An Assessment of Old and New Characters Using Traditional and Geometric Morphometrics for the Identification of Dendroctonus approximatus and Dendroctonus parallelocollis (Curculionidae: Scolytinae)

Abstract

There have been numerous changes in the nomenclature of species of Dendroctonus. The case of Dendroctonus parallelocollis Chapuis and Dendroctonus approximatus Dietz has been particularly noteworthy. These bark beetles colonize pine trees but are not aggressive species. The high morphological similarity, their overlapping geographical distribution, and the limited knowledge of these species have resulted in taxonomic confusion. The aim of the present study was to assess morphological characters reported in the literature and to scrutinize new characters that might be useful for the separation of these species. We evaluated 26 morphological attributes and performed geometric morphometrics analysis of the antenna, pronotum, spermatheca, and seminal rod to test if the variations in the shape of these structures allow the recognition of additional discrete characters to differentiate them. Our results show that five double-state characters of external morphology are useful to identify these species, and the shape of antenna, spermatheca, and seminal rod.

The genus Dendroctonus Erichson has been revised several times and consequently the species included has changed with each reviews (Hopkins 1909; Wood 1963, 1982). These reviews were mainly based on external and internal morphological characters (e.g., beetle size and color, and features of the head, pronotum, elytra, and male genitalia), behavioral characteristics (e.g., galleries, position and arrangement of egg niches, aspects of larval mines and feeding, and pupation), and biological attributes (e.g., life cycles, preferential hosts, and distribution range). Subsequent studies, on particular species incorporated attributes such as chromosome numbers, pheromones, cross breeding, molecular markers, and cuticular hydrocarbons. These attributes confirmed some species, described new species or suggested the presence of cryptic species (Lanier and Wood 1969; Lanier 1981; Bentz and Stock 1986; Lanier et al. 1988; Pajares and Lanier 1990; Furniss 1996; Kelley et al. 1999; Ruíz et al. 2009; Sullivan et al. 2012; Armendáriz-Toledano et al. 2014a, 2015). Currently, 20 species are recognized in Dendroctonus, one of them with two subspecies (Furniss 2001) of which 18 inhabit the American continent (Wood 1982; Armendáriz-Toledano et al. 2015, 2017) and two are native to Europe and Asia (Grégoire 1988).

Whereas some Dendroctonus species have been widely studied because they are major natural disturbing agents of coniferous forests during outbreaks, others such as Dendroctonus parallelocollis Chapuis and Dendroctonus approximatus Dietz are little known because they are not aggressive species (Six and Bracewell 2015). As currently understood in the most recent review (Wood 1982), these species are well defined. However, due to confusion of types (Wood 1969), most earlier publications have used incorrect identifications (Hopkins 1909, Schedl 1956, Wood 1963). Dendroctonus parallelocollis was described in 1869 for specimens from Durango, Mexico, whereas D. approximatus was recognized in 1890 for specimens from New Mexico and Colorado. Wood (1969) concluded that all previous usage of these names except for the original descriptions was incorrect. Dendroctonus approximatus has been treated as a member of the Dendroctonus frontalis complex (Wood 1982).

Since then, a minimal character set has been available to identify both species, including frontal sculpture, pronotum shape, elevation of epistomal process, and the elytral vestiture (Wood 1982). In addition, phylogenetic and chromosomal analyses have confirmed and recognized both taxa as independent and valid species (e.g., Lanier 1981, Kelley and Farrell 1998, Zúñiga et al. 1998, Víctor and Zúñiga 2016). Nevertheless, the broad morphological similarity of both species, the geographical variation of some morphological attributes (e. g., shape of pronotum), and the overlapping of their range distribution (Salinas-Moreno et al. 2004, 2010) frequently lead to erroneous taxonomic identification of both species.

In this study, we assess the consistency of morphological attributes previously used to identify these species (Lanier 1981, Wood 1982, Armendáriz-Toledano and Zúñiga 2017) using traditional morphometrics. In addition, we look for new discrete characters, as well as their usefulness in the identification of both taxa. Further, we performed a geometric morphometric analyis to determinate whether the variation in shape of the antenna, pronotum, seminal rod, and spermatheca is useful for this purpose.

Materials and Methods

In total, 144 specimens 39 D. approximatus and 105 D. parallelocollis were analyzed from 12 and 17 geographical sites, respectively (Table 1). Specimens were collected directly from naturally infested pines. Sex of the specimens was determined by the presence of stridulatory apparatus in males (Hopkins 1909, Lyon 1958). The preliminary identification of species was performed using Wood’s (1982) key. The removal, dissection, and mounting of the antenna, elytra, seventh tergite, and genitalia were performed following López et al. (2014) and Armendáriz-Toledano et al. (2015).

