dependence arises because individuals in a better condition have a larger resource pool and are thus better able to afford exaggerations of the signal than individuals in worse condition (Rowe & Houle, 1996). This leads di- rectly to the third assumption that the marginal cost of the signal must be greater for signallers in bad condition than for signallers in good condition (Grafen, 1990; Rowe & Houle, 1996). Providing these assumptions are satis- fied, the signal is likely to be differentially expressed in relation to the condi- tion of the signaller and a well-developed signal indicates honestly the po- tential for high fitness returns to the receiver from choosing such a partner. Although a wealth of studies report a diverse array of costs associated with the production of sexually selected signals (see Kotiaho 2001 and refer- ences therein), only two have shown that such costs are differential and fit- ness reducing. In the barn swallow, Hirundo rustica, males with artificially elongated tail feathers have lower survival than males with naturally long tails (Møller & de Lope, 1994). In the wolf spider Hygrolycosa rubrofas- ciata, males maintained on a low-quality diet and that were forced to in- crease their drumming rate (a trait that is attractive to females), survived less well than males maintained on a high-quality diet with similarly increased drumming rates (Kotiaho, 2000). Further, among the many studies investi- gating general costs of sexual signals, only very few have been devoted to chemical signals. One possible explanation for the lack of research into the cost of chemical signals is that the traditional perception of pheromones is that they are cheap to produce (Cardé & Baker, 1984) and, since the discus- sion on honest signalling has been focused on costly signals, this has made pheromones a less promising field of research in this context. A problem faced by researchers exploring the honesty of signals is that because individuals in good condition are expected to have both well- developed sexual signals and high survival, it is notoriously difficult to show individual cost-susceptibility (Kotiaho, 2001). The cost of signalling is thus masked by the correlated longevity. If, however, lifespan is kept constant it is possible to compare the relative investment in signalling for individuals under a range of conditions. In paper V, we did this by statistical methods, and then related investment in pheromones to actual lifespan in males of D. grimshawi. We assigned 88 focal males of the wild stock to one of three treatment groups. Males in the “mate” treatment were given the opportunity to court in the presence of a female; males in the “rival” treatment were given the op- portunity to lek in the presence of another male; and “solo” males were given the opportunity to display in the complete absence of conspecifics (fig.

25 5). Focal males where exposed to these treatments during two days per week for the duration of their lives. To obtain an estimate of relative lifetime in- vestment in pheromone production, we divided the number of days that each male had deposited pheromones with the total number of days in his life- span.

Figure 5. Experimental set-up. The display arena consisted of two glass vials divided by a fine-meshed netting. a) Mate treatment, b) rival treatment, c) solo treatment.

In general, the probability of a D. grimshawi male depositing pheromones seems to be predicted quite well by lifespan: on average depositing males lived longer than males that did not deposit pheromones (fig. 6). The males in the worst condition, as reflected in their short lifespan, might not be able to produce any pheromone at all. However, males that encountered other males survived equally well whether they deposited pheromones or not. Thus, while the ability to deposit pheromones may be based ultimately on a male’s condition, an individual may also adjust his investment in such be- haviour depending on his social environment. In the presence of other males at a lek, a male may choose to parasitize the chemical signalling efforts of the rivals, particularly if they are signalling at a high rate. Such a strategy may permit a male to utilise his own energy reserves for other mating activi- ties such as courtship displays (Droney & Hock, 1998). This approximates the “hotshot” hypothesis that suggests that leks are formed because unattrac- tive males cluster around attractive males to gain access to females (Beehler & Foster, 1988), an idea previously proposed to explain lek formation in D. grimshawi (Droney, 1994). The question that is immediately raised is why this alternative strategy should be more pronounced when males encounter rival males than when they meet females? One possible explanation is that males that court females in the absence of rivals may need to deposit at least a minimum amount of pheromones to retain a female at a lek. A higher pro- portion of males deposited for a lesser proportion of their lives in the “mate”

26 Figure 6. Life span of males (days) that did and did not deposit pheromones in the three treatments. ż: Mate treatment; Ƒ: rival treatment; ¸: solo treatment. Means are given r SE. P < 0.05. N = 4 mate nondepositors, 10 rival nondepositors, 11 solo nondepositors, 25 mate depositors, 20 rival depositors, 18 solo depositors. treatment (25/29) as compared to the “rival” (20/30) and “solo” (18/29) treatments, which seems to support this line of reasoning. In D. grimshawi, the number of pheromone streaks deposited by individ- ual males is highly repeatable and dependent on the condition of the males (Droney & Hock, 1998). Our study further implies that males that deposit more pheromones and invest a greater proportion of their lives producing this chemical signal also live longer than males that invest less effort in pheromone production. This suggests that the expression of the chemical signal in D. grimshawi is differentially dependent on male condition. As- suming that condition is reflected in both the amount of pheromone depos- ited and lifespan, males in good condition seem to be able to allocate propor- tionally more days to pheromone deposition than males in worse condition. Furthermore, males that deposited for a large proportion of their lives also deposited more pheromones per day. Condition dependence of sexual characters has mostly been investigated by subjecting test animals to low-stress and high-stress treatments, where the stress factor usually has been related to the quality of food or immunocom- petence (see Cotton et al., 2004 and references therein). By this method it has been shown that the composition of the male pheromone in the cock- roach Nauphoeta cinerea varies with food quality (Clark et al., 1997), and that the attractiveness of the male pheromone of the grain beetle Tenebrio molitor varies with parasite load (Worden et al., 2000), immunocompetence

27 (Rantala et al., 2002) and food quality (Rantala et al., 2003). While these studies demonstrate correlations between pheromones and different aspects of condition, they do not investigate the heightened condition dependence expected for indicator signals (Iwasa & Pomiankowski, 1994; Cotton et al., 2004). By relating pheromone deposition to the range of life spans of male D. grimshawi, we showed that pheromone deposition increases with lifespan not only in absolute terms but also proportionally. Males that encountered other males, rather than females or no flies, de- posited pheromones for a greater proportion of their lives (fig. 7). This greater investment seemed to be costly since these males also had shorter life spans (fig. 6). Our results correspond well with theoretical requirements of honest signals: individuals in good condition should live longer and pay a proportionally lower cost for the signal than individuals in bad condition. To

Figure 7. Proportion of treatment days devoted to pheromone deposition in three treatments when lifespan is adjusted for. Means are given r SE. P < 0.05. N = 25 mates, 20 rivals, 18 solos. our knowledge this is the first evidence that a chemical signal has differen- tial fitness costs for the signaller. As with other sexual signals, there are a several ideas on how the costli- ness of pheromones might be achieved, ranging from increased predation (Hendrichs & Hendrichs, 1998) to social costs (Bushmann & Atema, 2000; Gosling, 1990). In contrast to studies dealing with acoustic or visual signals

28 however, the energetic costs of signalling is rarely adduced for pheromones. While not measured directly, it seems probable that there is an energetic cost of pheromone production in D. grimshawi since males can deposit more than 1000 pheromone streaks per day for up to 25 % of their lives, and the num- bers of streaks vary with food quality (Droney & Hock, 1998). A further potential cost may be realised not through the volatile components of the pheromone but rather from the matrix that contains them.

