Geometric morphometrics and ecological niche modelling for delimitation of Triatoma pallidipennis (Hemiptera: Reduviidae: Triatominae) haplogroups

Abstract

A recent phylogenetic analysis of Triatoma pallidipennis, an important Chagas disease vector in Mexico, based on molecular markers, revealed five monophyletic haplogroups with validity as cryptic species. Here, we compare T. pallidipennis haplogroups using head and pronotum features, environmental characteristics of their habitats, and ecological niche modeling. To analyze variation in shape, images of the head and pronotum of the specimens were obtained and analyzed using methods based on landmarks and semi-landmarks. Ecological niche models were obtained from occurrence data, as well as a set of bioclimatic variables that characterized the environmental niche of each analyzed haplogroup. Deformation grids for head showed a slight displacement towards posterior region of pre-ocular landmarks. Greatest change in head shape was observed with strong displacement towards anterior region of antenniferous tubercle. Procrustes ANOVA and pairwise comparisons showed differences in mean head shape in almost all haplogroups. However, pairwise comparisons of mean pronotum shape only showed differences among three haplogroups. Correct classification of all haplogroups was not possible using discriminant analysis. Important differences were found among the environmental niches of the analyzed haplogroups. Ecological niche models of each haplogroup did not predict the climatic suitability areas of the other haplogroups, revealing differences in environmental conditions. Significant differences were found between at least two haplogroups, demonstrating distinct environmental preferences among them. Our results show how the analysis of morphometric variation and the characterization of the environmental conditions that define the climatic niche can be used to improve the delimitation of T. pallidipennis haplogroups that constitute cryptic species.

Graphical abstract

Highlights

• •Triatoma pallidipennis haplogroups were compared using head and pronotum features and ecological niche modeling.
• •Differences were found in shape and environmental characteristics of the distribution areas of analyzed haplogroups.
• •Features of the head had a higher taxonomic value for differentiation of the haplogroups than those of the pronotum.
• •Some haplogroups showed ecological segregation, based on the climatic characteristics of their distribution areas.
• •Analysis of morphometric variation and environmental conditions can improve delimitation of T. pallidipennis haplogroups.

1. Introduction

The correct estimation and characterization of biodiversity are key elements for the conservation and use of natural entities. Since “species” is the operative unit in biology, defining species limits is a pillar of the quantification and management of biodiversity. For this reason, its conceptualization has been the subject of considerable intellectual effort (Sites & Marshall, 2004; de Queiroz, 2007). Species delimitation is a crucial step in identifying and distinguishing different species, with significant implications in the context of public health (de Paiva et al., 2022). Accurate identification of species can assist in mitigating the impact of various health issues caused by different vector species, leading to more effective control strategies (Cruz & Arellano, 2022). Proper species delimitation can also aid in identifying vector-borne diseases more accurately, making it an essential tool for understanding the ecology and behavior of vectors and mitigating the impact of diseases on public health (Justi et al., 2018).

Historically, diagnostic phenotypic characters have been used to carry out taxonomic delimitations, classifying distinct populations into species and subspecies, known as the “traditional taxonomic approach” (Matos-Maraví et al., 2019). In entomology, specifically, much attention has been paid to the morphology of various body parts to identify, name, and classify organisms. Thus, traditional taxonomy and systematics have used specific external morphological characteristics as diagnostic characters for the identification of taxa (Tatsuta et al., 2018). However, some entities are recognized as distinct species under other criteria (mainly phylogenetic) but are not (at least superficially) distinguishable based on external morphology and have therefore been recognized as a single species; these are known as cryptic species (Bickford et al., 2007) and are a case in which the traditional taxonomic approach fails to delimit taxa that should be considered separate species. Failure to correctly detect and delimit species results in an underestimation of true levels of biodiversity (Lefébure et al., 2006; Padial et al., 2010). Cryptic species have been shown to occur in many animal groups (e.g. Bickford et al., 2007; Pfenninger & Schwenk, 2007; Cháves-González et al., 2022).

Recently, the integrative taxonomy approach has been widely proposed in systematics studies (Dayrat, 2005; Sangster, 2018). Lines of evidence such as geometric morphometrics and ecological analyses have been useful to discriminate between genetically well-differentiated species (Martínez-Borrego et al., 2022). These integrative approaches are particularly useful in cases where traditional methods provide insufficient resolution (MacGuigan et al., 2017; Freitas et al., 2018).

Leaché et al. (2009) argued that the use of morphological and ecological data, in addition to molecular data, allows better discrimination among species. Geometric morphometrics analysis is a robust tool to highlight interspecific variation and corroborate the phylogenetic relationships within animal groups (e.g. Bogdanowicz et al., 2005; Pavan & Marroig, 2016). In addition, ecological niche studies are being increasingly used for these same purposes (e.g. Rissler & Apodaca, 2007; Rivera et al., 2018; Zhao et al., 2019; Masonick & Weirauch, 2020), as well as for inferences related to evolutionary questions of both historical distributions and speciation processes (Graham et al., 2004).

