From water striders to water bugs: the molecular diversity of aquatic Heteroptera (Gerromorpha, Nepomorpha) of Germany based on DNA barcodes

Introduction

Aquatic insects are the dominant invertebrate fauna element in most freshwater ecosystems and are enormously variable in morphology, development, physiology, and ecology (Lancaster & Downes, 2013; Dijkstra, Monaghan & Pauls, 2014). Among the hemimetabolous insects, the Heteroptera or true bugs comprise a significant and diverse component of the world’s aquatic insect biota (Polhemus & Polhemus, 2008). They are unique as a group because they comprise both aquatic and terrestrial species, whereas other taxa include only species that are aquatic during some life stage, e.g., mayflies, stoneflies, or dragonflies (Wesenberg-Lund, 1943; Lancaster & Downes, 2013; Gullan & Cranston, 2014). Two infraorders, the Gerromorpha and Nepomorpha, are considered as primarily aquatic (Polhemus & Polhemus, 2008; Lancaster & Downes, 2013; Gullan & Cranston, 2014; Henry, 2017). With more than 4,400 described species worldwide (Henry, 2017), aquatic Heteroptera are well-known for utilizing an exceptionally broad range of habitats, ranging from the marine and intertidal to the arctic and high alpine (Polhemus & Polhemus, 2008). They may be found in almost every freshwater biotope. Approximately 120 species of the Gerromorpha and 230 species of the Nepomorpha are known from the Palearctic region (Polhemus & Polhemus, 2008). For Germany, 47 species of the Nepomorpha and 22 species belonging to the Gerromorpha have been recorded so far (Wachmann, Melber & Deckert, 2006; Strauss & Niedringhaus, 2014).

Species of the Nepo- and Gerromorpha exhibit numerous morphological and ecological adaptations to their aquatic environment. For instance, nepomorphan true bugs have a streamlined body, natatorial legs and short antennas, whereas gerromorphan species are well-known for their long slender legs which operate as motive (middle leg) and rudder (hind legs), allowing them to operate on the water surface (Wesenberg-Lund, 1943; Andersen, 1982; Lancaster & Downes, 2013; Gullan & Cranston, 2014) (Fig. 1). Furthermore, a reduction, loss, and/or polymorphism of wings can be observed in many taxa, which is controlled by environmental conditions and genetic factors (Zera, Innes & Saks, 1983; Muraji, Miura & Nakasuji, 1989; Spence & Andersen, 1994). With the exception of the omnivorous Corixidae, all aquatic true bugs are predators, feeding on any organism that can be subdued by the injection of a venom cocktail consisting of various toxins and proteolytic enzymes (Polhemus & Polhemus, 2008). On the other hand they serve as important prey for numerous fish and other organisms at higher trophic levels (McCafferty, 1981; Peckarsky, 1982; Zimmermann & Spence, 1989; Hutchinson, 1993; Klecka, 2014; Boda et al., 2015).

Due to their general high abundance in many freshwater systems, their great value as bioindicators of water quality and their unique morphological and ecological specializations for exploiting specialized microhabitats, these groups has been in the focus of entomological and ecological research for a long time (Hufnagel, Bakonyi & Vásárhelyi, 1999; Polhemus & Polhemus, 2008; Whiteman & Sites, 2008; Skern, Zweimüller & Schiemer, 2010). Nevertheless, as a result of their highly similar morphology, the determination of various species is quite difficult and requires the help of experienced taxonomists. Furthermore, it is very challenging or even impossible to identify nymphal stages or females of some species, e.g., some species of the genus Sigara Fabricius, 1775. In term of males of the Corixidae, typical diagnostic traits include the shape and size of the tarsus of the first leg (pala), the arrangement of pala pegs, and the morphology of the genitalia (Jansson, 1986; Savage, 1989). Because aquatic Heteroptera are of high importance for ecological and conservational studies, however, the correct species identification is essential (Hufnagel, Bakonyi & Vásárhelyi, 1999; Whiteman & Sites, 2008; Skern, Zweimüller & Schiemer, 2010). This is especially true for juveniles and females which can, depending on the life history of a species, dominate within a population over a given period of a year (Barahona, Millan & Velasco, 2005; Pfenning & Poethke, 2006; Wachmann, Melber & Deckert, 2006).

