Heredity 78 (1997) 277—283 Received 12 April 1996

Genetic basis for different host use in Epilachna pustulosa, a herbivorous ladybird beetle

HIDEKI UENO*, NAOYUKI FUJIVAMAt & HARUO KATAKURAt Laboratory of Environmental Medicine and Inform at/cs, Division of Bioscience, Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060 and tLaboratory of Systematics and Evolution, Division of Biological Sciences, Graduate School of Science, Hokkaido University, Sapporo 060, Japan

Thegenetic basis for different host plant use was studied in a herbivorous ladybird beetle, Epilachna pustulosa, that exhibits interpoputational variation in host plant utilization. It usually depends on thistle, but one of the local forms occurs on both thistle and blue cohosh, which differ at the infraclass taxonomic level. In this local population, genetic association between the developmental performance on the two plants was neutral, suggesting genetic independ- ence across the host plants. The form of reaction norms indicated some changes in relative rank position. Genetic variation among individuals on each host plant was detected. These results suggest that different genotypes are selected on different host plants and that a substantial proportion of the overall phenotypic plasticity is contributed by genotype-depend- ent environmental effects.

Keywords:Epilachnapustulosa, genetic correlations, host plants, reaction norms.

ential on the second character, resulting in changes Introduction in their joint evolution (Lande, 1979; Via & Lande, Theevolution of herbivore—plant interrelation has 1985). Even in the absence of direct selection, been a particularly contentious problem in ecology characters can undergo evolutionary changes as a and genetics (Futuyma & Peterson, 1985; Jaenike, result of selection on correlated characters (Lande 1990; Via, 1990). Much attention has been paid to & Arnold, 1983; Falconer, 1989). how the host range of a herbivorous insect is deter- By regarding the fitness component on one host mined, for this is supposed to be a clear example of plant as a character and the fitness component on the adaptation process. Adaptation to host plants is another host as a different character, genetic corre- dependent on the presence and nature of genetic lation between the performance across the host variation in behaviour and physiological perform- plants can be estimated (Via, 1984; Falconer, 1989). ance. Therefore the elucidation of the genetic basis The evolution of insect—plant relationships is signifi- for difference in habitat use, and habitat-related cantly influenced by this genetic correlation; with a fitness components, is a key to understanding the negative genetic correlation, selection for an course of insect—plant interaction. increase in a fitness component character on one Because selection operates on the whole pheno- host plant would result in a decrease in the type rather than on individual characters, the genetic analogous component character on the other plant covariance among characters substantially influences species. This trade-off in the performance may the overall course of evolution. In the case where further lead to the evolution of host-specialized both characters are under selection pressure, selec- genotypes, which are more fit in one particular tion on one character can produce a selection differ- habitat than other genotypes. On the other hand, with a positive genetic correlation, selection would lead to a population constitution where genotypes *Correspondence. Present address: Laboratory of Biology, Faculty of Education, Niigata University, 8050 Ikarashi Ni-No- with high performance on one host would also show Cho, Niigata 950-21, Japan. E-mail: [email protected] a high performance on the other. Thus, a difference