Table 1.

Country, reference acronym, locality, coordinates, altitude, and host of D. parallelocollis (par) and D. approximatus (app) specimens examined in this study

Species	Locality	Coordinates	Altitude (m)	Host
Mexico
app	Coahuila, Saltillo, Sierra de Arteaga (CoSA)	25°25′42.79″, −100°50′34.66″	1,750	Pinus rudis
app	Durango, Guanacevi (DGU)	25°55′53.89″, −105°57′16.83″	2,069	Pinus sp.
app	Jalisco, Atenquique, El llanito (JAE)	19°31′56.97″, −103°26′41.01″	2,230	Pinus teocote
app	Jalisco, Mpio Tapalpa, Alpuyeque (JTA)	19° 59′ 5.87″, −103° 40′ 8.16″	2,270	Pinus michoacana
app	México, Axapusco, Francisco I. Madero (MAF)	19°46′44.62″, 98°38′19.52″	2,870	Pinus rudis
app	México, Villa Allende, Buenavista 23 (MVB)	19°28′37.02″, −100°8′8.014″	2,846	Pinus pseudostrobus
app	Querétaro, La Pingüica (QPI)	20°22′7″, −100°0′45″	1,963	Pinus sp.
app	Querétaro, Pto Oro (QPO)	21°8′6.67″, −99°37′31″	2,351	Pinus arizonica
app	Oaxaca, Portillo, Cajones (OPC)	16°14′20.21″, −96°31′21.08″	2,182	Pinus sp.
both	Durango, San Quintin Barrancas (DSQ)	23°50′44.62″, −104°14′32.8″	1,741	Pinus sp.
both	Jalisco Gómez Farías (JAL)	19°46′49″, −103°29′30.99″	1,521	Pinus sp.
both	Oaxaca, San Juan Tepeuxila (OST)	17°44′08″, −96° 49′53.2″	2,329	Pinus sp.
par	Michoacán Zirahuén (MZH)	19°26′40.09″, −101°43′32.39″	2,140	Pinus sp.
par	Michoacán, Ario, Puerta Pesada (MPP)	19°12′7.63″, −101°43′43.67″	1,974	Pinus sp.
par	Michoacán, Tingambato, El Ciprés (MTC)	19°30′33.8″, −101°54′7.3″	2,074	Pinus sp.
par	Michoacán, Uruapan, km 10, Carr Uruapan (MUU)	19°24′23″, −102°2′35″	1,615	Pinus pringlei
par	Michoacán, Uruapan, San Lorenzo (MUL)	19°31′32.70″, −102°6′40.53″	2,107	Pinus sp.
par	Michoacán, Zacapu, La Cofradía (MZC)	19°48′45.04″, −101°47′44.41″	2,600	Pinus leiophylla
par	México, Amatepec, La Goleta (MAG)	18°40′40.14″, −100°5′58.63″	2,200	Pinus leiophylla
par	México, Atlautla, 3 km NE San Miguel (ASM)	19°1′36.57″, −98°46′46.8″	2,367	Pinus sp.
par	Tlaxcala, Temetzontla (TTE)	19°21′1″, −98°17′19″	2,571	Pinus sp.
par	Veracruz, Jalacingo, Tepeyac (VJT)	19°47′45.09″, −97°18′22.76″	1,939	Pinus oaxacana
par	Oaxaca, Concepción, Cuicatlán (OBL)	17°51′04,8″, −96°54′56.5″	1,810	Pinus pringlei
par	Querétaro, El Tepozán (QET)	20°11′9″, −99°59′44″	2,383	Pinus leiophylla
Guatemala
par	Guatemala, Zacapa, San Antonio El Chico (GZUAC)	14°57′ 25.11″, −89°46′53.56″	321	Pinus sp.
Honduras
par	Honduras, Potrerillos, Zihuatepeque (HPZI)	14°35′13.161″, −87°49′48.32″	1,075	Pinus sp.

