Cryptic Species Discrimination in Western Pine Beetle, Dendroctonus brevicomis LeConte (Curculionidae: Scolytinae), Based on Morphological Characters and Geometric Morphometrics

The western pine beetle (WPB), Dendroctonus brevicomis LeConte, is a major mortality agent of pines in North America. A total of 706 adults of WPB from 81 geographical sites were analyzed with traditional and geometric morphometric methods to evaluate the variation of discrete and quantitative morphological characters with particular attention to the antenna, spermatheca, and seminal rod. Principal coordinates and canonical variate analyses supported three geographical groups in WPB: (1) West, from British Columbia to southern California along the Pacific coast, Idaho, and Montana; (2) East-SMOC, including Nevada, Utah, Colorado, Arizona, New Mexico, Texas, Chihuahua, and Durango; and (3) SMOR, including Coahuila, Nuevo Leon, and Tamaulipas. The pubescence length on the elytral declivity was a robust character for separating West specimens from the other groups. Additionally, the genitalia shape both female and male in dorsal view was a reliable character for discriminating among groups. Based on these results, which agree with genetic and chemical ecology evidence, we herein reinstate Dendroctonus barberi Hopkins (East-SMOC group) and remove it from synonymy with D. brevicomis (West group). Differences in the spermatheca and seminal rod shape of SMOR specimens suggest that these populations might be a different species from D. barberi and D. brevicomis.


Introduction
The western pine beetle, Dendroctonus brevicomis LeConte, is a North American species distributed along the Pacific coast from southern British Columbia to southern California, in Idaho and eastern

Frons sculpturing (FS) (scarce vs. abundant).
Males of some species of Dendroctonus have numerous and prominent granulate tubercles on the lateral areas of the frons [15]. This sculpturing consists of granules, punctures, and fused granules resembling small crenulations (Figure 2a,b). Some authors have used this attribute to discriminate reliably between some members of D. frontalis complex, such as D. vitei and D. mexicanus [6,15,22].

1.
Frons sculpturing (FS) (scarce vs. abundant). Males of some species of Dendroctonus have numerous and prominent granulate tubercles on the lateral areas of the frons [15]. This sculpturing consists of granules, punctures, and fused granules resembling small crenulations (Figure 2a,b). Some authors have used this attribute to discriminate reliably between some members of D. frontalis complex, such as D. vitei and D. mexicanus [6,15,22].
2. Degree of elevation of the epistomal process (DE)(elevated vs. not elevated). The epistomal process is a structure immediately above the oral cavity that comprises a pair of lateral elevations or "arms" and an interposing ridge (Figure 2c,d). Process width, elevation, and inclination display important differences that are useful for distinguishing species or species groups in Dendroctonus [12,14,15].
3. Epicranial Surface (ES) (smooth vs. rough). The sculpture of the epicranial surface has not been evaluated previously in the genus Dendroctonus (Figure 2e,f).
4. Length of pubescence on the elytral declivity (SPD) (variable vs. uniform). Length of elytral pubescence has been a useful character for separating D. brevicomis populations (Figure 2g,h) [2,12,13]. Specimens from British Columbia and California were found to possess pubescence on the elytral declivity that was variable in length but did not exceed the width of interestriae, whereas specimens from Arizona, Chihuahua, Durango, and Nuevo Leon displayed a more uniform pubescence length [2].
5. Thickness of the pubescence of the striae in relation to the pubescence of the interestriae (RSP) (thin vs. thick). The surface of the elytra is covered by abundant setae of different thicknesses. In Dendroctonus, the setae on the elytral interspaces usually differ in thickness from those in the interstriae. This character has not been evaluated previously (Figure 2g,h).
6. Striae on the elytral declivity (SED) (impressed vs. poorly or not impressed). The degree of impression of the striae on the elytral declivity has been used to distinguish some members in the genus Dendroctonus [12,14,15]. Hopkins [12] proposed this as useful character for separating D. barberi and D. brevicomis (Figure 2i,j).
7. Proportion of the nodulus covered by striae (PNCS) (50% vs. >50%). In the female spermatheca, the pattern of striations can cover different proportions of the nodulus (Figure 2k,l,m). This character has been shown to differ among species of the D. frontalis complex [7,20]. 8. Cornu shape (CS) (oval vs. rounded). The female spermatheca is divided into a nodulus and cornu (Figure 2l,m). The cornu is the distal portion of the spermatheca beyond the middle constriction, whereas the nodulus is the proximal portion [7,20]. Variation in cornu shape has shown to be a useful taxonomic character for separating some species within the D. frontalis complex.
9. Protuberance of cornu (PC) (absent vs. present). The proximal region of female spermathecae (nodulus) may possess a protuberance (Figure 2l). This is the first time that this character is described and shown to differentiate among D. frontalis complex members.
10. Length of frontal tubercles (FTL). This attribute was measured from the center of the median groove to the apex of the frontal tubercles (Figure 3a), in dorsal view.
11. Distance between frontal tubercles (DFT). This attribute was measured between the apices of the right and left frontal tubercles (Figure 3a), in dorsal view.
12. Epistomal brush length (EBL). This attribute was measured between the anterior and posterior edges of the epistomal brush (Figure 3b), in frontal view.
13. Epistomal process width (EPW). This attribute was measured from the left and right lateral margins in the epistomal process (Figure 3b), in frontal view.
14. Distance between the eyes (DBE). This attribute was measured between the internal margins of the central region of the eyes (Figure 3b), in frontal view.
15. Eye width (EW). This attribute was measured between the lateral margins of the central region of the right eye (Figure 3c), in lateral view.
16. Eye Length (EL). This attribute was measured between the dorsal and ventral margins of the right eye (Figure 3c), in lateral view.
17. Head-pronotum length (HPL). In the case of females, this attribute was measured in dorsal view from the frons (including frontal tubercles in males) to the posterior, lateral margin of the pronotum. The measurement could be slightly influenced by the angle that the head happens to be in, and therefore it must be measured based on the images of Figure 3d.
18. Pronotum length (PL). This attribute was measured along the median line of the pronotum from the anterior to the posterior margin (Figure 3d), in dorsal view.
19. Pronotum width (PW). This attribute was measured at the widest portion of the pronotum in dorsal view, from left to right margins, (Figure 3d).

