Reproductive interference determines persistence and exclusion in species interactions


  • Shigeki Kishi,

    Corresponding author
    1. Center for Ecological Research, Kyoto University, 520-2113, 2-509-3, Hirano, Otsu, Shiga, Japan
      Correspondence author. E-mail:
    Search for more papers by this author
  • Takayoshi Nishida,

    1. Laboratory of Insect Ecology, Graduate School of Agriculture, Kyoto University, 606-8502, Kitashirakawa-Oiwake cho, Sakyo, Kyoto, Japan
    Search for more papers by this author
  • Yoshitaka Tsubaki

    1. Center for Ecological Research, Kyoto University, 520-2113, 2-509-3, Hirano, Otsu, Shiga, Japan
    Search for more papers by this author

Correspondence author. E-mail:


1. Reproductive interference is a negative interspecific sexual interaction that adversely affects the fitness of males and females during reproductive process. Theoretical studies suggest that because reproductive interference is characterized by positive frequency dependence it is far more likely to cause species exclusion than the density dependence of resource competition. However, the respective contributions of resource competition and reproductive interference to species exclusion, which have been frequently observed in many competition studies, remain unclear.

2. We show that reproductive interference is a far more critical cause of species exclusion than resource competition in the competition between Callosobruchus bean weevil species. In competition experiments over several generations, we manipulated the initial relative abundance of the adzuki bean beetle, Callosobruchus chinensis, and the southern cowpea beetle, Callosobruchus maculatus. When the initial adult ratio of C. chinensis : C. maculatus were 6 : 2 and 4 : 4, C. chinensis excluded C. maculatus. However, when C. maculatus was four times more abundant than C. chinensis at the start, we observed the opposite outcome.

3. A behavioural experiment using adults of the two species revealed asymmetric reproductive interference. The fecundity and longevity of C. maculatus females, but not those of C. chinensis females, decreased when the females were kept with heterospecific males. Fecundities of females of both species decreased as the number of heterospecific males increased. In contrast, resource competition at the larval stage resulted in higher survival of C. maculatus than of C. chinensis.

4. These results suggest that the positive frequency-dependent effect of reproductive interference resulted in species exclusion, depending on the initial population ratio of the two species, and the asymmetry of the interference resulted in C. chinensis being dominant in this study, as in previous studies. Classical competition studies should be reviewed in light of this evidence for reproductive interference.


The effect of interspecific sexual interaction on population dynamics has long been overlooked, because many scientists have viewed species as discrete units of reproduction and have simply ignored the possibility that sexual interactions occur particularly between species pairs that never hybridize. Reproductive interference, which is a consequence of incomplete species recognition in the reproductive process, is defined as any kind of interspecific sexual interaction that reduces the fitness of individuals (Bull 1991; Kuno 1992; Gröning & Hochkirch 2008). Reproductive interference can occur at various reproductive stages between closely related species, and sometimes between distantly related species (Chow-Fraser & Maly 1988). Many types of the fitness costs have been reported, including sexual harassment (McLain & Pratt 1999), reduced opportunity to mate with conspecifics (Kiritani, Hokyo & Yukawa 1963; Hochkirch, Gröning & Bücker 2007), damage to the female genitalia (Kubota & Sota 1998), gamete dumping (Kiritani et al. 1963; Lessios & Cunningham 1990) and hybridization (Sota & Kubota 1998; Garcia-Vazquez et al. 2002; Pfennig 2007). Recently, increasing numbers of studies have documented reproductive interference between closely related species of various taxa (reviewed by Gröning & Hochkirch 2008). For example, several studies have described habitat partitioning (Panonychus, spider mites, Takafuji, Kuno & Fujimoto 1997; Skistodiaptomus, copepods, Thum 2007; Tetrix, pigmy grass hoppers, Gröning et al. 2007) and sexual displacement (Hemidactylus, geckos, Dame & Petren 2006; Bemisia, whiteflies, Liu et al. 2007) in animals. In plants, interspecific pollen transfer has been suggested to cause rapid exclusion of a reproductively inferior species (Waser 1978; Campbell 1986). In fact, interspecific pollen transfer often negatively affects seed set in several species (Polemonium, alpine wildflowers, Galen & Gregory 1989; Lythrum, loosestrife, Brown & Mitchell 2001; Taraxacum, dandelions, Takakura et al. 2008). Despite the potential importance of reproductive interference as a mechanism of species exclusion (Kuno 1992; Reitz & Trumble 2002), reproductive interference has not been considered in previous laboratory competition studies, which have mostly reported competitive exclusion, rather than coexistence, to be the usual outcome (e.g. Drosophila, fruit flies, Merrell 1951; Tribolium, flour beetles, Park 1948; and Callosobruchus, bean weevils, Utida 1953). With their focus on resource competition, these studies conclude that the dominant species utilizes resources more efficiently, and that the resource competition results in the exclusion of the subordinate species (Tilman 1982).