1997 The Genetical Society of Great Britain. 277 278 H. UENOETAL. in host habitat would cause little difference in at least three copulations for each male and female performance among genotypes. pair. This was to ensure last-male paternity; last- The form of reaction norms, patterns of pheno- male sperm precedence (P2) in this species was type distributions produced by a single genotype reported as 0.651—0.827 (Nakano, 1985). Assuming across environments, also has important implications the same sperm precedence pattern for the last for the course of insect—plant interaction. If geno- three successive copulations, the expected fertiliza- typic responses to the environment differ consider- tion success of previous males is 0.349 =0.0425at ably, then norms of reaction may cross, changing the most and 0.173 =0.00518at least. ranking of phenotypes (Stearns, 1992). Under direc- For each female, 30 eggs were collected from tional selection for higher growth performance, several egg clutches. In the following statistical some genotypes would therefore be favoured in one calculations they were regarded as full-sib progenies, plant environment and different genotypes in the assuming complete last-male paternity. Of these 30 other. Thus, in this way, different genotypes may be eggs, 15 were reared on thistle and 15 on blue selected in different host plant habitats. cohosh. Larvae were reared individually. Fresh host The present study uses the herbivorous ladybird plant leaves were supplied every other day. Larval beetle, Epilachna pustulosa, to investigate (1) development period was checked daily. On the day whether different genotypes are favoured in each of pupation, fresh weight was measured to 0.0001 g host plant environment, and (2) how characters are with a microbalance. After eclosion adult body genetically correlated across host plants, i.e. whether length was measured under a microscope to the their genetic correlations are negative and whether nearest 0.01 mm. possible trade-off relations are implicated. A mixed-model ANOVA was performed to detect Epilachna pustulosa exhibits interpopulational the contribution of factors to the total variation in variation in host plant utilization (Katakura, 1981; larval performance, setting family and interaction as Hoshikawa, 1983). It usually depends on the thistle random, and sex and host as fixed effects. A mixed (Cirsium kamtschaticum, Asteraceae), but one of the ANOVA decomposes observed variances into variance local forms, the Sapporo form, occurs on both thistle components as shown in Table 2. This analysis was and blue cohosh (Caulophyllum robustum, Berber- performed on three variables, larval period, pupal idaceae), which differ at the infraclass taxonomic weight and adult body length. The analysis also level. Of the previous studies concerned with insect provides statistical tests of genetic correlations for genetic systems on closely related plant species, few the traits across the hosts. With the present statis- detected negative genetic association (Rausher, tical procedure, the between-family variance directly 1984; Hare & Kennedy, 1986; Fox, 1993). However, detects the covariance association term of the trait different evolutionary consequences might be across the host environments. Mean squares with expected when host shifts involve plants from F-values above 97.5 per cent indicate significant different families (Bush, 1969). The particular inter- positive genetic covariation, whereas F-values below est in this study therefore was to ascertain the 2.5 per cent indicate significant negative genetic genetic relationship between growth performance on covariation, at the 5 per cent level (Fry, 1992). totally different host environments in Sapporo popu- Two further ANOVAS were performed to evaluate lations, where variation in host plant utilization is the between-family variance of performance on each most apparent. host plant separately. Between full-sib family variances contain nonaddi- Materialsand methods tive sources of genetic variation, including domi- nance and maternal effects. Furthermore, these Epilachnapustulosa is a univoltine species. Over- estimates may possibly be inflated by the variances wintered adults were collected on thistles in a suburb of the common microenvironment in the rearing of Sapporo City in May and June 1994. Rearing conditions. These values should therefore be conditions were 22°C and 16L-8D throughout this regarded as approximations of genetic properties but experiment. Pairs each consisting of one male and they provide an upper limit of genetic variation. one female were confined in separate plastic cases. Because genetic properties are not necessarily the Virtually all females would have copulated and same for males and females, and some phenotypic stored sperm before overwintering (Katakura, 1982) differences were actually observed for the relation- and additional copulations could have been possible ship between developmental period and adult body in late spring to early summer before collection. size, genetic correlations were estimated separately Hence we started the experiment after we confirmed for each sex. Genetic correlation of the same trait

