Identification of Nearctic black flies using DNA barcodes (Diptera: Simuliidae)

Abstract

DNA barcoding has gained increased recognition as a molecular tool for species identification in various groups of organisms. In this preliminary study, we tested the efficacy of a 615-bp fragment of the cytochrome c oxidase I (COI) as a DNA barcode in the medically important family Simuliidae, or black flies. A total of 65 (25%) morphologically distinct species and sibling species in species complexes of the 255 recognized Nearctic black fly species were used to create a preliminary barcode profile for the family. Genetic divergence among congeners averaged 14.93% (range 2.83–15.33%), whereas intraspecific genetic divergence between morphologically distinct species averaged 0.72% (range 0–3.84%). DNA barcodes correctly identified nearly 100% of the morphologically distinct species (87% of the total sampled taxa), whereas in species complexes (13% of the sampled taxa) maximum values of divergence were comparatively higher (max. 4.58–6.5%), indicating cryptic diversity. The existence of sibling species in Prosimulium travisi and P. neomacropyga was also demonstrated, thus confirming previous cytological evidence about the existence of such cryptic diversity in these two taxa. We conclude that DNA barcoding is an effective method for species identification and discovery of cryptic diversity in black flies.

Introduction

Accurate taxonomic identification is a key aspect in biological research. Correct identification not only allows critical access to the broad body of literature available on a particular taxon but also permits the implementation of adequate measures to contend with species of medical or agricultural importance (Miller & Rossman 1995). In contrast, misidentifications could lead to inadequate control measures, which could potentially increase the impact caused by a particular pest species. Studies in biodiversity, systematics, community ecology and biomonitoring also depend on proper taxonomic identification. The accelerated rate of habitat destruction by humans has prompted the scientific community to accelerate the global inventory of biodiversity. However, this increased demand for taxonomic expertise corresponds with a period of decline in systematic biology as reflected by the decreased availability of qualified taxonomists.

The proposal of Hebert et al. (2003a, b), to use a small portion (c. 658 bp) of the mitochondrial gene cytochrome c oxidase unit I (COI) as a DNA barcode for species identification, has reinvigorated efforts to document global biodiversity; however, this initiative has also generated vigorous debate in the scientific community (Moritz & Cicero 2004; DeSalle et al. 2005; Ebach & Holdrege 2005a, b; Hebert & Gregory 2005; Hajibabaei et al. 2007). DNA barcoding has proved to be an effective molecular identification system in many groups of animals (e.g. Hogg & Hebert 2004; Ball et al. 2005; Barrett & Hebert 2005; Janzen et al. 2005; Monaghan et al. 2005; Schindel & Miller 2005; Ward et al. 2005; Cywinska et al. 2006; Hajibabaei et al. 2006; Pegg et al. 2006; Hinomoto et al. 2007; Kelly et al. 2007; Kerr et al. 2007), gaining recognition as a reliable taxonomic tool in certain scientific circles (Blaxter 2006). However, limited success in other taxa has revealed limitations of this particular portion of the COI gene to serve as a universal DNA barcode (e.g. Kress et al. 2005; Rubinoff et al. 2006). Nonetheless, DNA barcoding has proved to be a versatile tool with a variety of applications, for example, by facilitating the association between different developmental stages in insects (Ahrens et al. 2007). The approach has also proved to be an effective auxiliary tool in the forensic sciences (Dawnay et al. 2007), in studies on feeding ecology (Bourlat et al. 2008; Garros et al. 2008; Kuusk & Agusti 2008) and habitat conservation initiatives (Neigel et al. 2007; Ward et al. 2008), among other applications. But perhaps most importantly, DNA barcoding has proved to be especially useful in the study of taxonomically challenging taxa, where morphology-based identifications are frustrated due to cryptic diversity (Hebert et al. 2004; Quicke et al. 2006; Smith et al. 2006; Witt et al. 2006; Yassin et al. 2008), or phenotypic plasticity (Derry et al. 2003; Adamowicz et al. 2004).