In the last few years, new molecular and genomic approaches have become more and more popular to overcome possible drawbacks of this traditional and morphology-based way of species assessment. Given the recent technological advancement of DNA-based methods, in particular in the field of modern high-throughput technologies (Heather & Chain, 2016), it is expected that such techniques will gradually replace traditional field and lab procedures in bioassessment studies over the coming 10–15 years (Leese et al., 2016). For example, the EU COST Action CA15219 on “Developing new genetic tools for bioassessment of aquatic ecosystems in Europe”—or DNAqua-Net (http://dnaqua.net/)—aims to accelerate the use of DNA-based approaches for the monitoring and assessment of aquatic habitats (Leese et al., 2016). Following these considerations, the analysis of single specimens, bulk samples or environmental DNA will be performed routinely as part of modern species diversity assessment studies (Yu et al., 2012; Scheffers et al., 2012; Cristescu, 2014; Shackleton & Rees, 2015; Kress et al., 2015; Creer et al., 2016). However, the effectiveness of all these approaches relies highly on comprehensive sequence libraries that act as valid references (Brandon-Mong et al., 2015; Creer et al., 2016; Morinière et al., 2016). In this context, DNA barcoding represents undoubtedly the most prominent and popular approach using sequence data for valid species identification (Hajibabaei et al., 2007; Miller et al., 2016). The concept of DNA barcoding relies on the postulate that the interspecific genetic variation is higher than the intraspecific variation of the selected marker (Hebert, Ratnasingham & de Waard, 2003; Hebert et al., 2003). As a consequence, every species is characterized by a unique DNA barcode cluster. For animals, an approximately 650 base pair (bp) fragment of the mitochondrial cytochrome c oxidase subunit I (COI) gene was proposed as the global standard for the identification of unknown specimens in terms of a given classification (sensu Hebert, Ratnasingham & de Waard, 2003; Hebert et al., 2003). However, it should be noted that various problems may affect the use of mitochondrial DNA, e.g., recent speciation events (Balvín et al., 2012; Raupach et al., 2014), heteroplasmy (Boyce, Zwick & Aquadro, 1989; Kavar et al., 2006; Kmiec, Woloszynska & Janska, 2006), incomplete lineage sorting (Petit & Excoffier, 2009), (introgressive) hybridization (Jansson, 1979a, 1979b; Calabrese, 1982; Spence & Wilcox, 1986; Wilcox & Spence, 1986; Savage & Parkin, 1998; Raupach et al., 2014), the presence of alpha-proteobacteria as Wolbachia within terrestrial arthropods (Werren, Zhang & Guo, 1995; Xiao et al., 2011; Werren, Baldo & Clark, 2008), and the existence of mitochondrial pseudogenes (Leite, 2012; Song, Moulton & Whiting, 2014). Nevertheless, a vast number of studies across a broad range of different animals demonstrate the efficiency of DNA barcoding (Spelda et al., 2011; Hausmann et al., 2013; Hendrich et al., 2015; Lin, Stur & Ekrem, 2015; Raupach et al., 2015; Barco et al., 2016; Coddington et al., 2016; Morinière et al., 2017).

Despite the fact that more than 45,000 species of true bugs have been described worldwide until now (Henry, 2017), the number of studies analyzing the usefulness of DNA barcodes to discriminate species of this highly diverse insect taxon is still low. Some studies focus on selected species (Rebijith et al., 2012; Zhou et al., 2012; Lis, Lis & Ziaja, 2013), other on specific families (Grebennikov & Heiss, 2014; Kaur & Sharma, 2017), whereas four publications provide a larger representation of various families (Park et al., 2011; Jung, Duwal & Lee, 2011; Raupach et al., 2014; Tembe, Shouche & Ghate, 2014). However, all these studies focused primarily on terrestrial species, analyzing just small number of species belonging to the Gerromorpha and/or Nepomorpha (Park et al., 2011; Jung, Duwal & Lee, 2011; Raupach et al., 2014). To our knowledge, only two publications analyzed aquatic true bugs specifically until now: Castanhole et al. (2013) investigated the variability of 17 barcode sequences of a few species from Brazil, whereas Ebong et al. (2016) successfully tested the usefulness of DNA barcodes to discriminate various species from Cameroon.

