Geographic distribution, host plants, and morphological variation of the currently radiating phytophagous ladybird beetle Henosepilachna diekei

Abstract To overview the morphological and ecological diversification within a species of phytophagous ladybird beetle Henosepilachna diekei, information concerning its distribution and host plant use was studied. This species was found widely in South East and South Asia from the Philippines through the Greater Sunda islands, a western part of Lesser Sundas, the Malay Peninsula to Sri Lanka. Mikania species (Asteraceae), Mikania micrantha in particular, were the most frequently used host plants, but some other plants belonging to Lamiaceae or Acanthaceae were also used by some local populations. Each population is possibly monophagous. According to morphological analyses, seven populations from four host plants at six sites on four Indonesian islands were grouped into two, one occurring on Java and Kalimantan (Borneo) and another on Sulawesi and Lombok, indicating that they were separated by the Wallace line. This morphological distinction was not correlated with the host plant use.


Introduction
Rapid lineage diversification in association with evolution of various ecological characters and differential niche utilization is often referred to as "adaptive radiation", which is one of the major processes to generate biodiversity (Schluter 2000). This rapid lineage diversification is often coupled with morphological divergence between geographically, ecologically and phylogenetically distinct populations in single species (Gould and Woodruff 1990). As morphological diversification is often associated with species diversification through enhancing reproductive isolation such as lock-and-key (Sota & Nagata 2008;Sota & Tanabe 2010;reviewed in Masly 2012) or size assortative mating (Brown 1990;Jiang et al. 2013), the investigation of morphological variation other than ecological characters within currently radiating species is critical to understand the process of adaptive radiation.
Henosepilachna diekei Jadwiszczak & Węgrzynowicz (referred to as Epilachna sp. 3 in Katakura et al. 2001; see also Kobayashi et al. 2009; for the taxonomic treatment) is a phytophagous ladybird beetle that was described based on specimens from the Philippines (Dieke 1947). In our survey on phytophagous ladybird beetles in South East Asia, this species is also found from various localities in Indonesia and adjacent countries. In addition, there seem to be host races in this species. In Java, for example, there are two types of populations, one depending on Leucas lavandulifolia Sm. (Lamiaceae), and the other on the climbing hemp weed, Mikania micrantha Kunth (Asteraceae) Matsubayashi et al. 2011;Fujiyama et al. 2013). A series of field and laboratory experiments demonstrated that beetle populations from L. lavandulifolia do not grow on M. micrantha and those from M. micrantha do not grow on L. lavandulifolia (Matsubayashi et al. 2011;Nakano unpublished data). In addition, there is no reproductive isolation between these two types of populations except differential host plant use (Matsubayashi et al. 2011(Matsubayashi et al. , 2013a(Matsubayashi et al. , 2013b suggesting that these two types are at the early stage of speciation driven by divergent host adaptation (i.e. "host race", sensu Drès & Mallet 2002). We have further found that there are populations that show specialization to plants other than M. micrantha and L. lavandulifolia (Matsubayashi 2008). Phylogenetic analysis based on partial nucleotide sequences of the mitochondrial cytoplasmic oxidase subunit I (COI) gene revealed that the beetle populations from each host plant are not monophyletic (Fujiyama et al. 2013), indicating that the host shift could occur several times in different localities in Java. Katakura et al. (2001) further observed that the H. diekei populations from Sumatra, a neighbouring island of Java, use other host plants belonging to Lamiaceae. Hence, H. diekei provides a good example with which to investigate adaptive radiation under various geographic settings. Preliminary morphological observation indicated that there was a certain degree of morphological variation (body shape, elytral spot pattern, minute structure of male genitalia) between populations of H. diekei sampled on different host plants.
In this paper, we investigate the distribution and host plants of H. diekei in the Asian tropics and further analyse the morphological divergence between geographically and ecologically different populations. Although our preliminary morphological observation indicates some degrees of inter-population morphological variation, it is still unknown whether the morphological divergence is associated with differential host plant use or not. This kind of knowledge is essential to understand the pattern and process of morphological diversification during adaptive radiation.

Beetle distribution and host plants
We surveyed the occurrence and host plant use of H. diekei from 1980 to 2013 in almost all major islands of Indonesia (Java, Sumatra, Kalimantan, Sulawesi, Bali, Lombok, Sumbawa, Flores, West Timor and Papua) and some other localities in South East Asia (the shaded areas in Figure 2). We recorded the absence or presence of the beetle, and if present, we collected the beetles and the host plants for morphological analysis or species identification. Species identification was made based on morphologies by K. Matsubayashi and H. Katakura for the beetles and by J. Yokoyama for the host plants.