The effect of reproductive interference on population dynamics definitely differs from that of resource competition. Some mathematical studies have predicted that reproductive interference should cause ‘sexual exclusion’ (Gröning & Hochkirch 2008) much more readily than resource competition causes competitive exclusion, because the per capita cost of reproductive interference increases as the relative frequency of heterospecific individuals increases (Ribeiro & Spielman 1986; Kuno 1992; Yoshimura & Clark 1994). This positive frequency-dependent effect of reproductive interference is therefore predicted to accelerate extinction of the species with lower population density if reproductive interference is symmetric between the species. Therefore, it is highly probable that the survival or extinction between the two competing species depends critically on the ratio of their initial population densities (Kuno 1992; Yoshimura & Clark 1994). If reproductive interference is asymmetric, however, the model predicts survival of the species less affected by reproductive interference (Kishi et al., unpublished data). In fact, empirical studies have shown that reproductive interference is generally asymmetric (reviewed by Gröning & Hochkirch 2008).

The species exclusion observed in many previous studies of competition between closely related species may be attributable to reproductive interference. However, the major concern of the previous laboratory competition studies has been to determine whether resource competition resulted in coexistence, or not (Ayala 1971). As a result, no experimental study has examined whether resource competition or reproductive interference is the main determinant of the competition outcome (Gröning & Hochkirch 2008; Takakura et al. 2008). Furthermore, few studies have examined the positive frequency-dependence of reproductive interference (but see Hochkirch et al. 2007; Takakura et al. 2008).

Interspecific competition between the adzuki bean weevil, Callosobruchus chinensis L. and the cowpea weevil, Callosobruchus maculatus F., has been described as a typical example of interspecific competition (e.g. Utida 1953). Adults of both species lay eggs on the surface of postharvest beans of the genus Vigna, and the hatched larvae grow to adults within the bean, feeding on it from the inside. Several studies have confirmed that C. maculatus larvae are superior competitors to C. chinensis larvae within a bean (Yoshida 1966; Fujii 1967; Bellows & Hassell 1984; Ishii & Shimada 2008). However, the outcome of the interspecific competition over generations has been confounding, because most studies have documented competitive exclusion of C. maculatus by C. chinensis within several generations (Yoshida 1966; Fujii 1967; Bellows & Hassell 1984), although the opposite results have also been reported (Utida 1953; Ishii & Shimada 2008). Therefore, it seems unlikely that superiority in larval resource competition determines the dominant species, but the mechanism determining the competition outcome is still controversial, despite many competition experiments (Ishii & Shimada 2008).

We hypothesized that reproductive interference can determine the dominant species in species interactions between these two Callosobruchus weevils, even though the dominant species is somewhat inferior to the vulnerable species in resource competition. We conducted a series of experiments to test our hypothesis. First, we carried out an interspecific competition experiment that started with different numbers of adults of each species and observed the outcomes after several generations. Then, we investigated reproductive interference at the adult stage and resource competition at the larval stage in the two species. By integrating our results, we judged whether the competition outcome between these two weevils was consistent with asymmetric reproductive interference, or with asymmetric resource competition.