29 Conclusions

Chemical signals are omnipresent in sexual communication in the vast ma- jority of organisms. However, their function has traditionally been seen as restricted to species and mate recognition. In paper I, we reviewed the evi- dence for pheromones as signals of mate quality, and in the following papers I investigate the function of the male pheromone of the Hawaiian fruit fly Drosophila grimshawi. This species has a lek mating system, where males aggregate to compete over, and court, females. The experiments presented clearly indicate that the pheromone of D. grimshawi has multiple meanings, and that these meanings vary with social context. Thus, we have evidence from paper V that lone males devote a lar- ger proportion of their life to pheromone deposition than males subjected to females, but slightly less than males subjected to rival males. Moreover, these ‘solo’ males seemed to commence pheromone deposition earlier in their lives than males in the other two treatments. Taken together, these re- sults suggest a ‘calling’, i.e. species/mate recognition function of the phero- mone. Calling signals are less likely to be costly since the honesty of the signal is guaranteed simply by its presence. Thus, there is no need to in- crease the rate of signalling as long as the threshold level necessary for at- traction is surpassed. This conclusion is also consistent with the weak effect of amount of pheromone on close-range attraction recorded in paper II. However, when the social context shifts from one calling male to two lekking males, males seem to increase the investment in pheromone produc- tion until it exceeds a critical value and the pheromone becomes a costly signal. In addition, the higher levels of pheromone deposition in lekking males compared to courting males reported in paper V indicate that the evo- lution of the pheromone is more likely to be driven by male contest competi- tion rather than direct female choice. Although females prefer to mate with males that deposit the largest quantities of pheromone and invest most in courtship activity (Droney & Hock, 1998), courting males devote less time to depositing pheromone than calling males (paper V). Likewise, in an ex- periment where lekking behaviour in male-biased compared to female- biased situations was studied, males deposited pheromones more frequently in the male biased groups even when corrected for number of males (Widemo & Johansson, unpublished). This interpretation of the function of the pheromone is further corrobo- rated by the mating trial in paper IV. In this experiment, large males had a

30 mating advantage whereas males that deposited large amounts of phero- mones did not have greater mating success. Thus, given that females control mating in D. grimshawi (Droney & Hock, 1998), females seem to prefer large males but, at best, ignore the amount of pheromone that males deposit. However, it might be necessary for a male to produce at least a minimal amount of pheromone to retain a female and so gain a mating, since a larger proportion of males tended to deposit in the presence of females than with males or when alone (paper V). It thus seems that the pheromone, with re- gard to females, acts primarily as a mate recognition signal. However, it cannot be excluded that females have preferences for pheromone of certain qualities rather than quantities. In paper III, we propose a function for the pheromone in male-male inter- actions. Given that males can distinguish between pheromone depositions from several males, which seem to be the case, males are able to estimate the size of a lek and assess the competitive capacities of rivals, information that should be useful for all males when optimizing sexual behaviour. Several lines of evidence suggest that this optimization is channeled through two distinctive mating strategies. In paper IV we found that large males depos- ited less pheromone than small males. In another study, Droney and Hock (1998) compared the number of pheromone streaks with courtship rate, and found that two levels of sexual effort seemed to exist: males that courted vigorously and deposited many pheromone streaks, and males that just courted vigorously. In spite of the negative correlation between male body size and phero- mone deposition reported in paper IV, there is evidence for both characters being dependent on male condition. Thus, long-lived males spend propor- tionally more days on pheromone deposition than short-lived (paper V). Males that are fed a low-protein-containing medium deposit fewer phero- mone streaks than males fed on a high-protein-containing diet (Droney & Hock, 1998). However, large males are also more long-lived than small (Droney, 1990). Moreover, the females in paper IV produced more offspring when mated to males that were large and deposited large amounts of phero- mone, indicating that both characters are related to male condition. Taken together, these results suggest that there might be a trade-off between body size and pheromone production. Males in good condition seem to allocate their resources to either pheromone production or body size, the former be- ing an advantage in male-male competition and the latter in courtship. Varia- tion in pheromone production might thus be maintained through frequency- dependent sexual selection among males. The pattern of sexual behaviours sketched above might differ somewhat between the two stocks of D. grimshawi that we used in the experiments. Thus, both males and females of the lab stock seems to be more attracted to the pheromone than wild flies are (paper II), and we did not find any nega- tive correlation between body size and pheromone production in lab males

31 (paper IV). Moreover, lab males increase their deposition rate in the pres- ence of pheromone from one or two strange males (paper III), and we have indications that the frequency of non-depositors is larger in wild males. Al- though comparisons between only two populations do not allow for any strong conclusions, and we know too little about differences in general life- history, this at least suggests that the pheromone might have slightly differ- ent functions in the two populations. Thus, it seems that wild males are more discriminating in their use of pheromone than lab males, perhaps because the latter has been adapted to an environment of unlimited food resources for hundreds of generations. Moreover, the population densities in the lab are probably much higher than in the wild, which might favour the selection of behaviours related to lekking as opposed to courtship in the lab stock. The majority of picture-winged Drosophila does not form communal leks, but rather perform solitary displays during which they use anal gland secretions as attractants (Spieth, 1986). The ancestral function of the chemi- cal signal in this taxa is thus probably for mate attraction. This may have become modified through sexual selection when communal lekking evolved in D. grimshawi and its sister species. The indications of male parasitizing on the chemical signals of rivals found in this and other studies (Droney, 1994; Droney & Hock, 1998) suggest that the “hotshot” mechanism could be responsible for this transition to communal lekking. The multi-functionality of the pheromone may thus be a result of the evolutionary history of this species where new functions are added without the losing of old ones.