Morphological analysis
Adult individuals collected were fixed in absolute alcohol and kept at −40°C. Six body characteristics (body length, pronotum length, pronotum width, elytral length, elytron width, elytral height; see Figure 1A) were measured using a Nikon SMZ1500 microscope (Nikon Instech, Co., Ltd.) by recording images with a Digital Sight camera system (DS-L1, Nikon Instech, Co., Ltd.). Image stacking for the dorsal and lateral views of habitus was performed with CombineZP (Hadley, 2010;http://www.hadleyweb.pwp.blueyonder.co. uk/). Then, males were dissected in distilled water, and genitalia were eviscerated, cleaned in alkali solution (0.1% KOH) for 12 hours, and then washed in 0.1% lactic acid. The number of hairs on the penis guide of the tegmen and the structure of the apical edge of the penis were examined and recorded using the same systems described above (see Figure 1B, C). According to a preliminary observation, the structure of the apical edge of the penis was classified into three types; emarginate (Type I), truncate (Type II) and convex (Type III) (see Figure 6).
Principal component analysis (PCA) was conducted with the six morphological characters using princomp functions in R 2.15.0 software (R Development Core Team R 2012). After the principal component analysis, we squeezed the principal components based on the importance of components especially for cumulative proportion (in general, more than 80% is useful). Further, we detected the relative contribution of each morphological character for the selected principal components based on the factor loadings, showing correlation of the factor to the principal components. We then tested statistical differences in the value of selected morphological characters among beetle populations using one-way analysis of variance (i.e. aov function) with subsequent multiple comparisons by Tukey HSD test.
Furthermore, we tested differences in the number of hairs on the penis guide and the types of apical edge of penis among populations using a Kruskal-Wallis test with multiple comparisons by Kruskal.wallis function in R 2.15.0 software.

Beetle distribution and host plants
Collection and observation records of Henosepilachna diekei and the host plant use were summarized ( Figure 2).
Beetle distribution: Henosepilachna diekei was widely found from the southern part of the Malay Peninsula (Ban Han in Songkhla, Thailand and Kuala Lumpur in Malaysia) through the Greater Sunda Islands (Borneo, Sumatra, Java) to the western part of the Lesser Sunda Islands (Bali and Lombok) ( Figure 2). This beetle was also recorded from Gannoruwa, Sri Lanka (not shown in Figure 2). On the other hand, this species was not found in the eastern part of the Lesser Sundas (Flores, Sumbawa, West-Timor), Papua, and the southern part of Thailand (excluding the Host plants: Henosepilachna diekei used plants of at least three families and four genera in the area surveyed ( Figure 2). Of these four genera, Mikania (Asteraceae) was the most widely used. The beetle populations using M. micrantha were observed widely on Java (see Fujiyama et al. 2013, for further information), Sumatra (Lampung, Padang and Lake Toba), Borneo (Balikpapan and Pontianak in Kalimantan, Kinabaru in Saba), and Kuala Lumpur. In Kuching in Sarawak, in addition, feeding traces were found on this plant although no individual was collected. In Sulawesi, by contrast, the populations using M. micrantha were found in rather restricted areas of the southern lowland (Siwa and Tinokari), but not in the northern part (Manado and the vicinities). No population was from M. micrantha in Lombok. Some beetle populations also occurred on Mikania sp. aff. cordata (Burm. F.) B. L. Robinson at high altitudes (approximately > 600 m altitude) in Java (e.g. Mt Gede, Mt Halimun and Mt Salak in the western part), Bali and southern Sulawesi (Palopo). In Sri Lanka, H. diekei was collected on Mikania scandens B. L. Rob.
The following plants were used in restricted areas.
Leucas lavandulifolia: Infestation of H. diekei was found only in west and central Java, although this plant was sporadically found in other regions, such as west Sumatra (Padang), east Borneo (Balikpapan) and south Thailand (in the Malay Peninsula).
Leucas mollisima: Found only at Purwodadi Botanic Garden in East Java. Leucas sp.: Found only at the foot of Mt Rinjani in Lombok. Plectranthus spp. (this genus was previously assigned as Coleus, Grayer et al. 2010): In Sulawesi, H. diekei was found infesting wild and cultivated Plectranthus plants in the northern and southern parts such as Manado, Palopo and Rantepao (Figure 2). In other regions, Plectranthus plants were also commonly observed, but H. diekei was not observed on this plant.

