Diorhabda, Weise, 1883

Tracy, James L. & Robbins, Thomas O., 2009, Taxonomic revision and biogeography of the Tamarix-feeding Diorhabda elongata (Brullé, 1832) species group (Coleoptera: Chrysomelidae: Galerucinae: Galerucini) and analysis of their potential in biological control of Tamarisk, Zootaxa 2101 (1), pp. 1-152 : 18-39

Diorhabda elongata- Group tamarisk beetles

Diagnosis. The following diagnostic characters for tamarisk beetles are partly based upon keys to Diorhabda species for central Asia (Gressitt and Kimoto 1963a, Lopatin 1977a) and the Mediterranean region ( Warchalowski 2003), and descriptions of certain south Asian Diorhabda species ( Jacoby 1886, 1894; Maulik 1936). Tamarisk beetles may be distinguished from other species of Diorhabda by the following combination of characters: (1) the pronotum and elytra are glabrous or with very sparse setae, except for the pubescence of the epipleura; (2) the lateral borders of the prothorax are narrow, not flattened, with the posterior angles obtuse and distinct, not rounded or quadrate; (3) the elytral punctation is denser and finer than that of the pronotum; and (4) the elytron bears only a distinct lateral carina extending from the humeral calus. The description of external characters for adult D. carinulata (as D. e. deserticola ) by Lewis et al. (2003b) applies generally to all species in the group, excepting measurements and coloration, which are discussed below in each species account along with genitalia.

Included species. We recognize five sibling genitalic morphospecies of tamarisk beetles as forming the D. elongata group: D. elongata , D. carinata , D. sublineata , D. carinulata and D. meridionalis ( Figs. 1–45 ).

Distribution. The D. elongata species group is broadly distributed over the majority of the distribution of Tamarix from North Africa, southern Europe to central Asia (Map 1). Although tamarisk is also native to southern Africa, Israel and Palestine, and Southeast Asia ( India, Bangladesh, and Burma) ( Baum 1978), tamarisk beetles are unreported from these areas. Detailed distribution data with maps are discussed under the heading Distribution in each species account.

Discussion. Taxonomy. Diorhabda elongata Brullé (as Galeruca ) was originally described from the Pelopónnisos peninsula of Greece in 1832 ( Brullé 1832). From 1858 to 1893, four other species described under Galeruca were synonymized under D. elongata (Reiche and Saulcy 1858, Weise 1893): (1) D. carinata ( Faldermann, 1837) (of the Transcaucasus), (2) Galeruca sublineata ( Lucas, 1849) (of Annaba, Algeria), which has been regarded as the subspecies D. e. sublineata (Lucas) (Gressitt and Kimoto 1963a) , (3) D. carinulata carinulata (of Sarepta, Russia), and (4) G. costalis ( Mulsant, 1852) (of southwest Turkey). We compare specimens of these four taxa collected from the vicinity of type localities to literature descriptions of type specimens. We similarly studied specimens corresponding to the subspecific taxa D. e. deserticola Chen (1961) (of Yuli, China; also examined paratypes), D. carinulata meridionalis Berti & Rapilly (of Minab, Iran), and D. e. carinata (Faldermann) (from central Asia) as regarded by Bechyné (1961). We characterize the variability of the genitalia and some external characters (size and elytral markings) for each taxon using specimens from the type localities (topotypes) where possible. We compare and match specimens of tamarisk beetles from throughout southern Europe, Africa and west and central Asia with the various taxa above, providing data on variability in genitalia across the entire distribution of each taxon.

In our comparisons of the above taxa of tamarisk beetles, we examined male and female genitalia of 784 specimens from 37 countries. We distinguish five qualitatively distinct endophallic morphotypes based on endophallic sclerites. We also find geographically corresponding female genitalic morphotypes with distinct differences in internal sternite VIII and vaginal palpi. These five fully diagnosable genitalic morphotype pairs have additional nondiscrete differences in ranges of body length for each sex and, in some cases, striping of the elytra. Intermediate forms between these species are not seen in the extensive material examined, even in several instances of sympatry and syntopy (sharing individual tamarisk trees as habitat). Most species appear to be at least parapatric or marginally sympatric and some are moderately sympatric. Among species with overlapping or abutting geographic ranges, the absence in nature of intermediate hybrid genitalic forms along a geographic cline towards areas of range contact constitutes a morphological hiatus that is inconsistent with a status of interbreeding subspecies (see Krysan et al. 1980, Patten and Unitt 2002) or geographic races. Although additional material would be desirable in the areas of abutting ranges of certain species such as in western Italy and and central Turkey, we believe the data available sufficiently document the lack of intermediate forms between species. The numbers of diagnostic genitalic differences between the few allopatric species are comparable with the number of diagnostic differences seen between moderately sympatric species. Consequently, a status of potentially interbreeding subspecies or races is also inconsistent for the few species living in allopatry (see Helbig et al. 2002). The maintenance of distinguishing genitalic characteristics in four species in culture under identical laboratory conditions is contraindicative of a status of intraspecific morphs, such as might occur in cases of genitalic polymorphism (e.g., Jocqué 2002) or phenotypic plasticity (e.g., Agrawal 2001). Distinct qualitative differences and morphometric discontinuities in genitalic structures distinguishing species are evidence of strong reproductive isolation in nature ( Helbig et al. 2002). We find that these morphotype pairs are a complex of five fully diagnosable sibling genitalic morphological species (morphospecies) of Diorhabda . Further evidence for reproductive isolation between the four species D. elongata , D. carinata , D. sublineata , and D. carinulata is also found in differences in component ratios of putative aggregation pheromones (discussed below under Biology - Aggregation Pheromones) and reduced laboratory F2 hybrid egg viability (discussed below under Experimental Hybridization).

We characterize the genitalia of D. elongata and corroborate the restoration of the species D. carinata and D. carinulata and their removal from synonymy with D. elongata by Berti and Rapilly (1973). Specimens identified by Bechyné as D. e. carinata are conspecfic with D. carinata . The following four taxonomic changes are made: (1) Galeruca sublineata Lucas is removed from synonymy with D. elongata and restored as the species D. sublineata (Lucas) ; (2) D. e. deserticola Chen is synonymized under D. carinulata ; (3) D. koltzei ab. basicornis Laboissière is synonymized under D. carinulata ; and (4) D. carinulata meridionalis Berti & Rapilly is elevated to species status as D. meridionalis Berti & Rapilly. The color variant D. e. ab. bipustulata Normand is synonymized under D. sublineata . Five sibling species (with no subspecies) are recognized as forming the D. elongata group: D. elongata , D. carinata , D. sublineata , D. carinulata and D. meridionalis . Each species of tamarisk beetle exhibits some degree of sympatry with at least one of the other species. Four of these five tamarisk beetle species were originally described in the 1800’s and one was described as a subspecies, but all have been identified and published under the name D. elongata by various workers (e.g., Wilcox 1971, Riley et al. 2003, Warchalowski 2003, Bie ṅkowski 2004, Lopatin et al. 2004).

The five species of the D. elongata group may be distinguished by a combination of eight discrete, or near discrete, genitalic characters, involving the forms of the male endophallic sclerites and female vaginal palpi ( Table 5), and the additional genitalic character of the form of female internal sternite VIII. Males of all species are distinguishable by the forms of one or more of three endophallic sclerites (PES, EES, and CES; Figs. 14–33 ). A combination of characters involving the forms of the vaginal palpi (VP) and internal sternite VIII (IS VIII) can be used to distinguish between females of three species, D. elongata , D. carinulata and D. meridionalis ( Figs. 34–43 ). Females of two species, D. carinata and D. sublineata , are distinguished from the three other species by these same characters, but only in some cases are they distinguishable from one another, and only by the form of internal sternite VIII ( Figs. 40–41 ). We find no differences in the morphology of the spermatheca ( Figs. 34–38 ; SP) or median lobe ( Figs. 14–18 ; ML) and tegmen (not shown) of the aedeagus that are sufficient for species diagnosis.

Significant patterns in clinal geographic variation in the five species are not seen and, therefore, no subspecies are recognized. We provide keys for the sexing of adults, followed by keys to species for each sex based upon the genitalia.

We are unable to distinguish the five sibling species of the D. elongata group solely on the basis of external characters. The presence of elytral vittae that extend into the basal half of the elytra (SMV and SSV; Figs. 5, 9 ) can be used to eliminate an identification of D. elongata , in which the elytral vittae, if present, are confined to the apical half of the elytra ( Fig. 1 ). Differences in ranges of body length for each sex ( Table 2) can aid in species identification of some individuals when used with distribution data (Map 1). Further investigation might reveal obscure external characters useful in separating some species such as external setation (e.g., Konstantinov and Lopatin 2000).

a Ranks of values followed by the same letter within the same column for each sex are not significantly different. The ranks of values for females are significantly larger than those of males with no sex by species interactions (P> 0.05; Two-Way Kruskal-Wallis Tests on ranks using PROC GLM-LSMEANS test; SAS Institute 2005).

b Species by sex interaction is significant and ranks of values followed by the same letter within the same column (including both sexes) are not significantly different (P> 0.05; Two-Way Kruskal-Wallis Tests on ranks using PROC GLM-LSMEANS test; SAS Institute 2005).

c Combined width of both elytra at widest point.

