Cladistic analysis of Sericania ( Coleoptera : Scarabaeidae : Sericini ) – implications for the evolution of the xerophilous fauna of the Himalaya

A cladistic analysis of the species of Sericania Motschulsky, 1860, was executed using fifty-six morphological characters of adults. The monophyly of the genus is supported by the phylogenetic trees generated. Among the three major lineages indicated by the strict consensus tree the East Asian Sericania fuscolineata lineage represents the genus Sericania as defined “originally” and adopted by subsequent authors. The second, the clade Sericania nepalensis group + Sericania sp. 2, is a sister group to the S. fuscolineata clade. Both constitute a sister group to the third major lineage, the Sericania kashmirensis clade, which is endemic in the drier North-West Himalaya where it is the most diverse monophyletic group of Sericini. Provided that the stem species of the S. kashmirensis clade was xerophilous, the origin of this clade can not predate the early Miocene. Based on paleoclimatical and geological data, two competing hypotheses are proposed to explain the evolution of the xerophilous Sericania lineage: (a) a basal splitting within Sericania occurred because of the altitudinal and climatic barrier posed by the Himalaya, which separated the xerophilous lineage in the north (Tibet) from the hygrophilous lineage in the south-east (S slope of Himalaya/ Tibet), or (b) it was a consequence of the increase in the climatic east-west contrast along the southern slope of the Himalaya, which strengthened with the onset of monsoons 8 Ma ago. 517 * Present address: Zoologische Staatssammlung München, Münchhausenstr. 21, D-81247 München, Germany; e-mail: ahrens.dirk_col@gmx.de Staatliches Museum für Tierkunde, Dresden (SMTD); Zoologisches Museum der Humboldt-Universität, Berlin (ZMHB). The choice of taxa included in the ingroup was mainly based on the current classification of the Sericania species (Nomura, 1976) and preliminarily hypothesized relatives (Ahrens, 2004). Fifty-six adult characters were scored. The character states are illustrated in Figs 1 and 2. Phylogenetic analysis The 56 characters (40 were binary and 16 multistate) were all treated as unordered and equally weighted. Inapplicable characters were coded as “–”, while unknown character states were coded as “?” (Strong & Lipscomb, 1999). The parsimony analysis was performed in NONA 2.0 (Goloboff, 1999) using the parsimony ratchet (Nixon, 1999) implemented in NONA with WINCLADA vs. 1.00.08 (Nixon, 2002) as a shell program. Two hundred iterations were performed (one tree held per iteration) repeating the search ten times. The number of characters resampled during the parsimony ratchet was determined to be five. All searches were done under the collapsing option “ambiguous”, which collapses every node with a minimum length of 0. State transformations were considered to be apomorphies of a given node only if they were unambiguous (i.e., without arbitrary selection of accelerated or delayed optimization) and if they were shared by all dichotomised most parsimonious trees. The use of resampling methods (particularly the bootstrap) is questioned as a means of assessing confidence limits of phylogenies (Carpenter, 1992, 1996; Kluge & Wolf, 1993). This is principally because these tests rely on the assumption that the characters are “independently and identically distributed” (Felsenstein, 1985) and that they have been randomly sampled – two assumptions violated by most phylogenetic data. Morpho518 ZMHB Ta-pai Shan (Shaanxi, China) Sericania sp. 2 SMTD Tatsienlu (Sichuan, China) Sericania sp. 1 CA Japan-Ussuri S. yamauchii Sawada, 1938 MHNG, HNHM NW Himalaya S. torva Ahrens, 2004 MHNG NW Himalaya S. swatensis Ahrens, 2004 SMTD NW Himalaya S. poonchensis Ahrens, 2004 MSNG NW Himalaya S. piattellai Ahrens, 2004 MNHN NW Himalaya S. pacis Ahrens, 2004 CA Central Nepal S. nepalensis (Frey, 1965) CA Central Nepal S. mela Ahrens, 2004 CA Central Nepal S. mara Ahrens, 2004 MHNG NW Himalaya S. loebli Ahrens, 2004 CA Japan S. lewisi Arrow, 1913 NHMB, CA NW Himalaya S. laeticula (Sharp, 1878) BMNH, CA Central Nepal S. kleebergi Ahrens, 2004 HNHM NW Himalaya S. khagana Ahrens, 2004 ZMHB, CA NW Himalaya S. kashmirensis (Moser, 1919) CA NW Himalaya S. heinzi Ahrens, 2004 NHMB, CA NW Himalaya S. hazarensis Ahrens, 2004 NHMB, CA NW Himalaya S. gilgitensis Ahrens, 2004 CA North-Eastern China S. fuscolineata Motschulsky, 1860 NHMB NW Himalaya S. dubiosa Ahrens, 2004 NHMB NW Himalaya S. dispar Ahrens, 2004 ZMHB, CA W Himalaya S. costulata (Moser, 1915) ZMHB ? NW Himalaya S. carinata Brenske, 1898 SMNS Eastern Nepal S. bhojpurensis Ahrens, 2004 MHNG NW Himalaya S. besucheti Ahrens, 2004 MNHN, CA NW Himalaya Sericania babaulti Ahrens, 2004 CA Himalaya to northern Indochina Chrysoserica auricoma (Brenske, 1896) CA USA, Canada Stilbolemma sericea (Illiger, 1802)† CA Manchuria, Korea Nipponoserica koltzei Reitter, 1897 CA Nepal N. procera rufescens Frey, 1965† CA Darjeeling N. goomensis Ahrens, 1999† CA Nepal Nepaloserica procera procera Frey, 1965 CA Europe, Siberia Maladera holosericea (Scopoli, 1772)† CA Southern Africa Pleophylla Erichson, 1847 spec. Depository Distribution Species TABLE 1. List of the species studied for the cladistic analysis († not included in the analysis).