Elytra length (EYL).
This attribute was measured from the anterior margin of the elytra to the posterior terminus of the elytral declivity (Figure 3d), in dorsal view.

Abdominal length (AL).
This attribute was measured between the intercoxal process and the posterior tip of the venter (Figure 3e), in ventral view.
22. Length of the midline of the metathorax (LMM). This attribute was measured from the sternellar area to sternellar piece in the metathorax (Figure 3e), in ventral view.
Characters were observed and measured at 100-400× with phase contrast microscopy (Carl Zeiss, Oberkochen, Germany), and photographed with a Coolpix 5000 camera (Nikon, Tokyo, Japan). Antenna, elytra, and genitalia were removed and mounted following the protocols previously reported for these structures in other species [7,23]. The elements of elytral sculpture (punctures, granules, pubescence, etc.) were measured and quantified directly in slides with an ocular micrometer under a phase contrast microscope (400×). Details of characters were photographed with an environmental scanning electron microscope (ESM Evo ® 40VP; Zeiss, Ontario, CA, USA).

Pronotum width (PW).
This attribute was measured at the widest portion of the pronotum in dorsal view, from left to right margins, (Figure 3d).

Elytra length (EYL).
This attribute was measured from the anterior margin of the elytra to the posterior terminus of the elytral declivity (Figure 3d), in dorsal view.

Abdominal length (AL).
This attribute was measured between the intercoxal process and the posterior tip of the venter (Figure 3e), in ventral view.
22. Length of the midline of the metathorax (LMM). This attribute was measured from the sternellar area to sternellar piece in the metathorax (Figure 3e), in ventral view.
Characters were observed and measured at 100-400× with phase contrast microscopy (Carl Zeiss, Oberkochen, Germany), and photographed with a Coolpix 5000 camera (Nikon, Tokyo, Japan). Antenna, elytra, and genitalia were removed and mounted following the protocols previously reported for these structures in other species [7,23]. The elements of elytral sculpture (punctures, granules, pubescence, etc.) were measured and quantified directly in slides with an ocular micrometer under a phase contrast microscope (400×). Details of characters were photographed with an environmental scanning electron microscope (ESM Evo ® 40VP; Zeiss, Ontario, CA, USA).