The genus Triatoma (Hemiptera: Reduviidae) is one of the most epidemiologically important insect groups since they are the main vectors of Chagas disease on the American continent. In addition to Triatoma spp. relevance as vectors, the genus is highly scientifically valuable because it contains many species with high rates of speciation, making the genus an interesting model to evaluate recent speciation processes. The cryptic speciation phenomenon within Triatoma seems to be a relatively common process, detected mainly by molecular techniques (Bargues et al., 2008; Ibarra-Cerdeña et al., 2014; Justi et al., 2014; Pech-May et al., 2019). However, geometric morphometry has been repeatedly used as a tool to delimit species within the subfamily Triatominae (Gurgel-Gonçalves et al., 2011; Dujardin, 2011; Vendrami et al., 2017; Cruz et al., 2020).

Ecological segregation could also play an important role in the process of lineage divergence in Triatoma (Gurgel-Gonçalves et al., 2011). If two subsets of individuals from the same species can adapt to different ecological niches, they may undergo divergent selection and develop independent evolutionary paths, which can lead to speciation (Schluter, 2001). Consequently, many sister taxa exploit contrasting niches. This makes quantitative approaches such as ecological niche modeling (ENM) useful to investigate recent speciation events that are likely to have been driven by ecological divergence (Warren et al., 2008).

One of the most epidemiologically important and most widely distributed endemic triatomines in Mexico is Triatoma pallidipennis Stal, 1872 (Lent & Wygodzinsky, 1979; Martínez-Ibarra & Katthain-Duchateau, 1999; Ramsey et al., 2015). Lent & Wygodzinsky (1979), after examining several specimens from different locations, concluded that this species does not present significant morphological variations among its populations. As a result of this wide geographical distribution, a large niche breadth has been reported for the species (Ramsey et al., 2015), which can be assumed to represent high ecological plasticity. However, a recent phylogenetic analysis of T. pallidipennis using mathematical methods for species delimitation based on the mitochondrial nad4 gene from 73 specimens and the ribosomal internal transcribed spacer (ITS2) from 33 specimens from different locations throughout its geographical distribution revealed five monophyletic haplogroups with potential validity as cryptic species (Cruz & Arellano, 2022).

It is therefore desirable to broaden the understanding of the genetic distinction between T. pallidipennis haplogroups by following an integrative taxonomy approach using alternative data sources such as geometric morphometrics and ecological attributes. The goal of this study is to compare T. pallidipennis haplogroups using head and pronotum features, environmental characteristics of their habitats, and ecological niche modeling, based on the premise that they constitute genetically well-differentiated species.

2. Materials and methods

2.1. Haplogroups analyzed

Morphometric and ecological analysis was performed on four and three of the five haplogroups (putative cryptic species), respectively, that were detected in the phylogenetic analysis of T. pallidipennis by Cruz & Arellano (2022) (Supplementary file 1: Figure S1; Tables S1 and S2). The putative cryptic species analyzed were: Haplogroup I (specimens from Morelos, Oaxaca and eastern Puebla); Haplogroup II (specimens from southern Morelos, southwestern Mexico State and eastern Guerrero); Haplogroup III (specimens from Mexico State); and Haplogroup V (specimens from Colima and Jalisco). The individuals of T. pallidipennis used in the phylogenetic analysis were obtained through direct donations from the National Institute of Diagnosis and Reference (INDRE), the Ministry of Health of Mexico State, the State Laboratory of Medical Entomology (LEEM) of Mexico State, as well as direct collections made by LEEM personnel in the state of Guerrero (see Cruz & Arellano, 2022 for details).

2.2. Geometric morphometrics data

To analyze variation in shape, images of the head and pronotum (Fig. 1A) of the specimens were obtained and analyzed using methods based on landmarks and semi-landmarks (Bookstein, 1991; Mitteroecker & Gunz, 2009; Zelditch et al., 2012; Adams & Otárola-Castillo, 2013). The relevance of the head as a structure for morphometric analysis has been repeatedly evaluated in the genus Triatoma (Gurgel-Gonçalves et al., 2011; Hernández et al., 2013; Oliveira et al., 2017; Nattero et al., 2017). The pronotum, on the other hand, has virtually not been explored in morphometric terms; in the context of species delimitation, the only known work that studies its potential to discriminate between cryptic species of Triatoma is the study by Nattero et al. (2017). Table 1 shows the total number of specimens, as well as the number of females and males obtained for each haplogroup analyzed. These included the specimens used in the phylogenetic analysis by Cruz & Arellano (2022) and other specimens from the same localities. The final total sample size was 113 individuals.

Fig. 1.

Location of the structures analyzed (A) and the landmarks and semilandmarks used for the description of the shape of the head (B) and pronotum (C) in haplogroups of Triatoma pallidipennis. Blue dots represent landmarks. Red dots represent semi-landmarks.

Table 1.

