Phenotypic convergence in Chagas disease vectors: Rhodnius neglectus from Copernicia prunifera palms may express the typical Rhodnius nasutus color phenotype

Abstract

Background

Selection for camouflage coloration can lead to phenotypic convergence when closely-related taxa adapt to microhabitats with similar substrate color. Such convergence can, in turn, complicate species diagnosis - an issue of practical relevance when human-disease vectors are involved. Here, we present evidence suggesting that the Chagas disease vectors Rhodnius neglectus and R. nasutus may express similar, substrate-matching color phenotypes when they occupy Copernicia prunifera palms - the primary habitat of R. nasutus in northeastern Brazil.

Methods

We studied triatomines from C. prunifera sampled across the semiarid Caatinga and the Caatinga-Cerrado savanna transition, comparing them with specimens from core-Cerrado Mauritia flexuosa palms. Triatomines were identified by phenotype, then genotyped using mitochondrial cytb and nuclear rDNA ITS2 clone-sequences.

Results

Adult-specimen color patterns distinguished core-Cerrado dark brown (typical) R. neglectus from core-Caatinga light brown (typical) R. nasutus, but were dubious in bugs caught in C. prunifera from the Caatinga-Cerrado transition and parts of the central-western Caatinga. Bayesian cytb genealogies placed dubious-phenotype specimens resembling typical R. nasutus in a sub-clade sister to typical R. neglectus from M. flexuosa - with ~2% mean cytb divergence between sub-clades. ITS2 genealogies recovered R. neglectus and R. nasutus as paraphyletic; however, a multispecies-coalescent analysis suggested that, after accounting for incomplete lineage sorting, dubious-phenotype bugs from C. prunifera cluster with typical R. neglectus.

Conclusions

Rhodnius neglectus dwelling in C. prunifera palms can be chromatically similar to typical, light brown R. nasutus. Mitochondrial cytb sequence analyses helped address the taxonomic uncertainty brought about by this apparent phenotypic convergence. ITS2 results were less clear-cut, with multispecies-coalescent analysis results suggestive of incomplete lineage sorting. Taken together, our results raise the possibility that the atypical, light brown hue of R. neglectus specimens inhabiting C. prunifera may stem from predator-driven natural selection favoring camouflage phenotypes that better match the palm-crown substrate color. These findings contribute to our understanding of color variation in triatomines and call into question the widespread use of overall body color as a key phenotypic character in the taxonomy of these medically important insects.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13071-026-07651-3.

Background

True bugs in the subfamily Triatominae Jeannel, 1919 are the vectors of Trypanosoma cruzi Chagas, 1909, the etiological agent of Chagas disease [1]. Many triatomine species display substantial phenotypic variation. Phenotypes may vary among geographic populations of a given species or among bugs fed on different hosts or occupying distinct microhabitats [1–5]. Some of the most patent cases of microhabitat-related phenotypic variation in triatomines involve body color [1, 6]. It has been suggested, after observations of wild triatomines in their natural microhabitats, that at least part of this variation is related to camouflage [6]. In general, if individual animals with hues better matching the color of the background are less likely to be killed by predators, then camouflage (or “cryptic”) phenotypes are expected to spread in the population (e.g., [7]). This can also be the case for co-occurring populations of distinct species – if they undergo similar selective pressures, then their body hues may converge to the cryptic phenotype (e.g., [8]).

Triatomines of the genus Rhodnius Stål, 1859 are major vectors of T. cruzi, with several species involved in human Chagas disease transmission [1]. Most Rhodnius species are primarily associated with palms, and most of those palm-dwelling Rhodnius species have been recorded from more than a single palm species [9–11]. There is evidence that genetically similar Rhodnius populations inhabiting distinct palm microhabitats can be chromatically distinct—and that, conversely, genetically distinct Rhodnius lineages inhabiting similar palm microhabitats can have similar hues [12–17]. Two closely related and phenotypically similar Rhodnius species occur in the open eco-regions of northeastern and central Brazil [9, 17, 18]. In the semiarid Caatinga, Rhodnius nasutus Stål, 1859 is a sylvatic triatomine species that is primarily found in Copernicia prunifera (Mill.) H.E. Moore palms but sporadically invades houses and peridomestic structures [9, 10]. Bugs from these populations display the light brown hue that is typical of R. nasutus—and that closely matches the background color of the C. prunifera leaf-bases and fibers that make up the bugs’ preferred microhabitat [1, 9, 10, 19] (Fig. 1; we note that in [10], p. 424, “Copernicia tectorum” should read “Copernicia prunifera”). Rhodnius neglectus Lent, 1954 occurs primarily in the seasonally dry Cerrado savanna, where it has often been collected from Mauritia flexuosa L.f. palms, but also frequently invades and sporadically colonizes artificial structures, including human dwellings [9, 20, 21]; these populations have the species’ typical dark brown body color, which also closely resembles the color of their microhabitat among the leaf-bases of M. flexuosa crowns [1, 21] (Fig. 1).

Fig. 1.