The endophalli (END) of the D. elongata group consist of an elongated tubular sac bearing one to three sclerites and various characteristic bulgings ( Figs. 14–18 ). The sacs widen distally in D. elongata , D. carinata and D. sublineata , but narrow distally in D. carinulata and D. meridionalis (see Fig. 19d–e of Berti and Rapilly 1973 for an illustration of the fully exerted narrow tip of the endophallus in D. meridionalis ). All species have a conspicuous palmate shaped dorsal sclerite, the palmate endophallic sclerite (PES), at the base of the sac that is armed subdistally or distally with one to several spines (or teeth) ( Figs. 29–33 ). All species also have an elongate shaped ventral sclerite, the elongate endophallic sclerite (EES), which runs along most of the length of the sac. The elongate sclerite also bears a raised blade which is armed with one to several spines (or teeth) ( Figs. 19–23 ). The shapes of these two sclerites and the arrangement of the spines are useful in species diagnosis. Only D. sublineata posseses a third lateral sclerite that connects the palmate and elongate sclerites, the connecting endophallic sclerite ( Figs. 16 , 21 , 31 ; CES). Our illustrations ( Figs. 14–18 ) are of uninflated endophalli. Studies of inflated endophalli, such as obtained from mating pairs freshly killed by hot water, could reveal that the conformations of bulgings of the endophallus are also useful in species diagnosis. Scanning electron micrography is being done for the endophallic sclerites of the D. elongata species group by Jessica Perez, Roxana Reyna-Islas, and Dave Thompson at NMSU using some of our dissected material.

Silfverberg (1974) found characters of the endophallic sclerites useful in helping differentiate between the genera Galerucella , Pyrrhalta , and Xanthogaleruca in the tribe Galerucini . Endophalli in the D. elongata group closely resemble those of several Galerucella spp. ( Figs. 1a, 4, 5 of Silfverberg 1974) which are also in the form of a tubular sac. Endophalli of Galerucella differ in bearing only a dorsal sclerite which is elongate, not palmate, in shape. The endophallus of D. lusca with its large serrated flagellum (Fig. 97 of Mann and Crowson 1996) is strikingly different from those in both the D. elongata group and Galerucella , and the placement of D. lusca in the genus Diorhabda should be reviewed. Spermathecae of the D. elongata group ( Figs. 34–38 ; SP) more closely resemble those of some Galeruca spp. ( Figs. 1d, 2d of Beenen 1999) than the more slender spermathecae of some Galerucella spp. ( Fig. 6 of Hippa and Koponen 1986), but differ from both in bearing a pointed appendage at the distal tip (similar to that in North American Ophraella , but more prominent; LeSage 1986). Additional studies are needed of the morphology of the endophalli and its sclerites and the female genitalia (vaginal palpi, spermathecae and internal sternite VIII; Figs. 34–38 ) in other species of Diorhabda and related genera. Such genitalic studies should be useful in any reviews of subtribal, generic, or specific classifications within the tribe Galerucini .

Experimental Hybridization. D. C. Thompson et al. (in prep.) have conducted laboratory crossing studies between D. elongata and each of the three sibling species D. carinulata , D. sublineata , and D. carinata , and between D. carinulata and D. carinata (also Dave C. Thompson and Beth Peterson, New Mexico State University, Las Cruces, NM, pers. comm.). Additional crossing studies between D. elongata and D. carinulata have been conducted by some of us (DeLoach and Tracy unpublished data) and Julie Keller and Dan Bean (pers. comm.). These experiments are the subject of pending publications and involve interspecific pure line crossings, hybrid crossings, and concurrent pure line conspecific control crossings of laboratory reared virgin adult male and female beetles in no choice situations confined in vials with fresh tamarisk leaves for food (n = ca. ten replications per type of cross). The mean percent viability of the F 1 eggs produced from crossing the parental females and viability of the F2– F 3 eggs from crossing progeny of the parental females were recorded. Diorhabda colonies used in these studies were found to be free of Wolbachia bacterial infections (D. Kazmer pers. comm.) that can reduce fertility in crosses between chrysomelid subspecies, such as is seen in the subspecies of Diabrotica virgifera LeConte ( Giordano et al. 1997) .

In a no-choice laboratory environment, matings readily occurred between pure line D. elongata and the three other Diorhabda spp. ( D. carinulata , D. carinata , and D. sublineata ) and between D. carinulata and D. carinata , producing F1 hybrid eggs with 50–100% viability of control crosses, depending on parental species and their sex. In ♀ D. carinata × ♂ D. carinulata crossings, F1 hybrid egg viability was reduced by over 50% from that of pure line control crosses. But no F 1 eggs were viable from ♀ D. carinulata × ♂ D. carinata crossings in which 90% of females died in copulo. A high incidence of female death in copulo also occurred in ♀ D. carinulata × ♂ D. elongata crossings and was associated with a 30% reduction in F 1 egg viability over that of controls, but percent F 1 egg viability was normal for ♀ D. elongata × ♂ D. carinulata crossings. Most surviving females failing to produce viable eggs in interspecific crosses probably failed to mate, but fertilization could also have produced inviable eggs. The larger size of male D. carinata and D. elongata , along with the larger more prominent distal basal facing spines of their palmate endophallic sclerites ( Figs. 14–18 ), may be a factor in their observed difficulty disengaging from copulation with female D. carinulata and the associated high in copulo death rate for female D. carinulata .

Reductions in F 2 egg viability of ca. 50–100% from that of controls were found in at least one of the male/female crossing combinations for all interspecific hybrid/hybrid progeny crosses. Observed reductions in F1 and/or F2 hybrid egg viabilities is a form of hybrid breakdown (the opposite of hybrid vigor) providing further evidence for reproductive isolation between all the crossed Diorhabda species. Laboratory hybridization studies between the most morphologically similar species, D. sublineata and D. carinata , have not yet been conducted. However, these species strongly and asymmetrically differ with respect to the F1 and F2 hybrid egg viabilities resulting from the two combinations of male/female crosses of each species with D. elongata (D.C. Thompson and B. Peterson, pers. comm.), providing additional evidence of reproductive isolation. For example, in ♀ D. elongata × ♂ D. sublineata crosses, F 1 egg viabilities are close to that of control crosses, but in ♀ D. elongata × ♂ D. carinata crosses, no F 1 egg are viable.

D. Thompson and B. Peterson (pers. comm.) found that hybrid breakdown in the form of reduced egg viability over that of controls is much less apparent in F2 and F3 progeny of laboratory produced hybrid/ hybrid crosses of D. sublineata × D. elongata (0–60% reductions) compared to that of F2 progeny produced from other hybrid/hybrid crosses in the D. elongata group (90% reductions). For example, egg viability of F2 D. elongata × D. carinulata hybrids is reduced by ca. 90–100% over that of pure line controls. However, egg viability of ♀ D. sublineata × ♂ D. elongata hybrids is reduced by about 60% from that of pure line controls for the F2 generation and it is reduced by about 50% for the F3 generation. In ♀ D. elongata × ♂ D. sublineata hybrids, egg viability is about the same as pure line controls in the F2 generation, but is reduced by about 25% in the F3 generation, possibly as a result of additional hybrid breakdown between generations. Colonies of D. sublineata × D. elongata hybrids can be easily maintained for at least six generations in the laboratory (D. C. Thompson and B. Peterson, pers. comm.) and F1 and F2 hybrids can have diagnostic anomalous hybrid character combinations (see D. sublineata × D. elongata Hybrid Morphology under species account for D. sublineata ). Egg viability has never been measured in F4 or later generation hybrids, and, over several more generations of hybrid inbreeding, additional hybrid breakdown might lead to severe reductions in egg viability. Egg viability in backcross D. sublineata / D. elongata hybrids has not yet been characterized. With D. carinulata / D. elongata hybrids, egg viability appears normal when hybrid females are backcrossed with pure line males but there is no egg viability when hybrid males are backcrossed with pure line females (DeLoach and Tracy unpublished data). Additional forms of hybrid breakdown may occur in field situations (discussed below).