Number of samples per haplogroup of Triatoma pallidipennis used for morphometric analysis.

Haplogroup	Head			Pronotum
	♂	♀	Total	♂	♀	Total
I	8	12	20	8	13	21
II	8	19	27	9	19	28
III	22	13	35	23	16	39
V	12	13	25	13	12	25

Note: Haplogroup I: specimens from Morelos, Oaxaca, and eastern Puebla; Haplogroup II: specimens from southern Morelos, southwestern Mexico State and eastern Guerrero; Haplogroup III: specimens from Mexico State; Haplogroup V: specimens from Colima and Jalisco.

All individuals were photographed under the same magnification and exposure settings and with a millimeter scale. To describe the shape of the head, we used 14 landmarks (Fig. 1B) representing a combination of those suggested by Gurgel-Gonçalves et al. (2011) and Oliveira et al. (2017). Six landmarks and 38 semi-landmarks were used for the pronotum (Fig. 1C) to describe the right pronotal contour of the specimens. As a result of digitization in TPS software (Rohlf, 2015), a matrix of coordinates (x; y) was obtained for each set of morphological points for each image processed for both structures. All landmarks used are listed in Table 2.

Table 2.

Definition of the landmarks used for the morphometric characterization of the head and pronotum in haplogroups of Triatoma pallidipennis. The numbers correspond to the position of the landmarks and semi-landmarks, as shown in Fig. 1.

Structure	Landmarks and semi-landmarks
Head	Intersection point between head and neck on the left side (1)
	Intersection between head and neck on the right side (2)
	Post-ocular point on left side (3)
	Post-ocular point on right side (4)
	Pre-ocular point on left side (5)
	Pre-ocular point on right side (6)
	External region of the right antenniferous tubercle (7)
	Insertion of the right antenniferous tubercle (8)
	Internal region of the right antenniferous tubercle (9)
	Insertion of the left antenniferous tubercle (10)
	Internal region of the left antenniferous tubercle (11)
	External region of the left antenniferous tubercle (12)
	Left intersection between the gena and antecapeum (13)
	Right intersection between the gena and antecapeum (14)
Pronotum	Middle insertion of the upper pronotal border (1)
	Inner base of the right spine (2)
	Upper end of the right spine (3)
	External base of the right spine (4)
	External contour of pronotum (5–42)
	Middle insertion of the lower pronotal border (43)
	Intersection point between pronotal lobes (44)

The morphometric analyses were performed using the package geomorph v. 3.3.1 (Adams & Otárola-Castillo, 2013) of the R library (R Core Team, 2022). Using the “define.sliders” function, we defined which points constitute the semi-landmarks that characterized the right edge of the pronotum. Generalized Procrustes analysis (GPA) (Rohlf & Slice, 1990) was performed to superimpose the obtained coordinates and extract the shape variables. During this procedure, all samples are translated about a common origin to remove the effect of position, scaled through centroid sizes, and finally rotated (using the least-squares criterion) until the coordinates of the corresponding points align as closely as possible, to minimize the effect of differences in orientation. In the case of the pronotum configuration, semi-landmarks were allowed to slide along their tangent vectors until reaching the minimum point of bending energy (Bookstein, 1997; Gunz et al., 2005; Gunz & Mitteroecker, 2013). As a result of the GPA, the Procrustes coordinates and the Procrustes distances were obtained, which were used as shape variables. In addition, the centroid size (CS) was estimated to obtain a shape-independent measure of size.

2.3. Statistical data processing

Principal components analyses (PCA) of the geometric configurations of the head and pronotum were performed. From these analyses, the scores of the first 10 principal components (PC) were extracted, which were used for the rest of the statistical analyses as shape variables. The morphological variance concerning the mean shape was visualized and described in the positive and negative directions, on the first two PC using deformation grids.

As part of the exploratory analysis, the presence or absence of sexual dimorphism was determined by considering the shape variables and the CS of both structures (head and pronotum). To detect whether there were significant differences between sexes, a Procrustes ANOVA (Goodall, 1991; Anderson, 2001) was performed for head and pronotum shape using the function “ProcD.lm”, and a factorial design with shape as the dependent variable, sex as the main factor, and CS as a covariate. Significant differences (P ​≤ ​0.05) in head and pronotum size between sexes were tested using the function “lm.rrpp” in the R package RRPP (Collyer & Adams, 2018). We found no significant sexual dimorphism in the size or shape of any structure analyzed; therefore, we pooled males and females in our analyses of interspecific differentiation. The allometric component (possible effect of size on shape variation) was analyzed using a Procrustes ANOVA and the factorial design indicated above. However, this allometric effect was relatively low for both the head and the pronotum, with values of 5.7% and 2.0%, respectively.

Differences in size and shape between haplogroups were evaluated for the head and pronotum separately, using a Procrustes ANOVA. A factorial design was used with shape as the dependent variable, haplogroups as the main factor, and CS as a covariate. The mean head/pronotum shape differences between species were quantified (Procrustes distances) and tested for statistical significance using a pairwise permutation test (P ​≤ ​0.05; resampling ​= ​1000; function “permudist”) in the R package Morpho 2.4 (Schlager, 2017).