Black flies (Diptera: Simuliidae) are notorious for the haematophagous habits of the adult females of most species. Besides constituting a nuisance for humans and domestic and wild animals, black flies are known to be vectors of diseases such as avian leucocytozoonosis, bovine onchocerciasis and vesicular stomatitis virus in livestock (Crosskey 1973; Adler et al. 2004). In tropical areas, anthropophilic species are implicated in the transmission of mansonelliasis (a filarial infection) and onchocerciasis or ‘river blindness’, the world's second leading infectious cause of blindness (Etya’alé 2001, 2002). The immature stages of black flies live in a wide variety of running water habitats, but typically require clean, unpolluted water. Due to their small size (typically between 2–4 mm) and structural homogeneity, black flies have proven to be taxonomically difficult. Accurate identification typically requires the analysis of large series of larvae, pupae and adults of both sexes, including micro-dissection of genitalia and slide preparations. In addition, when the presence of morphologically indistinguishable sibling species is suspected, analysis of the larva's polytene chromosomes provides the only means to accurately identify species. The study of polytene chromosomes, or ‘cytotaxonomy’, has played a central role in the taxonomy and systematics of the Simuliidae, and cytogenetic studies have been critical for revealing the presence of sibling species in many morphologically defined nominal species. Chromosomal rearrangements (typically inversions) and sex chromosomal and autosomal polymorphisms are used to diagnose sibling species, which are particularly abundant within the family (Adler et al. 2004). Many species of black flies are also the subject of pest management programmes and the larvae are important bioindicators in fresh water biomonitoring programmes. Accurate species-level identification is required for all such programmes to be successful.

Black flies provide an excellent example where species-level identification can be enhanced by implementation of a DNA-based identification system. The inherent difficulties with morphological and chromosomal identifications, with a concomitant shortage of qualified taxonomists and cytogeneticists, underscore the need for another approach. In this study, we present a preliminary assessment of the utility of DNA barcoding to discriminate among Nearctic black fly species.

Materials and methods

Collection, DNA extraction and sequencing

Larvae, pupae and adult black flies were collected at various localities throughout Canada and the USA in 2005 and 2006 (Table 1). Collected individuals were fixed in 95% ethanol and were maintained at a low temperature (≈ 5 °C) until taken to the laboratory for identification and molecular analysis. Specimens were identified using the keys in Adler et al. (2004). Larvae selected for molecular analysis had their digestive tract removed to reduce the prospect of contamination; specimens were then cut in half and the posterior halves (abdomens) were used for DNA extraction whereas the thorax and head (where most taxonomic features reside) were retained as vouchers. These latter were deposited in the entomological collection of the Royal Ontario Museum. When pupae and adults were selected for analysis, the abdomen and individual legs were used respectively for DNA extraction. All instruments used for dissection were sterilized by flame between specimen dissections in order to prevent the transfer of DNA from one sample to another. Approximately 30 µL of total DNA was extracted using a GenElute Mammalian Genomic DNA Miniprep Kit. Extracting protocol followed manufacturer specifications. Polymerase chain reaction (PCR) primers used for amplifying the c. 658-bp long target region of the COI gene were those developed by Folmer et al. (1994). These same primers are considered the standard for DNA barcoding by Hebert et al. (2003a): LCO1490 (5′-GGTCAACAAATCATAAAGATATTGG-3′) and HCO2198 (5′-TAAACTTCAGGGTGACCAAAAAATCA-3′). PCRs were conducted using 1 µL of template DNA in a total reaction volume of 25 µL. The PCR mix contained 1 µL of each primer, 1.5 µL of MgCl (50 mm), 2.5 µL of PCR buffer, 0.8 µL of dinucleotide triphosphates (dNTPs), 0.1 µL of Taq DNA polymerase (5 U/µL). PCR conditions were an initial 1 min at 96 °C (denaturation) followed by 1 min at 94 °C (denaturation), 1 min at 55 °C (primer annealing) and 1.5 min at 72 °C (amplification) for 35 cycles, and 7 min at 72 °C. Sequencing reaction was conducted using 4 µL of purified template DNA in a total reaction volume of 10 µL. Sequencing reaction mix contained 2 µL of BigDye, 1 µL of 5× sequencing buffer, 1 µL of primer and 2 µL of water. Final products were sequenced with an ABI377 or 3730 automated sequencer. Analysed sequences are available in GenBank (Accession nos FJ524384–FJ524846). BOLD accession nos are available in the online Appendix S1.