32 Sammanfattning

Kemiska signaler anses allmänt vara den mest ursprungliga och vittspridda formen av kommunikation (Bradbury & Vehrencamp, 1998; Wyatt, 2003). Över hela den levande världen utväxlas kemikalier mellan kontrahenterna vid varje steg i den komplicerade följd av beteenden som leder till reproduk- tion. Detta har varit uppenbart allt sedan begreppet ’feromon’ myntades för att beteckna en kemikalie utsöndrad för att ”utlösa en specifik reaktion” hos en artfrände (Karlson & Lüscher, 1959), och det första identifierade feromo- net var också mycket riktigt den alkohol som honor av silkesmasken, Bom- byx mori, avger för att attrahera hanar (Butenandt et al., 1959). Sedan dess har könsferomoner identifierats hos omkring 1300 insektsarter (Metcalf, 1998) och ett ständigt ökande antal ryggradsdjur.

Partnervalets tre nivåer Trots att feromoner är väldokumenterat ofrånkomliga i sexuell kommunika- tion är förvånande lite känt om vad de i själva verket kommunicerar. Sexuell kommunikation i vid mening syftar till att påverka partnerval, definierat som ”varje beteende som inskränker antalet potentiella partners” (Wiley & Pos- ton, 1996). Sådana beteenden är fördelaktiga när de leder till mer avkomma, och/eller avkomma av bättre kvalitet, för kräsna individer jämfört med slumpmässig parning. Emellertid nödvändiggör denna vida definition av partnerval några underkategorier. För det första kan vi skilja mellan ’direkt’ och ’indirekt’ partnerval. Direkt partnerval, med Wiley och Postons ord, ”kräver urskiljning mellan individuella attribut hos det motsatta könet” me- dan indirekt partnerval är resultatet av alla andra beteenden som begränsar antalet potentiella partners. För det andra kan antalet potentiella partners begränsas till rätt art, rätt partner-igenkänningssystem (sensu Paterson 1985) eller rätt individuella kvaliteter. Dessa tre nivåer av partnerval definieras både av de olika kogniti- va förmågor som krävs för att utföra dem och av de olika evolutionära pro- cesser som framkallat dem. Artigenkänning evolverar således för att undvika de kostnader som är förknippade med att para sig med fel art, partnerigen- känning för att koordinera könens sexuella beteende och individuell partner- bedömning för att öka framgången vid inomartskonkurrens om partners. Från evolutionär synpunkt är resultatet av selektion för art- och partnering-

33 enkänning likartat. Båda drivs av icke-inomartsspecifika processer, d.v.s. mellanartskonkurrens om kommunikationskanaler (Cardé & Baker, 1984) respektive miljöförändring (Paterson, 1985); båda är i grunden adaptiva (Löfstedt, 1993) och båda kan vara föremål för stabiliserande selektion (Cardé & Baker, 1984; Paterson, 1985). Individuell partnerbedömning, å andra sidan, evolverar genom fördelarna, räknat i antal avkomma, med att välja en partner som i något avseende är överlägsen sina artfränder. Sexuella signaler har utvecklats för att påverka partnerval på alla dessa nivåer, och vi kan följdaktligen skilja mellan artigenkänningssignaler, part- nerigenkänningssignaler och partnerkvalitetssignaler. Var och en av dessa tre signaltyper har sin egen karaktäristiska form, evolverad genom olika proces- ser och marknadsförande olika typer av information. Art- och partnerigen- känningssignaler kan således förväntas vara relativt likartade inom respekti- ve art och kön. Partnerkvalitetssignaler å andra sidan, måste torgföra både sändarens identitet och kvaliteter som partner, och till yttermera visso göra detta på ett sätt som inte kan förfalskas. Partnerkvalitetssignaler kan därför förväntas vara kostsamma och variera kvalitativt med sändarens kondition för att kunna överföra trovärdig information i ett stabilt signalsystem (Grafen, 1990; Rowe & Houle, 1996). De bör vidare variera mellan individer och blir lätt överdrivna på ett sätt som ofta förefaller direkt skadligt för sän- darens överlevnad. Även om artigenkänning, partnerigenkänning och part- nerbedömning inte är ömsesidigt uteslutande funktioner hos en signal, kan det vara till hjälp att hålla dessa distinktioner i sinnet när man undersöker vilken roll en specifik könssignal spelar. Särskilt i samband med kemisk kommunikation, där signalfunktion annars ofta summariskt beskrivs med ordet ’könsferomon’, kan ytterligare insikter ernås genom att undersöka på vilken nivå ett specifikt feromon agerar. Studier av feromoners roll i sexuellt beteende har i allmänhet ägnats åt partnerattraktion (Svensson, 1996) eller vikten av kemiska signaler i art- och partnerigenkänning (Löfstedt, 1993; Ptacek, 2000) medan partnerkvalitets- aspekten tenderat att förbises. Exempelvis citerar Malte Andersson, i sin inflytelserika översikt av könsselektionen (1994), bara nio studier vilka un- dersökt feromonbaserade partnerpreferenser; detta i slående kontrast till de åtskilliga hundra studier vilka ägnats åt visuella eller akustiska signaler. Mycken möda har vidare ägnats åt feromoners kemiska struktur (se Stevens, 1998; Arn et al., 1986, för översikter) medan jämförelsevis lite arbete vigts åt deras beteendemässiga betydelse på partnerbedömningsnivån. Emellertid finns det nu mycket som tyder på att feromoners roll vid partnerval är betyd- ligt större än vad man tidigare ansett. Ett antal aktuella studier har sålunda påvisat individuell partnerbedömning med ledning av kemiska signaler (se tabell 1). Andra studier har vidare avslöjat en omfattande individuell varia- tion i feromonproduktion (t.ex Schlyter & Birgersson, 1989; Buesching et al., 2002) och forskare har också börjat studera feromonproduktionens kost- nader (t.ex. Bushmann & Atema, 2000; Nahon et al., 1995).