The Genetical Society of Great Britain, Heredity, 78, 277—283. GENETIC BASIS FOR HOST USE 279 across the hosts was estimated as r,y = Coy (X,Y)! significantly larger in body length and heavier in [Var(X)Var(Y)]"2, where Cov(X,Y) is the covariance pupal weight when reared on cohosh than were their of family mean of the trait reared on one host plant siblings raised on thistles. This between-plant differ- and family mean of the same trait reared on the ence in overall tendency was observed in both males other host, and Var(X) is the variance of the family and females. Females were significantly larger and mean in one host environment (Via, 1984). Jack- heavier than male siblings, whereas larval periods knife procedure was used to calculate the correla- were not significantly different between the sexes. tions with their standard errors (Sokal & Rohif, Thus females achieved a larger body size than males 1981). Here the correlation was jackknifed with each although the developmental periods were similar. In family omitted once so that the total number of contrast, duration of the pupal period was not influ- iterations was equal to the number of families. The enced by either host plant or sex. Consequently, correlation coefficients were z-transformed for jack- differences in total developmental time were almost knifing procedures and transformed back to the all attributable to differences in the larval period. coefficient scale after obtaining 95 per cent confi- For all the characters estimated, larval period, dence limits for z-values. pupal weight and adult body length, mixed-model ANOVAS detected significant family x host interac- tions. By contrast, statistical tests for genetic correla- Results tion using calculated mean squares from mixed- Asignificant difference in overall survivorship of model ANOVAS detected no significant positive or ladybirds was observed between host plants (G-test negative associations (Table 2). The actual F-values of association between survivorship and host plant, were larger than 1, but not significantly so with a G, = 18.2, P.

Table 1 Percentage survival and means (standard errors) for developmental traits in the herbivorous ladybird beetle Epilachna pustulosa reared on thistle and blue cohosh Host

Thistle Blue cohosh

Larval duration (days) Male 25.0±0.3a (42) l9.8±0.2"(57) Female 25.5 (37) 20.1 (63) Pupal duration (days) Male 6.1 (42) 6.2±0.1 (57) Female 5.9±0.2 (37) 6.2±0.1 (63) Pupal weight (mg) Male 35.6 (42) 44.2±0.6" (57) Female 39.9±1.la (37) 47.2±0.D (63) Body length (mm) Male 6.116±0.044a (41) 6.532±0.032" (57) Female 6.351 (36) 6.777±0.029" (62) Survival to adult (per cent) 39.5 61.0 Letters indicate significant differences (P <0.01) between host plant species determined by an analysis of variance. Sample sizes are given in parentheses.

C The Genetical Society of Great Britain, Heredity, 78, 277—283. 280 H. UENO ETAL.

Table 2 Analysis of variance of developmental traits in the herbivorous ladybird beetle Epilachna pustulosa reared on thistle and blue cohosh

Larval duration Pupal weight Body length

Source EMS d.f. SS F d.f. SS F d.f. SS F

Sex +k1Q 1 8.368 3.129 1 379.602 18.075** 1 2.037 35.591** Family +k2o

0.06 0.06 Table3 Genetic correlation estimates in the herbivorous ladybird beetle Epilachna pustulosa between the same characters across different host plants with their 95 per cent confident limits in parentheses (see text for details)

0.05 0.05 Male Female Larval duration —0.001 —0.324 (—0.392, 0.391) (—0.911, 0.721) Pupal weight 0.07 1 0.190 0.04 0.04 (—0.276, 0.402) (—0.552,0.762) Body length 0.129 0.346 (—0.524, 0.686) (—0.216, 0.746) All estimates are significantly different from complete 0.03 0 03 correlation (—1 or 1), but not significantly different from 0. Malepupal weight Male pupal weight on blue cohosh (g) on thistle (g) (Table 3). All estimates were not significantly Fig. 1 Variation among full-sib families for male pupal different from zero, but were significantly different weight. Mean pupal weight for each full-sib family on blue from complete correlations (i.e. 1 or —1). cohosh is shown on the left axis with the corresponding From the analysis of variance on characters of mean of sibs reared on thistle on the right axis. Lines ladybirds reared on thistles and blue cohosh separa- connect the means for each family. Nonparallelism is the tely, significant genotypic variation was detected in expression of genotype x host plant interaction. each case (Table 4).