Table 1.  List of black fly species, collection sites and number of sequences used to create the DNA barcoding tree in Fig. 1. Maximum and mean intraspecific values of genetic divergence (Kimura 2-parameter pairwise distances) are shown (minimum value in all instances was zero). Values are given only to species in which three or more individuals were examined and had at least one nucleotide substitution (as implemented by BOLD); missing entries represent individuals that did not fulfil these criteria. Species belonging to recognized species complexes are treated in a broad sense (i.e. sensu lato) and are marked with a single asterisk (*); those marked with two asterisks (**) correspond to nominal species in which the existence of siblings is suspected but have yet to be confirmed. In both instances, two or more siblings might be represented under one name in our samples. Values of genetic divergence for (*) and (**) are provided in Table 2. Classification follows Adler et al. (2004)

Species	Locality	No. of specimens	% divergence [max. (mean)]
Parasimuliinae
Parasimulium crosskeyi	Washington	5	0.16 (0.06)
Simuliinae
Prosimuliini
Gymnopais dichopticoides	Yukon Territory	8	0.49 (0.2)
Gymnopais holopticus	Alaska	2	0.65/—
Twinnia nova	Montana	3	—/—
Helodon (Parahelodon) decemarticulatus	Ontario	1	2.49 (1.22)
	Manitoba	3
	Northwest Territories	2
Helodon (Parahelodon) gibsoni	Manitoba	16	1.98 (0.76)
Helodon (Helodon) alpestris	Yukon Territory	1	—/—
Helodon (Helodon) irkutense	Nunavut	3	0.16 (0.11)
Helodon (Helodon) onychodactylus*	Yukon Territory	1	See Table 2
	British Columbia	3
Prosimulium (hirtipes spp. group)
Prosimulium arvum	Ontario	1	—/—
Prosimulium frohnei	Wyoming	5	1.31 (0.99)
Prosimulium secretum	California	5	0.32 (0.16)
Prosimulium shewelli	Wyoming	9	0.65 (0.16)
Prosimulium travisi 1	California	3	3.36 (1.45)
	Montana	1
	Wyoming	1
	Alberta	4
	British Columbia	5
	Yukon Territory	1
Prosimulium travisi 2	Colorado	5	0.8 (0.4)
Prosimulium (macropyga spp. group)
Prosimulium neomacropyga 1	Yukon Territory	3	0.61 (0.04)
Prosimulium neomacropyga 2	Colorado	3	0.97 (0.66)
Prosimulium ursinum	Yukon Territory	3	0.48 (0)
Prosimulium (magnum spp. group)
Prosimulium dicum	California	1	—/—
Prosimulium exigens	Washington	1	1.81 (0.87)
	Alberta	7
Prosimulium flaviantennus	California	11	1.99 (0.97)
Prosimulium impostor	California	4	1.15 (0.28)
	British Columbia	1
Simuliini
Greniera abdita	Ontario	1	—/—
Greniera abditoides	Ontario	1	—/—
Stegopterna decafilis	Yukon Territory	7	0.32 (0.08)
Stegopterna mutata*	Ontario	5	See Table 2
	Newfoundland	1
Cnephia dacotensis	Ontario	5	1.48 (0.8)
	Manitoba	8