INTRODUCTION
Since the breaking up of Gondwana tectonic and orogenic processes have influenced climate and evolution in South Asia.The accretion of the Indian plate and its subduction under the Asian plate led not only to extensive volcanism and orogenesis but also changed extensively the distribution of sea and land, wich caused modifications in wind circulation and consequently the climate in Asia (Ramstein et al., 1997).
The most recent attempts (e.g.Dobremez, 1976;Martens, 1983Martens, , 1993) ) to investigate the patterns of biodiversity in the Himalayan orogenic belt, reveal a trend in impoverishment of the fauna and flora along the Himalayan chain from east to west.This trend also occurs in the beetle tribe Sericini.For any investigation of the evolution of the Himalayan fauna important information may be provided by subordinate patterns, such as those found in many xerophilous forms, invading the Himalaya from the west (Mani, 1968;Martens, 1993) and becoming dominant elements of the fauna and flora of the alpine belt.However, alpine elements appear to be missing among the Sericini (Ahrens, 2004).The upper altitudinal occurrence of Sericini is limited by the absence of a soil layer with humus and plant roots.For that reason, these beetles have not occupied high altitudes.
Although very poor in terms of species richness, a similar general pattern of xerophilous organisms is reported among the montane forms of Sericini (Ahrens, 2004), which achieved in the NW Himalaya a higher diversity than in other regions of the Himalaya.One of these groups is the genus Sericania, with one major lineage (S. kashmirensis group) occurring in the montane belt of North-West Himalaya of which some species extend their ranges further east into the central Himalaya.
Sericania is with 65 known species a comparatively diverse sericine taxon.The taxonomy and distribution of the Himalayan species was revised recently (Ahrens, 2004) and the taxa of the Asian mainland were reviewed by Nomura (1976).The cladistic analysis presented here was conducted to further the understanding of the effect of factors such as orogenesis, climate shifting or glaciation on evolution of organisms and in particular improve our knowledge of the relationships among the Sericini.
Staatliches Museum für Tierkunde, Dresden (SMTD); Zoologisches Museum der Humboldt-Universität, Berlin (ZMHB).The choice of taxa included in the ingroup was mainly based on the current classification of the Sericania species (Nomura, 1976) and preliminarily hypothesized relatives (Ahrens, 2004).Fifty-six adult characters were scored.The character states are illustrated in Figs 1 and 2.