34 Feromoners innebörd En sexuell signal måste förmedla någon form av förmån till mottagaren (Andersson, 1994). Fördelen med partnerval är lätt att uppfatta i parningssy- stem där individer erhåller omedelbara förmåner genom sitt val av partner (såsom t.ex. bröllopsgåvor, territorier eller äggläggningsplatser). Hos arter där kräsna individer inte erhåller något mer än det genetiska material nöd- vändigt för att producera avkomma, såsom t.ex. hos spelande arter eller vid utomparsparningar, är partnervalets orsaker mindre uppenbara (Höglund & Alatalo, 1995). De traditionella modellerna för att förklara evolutionen av partnerval undersöker konsekvenserna av partnerpreferenser genom de di- rekta eller indirekta förmåner en individ erhåller genom sitt val (Andersson, 1994). Dessa förmåner, kommunicerade genom ett brett spektrum av sinnes- förmögenheter, utgör den sexuella signalens ’mening’. I artikel I granskar vi beläggen för feromoner som indikatorer på partner- kvalitet och deras användning vid partnerbedömning. Vi konstaterar att fe- romoner, i och med att de består av materia, i vissa fall kan utgöra en resurs i sig. Många insekter vilka använder växtsubstanser som försvar mot predato- rer utnyttjar således dessa substanser också som feromoner (Gullan & Crans- ton, 1994; Landolt & Phillips, 1997). Till exempel samlar hanar av flera olika fjärilsfamiljer (Boppre, 1990; Dussourd et al., 1991) in växtsubstanser, av vilka en del avskiljs till spermatoforer, medan någon del avges som fero- mon. Honor föredrar att para sig med hanar som avger mycket feromon, och signalens styrka avspeglar den mängd substans som honorna får sig till del genom att ta emot spermatoforen. I de flesta fall är dock feromonets innebörd mera svårtolkad, och man får övergå till indirekta förmåner för att förklara signalens funktion. Dessa mo- deller bygger i de flesta fall på antagandet att signalen är kostsam att produ- cera. Om en signal är dyr för sändaren, bör dess styrka eller kvalitet avspegla sändarens fenotypiska och eventuellt också genotypiska kvalitet. Kräsna individer vinner alltså en ökad fitness hos sin avkomma genom sitt partner- val. Det finns ett antal studier vilka demonstrerar sådana ’bra gener’- förmåner vunna genom feromonbaserat partnerval (tabell 1). Den mekanistiska länken mellan den kemiska signalen och den kvalitet den anses avspegla är dock i de flesta fall dåligt känd, bl.a. eftersom få för- sök att uppskatta feromonets kostnad har gjorts. Feromonstudier förefaller överhuvudtaget lida av att kemisk ekologi så sällan förenas med beteende- ekologi. Förutom kemin är kunskap om den individuell variationen, ärftlig- heten och kostnaden för feromoner instrumentella för en djupare förståelse av feromoner som partnerkvalitetssignaler. Denna avhandling är ett försök att undersöka dessa tre faktorer genom att använda ett hanligt feromon hos fruktflugan Drosophila grimshawi.

35 Studieart och metoder Drosophila grimshawi är en av de mer än 600 arter av fruktflugor vilka är endemiska för ögruppen Hawaii. Den har ett spelbaserat parningssystem i vilket hanar samlas på blad eller trädstammar för att konkurrera om, och uppvakta, honor. Honor besöker spelen enbart för att para sig eftersom föda och äggläggningsplatser finns på annat håll (Spieth, 1986; Droney, 1990). Hanarna har en bred repertoar av taktila, visuella, kemiska och förmodligen också akustiska signaler vilka används i interaktioner med honor och andra hanar (Ringo & Hodosh, 1978; Spieth, 1986; Hoikkala et al., 1989). Under spelets gång stryker hanarna upprepade gånger bakkroppsspetsen mot under- laget, varvid en klibbig substans avsätts. Dessa avsättningar formar sig till distinkta spår, ett par mm långa, vilka avger ett feromon som attraherar båda könen till spelplatsen (Droney, 1994). Det antal spår en hane producerar korrelerar, tillsammans med hans allmänna uppvaktningsintensitet, med hans parningsframgång (Droney & Hock, 1998). I motsats till många andra spe- lande arter försvarar hanar av D. grimshawi inga revir (Droney, 1992) och feromonspåren kan vara spridda över hela spelplatsen. Feromonets kemiska struktur är inte känd, men eftersom hanarnas sekret formar distinkta spår vilka är synliga under flera månader förefaller det troligt att dess mera flyk- tiga fraktioner konserveras i en matris av någon mindre flyktig substans. För att uppskatta hanarnas feromonproduktion klädde jag insidorna av provrör med hoprullade, rektangulära stycken genomskinlig plast (900 × 850 mm). Ensamma hanar placerades i provrören och lämnades att avsätta fero- mon under 24 timmar. Plastfilmerna fotograferades därefter i en scanner, och de resulterande digitala bilderna vidarebehandlades i bildhanteringspro- grammet ImageJ. I och med att feromonspåren var ljusare än bakgrunden kunde jag uppskatta den area de täckte genom att mäta alla pixlar ljusare än ett visst tröskelvärde.

Resultat och diskussion De försök som presenteras i denna avhandling tyder på att det hanliga fero- monet hos D. grimshawi har flera betydelser, och att dessa betydelser varie- rar med social kontext. Från artikel V har vi således indikationer på att hanar avsätter feromon under en större andel av sitt liv när de är ensamma i jämfö- relse med om de har sällskap av honor. I jämförelse med hanar som har säll- skap av andra hanar ägnar de däremot mindre tid till feromonavsättning i proportion till sin livslängd. Dessa ensamma hanar förefaller vidare påbörja feromonavsättningen något tidigare än hanarna i de övriga två behandlingar- na. Sammantaget tyder detta på en ’anropande’, d.v.s. art- /partnerigenkänningsfunktion hos feromonet. Eftersom sådana signaler är trovärdiga genom sin blotta närvaro behöver de inte vara dyra att producera.