Material and methods

Two bean weevil species, C. chinensis and C. maculatus, are both worldwide pests that infest postharvest beans of the genus Vigna (Fabaceae) (Tuda et al. 2006). Both species spend c. 3 weeks from egg to adult emergence. If mated females of both species do not feed, then they lay 60–80 eggs on the surface of beans during a life span of c. 1 week. Many strains of C. chinensis (Harano & Miyatake 2005) and C. maculatus (Toquenaga & Fujii 1991) are kept in several institutes and laboratories. We used the jC strain of C. chinensis, which has been maintained for more than 70 years at Kyoto University (Utida 1953; Kuno 1995), and the hQ strain of C. maculatus, a gift from Okayama University (Miyatake & Matsumura 2004). All experiments were carried out under fixed conditions (temperature, 30 °C; humidity, 70%; 16 hours light : 8 hours dark regime). No hybrids have been reported between these two species.

Interspecific competition in a Petri dish

We observed interspecific competition between C. chinensis and C. maculatus in semi-continuous generation system (Ishii & Shimada 2008). Competition experiments using discrete generations, as carried out by most studies (Utida 1953; Fujii 1965; Yoshida 1966; Bellows & Hassell 1984), encounter a problem caused by the different developmental periods of the two species (Bellows & Hassell 1984). The shorter egg oviposition interval between generations of C. chinensis gives that species the advantage that it is able to oviposit more eggs, whereas the longer interval of C. maculatus gives an advantage to the later emerging adults of that species, because the fecundity of C. chinensis deteriorates before that of C. maculatus (Fujii 1967; Bellows & Hassell 1984). Therefore, to overcome this problem, we set up competition experiments that continued through continuous generations (Ishii & Shimada 2008). We used four-compartment Petri dishes (100 × 15 mm; BD Falcon, Mississauga, ON, Canada). In each dish, the space between the partition tops and the cover is large enough for adult beetles to move freely across compartments, though larvae remain in the individual compartments. The four sections of each Petri dish were numbered clockwise from 1 to 4. Adult pairs of the two species and 5·0 g of adzuki beans, Vigna angularis, were introduced into sections 1, 2 and 3 on the first, the eighth and 15th day respectively. Because adults of all sections were allowed to mix, this setup created continuous generations rather than discrete cohorts. We carried out four different treatments, in each of which different numbers of adult pairs of each species were introduced: the ratio of adult pairs of C. chinensis to those of C. maculatus were 6 : 2, 4 : 4, 2 : 6 or 2 : 8 in the four treatments. We performed three replicates of each treatment. On the 22nd day, only 5·0 g of adzuki beans and no adult weevils were introduced into section 4. On the same day, after lightly anaesthetizing the emerged weevils from section 1 with diethyl ether, we also sieved out all individuals from the Petri dish, and counted the numbers of dead and live adults. Dead individuals were discarded, and living adults were put back into the Petri dish. On the 29th day, we removed the exhausted beans from compartment 1 and replenished them with 5·0 g of new beans, because the adults that had emerged from the beans in compartment 1 had laid most of their eggs on the adzuki beans in compartment 4 for 1 week (Yoshida 1966; Ishii & Shimada 2008). We again counted the number of dead and living adults, and put the living adults back into the Petri dish. Every seventh day thereafter, we repeated these procedures until no living adult of either species had been observed for four consecutive weeks. The number of adults that emerged per week was calculated according to Utida (1941).

Promiscuity of male mating attempts

We examined to what degree males of both species discriminated between conspecific and heterospecific females in terms of the number of mating attempts. First, we obtained virgin males and virgin females every day by collecting newly emerged adults, each of which was reared individually in a well of a 24-well plastic plate. We allowed the virgin females to mate once with conspecific males in small Petri dishes (60 × 15 mm; BD Falcon). Using mated females, we aimed to make male mating attempts traumatic for females (Crudgington & Siva-Jothy 2000; Yanagi & Miyatake 2003), and to avoid mating during the experiment. Then, we arbitrarily selected one-mated female of C. chinensis and one-mated female of C. maculatus and put them both together with two virgin males of C. chinensis or C. maculatus in a small Petri dish. We recorded the total number of male mating attempts in each Petri dish for 30 min. Ten replicates of each treatment were carried out. We compared the number of male mating attempts between female species by using the paired Wilcoxon test. All statistical analyses were performed with r software version 2.6.0 (R Development Core Team 2007).