The external morphology and specific microstructures (antenna, spermatheca, and seminal rod) of each specimen were examined and photographed with a Nikon DXM1200F digital camera on a NIKON SMZ800 stereo microscope. Measurements of quantitative characters of the body were taken with an optical micrometer at 20–63× and those of mounted structures with a optical micrometer at 400× directly from slides. The images of the pronotum of each insect were obtained by placing the insects in dorsal view in a Petri dish with fine sand and 70% alcohol and photographed at 40× with an Olympus C-5060 camera. The seminal rod (distal process pointing up), spermatheca, (cornu oriented down), and antenna (scape pointing down) were always arranged in the same position. These structures were photographed with a Panasonic DMC-FS3 camera with a phase contrast microscope (Carl Zeiss, Germany) at 400×.

Characters Analyzed

In total, 26 morphological characters from head, antenna, pronotum, elytral declivity, and genitalia were selected, of which 10 were double-state (DB, binary characters that have two states: presence/absence, yes/no, male/female) and 16 quantitative (QC) characters. The coding and description of these characters are listed below from anterior to posterior.

• 1. Pubescence on head (PH): The surface of the frons in Dendroctonus species is covered with pubescence that varies in density and distribution (Hopkins 1909, Wood 1982). Dendroctonus parallelocollis has inconspicuous pubescence in this region (Wood 1963, 1982), whereas D. approximatus has a rather long and sparse pubescence (Supp Table 1 [online only]; Figs. 1 and 2)

• 2. Frontal sculpture of males (FS): Males of some species of Dendroctonus have numerous and prominent granular tubercles on the lateral areas of the frons (Wood 1982, Lanier et al. 1988). The frons in D. parallelocollis is deeply punctured with rather abundant, isolated, and scattered granules (Wood 1963, 1982), whereas in D. approximatus the frons is coarsely granulate with a pair of prominent lateral tubercles (Dietz 1890; Wood 1963, 1982) (Supp Table 1 [online only]; Figs. 1–4).

• 3. Head surface (HS). Species of D. frontalis complex show a median groove in this region, but not the other species from this genus (Wood 1963, 1982). The head of D. parallelocollis is strongly and evenly convex, whereas the D. approximatus head displayed a deep median groove (Dietz 1890, Hopkins 1909, Wood 1982) (Supp Table 1 [online only]; Figs. 1 and 2).

• 4. Shape of epistomal process (SEP): The epistomal process is a structure composed of a pair of lateral elevations (arms) along of the epistomal margin. The relative width of the elevation from the epistomal margin and curvature of the arms show differences that are useful for the identification of Dendroctonus species or species groups (Hopkins 1909; Wood 1963, 1982). The epistomal process of D. parallelocollis is broad and flat, without elevated arms (Wood 1963, 1982), whereas D. approximatus has arms elevated and oblique almost 40° from horizontal (Dietz 1890; Wood 1963, 1982; Supp Table 1 [online only]; Figs. 3 and 4).

• 5. Distribution of sensillae on third sensorial band of the antennal club (DST): The distribution of sensillae has been taxonomically informative in some Dendroctonus species (López et al. 2014, 2018; Armendáriz-Toledano et al. 2017). This character had not been evaluated in D. parallelocollis and D. approximatus (Supp Table 1 [online only]; Figs. 5 and 7).

• 6. Organization of sensillae in the sensorial bands on the antennal club (OSB): The organization of sensillae on sensorial bands differs among members of the genus Dendroctonus (López et al. 2014, 2018). This character had not been evaluated in the species studied here (Supp Table 1 [online only]; Figs. 5 and 7).

• 7. Disposition of sensorial bands on antennal club (DSA): Differences in the curvature of the sensorial bands have been recognized in several species of Dendroctonus (Hopkins 1909). This character had not been previously evaluated in species studied here (Supp Table 1 [online only]; Figs. 5, 7, and 8).

• 8. Surface of preepisternal area (SPA): The antero-lateral area of the pronotum is divided into pre-episternal, episternal, and epimeral areas. In some members of the D. frontalis complex, the small grooves on the pre-episternal area differentiate some species (Armendáriz-Toledano and Zúñiga 2017). This feature has not been evaluated before in the species studied here (Supp Table 1 [online only]; Fig. 9).

• 9. Abundance of pronotal vestiture (PV): The vestiture of the pronotum varies in density and distribution. The vestiture in D. parallelocollis is generally scanty but more abundant and coarse in the anterolateral region of pronotum, where it becomes longer, and rather coarse. The vestiture on both regions of the pronotum of D. approximatus is sparse and long (Wood 1963, 1982; Supp Table 1 [online only]; Figs. 10 and 11).

• 10. Color of the vestiture on elytral declivity (VCD): This character was considered in the phylogenetic analysis of the genus Dendroctonus. All species had yellowish setae except for D. approximatus with dark setae (Víctor and Zúñiga 2016); however, the character has not been evaluated geographically in Dendroctonus species (Supp Table 1 [online only]; Figs. 12 and 13).