36 Därmed behöver inte signalfrekvensen höjas så länge som den tröskelnivå som krävs för attraktion överskrids. Att mängden feromon hade så svag ef- fekt på attraktion över korta distanser i artikel II verkar också bekräfta denna tolkning. När det sociala sammanhanget förändras från en anropande hane till två spelande förefaller emellertid dessa båda öka sin investering i feromonpro- duktion tills denna överskrider ett kritiskt värde och feromonet blir en kost- sam signal. Dessutom antyder den större feromonavsättning hos spelande hanar jämfört med uppvaktande hanar som vi påvisade i artikel V att fero- monets evolution drivs av hanlig konkurrens snarare än direkt honligt part- nerval. Även om honor föredrar att para sig med hanar som avsätter mycket feromon och investerar mest i uppvaktning (Droney & Hock, 1998), avsätter uppvaktande hanar mindre tid till feromonavsättning än spelande (artikel V). Denna tolkning av feromonets funktion får ytterligare stöd av parnings- försöket i artikel IV. I detta försök hade stora hanar ett parningsförsteg me- dan hanar som avsatte mycket feromon inte hade större parningsframgång. Givet att honor kontrollerar parningen hos D. grimshawi (Droney & Hock, 1998) förefaller de alltså föredra stora hanar men, i bästa fall, ignorera mängden feromon de avsätter. Det kan emellertid vara nödvändigt för en hane att avsätta åtminstone en mindre mängd feromon för att attrahera en hona, eftersom en större andel hanar föreföll avsätta feromon i sällskap med honor jämfört med om de hade sällskap av hanar eller var ensamma (artikel V). Det förefaller således som om feromonet, med avseende på honor, fun- gerar som en partnerigenkänningssignal mer än något annat. Det kan emel- lertid inte uteslutas att honor reagerar på kvalitativa skillnader i feromonet snarare än de kvantitativa som vi laborerat med i dessa försök. I artikel III föreslår vi en funktion för feromonet i han-han interaktioner. Givet att hanar kan skilja mellan feromonavsättningarna från flera rivaler, vilket förefaller vara fallet, kan de bedöma spelets storlek liksom även riva- lernas konkurrensförmåga, information som bör användbar för alla hanar för att optimera det sexuella beteendet. Flera bevislinjer antyder att denna opti- mering kanaliseras genom två distinkta parningsstrategier. I artikel IV fann vi att stora hanar avsatte mindre feromon än små. I en annan studie jämförde Droney och Hock (1998) antalet feromonspår med uppvaktningsfrekvensen, och fann att det föreföll finnas två sexuella prestationsnivåer: hanar som uppvaktade intensivt och avsatte mycket feromon, och hanar som bara upp- vaktade intensivt. Trots den negativa korrelation mellan hanlig kroppsstorlek och feromon- avsättning som rapporteras i artikel IV finns det belägg för att båda karaktä- rerna är beroende av hanlig kondition. Långlivade hanar är till exempel både större (Droney, 1990) och avsätter feromon under en större andel av sitt liv än kortlivade hanar (artikel V). Hanar uppfödda på lågkvalitativ föda avsät- ter mindre feromon än hanar som fått bättre kost (Droney & Hock, 1998). Honorna i artikel IV producerade vidare mer avkomma när de parades med

37 hanar som var stora och avsatte mycket feromon. Sammantaget tyder detta på en avvägning mellan kroppsstorlek och feromonproduktion. Hanar i god kondition allokerar sina resurser till antingen kroppsstorlek eller feromon- produktion, det förra en fördel vid uppvaktning och det senare vid han-han konkurrens. Variation i feromonproduktion kan därmed upprätthållas genom frekvensberoende selektion av hanar. Merparten av Hawaiis fruktflugor har inte gemensamma spel utan solitära uppträdanden under vilka de använder feromon för att attrahera partners (Spieth, 1986). Den kemiska signalens ursprungliga funktion i detta taxa är således troligen art-/partnerigenkänning. Denna funktion kan ha modifierats när gemensamma spel evolverade hos D. grimshawi och dess systerarter. Indikationer på att hanar kan parasitera på rivalers feromonavsättning från denna och andra studier (Droney, 1994; Droney & Hock, 1998) tyder på att en ’hotshot’-mekanism kan vara orsaken till denna förändring. ’Hotshot’- modellen föreslår att spel formas genom att oattraktiva hanar samlas kring attraktiva för att få tillgång till honor (Beehler & Foster, 1988). Feromonets multifunktionalitet kan således vara ett resultat av D. grimshawis evolutionä- ra historia, under vilken nya funktioner har adderats utan att gamla förlorats.

38 Acknowledgments

No scientific, or indeed personal, achievements are possible without the sup- port from others. And when it comes to this work, the roots run deep to many people that have pushed, influenced and helped me during many years. I can only hope that you appreciate the result of your efforts! First, I would like to thank the triumvirate of supervisors that I have ac- cumulated during my years at the department of Animal Ecology: Fredrik Widemo who took me on as a PhD-student; Göran Arnqvist who guided me through the last years, and Anders Berglund for being always jovial and a saving grace in foreign lands. None the less important as a discussion partner and idea-giver has Therésa Jones of Melbourne University been. Moreover, she set up our rearing lab and brought home the wild stock from the Hawai- ian wilderness. We stand on some giants’ shoulders more than others. In the case of D. grimshawi-studies, these are Kenneth Kaneshiro of the University of Hawaii and David Droney of Hobart & William Smith Colleges. Kenneth Kaneshiro kindly provided the lab stock and gave advice on rearing techniques. David Droney gave further advice on rearing and useful tips on the bioassays. Stefan Gunnarsson of the EBC Microscopy Unit untiringly helped and in- structed when I was experimenting with the imaging techniques. The history of most research projects is, I suspect, full of experiments that did not work. Thus, I spent a couple of nice weeks under the aegis of Christer Löfstedt at the Pheromone group of Lund University, trying to get somewhere with the pheromonal chemistry. Likewise, Bo G. Svensson helped me dissect the flies when I was looking for the anal glands. Olle Håstad and Anders Ödeen and the cool gadgets of theirs convinced me that the pheromone streaks have nothing whatsoever to do with UV-light. Jens Olsson, Johanna Arrendal, Sonia Fontana and Ellen Hultman helped me in the arduous task of keeping the flies alive. I wish you good luck with your own projects! Ellen Hultman also assisted admirably with some of the experiments. At the department, Mats Björklund and Ingela Ericsson have always been very helpful and kept things running smooth in their respective roles as head of dep and administrator. My roommates Claudia Fricke and Niclas Kolm have made my office-days infinitely more fun apart from generously helping me with practical and scientific problems. And of course, a big thank you to

39 all the people of the department who made these four and a half years such a great time! Many people helped me to get to Uppsala. Thus, Arja Kaitala, then at Stockholm University, got me started on mate choice when I was a grad student. Olle Anderbrant at the Pheromone group, Lund University, intro- duced me to pheromones and generously let me write two papers when I was working for him after my graduation. Nils Uddenberg and Jens Rydell has been important role models for a very long time, showing me in rather dif- ferent ways what it might mean to be a scientist. Off course, the journey started with my parents, Inger and Stig, who has always encouraged my interest in nature, and that glorious NGO, Fältbiolo- gerna, where I learnt more about biology (not to mention a lot of other stuff) than I am likely to ever learn again. Petronella Eriksson worked with the formatting of the papers, assisted with some of the experiments and discussed the results, found a printer and, most important, gave me a never-ceasing flow of love.