Metacnephia borealis	Nunavut	13	1.48 (0.70)
Metacnephia saileri	Yukon Territory	8	1.99 (0.96)
	Nunavut	8
Metacnephia saskatchewana	Nunavut	9	0.49 (0.16)
Metacnephia sommermanae	Yukon Territory	8	0.32 (0.15)
Metacnephia villosa	California	4	0.49 (0.27)
Simulium (Hellichiella) anatinum	Manitoba	2	—/—
Simulium (Hellichiella) curriei	Wyoming	8	0.49 (0.19)
Simulium (Hellichiella) nebulosum/minus?	California	2	4.58 (3.59)
	Montana	1
Simulium (Boreosimulium) balteatum	British Columbia	9	1.48 (0.91)
Simulium (Boreosimulium) emarginatum	Ontario	6	3.18 (2.07)
Simulium (Boreosimulium) joculator	California	7	1.31 (0.46)
Simulium (Boreosimulium) baffinense	Nunavut	7	—/—
Simulium (Boreosimulium) johannseni	Manitoba	4	0.48 (0.53)
Simulium (Eusimulium) bracteatum	Ontario	12	0.65 (0.23)
Simulium (Nevermannia) aestivum	Ontario	2	—/—
Simulium (Nevermannia) burgeri	Manitoba	4	1.14 (0.60)
Simulium (Nevermannia) carbunculum	Colorado	3	1.34 (0.56)
	Wyoming	4
Simulium (Nevermannia) conicum	British Columbia	2	—/—
Simulium (Nevermannia) craigi**	Montana	1	See Table 2
	Alberta	1
	Manitoba	2
	Ontario	1
	Nunavut	2
Simulium (Nevermannia) croxtoni	Manitoba	8	0.98 (0.69)
Simulium (Nevermannia) gouldingi	Ontario	7	1.15 (0.68)
Simulium (Nevermannia) pugetense	British Columbia	2	—/—
Simulium (Nevermannia) quebecense**	Ontario	3	See Table 2
Simulium (Nevermannia) silvestre	Colorado	1	3.51 (1.45)
	Wyoming	7
	Manitoba	6
	Nunavut	6
Simulium (Schoenbaueria) furculatum	Nunavut	8	1.64 (0.71)
Simulium (Psilopelmia) venator	California	3	0.49 (0.32)
Simulium (Psilozia) argus	California	2	—/—
Simulium (Psilozia) encisoi	Arizona	1	—/—
	Wyoming	1
Simulium (Psilozia) vittatum	California	1	2.8 (1.3)
	Washington	1
	Manitoba	2
	British Columbia	1
	Ontario	1
	Nunavut	3
	Northwest Territories	1
Simulium (Aspathia) hunteri	Washington	9	0.32 (0.03)
	British Columbia	11
Simulium (Aspathia) piperi	California	1	2.31 (1)
	British Columbia	10
Simulium (Hemicnetha) canadense	California	8	2.83 (1.34)
	British Columbia	9
Simulium (Simulium) arcticum*	California	3	See Table 2
	Colorado	2
	British Columbia	1
	Alaska	1
Simulium (Simulium) decimatum	Nunavut	7	0.82 (0.35)
Simulium (Simulium) decorum	Wyoming	1	0.98 (0.45)
	Manitoba	3
	Ontario	5
Simulium (Simulium) noelleri	Nunavut	13	0.49 (0.39)
Simulium (Simulium) murmanum**	Manitoba	2	See Table 2
	Ontario	1
Simulium (Simulium) rostratum	Ontario	4	3.84 (1.03)
	Nunavut	3
Simulium (Simulium) tuberosum*	Wyoming	1	See Table 2
	Manitoba	2
	British Columbia	2
	Ontario	1
	Nunavut	3
Simulium (Simulium) venustum*	British Columbia	3	See Table 2
	Ontario	5
	Manitoba	2
	Yukon Territories	1
	Nunavut	3
		Total = 463

