Geographic variation in body and ovipositor sizes in the leaf beetle Plateumaris constricticollis ( Coleoptera : Chrysomelidae ) and its association with climatic conditions and host plants

Plateumaris constricticollis is a donaciine leaf beetle endemic to Japan, which lives in wetlands and uses Cyperaceae and Poaceae as larval hosts. We analyzed geographic variation in body size and ovipositor dimensions in three subspecies (constricticollis, babai, and toyamensis) in different climatic conditions and on different host plants. In addition, the genetic differentiation among subspecies was assessed using nuclear 28S rRNA gene sequences. The body size of subspecies toyamensis is smaller than that of the other subspecies; mean body size tended to increase towards the northeast. Ovipositor length and width are smaller in subspecies toyamensis than in the other subspecies. Although these dimensions depend on body size, ovipositor length still differed significantly between toyamensis and constricticollis-babai after the effect of body size was removed. Multiple regression analyses revealed that body size and ovipositor size are significantly correlated with the depth of snow, but not temperature or rainfall; sizes were larger in places where the snowfall was greatest. Haplotypes of the 28S rRNA gene sequence were not shared among the subspecies. Subspecies constricticollis and babai each had a unique haplotype, whereas subspecies toyamensis had four haplotypes, indicating differentiation among local populations within toyamensis. The evolution of body and ovipositor size in relation to habitat conditions and host plants is discussed.


INTRODUCTION
Geographic variation in insect body size reflects the differential adaptation of local populations to local environmental conditions such as climatic factors, food availability, and the presence of related species (Masaki, 1967(Masaki, , 1978;;Roff, 1980;Mousseau & Roff, 1989;Blanckenhorn & Fairbairn, 1995;Sota et al., 2000a, b).As body parts are particularly important in the adaptation to habitat and food conditions, the ovipositor and associated structures also show marked geographic variation in response to abiotic conditions (Masaki, 1979(Masaki, , 1986;;Bradford et al., 1993;Mousseau & Roff, 1995) and host plant morphology (e.g., Toju & Sota, 2006a, b).Analyses of geographic variation in body size and the size of other functional parts are important for understanding locally variable adaptations, which produce morphological diversity within insect species and potentially lead to speciation.
Here, we analyze the geographic variation in body and ovipositor sizes of a wetland leaf beetle, Plateumaris constricticollis (Coleoptera, Chrysomelidae), a Donaciinae species endemic to the Japanese archipelago the larvae of which feed on the roots of monocotyledonus plants.Fossils of this species are known from the late Pliocene, along with another endemic species P. akiensis, although few extant species occur in late Pliocene and early Pleistocene deposits (Hayashi, 2002b(Hayashi, , 2004;;Hayashi & Shiyake, 2002).Thus, P. constricticollis represents one of the few ancient insect species persisting since the late Pliocene.Whereas P. akiensis is confined to a small area in western Honshu, P. constricticollis is distributed widely from western Honshu to Hokkaido and is polytypic with three subspecies (Hayashi, 2004;Hayashi & Shiyake, 2004) characterized by external and internal (ovipositor) characters.The two northern subspecies, constricticollis and babai, have equally long ovipositors with an acutely angled apex (Fig. 1), but differ in body coloration.The southwestern subspecies, toyamensis, has a shorter ovipositor with a right-angled apex and a smaller body than the other two subspecies (Fig. 1).The different morphological characters of the P. constricticollis subspecies may reflect local adaptations to different host plants or climatic conditions; the length of its ovipositor appears to match the stem thickness of its host plants (T.Yagi, unpubl.observ.),and larger body may have resulted from an adaptation to a climate with higher snowfall (Ego et al., 1988;Tominaga, 1988).
Here, we describe the variation in body and ovipositor size among local populations and analyze the effects of environmental variables on morphological variation.Further, we examine the relationship between morphological (subspecies) differentiation and genetic differentiation using partial sequence data from the nuclear 28S rRNA gene.