40 References

Andersson, M. 1994. Sexual Selection. Princeton University Press. Archer, M. S. & Elgar, M. A. 1999. Female preference for multiple partners: sperm competition in the hide beetle, Dermestes maculatus (DeGeer). Animal Behav- iour, 58, 669-675. Arn, H., Toth, M. & Priesner, E. 1986. List of sex pheromones of lepidoptera and related attractants. Paris: OILB-SROP. Beehler, B. M. & Foster, M. S. 1988. Hotshots, hotspots, and female preference in the organization of lek mating systems. American Naturalist, 131, 203-219. Beeler, A. E., Rauter, C. M. & Moore, A. J. 2002. Mate discrimination by females in the burying beetle Nicrophorus orbicollis: the influence of male size on attrac- tiveness to females. Ecological Entomology, 27, 1-6. Boppre, M. 1990. Lepidoptera and Pyrrolizidine Alkaloids - Exemplification of Complexity in Chemical Ecology. Journal of Chemical Ecology, 16, 165-185. Bradbury, J. W. & Vehrencamp, S. L. 1998. Principles of Animal Communication. Sunderland: Sinauer Associates. Buesching, C. D., Waterhouse, J. S. & Macdonald, D. W. 2002. Gas- chromatographic analyses of the subcaudal gland secretion of the European badger (Meles meles) Part I: Chemical differences related to individual parame- ters. Journal of Chemical Ecology, 28, 41-56. Bushmann, P. J. & Atema, J. 2000. Chemically mediated mate location and evalua- tion ln the lobster, Homarus americanus. Journal of Chemical Ecology, 26, 883- 899. Butenandt, A., Beckman, R., Stamm, D. & Hecker, E. 1959. Über den Sexual Lock- stoff des Seidenspinners Bombyx mori. Reindarstelling und Konstitution. Zeitschrift für Naturforschung., 14b, 283-284. Cardé, R. T. & Baker, T. C. 1984. Sexual communication with pheromones. In: Chemical Ecology of Insects (Ed. by Bell, W. J. & Cardé, R. T.), pp. 355-383. London: Chapman & Hall. Clark, D. C., DeBano, S. J. & Moore, A. J. 1997. The influence of environmental quality on sexual selection in Nauphoeta cinerea (Dictyoptera:Blaberidae). Be- havioral Ecology, 8, 46-53. Cotton, S., Fowler, K. & Pomiankowski, A. 2004. Do sexual ornaments demonstrate heightened condition-dependent expression as predicted by the handicap hy- pothesis? Proceedings of the Royal Society of London Series B, 271, 771-783. Deutsch, J. C. & Nefdt, R. J. C. 1992. Olfactory cues influence female choice in two lek-breeding antelopes. Nature, 356, 396-397. Diaz, E. R. & Thiel, M. 2004. Chemical and visual communication during mate searching in rock shrimp. Biological Bulletin, 206, 134-143. Drickamer, L. C. 1992. Oestrous female house mice discriminate dominant from subordinate males and sons of dominant from sons of subordinate males by odour cues. Animal Behaviour, 43, 868-870.

41 Droney, D. C. 1990. Lek mating dispersion and mate choice in a Hawaiian droso- philid. PhD thesis, Syracuse University. Droney, D. C. 1992. Sexual selection in a lekking Hawaiian Drosophila: the roles of male competition and female choice in male mating success. Animal Behaviour, 44, 1007-1020. Droney, D. C. 1994. Tests of hypotheses for lek formation in a Hawaiian Droso- phila. Animal Behaviour, 47, 351-361. Droney, D. C. 1996. Environmental influences on male courtship and implications for female choice in a lekking Hawaiian Drosophila. Animal Behaviour, 51, 821-830. Droney, D. C. & Hock, M. B. 1998. Male sexual signals and female choice in Dro- sophila grimshawi (Diptera : Drosophilidae). Journal of Insect Behavior, 11, 59- 71. Dussourd, D. E., Harvis, C. A., Meinwald, J. & Eisner, T. 1989. Paternal allocation of sequestered plant Pyrrolizidine alkaloid to eggs in the Danaine butterfly, Danaus gilippus. Experientia, 45, 896-898. Dussourd, D. E., Harvis, C. A., Meinwald, J. & Eisner, T. 1991. Pheromonal adver- tisement of a nuptial gift by a male moth (Utetheisa ornatrix). Proceedings of the National Academy of Sciences of the United States of America, 88, 9224- 9227. Ehman, K. D. & Scott, M. E. 2001. Urinary odour preferences of MHC congenic female mice, Mus domesticus: implications for kin recognition and detection of parasitized males. Animal Behaviour, 62, 781-789. Eisner, T., Smedley, S. R., Young, D. K., Eisner, M., Roach, B. & Meinwald, J. 1996. Chemical basis of courtship in a beetle (Neopyrochroa flabellata): Can- tharidin as a precopulatory ''enticing'' agent. Proceedings of the National Acad- emy of Sciences of the United States of America, 93, 6494-6498. Falconer, D. S. & Mackay, T. F. C. 1996. Introduction to Quantitative Genetics. Harlow: Longman. Fisher, H. S., Swaisgood, R. R. & Fitch-Snyder, H. 2003. Counter-marking by male pygmy lorises (Nycticebus pygmaeus): do females use odor cues to select mates with high competitive ability? Behavioral Ecology and Sociobiology 53, 123- 130. Fisher, R. A. 1958. The Genetical Theory of Sexual Selection. Clarendon Press, Oxford. Garner, T. W. J. & Schmidt, B. R. 2003. Relatedness, body size and paternity in the alpine newt, Triturus alpestris. Proceedings of the Royal Society of London Se- ries B, 270, 619-624. Gosling, L. M. 1990. Scent marking by resource holders: alternative mechanisms for advertising the costs of competition. In: Chemical Signals in Vertebrates. (Ed. by Macdonald, D. W., Müller-Schwarze, D. & Natzynczuk, S. E.), pp. 315-328. Oxford University Press. Gosling, L. M., Atkinson, N. W., Dunn, S. & Collins, S. A. 1996. The response of subordinate male mice to scent marks varies in relation to their own competitive ability. Animal Behaviour, 52, 1185-1191. Gosling, L. M. & Roberts, S. C. 2001. Scent-marking by male mammals: Cheat- proof signals to competitors and mates. In: Advances in the Study of Behavior, Vol 30, pp. 169-217. San Diego: Academic Press Inc. Grafen, A. 1990. Biological signals as handicaps. Journal of Theoretical Biology, 144, 517-546. Gullan, P. J. & Cranston, P. S. 1994. The Insects: an Outline of Entomology. Lon- don: Chapman & Hall.