Figs. 1–13.

Anatomical structures of D. parallelocollis (1, 3, 5, 7, 8, 9, 11, and 13) and D. approximatus (2, 4, 6, 10, and 13). (1 and 2) male head in dorsal view, (3 and 4) male frons in frontal view, (5–8) anterior side of antennal club, (9) preepisternal area, (10 and 11) pronotum in dorsal view, (12 and 13) color of vestiture on elytral declivity. ep, epistomal process; ft, frontal tubercles; in, intersected; mg, median groove; pb, pubescence; sb, sensorial band; st, striae.

Quantitative characters include the following: (11) width of epistomal process (WEP), (12) width of antennal club (WAC, Fig. 14), (13) length of antennal club (LAC, Fig. 14), (14) eye width (EW), (15) distance between eyes (DBE), (16) head width (HW), (17) head length (HL), (18) pronotum length (PL), (19) pronotum width (PW), (20) elytral length (EL), (21) total length of the body (TLB), (22) nodulus width of the spermatheca (NW, Fig. 16), (23) cornu width of the spermatheca (CW, Fig. 16), (24) distance between nodulus and cornu (DCN, Fig. 16), (25) seminal rod length (SRL, Fig. 17), and (26) seminal rod width (SRW, Fig. 17). Head (WEP, EW, DBE, HW, HL), pronotal (PL, PW), elytral (EL), and TLB attributes were measured according to Armendáriz-Toledano et al. (2017; Figs. 1, 3, and 4).

Figs. 14–17.

Anatomy, measures, position of landmarks with respective combs or fans, and resulting configuration of the structures evaluated, that apply to both species. (14) Antenna, (15) pronotum, (16) spermatheca, and (17) seminal rod. a, antenomere; co, cornu; no, nodulus; sb, sensorial band; srb, seminal rod body; and sv, seminal valve.

Traditional Morphometric Analysis

Univariate Statistics

Because the statistical tests that are used assume a normal distribution of the data, the homogeneity of variances and normal distribution for continuous quantitative characters were evaluated with the F-test and Shapiro–Wilk test, respectively (Zar 2010). Differences between quantitative characters were estimated by comparing mean values with the Student’s t-test (Zar 2010) and a Guillaumin profile (Guillaumin 1972). Statistical differences between qualitative characters of both species were estimated using the Mann–Whitney U-test (Zar 2010).

Multivariate Statistics

Pairwise similarities among females, males, and both sexes of each species were calculated using Gower index (Gower 1971). To explore whether the variation pattern in a multidimensional space segregates specimens according to their species, we performed separated principal coordinates analyses (PCoAs) using the pairwise similarity matrices of females, males, and females–males of both species. Likewise, to test whether specimens of one species are more similar to each other than with any member of another species, cluster analyses with the unweighted pair-group method with arithmetic average were carried out using the corresponding similarity matrices (Legendre and Legendre 1998).

The characters included in each analysis were different due to the differences in numbers of sex-specific characters examined. The comparison between females of both species was performed on 50 D. parallelocollis and 21 D. approximatus using 23 characters (PH, HS, SEP, DST, OSB, DSA, SPA, PV, VCD, WEP, WAC, LAC, EW, DBE, HW, HL, PL, PW, EL, TLB, NW, CW, and DCN); the comparison between males on 55 D. parallelocollis and 18 D. approximatus using 23 characters (PH, FS, HS, SEP, DST, OSB, DSA, SPA, PV, VCD, WEP, WAC, LAC, EW, DBE, HW, HL, PL, PW, EL, TLB, SRL, and SRW) and the comparison of both sexes were performed on 105 D. parallelocollis and 39 D. approximatus using 20 characters (PH, HS, SEP, DST, OSB, DSA SPA, PV, VCD, WEP, WAC, LAC, EW, DBE, HW, HL, PL, PW, EL, and TBL).

Geometric Morphometrics

Patterns of two-dimensional shape variation of antennae, pronotum, seminal rod, and spermatheca of both bark beetles were quantified using appropriate landmarks (lm) and semilandmarks (sml) (Figs. 14–17; Bookstein 1991, Zelditch et al. 2004). The antenna club shape was defined by eight type I lm, and 26 sml (Fig. 14), the pronotum with two type I lm, two type II lm, and 38 sml (Fig. 15), the spermatheca with two type II lm and 24 sml (Fig. 16), and the seminal rod with five type II lm and 16 sml (Fig. 17).