Palm-crown microhabitat and overall body color of Rhodnius spp. from central and northeastern Brazil. In the Cerrado savanna, typical, dark brown R. neglectus are common in the crowns of Mauritia flexuosa palms, where the palm-leaf base background color is dark brown. In the semiarid Caatinga, typical, light brown R. nasutus are common in the crowns of Copernicia prunifera palms, where the palm-leaf base background color is light brown. In parts of the Caatinga–Cerrado transition and central-western Caatinga, some C. prunifera palms harbor light brown Rhodnius bugs resembling R. nasutus. Note that the four specimens in this Figure were pictured together in the same photograph shortly after field collection; the two dubious-phenotype bugs were caught in the same individual C. prunifera palm in the state of Piauí, Brazil. Light brown specimens showed no signs of having undergone recent ecdysis—they moved normally and had a hardened cuticle

Putative R. neglectus populations have also been found in C. prunifera palms in some parts of the central-western and southern Caatinga and in the transitional eco-regions between the Caatinga and the Cerrado [18] (Fig. 2). Here, we set to assess the taxonomic status of Rhodnius specimens collected from C. prunifera and M. flexuosa palms in the Caatinga, Cerrado, and Caatinga–Cerrado transitional eco-regions. Our sample comprises (i) dark brown bugs resembling typical R. neglectus and caught in M. flexuosa; (ii) light brown bugs resembling typical R. nasutus and caught in C. prunifera; and (iii) a set of bugs with dubious phenotypes, also caught in C. prunifera (Fig. 1). By genotyping (via DNA sequencing) dubious and ‘reference’ bugs with diverse color phenotypes, we show that some C. prunifera-dwelling R. neglectus have a R. nasutus-like, light brown hue while retaining overall genetic similarity to typical, dark brown R. neglectus. The patent match between bug color and microhabitat-background color raises the hypothesis that sight-guided predators may contribute to selecting camouflage body-color phenotypes in closely related Rhodnius species. Reliable differentiation of these two species is essential for accurately assessing Chagas disease transmission risk and informing targeted vector control strategies.

Fig. 2.

Sampling sites and the known range-limits of Rhodnius neglectus and R. nasutus. Point shapes and colors: orange circles, typical light brown R. nasutus; dark pentagons, typical dark brown R. neglectus; yellow squares, dubious phenotypes; diamonds, sites where only nymphs were found (yellow, light brown; brown, dark brown). Point-shape outer lines are black for bugs caught in Mauritia flexuosa palms and orange for bugs caught in Copernicia prunifera palms. Biomes: Caatinga, dark-grey area; Cerrado, light-grey area; white areas, other eco-regions (dry forests, babaçu forests, restingas, mangroves, Atlantic moist forest, and Bahia and Pernambuco interior forests). Numbers represent localities in (i) the Caatinga–Cerrado transition (light-blue shapes): northern transition (NT, 1–4), central-northern transition (CNT, 5–8), central-southern transition (CST, 9 and 10), and southern transition (ST, 11 and 12); (ii) core Caatinga (13–19); and (iii) core Cerrado (20–26). The brown contour follows the approximate eastern limit of the known distribution of R. neglectus. The orange-shaded area (with dotted red outline) covers the approximate known distribution of R. nasutus in the northern Caatinga and transition with the Cerrado. Bugs from C. prunifera: State of Piauí (PI): Parnaíba (1), Cocal de Telha (2), Campo Maior (3), Altos (4), Oeiras (5), Colônia do Gurguéia (6), Canto do Buriti (7), São Raimundo Nonato (8), Jaicós (16). State of Ceará (CE): Jaguaruana (13). State of Rio Grande do Norte (RN): Carnaúba dos Dantas (14). State of Paraíba (PB), Olivedos (15). State of Pernambuco (PE): Serra Talhada (17). State of Bahia (BA): Xique-Xique (9), Ibotirama (10), Curaçá (18), Juazeiro (19). State of Maranhão (MA): Vargem Grande (20), Loreto (21). Samples from M. flexuosa: State of Minas Gerais (MG): Januária (11), Buritizeiro (12). State of Piauí (PI): Monte Alegre do Piauí (23). State of Tocantins (TO): Palmeirante (22), Taguatinga (24). State of Bahia (BA): São Desidério (25). State of Goiás (GO): Mambaí (26). Dark-red numbers: sites from which cytb and ITS2 sequences were determined; black numbers: sites from which only cytb sequences were determined; grey numbers: sites from which bugs were used only for phenotypic comparisons (see Table S1). The white star in southern Ceará indicates the approximate location of the Chapada do Araripe, where Dias et al. [19] conducted fieldwork. All locations are approximate

Methods

We sampled wild triatomines from palms at 26 sites in 10 Brazilian states along the eastern Cerrado; across the central-northern Caatinga; and covering parts of the transitional Atlantic dry forests and Maranhão babaçu forests (Fig. 2). Most samples were from C. prunifera, but we also included bugs from M. flexuosa in comparative analyses. Bug sampling in palms was done as described elsewhere [21] and approved by the Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio, Ministry of the Environment, Brazil, permit #22404-1). Palms were identified after [22].