42 Hendrichs, M. A. & Hendrichs, J. 1998. Perfumed to be killed: Interception of Mediterranean fruit fly (Diptera : Tephritidae) sexual signaling by predatory foraging wasps (Hymenoptera : Vespidae). Annals of the Entomological Society of America, 91, 228-234. Hodosh, R. J., Keough, E. M. & Ringo, J. M. 1979. The morphology of the sex pheromone gland in Drosophila grimshawi. Journal of Morphology, 161, 177- 184. Hoikkala, A., Hoy, R. R. & Kaneshiro, K. Y. 1989. High-frequency clicks of Hawai- ian picture-winged Drosophila species. Animal Behaviour, 37, 927-934. Höglund, J. & Alatalo, R. V. 1995. Leks. Princeton: Princeton University Press. Iwasa, Y. & Pomiankowski, A. 1994. The evolution of mate preferences for multiple sexual ornaments. Evolution, 48, 853-867. Iyengar, V. K., Rossini, C. & Eisner, T. 2001. Precopulatory assessment of male quality in an arctiid moth (Utetheisa ornatrix): hydroxydanaidal is the only crite- rion of choice. Behavioral Ecology and Sociobiology, 49, 283-288. Jennions, M. D. & Oakes, E. J. 1994. Symmetry and sexual selection. Trends in Ecology & Evolution, 9, 440-440. Jones, T. M. & Hamilton, J. G. C. 1998. A role for pheromones in mate choice in a lekking sandfly. Animal Behaviour, 56, 891-898. Jones, T. M., Quinnell, R. J. & Balmford, A. 1998. Fisherian flies: benefits of fe- male choice in a lekking sandfly. Proceedings of the Royal Society of London Series B-Biological Sciences, 265, 1651-1657. Jones, T. M. & Widemo, F. In press. Survival and reproduction when food is scarce: implications for a lekking Hawaiian Drosophila. Ecological Entomology Karlson, P. & Lüscher, M. 1959. Pheromone, ein Nomeklatur vorschlag für ein Wirkstoffklasse. Naturwissenschaffen, 46, 63-64. Kotiaho, J. S. 2000. Testing the assumptions of conditional handicap theory: costs and condition dependence of a sexually selected trait. Behavioral Ecology and Sociobiology, 48, 188-194. Kotiaho, J. S. 2001. Costs of sexual traits: a mismatch between theoretical consid- erations and empirical evidence. Biological Reviews, 76, 365-376. Lande, R. 1981. Models of speciation by sexual selection on polygenic traits. Pro- ceedings of the National Academy of Sciences of the United States of America, 78, 3721-3725. Landolt, P. J. & Phillips, T. W. 1997. Host plant influences on sex pheromone be- havior of phytophagous insects. Annual Review of Entomology, 42, 371-391. Larsson, M. C., Hedin, J., Svensson, G. P., Tolasch, T. & Francke, W. 2003. Charac- teristic odor of Osmoderma eremita identified as a male-released pheromone. Journal of Chemical Ecology, 29, 575-587. Lemaster, M. P. & Mason, R. T. 2002. Variation in a female sexual attractiveness pheromone controls male mate choice in garter snakes. Journal of Chemical Ecology, 28, 1269-1285. Lessels, C. M. & Boag, P. T. 1987. Unrepeatable repeatabilities: a common mistake. Auk, 104, 116-212. Lewis, S. M. & Austad, S. N. 1994. Sexual selection in flour beetles: the relation- ship between sperm precedence and male olfactory attractiveness. Behavioral Ecology, 5, 219-224. Löfstedt, C. 1993. Moth pheromone genetics and evolution. Philosophical Transac- tions of the Royal Society of London Series B-Biological Sciences, 340, 167- 177. Madsen, T., Shine, R., Loman, J. & Hakansson, T. 1992. Why do female adders copulate so frequently? Nature, 355, 440-441.

43 Marco, A., Chivers, D. P., Kiesecker, J. M. & Blaustein, A. R. 1998. Mate choice by chemical cues in western redback (Plethodon vehiculum) and Dunn's (P. dunni) salamanders. Ethology, 104, 781-788. Martin, J. & Lopez, P. 2000. Chemoreception, symmetry and mate choice in lizards. Proceedings of the Royal Society of London Series B-Biological Sciences, 267, 1265-1269. Mays Jr, H. L. & Hill, G. E. 2004. Choosing mates: good genes versus genes that are a good fit. Trends in Ecology & Evolution, 19, 554-559. Metcalf, R. L. 1998. Ultramicrochemistry of insect semiochemicals. Mikrochimica Acta, 129, 167-180. Milinski, M. 2003. The function of mate choice in sticklebacks: optimizing MHC genetics. Journal of Fish Biology, 63, 1-16. Moore, A. J. 1994. Genetic evidence for the good genes process of sexual selection. Behavioral Ecology and Sociobiology, 35, 235-241. Moore, A. J., and & Breed, M. D. 1986. Mate assessment in a cockroach, Nauphoeta cinerea. Animal Behaviour, 34, 1160-1165. Moore, A. J. & Moore, P. J. 1999. Balancing sexual selection through opposing mate choice and male competition. Proceedings of the Royal Society of London Series B-Biological Sciences, 266, 711-716. Møller, A. P. & de Lope, F. 1994. Differential costs of a secondary sexual character - an experimental test of the handicap principle. Evolution, 48, 1676-1683. Nahon, E., Atzmony, D., Zahavi, A. & Granot, D. 1995. Mate selection in yeast - a reconsideration of the signals and the message encoded by them. Journal of Theoretical Biology, 172, 315-322. Pagel, M. 1993. Honest signaling among gametes. Nature, 363, 539-541. Palmer, A. R. & Strobeck, C. 1986. Fluctuating assymetry: measurement, analysis, patterns. Annual Reveiw of Ecological Systems, 17, 391-421. Paterson, H. E. H. 1985. The recognition concept of species. In: Species and speci- ation (Ed. by Vrba, E. S.), pp. 21-29. Pretoria: Transvaal Museum Monograph. Penn, D. & Potts, W. K. 1998. Untrained mice discriminate MHC-determined odors. Physiology & Behavior, 64, 235-243. Ptacek, M. B. 2000. The role of mating preferences in shaping interspecific diver- gence in mating signals in vertebrates. Behavioural Processes, 51, 111-134. Rantala, M. J., Jokinen, I., Kortet, R., Vainikka, A. & Suhonen, J. 2002. Do phero- mones reveal male immunocompetence? Proceedings of the Royal Society of London Series B-Biological Sciences, 269, 1681-1685. Rantala, M. J., Kortet, R., Kotiaho, J. S., Vainikka, A. & Suhonen, J. 2003. Condi- tion dependence of pheromones and immune function in the grain beetle Tene- brio molitor. Functional Ecology, 17, 534-540. Reusch, T. B. H., Haberli, M. A., Aeschlimann, P. B. & Milinski, M. 2001. Female sticklebacks count alleles in a strategy of sexual selection explaining MHC polymorphism. Nature, 414, 300-302. Rich, T. J. & Hurst, J. L. 1998. Scent marks as reliable signals of the competitive ability of mates. Animal Behaviour, 56, 727-735. Ringo, J. M. & Hodosh, R. J. 1978. A multivariate analysis of behavioral divergence among closely related species of endemic Hawaiian Drosophila. Evolution, 32, 389-397. Roelofs, W. L. & Cardé, R. T. 1974. Sex pheromones in the reproductive isolation of lepidopterous species. In: Pheromones (Ed. by Birch, M. C.), pp. 96-114. Amsterdam: North-Holland Publishing Company. Roff, D. A. 1996. The evolution of genetic correlations: An analysis of patterns. Evolution, 50, 1392-1403.