Morphological analysis
A total of 235 female beetles were collected from 20 sites for morphological analysis (Table 1, Fig. 1).To measure pronotum width, right elytral length, and ovipositor length and width, a dorsal image of the whole body and dorsal and lateral images of the ovipositor were taken using a CCD camera mounted on a microscope.The software ImageJ 1.34s (National Institute of Health, USA) was used to determine the dimensions from the images (see Fig. 2).Ovipositor length was measured along the curved line from the apex to the base on the lateral image of the chitinized ovipositor.The maximum width on the dorsal image was used as ovipositor width.
In the statistical analysis, all millimeter dimensions were log10 transformed to ensure normality and homoscedasticity.All statistical analyses of the morphological data were performed using JMP 5.0.1J(SAS Institute Inc.).To reveal the climatic factors affecting body size variation, annual mean temperature, annual rainfall, and snowfall data from Mesh Climatic Data 2000 (Japan Meteorological Agency, 2002) were used in a multiple regressions with stepwise model selection.For snowfall, the maximum depth of snow in February, among others (maximum depth of snow in December-March, and the average and maximum of these monthly data), because the F-values were highest for the effect of this variable on body and ovipositor sizes.In the multiple regression, subspecies (0 = toyamensis, 1 = babai + constricticollis) were included as an independent variable.

DNA sequencing and phylogenetic analysis
To determine the genetic relationship among the subspecies, partial sequences of the nuclear 28S rRNA gene (28S) were analyzed.A total of 20 ethanol-fixed specimens from 20 localities were used.In addition, samples of the four other Japanese Plateumaris species were used: P. weisei (n = 1 from Higashikawa, Hokkaido), P. akiensis (n = 1 from Geihoku, Hiroshima), P. sericea (n = 1 from Shimane) and P. shirahatai (n = 1 from Iwaizumi, Iwate).
Total genomic DNA was extracted from muscles using an AquaPure Genomic DNA Kit (Bio-Rad Laboratories, Hercules, CA, USA).An 804-bp fragment of 28S was PCR-amplified using the primer pair 28S-01/28SR-01 (Kim et al., 2000).For direct sequencing of the PCR products, a dye terminator cyclesequencing reaction was performed using an ABI PRISM BigDye Terminator Cycle Sequencing FS Ready Reaction Kit, followed by electrophoresis on an ABI 377 sequencer (Applied Biosystems, Foster City, CA, USA).
Sequence alignment was done manually without gaps.A maximum-likelihood tree was obtained using PHYML version 2.4.4 (Guindon & Gascuel, 2003) with a GTR+I+G substitution model.Node support was obtained using 1000 replicates of nonparametric bootstrapping.A statistical parsimony network was 166 a C -Carex (Cyperaceae); E -Eleocharis (Cyperaceae); P -Phragmites (Poaceae).
Stepwise multiple regression analysis of body and ovipositor dimensions with three climatic variables commonly resulted in models with maximum depth of snow in February, together with the effect of subspecies difference (toyamensis vs. babai + constricticollis) in body size and ovipositor dimensions (Table 3).Only for OL was a model selected that included rainfall, but the effect of rainfall was not significant.Body and ovipositor size increased with depth of snow (Fig. 3c, d).
The 28S sequences were aligned unambiguously without requiring gaps.The 28S haplotype (sequence) was monomorphic within subspecies constricticollis (n = 7) and subspecies babai (n = 8), but four haplotypes were found for subspecies toyamensis (n = 5; Fig. 5).The statistical parsimony network revealed that the haplotypes from different subspecies were connected via missing haplotypes.In particular, subspecies constricticollis was separated by three or more steps from subspecies babai 168 ***, P < 0.0001.and toyamensis, indicating a deep coalescence between these subspecies.