Based on the scores of the first 10 ​PCs, a linear discriminant function analysis (LDFA) was performed with the mass library (Venables & Ripley, 2002) to estimate the percentage of correct classification. Finally, an analysis of canonical variables was performed and, as with the PCs, the variation of the mean shape in the positive and negative directions along each canonical axis was visualized using deformation grids.

2.4. Ecological niche data

The occurrence data used for the ecological analyses were the coordinates of the localities where the individuals of each haplogroup were collected. We obtained additional data to reach the minimum number of locations (30) from the database published by Ramsey et al. (2015), which is freely available (https://doi.org/10.5061/dryad.rq120), and the international database GIBIF (Global Information Biodiversity Facility, www.gbif.org). We used only occurrence data collected after 1950, and data for specimens with inconsistent information were not included in the analysis. Bioclimatic variables from the Worldclim 2.0 database (Fick & Hijmans, 2017) were used as descriptive variables (temperature and rainfall), with a spatial resolution of approximately 1 ​km².

2.5. Data processing and generation of ecological niche models

To detect differences between the ecological niches of the haplogroups previously defined by the phylogenetic analysis, ENM was performed for each one of them in the MaxEnt program ver.3.4.1 (Phillips et al., 2006, 2017). The subset of bioclimatic variables was composed of those that were not strongly correlated (r ​< ​0.8, following Moo-Llanes et al., 2020; Tarquino-Carbonell et al., 2020). Based on this previous analysis, the variables selected to obtain the ENMs were: Bio2, mean diurnal range; Bio3, isothermality; Bio4, temperature seasonality; Bio13, precipitation of wettest month; Bio15, precipitation seasonality; and Bio16, precipitation of wettest quarter.

Since the performance of ENM is sensitive to the specifications of the modeling process (Araujo & Guisan, 2006; Araújo & Peterson, 2012; Radosavljevic & Anderson, 2014), before the generation of the ENM, we evaluated the program settings for each haplogroup. We used the Wallace package (Kass et al., 2018) in R (R Core Team, 2022) to clean occurrence data, reduce spatial autocorrelation, and select the calibration area and the background samples. Spatial filtering of the occurrence data was carried out for each haplogroup to be modeled using the spThing package in R (Aiello-Lammens et al., 2015). The minimum distance for filtering was 5 ​km. Calibration areas were determined by a convex minimum polygon of the presence points resulting from spatial filtering. From these, 10,000 background points were generated to obtain the ENM.

To define an appropriate level of complexity, we evaluated MaxEnt configurations to compare the performances of different test models. These models were created using the checkerboard method to partition the training and test data. In addition, 12 values were used for the regularization multiplier (0.5–6.0, with steps of 0.5) and six configurations of feature classes: L, LQ, H, LQH, and LQHP (where L ​= ​linear, Q ​= ​quadratic, H ​= ​“hinge” and P ​= ​product). These analyses were performed using the ENMeval package in R (Muscarella et al., 2014). The best configurations were selected based on the lowest values of the corrected Akaike information criterion (AICc).

The final models were calibrated with the full set of filtered records and the selected set of variables. We performed 50 replicates per modeled haplogroup using the best configurations obtained during the evaluation process. In addition, they were projected onto the entire Mexican territory to estimate new areas of environmental suitability for each of the lineages. Occurrence data, calibration, and parameter results are shown in Supplementary file 1: Tables S3 and S4.

2.6. Ecological segregation between phylogenetic lineages

To estimate whether there is ecological segregation between haplogroups, we obtained the ellipsoids and centroids that characterize the environmental space of each one (Van Aelst & Rousseeuw, 2009). We did this using a principal components analysis (PCA) with 15 bioclimatic variables (excluding Bio8, Bio9, Bio18 and Bio19), to characterize the environmental space in which each haplogroup is distributed. From this analysis, the fundamental niche of T. pallidipennis was obtained. In this fundamental niche, we then represented the realized niche of the populations composing each haplogroup using their presence data. All analyses were performed in the NicheA program (Qiao et al., 2016).

From the PCA, the scores of the components with the highest contribution to the total variance were used to compare the environmental niche of each haplogroup using Kruskal-Wallis tests, since the data did not meet the assumptions of normality and homogeneity of variance. This analysis was performed in the program Statistica v. 8.

3. Results

3.1. Geometric morphometrics

The ordination based on the exploratory analysis of PCA was not consistent with the separation between haplogroups for any of the structures analyzed (Fig. 2). The first two PCs explained approximately 50% and 60% of the total cumulative variance for the head and pronotum, respectively. In the deformation grids for the head in the first two PCs, a slight displacement towards the posterior region of the head of the landmarks characterizing the pre-ocular region at the negative end of PC1 was observed. This shift was more evident at the negative end of PC2. For the positive ends of both axes, the greatest change in shape was observed in PC2, which showed a strong displacement towards the anterior region of the head of the landmarks that characterize the antenniferous tubercle.