Reduced F 2 egg viability is also seen in laboratory hybrids of the chrysomelids Altica carduorum Guerin- Meneville and A. cirsicola Ohno ( Laroche et al. 1996) . The chrysomelids Diabrotica longicornis (Say) and D. virgifera will also hybridize in the laboratory and produce fertile F1 and F2 hybrids. However, interspecific matings between these Diabrotica spp. are very rare in the field and prezygotic reproductive isolation apparently is operating through differences in adult feeding habits, pheromones, and phenologies (Hintz and George 1979). Interspecific laboratory crossings of sympatric populations of Galerucella nymphaeae and “ G. sagittariae ” from different host plants in Finland produce fertile F1 hybrids, but there is no cytological evidence of hybridization in nature (Nokkala and Nokkala 1998). The production of laboratory hybrids able to reproduce for several generations is not evidence for extensive hybridization and gene flow in nature (see Helbig et al. 2002). We have seen no morphological intermediates in nature that give evidence of hybridization between any species of the D. elongata group (for further discussion and characterization of male genitalia in laboratory produced hybrids, see D. sublineata × D. elongata Hybrid Morphology under D. sublineata species account). In addition, the genitalic morphology of laboratory produced male hybrids of D. sublineata and D. elongata cannot be attributable to other known species of Diorhabda . The lack of intermediate genitalic morphotypes between these species in nature, especially in areas of sympatry and parapatry, is strong evidence of reproductive isolation.

Reproductive isolation in sympatric Diorhabda is probably maintained by prezygotic isolating mechanisms such as differing mate-recognition systems which may be reinforced by differences in aggregation pheromones (see Biology – Aggregation Pheromones below). It is possible that some of these Diorhabda species may hybridize rarely in nature and have limited flow between certain genes, as is seen in some other closely related animal species ( Helbig et al. 2002, Mallet 2005, Bull et al. 2006). For example, several species of passion-vine butterflies, Heliconius ( Lepidoptera : Nymphalidae : Heliconiinae ), will interbreed and produce fertile hybrids in hybrid zones, but genetic analyses reveal strong barriers to interspecific gene flow, probably as a result of differing mate recognition systems ( Jiggins et al. 1997, Bull et al. 2006). A limited amount of interspecific hybridization resulting in fertile hybrids also occurs between natural populations of some sympatric nonsibling Hawaiian species of Drosophila fruit flies ( Diptera : Drosophilidae ) which are strongly reproductively isolated by differing mate-recognition systems (Ahearn and Templeton 1989). The natural phenomenon of uncommon interspecific hybridization leading to fertile offspring is becoming more widely known among a variety of animal taxa ( Mallet 2005).

Diorhabda hybrids in the field might exhibit hybrid breakdown in areas additional to reduced egg viability, including behavioral disruptions in foraging (e.g., Godoy-Herrera et al. 1994) and pheromonal communication as a result of changes in hybrid pheromone composition (e.g., Wee and Tan 2005) and response (e.g., Lanier 1970). Laboratory or cage produced Diorhabda hybrids should not be considered as candidates for tamarisk biological control because of known and potential problems of hybrid breakdown in the field which would probably lead to low persistence and low efficacy. Further investigation is needed with genetic techniques to search for various types of hybrids and backcross hybrids and genetic introgression between Diorhabda species in the Palearctic.

Morphometrics. Ranges for variables of all the external measurements overlap among males and females of all species in the D. elongata group ( Table 2). Certain individuals of extremely small or large size can be identified as being of one or two species. For example, only females of D. carinata reach over 8.0 mm in length. Several significant differences in mean ranks of morphometric variables are seen. The mean lengths of the body and left elytron are significantly greater in females than males in each of the five species in the group. Mean lengths of the body and left elytron of both males and females of D. carinata are significantly greater than those of all other species in the D. elongata group ( Table 2). Mean length of the body and left elytron are significantly smaller in males and females of D. carinulata and D. meridionalis compared to all other members of the D. elongata group. The combined width of the elytra at the widest point is significantly greater in females of D. elongata , D. sublineata , and D. carinata , than in females of D. carinulata and D. meridionalis and males of all species. The width of the elytra is significantly greater in male D. carinata compared to males of all other species and females of D. carinulata and D. meridionalis . The mean ratio of the width of the elytra to the length of the left elytron is significantly smallest in D. carinata making it the most elongate species in the group. Additional external morphometric variables are being evaluated for use in classification of the D. elongata group (Joaquin Sanabria, International Fertilizer Development Center [IFDC], Muscle Shoals, AL, pers. comm.).

Milbrath and DeLoach (2006b) found the mean lengths of the left elytron (mm) in D. elongata (from Crete) reared on T. ramosissima × T. chinensis (4.57 ± 0.05 in 29 males, 5.19 ± 0.04 in 47 females) are slightly but significantly larger than that of individuals reared on T. aphylla (4.46 ± 0.04 in 42 males, 5.09 ± 0.05 in 33 females). These mean values for the length of the left elytron were similar to those observed for field material of D. elongata in this revision ( Table 2), but the standard deviations they observed were much smaller than we observed. The wide geographic range of specimens measured in our revision probably accounts for the wider variability in measurements.

Significant interspecific differences are seen in ranks of measurements of male endophallic sclerites ( Table 3) and female vaginal palpi and internal sternite VIII ( Table 4). All species of the D. elongata group significantly differ in the mean values of three variables: (1) the length of the blade of the elongate endophallic sclerite, (2) the ratio of the length of the blade of the elongate sclerite to the total length of the elongate sclerite, and (3) the width of the stalk of female internal sternite VIII. We find nine instances of discrete breaks or near breaks in size ranges of genitalic structures between species of the group, some of which are used in the keys to species (below): (1) both the range in width and width to length ratio of the palmate endophallic sclerite is nearly separate between D. carinulata and D. meridionalis , (2) the length of the elongate sclerite is largest in D. carinata and the range separates D. carinata from D. carinulata and D. meridionalis (see also scatter plot of Fig. 48 ); (3) the ratios of both the blade length and length of the spined area of the blade to the length of the elongate sclerite is smallest in D. elongata and the range separates D. elongata from all other species (see also scatter plot of Fig. 48 ), (4) the range in the ratio of the blade length to the length of the elongate sclerite separates D. carinulata and D. meridionalis ( Fig. 48B ); (5) both the range of width and width to length ratio of the vaginal palpus in D. carinulata and D. meridionalis is nearly separate from all other species in the group; (6) the range in width of the stalk of female internal sternite VIII separates D. meridionalis from all other members of the group except D. elongata ; (7) the range in ratio of the width of stalk to width of apical lobe of female internal sternite VIII separates D. meridionalis from all other members of the group except D. elongata ; and (8) the width of the lobe of the stalk in female internal sternite VIII separates D. meridionalis and D. carinata .

a Ranks of values followed by the same letter within the same column are not significantly different (P> 0.05; Two-Way Kruskal-Wallis Tests on ranks using PROC GLM-LSMEANS test; SAS Institute 2005) (see Figs. 14–33 for illustrations of measured structures; see Fig. 48 for scatter plots of measurements of the elongate endophallic sclerite).

a Ranks of values followed by the same letter within the same column are not significantly different (P> 0.05; Two-Way Kruskal-Wallis Tests on ranks using PROC GLM-LSMEANS test; SAS Institute 2005). See Figs. 34–38 for illustrations of measured structures.

b Width of widest lobe.

keys of the Diorhabda elongata species group. a a Increasing degree or presence of a character state is denoted by a higher number. For details on diagnostic genitalic character states, see illustrated taxonomic keys, genitalic characters under each species, and Tables 2, 3 and 4. Character states are standardized to range from zero to one. Differences in character states between species are discrete with the exceptions noted below for three characters with near discrete differences for separating D. elongata from some species. b Differences in spination and narrowing of the palmate endophallic sclerite are near discrete between D. elongata and the two species D. carinata and D. sublineata .

c Differences in elongation of vaginal palpi are near discrete between D. elongata and D. carinulata (difference is discrete if used in combination with width of widest lobe of female internal sternite VIII).

a Matrices produced with NTSYSpc Interval Data (SIMINT) module ( Rohlf 2006).