Semilandmarks were defined using fans (on spermatheca) and combs (on pronotum, seminal rod, and antenna) on photographs of these structures with MakeFan6 of Integrated Morphometrics Package (IMP; Sheets 2003). Semilandmarks were specific sites located in every intersection between outline of each structure and the projected lines from reference points (putative homologous points) easily recognized in all images. In the antenna, the intersection points between the second sensorial band and their respective lateral margins were used to project nine perpendicular and equidistant lines (Fig. 14). Two combs of 12 parallel lines each were located in the pronotum; the first (perpendicular) located from the points of maximum curvature of posterior sides of pronotum and second (longitudinal) from points of maximum curvature of anterior and posterior median curvatures (Fig. 15). Points of maximum curvature in spermatheca, both nodulus and cornu, were used to project a fan of 25 radiating lines at equal angular displacements (Fig. 16). In seminal rod, 10 perpendicular lines were projected from the point of maximum curvature of distal process and the point of maximum curvature between the seminal valve and seminal rod body (Fig. 17).

Landmarks and semilandmarks were digitalized as two-dimensional coordinates with tpsDig 1.40 software (Rohlf 2004) and together constituted the shape configurations. A generalized procrustes analysis (Zelditch et al. 2004) was performed to remove scale effects, position, and orientation with CoordGen6 program of IMP (Sheets 2003). To minimize the tangential variation of curvatures in the shape configurations produced by the use of sml, the minimum Procrustes distance criterion (Perez et al. 2006) was used in all morphological structures, allowing an optimal superimposition with SemiLand6 of IMP (Sheets 2003). These coordinates were transformed to new variables (relative warps, RW) by means of relative warps analysis (Zelditch et al. 2004), which quantify highest variation of shape. Shape variation of each structure was analyzed with the first five RWs and shape changes were visualized by mean of deformation grids obtained with Thin-Plate Spline technique. Patterns of shape variation between species were visualized by means of scatter plots between RW1 and RW2 with the help of their respective deformation grid diagrams.

A multivariate analysis of variance (MANOVA) with the first five RWs was performed for each morphological structure. In particular, for pronotum shape analysis, a canonical variate analysis (CVA, Legendre and Legendre 1998) was performed with these RWs using species and sex as a priori groups. The resulting groups were contrasted with a MANOVA test and its respective pairwise Hotelling’s T-squared comparisons.

Results

Analysis of Characters

Five double-state characters showed differences between D. approximatus and D. parallelocollis: FS (Figs. 1 and 2), HS (Figs. 1 and 2), SEP (Figs. 3 and 4), OSB (Figs. 5 and 6), and DSA (Figs. 5 and 7). Males of D. parallelocollis do not have frontal tubercles (Fig. 1), and both sexes have a uniformly convex frons (Fig. 1). The lateral arms of their epistomal process are flat and straight (Fig. 3), and the antennal club has only one row of sensillae on each sensorial band (Fig. 6). In some specimens, the sensorial bands intersect at some point on the club (Fig. 7). Males of D. approximatus have prominent frontal tubercles (Fig. 2), and in both sexes, the frons has a median groove from the upper level of the eyes to the epistomal area (Fig. 2). The lateral arms of their epistomal process are strongly elevated and curved (Fig. 4); the antennal club has three rows of sensillae in the first and second sensorial bands (Fig. 6); and all sensorial bands are distributed independently, i.e., they do not intersect in anterior surface (Fig. 5).

Student’s t-test showed differences in 12 quantitative continuous characters between these species (Supp Table 2 [online only]): WEP, WAC, LAC, EW, DBE, HW, PL, EL, TLB, NW, SRL, and SRW. The characters WEP, WAC, LAC, EW, HW, NW, SRL, and SRW had higher mean values in D. approximatus than in D. parallelocollis, whereas DBE, PL, EL, and TLB values were higher in the second species than in the first (Supp Table 1 [online only], Fig. 18). The Guillaumin profile showed that the most different attributes were WEP, LAC, NW, and SRW (Fig. 18); however, despite these differences, their characters were overlapping in their maximum and minimum values between species.

Fig. 18.

Guillaumin profile based on quantitative continuous characters. t-test: * P ≤ 0.05; ** P ≤ 0.001. For each variable the points correspond to standarized differences between species (circle for D. parallelocollis and square are represented to D. approximatus).