Fig. 2.

Principal components analysis (PCA) for the shape of the head (A) and pronotum (B) among haplogroups of Triatoma pallidipennis. Deformation grids show the minimum and maximum variation in shape along the axis. Abbreviations: PC, principal component; H I, Haplogroup I (specimens from Morelos, Oaxaca, and eastern Puebla); H II, Haplogroup II (specimens from southern Morelos, southwestern Mexico State and eastern Guerrero); H III, Haplogroup III (specimens from Mexico State); H V, Haplogroup V (specimens from Colima and Jalisco).

The changes in the shape of the pronotum were most evident at the negative extremes of the PC, associated with the right lobe. At the negative end of PC1, there was a marked contraction of the points that characterize this region. This is different from the negative end of PC2, in which there was a displacement of these points towards the anterior region of the pronotum (Fig. 2B).

3.2. Differences in shapes and centroid size between haplogroups

The Procrustes ANOVA (Table 3) and pairwise comparisons showed differences in mean head shape among all haplogroups (Fig. 3A–D) except between haplogroups in the contrasts H I vs H II and H I vs H III. Pairwise comparisons of mean pronotum shape only showed differences between haplogroups in the contrasts H II vs H V and H II vs H III (Fig. 3E and F). The greatest differences in the shape of the head were mainly from the region of the antenniferous tubercle, which was smaller in haplogroup III than in haplogroups I and II (Fig. 3A and B). However, concerning haplogroup III, this region was larger (Fig. 3C). The ocular region and the base of the head was broader in haplogroup III than in haplogroups I and II but smaller than in haplogroup V. Finally, the most similar heads in shape were the middle forms of haplogroups II and V; the greatest differences between them were in the insertion of the antenniferous tubercle and the post-ocular region.

Table 3.

Differences in the shape of the head and pronotum between haplogroups of Triatoma pallidipennis using analysis of variance (ANOVA).

Variable	Statistic	SS	MS	R2	F	Z	P-value
Head	Haplogroups	0.0198	0.0066	0.0930	3.5215	5.0964	0.001∗
	Residuals	0.1934	0.0018	0.9069
	Total	0.2132
Pronotum	Haplogroups	0.0129	0.0043	0.0566	2.1825	2.7106	0.01∗
	Residuals	0.2153	0.0019	0.9433
	Total	0.2283

Abbreviations: SS, sum of squares; MS, mean squares.

Note: Asterisks indicate significant differences (P ​< ​0.05).

Fig. 3.

Differences in mean shape (Procrustes distances) between haplogroups of Triatoma pallidipennis. Differences were magnified by a factor of 3. A-D Head. E, F Pronotum. Haplogroup I: specimens from Morelos, Oaxaca, and eastern Puebla; Haplogroup II: specimens from southern Morelos, southwestern Mexico State and eastern Guerrero; Haplogroup III: specimens from Mexico State; Haplogroup V: specimens from Colima and Jalisco. Abbreviation: dP, Procrustes distance between the mean shape of each pairwise comparison. Asterisks indicate significant differences (P ​≤ ​0.05), based on pairwise permutation tests.

For the pronotum, there were differences in the shape of the right dorsal spine, in the middle insertion of the upper pronotal edge, in the insertion point of the pronotal lobes, and the shape of the right lobe, for the two pairwise comparisons made (Fig. 3E and F). The main differences in the mean shapes of the pronotum between haplogroups II and III were in the right lobe, the humeral angle, and the lower margin of the pronotum (Fig. 3E). The right lobe in haplogroup V was wider than in haplogroup II. The humeral angle was markedly truncated relative to the humeral angle of haplogroup II, which had a more pointed shape. Significant variation in the lower margin of the pronotum was also observed between these haplogroups. The differences in the mean shape of the posterior lobe of the pronotum between haplogroups II and V were not as marked as in the previous comparison. However, haplogroup V presented a wider right lobe. As for the lower margin, the variation was much smaller (Fig. 3F). Although there were differences in shape between haplogroups, there were no differences in the size of either the head or the pronotum (Fig. 4).

Fig. 4.

Comparison of centroid size for the head (A) and pronotum (B) between haplogroups of Triatoma pallidipennis. Haplogroup I: specimens from Morelos, Oaxaca, and eastern Puebla; Haplogroup II: specimens from southern Morelos, southwestern Mexico State and eastern Guerrero; Haplogroup III: specimens from Mexico State; Haplogroup V: specimens from Colima and Jalisco.

3.3. Correct classification of haplogroups

It was not possible to correctly classify any of the haplogroups analyzed using the LDFA. The global correct classification percentages obtained for both structures were similar, with 49.5% of correct classification for the head and 50.4% for the pronotum. However, the percentage of specific correct classification in some haplogroups was relatively high. For the head, the best classified haplogroups were III and V, while for the pronotum, the best classified were haplogroups II and III (Table 4).