Stenophenetic Analysis. The data matrix of genitalic character state profiles for the D. elongata group is shown in Table 5 and the derived average taxonomic dissimilarity matrix and a Pearson product-moment correlation similarity matrix are shown in Table 6. The genitalic dissimilarity and similarity phenograms produced from these matrices using the UPGMA clustering method ( Figs. 49–50 ) were of the highest quality having the greatest r coph values of 0.817 and 0.885, respectively. Genitalic phenograms produced from complete-linkage clustering and single-linkage clustering (not shown) have lower r coph values of 0.745 and 0.811, respectively, for dissimilarity phenograms, and 0.882 and 0.862, respectively, for similarity phenograms. The genitalic similarity phenograms produced from any of the three clustering methods are of higher quality (have higher r coph values) than genitalic dissimilarity phenograms produced from any of the clustering methods. Both UPGMA dissimilarity and similarity phenograms ( Figs. 49–50 ) reveal D. carinata and D. sublineata are the most similar in genitalia, sharing seven of eight genitalic character states. D. carinata and D. sublineata form a group separate from the other three species in the genitalic dissimilarity phenogram while they form a group with D. elongata in the genitalic similarity phenogram. Diorhabda carinulata and D. meridionalis share five of eight genitalic character states and form a group with D. elongata in the genitalic dissimilarity phenogram but are separate from the other three species in the genitalic similarity phenogram. Similarities in genitalia among these Diorhabda species that are seen in the genitalic phenograms may or may not be correlated with genetic similarity.

Berti and Rapilly (1973) originally described D. meridionalis as a subspecies of D. carinulata . Diorhabda sublineata differs from D. carinata by only a single genitalic character in each sex. Therefore, we suspect the D. carinulata / D. meridionalis and D. carinata / D. sublineata groupings do represent genetic similarity. The placement of D. elongata with either D. carinulata / D. meridionalis or D. carinata / D. sublineata has less support. A morphometric phenogram for the D. elongata group based upon external and genitalic data are in preparation (J. Sanabria, pers. comm.). Studies of genetic similarity in mitochondrial and nuclear DNA among D. elongata , D. sublineata , D. carinata , and D. carinulata are in progress (D. Kazmer, pers. comm.).

Common Name. The vernacular name “tamarisk leaf beetle” has been applied in the Bulgarian, Russian and North American biological literature to D. elongata ( Tomov 1974) as well as D. carinulata (as D. elongata: Sinadsky 1957, 1963 , 1968; as D. e. deserticola: Bean et al. 2007b ). The name “elongate tamarisk leaf beetle” was applied by Lozovoi (1961) to D. carinata (as D. elongata ) in Georgia. The name “saltcedar leaf beetle” has been applied in North American biological control literature primarily to both D. elongata and D. carinulata (as D. elongata ; DeLoach and Carruthers 2004 b, Dudley and DeLoach 2004, Dudley 2005 a, USDA Animal and Plant Health Inspection Service 2005, Carruthers et al. 2006, Dudley et al 2006, Herr et al. 2006, DeLoach et al. in prep., Moran et al. in press), but also to D. carinata and D. sublineata (as D. elongata ; DeLoach and Carruthers 2004b). We propose the term “tamarisk beetles” to refer to all five species of the D. elongata group. Because Diorhabda are Old World species, in the common name we conserve the term “tamarisk”, used for the entire genus Tamarix in much of its native distribution ( Baum 1978) and often in North America (e.g., Baum 1967 a, Birken and Cooper 2006). In contrast, the competing term “saltcedar” is generally confined to use in the United States where it applies to certain weedy deciduous species ( T. ramosissima / T. chinenis , T. canariensis / T. gallica , and T. parviflora ; DiTomaso 1998). Members of the D. elongata group are probably the most important coleopterous defoliaters of tamarisk, making this group worthy of the general name “tamarisk beetle”. Consequently, the term “leaf” is dropped from the group vernacular name in order to shorten proposed common names for each species. Species common names are formed by adding a term referring to some unique intrinsic biogeographical or morphological characteristic before the term “tamarisk beetle”. Other species of the numerous beetles specializing on tamarisk outside the tribe Galerucini ( Kovalev 1995) could be distinguished in their common name from “tamarisk beetle” by a name referring to their differing tribe (e.g., “tamarisk flea beetle” for Alticini), subfamily (e.g., “tamarisk casebearer” for Cryptocephalinae) or family (e.g., “tamarisk weevil” for Curculionidae ). Vernacular names are listed following the Latin names in the section header for each species account.

Biology. Host Plants. Only shrubs and trees of the Old World family Tamaricaceae (order Tamaricales ), including Tamarix and Myricaria , serve as hosts for any of the five species of tamarisk beetles ( Table 1). Among tamarisks in North America, T. aphylla (Linnaeus) Karsten (athel tamarisk) is considered a non-target for biological control by Diorhabda because of its value as a shade tree and for windbreaks in parts of the southwestern U.S. and northern Mexico ( DeLoach 1990; Map 7) and because it is only a minor invasive. However, T. aphylla is a major invasive along the Finke River in Australia ( Griffin et al. 1989), and could potentially become more invasive in North America, where it is now known to invade by seed propagation in addition to vegetative propagation in southern Nevada ( Walker et al. 2006) and northern Mexico (unpublished data; Map 7). Tamarix aphylla is a natural, but probably minor, host for D. carinata in Pakistan and D. sublineata in Tunisia ( Table 1). In no-choice, small field cages, D. elongata , D. carinata , and D. sublineata , accepted T. aphylla for oviposition to the same degree as they accepted T. ramosissima × T. chinensis (Milbrath and DeLoach 2006b) . In more sensitive multiple-choice, field cages studies, D. sublineata and D. carinulata generally prefer other North American tamarisks to T. aphylla for oviposition (egg-laying). However, D. elongata and D. carinata selected T. aphylla equally as well as other Tamarix species for oviposition in some tests, but oviposited on T. aphylla significantly less in other tests (Milbrath and DeLoach 2006a, 2006b), laying only ca. 22–30% as many eggs on T. aphylla as on some other tamarisks. In open field multiple-choice tests, D. elongata preferred ovipositing on T. chinensis × T. canariensis / T. gallica when compared with T. aphylla (Moran et al. in press). Tamarix aphylla appears to be at low to moderate risk to damage by the four tested Diorhabda spp. in the field. Potential damage by these Diorhabda to T. aphylla in a no-choice open field setting (especially where other Tamarix spp. are absent) is difficult to predict (Milbrath and DeLoach 2006b). The degree of potential damage to T. aphylla is still under investigation in field studies.

Frankenia spp. (family Frankeniaceae ) subshrubs and herbs are also in the order Tamaricales and occur throughout the native range of the D. elongata group ( Jäger 1992) and in western North America ( Whalen 1987). Native North American Frankenia spp. are of major concern as non-targets of tamarisk biological control whose safety must be insured ( Lewis et al. 2003a; Milbrath and DeLoach 2006a; Herr et al. 2006, in prep.). Frankenia spp. can serve as a suitable but generally less favorable host in development from larvae to adult for D. elongata , D. carinata , D. sublineata and D. carinulata in caged no-choice studies (further discussed later; DeLoach et al. 2003b; Lewis et al. 2003a; Milbrath and DeLoach 2006a; Herr et al. 2006, in prep.). However, Frankenia spp. are never reported to serve as a hosts for any Diorhabda spp. under natural conditions in the open field ( Table 1), even in specific surveys of Frankenia adjacent to where Diorhabda is found in Tunisia ( DeLoach et al. 2003b). Three North American Frankenia spp. , F. salina (Molina) I.M. Johnston , F. johnstonii Correll , and F. jamesii Torrey ex. A. Gray, provide almost no attraction for oviposition compared to Tamarix in field cage multiple-choice studies with D. carinulata ( Lewis et al. 2003a, Milbrath and DeLoach 2006a) and D. elongata , D. carinata , and D. sublineata (Milbrath and DeLoach 2006a) . In field cage no-choice studies with D. carinulata , D. sublineata and D. carinata , oviposition on F. jamesii and F. johnstonii was not different from non-host coyote willow ( Salix exigua Nutall ) and adults experienced increased mortality compared to T. ramosissima × T. chinensis treatments (Milbrath and DeLoach 2006a). However, in laboratory cage no-choice tests, differences in oviposition by D. elongata on F. salina (inland variety) and T. ramosissima , T. parviflora , and T. aphylla were not significant ( Herr et al. 2006). In open field testing of Frankenia salina transplanted among stands of Tamarix in Nevada and Wyoming, D. carinulata did not oviposit on F. salina which only sustained light (<4%) foliage loss from feeding of Diorhabda larvae that had crawled from nearby Tamarix (Dudley and Kazmer 2005). In open field tests with F. salina , T. ramosissima / T. chinensis and T. aphylla transplanted together in the field at Big Spring, Texas, D. elongata oviposited only a single egg mass on F. salina compared with hundreds on T. chinensis × T. canariensis / T. gallica ( Herr et al. 2006) . Lewis et al. (2003a) and Milbrath and DeLoach (2006a) concluded that Frankenia is at very low risk of damage from any of the four tested Diorhabda species. Attack on non-preferred host plants by insect biological control agents is generally reduced as distance to the preferred host increases ( Blossey et al. 2001). Consequently, we expect that damage by these four Diorhabda species to both Frankenia and T. aphylla is more unlikely the further away these plants are growing from preferred Tamarix spp. hosts. Host preferences of D. meridionalis remain to be studied. Further information on the host range of the D. elongata group as well as additional biological data are reviewed later for each species.