Multivariate Analysis

The first three principal coordinates of PCoAs of each morphological comparisons explained >59% of variation: males–females (PCo1_♀♂ = 44.4%, PCo2_♀♂ = 8.1%; PCo3_♀♂=7.1%; Fig. 19), females (PCo1_♀= 48.0%; PCo2_♀= 12.0%; PCo3_♀= 6.6%; Fig. 20), and males (PCo1_♂ = 44.14%; PCo2_♂= 11.83%; PCo3_♂=7.42%; Fig. 21). Scatterplots of the first two principal coordinates (PCo1 vs PCo2) of these comparisons showed the special segregation of discrete phenotype groups according to each species (Figs. 19–21). Likewise, dendrograms showed clustering of D. parallelocollis specimens separated of those of D. approximatus. However, in the case of female–male and female comparisons, the specimens of D. approximatus were separated in two clusters, one of them more similar with respect to the cluster of D. parallelocollis (Figs. 19–21).

Figs. 19–21.

Principal coordinate analysis and dendograms from cluster analysis of morphological characters of D. parallelocollis and D. approximatus. (19) Males and females combined. (20) Males. (21) Females. Bootstrap values after 1,000 pseudoreplicates are indicated at the nodes. The scale of the cluster of male and cluster of female is shared (downward).

Geometric Morphometrics

Antenna

The superimposition of antenna configurations of D. approximatus (13 ♀ and 11 ♂) and D. parallelocollis (33 ♀ and 34 ♂) specimens showed that shape variation is located on the sensorial bands and the relative length of the antennal club. The first two RW explained 62.7% of total variation (RW1 = 35.07%; RW2 = 21.37%). The respective two-dimensional scatterplot of these RWs revealed two discrete groups corresponding to each species (Fig. 22). The deformations in the components RW1 and RW2 were related with the elongation of antennal club and the curvature of sensorial bands (Fig. 22). The MANOVA supported statistically significant differences in antennal shape between species (λ_Wilk’s = 0.19; F = 42.42; P ≥ 0.001).

Figs. 22–23.

Scatter plots between the first and second relative warps with its respective deformation grids ±3 SD, corresponding to shape analysis of antennal club (22) and pronotum in dorsal view (23). A scatter plot between the first and third axis obtained from CVA using 10 first relative warps is shown to the right of the scatter plot of pronotum (RW1 versus RW2).

Pronotum

The superimposition of pronotum configurations of D. approximatus (23 ♀ and 21 ♂) and D. parallelocollis (29 ♀ and 25 ♂) specimens showed that the sites with a high variation are the lateral profiles. The first two components of the relative warp analysis explained 97.02% of variation (RW1 = 81.55%; RW2 = 15.87%). Two-dimensional scatterplot of these RWs showed the segregation of four overlapped groups corresponding to species and sexes (Fig. 23). Deformations in the two first components RW1 and RW2 were related to the variation in the curvature of lateral sides, which can be strongly curved or straight (Fig. 23). The MANOVA with its respective post hoc pairwise comparisons using Hotelling’s test showed statistically significant differences in pronotum shape (λ_Wilk’s = 0.25; F = 7.54; P ≥ 0.001) between sexes within each species and between species (P ≤ 0.05). The CVA from first five relative warps among sexes and the species (scatter plot of CV1 = 67.76% vs CV3 = 7.35%) recovered the specimens clustered in four discrete groups corresponding to sexes and species (Fig. 23).

Spermatheca

The superimposition of spermatheca configurations of D. approximatus and D. parallelocollis females showed that the sites with high variation are the curvature of spermatheca and length–width of nodulus and cornu. The first two components of the RWs explained 61.7% of variation (RW1 = 45.02%; RW2 = 21.72%). Two-dimensional scatterplots of these RWs segregated two discrete groups corresponding to each species (Fig. 24). The RW1 deformation was associated with the curvature of spermatheca and RW2 deformation with the width of the entire spermatheca (Fig. 24). The MANOVA showed statistically significant differences in spermatheca shape between the species (λ_Wilk’s = 0.17; F = 30.07; P ≤ 0.001).

Figs. 24–25.

Scatter plots of the first and second relative warps with respective deformations grids at ±3 standard deviation (SD). (24) Spermatheca. (25) Seminal rod.

Seminal Rod

The superimposition of seminal rod configurations of D. approximatus and D. parallelocollis males showed that sites with concentrated variation were the seminal rod body and seminal valve. The first two relative warps explained 57.8% of the variation (RW1 = 34.03%; RW2 = 23.83%). The two-dimensional scatterplot between RWs separated specimens according to species (Fig. 25). The MANOVA supported statistically significant differences in seminal rod shape between species (λ_Wilk’s = 0.38; F = 10.52; P ≤ 0.001).