44 Rowe, L. & Houle, D. 1996. The lek paradox and the capture of genetic variance by condition dependent traits. Proceedings of the Royal Society of London Series B-Biological Sciences, 263, 1415-1421. Schlyter, F. & Birgersson, G. 1989. Individual variation in bark beetle and moth pheromones - a comparison and an evolutionary background. Holarctic Ecol- ogy, 12, 457-465. Shohet, A. J. & Watt, P. J. 2004. Female association preferences based on olfactory cues in the guppy, Poecilia reticulata. Behavioral Ecology and Sociobiology, 55, 363-369. Spieth, H. T. 1986. Courtship Behaviors of the Hawaiian Picture-Winged Droso- phila. Berkeley: University of California Press. Stevens, I. D. R. 1998. Chemical aspects of pheromones. In: Insect pheromones and their use in pest management (Ed. by Howse, P. E., Stevens, I. D. R. & Jones, O. T.), pp. 133-260. London: Chapman & Hall. Svensson, M. 1996. Sexual selection in moths: The role of chemical communication. Biological Reviews of the Cambridge Philosophical Society, 71, 113-135. Teale, S. A., Hager, B. J. & Webster, F. X. 1994. Pheromone-based assortative mat- ing in a bark beetle. Animal Behaviour, 48, 569-578. Thornhill, R. 1992. Female preference for the pheromone of males with low fluctu- ating asymmetry in the Japanese scorpionfly (Panorpa japonica, Mecoptera). Behavioral Ecology, 3, 277-283. Thornhill, R. & Gangestad, S. W. 1999. The scent of symmetry: A human sex pheromone that signals fitness? Evolution and Human Behavior, 20, 175-201. Thornhill, R., Gangestad, S. W., Miller, R., Scheyd, G., McCollough, J. K. & Frank- lin, M. 2003. Major histocompatibility complex genes, symmetry, and body scent attractiveness in men and women. Behavioral Ecology, 14, 668-678. Tompkins, L., McRobert, S. P. & Kaneshiro, K. Y. 1993. Chemical communication in Hawaiian Drosophila. Evolution, 47, 1407-1419. Tregenza, T. & Wedell, N. 2000. Genetic compatibility, mate choice and patterns of parentage. Molecular Ecology, 9, 1013-1027. Trivers, R. 1972. Parental investment and sexual selection. In: Sexual Selection and the Descent of Man (Ed. by Campbell, B. G.), pp. 136-179. Chicago: Aldine Press. Wedekind, C. & Furi, S. 1997. Body odour preferences in men and women: do they aim for specific MHC combinations or simply heterozygosity? Proceedings of the Royal Society of London Series B-Biological Sciences, 264, 1471-1479. Widemo, F. 1997. The social implications of traditional use of lek sites in the ruff Philomachus pugnax. Behavioral Ecology, 8, 211-217. Widemo, F. & Owens, I. P. F. 1995. Lek size, male mating skew and the evolution of lekking. Nature, 373, 148-151. Widemo, F. & Owens, I. P. F. 1999. Size and stability of vertebrate leks. Animal Behaviour, 58, 1217-1221. Wiley, R. H. & Poston, J. 1996. Perspective: Indirect mate choice, competition for mates, and coevolution of the sexes. Evolution, 50, 1371-1381. Williams, G. C. 1966. Adaptation and Natural Selection: A Critique of Some Cur- rent Evolutionary Thought. Princeton University Press. Vonnickischrosenegk, E. & Wink, M. 1994. Influence of previous feeding regimes and ambient-temperatures on degradation and storage of pyrrolizidine alkaloids in the moth species Creatonotos transiens (Lepidoptera, Arctiidae). Entomolo- gia Generalis, 19, 157-170. Worden, B. D., Parker, P. C. & Pappas, P. W. 2000. Parasites reduce attractiveness and reproductive success in male grain beetles. Animal Behaviour, 59, 543-550.

45 Wyatt, T. D. 2003. Pheromones and Animal Behaviour: Communication by Smell and Taste. Cambridge University Press. Zahavi, A. 1975. Mate selection - a selection for a handicap. Journal of Theoretical Biology, 53, 205-214. Zala, S. M., Potts, W. K. & Penn, D. J. 2004. Scent-marking displays provide honest signals of health and infection. Behavioral Ecology, 15, 338-344. Zeh, J. A., Newcomer, S. D. & Zeh, D. W. 1998. Polyandrous females discriminate against previous mates. Proceedings of the National Academy of Sciences of the United States of America, 95, 13732-13736.

46

  , *&)   ) ,-) ) )                    

. &' / 0  1  2  3 1     /  43

  &  ))&   1&  2  3 1     /  435 ,--) , *&) 35 ) ) 3 ) &3 1  (& 1 - -&)6  17 - ) 1   -   ))&   & 8-  9& 7 ) &) &  (& & )5 7   ) &3  )  )& ( &   3 &4  )& ) 0 4   -&) *  & ) 1 ,--) 0 ))&  ) 1&  2  3 1     /  436 :;& &    &35 ++ 5  )& ) 7 ) -( ) &     < -&) *  & ) 1 ,--) 0 ))&  ) 1&  2  3 1     /  43<6=

      0 )& ( ' -(  )66)   &'(')'' * !"$"+