Table 4.

Results of the linear discriminant function analysis (LDFA) to classify specimens from haplogroups of Triatoma pallidipennis using shape information for the head and pronotum.

	Haplogroup	I	II	V	III	Total	% CC
Head	I	3	8	4	5	20	15.0
	II	7	9	1	10	27	33.3
	III	2	6	3	24	35	68.5
	V	3	2	17	3	25	68.0
Pronotum	I	8	6	4	3	21	38.1
	II	1	18	5	4	28	64.3
	III	2	8	7	22	39	56.4
	V	8	2	9	6	25	36.0

Note: Haplogroup I: specimens from Morelos, Oaxaca, and eastern Puebla; Haplogroup II: specimens from southern Morelos, southwestern Mexico State and eastern Guerrero; Haplogroup III: specimens from Mexico State; Haplogroup V: specimens from Colima and Jalisco.

Abbreviation: % CC, correct classification percentage.

In the ordination graph of the analysis of canonical variables for the shape of the head considering 10 ​PCs, haplogroup V was the most separated from the rest (Fig. 5A). The rest of the haplogroups analyzed showed high overlap. The deformation grids obtained for each canonical axis showed variations in head shape, mostly towards the negative end of the axis of canonical axis 1 (CV1). These variations were mainly due to contractions in the pre-ocular regions and the antenniferous tubercles. The variations in shape were subtler along the canonical axis 2 (CV2), given by contractions in the ocular region of the head (Fig. 5A).

Fig. 5.

Ordination plot of the first two axes of the analysis of canonical variables for the shape of the head (A) and pronotum (B), among haplogroups of Triatoma pallidipennis. Deformation grids represent the minimum and maximum variation in shape along the axis. Abbreviations: H I, Haplogroup I (specimens from Morelos, Oaxaca, and eastern Puebla); H II, Haplogroup II (specimens from southern Morelos, southwestern Mexico State and eastern Guerrero); H III, Haplogroup III (specimens from Mexico State); H V, Haplogroup V (specimens from Colima and Jalisco).

Regarding the shape of the pronotum, the haplogroups overlapped almost completely in the analysis of canonical variables (Fig. 5B). The shape did not present marked variations, as evidenced by the deformation grids obtained for both canonical axes. The slight variations observed were mainly in the shape of the right lobe and the humeral angle.

3.4. Ecological niche modeling

When obtaining the ecological niche models, the climatic suitability areas (CSA) were recovered for each evaluated haplogroup (Fig. 6). Haplogroup V had a larger potential distribution (Fig. 6C) than haplogroups I (Fig. 6A) and III (Fig. 6B). When projecting the conditions of these regions onto all of Mexico, the ecological niche of each haplogroup did not predict the CSA of the rest in the same way. Haplogroup V predicted a large part of the CSA obtained for haplogroup I, but only a small part of the known distribution area for haplogroup III (Fig. 6C). The models for haplogroup I and haplogroup III very poorly predicted the CSA of haplogroup V (Fig. 6A and B). In turn, the models of these last two haplogroups predicted very few CSA of each other.

Fig. 6.

Climatic suitability areas (A–C) and three-dimensional models in the environmental space (D) of the presence records for three phylogenetic haplogroups of Triatoma pallidipennis (Hemiptera: Reduviidae). A Haplogroup I. B Haplogroup III. C Haplogroup V. D Blue ellipsoids correspond to the populations of Haplogroup I; black ellipsoids correspond to the populations of Haplogroup III; red ellipsoids correspond to the populations of Haplogroup V.

The ellipsoids that describe the environmental space of the populations for each haplogroup showed that haplogroup V has a broader environmental niche than the rest of the analyzed haplogroups (Fig. 6D). In addition, part of the environmental niche of haplogroup I and II overlapped with those of haplogroup V. The greatest differentiation between ecological niches was observed between haplogroup I and II, where only a small part of the environmental conditions of the distribution of both overlaps.

In the principal components analysis, the first two components explained more than 95% of the variance of the environmental data. There were significant differences between at least two haplogroups along the first two components (Fig. 7).

Fig. 7.

Statistical differences in the scores of the first two principal components among haplogroups of Triatoma pallidipennis (Hemiptera: Reduviidae), obtained by Kruskal-Wallis tests. A Differences between haplogroups according to principal component 1 (PC1) score. B Differences between haplogroups according to principal component 2 (PC2) score. Histograms, data density polygons, and data points (data used for statistical analysis) are plotted. Abbreviations: H I, Haplogroup I (specimens from Morelos, Oaxaca, and eastern Puebla); H III, Haplogroup III (specimens from Mexico State); H V, Haplogroup V (specimens from Colima and Jalisco).