Morphometrics Analysis
To characterize the main geographical trends of morphological variation among specimens, we performed three principal coordinates analyses (PCoA): Females and males together (n = 706), and females (n = 340) and males (n = 366) alone (Table S1). These analyses used pairwise Gower's distance [24] estimated from corresponding qualitative and quantitative characters (17 for sexes combined, 17 for females, and 19 for males). Quantitative characters were log transformed because they did not meet the criteria of normality. Additionally, we used a canonical variate analyses (CVAs) to determine to what extent these attributes explained the possible geographical regionalization of specimens. For this analysis, we calculated average values and variances for each character clustered by the putative geographic groups defined with PCoA. Lastly, we looked for statistical differences among geographical groups with an analysis of similarities (ANOSIM) and pairwise Hotelling's T non-parametric tests [25].
Lastly, to assess the relative taxonomic weight of qualitative and quantitative characters employed, we examined statistical differences with chi-square contingency tables and ANOVA tests, respectively [25]. As with PCoA, we analyzed males and females together and separately within the putative geographical groups.

Morphometrics Analysis
To characterize the main geographical trends of morphological variation among specimens, we performed three principal coordinates analyses (PCoA): Females and males together (n = 706), and females (n = 340) and males (n = 366) alone (Table S1). These analyses used pairwise Gower's distance [24] estimated from corresponding qualitative and quantitative characters (17 for sexes combined, 17 for females, and 19 for males). Quantitative characters were log transformed because they did not meet the criteria of normality. Additionally, we used a canonical variate analyses (CVAs) to determine to what extent these attributes explained the possible geographical regionalization of specimens. For this analysis, we calculated average values and variances for each character clustered by the putative geographic groups defined with PCoA. Lastly, we looked for statistical differences among geographical groups with an analysis of similarities (ANOSIM) and pairwise Hotelling's T non-parametric tests [25].
Lastly, to assess the relative taxonomic weight of qualitative and quantitative characters employed, we examined statistical differences with chi-square contingency tables and ANOVA tests, respectively [25]. As with PCoA, we analyzed males and females together and separately within the putative geographical groups.

Geometric Morphometrics
We evaluated whether shape variation of the antennal club (n = 325), seminal rod (n = 202 dorsal view, 247 lateral view), and spermatheca (n = 203) (Table S1) segregated specimens into geographical groups. Structures were oriented in the same direction for photographs ( Figure S1). Few homologous points were available in these features for landmarks (lds), thus we also used semilandmarks (smlds), which are points assigned along a geometric curve, edge, or surface of a feature [26,27]. Smlds were defined by superimposing radial lines ("fans") or parallel lines ("combs") onto digitized photographs of the antennal club, seminal rod, and spermatheca using MakeFan6 of the Integrated Morphometrics Package (IMP) [28].
A total of eight lds (2-3; 5, 10; 11, 16; 17, 22) and 16 smlds (1-4; 7-9; 12-15; 18-21; 23-24) were used to characterize the antennal club configuration ( Figure S2a). To define smlds, we traced a straight line between lds 11 and 16 located at the intersection of the extremes of the second sensorial band and outline of this structure, then by applying a comb overlay, we projected from it six equidistant, perpendicular lines. Smlds were located at every intersection between these lines and sensory bands, as well as the outline of antennal club ( Figure S2a).