We first tentatively identified adult triatomines as R. neglectus or R. nasutus using Lent and Wygodzinsky’s [1] key and descriptions, which indicate that these two species can be distinguished based on 7 characters as summarized in Table 1. We used a calibrated stereomicroscope to measure total body length (anteclypeus–terminalia) in 7 typical dark brown R. neglectus from the core Cerrado, 26 typical light brown R. nasutus from the core Caatinga, and 46 dubious-phenotype bugs caught in Copernicia prunifera palms from the Caatinga–Cerrado transition. In a sample of 18 adult bugs from 12 sites in 5 states, we assessed, in addition to overall body color, the color patterns of antennal segments, legs, and abdomen (Table 1 and Table S2); in nine males, we removed the parameres and, using a stereomicroscope, qualitatively assessed the shape of the median process of the pygophore [1].

Table 1.

Phenotypic differences between adult Rhodnius neglectus and R. nasutus, according to Lent and Wygodzinsky [1]

Character	Rhodnius neglectus	Rhodnius nasutus
Body length	Larger (males, 17.5–19.0; females, 18.5–20.5 mm)	Smaller (males, 12.5–16.5; females, 14.0–18.0 mm)
Overall color	Dark brown	Light brown
Third antennal segment	Basally dark, apically light	Basally light, apically dark
Trochantera	Light colored, strongly contrasting with dark femora	Not contrasting with femora
Connexival spots	Dark and well-delimited; also evident on ventral surface	Faint and blurry, especially on ventral surface
Basal portion of abdomen	Wide, elongate light-colored central area extending onto metasternum	No such light area
Median process of the pygophore	Narrow at base	Wide at base

In our initial assessment of overall phenotype patterns, we found some bugs with ‘dubious phenotypes’, which we define as follows: (i) light brown nymphs caught in C. prunifera palms from within the known range of R. neglectus; (ii) one dark brown nymph caught in a C. prunifera from within the known range of R. nasutus; and (iii) light brown adult bugs (overall resembling R. nasutus) caught in C. prunifera from within the known range of R. neglectus (Fig. 2; Table S1). In these dubious cases, we adopted a molecular taxonomic approach and sequenced a fragment of the mitochondrial cytochrome b gene (cytb; 682 base pairs, bp; 37 dubious-phenotype bugs and 8 typical R. neglectus; Table S1) and the nuclear ribosomal second internal transcribed spacer (ITS2; 703–741 bp; 12 dubious-phenotype bugs, 2 typical R. neglectus, and 3 typical R. nasutus; Table S1) [15, 23]. DNA was extracted from bug legs using a Real Genomics kit (RBC Bioscience, Taiwan), and target sequences were amplified as described in Monteiro et al. [15] and Marcilla et al. [23]. Amplicons were purified with the Real Genomics Hi Yield Gel/PCR DNA Extraction Kit (RBC Bioscience). ITS2 amplicons from 12 specimens were cloned with the pGEM T Easy Vector System kit (Promega, USA). The plasmid DNA was prepared with the Flexiprep kit (GE Healthcare, USA) in 96-well micro-plates and filtered in Multiscreen-filter plates (Merck Millipore, USA). Up to 8 clones were sequenced for each specimen. Both strands of each marker were Sanger-sequenced with the ABI Prism BigDye Terminator v3.1 Cycle Sequencing Kit and run on an ABI 3730 sequencer (Applied Biosystems, USA) in the RTP-01 core facility of the PDTIS/FIOCRUZ DNA Sequencing at Instituto Oswaldo Cruz, Fiocruz (Rio de Janeiro, Brazil). Sequences were edited and consensus sequences generated using Seqman Lasergene v. 7.0 (DNAStar, USA). All cytb sequences were checked for stop codons, indels, and any other signs of pseudogenes. To check ITS2 sequence orthology, we predicted secondary fold structures and their stabilities with RNAfold [24] as previously described [25].

As a benchmark for our molecular taxonomic analyses, we used “reference” sequences from (i) typical, dark brown R. neglectus adults collected in M. flexuosa palms: 7 cytb sequences (including 1 from our previous works [26, 27]) and 16 ITS2 sequences from two bugs (8 clones each); and (ii) typical, light brown R. nasutus adults collected in C. prunifera palms: 5 cytb sequences from our previous works [26, 27], and 10 ITS2 sequences from two bugs (2 and 8 clones, respectively) (Fig. 2; Table S1).

We used the Tamura–Nei (TrN) model [28] and the Maximum-Composite Likelihood algorithm [29] to construct distance matrices for cytb and ITS2 sequences in MEGA 5 [30]; standard errors (SE) were computed via bootstrapping with 1000 pseudo-replicates. Standard genetic diversity metrics, such as nucleotide diversity and number of polymorphic and parsimony-informative sites, were computed with DNASP 5.10 [31].

To determine species identity, we manually aligned consensus cytb sequences from both dubious and ‘reference’ specimens with published orthologous sequences from Rhodnius robustus Larrousse, 1927 groups I to V sensu Monteiro et al. [15] (i.e., including R. montenegrensis Rosa et al. 2012 and R. marabaensis Souza et al. 2016 [17, 32, 33]), R. prolixus Stål, 1859, R. neglectus, and R. nasutus; R. barretti Abad-Franch, Palomeque & Monteiro 2013 was used to root cytb phylograms [26]. ITS2 sequences were first aligned with CLUSTALW v. 2.0 [34] and then visually checked and manually edited. Three published orthologous sequences of R. prolixus [32] were used to root ITS2 phylograms. GenBank accession codes for all the sequences used in this study are provided in Table S1.