The aim of this study was to build-up a baseline for a comprehensive library of DNA barcodes for aquatic Heteroptera (Gerromorpha, Nepomorpha) of Central Europe with a focus on the German fauna and to test the efficiency of DNA barcodes to discriminate the analyzed species. Moreover, our study provides the first thorough molecular study of the aquatic Heteroptera of Germany. In doing so, we analyzed more than 700 DNA barcodes representing more than 60 species. Our library included various morphological similar taxa, e.g., species of the genera Sigara Fabricius, 1775 and Notonecta Linnaeus, 1758 as well as water striders of the genus Gerris Fabricius, 1794 from different localities in Germany. In addition to this we added various specimens from other European countries for comparison, e.g., specimens of the expansive small-bodied backswimmer Anisops sardeus Herrich-Schaeffer, 1849 (Soós et al., 2010; Berchi, 2011; Klementová & Svitok, 2014).

Materials and Methods

Species collection and identification

All analyzed Gerromorpha and Nepomorpha were collected between the years 2003 and 2017. Most of them were adults (n = 584; 96.8%). Specimens were stored in ethanol (96%) immediately after collection and identified by some of the authors (NH, MMG, MJR, PS, RN) using various keys (Nieser, 1982; Jansson, 1986; Savage, 1989; Stoffelen et al., 2013; Strauss & Niedringhaus, 2014) based on the most recent taxonomic classification (Aukema & Rieger, 1995). All specimens were carefully checked multiple times by some of the authors in order to prevent a misidentification. For our analysis we also included 109 DNA barcodes of aquatic bugs that were part of a previous barcoding study of true bugs of Central Europe and in which species identification was verified by the authors for comparison (Raupach et al., 2014). Most of the analyzed bug specimens were collected in Germany (n = 616: 86.5%), but various individuals were sampled in Austria (37; 5.2%), Greece (20; 2.8%), Spain (16; 2.3%), Switzerland (8; 1.1%), Italy (7; 1.0%), Poland (6; 0.8%), and Portugal (2; 0.3%) for comparison (Fig. 2). In this context we also included specimens of four species that are not recorded for Germany: I. Anisops sardeus Herrich-Schaeffer, 1849 (n = 5) from Greece, II. Mesovelia vittigera Horváth, 1895 (n = 4) from Greece, III. Sigara dorsalis (Leach, 1817–1818) (n = 1) from Switzerland, and IV. Velia currens (Fabricius, 1794) (n = 3) from Switzerland. The total data set consisted of 712 DNA barcodes with 63 species that are documented for Germany. Furthermore, the number of analyzed specimens per species ranged from one (eight species) to a maximum of 41 for Notonecta glauca Linnaeus, 1758.

Results

Our analyzed DNA barcode library comprised 63 species that are documented for Germany, representing 91% of the known aquatic bug species diversity of this country (Nepomorpha: n = 43 (92%); Gerromorpha: n = 20 (91%)), and additional four species that were collected in other countries and not recorded for Germany. In total, we generated 603 new barcodes of 64 species. The complete alignment of all analyzed sequences (n = 712) had a length of 658 bp, with fragments lengths ranging from a minimum of 366 bp to the full barcode fragment size of 658 bp. For some studied specimens of Cymatia coleoptrata (Fabricius, 1777) (n = 22), our analysis revealed two characteristic deletions of 39 (alignment position: 110–148) (Fig. S1) and nine nucleotides (629–637) for all studied specimens. Average base frequencies were A = 32%, C = 17%, G = 16%, and T = 35%. For eight species only one barcode sequence was generated (Table 1). Intraspecific distances ranged from zero to maximum values of 8.3% (Plea minutissima Leach, 1817) and 9.44% (C. coleoptrata) (Table 1). Maximum intraspecific pairwise distances with values >2.2% were found for 11 species (Table 1). In terms of interspecific divergence, values ranged from zero to 18.58%, with 18 species pairs having values <2.2% (Table 1). We found interspecific distances below 1% for nine species. Unique BINs were recorded for 55 species, whereas two BINs were identified for 10 species (Table 1). For two species that were represented only by one specimen, namely Arctocorisa germari (Fieber, 1848) and Corixa dentipes Thomson, 1869, our sequences did not have he required fragment length of at least 400 bp to fulfill the criteria for BIN assignment. As consequence, no BINs were available for these two species.