Geometric Morphometrics
We evaluated whether shape variation of the antennal club (n = 325), seminal rod (n = 202 dorsal view, 247 lateral view), and spermatheca (n = 203) (Table S1) segregated specimens into geographical groups. Structures were oriented in the same direction for photographs ( Figure S1). Few homologous points were available in these features for landmarks (lds), thus we also used semilandmarks (smlds), which are points assigned along a geometric curve, edge, or surface of a feature [26,27]. Smlds were defined by superimposing radial lines ("fans") or parallel lines ("combs") onto digitized photographs of the antennal club, seminal rod, and spermatheca using MakeFan6 of the Integrated Morphometrics Package (IMP) [28].
A total of eight lds (2-3; 5, 10; 11, 16; 17, 22) and 16 smlds (1-4; 7-9; 12-15; 18-21; 23-24) were used to characterize the antennal club configuration ( Figure S2a). To define smlds, we traced a straight line between lds 11 and 16 located at the intersection of the extremes of the second sensorial band and outline of this structure, then by applying a comb overlay, we projected from it six equidistant, perpendicular lines. Smlds were located at every intersection between these lines and sensory bands, as well as the outline of antennal club ( Figure S2a).
For the spermatheca configuration we used two lds (1,14) and 31 smlds (2-13, 15-33) ( Figure S2b). Smlds were defined by drawing a straight line between landmarks 1 and 14 (the apices [curvature maxima] of the nodulus and cornu, respectively), and from the midpoint of this line, projecting a fan with 21 radiating lines at equal angles ( Figure S2b).
For seminal rod configuration in dorsal view, we used 10 lds (1-2, 8, 13, 21, 29, 34, 40-42) and 32 smlds (3-7, 9-12, 14-20, 22-28, 30-33, 35-39) (Figure S2c). Three straight lines were drawn between well-defined lds to define smlds. The first line was between the apex of the right arm of the seminal valve and the intersection of the outline of the seminal rod body and right arm (lds 8 and 13, respectively); the second line, between the apex of the seminal rod body and the maximum curvature of the seminal valve (lds 21 and 41, respectively); the third line was the left-side equivalent of the first line (lds 29 and 34). From these lines, we projected 6, 10, and 6 perpendicular lines, respectively.
For each specimen, the x, y coordinates of lds and smlds from each morphological structure were captured using the software TpsDig ver.140 [29]. To minimize the tangential variation of smlds, a coordinate adjustment was performed in SemiLand 6 [28]. Procrustes superimposition was performed in CoordGen6 to remove the effect of size, position, and rotation on position of coordinates of the individual configurations [26,27].
To obtain new variables that quantified the highest percentage of shape variation of these structures, relative warps analysis (RWA) was performed in PAST 3.12 [30] using the adjusted x, y coordinates matrix for both specimens and localities [27]. RWAs were performed using paired variance-covariance matrices among specimens or localities. Shape variation was plotted using the first two relative warps (RW1 vs. RW2; data not shown), and the change in the structure's configuration was visualized by thin-plate spline deformation grids in PAST 3.12.
Canonical variate analyses (CVAs) were performed to test whether the shape variation of these structures could be used to discriminate among possible geographic groups. Based on Procrustes results of the antennal club, seminal rod, and spermathecae, the specimens of different localities analyzed were a priori assigned to possible geographic groups. CVAs of specimens were performed from the matrix x, y configurations. Non-parametric multivariate analyses of variance (PERMANOVA) and post-hoc pairwise Hotelling's T-tests [27] were performed to evaluate differences among the groups recovered by CVAs. All multivariate analyses were performed using PAST 3.12.

Multivariate Analysis of Non-Geometric Morphological Data
The first three coordinates of the PCoA considering the 17 characters common to both sexes explained 43.8% of total variation (PCo1, 16.7%; PCo2, 14.1%; PCo3, 13%). The scatter plot of the two first components (PCo1 vs. PCo2) showed partial separation of the specimens into three recognizable geographical groups (Figure 4a). The first included specimens of the populations located in the western coastal states of the USA and British Columbia including Idaho, Montana, and southern Nevada (hereafter, "West group"); the second included specimens of the states adjacent to those of the west coast (Utah, Colorado, Arizona, New Mexico, and Texas), as well as those from Chihuahua and Durango states in the Sierra Madre Occidental (SMOC) in the North of Mexico (East-SMOC group); the third included specimens from Nuevo Leon, Coahuila, and Tamaulipas states in the Sierra Madre Oriental (SMOR) in northern Mexico (SMOR group). The specimens of both West and SMOR groups presented a slight overlap with those of the East-SMOC group (Figure 4a). Significant differences were found among these geographical groups (ANOSIM, R = 0.501; p ≤ 0.001). Pairwise Hotelling's T-tests support differences between groups: West vs. East-SMOC groups (R = 0.440, p ≤ 0.001), West vs. SMOR (R = 0.706, p ≤ 0.001), and East-SMOC vs. SMOR (R = 0.336, p ≤ 0.001).