Discussion were detected (male larval period: r, =0.200, P>.0.05; pupal weight: r, =0.119,P>.0.05; adult Thepurpose of this study was to see how fitness body length: r =0.077,P>0.05; female larval components on one host are genetically correlated period: r, =0.182,P>0.05; pupal weight: r =0.105, with analogous components on a different host, and P>0.05; adult body length: r, =0.434,P>0.05). whether different genotypes are favoured in each The calculated genetic correlation was in agree- different host plant environment. With the expected ment with the above results. Estimated values of mean squares in the present statistical model, the correlation coefficients based on family means with family main effect variance provides a statistical test their 95 per cent confident limits are presented for the covariance term of the fitness components

The Genetical Society of Great Britain, Heredity, 78, 277—283. GENETIC BASIS FOR HOST USE 281

Table 4 Analysis of variance of developmental traits in the herbivorous ladybird beetle Epilachna pustulosa reared on thistle (A) or blue cohosh (B) and tested on each host plant separately

Larval duration Pupal weight Body length

Source d.f. SS F d.f. SS F d.f. SS F

A. Thistle Sex 1 14.708 3.379 1 1076.196 3493* 1 0.502 6.505* Family 12 230.318 4.410** 12 174.806 6.808** 12 3.936 4.252** Error 64 278.571 64 25.675 62 4.783 B. Blue cohosh Sex 1 0.295 0.181 1 0.021 11.071** 1 1.586 34.807** Family 12 94.624 4.827** 12 0.106 4747** 12 1.492 2.728** Error 106 173.157 106 0.197 106 4.784 *p<005, **P<0.QL

across habitat environments (Fry, 1992). No signi- Because the ability to exploit hosts is unlikely to ficant effect was found in this study. The actual esti- be constrained by the present physiological charac- mates of genetic correlations were close to zero and ters in this population, host plant utilization and differed significantly from —1 and +1. Thus, both host range determination will primarily be influ- analyses indicate no genetic correlation between the enced by behavioural and ecological factors such as fitness components on the two different host oviposition preference (if one exists), migration environments. With these genetic properties, selec- among host resources and abundance of host resour- tion on the present physiological mechanisms would ces. This view is consistent with studies that stress neither facilitate nor reduce the ability to exploit the the role of behaviour in host determination other host plant. (Jaenike, 1978, 1990; Courtney, 1983; Iwasa et al., The form of the reaction norms suggests some 1984; Singer et al., 1988, 1989; Conner, 1991; Leder- changes in relative rank position when genotypes house et a!., 1992). The genetic basis for these were reared on the two hosts. Rank correlations behavioural factors and their associations with based on family mean variance were not significant physiological performance need to be investigated. for any variable. On the other hand, among-family Genetic correlation between performance on variation was detected for all variables when statis- different hosts has been estimated in many species tical analyses were made for each host resource. (Rausher, 1984; Via, 1984; Hare & Kennedy, 1986; Therefore, the present results would support the Futuyma & Philippi, 1987; James et a!., 1988; idea that under directional selection for higher Jaenike, 1989; Fox, 1993), but only a few revealed values of the characters, different genotypes are the expected negative correlations (Via, 1991). One selected on different host plants, and substantial explanation for the lack of negative correlation is genetic variation for growth performance is main- that the antagonistic relation is generally absent tained among the individuals on each host plant. between the performance on different hosts (Butlin, These lines of reasoning indicate that substantial 1987; Futuyma & Philippi, 1987; Bernays & portions of phenotypic plasticity result from geno- Graham, 1988), no matter how different the chemi- type-dependent environmental effects. cal characteristics and physiological systems. One aim of this study was to see whether negative A negative genetic correlation is not necessarily genetic correlations are observed between growth expected if the population has only recently shifted performance on totally different food plants in an to a host plant and the plant environment is still insect species that varies in host plant utilization. novel to the population. At the beginning of the The results indicate that the physiological systems insect—plant relationship, alleles affecting the growth responsible for the growth performance on different performance of insects on the novel host are plants are genetically rather independent. Despite expected to be randomly associated with alleles the difference in host plant environments, trade-off affecting the performance on the established hosts, relationships were not found. and negative genetic correlation is expected only