We separately inferred Bayesian genealogies for cytb and ITS2 sequences in BEAST 1.8 [35]. Bayesian information criterion (BIC) scores from jMODELTEST 2 [36] favored the TrN model [28] with a proportion of invariable sites (+ I) for cytb data, and the Hasegawa–Kishino–Yano (HKY) model [37] with 4 gamma categories (+ Γ) for ITS2 data. We randomly generated tree priors and imposed the Yule process of speciation for all phylogenetic reconstructions. The Yule model is the simplest birth–death model of speciation and hence the best choice for phylogenies based on one or a few molecular markers [38]. Due to the multicopy nature of ITS2 sequences (see Results for further details), we also used the *BEAST 0.15 package of the BEAST 2.6 platform [39, 40] to test the phylogenetic signal of this ribosomal marker through the reconstruction of a multispecies-coalescent tree [39]. For this, cloned ITS2 sequences were assigned to 3 groups based on cytb results and the bugs’ chromatic patterns: typical R. neglectus, typical R. nasutus, and dubious specimens—nymphs and light brown adult bugs caught in C. prunifera, but with R. neglectus-like cytb sequences. For each tree (cytb, ITS2, and ITS2 multispecies-coalescent), we ran 3 independent runs for 2 × 10⁷ generations, excluding the initial 10% as burn-in. Convergence and proper mixing were confirmed by calculating the Effective Sample Size (ESS) of each run in TRACER 1.5 [41]; all parameters had ESS > 10⁴. Runs were combined using LOGCOMBINER, and a maximum-credibility tree (based on the 30,000 trees generated and a posterior probability limit of 0.5) was built using TREE ANNOTATOR, a package of the BEAST platform [35]. We used posterior probabilities (PP) to assess the statistical support for tree nodes; trees were visualized in FIGTREE v. 1.4 (https://github.com/rambaut/figtree/). A multispecies-coalescent density tree was then constructed based on the 1000 trees generated in DensiTree [42].

Results

Morphology

We found both dark brown and light brown adult bugs; the latter tended to be smaller (males: 13–17 mm; females: 16–17 mm) than typical R. neglectus, and similar in size to typical R. nasutus (Fig. 3A). Overall, we identified 9 different combinations of the chromatic characters used to separate R. neglectus from R. nasutus (Table 1); Fig. 3B illustrates those 9 chromatic patterns. Some specimens caught in C. prunifera from the core Caatinga had the typical R. nasutus chromatic pattern (Table 1), and some specimens caught in M. flexuosa from the core Cerrado and the southern Caatinga–Cerrado transition had the typical R. neglectus chromatic pattern (Fig. 3B). Some overall dark brown bugs, however, had atypical chromatic patterns relative to R. neglectus, including variations in the patterning of trochantera/femora and basal portion of the abdomen (Fig. 3B). Further, some overall light brown bugs had atypical chromatic patterns relative to R. nasutus, including variations in the patterning of antennae, trochantera/femora, and abdomen (Fig. 3B).

Fig. 3.

Phenotypic variation in adult Rhodnius neglectus and R. nasutus. A Size variation (mean and range) in adult R. neglectus and R. nasutus according to [1] (white symbols) and in wild bugs collected from palms and displaying distinct color patterns (7 typical dark brown R. neglectus from the core Cerrado, 26 typical light brown R. nasutus from the core Caatinga, and 46 dubious-phenotype bugs caught in Copernicia prunifera palms from the Caatinga–Cerrado transition). B: Alluvial plot illustrating the various chromatic patterns found in field-collected adult R. neglectus and R. nasutus. Each colored ribbon represents one chromatic pattern for bugs with either R. neglectus (Rng) or R. nasutus (Rns) mitochondrial cytochrome b haplotypes (“cytb” grey vertical bar). The dark brown and yellow ribbons highlight the patterns described as typical of R. neglectus or R. nasutus in the literature (e.g., [1]; see Table 1); the 7 more transparent ribbons are ‘atypical chromatic patterns.’ Chromatic patterns combine 5 characters (grey vertical bars): body color (“Color”: DB, dark brown; LD, light brown); color pattern of third antennal segments (“Antenna”: DL, dark base/light apex; LD, light base/dark apex; LDL, light/dark/light); color pattern of connexivum spots (“Connexivum”: DS, dark/sharp spots; FB, faint/blurry spots); color of trochantera, relative to color of femora (“Trochantera”: CF, contrasting with femora; NCF, not contrasting with femora); and color pattern of the basal portion of abdomen (“Abdomen”: LCA, light central area; NLA, no light area). For each chromatic pattern, we indicate the region(s) where we recorded them (CER, Cerrado; CAA, Caatinga; ST, southern transition; CST, central-southern transition; CNT, central-northern transition; NT, northern transition; see Fig. 1)