Morphological analysis
Six body-size-related characters were measured for a total of 67 females and 67 males from seven populations at six localities in four major islands of Indonesia (see Figure 2): three populations from Mikania micrantha (Bogor, Balikpapan and Tinokari), two from Leucas spp. (Bogor and Lombok), and one population from each of Dicliptera sp. (Patuha) and Plectranthus spp. (Rantepao). Of these, two populations from Bogor, West Java, one from Mikania and the other from Leucas occurred side by side; i.e. they were sympatric. All characters were highly variable among and within populations (Table 1). In the principal component analysis, PC1 occupied 64.7% of the total variance, and PC2 occupied 14% (Table 2). Body and elytra length showed high factor loadings to PC1, and elytra height showed a high loading to PC2 ( Table 2). The seven populations were not clearly discriminated by PC1 and PC2 (see Supplementary material, Fig. S1). Hence, we simply compared body length and elytral height across populations for the representatives of PC1 and PC2, respectively ( Figure 3). Body length was significantly variable among populations in females (df = 6, F = 2.624, P = 0.025) but not in males (df = 6, F = 0.837, P = 0.546) (Figure 4). Females of the Balikpapan population had much smaller bodies than other populations. On the other hand, elytra height was variable among populations both in females (df = 6, F = 5.589, P < 0.001) and in males (df = 6, F = 7.462, P < 0.001) ( Figure 5), whereas host use seemed to be unrelated to this morphological divergence (Table 3, Figures 4, 5). In summary, there was a certain morphological divergence among populations specifically in the elytra height for both sexes, but there was no clear morphological distinction among populations with respect to geographic or ecological divergence.
There were significant among-population differences in the number of hairs on the penis guide (df = 6, Kruskal-Wallis χ 2 = 38.127, P < 0.001) and the types of apical edge of penis (df = 6, Kruskal-Wallis χ 2 = 54.121, P < 0.001). Specifically for the latter, there seemed to be a geographic divergence; i.e. populations from Java and Kalimantan, and those from Sulawesi and Lombok. The former population group possessed the emarginate (Type I) structure more frequently, whereas the latter population group possessed the convex (Type III) structure more frequently (Figure 6). The intermediate, truncate (Type II) structure was found in both population groups (see Figure 6). Here again, host utilization seemed unrelated to this morphological divergence (Table 3, Figures 5, 6).

Discussion
The present study showed that Henosepilachna diekei is distributed widely in South East and South Asia, ranging from the Philippines through the Greater Sunda Islands, the western part of the Lesser Sundas, the southern part of the Malay Peninsula, to Sri Lanka. It also appeared that this species uses various plants belonging to Asteraceae, Lamiaceae and Acanthaceae. At the population level, however, this species is monophagous, i.e. each population depends on a single or some closely-related plant species. In consequence, two different populations that depend on different host plants sometimes occur in the proximity. For example, we observed that two populations, one depending on M. micrantha and the other on L. lavandulifolia, occurred side by side in Bogor, West Java. Our laboratory and field-cage experiments revealed that each of these two populations has a very strict host preference to its original plant, M. micrantha or L. lavandulifolia, and there is no reproductive isolation between these two populations except differential host specialization (Fujiyama et al. 2001;Matsubayashi et al. 2011Matsubayashi et al. , 2013aMatsubayashi et al. , 2013b. In our preliminary studies (K. Matsubayashi et al. unpublished), it was shown that beetle populations collected from other plant species also showed very strict preference for the original plants Hence, H. diekei seems to be in the process of radiation in South East Asia in association with the host plant shift.
It is uncertain which plants the ancestral H. dikei population used. At least, M. micrantha, a most widely used plant, and M. scandens, a plant used in Sri Lanka, are not ancestral host plants, because they were recently introduced to South East Asia from South America (Kostermans et al. 1987;Whitten et al. 1996) or North America. On the other hand, the original distributions of other host plants, Mikania sp. aff. cordata, Leucas, Dicliptera and Plectranthus species are still unclear. The speciation and colonization histories of these host plants should be studied. Henosepilachna diekei exhibited geographic variation among populations in elytral height, number of hairs on the penis guide, and the types of apical edge of penis. According to the present morphological analysis, specifically for the types of   apical edge of penis, these seven populations were grouped into two, one including four populations from Java and Kalimantan and the other including three populations from Sulawesi and Lombok ( Figure 6). In addition, males from Bali (10 individuals), Padang in Sumatra (four individuals) and Sri Lanka (two individuals) possessed emarginate or truncate apical edge of penis, a characteristic of the former  group. Hence, it seems that the two population groups of H. diekei are separated by the Wallace line (c.f. Stelbrink et al. 2012; see Figure 6). In our preliminary molecular phylogenetic analysis, it has also been shown that H. diekei populations from the east side of this line and those from the west side form distinct clusters (Matsubayashi 2008;Matsubayashi et al. unpublished data).The Wallace line is assumed to act as a barrier for dispersal in H. diekei populations, as this line acts as a barrier for movements of large animals (Tänzler et al. 2014;Barley et al. 2015).
On the other hand, this line seemed not to affect the evolution of host plant use since diversification in host utilization was commonly found in areas to both the east and west of the Wallace line ( Figure 2). Further population sampling and detailed phylogenetic work are necessary to provide a comprehensive scheme for the evolutionary history of diversification in the morphological characters and host plant utilization in this beetle species.