DISCUSSION
The three subspecies of P. constricticollis were distinguished by their nuclear 28S haplotypes.The 28S sequence was monotypic for both constricticollis and babai subspecies, whereas four distinct haplotypes were found for toyamensis.Although the sample size was small, the 28S gene sequences generally showed very little variation within or between populations.Indeed, the sequences of P. weisei, P. sericea and P. shirahatai are shared by continental populations of the same species (T.Sota, unpubl. data).Thus, differentiation among the subspecies of P. constricticollis has a substantial historical basis.Subspecies constricticollis exhibited a rather large genetic differentiation from subspecies babai, despite their small morphological differences (body colour), indicating a long period of differentiation between the two subspecies without strong divergent selection on morphology.The monotypic haplotypes of these subspecies may be attributed to bottleneck events following range contraction.In contrast, the toyamensis subspecies exhibited haplotype divergence among localities, suggesting the relatively long persistence of segregated local populations.A more detailed analysis of P. constricticollis's historical biogeography using mitochondrial cytochrome oxidase subunit I gene sequences will be reported elsewhere (Sota & Hayashi, unpblished data).
Fossils of P. constricticollis are recorded from the late Pliocene to the late Pleistocene (e.g., Fossil Insect Research Group for Nojiri-ko Excavation, 1987Excavation, , 1990;;Hayashi, 1999aHayashi, , b, 2004b) ) and provide insight into the evolution of body size in subspecies babai.The ovi- positor shape of early Pleistocene fossils of P. constricticollis from the Uonuma Formation in Niigata, Honshu (within the present distribution of subspecies babai) is similar to that of the current subspecies babai, but the body size is small like that of subspecies toyamensis, suggesting that the ancestral body size of subspecies babai may have been smaller than it is today (Hayashi, 1999a).Similarly, P. constricticollis fossils obtained from Plio-Pleistocene strata in Saitama resemble subspecies babai and constricticollis, but the body size is smaller than in the present subspecies babai (Hayashi, 1999b).In addition, fossils of P. constricticollis found in strata from the last glacial period at Nojiriko, Nagano, Honshu, identified as subspecies babai, are smaller than beetles in a nearby extant population, suggesting an increase in body size during the postglacial period (Fossil Insect Research Group for Nojiri-ko Excavation 1987Excavation , 1990)).Thus, the body sizes of subspecies babai and constricticollis may have changed along with climatic changes.Fossils identified as subspecies toyamensis, based on ovipositor shape and external morphology, occur in the mid-Pleistocene (0.3 Ma) deposits in Gifu, central Honshu (Ego et al., 1988); the body size of these fossils is comparable to the extant toyamensis in the same region, indicating there has not been a body size change in this subspecies.
The body size of P. constricticollis did not show simple clinal variation, as is expected based on local adaptation to temperature conditions and a fixed life cycle (voltinism; Masaki, 1967;Roff, 1980).Because Tominaga (1988) mentioned the possibility of selection for large body size in areas with deep snow (see also Ego et al., 1988), we tested the effect of climatic factors, including temperature, rainfall, and snowfall, on body and ovipositor dimensions, and found that only depth of snow had a consistent positive effect on these dimensions.Heavy snowfall in the central northern region of Honshu, facing the Sea of Japan, is associated with the inflow of the warm Tsushima Current from the western Tsushima Channel.The Tsushima Channel was closed until the beginning of the Pleistocene and has been open since, except during glacial periods (Kitamura et al., 2001;Kitamura & Kimoto, 2004).The correlation of body size with snowfall and the probable change in snowfall pattern after a EL -elytral length; PW -pronotum width; OL -ovipositor length; OW -ovipositor width NS, P > 0.05; *, P 0.05; **, P < 0.01; ***, P < 0.001.the early Pleistocene may explain the aforementioned change in the body size of P. constricticollis detected by comparing fossil and extant specimens.Thus, with the inflow of the Tsushima Current, the body size of P. constricticollis in the north, on the Sea of Japan, would have increased in response to changes in habitat conditions due to heavy snowfall.
The differences in body and ovipositor sizes among the subspecies can be associated with differences in lifehistory traits.The two northern subspecies, constricticollis and babai, emerge as adults in June and early July.The adult beetles mate on Carex spp.and other plants, but have never been observed feeding.Therefore, females are thought to deposit eggs immediately after emergence and mating, without feeding (Fossil Insect Research Group for the Nojiriko Excavation, 1985).In contrast, adult beetles of subspecies toyamensis emerge in late May and June, and feed on pollen and leaves of Carex and Scirpus spp.Ego et al. (1988) and Tominaga (1988) hypothesized an association between large body size and a climate with heavy snowfall, and suspected that the longer larval period under snow cover results in delayed adult emergence and asynchrony with host plant phenology, favouring both the enlargement of body size and gonad maturation without adult feeding.In the region occupied by subspecies toyamensis, the relatively mild climate allows for a short larval period, and natural selection may favour early adult emergence at a smaller body size and active feeding for gonad maturation.In the region of subspecies babai and constricticollis, however, the cool climate would delay larval development and the start of adult activity, and selection would favour adult emergence at a larger size and immediate oviposition without feeding.
Although ovipositor length largely depended on body size, the difference in ovipositor shape between subspecies toyamensis and constricticollis-babai can be related to differences in their host plants.Some of the larval host plants of P. constricticollis are known: Carex sp. for subspecies constricticollis (Narita, 2003); Carex thunbergii, Carex ampliata and Phragmites australis for subspecies babai (Hayashi, 2002a;Narita, 2003); and Carex sp. and Eleocharis sp. for subspecies toyamensis [Narita, 2003;Hayashi, 2005;Narita (2003) records Scirpus sp., but it was actually Eleocharis sp.; Y. Narita, pers. comm.].At some of the locations where subspecies constricticollis and babai were collected, only reeds of Phragmites australis were available and likely to be the sole host plant (Table 1).Ovipositor morphology may represent an adaptation to different host plants because the importance of matching ovipositor length and host plant stem diameter is indicated by laboratory observations (T.Yagi, unpubl.data).Females of subspecies toyamensis with short ovipositors laid eggs in the thin stems of Eleocharis and Carex, whereas females of subspecies babai from Niigata and Nagano, which have long ovipositors, could not use these hosts and laid eggs in the thick stems of reeds.Interestingly, the long ovipositor of subspecies babai went through the stem of Eleocharis and the eggs were laid on the outside of the stem.Because reeds are more common in the habitats of subspecies babai and constricticollis than in the habitats of toyamensis (Table 1), the elongated ovipositor in the former subspecies may an adaptation allowing the use of this plant.
In conclusion, the differentiation in body size and ovipositor size and shape among the subspecies of P. constricticollis can be attributed to differential adaptation to host plant and climatic conditions.However, the evolutionary processes remain to be explored further, based on the subspecies' life histories in the field.In particular, more information is needed on host plant use and the life cycle, especially during the larval period.