Fitness costs of reproductive interference

To quantify the fitness costs to females of reproductive interference, we compared the fecundity (the number of eggs laid during the first 3 days) and the longevity (survival period in days) of females when heterospecific males, conspecific males or no males were present. First, virgin females of C. chinensis and C. maculatus, within 24 hours after emergence, were allowed to mate once with conspecific males by putting one female and one male together in a small mating arena (ϕ15 × 30 mm). Then, each of the mated females was put into a small Petri dish (60 × 15 mm; BD Falcon), along with a conspecific male, a heterospecific male or no male, together with 10 adzuki beans as oviposition substrates. Each male treatment was replicated for 10 females of each species. We recorded the fecundity of the females after 3 days and thereafter measured the female longevity. Next, to demonstrate the frequency-dependence of reproductive interference, we examined the enhanced costs of reproductive interference as the number of heterospecific males increased. We measured the fecundity of females that had mated once with a conspecific male when those females were placed with 0, 1, 2 or 3 heterospecific males for 3 days, but without conspecific males, in a way similar to that described above. Each treatment was replicated for 10 females of each species. We used two-way analysis of variance (two-way anova) to examine the effects of male species, female species and their interaction on the fecundity of females. Then we used anova to compare the fecundity between the male treatments within each species. We compared the longevity of females between the male treatments by using a log-rank survival test based on the Kaplan–Meier method. To examine the effect of increasing numbers of heterospecific males on the fecundity of females, we performed a linear regression analysis for each of the two species. Then, we examined the effects of male number, female species and their interaction on female fecundity by using analysis of covariance (ancova).

Larval resource competition

To confirm that C. maculatus larvae were superior to C. chinensis larvae in resource competition within a bean, as has been documented by most previous studies (Yoshida 1966; Fujii 1967; Bellows & Hassell 1984), we examined whether larvae could grow to adulthood together with heterospecific larvae in the same bean. First, we allowed more than 100 mated females of C. chinensis to lay eggs freely on the surface of more than 50 adzuki beans in a plastic container for 1 hour. Then, after removing beans with less than three eggs, we used a pen to encircle three eggs on each adzuki bean to mark them, and scraped the other eggs off with a paper cutter. We next provided more than 100 C. maculatus females with the marked adzuki beans (each with three circled C. chinensis eggs). After an oviposition period of 1 hour, we scraped off all but three C. maculatus eggs and the three circled C. chinensis eggs. Then, we chose 24 adzuki beans, each with three eggs of C. chinensis and three eggs of C. maculatus. In a similar way, but reversing the oviposition order, we prepared another set of 24 adzuki beans; on each of these first C. maculatus females and later C. chinensis females had laid three eggs. We put each treated bean into a single well of a 24-well plastic plate. The number of emerged adults was recorded by species for each bean. We performed a paired Wilcoxon test on the data from each 24-bean set.


Interspecific competition in a Petri dish

The interspecific competition always resulted in the exclusion of one species, usually in the extinction of C. maculatus (Fig. 1). When six pairs of C. chinensis and two pairs of C. maculatus were put together at the start, C. maculatus went rapidly extinct within 119 days (Fig. 1a–c). Similarly, C. chinensis eliminated C. maculatus within 168 days in all replicates of the treatment of C. chinensis : C. maculatus = 4 : 4 (Fig. 1d–f). In the treatment of C. chinensis : C. maculatus = 2 : 6, C. chinensis was the sole surviving species in two dishes, after 133 and 245 days, but C. maculatus was the surviving species in one dish after 301 days (Fig. 1g–i). In the treatment of C. chinensis : C. maculatus = 2 : 8, C. maculatus displaced C. chinensis within 189 days in all replicates (Fig. 1j–l).

Figure 1.

 Population dynamics of two competing bean weevil species. The number of emerged adults per week of Callosobruchus chinensis (solid lines) and Callosobruchus maculatus (dotted lines) is plotted for different initial numbers of adults: C. chinensis (C) : C. maculatus (M) = 6 : 2 (a–c), 4 : 4 (d–f), 2 : 6 (g–i) and 2 : 8 (j–l).