The Genetical Society of Great Britain, Heredity, 78, 277—283. 282 H. UENOETAL. when a population has reached genetic equilibrium BUTLIN, R. 1987. A new approach to sympatric speciation. on two hosts (Joshi & Thompson, 1995). Although Trends Ecol. Evol., 2, 3 10—311. at least 40 years (=generations)have passed since CONNER, E. F. 1991. Colonization, survival, and causes of the Sapporo population was first observed to utilize mortality of Camera nahamadiyadella(Lepidoptera: both of the plant species (Kurosawa, 1953; Kata- Gracillariidae) on four species of host plants. Ecol. Entomol., 16,315—322. kura, 1974), it is not certain whether the present COURTNEY, S. i'. 1983. Models of host plant location by population has experienced sufficient selection in butterflies: the effect of search images and searching both plant environments and reached a genetic efficiency. Qecologia, 59, 3 17—321. equilibrium. FALCONER, a s. 1989. Introduction to Quantitative Genet- Another explanation for the apparent lack of ics, 3rd edn. John Wiley and Sons, New York. negative genetic correlations might lie in the experi- FOX, C. W. 1993. A quantitative genetic analysis of oviposi- mental conditions used. Our conditions may have tion preference and larval performance on two hosts in unintentionally imposed selection on the experi- the bruchid beetle, Callosobruchus maculatus. Evolu- mental population for overall viability in the novel tion, 47, 166—175. conditions (Service & Rose, 1985). This would yield FRY, .i. D. 1992. The mixed-model analysis of variance overestimates of genetic correlations, and any applied to quantitative genetics: biological meaning of discussion based on these estimated genetic proper- the parameters. Evolution, 46, 540—550. FUTUYMA,D. J. AND PETERSON, S. C. 1985. Genetic varia- ties may lead to erroneous conclusions. For this tion in the use of resources by insects. Ann. Rev. Ent., reason, the results obtained should be interpreted 30, 217—238. cautiously. FUTUYMA, D. J. AND PHILIPPI, T. E. 1987. Genetic variation Variation in habitat use among local populations and covariation in responses to host plants by Alsophila is known in E. pustulosa. The population collected pometaria (Lepidoptera: Geometridae). Evolution, 41, from a suburb of Sapporo City indicates a close 269—279. association with blue cohosh, as shown in the HARE, J. D. AND KENNEDY, G. G. 1986. Genetic variation in present study. Populations from other localities plant-insect associations: survival of Leptinotarsa decem- appear to be dependent mainly on thistle, even lineata populations on Solanum carolinense. Evolution, though blue cohosh is present in the locality (Kata- 40, 1031—1043. kura, 1981; Hoshikawa, 1983). Thus, the insect— HOSHIKAWA, K. 1983. Host-race formation and speciation plant relationship seems to be specific to each local in the Henosepilachna vigintioctomaculata complex (Coleoptera, Coccinellidae) I. Host-plant ranges and population. If the genetic structure of ladybird popu- food-preference types. KontyO, 51, 254—264. lations reflects host plant variation, comparative IWASA, Y., SUZUKI, Y. AND MATSUDA, H. 1984. Theory of study of genetic properties among different popula- oviposition strategy of parasitoids.I.Effects of tions may provide us with evidence of local adapta- mortality and limited egg number. Theor Pop. Biol., 14, tions and have implications for the potential of 350—356. herbivorous insects to coevolve with their host plants JAENHCE, i. 1978. On optimal oviposition behavior in (Michalakis et al., 1993). phytophagous insects. Theor Pop. Biol., 14, 350—356. JAENIKE, j.1989.Genetic population structure of Droso- phila tripunctata: patterns of variation and covariation Acknowledgements of traitsaffecting resource use. Evolution, 43, 1467—1482. Weexpress our sincere thanks to A. Osugi for assis- JAENIKE, .1990.Host specialization in phytophagous tance throughout this experiment and to K. Tsuji insects. Ann. Rev. EcoL Syst., 21, 243—273. and S. Higashi for comments on the manuscript. JAMES, A. C., JACKUBCZAK, J., RIDLEY, M. P. AND JAENIKE, J. Two anonymous referees improved the manuscript 1988. On the causes of monophagy in Drosophila very much. This work was supported by JSPS quinaria. Evolution, 42, 626—630. Research Fellowships to H. Ueno. JOSHI, A. AND THOMPSON, J. N. 1995. Trade-offs and the evolution of host specialization. Evol. Ecol., 9, 82—92. KATAKURA, H. 1974. Henosepilachna pustulosa feeding on References Caulophyllum thalictroides robustum in Sapporo. Kontyui, 42, 394 (in Japanese). BERNAYS,E. AND GRAHAM, M. 1988. On the evolution of KATAKURA, H. 1981. Classification and evolution of the host specificity in phytophagous arthropods. Ecology, phytophagous ladybirds belonging to Henosepilachna 69, 886—892. vigintioctomaculata complex (Coleoptera, Coccinelli- susi-i, G. L. 1969. Sympatric host race formation and speci- dae). J. Fac. Sci. Hokkaido University, VI (Zool.), 22, ation in fungivorous flies of the genus Rhagoletis 301—378. (Diptera, Tephritidae). Evolution, 23, 237—251. KATAKURA, H. 1982. Long mating season and its bearing