Fig. 1 .
Fig. 1.Sampling localities of Plateumaris constricticollis (closed and open circles).Open circles represent localities where DNA samples only were collected.See Table1for locality numbers beside the circles.Photographs are of adult beetles and the ovipositors of three subspecies.

Fig. 3 .
Fig. 3. Variation in elytral length (a, c) and ovipositor length (b, d) among local populations of Plateumaris constricticollis, relative to latitude (a, b) and maximum depth of snow in February (c, d).Values are mean ± SD.
Model selected by the stepwise multiple regression analysis.Explanatory variables were annual mean temperature, mean annual rainfall, maximum depth of snow in February and subspecies (0 = toyamensis; 1 = babai + constricticollis).

Fig. 5 .
Fig. 5. Phylogram resulting from a maximum-likelihood analysis of the nuclear 28S rRNA gene (upper) and a statistical parsimony network for the same sequence data (lower).Gen-Bank accession numbers for the sequence data are shown in the upper panel.

TABLE 1 .
(Clement et al., 2000)d number of specimens of Plateumaris constricticollis used in the morphological analysis of female beetles and molecular phylogenetic analysis.constructedusingTCS version 1.21(Clement et al., 2000)and a 95% parsimony connection limit.

TABLE 2 .
Nested analysis of variances for the female body and ovipositor sizes.