Promiscuity of male mating attempts

Callosobruchus chinensis males indiscriminately attempted to mate with C. chinensis females (2·9 ± 1·18 times; mean ± standard error) and C. maculatus females (2·5 ± 0·85) (paired Wilcoxon test: = 16·5, = 0·88). Similarly, C. maculatus males attempted to mate with both C. chinensis females (5·1 ± 2·18) and C. maculatus females (6·0 ± 2·14) (= 23, = 1·00). The number of male mating attempts was not significantly different among males of the two species (Wilcoxon test: = 158, = 0·25).

Fitness costs of reproductive interference

Female fecundity did not differ between the species, but depended significantly on the species of the male, and the interactive effect of the male and female species was also significant (two-way anova: male, F2,54 = 17·34, < 0·0001; female, F1,54 = 3·10, = 0·084; interaction, F2,54 = 5·49, = 0·0068; Fig. 2a). Thus, the presence of C. chinensis males reduced the fecundity of C. maculatus females (anova for C. maculatus females: F2,27 = 18·51, < 0·0001, Fig. 2a), but did not reduced the fecundity of C. chinensis females (anova for C. chinensis females: F2,27 = 1·18, = 0·32; Fig. 2a). The presence of C. maculatus males did not reduce the fecundity of either conspecific or heterospecific females. The longevity of C. maculatus females was significantly shorter when they were placed with C. chinensis males (average: 4·6 days) than with C. maculatus males (8·9 days), or with no males (9·4 days) (log-rank test, Kaplan–Meier method, χ2 = 30·6, < 0·0001). In contrast, the species of the male did not affect the longevity of C. chinensis females (log-rank test, Kaplan–Meier method, χ2 = 0·2, = 0·90).

Figure 2.

 (a) The total number of eggs laid by a female on 10 adzuki beans over 3 days when the female was placed with a Callosobruchus chinensis male (white bars), or with a Callosobruchus maculatus male (grey bars), or was left alone (black bars). (b) The total number of eggs laid by C. chinensis females (grey bars) or C. maculatus females (white bars) on 10 adzuki beans over 3 days when with 0, 1, 2 or 3 heterospecific males. Different letters (a or b) above the bars indicate a significant difference by Tukey–Kramer’s HSD (< 0·05). Error bars are standard errors.

Fecundity of both species decreased as the number of heterospecific males increased, clearly indicating a positive frequency-dependence of reproductive interference (linear regression analysis: C. chinensis, = −2·94, = 0·0056; C. maculatus, = −5·55, < 0·0001; Fig. 2b). The positive frequency-dependence was more pronounced in C. maculatus (ancova: male number, = 4·13, < 0·0001; female species, = −5·21, < 0·0001; interaction, = 6·23, < 0·0001; Fig. 2b). In the absence of males, both species exhibited similar fecundity: 53·9 ± 2·64 of C. chinensis females and 49·8 ± 2·72 of C. maculatus females (t-test: = 1·08, = 0·29). However, the presence of three heterospecific males halved the average fecundity of C. maculatus females (19·1 ± 3·33) compared with that of C. chinensis females (42·0 ± 3·45) (t-test: = 4·78, = 0·00015; Fig. 2b), whose fecundity decreased moderately as the number of heterospecific males increased.

Resource competition within a bean

From the adzuki beans infested with the same number of C. maculatus eggs and C. chinensis eggs, a greater number of C. maculatus adults emerged, irrespective of the oviposition order (paired Wilcoxon test: C. chinensis first, total number of C. chinensis = 33, C. maculatus = 49, = 51, = 0·041; C. maculatus first, total number of C. chinensis = 17, C. maculatus = 31, = 40, = 0·043).


This study revealed that C. chinensis was dominant to C. maculatus in terms of reproductive interference, in contrast to the dominance of C. maculatus in larval resource competition. The fact that species interaction generally resulted in sexual exclusion of C. maculatus indicates that reproductive interference largely determined which species of the two weevils persisted or went extinct.