Phenology. Diorhabda carinulata has two to four generations from April through September and overwinter as adults in ground cover in western China and central Asia ( Sinadsky 1968; Tian et al. 1988; Bao 1989; Sha and Yibulayin 1993; Mityaev and Jashenko 1998, 2007; Chen et al. 2000; Li et al. 2000. Diorhabda carinulata has two generations north of 38° in North America (DeLoach and Carruthers 2004b). Diorhabda elongata , D. carinata , and D. sublineata can have up to five generations from March to October in field cages at Temple, Texas ( Milbrath et al. 2007). Five generations have also been observed for D. carinata (as D. elongata ) at Ashgabat, Turkmenistan from March to September ( Myartseva 1999). Diorhabda sublineata is collected from mid-January to mid-December in Tunisia (from examined material), where it may have more than five generations. The voltinism of D. meridionalis is yet to be studied, but adults are found from March to October in southern Iran (from examined material).

Bean et al. (2007a) found the critical photoperiod (inducing diapause in 50% of the population) for adult D. carinulata from Fukang, China (44°N) is ca. 14 hours 55 minutes at 28°C. In comparison to D. carinulata from Fukang, the critical photoperiod for diapause induction is more influenced by temperature and much shorter in D. elongata from Crete (35°N), D. carinata from Qarshi, Uzbekistan (38°N), and D. sublineata from Sfax, Tunisia (35°N) ( D. sublineata having the shortest critical photoperiod) (Bean and Keller in prep.).

Bean and Keller (in prep.) also found intraspecific differences in critical photoperiod between two populations of D. carinulata from a similar latitude but widely varying elevations in China. Diorhabda carinulata from a low elevation at Turpan (42.8°N, - 3 m elevation) have a critical photoperiod ca. 1 hr shorter than populations from Fukang (44.2°N, 552 m elevation) under laboratory conditions of ca. 25°C. These populations may represent two climatypes of D. carinulata adapted to differing seasonal progressions as influenced by elevation. Climatypes with intraspecific differences in critical daylength as influenced by seasonal differences across wide latitudinal and elevational gradients can probably be found among all the species of the D. elongata group. Additional observations on phenology are discussed in the species accounts.

Aggregation Pheromones. Cossé et al. (2005) identified a male-produced aggregation pheromone in D. carinulata (as D. elongata ) in laboratory and field studies. The pheromone consists of two components, the aldehyde (2 E,4 Z)-2,4-heptadienal and the alcohol (2 E,4 Z)-2,4-heptadien-1-ol. Robert Bartelt (USDA/ARS, Peoria, IL, pers. comm.) found that these same chemical components are also emitted by males of D. elongata , D. carinata and D. sublineata , but that the ratios of aldehyde to alcohol are not necessarily the same as in D. carinulata . In species other than D. carinulata , these chemicals are considered as putative pheromones until they can be field tested. The component ratio of alcohol to aldehyde in the putative pheromone of D. elongata of Sfakaki, Crete, Greece, is almost identical to that of D. carinulata . However, field testing revealed the D. carinulata pheromone is ineffective for D. elongata established at Big Spring, Texas (Allen Knutson, The Texas AgriLIFE Extension Service, Dallas, TX, pers. comm.). Diorhabda elongata from Posidi Beach, Greece are unusual in producing a higher proportion of alcohol in the pheromone, up to ca. 4 times higher than in the Crete population, as more pheromone is released. It is unclear whether the Crete D. elongata population may also be able to release higher amounts of pheromone under the proper conditions and prove to be more similar in pheromone composition to the Posidi Beach population than to D. carinulata (R. Bartelt, pers. comm.). If the Crete D. elongata pheromone is more similar to that of the Posidi Beach D. elongata , this would explain why the D. carinulata pheromone was ineffective for D. elongata . Proportions of alcohol in putative pheromones of D. carinata were intermediate between those of D. elongata from Crete and Posidi Beach. The proportion of alcohol in the putative pheromone of D. sublineata is about 10 to 20 times higher than that found in D. carinata , D. elongata from Crete, and D. carinulata , and about 5 times higher than that found in D. elongata from Posidi Beach. In some sympatric and syntopic species of bark beetles ( Coleoptera : Scolytidae ), pheromones of one species may disrupt conspecific pheromonal response of another species ( Poland and Borden 1998), and the possibility of such competitive interactions between some sympatric and syntopic species of tamarisk beetles may warrant investigation.

a One of several coastal locations with minimum values (See Figs. 51–52 for descriptive statistics of biogeographic characteristics summarized over 5 minute resolution grids).

b Biomes ordered by percentage collections; biomes listed until cumulative percentage of collections from biomes equals at least 80% ( Fig. 52B ).

c Based on descriptive statistics of biogeographic characteristics ( Figs. 51–52 ).

Biogeography. General. The native range of the D. elongata species group is primarily restricted to the Palearctic realm (Map 1). However, D. sublineata extends marginally into the Afrotropical realm in Senegal and Yemen and D. carinata extends marginally into the Indo-Malayan realm in northern Pakistan. All species of the D. elongata group are at least marginally sympatric with at least one other species in the group (Map 1; Table 8). Diorhabda carinulata and D. carinata are the most widely sympatric among the group, overlapping in distribution over a large portion of central Asia. The D. elongata group is distributed from 48°N in Kazakhstan ( D. carinulata ) to 16°S in Africa ( D. sublineata ) (Map 1, Table 7, Fig. 51B ). Diorhabda carinulata is primarily found further north than all other species in the group while D. meridionalis is primarily found further south than all other species. The primary distributions of D. elongata , D. carinata , and D. sublineata are intermediate in latitude within the D. elongata group, but D. sublineata is primarily found further south than D. elongata and D. carinata (Map1, Fig. 51B ). All species may be found at or below sea level and D. carinata and D. sublineata can range to 2,903 and 2,606 m elevation, respectively ( Table 7, Fig. 51A ). Diorhabda elongata , D. sublineata , and D. meridionalis are primarily found at elevations below 400 m and in maritime regions. In contrast, D. carinata and D. carinulata are primarily found above 400 m and are more continental in distribution, occurring the furthest inland from the oceans (Map 1, Table 7, Figs. 51A , 52A ).

Eight desert, grassland and forest biomes characterize the biogeography of differing Diorhabda species ( Table 9, Maps 2–6, Fig. 52B ). Within these biomes, Diorhabda are most likely to be found in the primarily riparian, spring and maritime habitats of their host Tamarix species. The biomes inhabited vary widely among Diorhabda and host Tamarix species ( Tables 9 and 11; Fig. 52B ). Diorhabda elongata is usually collected in Mediterranean and Temperate Broadleaf forest biomes, and it is not reported from desert or grassland biomes as are many other species in the D. elongata group. The bioclimatic conditions to which D. elongata is adapted for Mediterranean maritime and riparian habitats are likely very different from conditions found in riparian habitats of the desert biome in which D. carinulata and D. meridionalis are primarily found. a Biomes of Olson and Dinerstein (2002). States of occurrence in biome: 0—no record; 0.5—single record (minor presence); 1—multiple records (major presence).

Similarly, desert riparian habitats and grassland riparian habitats differ in bioclimatic and biogeographic characteristics. For instance, different salicaceous trees and shrubs generally characterize North American grassland riparian habitats, such as plains cottonwood ( Populus deltoides Barton ex Marshall subsp. monilifera [Aiton] Eckenwalder), and black willow ( Salix nigra Marshall ), and various desert riparian habitats, such as Fremont cottonwood ( P. fremontii Watson subsp. fremontii ), Meseta cottonwood ( P. fremontii Watson subsp. mesetae Eckenwalder ), Rio Grande cottonwood ( P. deltoides Barton ex Marshall subsp. wislizenii [Watson] Eckenwalder), and Goodding willow ( S. gooddingii Ball ) ( Eckenwalder 1977, Powell 1998). Diorhabda carinata is also most often found in the desert biome but it occurs in the temperate grassland biome to a higher degree than other species of the D. elongata group. Like D. elongata , D. sublineata is also most common in the Mediterranean biome, but it also is common in the desert biome in which D. elongata is absent and has a strong presence in the Flooded Grasslands and Savannas biome in which D. elongata is uncommon ( Fig. 52B ). Diorhabda elongata and D. meridionalis occur in the fewest biomes and are the most stenobiomic species with low biomic specialization index (BSI) values of 3.0 ( Table 9). Diorhabda carinata occurs in the largest number of biomes and is the most eurybiomic species, with the largest BSI value of 5.5.

a Matrix produced with NTSYSpc Interval Data (SIMINT) module ( Rohlf 2006). Raw data matrix normalized with log (x + 1) transformation before dissimilarity matrix calculated.