Data analysis

Electropherograms were edited and aligned using Sequencher version 4.5. Sequences were trimmed to a final length of 615 bp. All taxa were subjected to pairwise nucleotide sequence divergence calculations using the Kimura 2-parameter (K2P) model (Table 1) because this model provides the best metric when genetic distances are low, as in closely related species (Nei & Kumar 2000). Intraspecific levels of genetic divergence was estimated in species represented by ≥ 3 specimens and with a minimum of one nucleotide substitution as implemented by the online workbench of the Barcode of Life Data Systems (BOLD). Calculations were additionally cross-checked using paup* version 4.0b10 (Swofford 2003). A total of 693 specimens representing 65 nominal species or species complexes and 10 genera were analysed (a taxonomic list of the studied taxa is presented in Table 1). Levels of genetic divergence within and between genera and species were performed using all specimens (Tables 1–3). A neighbour-joining analysis using paup* was conducted to represent genetic distances among taxa. However, in order to generate a more manageable tree, only 463 of 693 specimens were used to create the preliminary standard DNA barcode profile tree for the Nearctic black fly exemplars as shown in Fig. 1 (see below). The 463 specimens selected were chosen according to their geographical distribution and genetic variability; omission of the other 170 specimens had no significant effect on the topology of the tree.

Table 2.  Levels of genetic divergence in known and suspected species complexes and number of individuals analysed per taxon (full data set); n, number of specimens examined. Since each species likely includes two or more sibling species, mean values of genetic divergence are uninformative and therefore have been omitted, as these do not represent the actual average divergence of each sibling

Known species complexes	n	% divergence (max.)
Helodon (Helodon) onychodactylus	5	4.91
Stegopterna mutata	23	5.47
Simulium (Simulium) arcticum	62	6.50
Simulium (Simulium) venustum	83	5.31
Simulium (Simulium) tuberosum	64	5.44
Suspected species complexes	n	% divergence (max.)
Simulium (Nevermannia) craigi	27	4.93
Simulium (Nevermannia) quebecense	3	5.79
Simulium (Simulium) murmanum	15	4.58
	Total = 282

Table 3.  Intrageneric levels of genetic divergence (%). Missing entries indicate that only one species of a particular genus was analysed. Range of genetic variation in Greniera could not be estimated as only two species with one individual each was analysed; n, number of specimens examined

Genus	n	% divergence
		Min–Max	Mean
Parasimulium	5	—	—
Gymnopais	10	9.64–10.03	9.99
Twinnia	3	—	—
Helodon	26	9.85–19.38	14.29
Prosimulium	75	3.35–18.56	12.42
Cnephia	13	—	—
Greniera	2	6.53	6.53
Stegopterna	7	—	—
Metacnephia	50	5.1–12.70	8.57
Simulium s. l.	215	2.83–20.33	15.33

Figure 1. A Kimura 2-parameter neighbour-joining tree showing the DNA barcoding profile for 57 nominal species and eight species complexes of Nearctic black flies. The limit of each terminal branch representing a given species is indicated by one or two dashes, the latter indicates a recognized species complex.

Given the methodological constraints posed by the identification of sibling species, species complexes are considered in their broad sense. For example, although the Simulium tuberosum complex includes 10 reproductively isolated sibling species in North America, individuals were all referred to as ‘Simulium tuberosum complex’. The species complexes included in this study are: Helodon (Helodon) onychodactylus complex, Stegopterna mutata complex, Simulium (Simulium) arcticum complex, Simulium (Simulium) tuberosum complex and Simulium (Simulium) venustum complex. Pairwise distances in species complexes were calculated separately. This approach helps to evaluate the levels of genetic divergence within each species complex, thereby testing cytological evidence that implies a high degree of cryptic diversity. Such an approach prevents overestimating the level of intraspecific genetic divergence by underestimating the actual diversity within each species complex (as constituent siblings were not a priori identified). Similarly, three additional taxa, each suspected of being a species complex, were included (Adler et al. 2004): Simulium (Nevermannia) craigi, Simulium (Nevermannia) quebecense and Simulium (Simulium) murmanum. In addition, we included populations of Prosimulium travisi and Prosimulium neomacropyga from Colorado. Previous cytological research (Adler et al. 2004) suggested that these populations may constitute reproductively isolated sibling species. The present study offers an opportunity to test the hypothesis of specific distinctiveness by measuring the level of genetic divergence among populations assigned to these two species. Exemplars of both morphologically distinct species and sibling species are represented in the NJ tree; however, only a few exemplars of the latter were included in order to establish their relative position on the tree (this explains the lower number of individuals included in the DNA barcode profile tree compared to the actual sample size used to estimate their divergence values). Two mosquito species (Diptera: Culicidae), Aedes canadensis and Culex pipiens, were used as outgroups.

Results

The COI sequences from the sampled species showed an A + T bias in nucleotide content (mean = 0.6432) relative to the C + G content (mean = 0.3569), as is typical of arthropods (Crease 1999). Individual mean nucleotide content was: A = 0.2748, G = 0.1668, C = 0.1901 and T = 0.3684.