Molecular analyses

Mitochondrial cytb sequences

No sign of pseudogenes (double electropherogram peaks, indels, or stop-codons) was observed after sequencing 37 dubious-phenotype bugs including 8 adults and 29 nymphs (Table S1). We found substantial haplotype diversity in both R. neglectus and R. nasutus (Hd = 0.84 and 0.93, respectively), but nucleotide diversity was slightly higher in R. neglectus (π = 0.009) than in R. nasutus (π = 0.004; Table S3). Mean pairwise divergences between R. neglectus and R. nasutus ‘reference’ sequences were ~11–12% (SE 1.2). In contrast, the sequences from dubious-phenotype specimens diverged from ‘reference’ R. neglectus sequences by 1.2–2.7% (mean = 1.96%, SE 0.56); below we refer to these bugs as “putative R. neglectus” (Table 2). Finally, sequences from typical, dark brown R. neglectus from the core Cerrado diverged by 0.0–1.0%, and sequences from typical, light brown R. nasutus by 0.0–0.8% (Table 2, Table S4).

Table 2.

Mean pairwise cytb-sequence divergence based on Tamura–Nei model of nucleotide substitution within (diagonal) and between (off-diagonal) dubious-phenotype Rhodnius specimens, dark brown R. neglectus, and light brown R. nasutus. Standard errors (calculated with 1000 bootstrap pseudo-replicates) are in parentheses

Groups	Putative R. neglectus*	Rhodnius neglectus	Rhodnius nasutus
Putative R. neglectus*	0.38% (0.16)
Rhodnius neglectus	1.96% (0.56)	0.68% (0.21)
Rhodnius nasutus	11.37% (1.70)	11.29% (1.68)	0.39% (0.19)

*These putative R. neglectus include (i) light brown nymphs caught in Copernicia prunifera palms within the known range of R. neglectus; (ii) one dark brown nymph caught in a C. prunifera from within the known range of R. nasutus; and (iii) light brown adult bugs (overall resembling R. nasutus) caught in C. prunifera from within the known range of R. neglectus (Fig. 2; Table S1)

Bayesian cytb phylogenetic analyses recovered a 100%-supported clade including (i) all the sequences from putative R. neglectus (i.e., dubious-phenotype bugs); (ii) the sequences from typical, dark brown R. neglectus including ‘reference’ sequences; and (iii) the sequence from 1 dark brown nymph collected in a C. prunifera palm located outside the known distribution range of R. neglectus (#4 in Fig. 2; sample “OL01 (CA)” in Fig. 4A).

Fig. 4.

Molecular taxonomy of typical, dark brown Rhodnius neglectus, typical, light brown R. nasutus, and putative R. neglectus. Bayesian majority consensus phylogenetic trees of mitochondrial cytb (A) and nuclear rDNA ITS2 (B) sequences. Tip labels are the codes of individual bugs (see also Fig. 2 and Tables S1–S6). In A and B, letters in parenthesis indicate geographical origin: CAA, core Caatinga; CE, core Cerrado; NT, northern Caatinga–Cerrado transition; CNT, central-northern Caatinga–Cerrado transition; CST, central-southern Caatinga–Cerrado transition; ST, southern Caatinga–Cerrado transition (see also Fig. 2). In B, numbers in parentheses indicate the number of clone sequences determined for each specimen. C: Multispecies-coalescent density tree based on ITS2 sequences; individual consensus trees with differing topologies are in blue, red, and green. Bayesian posterior probabilities > 0.8 are shown near the key nodes in all trees

Nuclear ribosomal ITS2 sequences

We sequenced 2 to 8 clones of the rDNA ITS2 for each of 12 putative R. neglectus and used ITS2 sequences from typical R. nasutus and R. neglectus as ‘reference’ sequences (Table S1). The 97 ITS2 clone sequences ranged from 703 to 741 bp (mean, 722 bp), and were AT-biased (78.7–79.5%; Table S5). ITS2 clone sequences from individual bugs often differed: we found 81 unique clone sequences (or ‘ribotypes’) in the 97-sequence ITS2 dataset; the 772-character alignment (including gaps) of these ribotypes revealed 144 variable sites, 47 of which were parsimony-informative. Most gaps were associated with single nucleotides or short-tandem repeats, but there was also a 17-bp indel (Table S5; Additional file 2). All estimated ITS2 secondary structures had an overall similar shape and comparable stabilities, with only negligible differences in the number and size of loops inside helices (Additional file 3). Overall, ribotype sequences and their inferred secondary structures provide no evidence for pseudogenization. We found no ITS2 diagnostic character to unambiguously distinguish R. neglectus as defined by cytb sequences (i.e., including putative R. neglectus) from R. nasutus; instead, intra-individual ribotype pairwise divergence (0.0–2.8%) often exceeded (putative) interspecific values (0.3–3.0%; Table S6). Mean pairwise ITS2 sequence divergence levels were only slightly larger for between-group (~ 1.4–1.8%) than for within-group comparisons (~ 1.3–1.5%), and broadly overlapped when taking the associated standard errors into account (Table 3).

Table 3.