Our knowledge of the biogeographic characteristics of the D. elongata group is limited by potentially under-collected areas such as eastern Turkey, Syria, Pakistan, Afghanistan and the Arabian Peninsula. Additional comparative and descriptive information on the biogeography of each Diorhabda species , including the primary ecoregions they inhabit and their biogeographic associations with indigenous host Tamarix species , is discussed under the heading Biogeography in the species accounts.

As an aid to matching Diorhabda species to target tamarisk species in North America, we compare the similarities in biomic profiles of the D. elongata group with those of tamarisk species invasive in North America below (see Biomic Analysis). Among these tamarisk species, T. aphylla and T. ramosissima are the most eurybiomic with BSI values of 8.0 and 6.0, respectively ( Table 11). Tamarix chinensis and T. parviflora are the most stenobiomic tamarisks with BSI values of 1 and 2.5, respectively.

Tamarix austromongolica is closely related to T. chinensis ( Zhang 2004) , and our comparison of the native ranges and biomic profiles of these tamarisks prompted us to examine the potential presence of T. austromongolica in North America. Tamarix chinensis is widely cultivated throughout China, but it is indigenous only to eastern China (Zhang and Zhang 1990), east of 115°E in the Temperate Broadleaf and Mixed Forests biome. Tamarix chinensis has long been considered as widely invasive in the western U.S. ( Baum 1967a; Gaskin and Schaal 2002, 2003), but the native biomic profile of T. chinensis does not include desert and grassland biomes from which it is reported in the western U.S. ( Table 11). However, the native biomic profile of T. austromongolica does include desert and grassland biomes to which it is indigenous in the area between Lanzhou and Hohhot in north central China (Ma and Liu 1988; DeLoach, unpublished data; Zhang, Peng-yun, Lanzhou University, Lanzhou, China, pers. comm.; Map 5). Tamarix austromongolica was once considered a subspecies of T. chinensis ( Zhou 1989) , but recent studies support its status as a separate species ( Hua et al. 2004, Zhang 2004). In response to our interest in the potential presence of T. austromongolica in North America, J. Gaskin haplotyped intron 4 of the nuclear phophoenolpyruvate carboxylase (pepC) gene (Gaskin and Schaal 2002) from specimens of cultivated T. austromongolica from China and found that some genetically match that of common T. chinensis and T. ramosissima / T. chinensis hybrids (J. Gaskin, pers. comm.). The fairly narrow distribution of T. austromongolica from ca. 103– 112°E in China is situated at the eastern edge of the distribution of T. ramosissima and approaching the westernmost edge of the distribution of T. chinensis (115°E) where a T. ramosissima / T. chinensis hybrid zone would be expected to occur (Map 5). In view of the biomic profile, DNA data, and distribution of T. austromongolica , the possibilities that this species may be a T. ramosissima / T. chinensis hybrid and that it is contributing to the genome of invasive T. ramosissima / T. chinensis in western North America warrant further investigation.

Biomic Analysis. The biomic profile for the D. elongata species group ( Table 9) is used to generate a biomic Bray-Curtis dissimilarity matrix ( Table 10) in order to compare similarities in biomic profiles within the group. Of the several clustering methods used to generate biomic dissimilarity dendrograms from the dissimilarity matrix, the UPGMA clustering method produces the biomic dendrogram with the greatest r coph value of 0.879 ( Fig. 53 ). Biomic dendrograms produced from complete-linkage clustering and single-linkage clustering (not shown) have lower r coph values of 0.850 and 0.855, respectively. In the UPGMA biomic dendrogram, D. carinata and D. carinulata are the most similar in biomic profiles. The close relationship in biomes inhabited by D. carinata and D. carinulata can be related to these species having the greatest degree of sympatry and syntopy in the D. elongata group ( Table 8, Map 1). Diorhabda meridionalis is partially sympatric with D. carinata and is grouped next followed by D. sublineata . Diorhabda elongata is grouped separately from all other tamarisk beetles.

We also test for a positive correlation in the dissimilarities in biomic profiles of the D. elongata group ( Table 10) to dissimilarities in key morphological characters of the genitalia ( Table 6, discussed previously). The biomic Bray-Curtis dissimilarity matrix of the D. elongata group ( Table 10) is not significantly positively correlated with the genitalic average taxonomic dissimilarity matrix ( Table 6) (r M = -0.40559, P[random r M ≥ observed r M] = 0.9211 with 9,999 permutations, Mantel Test; NTSYS Matrix Comparison Plot [MXCOMP] module [ Rohlf 2006]). In other words, we find no significant positive correlation between the similarity in biomes inhabited by the five Diorhabda species and similarities in their genitalia. Instead, genitalic similarity tends to be negatively correlated with biomic similarity. This negative correlation is seen in species pairs such as D. carinata / D. carinulata , which are similar in biomic profiles but dissimilar in genitalic character profiles, and D. carinata / D. sublineata , which are different in biomic profiles, but similar in genitalic character profiles ( Figs. 49 and 52 ; see Discussion — Stenophenetic Analysis above). An exception to genitalically similar species more greatly differing biomically is found in the species pair D. carinulata / D. meridionalis , which are both common in the Deserts and Xeric Shrublands biome. However, D. carinulata and D. meridionalis are parapatric (Map 4), and the apparent biogeographic distinctiveness of these species probably results from different preferences for latitude and distance to the ocean within the desert biome (discussed later; see Figs. 51–52 ). Further ecogeographic studies in the D. elongata group should include testing for a negative correlation between morphological similarity and a profile of a variety of climatic variables.

The combined biomic profiles of the D. elongata group ( Table 9) and tamarisks invasive in North America ( Table 11) are used to generate a Bray-Curtis dissimilarity matrix ( Table 12) in order to assess similarities in biomic profiles among Diorhabda and Tamarix together. Two biomic dissimilarity dendrograms from the dissimilarity matrix that are produced using the UPGMA clustering method have the greatest r coph value of 0.881 (one dendrogram is shown in Fig. 54 ). Several biomic dendrograms produced from completelinkage clustering and single-linkage clustering (not shown) have lower r coph values of 0.841 and 0.808, respectively. The two UPGMA biomic dendrograms differ only in alternating the places of T. gallica and D. elongata ( Fig. 54 ). In these dendrograms, D. carinulata , D. carinata , T. ramosissima , and T. austromongolica form a group similar in biomic profiles. Diorhabda elongata , T. parviflora , and T. gallica also form a group associated with T. chinensis . Diorhabda meridionalis and T. canariensis form a group from which D. sublineata occurs next up the tree.

In order to display spatial relationships in biomic profiles among the Diorhabda and Tamarix species , a three dimensional biomic principle coordinate analysis (PCoA) scatter plot ( Fig. 55 ) is computed from the biomic Bray-Curtis dissimilarity matrix for Diorhabda and Tamarix ( Table 12). Visual groupings of the Diorhabda and Tamarix species generally coincide with relationships in the biomic dissimilarity dendrogram ( Fig. 54 ). The two groupings of D. elongata / T. gallica / T. parviflora and D. meridionalis / T. canariensis are the most distinct in the biomic PCoA scatter plot. The group of D. carinata / T. ramosissima / D. carinulata / T. austromongolica is also evident, but less distinct. The taxa of D. sublineata and T. aphylla are more isolated as in the biomic dendrogram. The influence or loadings of the different biomes on PCoA axes are computed as Spearman rank-order coefficients ( Table 13 —Spearman rank-order correlation coefficients) for ranks of species in biomes (from Tables 9 and 11) versus ranks of species in PCoA axes (from Table 13 —Eigenvectors). The D. elongata / T. gallica / T. parviflora group has high eigenvectors along PCoA axis one ( Table 13 —Eigenvectors, Fig. 55 ) which is strongly negatively correlated with the Desert and Xeric Shrublands, Temperate Grasslands, Savannas and Shrublands, and Montane Grasslands and Shrublands biomes ( Table 13 - Spearman rank-order correlation coefficients), but positively correlated for Mediterranean and Temperate Broadleaf and Mixed Forests biomes. In other words, deserts and grasslands are strongly contraindicative for the presence of species favoring Mediterranean and Temperate Broadleaf and Mixed Forests biomes, such as D. elongata , T. gallica and T. parviflora . In contrast, eigenvectors for D. carinulata , D. carinata , T. ramosissima , and T. austromongolica , are very low along axis one as these species have a strong presence in deserts and temperate and montane grasslands. Eigenvectors for T. aphylla and D. sublineata are highest along PCoA axis two which is strongly positively correlated with Flooded Grasslands and Savannas. The eigenvector for D. carinata is the highest along axis three which is strongly positively correlated with the Temperate Conifer Forests biome.

a Matrix produced with NTSYSpc Interval Data (SIMINT) module ( Rohlf 2006). Raw data matrix normalized with log (x

+ 1) transformation before dissimilarity matrix calculated.