Phylogenetic analysis
The 56 characters (40 were binary and 16 multistate) were all treated as unordered and equally weighted.Inapplicable characters were coded as "-", while unknown character states were coded as "?" (Strong & Lipscomb, 1999).The parsimony analysis was performed in NONA 2.0 (Goloboff, 1999) using the parsimony ratchet (Nixon, 1999) implemented in NONA with WINCLADA vs. 1.00.08 (Nixon, 2002) as a shell program.Two hundred iterations were performed (one tree held per itera-tion) repeating the search ten times.The number of characters resampled during the parsimony ratchet was determined to be five.All searches were done under the collapsing option "ambiguous", which collapses every node with a minimum length of 0. State transformations were considered to be apomorphies of a given node only if they were unambiguous (i.e., without arbitrary selection of accelerated or delayed optimization) and if they were shared by all dichotomised most parsimonious trees.
The use of resampling methods (particularly the bootstrap) is questioned as a means of assessing confidence limits of phylogenies (Carpenter, 1992(Carpenter, , 1996;;Kluge & Wolf, 1993).This is principally because these tests rely on the assumption that the characters are "independently and identically distributed" (Felsenstein, 1985) and that they have been randomly sampledtwo assumptions violated by most phylogenetic data.Morpho-  1. List of the species studied for the cladistic analysis ( † not included in the analysis).logical data sets are particularly susceptible to these problems as they are often replete with examples of redundant (inapplicable) and correlated characters, although molecular data sets are not immune from this (e.g.ribosomal RNA genes, which code for a secondary structure).However, as an estimate of the robustness of a data set, these resampling techniques can be useful for discovering ambiguity between characters.
Bremer support (Bremer, 1988(Bremer, , 1994) ) and bootstrap values (Felsenstein, 1985) were evaluated using NONA.Bootstrap analyses of data were performed with 200 replicates using TBR branch swapping.The bootstrap values did not augment significantly when uninformative characters were deleted from the matrix.The search was set to a Bremer support level of 12, with seven runs (each holding a number of trees from 100 to 500 times multiple of suboptimal tree length augmentation) and a total hold of 8000 trees.Character changes were mapped on the consensus tree using WINCLADA.
Successive weighting (Farris, 1969) was used to evaluate further phylogenetic relationships.This method uses post hoc character weighting based on the fit of each character as applied to the trees currently in the memory.Thus, the 'quality' of the character data is used rather than intuitive feeling regarding weighting of characters.Although this method increases the assumptions in the analysis (Siebert, 1992), it is useful for ana-lysing phylogenetic pattern when characters exhibit a high level of homoplasy.Characters were reweighted based on the consistency index.The base weight was set at 100.Weights were inserted manually into NONA parsimony ratchet search via WINCLADA.Tree searches continued until the character weights no longer changed (Farris, 1988) or until identical trees were found in consecutive searches (indicating stability in the trees).

RESULTS
The analysis of 56 adult characters with the parsimony ratchet, the above mentioned settings and repeating the search ten times yielded 102 equally parsimonious trees of 171 steps [ensemble consistency index (CI): 0.43, ensemble retention index (RI): 0.72].Characters 22, 45, and 49 proved uninformative.The strict consensus of these trees is presented in Fig. 3 with areas of conflict in topology shown as polytomies.The strict consensus tree (Fig. 3) of equally weighted characters exhibits a high level of polytomy in most nodes of the Sericania kashmirensis group.
Within the monophyletic clade Sericania (node A) the strict consensus tree (Fig. 3) indicates three major lineages: (1) the Sericania fuscolineata group; (2) the Sericania nepalensis group + Sericania sp.2; and (3) the Sericania kashmirensis group.The topology of the strict consensus tree did not reveal a sister taxon of Sericania .In comparison to the apical nodes of the genus, the sup-524 port for the node Sericania is relatively low (node A: Bremer support: 1, bootstrap value: 46%).The information in the high number of equally parsimonious trees that resulted from the parsimony ratchet with equally weighted characters, is summarized by the majority rule consensus tree (Fig. 4).
The use of successive approximations character weighting (SACW, Farris, 1969) based on consistency index greatly reduced the number of most parsimonious trees (MPTs), producing a significantly more resolved strict consensus tree, especially among the representatives of the Serica kashmirensis group.In this approach weights no longer changed after two iterations.Six MPTs of 172 steps (CI: 0.55, RI: 0.78) resulted from the analysis, whose strict consensus tree was one step longer (Fig. 5).Although numerous nodes are well supported, especially those of the species groups indiated by the analysis with unweighted characters (Fig. 3, nodes B, C, D), the bootstrap values (calculated using character weights) for the several basal and apical nodes are low even after SACW.

Characters and computer analysis
The overall structure of the equally weighted majority rule consensus tree (Fig. 4) is the same as that indicated by the SACW analyses (Fig. 5), with one separate clade including Nepaloserica, Chrysoserica, Nipponoserica and Serica, and a second comprising all species of Sericania.Only in the latter clade is the topology altered slightly.The evolution of characters and evolutionary diversification of Sericania are discussed using the SACW strict consensus tree.Figure 6 illustrates the character changes along each branch under deltran optimization.
Characters of external morphology are consistent with the tree topology of the similar analyses based on male genital characters, showing both similar consistency index (ci) and retention index (ri) means.