4. Discussion

4.1. Geometric morphometry

Species delimitation is frequently a controversial and complicated task, especially when the focal organisms are considered cryptic taxa (Tatsuta et al., 2018). In the genus Triatoma, the presence of cryptic species has been extensively addressed, mainly using molecular methods (Bargues et al., 2008; Ibarra-Cerdeña et al., 2014; Justi et al., 2014; Pech-May et al., 2019). Geometric morphometrics has proven to be a useful tool in the context of species delimitation and taxonomic studies in triatomines (Dujardin et al., 1999; Matias et al., 2001; Jaramillo et al., 2002; Villegas et al., 2002; Gumiel et al., 2003; Lehmann et al., 2005; Vargas et al., 2006; Feliciangeli et al., 2007; Nattero et al., 2017; Cruz et al., 2020). The use of morphological characters derived from novel techniques such as geometric morphometrics, as a reflection of genetic variability, is an efficient way to estimate the differentiation between phylogenetically differentiated groups (Patterson et al., 2001).

Our results show significant differences in shape (but not size, as characterized by the centroid size, CS) among haplogroups. Previous studies have shown differences in head size between species (Gurgel-Gonçalves et al., 2011; Oropeza et al., 2017). Patterson (2007) hypothesized that differences in the head may have an evolutionary cause related to feeding strategies, with different morphologies allowing the ingestion of blood from different specific food sources, as well as different growth patterns. If we consider the hypothesis of Patterson (2007), a possible explanation for the fact that there is no difference in the size of the head between haplogroups is that the recent speciation processes that have occurred in the populations of T. pallidipennis may not have led to a segregation of feeding niches. For these reasons, the general size pattern of T. pallidipennis may be maintained in each of the haplogroups analyzed. However, the observed variation in the shape of the head among haplogroups could be an indication that modifications in the structure may be occurring based on the microhabitat characteristics and by the hosts that these haplogroups use as a food source, as analyzed by Abad-Franch & Gurgel-Gonçalves (2021). However, there is very little information available on this topic, so future studies could clarify whether there are differences in the microhabitats and food sources of each of the analyzed haplogroups. This information could complement the results obtained in this study.

Although significant differences were found in the shape (Procrustes distances) of the head among almost all the haplogroups, there was relatively low discrimination among haplogroups using the LDFA. Thus, the differences found in the shape of the head and the pronotum should be considered an indication that the haplogroups of T. pallidipennis are following different evolutionary divergences in which geographical isolation seems to be having an important effect. This phenomenon has been described for other triatomines, as is the case of the Triatoma dimidiata species complex, in which morphological differences in head shape were a phenotypic consequence of isolation among their populations (Bustamante et al., 2004).

The fact that the best-discriminated haplogroup based on head shape was from Colima and Jalisco (Haplogroup V) may be an indication that the geographical isolation between the populations of T. pallidipennis is acting as a modulator of the expression of differentiated morphological characters, concerning the other haplogroups. Geometric morphometrics is a complement to molecular methods since genetic differentiation very often results in morphometric variations that allow groups from different regions to be distinguished (Bradshaw et al., 2000; Stacey & Fellowes, 2002).

The capacity of geometric morphometrics to discriminate phylogenetically differentiated groups has been demonstrated for other groups of insects (De la Riva et al., 2001). In our previous phylogenetic results (Cruz & Arellano, 2022), haplogroup V showed high values of genetic distances (K2P model) based on the nad4 gene, concerning the other haplogroups analyzed (4.91%, 5.82%, and 6.45% with haplogroups I, II and III, respectively). Therefore, the geometric morphometrics analysis was able to differentiate this evolutionary unit from the others (correct classification of 68.0%). This was also the case of haplogroup III, showing a similar correct classification value (68.5%) to haplogroup V, and like these, had high genetic distances concerning the other haplogroups analyzed (6.61% and 6.76% with haplogroups I and II, respectively). At the same time, the near-total overlap between haplogroups I and II in terms of head shape reflects the smaller genetic distance between these two haplogroups (3.19%) (see Cruz & Arellano, 2022).

Regarding the pronotum, the global values of correct classification were lower than those found for the head. However, haplogroup II was the best discriminated against based on pronotum shape, reaching 64.3% correct classification. Although statistically significant differences were only found in the shape of the pronotum in the paired comparison between haplogroup II and haplogroups III and V, these differences were marked, as can be seen in Fig. 5. Similar results were obtained by Nattero et al. (2017); they analyzed the morphological variability of the T. sordida species subcomplex and obtained the lowest discrimination values between species from the pronotum shape information. These authors only used four landmarks for the morphometric characterization of this structure, which could explain their low discrimination values. The pronotum has been very seldom used in the context of geometric morphometrics (Nattero et al., 2017), and in triatomines, it has mainly been used in traditional morphological descriptions (Lent & Wygodzinsky, 1979).