Behavioural mechanisms of asymmetric reproductive interference

Mating attempts by C. chinensis males reduced the fecundity of C. maculatus females much more than mating attempts by C. maculatus males reduced that of C. chinensis females. This asymmetric reproductive interference is probably due to interspecific differences in male mating behaviour and female tolerance, because males of neither species discriminated between conspecific and heterospecific females. In the experiment of male mating attempts, C. chinensis males repeatedly tried to insert their penis into females, whereas C. maculatus males simply grasped the female abdomen. Females of C. maculatus behaviourally avoid the mating attempts of conspecific males after the first traumatic mating (Crudgington & Siva-Jothy 2000), though another study has suggested that females benefit from multiple mating (Fox 1993). Our preliminary observations of mating behaviour suggested that mating attempts by C. chinensis males more frequently triggered escape behaviours by C. maculatus females, eventually disturbing their reproductive behaviours, such as searching for oviposition sites and laying eggs, but less frequently induced escape behaviours in C. chinensis females (Kishi et al., unpublished data). The higher tolerance of mated C. chinensis females to the mating attempts of conspecific males seems to be due to their single mating (Miyatake & Matsumura 2004). Callosobruchus chinensis females kicked out courting males, contrast to C. maculatus females, who often fled from courting males (Kishi et al., unpublished data). In contrast to the frequent behavioural interference, interspecific copulation is rarely observed between C. maculatus males and C. chinensis females. Interspecific copulation has been observed occasionally between C. chinensis males and C. maculatus females, but without transfer of seminal fluid (T. Yamane, personal communication). However, this asymmetry in the occurrence of interspecific copulation probably has little importance with regard to the mechanism of reproductive interference, because neither hybrids nor unfertilized eggs are produced. Thus, reproductive interference in these two bean weevil species occurs exclusively through behavioural interactions, and not through hybridization or gamete dumping as a result of interspecific copulation. As with intraspecific mating, several steps precede interspecific copulation, so females (or males) can avoid such mating by changing their behaviour (Gröning & Hochkirch 2008).

Positive frequency-dependence of reproductive interference

In most cases, C. chinensis excluded C. maculatus competitively, depending on the initial ratio of adults, as previous studies have reported (Yoshida 1966; Fujii 1967; Bellows & Hassell 1984). The fecundity of females of both species decreased as the number of heterospecific males increased. The asymmetric and frequency-dependent competition outcomes are consistent with the theoretical predictions of asymmetric reproductive interference (Ribeiro & Spielman 1986; Yoshimura & Clark 1994). However, only a few laboratory studies have analysed frequency dependence in interspecific competition. In competition studies using fruit flies (Drosophila), only Narise (1965) has suggested that the mating stage was responsible for the observed positive frequency dependence of the competition between Drosophila melanogaster and Drosophila simulans. In contrast, other studies considered the detected positive frequency dependence to be a consequence of natural selection (Ayala 1971; De Benedictis 1977; Antonovics & Kareiva 1988). The competition results in flour beetles (Tribolium), which are also used as typical examples of interspecific competition, also imply the presence of asymmetric reproductive interference. Birch, Park & Frank (1951) reported that males of Tribolium confusum reduce the fecundity of Tribolium castaneum females, and that this reduction is not caused by egg cannibalism. In fact, T. confusum is dominant over T. castaneum in competition experiments, depending on the initial ratio of adults of the two species (Park 1948). Whether reproductive interference occurs, and its effect if it does occur, between species of fruit fly and between species of flour beetles needs to be examined. Further studies are necessary to review the competition results of previous laboratory competition studies in light of possible reproductive interference.

Our competition experiment showed the alternate population growth in the competition process (particularly Fig. 1g–i). This result may indicate the presence of a positive frequency-dependent effect of reproductive interference on the population dynamics of the heterospecific populations. We argue that the temporal fluctuations in the density of C. chinensis males, and also the density of C. maculatus males, though to a lesser extent, are likely to translate into temporal fluctuations in the magnitude of reproductive interference by turns, thereby naturally resulting in the alternate population growth. The similar alternate population growth has also been reported by Ishii & Shimada (2008). In contrast, it is notable that the assumed density-dependent effect in resource competition would be expected to synchronize the population dynamics of two and more species (Miller & Epstein 1986; Watson, Moss & Rothery 2000; Raimondo et al. 2004). To determine the relative contributions of reproductive interference and resource competition, it is thus necessary to separate positive frequency-dependent effects from density-dependent effects in competition studies.