Mean pairwise ITS2-sequence divergence calculated with the maximum-composite likelihood algorithm within (diagonal) and between (off-diagonal) putative R. neglectus, typical dark brown R. neglectus, and typical light brown R. nasutus. Standard errors (calculated with 1000 bootstrap pseudo-replicates) are in parentheses

Groups	Putative Rhodnius neglectus	Rhodnius neglectus	Rhodnius nasutus
Putative Rhodnius neglectus	1.34% (0.42)
Rhodnius neglectus	1.41% (0.44)	1.31% (0.39)
Rhodnius nasutus	1.60% (0.48)	1.79% (0.52)	1.49% (0.45)

ITS2 ribotypes from R. neglectus (either putative, typical, or ‘reference’) and R. nasutus appeared as paraphyletic in the ITS2 genealogy (Fig. 4B), which recovered 2 main clades—each including clones from R. neglectus and from R. nasutus. In many cases, same-bug ribotypes clustered in separate clades, regardless of the specimen’s geographic origin and irrespective of whether the bug was a putative R. neglectus (e.g., bug CG1880), a core-Cerrado typical R. neglectus (e.g., bug PA016), or a core-Caatinga typical R. nasutus (e.g., bug CR25). Within each main clade, however, ribotype sequences from putative R. neglectus consistently clustered together with those from typical and ‘reference’ R. neglectus, and were separated from R. nasutus sequences (Fig. 4B).

In contrast with the overall paraphyletic assemblage of ribotype sequences in the ITS2 genealogy (Fig. 4B), but in close agreement with the cytb genealogy (Fig. 4A), all R. neglectus sequences (putative, typical, and ‘reference’ bugs) segregated in a 100%-supported clade in the multispecies-coalescent density tree (Fig. 4C). In other words, when incomplete lineage sorting was explicitly modeled in a multispecies-coalescent ITS2 analysis [39], all dubious-phenotype bugs from C. prunifera of the Caatinga–Cerrado transition (i.e., all putative R. neglectus) firmly clustered with typical and ‘reference’ R. neglectus (Fig. 4C)—just as they did in a genealogy of the fast-evolving, matrilineal cytb marker (Fig. 4A).

Discussion

This study provides evidence that bugs genetically close to the typically dark brown R. neglectus can express light brown phenotypes when inhabiting C. prunifera palms; those lighter bugs strikingly resemble typical R. nasutus, for which C. prunifera represents primary habitat (Fig. 1). Phylogenetic analyses of mitochondrial cytb sequences indicate that light brown bugs similar to R. nasutus and collected in C. prunifera palms of the central Caatinga and the central-southern Caatinga–Cerrado transition were genetically closer to R. neglectus than to R. nasutus (Figs. 2 and 4A). While ITS2 ribotypes did not segregate into reciprocally monophyletic clades in Bayesian genealogies, the results of a multispecies-coalescent analysis of nuclear ITS2 ribotypes were broadly consistent with cytb-based findings (Figs. 2, 4B and C). Taken together, these results raise the possibility that populations of closely related Rhodnius species may converge toward a similar color phenotype when they occupy the same palm-crown microhabitats—and that, in addition, the phenotype toward which the bugs converge is one that closely matches the background color of their shared microhabitat (Fig. 1).

The fact that the light brown color of both typical R. nasutus and putative R. neglectus visibly matches the background color of C. prunifera crowns (Fig. 1) suggests selection for camouflage or cryptic color phenotypes, perhaps mediated by pressure from sight-guided predators, as a plausible hypothetical mechanism [6–8, 43]. Testing this hypothesis was out of the scope of the present study, but we note that observational and experimental approaches can be used to that end. Observational studies, for example, may quantitatively assess color matching (e.g., [43, 44]) or use predator-vision modeling (e.g., [45])—which in our case would come with the caveat that many, widely different taxa (insects, arachnids, myriapods, reptiles, amphibians, birds, and mammals) can prey on triatomines [46, 47]. In predation experiments, model prey of different colors or color patterns are set on backgrounds of different colors or color patterns (e.g., [45, 48, 49]). Physical prey models set on natural backgrounds test for natural-predator effects in the field [45, 48], whereas digital prey and backgrounds test for “prey” detection by humans acting as sight-guided “predators” [49]. For several organisms, detailed studies have found support for the hypothesis that predation pressure may lead to color divergence among same-species populations occupying distinct habitats [7, 8, 12, 50] or to color convergence between populations of distinct species occupying similar habitats [8, 45, 48, 51]. In either case, the mismatch between phenotype and genotype has the potential to confound phenotype-based species identification—an issue of practical relevance when disease vectors are involved [12, 17].