In this investigation, the greatest differences in pronotum shape were found in the right posterior lobe and the humeral angle, which partially coincides with the results obtained by Cruz et al. (2020) using elliptic Fourier descriptors to discriminate haplogroups of the T. dimidiata complex. Therefore, although we did not find a high percentage of discrimination using the pronotum in the present study, the exploration of this structure constitutes an important step in any protocol of geometric morphometrics in triatomines to achieve a better characterization of its variation in this group of insects. It is possible that by describing the shape of the entire structure, using landmarks and semi-landmarks, as well as other morphometric methods such as Fourier elliptic descriptors, better discrimination percentages could be achieved. Fourier elliptic descriptors have been particularly efficient in discriminating species and haplogroups within the genus Triatoma (Santillán-Guayasamín et al., 2017; Cruz et al., 2020).

4.2. Ecological niche

In this study, environmental constraints for the current distribution of members of three T. pallidipennis haplogroups were identified by applying ecological niche modeling techniques. The distribution range of these haplogroups in Mexico reflects the diverse environmental conditions to which they have probably adapted. They may also reflect physiological differences among them (Zhao et al., 2019).

The ecological niche model for each haplogroup, both in their strict distribution areas and in the projections obtained for the rest of Mexico, showed considerable independence from environmental conditions. However, the prediction of some climatic suitability areas of one haplogroup from the occurrence data of another demonstrated a certain degree of niche conservatism, especially between haplogroups I and V.

Our results have shown that the distribution of each haplogroup analyzed is limited by a set of environmental conditions. This was corroborated by obtaining the environmental space occupied by each hypervolume. Although there is still some overlap in the environmental conditions related to the distribution of the haplogroups, this is not of equal intensity for all. This could be a sign that ecological segregation has indeed played an important role in the differentiation of these monophyletic lineages, which is supported by the significant differences found for at least two of the haplogroups analyzed.

The ellipsoids indicating the environmental space of the populations showed differences in the use of the existing environmental space among the three haplogroups analyzed. This differentiation pattern has been reported for T. dimidiata haplogroups (Gómez-Palacio et al., 2015), so this phenomenon may be more common in triatomines than currently recognized. However, although these results partially corroborate the role of ecological segregation as a speciation phenomenon in at least these two species complexes, for the T. pallidipennis haplogroups it is necessary to evaluate hypotheses of niche conservatism to reach more solid conclusions.

Allopatric divergence has been considered the predominant mode of geographical speciation for most animal taxa (Barraclough et al., 1999; Schluter, 2001; Levin, 2004) and does not account for significant niche divergence between closely related taxa (Peterson et al., 1999; Graham et al., 2004; Kozak & Wiens, 2006; Broennimann et al., 2007; Peterson & Nyári, 2008; Warren et al., 2008). However, the distribution of the populations of the haplogroups analyzed in different biogeographical provinces is an important element to have in mind when considering ecological segregation. According to the classification of biogeographical provinces (Morrone, 2005), the populations belonging to haplogroup V are distributed in the Pacific Coast and Volcanic Axis provinces. Being present in two biogeographical provinces with different environmental characteristics explains the fact that this haplogroup presented a broader environmental niche than the rest. Haplogroup III has populations in the Balsas Depression and a small part of the Volcanic Axis provinces, while populations of haplogroup I are only distributed within the Balsas Depression. The presence of shared populations in the Balsas Depression between haplogroups I and III could explain the slight overlap observed between ellipsoids of these haplogroups.

Although all these results suggest that the genetic differentiation between haplogroups of T. pallidipennis could reflect recent ecological segregation processes, the potential effect of genetic isolation of the populations of each haplogroup cannot be denied. Future analyses should be aimed at resolving this question, to provide more elements to elucidate the speciation processes that are occurring within T. pallidipennis. However, these results provide support from an ecological perspective that this species should be considered a cryptic species complex.

5. Conclusions

Our results show how the analysis of morphometric variation and the characterization of the environmental conditions that define the climatic niche can be used to improve the delimitation of haplogroups that may constitute cryptic species. Analysis of multiple lines of evidence (i.e. the integrative taxonomy approach) can increase our ability to detect intraspecific variation and provide strong evidence for species separation. In the case of triatomine species and other disease vectors, this information has implications for epidemiology, management, and vector control.

Funding

This study was supported by the CONACYT scholarship programme 2018-000012-01NACF-11846.

Ethical approval

Not applicable.

CRediT author statement

Daryl D. Cruz: Conceptualization, investigation, formal analysis, writing - original draft; writing - review & editing, visualization, supervision. Sandra Milena Ospina: Conceptualization, formal analysis, writing - original draft, writing - review & editing, visualization. Elizabeth Arellano: Conceptualization, writing - original draft, writing - review and editing, visualization. Carlos N. Ibarra-Cerdeña: Writing - original draft, writing - review & editing, visualization. Elizabeth Nava-García: Writing - original draft, writing - review & editing, visualization. Raúl Alcalá-Martínez: Writing original draft, writing - review & editing, visualization. All authors read and approved the final manuscript.

Declaration of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

The datasets used and/or analysed during the present study are available from the corresponding author upon request.