Monophyly of Sericania
Although not exhaustive, this study is the most comprehensive phylogenetic analysis of Sericania ever conducted.Due to the little variability in the genital features (Nomura, 1976), many of the Far East species of Sericania were not included in the cladistic analyses.All of them, however, share the same apomorphic states at node B as the taxa included.The results improve the current understanding of the relationships among the Asian Sericini.
The genus Sericania established by Motschulsky (1860) for a single species, S. fuscolineata, was comprised until recently only East Asian species, mainly from the Japanese Archipelago (Medvedev, 1952;Nomura, 1976).However, some of these species were reported recently from northern China, and moreover, taxa from the Himalaya were also assigned to this genus (Ahrens, 2004).
The search for the sister taxon of Sericania, using the tree topology of the present analyses was unsuccessful.This might be a consequence of our poor knowledge of the Asian sericine fauna.But it is also evident from other similar studies (Ahrens, 2006 and unpubl. results), that the morphological characters of Sericini are only little suitable to resolve basal nodes of major species groups or 526  Its monophyly is best supported by the following apomorphies (see node B, Fig. 6): ( 1) metacoxa transversely impressed (19 : 1); and (2) basal metatarsomere as long as the subsequent tarsomere (28 : 1).The species of this clade are so far only reported from the Far East Asia, but a still undescribed taxon (Sericania sp. 1) is reported here also from the eastern margin of the Tibetan Plateau.This species is nested in a node among the representatives of this group.
The sister taxon relationship of the Sericania fuscolineata group with the second major clade of Sericania, the Sericania nepalensis group + Sericania sp. 2, is based upon (1) the scutellum compared to elytral base narrow (18 : 1); and, as apparent under DELTRAN and ACC-TRAN optimization (Fig. 6), ( 2) the metatibia at apex interiorly close to tarsal articulation sharply incised and strongly truncate (25 : 2).The male genitalia of Sericania sp. 2 from Shaanxi (China) have in several apomorphies that differ notably from the basic pattern of the Sericania fuscolineata and Sericania nepalensis groups, which is reflected in the little support for the branch of Sericania nepalensis group + Sericania sp. 2. Consequently, I prefer not to assign Sericania sp. 2 to the formally established and well supported Sericania nepalensis group (node C).The areas, where Sericania sp. 1 and sp. 2 occur, are still poorly investigated, and additional representatives of further lineages are likely to be discovered.The representatives of the Sericania nepalensis group are restricted to the central Himalaya, where they diversified little compared to other groups endemic to the Himalaya (Fig. 7).
According to the phylogeny indicated by this cladistic analysis, the Himalayan Sericania species are not a monophyletic group, but two separate lineages.The S. nepalensis clade, is nested within a clade composed in major part of East Asian species (Fig. 8), and which is sister taxon to the second Himalayan clade, the Sericania kashmirensis group.This reveals that faunal exchange must have occurred at least twice, but fossils and a better knowledge of the Chinese sericine fauna are needed to determine the direction.
The Sericania kashmirensis group species are mainly restricted to the Indus Himalaya (Troll, 1972).A few further east.Ranges of many of its taxa are based on a fragmentary exploration of the areas which results in widely isolated occurrences (Fig. 9), such as in S. babaulti, S. laeticula, and S. hazarensis.In several cases closely related species (sister taxa) occur sympatrically (Fig. 9), such as S. laeticula and S. poonchensis, or are close neighbours, as in S. gilgitensis and S. piattellai, S. heinzi and S. kashmirensis, S. costulata and S. torva, and S. besucheti and S. swatensis.There are no endemic forms east of the Indus Himalaya.The Indus Himalaya there is an arboreal belt, unlike in all other regions of the Himalaya, which is delimited at high altitude by the alpine vegetational belt, and at lower altitude (below ~ 2000 m) by desert steppe (Troll, 1972;Fig. 10).Most of the species of Sericini dwell in arboreal habitats (Ahrens, 2004) and avoid the alpine belt.This upper and lower altitudinal vegetational delimitation is supposed to be important for survival during periods of climatic deterioration, or dispersal to adjoining areas.This might explain, why species adapted to aridity are restricted to the north-western Himalaya and found not in drier regions of the Tibetan Highlands or inner valleys of the central Himalaya, and why closely related species occur in neighbouring areas.However, in extremely arid habitats the pressure of com-petition from hygrophilous eastern elements of Himalayan sericine fauna is absent.