Discussion

Analysis of Characters

Our results support that 5 of the 10 double-state characters of external morphology are useful in identifying these species (FS, HS, SEP, OSB, and DSA). OSB and DSA are noteworthy because they are unamiguous characters, which have not been previously reported. Antennal characters have also been reported as diagnostic in other sibling species of the genus (e.g., Dendroctonus mexicanus Hopkins–D. vitei Wood, and D. rhizophagus Thomas and Bright–D. valens LeConte) (López et al. 2014, Armendáriz-Toledano et al. 2017).

Our results indicate that characters FS and HS, previously reported in other studies to identify these species (Wood 1982, Lanier et al. 1988, Armendáriz-Toledano and Zúñiga 2017), are useful for discriminating males of D. approximatus from D. parallelocollis. The presence of prominent frontal tubercles in males is not exclusive of D. approximatus, because this character is shared with other members of the D. frontalis complex species (Víctor and Zúñiga 2016, Armendáriz-Toledano and Zúñiga 2017). However, the tubercles and median groove upper level of eyes to epistomal area are useful for distinguishing D. approximatus from D. parallelocollis.

Whereas the amplitude of epistomal process (WEP) and the elevation of margins (SEP) have been recognized and proposed as diagnostic characters to identify D. parallelocollis (Wood 1982), our results show that only the elevation of margins is useful for differentiating these species , because the WEP overlaps frequently between specimens of both species.

Other characters that were considered nondiagnostic, such as OSB, SPA, PV, and VCD, showed significant differences between species; however, excluding OSB, they are not recommended for the discrimination of these species, because character states overlap between them. OSB can be partially used to identify specimens of D. parallelocollis, because about 43% specimens analyzed sensorial bands intersect at some point on the club (Supp Table 1 [online only], Fig. 7). This is a novel character absent in other species of the genus (Payne et al. 1973, Dickens and Payne 1978, Chen et al. 2010, López et al. 2014, Armendáriz-Toledano et al. 2017).

These bark beetles also show marked differences in 12 quantitative continuous characters. Dendroctonus approximatus has WAC, LAC, HW, EW, WEP, and SRL, SRW, and wider female genitalia (NW) than D. parallelocollis has. Despite this, the latter species is larger in body size than D. approximatus (PL, EL). These quantitative measures of almost all these characters have been successfully used in the identification of other Dendroctonus species, such as D. ponderosae Hopkins and D. jeffreyi Hopkins (Lanier and Wood 1968), D. mesoamericanus Armendáriz-Toledano and Sullivan and D. frontalis Zimmermann (Armendáriz-Toledano et al. 2014a), D. mexicanus Hopkins and D. vitei Wood (Armendáriz-Toledano et al. 2017), and other members of D. frontalis complex (Lanier et al. 1988).

Multivariate Analysis

Multivariate analyses of the quantitative continuous and discrete characters (PCoA) show that females, males, and females–males of these bark beetles present strongly differentiated phenotypes (Figs. 19–21). These analyses also show that the males and females of D. approximatus display much greater phenotypic variation than the sexes of D. parallelocollis (Figs. 19–21), which suggest phenotypic limitations operate less severely in D. approximatus than those present in D. parallelocollis. One of these factors may be the presence of sibling or phylogenetically close species in the same habitat. For instance, it is very common to find D. parallelocollis and D. valens in a Mexican forest in the same tree, but it not so for D. approximatus and other species of the D. frontalis complex.

Geometric Morphometrics

Quantitative analyses of the shape of the antennal club, pronotum, spermatheca, and seminal rod have not been performed previously in these bark beetles. However, the previous geometric morphometric analyses have shown that both antennal club (e.g., D. valens vs D. rhizophagus) and seminal rod shape (e.g., D. frontalis complex species) differed significantly between sibling species despite their wide range distribution (Armendáriz-Toledano 2014b, López et al. 2014, Valerio-Mendoza et al. 2017), which supports the utility of these characters for identification.