Our evaluation of phenotypes in genotyped specimens indicates that other phenotypic characters thought to distinguish R. neglectus from R. nasutus may not be useful. Lent and Wygodzinsky [1], for example, suggest that R. neglectus are consistently larger than R. nasutus (Table 1). We found that light-colored bugs caught in C. prunifera and bearing R. neglectus genotypes were smaller than expected for the species, with male and female body-length ranges not overlapping those given by [1] for R. neglectus, but within the ranges given for R. nasutus (Fig. 3A, Table 1). The bugs’ light brown body color and small overall size, together with the fact that they were breeding in C. prunifera palm-crowns of the Caatinga or Caatinga–Cerrado transition, all suggested at first sight that these bugs were probably R. nasutus—yet, as our DNA-sequence data and analyses show, they were most likely R. neglectus. Remarkably, a small but geographically inclusive sample of field-caught adult bugs yielded 7 atypical chromatic patterns (Fig. 3B and Table 1). Although a more detailed assessment in a larger sample would be required to quantitatively investigate possible genotype–phenotype associations, these observations underscore the need to establish diagnostic characters for closely related and phenotypically similar triatomine species [17]. Our study outlines a path toward that end: the use of specimens of known genotype as a starting point in the search for reliable phenotypic differences—broadly along the lines of previous work on, for example, the Lutzomyia longipalpis (Lutz & Neiva, 1912) and the Anopheles gambiae Giles, 1902 species complexes [52–54].

There are a few well-documented examples of color variation in natural populations of Rhodnius spp. Marcilla et al. [23] reported that pale (i.e., typical) and dark R. stali Lent, Jurberg & Galvão, 1993 “varieties” from the Alto Beni, Bolivia had identical ITS2 sequences, but we lack data on mitochondrial DNA, ITS2 ribotype diversity, or habitat use in these varieties. Dark R. pallescens Barber, 1932 were discovered in Attalea Khunt palms of the Talamancan montane forests in western Panama [13]. A high-resolution population genetics study suggested that those dark-colored populations are only weakly connected by gene flow with typical, light-colored bugs from Attalea palms of the warmer central Panamanian lowlands ~150 km eastwards [14]. Mitochondrial DNA sequence analyses, on the other hand, indicate that dark and light populations most likely belong, together with northwestern-Colombian bugs, in the same sub-clade (dubbed “R. pallescens I”) within the “R. pallescens complex” [9, 14, 55]. In line with our findings for R. neglectus color variants (Fig. 4), cytb sequence data also suggest incipient divergence of dark and typical R. pallescens populations within R. pallescens I [14, 55]. This is in contrast with results showing that R. ecuadoriensis (Lent & León 1958) populations with sharply distinct phenotypes (including size and color) and occupying distinct microhabitats (including highland and lowland palms or vertebrate tree-nests) can share cytb and ITS2 haplotypes across the range of the species in Ecuador [12]. We finally note that one study suggested that R. nasutus inhabiting different palm species can express different color phenotypes [19]; these bugs, however, were collected in an area where both R. neglectus and R. nasutus are known to occur (white star in Fig. 2) and were not genotyped [19], so some uncertainty remains as to their taxonomic status. Our current results are overall consistent with the hypothesis that microhabitat features, and in particular microhabitat-background color, may be associated with body-color convergence in closely related Rhodnius species [12]. However, other processes, including phenotypic plasticity, genetic drift, or selection at loci unrelated to camouflage, could also contribute to variation in body color.

Our results also illustrate the nontrivial challenges of molecular taxonomy. First, Bayesian genealogies based on two often-used molecular markers (cytb and ITS2; e.g., [12, 15–17, 23, 27, 32, 56]) had different topologies (Fig. 4A, B). Apart from the clustering of R. neglectus mitochondrial DNA sequences with those of the R. prolixus–R. robustus species group (a phenomenon that has been observed previously, e.g., by [17, 26, 27, 33], but remains to be investigated), the most salient feature underpinning this topological incongruence is the widespread paraphyly of ITS2 ribotype sequences (Fig. 4B). When we used the multispecies-coalescent [39] to explicitly model incomplete lineage sorting, however, we recovered R. neglectus and R. nasutus as reciprocally monophyletic, sister taxa while allowing for gene tree/species tree discordance (Fig. 4C). This result points toward incomplete lineage sorting (ILS) as a likely source of ITS2 non-monophyly, but other factors, including imperfect concerted evolution, limited clone sampling, or PCR/cloning artifacts, might also play some role. ILS seems to be common in recently diverged triatomine species exhibiting mito-nuclear phylogenetic discordance, as reported for the Triatoma phyllosoma vector complex [53], and this hypothesis warrants further investigation.

We also note that limited interbreeding data suggest that (female) R. neglectus × (male) R. nasutus crosses may yield fertile offspring [57], and limited DNA-sequence data suggest that bugs with R. nasutus nuclear DNA and R. neglectus mitochondrial DNA may occur in nature [33]. We, therefore, cannot rule out the possibility that some of our bugs, and especially some of those collected at sites where both species occur, were of hybrid ancestry. Determining whether and to what extent R. neglectus and R. nasutus can hybridize at present or did so in the past will require finer-grained genetic analyses [16, 33, 58, 59], ideally using field-caught bugs from putative hybrid zones (e.g., the northern Caatinga–Cerrado transition or the easternmost tip of R. neglectus’ range in southern Ceará, Pernambuco and Paraíba; Fig. 2) and perhaps in combination with carefully designed (e.g., allowing for mate choice) intercrossing experiments.