Interiors of large continents are normally arid because of their great distance from sources of ocean moisture, but sediments deposited in the western Loess plateau of East Asia indicate these areas were fluvial-lacustrine during the Oligocene epoch (Wang, 1985) and had a more humid climate.One source of moisture during the Oligocene in the Asian interior may have been the Paratethys sea in western Central Asia (Ramstein et al., 1997).As this inland sea shrank during the late Oligocene and early Miocene, the Asian interior have become more arid and the climate more continentaly, with hotter summers and colder winters.The onset of greater aridity recorded by the early Miocene loess (Guo et al., 2002) in the Qinan deposits in northern China (Gansu province) is broadly consistent with this change in land-sea distribution.These Qinan deposits indicate that by the early Miocene the winter wind pattern was similar to the modern Asian winter monsoon.
Based on the present distribution of the Sericania kashmirensis clade in the drier parts of the North-West Himalaya it is hypothesized that the stem species of this clade was also xerophilous and less adapted to the humid conditions, which today prevail on the southern slopes of the central and eastern Himalaya.Provided that the stem species (Fig. 6, node D) of the S. kashmirensis clade was xerophilous, this clade will not predate the early Miocene.One conceivable hypothesis is that the extension of distribution of the species of the S. kashmirensis clade was former wider along the northern margin of the rising Tibet, which possibly achieved a high elevation much later (Pliocene to Quaternary) than the southern margin (Tapponier et al., 2001), which reached considerable elevation already in the Eocene / Oligocene.The climatic and altitudinal divide along the Tertiary orogenic belt (southern Tibet, Himalaya) might have been an effective barrier separating the two basal lineages of Sericania.The highest elevation could have been achieved by early Miocene, since fossil plant assemblages provide evidence that the altitude in parts of the southern Tibetan Plateau probably has remained unchanged for the past 15 Ma (Spicer et al., 2003).The present range of the S. kashmirensis clade would constitute just a relic of a formerly more extensive northern distribution.
On the other hand, the occurrence of taxa of the sister lineage of the S. kashmirensis clade (S. nepalensis group) on the southern slopes of the Himalaya, indicate that the separation of the basal Sericania lineages may have occurred along the southern slope of the Himalaya / Tibet and probably coincided with the onset of monsoons arround 8 Ma (Prell et al., 1992, Quade et al., 1989).The east-west contrast in climate along the southern slope of the Himalaya have strengthened significantly since that time (Ramstein et al., 1997, Ruddiman & Kutzbach, 1989;Manabe & Broccoli, 1990;An et al., 2001).
Considering the stem lineage taxa of Sericania more hygrophilous, the sister taxon to the S. kashmirensis clade probably evolved close to the south-eastern margin of the present Tibetan Highlands and the stem lineage of the S. nepalensis group possibly dispersed westards along the southern Himalayan slope, and the ancestral taxa of the S. fuscolineata group moved to the north-east.The latter have strongly diversified on the Japanese Archipelago.Troll, 1972) where the vast majority of the species of the Sericania occur.S. = Sericania.
The preliminary results and discussions presented here are part of a wider study of the sericine beetle component of the fauna of the Tertiary orogenic belt.Further exploration of the Asian Fauna of Sericini, the discovery of new taxa and the identification of the sister taxon of Sericania (in further phylogenetic studies) will help to determine the origin of the stem species of Sericania and, in particular, provide evidence for one of the two hypotheses for the evolution of the xerophilous S. kashmirensis lineage.

Fig. 7 .
Fig. 7. Phylogeny of the Sericania nepalensis group in its geographical framework, with indicated distribution of the species, and 2000 meter altitude line.S. = Sericania.

Fig. 8 .
Fig. 8. Phylogeny of Sericania in its geographical framework.Cumulative generalized distribution of the principal lineages (with bold tree branch lines)is shown, with the respective ranges indicated as follows: S. kashmirensis-group -continuous line and shaded from right to left; S. nepalensis-group -continuous line and shaded from left to right; S. fuscolineata-group -dotted line.S. = Sericania. 529

Fig. 10 .
Fig. 10.Altitudinal distribution of the taxa of the Sericania kashmirensis clade (S. gilgitensis omitted due to lack of data) in the context of the climatic and vegetational zonation of the Indus Himalaya (according toTroll, 1972) where the vast majority of the species of the Sericania occur.S. = Sericania.

TABLE 2 .
Data matrix for the cladistic analysis.Inapplicable characters are coded as "-" and missing character states as "?".All characters are equally weighted and nonadditive.