Habitat Suitability Index Models. Descriptive statistics of the four biogeographic variables (biome, latitude, elevation and distance to ocean) for each Diorhabda species are displayed in Figures 51–52 . These four descriptive statistics are used in calculating the five suitability indices (SI 1 –SI 5; Figs. 56–58 ), and the final habitat suitability index (HSI; the geometric mean of SI 1 –SI 5) in hand-fitted HSI models for each Diorhabda species We subjectively adjusted parameters of the suitability indices in preliminary HSI models (not shown) to reduce the visually assessed overall error in the final models (see Materials and Methods). Our sensitivity and elasticity analysis of the five suitability indices revealed that the categorical biomic indices SI 4 alone and both SI 4 and SI 5 together produce significantly higher model elasticities than do the continuous linear variable indices SI 1, SI 2, and SI 3 ( Table 14).

a PCoA eigenvectors and eigenvalues computed (NTSYSpc Eigenvetors [EIGEN] module; Rohlf 2006) from double centered symmetric dissimilarity matrix (not shown; NTSYSpc Dcenter module). Spearman rank-order correlation coefficients with asterisk are significant (P <0.05; Proc Corr; SAS Institute 2005). See Figure 55 for biomic PCoA scatter plot of the first three axes.

Our final HSI models for the D. elongata group in the Palearctic (Maps 8 and 9) are generally accurate in estimating the optimal native range for each Diorhabda species. However, the models overestimate the optimal native ranges in some cases such as for D. elongata in France, Spain, and northwest Africa where D. sublineata dominates (Map 8a) and for D. sublineata in southeastern Europe from which only D. elongata is known (Map 8c). The domination of either D. elongata or D. sublineata in parapatric areas of similar suitability could be related to potentially strong competitive interactions between these species.

5

a Means of a given variable and the same sample size followed by the same later are not significantly different (P> 0.05; Fisher’s Protected LSD Test using PROC GLM-LSMEANS test; SAS Institute 2005). Sensitivities and elasticities are calculated across eight possible values ranging from 0.0 to 1.0 for the analyzed suitability index while the other indices are held constant at 0.5. For the continuous variables SI 1, SI 2 and SI 3, eight values are chosen at equal intervals of 0.125 (0.0, 0.125, 0.250,…1.0) and these do not vary between species. For the categorical variables SI 4 and SI 5, the eight fixed possible values (from Figs. 57B and 58 ) vary between species, and SI 4 and SI 5 vary directly with one another and are considered together in calculating sensitivity and elasticity.

b Total sample size (n) = five species × four indices × eight values per index = 160.

The estimated North American ranges of the various Diorhabda spp. vary widely (Maps 10 and 11). The HSI model for D. elongata correctly predicts both the high suitability of the Cache Creek, California site and the low suitability of the Artesia, New Mexico and Lake Meredith, Texas sites. However, our D. elongata HSI model underestimates the suitable habitat at sites in west Texas at Big Spring, Imperial, and Pecos, where D. elongata has established well but where the model continentality suitability index and biomic suitability indices are near zero (Map 10a). The HSI model for D. carinulata correctly predicts the high suitability of northern desert habitats in Nevada, Colorado and Wyoming, and additionally predicts suitable habitat in the northern Chihuahuan Desert. Our HSI models estimate that D. carinata should have the widest area of suitable habitat in North America (Map 10b), while D. elongata is estimated to have the smallest area (Map 10a). The wider area of suitability for D. carinata is related to its being the most eurybiomic species (see Biogeography — General above), giving it a wide tolerance in the biomic suitability indices ( Figs. 57B and 58 ), and having a relatively wide tolerance in both the elevational and continentality suitability indices ( Figs. 56B and 57A ). The smaller estimated area of suitability for D. elongata in North America is related to it being among the most stenobiomic species, giving it a smaller tolerance for the biomic suitability indices ( Fig. 57B and 58 ), and having a very narrow tolerance in the continentality index. D. sublineata is estimated to have a similar area of optimal suitability to D. elongata in Mediterranean biome along the Pacific coast, but has a higher suitability for southern desert biomes (Map 10c). D. meridionalis is predicted to have the highest suitability in southern maritime desert habitats such as the Sonoran Desert and Tamaulipan Mezquital in southern Texas (Map 11b).

The composite maps displaying which Diorhabda spp. score within the top 15% of the maximum HSI value of any Diorhabda spp. are generally accurate in estimating which single species or multiple species dominate in any given area of their native range (Map 12). For example, both the dominance of D. carinulata in China and the northern part of its range, and the dominance of D. carinata is in some grasslands of Central Asia are depicted fairly accurately. The composite HSI maps fail to depict the dominance of D. sublineata in France, Spain, and the coast of northwest Africa and the dominance of D. elongata in the northeastern Mediterranean, instead depicting both species as equally dominant in these areas.

The composite HSI maps for the D. elongata group for North America (Map 13) estimate D. carinulata as dominant or co-dominant with D. carinata in areas where it has established well above 38°N in temperate cold deserts such as the Great Basin Shrub Steppe, Colorado Plateau Shrublands and Wyoming Basin Shrub Steppe. Diorhabda carinulata and D. carinata are estimated as potentially dominating or co-dominating in temperate warm desert areas areas below 38°N such as the northern Mojave Desert, southern Colorado Plateau Shrublands, and the Trans-Pecos region of the Chihuahuan Desert. Diorhabda carinata is estimated to dominate in large areas of temperate grasslands in the Great Plains, such as the Western Short Grasslands (including where D. carinulata has established at Pueblo, Colorado) and Central and Southern Mixed Grasslands, and in temperate conifer forests, such as the Arizona Mountains Forests in Arizona and New Mexico (Map 13). Although D. carinulata is established in the Western Short Grasslands at Pueblo, Colorado, our model predicts that D. carinata is better suited to this habitat for which it is in a more optimal range in terms of biome (being a grassland), latitude, and distance to the ocean. Pueblo, Colorado falls within the HSI environmental envelope suitability score (SI 5) for D. carinulata , but the model map only displays D. carinata for Pueblo because this species is more than 15% higher in the overall optimality (SI 10) score. Introduction of southern climatypes of D. carinulata and D. carinata may speed their adaptation to areas south of 34°N (see further discussion in species accounts under Potential in Tamarisk Biological Control). Diorhabda elongata is correctly estimated as dominating the Mediterranean biome of northern California, including where it is established vigorously at Cache Creek. However, our models estimate that areas of west Texas, where D. elongata has also established well, are more suitable for D. carinata and D. carinulata . West Texas falls outside the environmental envelope suitability scoring (SI 5) for D. elongata in both biome (being in temperate grasslands) and distance to ocean limits (over 600 km from ocean versus a 255 km limit) ( Table 7, Fig. 52 ). The environmental envelope suitability score portion of our HSI model underestimates the potential range of D. elongata in grasslands and desert of west Texas, but our overall HSI model score may still be accurate in predicting that D. carinata and D. carinulata are better suited to these areas. Diorhabda sublineata is estimated as dominating in the Mediterranean biome in California below 37°N and the Chihuahuan Desert below 29°N. Diorhabda meridionalis , a species with a higher preference for maritime subtropical deserts between 26– 31°N, is estimated as dominating over most of the southern Sonoran desert and Tamaulipan Mezquital xeric shrubland.

We consider the estimations of these hand-fitted HSI models as first rough approximations of the optimal tamarisk beetle species for various Old World and New World tamarisk habitats. The descriptive statistics used in calculating HSI grids could change substantially for some Diorhabda species with additional distribution data from under-collected areas. Complex interactions of the biogeographic variables we examined between both one another and unexamined bioclimatic variables probably occur, especially near environmental extremes for each species or in novel environments such as North America. Bioclimatic variables are among the best predictors of species distributions and further studies of the distributions of tamarisk beetles are planned using climatic data with species distribution models designed for presence-only data (e.g., Elith et al. 2006).

More detailed comparisons of the descriptive statistics for biogeographic variables among Diorhabda species are found in the species accounts below under the heading Biogeography - Comparative. Discussion of these results regarding biological control of tamarisk appear under the heading Potential in Tamarisk Biological Control for each species and in the concluding section, Implications Regarding Biological Control of Tamarisk

Figures

FIGURE 48. Scatter plots of length of blade of elongate endophallic sclerite (A) and length of spined area of elongate endophallic sclerite (B) versus the length of the elongate endophallic sclerite for field collected material of Diorhabda elongata species group and laboratory produced F1 and F2 D. elongata/D. sublineata hybrids (see Figs. 45–47 for illustrations of endophallic sclerites for individual hybrids identified by alphanumeric codes in plots; see Table 3 for sample sizes and statistics for measurements).