Relative Taxonomic Weight of Individual Characters
The qualitative characters FS, SPD, RSP, SED, and PSS differed significantly (p < 0.05) among specimens from different geographical groups; however, only SPD was an exclusive character of the West group (Table S2). Beetles from this group possessed 2-3 sizes of pubescence (SPD) on the elytral declivity, whereas specimens from the other groups possessed only one (Figure 2g,h).
The ANOVA and Tukey tests showed that ten quantitative characters common to females and males (EBL, EPW, EL, DBE, HPL, PL, PW, EYL, AL, LMM) displayed statistically significant differences in at least one geographical group; however, these were not always presented in the same groups ( Table S3a). The individuals from SMOR showed higher average values in these characters than the specimens of other geographical groups. Specimens from East-SMOC group had lower average values of epistomal brush length (EBL), epistomal process width (EPW), eye length (EL), and pronotum width (PW) than individuals of West and SMOR groups; and they were smaller than individuals of these groups, except in the EYL and LMM characters. Specimens of the West group typically displayed intermediate average values between East-SMOC and SMOR groups.
Seven (EL, HPL, PL, PW, EYL, AL, MML) of 11 characters were statistically different among females ( Table S3b). The differences were also concentrated in measures of size, with SMOR females being larger than those from West and EAST-SMOC groups. The females from West and East-SMOC were very similar, showing significant differences in only four characters (EL, HPL, PW, and AL). In the case of males, seven (DFT, EBL, EPW, EL, PW, AL, MML) of 13 characters were statistically different (Table S3c); however, only three of them were different (PW, AL, MML). SMOR males were slightly larger than those of males from West and East-SMOC groups, and similar to females, the West males displayed intermediate mean values between East-SMOC and SMOR groups.
The configurations for antennae (atn), seminal rod in dorsal view (srdv), seminal rod in lateral view (srlv), and spermathecae (spmt) for D. brevicomis showed several anatomical regions where the shape variation was concentrated. RW1 of the deformation grid of atn corresponded to changes in the degree of curvature of the sensory bands in the antennal club and in the distance between the second and third sensory bands (Figure 5a and Figure S2a). RW2 corresponded to deformations in the proximal margin of the antennal club, the distance between the proximal margin of the antennal club and the first sensory band, and deformations of the lateral margins of the antennal club. The bivariate scatter plot of the RWs of atn, recovered specimens in West, East-SMOC, and SMOR as three partially overlapping clusters (Figure 5a and Figure S2a). In the CVA, all groups had the highest correct classification of individuals (100%) (Figure 6a). PERMANOVA supported statistical differences in the shape of this structure between West vs. SMOR (p ≤ 0.0001), West vs. East-SMOC (p ≤ 0.001), and East-SMOC vs. SMOR (p ≤ 0.001).
RW1 of the deformation grid of spmt corresponded to deformations in the degree of curvature in the nodulus margins, nodulus length, and curvature and symmetry of the cornu (Figure 5b and Figure  S2b). RW2 corresponded to variation in the length of the nodulus, the distance between nodulus and cornu, and the length and symmetry of the cornu. The plot of RWs of spmt separated all three geographical groups (Figure 5b). In the CVA, there was a slightly high percentage (80-90%) of correct classification of individuals from West and East-SMOC, but full classification of all SMOR specimens correctly classified (100%) (Figure 6b). PERMANOVA supported statistical differences in the shape of this structure between West vs. SMOR (p ≤ 0.0001) and East-SMOC vs. SMOR (p ≤ 0.002), but not between West vs. East-SMOC. RW1 of the srdv deformation grid corresponded to changes in the length of the lateral arms of the seminal rod valve and in the thickness at the center of the seminal rod body. RW2 corresponded to changes in the length of the seminal rod valve and symmetry of the lateral arms of the seminal rod valve ( Figure S2c). The plot of RWs of srdv indicated slight overlapping among all three geographical groups (Figure 5c). However, CVA indicated 100% accurate classification of specimens to groups (Figure 6c). PERMANOVA supported statistical differences in the shape of this structure among all geographic groups: West vs. East-SMOC (p ≤ 0.032), West vs. SMOR (p ≤ 0.0002), and East-SMOC vs. SMOR (p ≤ 0.0001).
Insects 2019, 10, x FOR PEER REVIEW 11 of 24 RW1 of srlv corresponded to variation in width at the center of the seminal rod body and the seminal valve. RW2 corresponded to variation in the degree of curvature of the dorsal and ventral margins of the seminal rod body, as well as in the length and thickness of the seminal valve ( Figure  S2d). The plot of RWs of srlv showed some overlap among all three geographical groups (Figure 5d). The CVA analysis correctly classified 89.5% of specimens to geographic group (Figure 6d). PERMANOVA supported statistical differences in the shape of this structure between West vs. SMOR (p ≤ 0.0001), and East-SMOC vs. SMOR (p ≤ 0.0001).     Figure  S2d). The plot of RWs of srlv showed some overlap among all three geographical groups (Figure 5d). The CVA analysis correctly classified 89.5% of specimens to geographic group (Figure 6d). PERMANOVA supported statistical differences in the shape of this structure between West vs. SMOR (p ≤ 0.0001), and East-SMOC vs. SMOR (p ≤ 0.0001).   RW1 of srlv corresponded to variation in width at the center of the seminal rod body and the seminal valve. RW2 corresponded to variation in the degree of curvature of the dorsal and ventral margins of the seminal rod body, as well as in the length and thickness of the seminal valve ( Figure S2d). The plot of RWs of srlv showed some overlap among all three geographical groups (Figure 5d). The CVA analysis correctly classified 89.5% of specimens to geographic group (Figure 6d). PERMANOVA supported statistical differences in the shape of this structure between West vs. SMOR (p ≤ 0.0001), and East-SMOC vs. SMOR (p ≤ 0.0001).