The Genetical Society of Great Britain, Heredity, 78, 277—283. GENETIC BASIS FOR HOST USE 283

on the reproductive isolation in a pair of sympatric environments. Evolution, 39, 943—945. phytophagous ladybirds (Coleoptera, Coccinellidae). SINGER, M. C., NG, D. AND THOMAS, C. D. 1988. Heritability Kontyâ, 50, 599—603. of oviposition preference and its relationship to KUROSAWA,T. 1953.Notes on the food-plants of Epilachna offspring performance within a single insect population. pustulosa Kôno (Coleoptera, Coccinellidae). Oyô-Kon- Evolution, 42, 977—985. tyâ, 9, 12—15 (in Japanese). SINGER, M. C., THOMAS, C. D., BILLINGTON, H. L. AND LANDE,R.1979. Quantitative genetic analysis of multi- PARMESAN, C. 1989. Variation among conspecific insect variate evolution, applied to brain:body size allometry. populations in the mechanistic basis of diet breadth. Evolution, 33, 402—416. Anim. Behav., 37, 751—759. LANDE, R. AND ARNOLD, S. 1983. The measurement of SOKAL, R. R. AND ROULE, F. .i. 1981. Biomet,'y, 2nd edn. selection on correlated characters. Evolution, 37, Freeman, San Francisco. 1210—1226. STEARNS, S. C. 1992. The Evolution of Life Histories. Oxford LEDERHOUSE, R. C., AYRES, M. P., NITAO, J. K. AND SCRIBER, University Press, New York. J. M. 1992. Differential use of lauraceous hosts by swal- WA, s. 1984. The quantitative genetics of polyphagy in an lowtail butterflies, Papilio troilus and P palamedes insect herbivore. II. Genetic correlations in larval (Papilionidae). Oikos, 63, 244—252. performance within and among host plants. Evolution, MICHALAKIS, Y., SHEPPARD, A. W., NOEL, V. AND OLIVIERI, 1. 38, 896—905. 1993. Population structure of a herbivorous insect and VIA, S. 1990. Ecological genetics and host adaptation in its host plant on a microgeographic scale. Evolution, 47, herbivorous insects: the experimental study of evolution 1611—1616. in natural and agricultural systems. Ann. Rev. Ent., 35, NAKANO, s. 1985. Sperm displacement in Henosepilachna 421—446. pustulosa (Coleoptera, Coccinellidae). Kontya, 53, VIA, S. 1991. The genetic structure of host plant adapta- 516—519. tion in a spatial patchwork: demographic variability RAUSHER, M. D. 1984. Tradeoffs in performance on among reciprocally transplanted pea aphid cloñès. different hosts: evidence from within- and between-site Evolution, 45, 827—852. variation in the beetle Deloyala guttata. Evolution, 38, VIA, S. AND LANDE, R. 1985. Genotype-environment inter- 582—595. action and the evolution of phenotypic plasticity. Evolu- SERVICE, P. M. AND ROSE, M. R. 1985. Genetic covariation tion, 39, 505—522. among life history components: the effect of novel

TheGenetical Society of Great Britain, Heredity, 78, 277—283.