Our findings of the geometric morphometric analysis indicate that the shape of the structures differ among species. The lack of overlap in the shape configurations of the antennal club (Fig. 22), spermatheca (Fig. 24), and seminal rod (Fig. 25) of both species confirms that they constitute a robust diagnostic character useful in their identification. The shape pattern of D. approximatus’ antennal club is quadrangular with sensorial bands almost equally distributed, whereas the D. parallelocollis’ antennal club dorso-ventrally elongates with sensorial bands very closely spaced. The spermatheca shape in D. approximatus is wide and with a pronounced curvature between nodulus and cornu, nodulus and cornu oblique, and length of the cornu equal or more than a half of nodulus whereas, D. parallelocollis’ spermatheca is narrow with a poorly developed curvature between nodulus and cornu, the nodulus and cornu perpendicular, and the length of the cornu shorter than of nodulus, approximately a third of nodulus. Lastly, the seminal rod of D. approximatus is slightly curved or straight in its distal-ventral edge with a short seminal valve, whereas seminal rod D. parallelocollis is sinuous in its distal-ventral edge with a long prolongation of the seminal valve.

With respect to the pronotum, our findings show that the shape variation is complex between the sexes and the species. The scatterplot between RW1 vs RW2 do not segregate discrete groups based on their shape configurations according to their sex and species due to high intraspecific variation (Fig. 23). However, since the first five relative warps are included in this analysis, the species and sexes are different statistically and segregated as discrete groups (Fig. 23). Based on the inclusion of more deformation components, our results indicate that D. parallelocollis’ males have the basal lateral margins of pronotum slightly curved to almost parallel, whereas the margins in females are parallel from the median half to the anterior region and the anterior third is slightly narrower than posterior third (Fig. 26). Dendroctonus approximatus’ males have shapes similar to those of both males and females of D. parallelocollis and the pronotum of females of that species is shorter than that of D. parallelocollis due to the presence of mycangium (Fig. 27), an attribute absent in latter species. The etymology of D. parallelocollis, as have been referred by some authors (Chapuis 1869; Wood 1963, 1982), assumes that this species has lateral margins less curved and almost parallel; however, our results indicate that this assertion is only partly true, because only the females have this pattern (Fig. 26). In addition, since D. approximatus’ males do not have mycangium (Dietz 1890; Hopkins 1909; Wood 1963, 1982), they can be confused with males and females of D. parallelocollis, because the variation in shape of this structure overlaps in these species.

Figs. 26–27.

Variation in profiles of dorsal pronotum from different localities. (26) Dendroctonus parallelocollis. (27) Dendroctonus approximatus. Acronyms for localities below images are from Table 1.

Geographical Variation

Among specimens of D. parallelocollis from different locations, we found some differences in shape of antenna, pronotum, and seminal rod. In particular, the antenna in southern locations (Oaxaca and Central America) show pronounced variation in the sensorial bands, because specimens from these regions present a third sensorial band in discontinuous form and in some cases, these intersect at some point. Specimens from the northern locations (states of Coahuila and Durango) do not display these features. We also found marked differences in the pronotum because in specimens from southern locations, the lateral edges are more arched, and in some cases, more similar to D. approximatus (Fig. 26). Lastly, we found that the seminal rod from the specimens from the northern locations, has the distal-ventral edge of this structure almost straight as compared with those from the central locations (Jalisco, Michoacán, Queretaro, Tlaxcala, Veracruz, and México states). These present a seminal rod body elongate with a bigger seminal valve and with southern localities and Guatemala specimens, which have the distal-ventral edge with two maximum points of curvature (Fig. 28).

Figs. 28–29.

Variations in lateral view of seminal rod from different localities. Acronyms for localities below images are from Table 1. (28) Dendroctonus parallelocollis. (29) Dendroctonus approximatus.

These morphological differences among geographical regions suggest the presence of a cryptic species within D. parallelocollis. In fact, a previous phylogenetic analysis found that the phylogenetic position of this species is ambiguous (Víctor and Zúñiga 2016) suggesting that other species might have not been discovered. Furthermore, we have observed behavioral differences in the colonization pattern of D. paralellocollis in the northern and southern populations. Specimens from different localities in the Trans-Mexican Volcanic Belt arrive frequently to galleries built previously by D. valens but the specimens from other localities in Oaxaca and Guatemala colonize both old and young trees. In the latter ones, D. parallelocollis is able to build galleries in the roots. Taxonomic integrative analyses need to be conducted and a more intensive collective effort must be undertaken to evaluate the existence of possible cryptic species in D. parallelocollis.

In brief, this study defines a set of old and new features that facilitate the discrimination of D. approximatus and D. parallelocollis. These features are consistent and robust thorough of the distribution range of these species and provide a more precise framework of intra- and interspecific morphological variation. In the case of D. parallelocollis, the inclusion of molecular markers should establish the basis to test the presence or not of a new taxon within this species.