Our NJ analysis based on K2P distances revealed two large and distinct clusters, separating all analyzed Gerromorpha and all Nepomorpha specimens from each other (Fig. S2). For a better presentation, the topology has been split on this basis and shown in two figures (Gerromorpha: Fig. 3, Nepomorpha: Fig. 4). We found non-overlapping clusters with bootstrap values >90% for 57 species (85%) (Figs. 3 and 4). Of the analyzed 59 species with more than one specimen, 52 (88%) were monophyletic, three (5%) paraphyletic, and four (7%) polyphyletic (Table 1; Fig. S2).

The statistical maximum parsimony network analysis of species with interspecific distances below 1% revealed a close relationship between Gerris asper (Fieber, 1860) (n = 1) and Gerris lateralis Schummel, 1832 (n = 2) (Fig. 5). We found three haplotypes with a frequency of one (singletons) that were separated by only one or two mutational steps, with haplotype h1 (G. asper) connected with h2 (G. lateralis), which was in turn connected with haplotype h3 (G. lateralis). A similar situation was observed for Sigara limitata (Fieber, 1848) (n = 2) and Sigara semistriata (Fieber, 1848) (n = 5) (Fig. 5). Three different haplotypes were identified, with h1 representing all studied specimens of S. semistriata. Both unique haplotypes of S. limitata (h2, h3) were directly connected to this haplotype by two or three mutational steps. In the case of Callicorixa praeusta (Fieber, 1848) (n = 23) and Callicorixa producta (Reuter, 1880) (n = 1) we found five different haplotypes (Fig. 5), with h1 representing the dominant haplotype which includes 19 specimens of C. praeusta and the only specimen of C. producta. All other four haplotypes (h2–h5) were only scored in one specimen and connected with h1 by one or two mutational steps. A much more complex network was revealed for Sigara distincta (Fieber, 1848) (n = 7), Sigara falleni (Fieber, 1848) (n = 12), and Sigara iactans Jansson, 1983 (n = 12) (Fig. 6). We identified 16 different haplotypes in total, with six haplotypes (h1–h6) shared by more than one specimen. Three of these haplotypes (h2, h3, h5) were shared by specimens of S. falleni and S. iactans. Furthermore, haplotypes of both previously mentioned species were randomly distributed within the network. In many cases, haplotypes of S. falleni were separated merely by two mutational steps from haplotypes of S. iactans (e.g., h6 and h13) and vice versa. We found four singletons for S. falleni and five for S. iactans. In contrast to these two species, we identified only two haplotypes (h1, h8) for the seven analyzed specimens of S. distincta. Moreover, most specimens (n = 6) were identical (h1) and located at the periphery of the network. The other haplotype (h8), a singleton collected among others at Apen (Lower Saxony), was separated by more than 25 mutational steps from the network and represents the most isolated haplotype in this network by far. Therefore, S. distincta shared no haplotypes with other species.

Discussion

Our comprehensive DNA barcode library represents an important step for the molecular characterization of the freshwater fauna in Central Europe and adjacent regions. As COI sequences are used routinely in phylogeographic, phylogenetic, and evolutionary studies as well, our data can be also implemented in projects analyzing the genetic variation of species in relation to historical, geographical, and ecological factors (Galacatos, Cognato & Sperling, 2002; Damgaard, 2005, 2008b; Gagnon & Turgeon, 2010; Ye et al., 2016). Unique BINs were found for 55 species, allowing a valid identification of 82% of the analyzed 67 species. Distinct and monophyletic lineages were revealed for 52 species (78%). Our study also indicates the need of further detailed taxonomic revisions, using state-of-the-art methods for a fine-scaled characterization (Raupach et al., 2016). This is especially true for the species-rich family Corixidae. In the following we will discuss noticeable species with high intraspecific and/or low interspecific distances more in detail.