Discussion
Our PCoA of quantitative and qualitative with sexes combined indicates continuous phenotypic variation in populations of D. brevicomis, as indicated by Wood [14]. However, when females and males were analyzed separately, we recovered three geographical groups defined by discontinuities in morphological variation corresponding to West (females distinct), SMOR (males distinct), and East-SMOC (geographic zone occupied exclusively by the remaining specimens) (Figure 4). These findings indicate that variation is, in fact, not continuous; rather, natural division into groups is dependent on the sex examined. One other study has explored the relative contribution of each sex in the quantitative morphological differentiation of Dendroctonus species; however, no difference was observed [7].
The evaluation of nine discrete characters indicated that only one (size of the pubescence of the elytral declivity, SPD) was useful for separating West specimens from the other groups. West specimens possessed pubescence of differing lengths in the interstriae of the elytral declivity, whereas insects from other groups had only one length. Since this feature is external and readily observed at low magnification, it represents a taxonomic character that is particularly suitable for identification in the field. Although Wood [14] reviewed few specimens from Mexico, he found no reliable ways to distinguishing D. brevicomis (West group) from D. barberi (East specimens within East-SMOC Group).
Quantitative morphological characters showed considerable geographic variation among the three geographic groups. However, quantitative characters corresponding to body size possessed significant differences in average values among groups, with SMOR specimens being on average larger than those of the other groups. Furthermore, there was greater similarity in size between West and East-SMOC specimens than between either of these groups and SMOR. Univariate and multivariate statistical analyses have frequently been used to identify morphological differences among species of scolytines including Dendroctonus: D. monticolae Hopkins and D. ponderosae Hopkins [31], D. valens LeConte and D. terebrans Olivier [32], D. parallelocollis and D. approximatus [33], as well as species of the D. frontalis complex [6,34].

Geometric Morphometrics
Our findings on shape variation of the antennae and seminal rod in lateral view partially confirm the presence of three geographical groups, because scatter plots of the relative warps of these structures show significant segregation of specimens within the three geographic groupings ( Figure 5). Antennal club shape has received little attention as a taxonomic character in other scolytids including Dendroctonus species. In those studies where antennal club morphology has been used, significant differences in antennal club shape have been found between sibling species, including D. valens Le Conte and D. rhizophagus Thomas & Bright [23], D. frontalis Zimmermann and D. mesoamericanus Armendáriz-Toledano & Sullivan [7], and D. approximatus Dietz and D. parallelocollis Chapuis [33]. In addition, in D. vitei Wood, the presence of sensillae clustered into pit craters is a taxonomic attribute unique within the Dendroctonus frontalis species complex [35].
Likewise, despite substantial intraspecific variation, seminal rod shape has been used to separate members of the D. frontalis complex [2,7], confirming the taxonomic value of this structure [6,15,36,37]. In the present study, the scatter plot of seminal rod shape in dorsal view only partially recovered three geographic groups (Figure 5c,d). East-SMOC individuals exhibited broad variation in shape of the seminal rod body that overlapped with that of the West and SMOR groups. However, there were consistent differences among groups focused in the seminal rod valve (Figure 6d), and SMOR and West were well resolved in dorsal view.
Additionally, the three geographical groups could be discriminated by shape of the female genitalia (Figure 5b). Spermathecae of West females possessed a nodulus that was wider than the cornu and had an oval shape, East-SMOC females had a nodulus with rounded anterior and posterior edges and a distally-widened, ovate cornu, and SMOR females possessed a nodulus that was slightly wider than the cornu, and a cornu similar to East-SMOC specimens. The female genitalia have rarely been used in the taxonomy of the genus Dendroctonus; however, some discrete attributes have proven to be useful for separating D. frontalis from D. mexicanus [38], D. frontalis from D. mesoamericanus [7], D. vitei from D. mexicanus [39], and D. approximatus from D. parallelocollis [33].