FIGURE 51. Schematic box plots for elevations (A) and latitudes (B) from native presence-only field collection data for the Diorhabda elongata species group at 5 minute grid resolution. Box plots for each species depict the mean, median, first quartile (Q1), third quartile (Q3), interquartile range (IQR, Q1–Q3), low whisker (LW; lowest point at or above 1.5*IQR lower than Q1), high whisker (HW, highest point at or below 1.5*IQR higher than Q3), mild outliers (MO, points between the low or high whisker and 3*IQR from Q1 and Q3, respectively), and extreme outliers (EO, points below or above 3*IQR from Q1 and Q3, respectively) (Proc Boxplot; SAS Institute 2005). Sample sizes of field localities (N) and a table summary statistics and plotted values are inset in each chart.

FIGURE 52. Schematic box plots for distances to the ocean (A) (Proc Boxplot [Boxstyle=Schematic]; SAS Institute 2005) and a bar chart of frequency percentages for distribution of each Diorhabda species across biomes (B) (Proc Freq; SAS Institute 2005) from native presence-only field collection data for the Diorhabda elongata species group at 5 minute grid resolution. Percentage of a species in a biome is number of collections for that species in the biome divided by the number of collections of that species across all biomes. See Figure 51 for explanation of box plots. Sample sizes of field localities (N) and a table summary statistics and plotted values are inset in each chart.

FIGURE 55. Three dimensional biomic principal coordinate analysis (PCoA) scatter plot for the Diorhabda elongata species group and Tamarix spp. invasive in North America for the first three eigenvectors (cumulative axis loading of 80.92%; Table 13) computed from a biomic Bray-Curtis dissimilarity matrix (Table 12). Biomes in which ranks of species are statistically significantly positively or negatively correlated with ranks of species in each PCoA axis are indicated in parentheses (see Table 13). Plotted with Mod3D module of NTSYSpc (Rohlf 2006).

FIGURE 56. Linear models for latitudinal (A) and elevational (B) habitat suitability indices for the Diorhabda elongata species group. Model parameters calculated from descriptive statistics (Fig. 51), including interquartile range (IQR), minimum (MIN) value minus 1% or 10% of range (R), and maximum value (MAX) plus 1% or 10% of range (labeled for D. elongata).

FIGURE 57. Linear and categorical models for continentality (A) and biomic (B) habitat suitability indices for the Diorhabda elongata species group. Model parameters calculated from descriptive statistics (Fig. 52), including interquartile range (IQR), minimum (MIN) value minus 20% of range (R), and maximum value (MAX) plus 20% of range (labeled in A for D. carinulata).

FIGURE 58. Categorical model for biomic relative suitability indices for the Diorhabda elongata species group. Model parameters calculated from descriptive statistics (Fig. 52B).

FIGURES 1–9. Dorsal habitus of males (dead - top row, living - bottom row) of the Diorhabda elongata species group (tamarisk beetles): 1–2—D. elongata (Mediterranean tamarisk beetle), 3–4—D. carinata (larger tamarisk beetle), 5–6—D. sublineata (subtropical tamarisk beetle), 7–8—D. carinulata (northern tamarisk beetle), 9—D. meridionalis (southern tamarisk beetle). SSV—subsutural elytral vitta, SMV—submarginal elytral vitta. Scale bar 5.0 mm.

FIGURES 10–13. Sexual dimorphism in Diorhabda elongata: 10—male last visible sternite, 11—female last visible sternite, 12—male foretarsus, 13—female foretarsus. AN—apical notch, T3—third bi-lobed tarsal segment, T5—fifth tarsal segment. Scale bars 1.0 mm.

FIGURES 14–18. Male genitalia: median lobe (ML) of aedeagus with everted endophallus (END; uninflated) (lateral and dorsal views).14—Diorhabda elongata, 15—D. carinata, 16—D. sublineata, 17—D. carinulata, 18—D. meridionalis. BF—basal foramen (or basal orifice), CES—connecting endophallic sclerite, DO—dorsal ostium (or apical orifice); EES—elongate endophallic sclerite, GP—gonopore, LA—lateral appendage, LB—length of blade, PES—palmate endophallic sclerite, SL—length of spined area of blade, VM—ventral membrane. Scale bar 1.0 mm.

FIGURES 19–23. Elongate (ventral) endophallic sclerite (lateral view) with connecting (lateral) endophallic sclerite (dorsal view, when present). 19—Diorhabda elongata, 20—D. carinata, 21—D. sublineata, 22—D. carinulata, 23—D. meridionalis. CES—connecting endophallic sclerite, EES—elongate endophallic sclerite, LA—lateral appendage, LB—length of blade, LN—lateral notch, SL—length of spined area of blade. Scale bar 1.0 mm.

FIGURES 24–28. Elongate (ventral) endophallic sclerite (dorsal views and a single dorso-lateral view [DLV]). 24—Diorhabda elongata, 25—D. carinata, 26—D. sublineata, 27—D. carinulata, 28—D. meridionalis. BL—blade, HA —hooked apex, CES—connecting endophallic sclerite (partial), EES—elongate endophallic sclerite, LA—lateral appendage, LB—length blade, LN—lateral notch. Scale bar 1.0 mm.

FIGURES 29–33. Palmate (dorsal) endophallic sclerite (dorsal view). 29—Diorhabda elongata, 30—D. carinata, 31—D. sublineata, 32—D. carinulata, 33—D. meridionalis. CES—connecting endophallic sclerite, DS—distal spines, EES—elongate endophallic sclerite, LA—lateral appendage, LN—lateral notch, PES—palmate endophallic sclerite, SDS—subdistal spines. Scale bar 1.0 mm.

FIGURES 34–38. Female genitalia: internal sternite VIII (IS VIII), vaginal palpi (VP), and spermatheca (SP). 34—Diorhabda elongata, 35—D. carinata, 36—D. sublineata, 37—D. carinulata, 38—D. meridionalis. AL—apical lobe of IS VIII, LP—length vaginal palpus, PA—pointed appendage of SP, ST—stalk of IS VIII, TL—tips of lobes of IS VIII, WAL—width apical lobe of IS VIII, WLS—width lobe of stalk of IS VIII, WP—width vaginal palpus, WST—width stalk of IS VIII. Scale bar 1.0 mm.

FIGURES 39–43. Female internal sternite VIII (IS VIII). 39—Diorhabda elongata, 40—D. carinata, 41—D. sublineata, 42—D. carinulata, 43—D. meridionalis. AL—apical lobe, ST—stalk, TL—tips of lobes of stalk, *—IS VIII distinctly identifiable as D. carinata, **—IS VIII distinctly identifiable as D. sublineata. Scale bar 1.0 mm.

FIGURES 44–47. Palmate, connecting (dorsal views), and elongate (dorsal and lateral views) endophallic sclerites with abnormal sclerites in laboratory pure lines of Diorhabda elongata and D. sublineata and laboratory produced D. sublineata × D. elongata hybrids from New Mexico State University (NMSU). 44—♂ D. elongata pure line, 45—♂ D. sublineata pure line, 46—♀ D. sublineata × ♂ D. elongata hybrids (F1 and F2), 47—♀ D. elongata × ♂ D. sublineata hybrids (F1 and F2). AS—abnormal sclerite, CES—connecting endophallic sclerite, DS—distal spines, EES—elongate endophallic sclerite, LA—lateral appendage, LB—length of blade, LN—lateral notch, PES—palmate endophallic sclerite, SDS—subdistal spines, SL—length of spined area of blade. Scale bar 1.0 mm.

FIGURES 49–50. Genitalic phenograms for the Diorhabda elongata species group. 49–dissimilarity phenogram based on average taxonomic dissimilarity matrix (Table 5), 50—similarity phenogram based on and Pearson product-moment correlation similarity matrix (Table 6). Phenograms produced with NTSYSpc Tree plot module (Rohlf 2006) from clusters formed with unweighted arithmetic average clustering (UPGMA) (NTSYSpc SAHN module).

FIGURES 53–54. Biomic dissimilarity dendrograms based on biomic Bray-Curtis dissimilarity matrices. 53—Diorhabda elongata species group (from Table 10), 54—both the D. elongata group and Tamarix species invasive in North America (from Table 12). Dendrograms produced with NTSYSpc Tree plot module (Rohlf 2006) from clusters formed with unweighted arithmetic average clustering (UPGMA) (NTSYSpc SAHN module). Line connecting D. elongata and T. gallica at left signifies that the positions of these taxa are interchangeable in an alternate dendrogram of equal rcoph value.