Small and cryptic: two highly distinct DNA barcode clusters within Plea minutissima Leach, 1817

Pygmy backswimmers are small bugs, usually less than 3.5 mm in length and confine themselves to the vegetation in which they hide and where they prey on mosquito larvae and other small arthropods (Schuh & Slater, 1995). For Europe, only one species of the Pleidae is recorded, namely P. minutissima. As part of our study we found two distinct lineages within the 16 analyzed specimens with high distances ranging from 8.1% to 8.3%. Both lineages were supported by high bootstrap values (99%) (Fig. 7). Most specimens of lineage A (n = 8) were found in Brandenburg and Bavaria, but also two specimens were collected in Lower Saxony (Jaderberg). In contrast to this, all specimens of lineage B (n = 8) were collected in Lower Saxony (Jaderberg, Lingen, Norderney). Whether this surprisingly high molecular diversity is a result of effects as incomplete lineage sorting (Damgaard, 2008a) or whether we found evidence for the existence of two sibling species (Damgaard, 2005), is not within the scope of this study but clearly needs further investigation.

A currently unknown species of the genus Cymatia Flor, 1860?

For the genus Cymatia, three European species are documented so far: C. coleoptrata (Fabricius, 1777), C. bonsdorffii (Sahlberg, 1819), and C. rogenhoferi (Fieber, 1864). In terms of a morphological identification, all species can be identified according to their size and hemelytral patterns without doubt (Jansson, 1986, Stoffelen et al., 2013). Our study revealed two distinct lineages within the analyzed specimens of C. coleoptrata (lineage A and B), with a K2P distances ranging from 9.13% to 9.42% and bootstrap support values of 99% (Fig. 8). Whereas lineage A includes 22 specimens from Lower Saxony (n = 21, Lingen) and Baden-Württemberg (n = 1, Wolperstwende), lineage B contains two specimens that were collected in Brandenburg (Voßberg). Both specimens of lineage B were small adult males with a body size between 4.3 and 4.5 mm and were identified using morphological traits as C. coleoptrata at first sight. Interestingly, their barcode sequences did not have the characteristic nucleotide deletions of this species (Fig. S1). Furthermore, we found no other similar sequences using the BOLD identification engine (Best ID: C. coleoptrata) (date of request: 2017-11-20). Unfortunately, both Cymatia vouchers were lost, preventing a closer reanalysis of the specimens. Nevertheless, our results should motivate heteropterologists to study more specimens of this genus, in particular from the Eastern parts of Germany, in order to verify the presence of this putative new species.

Conclusion

In our study we lay the foundations for a comprehensive DNA barcode data set for the aquatic Heteroptera in Central Europe and adjacent regions, which will act as useful reference library for freshwater bioassessment studies using modern high-throughput sequencing technologies. Unique BINs were revealed for 55 species, representing 82% of the analyzed 67 species. Furthermore, monophyletic lineages were found for 52 species (78%). Nevertheless, our molecular data highlights discordance between the generally accepted but exclusively morphologically based taxonomy and observed molecular diversity within some species of the Gerromorpha and Nepomorpha. The analysis of additional specimens from other localities and of other molecular markers, e.g., microsatellites or SNPs, will give us more insights into the taxonomic status of these species as well as in the eco-evolutionary processes underlying the observed genetic patterns. However, it should be kept in mind that the traditional aims of taxonomy are unchanged and include various aspects, e.g., detailed high-quality descriptions and delimitation of species, a classification that reflects evolution, a dynamic nomenclature, and fast and reliable identification tools. Therefore, our DNA barcode library may be considered as a promoter for such studies.

Supplemental Information