Taxonomic Considerations
Traditionally, the species status of members of the genus Dendroctonus has been supported by biological and ecological data, together with external morphological characters [12,15,40]. In addition, cytogenetic evidence, crossbreeding tests, molecular and biochemical data, and comprehensive analysis of morphological variation, e.g., [6,7,39,41], have been used to confirm, segregate, or describe new species.
Previous morphological [12], biological [6], genetic [16], pheromone [17], and molecular [18] studies support the conclusion that D. brevicomis is constituted by two distinct taxa corresponding to our West and East-SMOC groups within the USA; however, these studies did not analyze specimens across the entire range of D. brevicomis. Taking into consideration the genetic data of Kelley et al. [16] and available data on distribution, Bright [42] designated populations corresponding to our West and East-SMOC groups within the USA as subspecies of D. brevicomis in his taxonomic monograph of the Bark and Ambrosia Beetles of the West Indies.
Our research further demonstrates that D. brevicomis can be morphologically differentiated into a third geographically-delimited group (SMOR) in northeastern Mexico that appears to be distinct from the groups whose ranges are limited to or include the USA (i.e., West and East-SMOC, respectively). The specimens from these geographic areas displayed non-overlapping quantitative characteristics of external and reproductive morphology. However, we discovered only one qualitative character (pubescence length of the elytral declivity) that allowed reliable separation of the West group from the two other groups, and we found no such character to reliably separate East-SMOC from SMOR. These morphological differences between specimens from West and East-SMOR/SMOC support the hypothesis that D. brevicomis is integrated by at least two different species existing on either side of the Great Basin, Mojave, and Sonoran deserts: D. brevicomis and D. barberi, as originally designated by Hopkins [12] and then synonymized by Wood [14]. Several lines of evidence support this hypothesis. First, genetic differences between populations on either side of the three western American deserts [16,18] are similar in degree to those existing between recognized sibling species or subspecies of Dendroctonus or other bark beetles [43][44][45][46]. Second, the composition of the aggregation pheromone differs between these populations [17,19], with the major female-produced component of the aggregation pheromone of populations west of the Great Basin being exo-brevicomin, and that of populations east of the Great Basin being endo-brevicomin [17,19]. Third, our observations in the field indicate that West specimens produce subtransversely-winding parental tunnels, whereas parental tunnels of East-SMOC specimens are distinctly transversely winding and more aligned with the bole, as was reported previously for D. brevicomis and D. barberi, respectively, by Hopkins [12].
Therefore, based on morphometric analyses in this paper, our observations of gallery construction, and the earlier genetic and chemical ecology evidence [2,[16][17][18][19], we removed Dendroctonus barberi Hopkins from synonymy with Dendroctonus brevicomis LeConte and updated the descriptions for both taxa. In addition, due to broad morphological similarity observed between populations of East-SMOC both in Mexico and USA, we included both in D. barberi, although Mexican populations of this group were not analyzed in previous genetic or chemical ecology studies. The type specimens for D. brevicomis (the West Group of the present study) were deposited at the Harvard University Museum of Comparative Zoology, whereas those of D. barberi (East-SMOC Group of the present study) were deposited at the Smithsonian National Museum of Natural History.

Dendroctonus brevicomis LeConte 1876
The redescription of this species is based on a sample of 133 females and 133 males. Photos and illustrations were taken and drawn directly from these specimens, and the external morphology cross-checked with type specimens.
Distribution. It occurs in the USA and Canada: From southern British Columbia along the western coastal states of the USA to southern California, as well as in the states of Idaho and Montana.

Dendroctonus Barberi Hopkins 1909
The redescription of this species is based on a sample of 119 females and 119 males. Photos and illustrations were taken and drawn directly from these specimens and the external morphology crosschecked with type specimens. Gallery. Parental tunnels subtransversely-winding (Figure 8f) [12]. Larval galleries short, narrow, and on both sides of parental gallery.
Distribution. It occurs in the USA and Canada: From southern British Columbia along the western coastal states of the USA to southern California, as well as in the states of Idaho and Montana.

Dendroctonus Barberi Hopkins 1909
The redescription of this species is based on a sample of 119 females and 119 males. Photos and illustrations were taken and drawn directly from these specimens and the external morphology cross-checked with type specimens.