Explaining the inconsistent outcomes of previous studies

Asymmetric reproductive interference between these two bean weevil species can convincingly explain previous findings of dominance by C. chinensis (Yoshida 1966; Fujii 1967; Bellows & Hassell 1984). Yoshida (1966) reported the persistent dominance of C. chinensis in the competition experiments, but that C. chinensis did not survive when the relative frequency of C. maculatus was greater than a threshold, and also that the longevity of C. maculatus females decreased when they were placed together with C. chinensis individuals. However, he did not suggest a mechanism to explain these phenomena. Our results seem to suggest that reproductive interference can readily account for these phenomena. However, some conflicting results among other studies remain unexplained. Although most studies have documented the survival of C. chinensis (Fujii 1965; Yoshida 1966; Bellows & Hassell 1984), two other studies have reported opposite results (Utida 1953; Ishii & Shimada 2008). With regard to these contradictory results, Fujii (1967) showed that the competition outcome depended in part on environmental conditions, such as humidity, ventilation and temperature. In fact, in flour beetles (Tribolium), the competition outcome can be affected by humidity, temperature and the presence of parasites (Park 1948; Neyman, Park & Scott 1956). However, differences in environmental conditions among studies can hardly provide a general explanation for the results of Utida (1953) and Ishii & Shimada (2008), because their environmental conditions and experimental designs were very similar to those of our study. Furthermore, we used the same strains of these two species as Ishii & Shimada (2008) used. Our results indicate that reproductive interference can more critically affect the competition outcome than resource competition, if environmental conditions are considered as sharing resources. We hypothesize that unintentional artificial selection changed some reproductive traits, which are closely associated with reproductive interference, in these two weevil strains. In our observations, the individuals of the jC strain of C. chinensis used in this study were very active, as suggested by the frequency of male mating attempts, whereas the jC strain individuals used by Ishii & Shimada (2008) were less active. In fact, we conducted a preliminary competition experiment using the same jC strain of C. chinensis as that used by Ishii & Shimada (2008) and the hQ strain of C. maculatus, and confirmed the extinction of the former in the 4 : 4 pair ratio treatment under the same experimental settings as in this study (Kishi et al., unpublished data). Moreover, the fecundity of C. maculatus females remained higher even when they were placed with the jC strain of C. chinensis males derived from the stock population used by Ishii & Shimada (2008). We argue that differential degrees of reproductive interference can explain the controversial and confounding results described above of the previous competition experiments in these two bean weevils.

Why reproductive interference has been overlooked

Reproductive interference in these two bean weevil species occurs through behavioural interactions, without leaving unfit hybrids or unfertilized eggs (Kishi et al., unpublished data). As a number of species discrimination processes may proceed before interspecific copulation occurs in most animal species, interspecific copulation should be a final behavioural consequence. Consequently, if any of these discrimination processes works successfully, virtually no detectable evidence of reproductive interference is left. In other words, absence of hybridization can never guarantee the absence of reproductive interference. We suggest that a lack of detectable evidence is why reproductive interference has long been overlooked, though hybridization and the resultant genetic invasion have been much reported (Gröning & Hochkirch 2008). It should be noted that reproductive interference itself reduces the occurrence of reproductive interference by promoting spatial segregation of interacting species. In the real world, resource partitioning between closely related species is expected to arise as a consequence of reproductive interference, minimizing negative interactions between those species (Gröning et al. 2007). Therefore, it may be rather difficult to observe interspecific behavioural interactions in the field after the establishment of habitat or resource partitioning or sexual exclusion that is sometimes caused by species invasion (Reitz & Trumble 2002; Liu et al. 2007). We expect that further studies will reveal the factors determining the differences in mating behaviour and mating systems that produce asymmetric reproductive interference between closely related species.


We greatly appreciate T. Harano and T. Miyatake for giving us the hQ strain of C. maculatus. We are also grateful to A. Honma and K.I. Takakura for comments on an early draft, and to E. Kuno, O. Kishida and T. Nakazawa for valuable discussions. We also thank two reviewers for numerous constructive comments. YT was funded by a grant-in-aid from the Japanese Ministry of Education, Science and Culture (no. 19201047).