ITS2 sequence analyses were also challenging due to the high levels of intragenomic variation detected (Fig. 4B, Table 3, Tables S5 and S6). This variation is most likely due to inefficient concerted evolution, whereby distinct ribotype copies are retained within individual genomes, and may result in intra-individual sequence-divergence values similar to, or even exceeding, typical interspecific values [60]. Multiple ITS2 ribotypes have been described in individuals of several arthropod taxa including mosquitoes [61, 62], black flies [63], and butterflies [64]. Concerted evolution, on the other hand, seems to efficiently homogenize rDNA tandem-repeat sequences in Triatoma infestans (Klug, 1834), T. dimidiata (Latreille, 1811), T. rubrovaria (Blanchard, 1843), or Panstrongylus megistus (Burmeister, 1835), thus allowing these multicopy sequences to function as a single, reliable orthologous marker [23, 65–67]. Because rDNA pseudogenization is another potential source of intragenomic ribotype diversity [25], we checked our ITS2 sequences for signs of paralogy—which in Triatoma species include (i) unstable secondary structures that are visually distinct from those of orthologous ITS2 sequences, and (ii) the clustering of pseudogene sequences away from their ingroup sequences in phylogenetic analyses [25]. The sequences differed by only a few SNPs and indels (Additional file 2), and their predicted secondary RNA structures were highly similar and stable (Additional file 3), suggesting that we likely sequenced orthologous ITS2 copies. We note, however, that the absence of clear signs of pseudogenization does not fully resolve the orthology/paralogy relationships expected in multicopy rDNA arrays. Indeed, the methods used here may have missed recently duplicated functional copies that are still structurally identical to the original gene but that could still confound the phylogenetic signal.

Finally, our analyses raise the possibility (without providing conclusive evidence) of incipient divergence between typical dark brown and atypical light brown R. neglectus inhabiting distinct palm species (Table 2, Figs. 1 and 4). In Rhodnius species, cytb population-level divergence values are typically below 1.5%, and interspecific values above 2.0% [12, 15]. When comparing our putative R. neglectus (including light brown adult bugs) with dark brown R. neglectus (including ‘reference’ and other core-Cerrado bug sequences), we scored cytb distance values ranging from 1.2% to 2.7% (Table 2, Table S4). In addition, both cytb genealogies (Fig. 4A) and the multispecies-coalescent ITS2 tree (Fig. 4C) overall favored the hypothesis that the two groups are reciprocally monophyletic. This might indicate incipient ecological speciation [68] in R. neglectus, with populations adapted to distinct palm microhabitats becoming (i) phenotypically distinct, and (ii) increasingly isolated, so that the two gene pools start diverging even in the face of some level of gene flow [68, 69]. Based on previously reported evolutionary rates for mitochondrial markers in Triatominae [70], divergence could have started as early as the beginning of the Pleistocene (2.45–0.67 million years ago). Wider sampling covering the species’ range, together with detailed phenotypic and genetic analyses (e.g., [14, 16, 71]), would be required to test this hypothesis.

Conclusions

Our results indicate that R. neglectus dwelling in C. prunifera palm crowns may express color and body-size phenotypes similar to those of typical R. nasutus—for which C. prunifera palm crowns represent primary habitat. Such phenotypic similarity is a potential source of taxonomic confusion, especially when similarly looking bugs also share a common microhabitat. Fast-evolving DNA markers such as mitochondrial cytb sequences can help determine specific status in such cases. Nuclear rDNA ITS2 sequence results were less clear-cut, with Bayesian multispecies-coalescent analyses suggestive of incomplete lineage sorting. We also found evidence that light- and dark-colored R. neglectus might be on their way to (ecological) speciation—a possibility that merits further scrutiny. Together with results from other organisms [7, 8, 48, 50, 51, 72], our findings raise the possibility that predator-driven selection pressure may play a role in the evolution of background-matching hues in Rhodnius spp. and perhaps in other triatomines. We cannot rule out the possibility that some of the specimens included in this study, especially those collected from localities where both species occur sympatrically, are of hybrid origin. Assessing whether R. neglectus and R. nasutus are capable of hybridization under natural conditions, or have experienced historical introgression, will require more comprehensive genomic investigations. Our findings contribute to the understanding of color variation in the Triatominae and call into question the widespread use of overall body color as a key phenotypic character in the taxonomy of these medically important insects.

Author contributions

Conceptualization: MGP, RG-G, FAM, FA-F; methodology: MGP, RG-G, JCA, KAM, ACB, FA-F; formal analysis: MGP, RG-G, FA-F; data curation: MGP, RG-G, FA-F; investigation: MGP, RG-G, JCA, KAM, ACB, FA-F, FAM; project administration: MGP, RG-G, FAM; resources: RG-G, FAM; supervision: MGP, RG-G, FA-F, FAM; validation: MGP, RG-G, FA-F, FAM; writing—original draft: MGP, RG-G, FA-F, FAM; writing—review and editing: MGP, RG-G, JCA, KAM, ACB, FA-F, FAM.

Funding

MGP and JCA were supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, Brazil, finance code 001). RG-G and FAM were supported by fellowships from the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Brazil). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability

All data generated and analyzed during this study are included in this published article and its supplementary information files. All cytb and ITS2 sequences were deposited in GenBank.

Declarations

Ethics approval and consent to participate

Triatomine sampling in palms was approved by the Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio, Ministry of the Environment, Brazil, permit #22404-1).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.