Description of the Female
Total length, 2.6-4.7 mm (X = 3.91 mm); head black to dark brown; pronotum with somewhat lighter color, elytra same color as pronotum (Figure 9a (Figure 10a). Pronotum: Similar to females, except in length (0.64-1.15 mm, X = 0.97mm) and width (1.02-1.72 mm, X = 1.42); widest on posterior third, anterior region slightly constricted. Lacking pronotal callus of female. Elytra: Similar to female specimens, except in the length (1.69-3.04 mm, X = 2.39 mm), 1.6 times longer than wide, 2.7 times longer than pronotum; sides parallel on the proximal sides, rather broadly rounded in the distal region; declivity convex with striae strongly impressed (Figures 9f-h and 10b). Genitalia: Seminal rod with the same elements as D. brevicomis male specimens; this structure displays two curved lateral arms in dorsal view, which are joined to the seminal rod base.  Dorsal process of the seminal rod (seminal rod body) enclosed within a triangular plate (Figure 10d,e). Seminal rod entire and slightly longer in the seminal rod body than D. brevicomis with a seminal rod body that ends in an apex; in dorsal view, length of the seminal rod body from four to five times the length of seminal valve ( Figure 10d); lateral arms of seminal rod curved and not fused. The lateral view of the seminal rod is thickened distally; seminal valve of the seminal rod body without a projection (Figure 10e).
Distribution. This species occurs in the USA and Mexico: From Utah, Nevada, Colorado, Arizona, New Mexico, western Texas, Chihuahua, and Durango. Gallery. Transversely winding (Figure 10f) [12].
Distribution. This species occurs in the USA and Mexico: From Utah, Nevada, Colorado, Arizona, New Mexico, western Texas, Chihuahua, and Durango.

Final Considerations
Currently, D. brevicomis is geographically isolated from D. barberi by the Great Basin, and Mojave and Sonoran deserts in North America. Based on the results, we hypothesize that the distribution of D. brevicomis occurs from British Columbia, Canada throughout western USA to southern California [12,15], including Idaho and Montana; whereas D barberi occurs in Nevada, Utah, Colorado, Arizona, New Mexico, and Texas, USA, and in Chihuahua and Durango, Mexico. This geographic partitioning of the insect taxa reflects that of their host species [48]: The D. brevicomis range coincides with that of host species P. ponderosa, P. benthamiana, and P. coulteri; whereas D. barberi occurs with hosts P.

Final Considerations
Currently, D. brevicomis is geographically isolated from D. barberi by the Great Basin, and Mojave and Sonoran deserts in North America. Based on the results, we hypothesize that the distribution of D. wbrevicomis occurs from British Columbia, Canada throughout western USA to southern California [12,15], including Idaho and Montana; whereas D barberi occurs in Nevada, Utah, Colorado, Arizona, New Mexico, and Texas, USA, and in Chihuahua and Durango, Mexico. This geographic partitioning of the insect taxa reflects that of their host species [48]: The D. brevicomis range coincides with that of host species P. ponderosa, P. benthamiana, and P. coulteri; whereas D. barberi occurs with hosts P. arizonica var. stormiae, P. brachyptera, P. scopulorum, and P. arizonica Shaw and P. engelmannii Carrier Mexico [2,48].
Lastly, an unexpected finding in our study was the strong morphological differentiation of SMOR, a population of D. brevicomis occurring only in Mexico (Coahuila, Nuevo León, and Tamaulipas states). Morphometric differences of the spermatheca and seminal rod in dorsal view were more pronounced between SMOR and both D. barberi (East-SMOC) and D. brevicomis (West) than differences between these latter two groups. Nonetheless, we were unable to identify diagnostic external characters for differentiating SMOR from East-SMOC. Thus, we conclude that the SMOR specimens might be a distinct species from D. barberi and D. brevicomis. However, molecular and additional studies are necessary before a formal description of this taxon can be justified. For now, SMOR specimens should be considered as D. barberi.

Conclusions
In this study, we provide lineal and geometric morphometric evidence to reinstate Dendroctonus barberi, and remove it from synonymy with D. brevicomis. The pubescence length on the elytral declivity, the genitalia shape both female and male in dorsal view, antennae shape and curvature of the three sensory bands, and the parental tunnels shape are robust characters for segregation of these species (Figures 8 and 10). In addition, molecular and chemical ecology data published and in progress support this decision. On the other hand, differences in some quantitative and qualitative characters, as well as the spermatheca and seminal rod shape of SMOR specimens, suggest that these populations might be a different species from D. barberi and D. brevicomis, which should be confirmed in future studies.
Supplementary Materials: The following are available online at http://www.mdpi.com/2075-4450/10/11/377/s1: Table S1. Localities of D. brevicomis samples analyzed; Table S2. Qualitative morphological characters analyzed among geographical groups; Table S3. ANOVA and Tukey test results of continuous morphological characters among females and males, males alone, and females alone of different geographical groups; Figure S1. Landmarks and semilandmarks in antennae, spermathecae, and seminal rod in dorsal and lateral view; Figure S2