The following genera and subgenera were identified in the sampled material and barcoded using COI sequences: Parasimulium, Prosimulium, Helodon (subgenera Parahelodon and Helodon s. s.), Gymnopais, Twinnia, Cnephia, Greniera, Metacnephia, Stegopterna and Simulium (subgenera Hellichiella, Nevermannia, Eusimulium, Schoenbaueria, Boreosimulium, Psilopelmia, Hemicnetha, Psilozia, Aspathia and Simulium s. s.). This represents 75% of the North American simuliid genera (10 of 13) and 90% of the subgenera (10 of 11) within the highly diverse genus Simulium s. l. Individuals of the same species grouped together, even when samples were obtained from geographically disparate areas. Similarly, members of the same genus or subgenus tended to group together, although this did not hold true in all instances. Levels of sequence divergence were variable across the taxa sampled. Thus, while conspecific individuals collected from a single site often exhibited zero or modest divergence values, conspecific individuals collected from numerous and geographically distant locations also exhibited limited variability in some instances (Table 1). Mean intraspecific genetic divergence for the 58 evaluated morphologically distinct species (i.e. members of species complexes not included) was 0.76%. The maximum intraspecific divergence value (3.84%, mean = 1.03%) was observed in Simulium (Simulium) rostratum, followed by Simulium (Nevermannia) silvestre (3.51%, mean = 1.45%) and Prosimulium travisi (3.36%, mean = 1.76%; Table 2). Higher levels of divergence were found among members of cytologically defined sibling complexes: Helodon (Helodon) onychodactylus complex (max. 4.91%), Stegopterna mutata complex (max. 5.47%), Simulium (Simulium) arcticum complex (max. 6.5%), Simulium (Simulium) tuberosum complex (max. 5.44%) and Simulium (Simulium) venustum complex (max. 5.31%; Table 2). Putative species complexes also exhibited higher levels of intraspecific divergence as in Simulium (Nevermannia) craigi (max. 4.93%), Simulium (Nevermannia) quebecense (max. 5.79%) and Simulium (Simulium) murmanum (max. 4.58%). Along similar lines, specimens identified as Simulium (Hellichiella) nebulosum also exhibit a high intraspecific genetic divergence (max. 4.58%; Table 1); however, it is possible that our sample also included members of Simulium (Hellichiella) minus. These two species are indistinguishable in the immature stages, and they occur sympatrically in the Sierra Nevada of California, from where some of the studied specimens were obtained. Accordingly, the high level of divergence exhibited by Simulium (Hellichiella) nebulosum might be the result of a mixed sample. Mean intraspecific and interspecific divergence were 0.76% and 14.93%, respectively (species complexes not included; Fig. 2). At the generic level, average DNA divergence varied across taxa, with a mean interspecific divergence value of 14.99% (minimum and maximum values were 2.83% and 20.02%, respectively; Table 3).

Figure 2. Histograms showing intraspecific (A) and interspecific (B) genetic divergences (%).

Discussion

Species identification

The results obtained indicate that the portion of COI used as a DNA barcode effectively discriminates among black fly species. DNA barcoding studies on insects and invertebrates have shown maximum intraspecific variation ranging from 3 to 3.9% (Ball et al. 2005; Cywinska et al. 2006; Smith et al. 2006; Carew et al. 2007). However, smaller maximum values have been observed in other taxa, although in most cases, these were characterized by limited sampling (e.g. Hogg & Hebert 2004; Monaghan et al. 2005; Scheffer et al. 2006). We found that, at least in morphologically distinct species, the maximum values of intraspecific divergence ranged from 3.36 to 3.84%. Data from Prosimulium travisi is of particular interest because levels of intraspecific genetic divergence were derived from 282 additional individuals (unpublished; J. Rivera and D. C. Currie, in preparation) collected across the entire range of the species except for populations from Colorado (see below). Similarly, samples of Simulium (Nevermannia) silvestre and Simulium (Simulium) rostratum, although far less numerous, encompass a wide geographical area. The level of intraspecific genetic variation in species with large geographical distributions, and its significance for DNA-based identification, has not yet been evaluated systematically in a particular species; however our preliminary data suggests a positive correlation between geographical distance and intraspecific genetic divergence, at least in some species. Given the geographical scope of P. travisi samples, it is likely that the level of intraspecific divergence observed in this species is representative of the entire gene pool. The same might be true for values observed in Nearctic populations of S. (S.) rostratum (max. = 3.51%; mean = 1.45%) and S. (N.) silvestre (max. = 3.84%; mean = 1.03%), and similar maximum values can be expected in other widespread species of Nearctic black flies. Simulium (Nevermannia) silvestre is possibly a species complex, although cytological and distributional data alternatively suggests that it might represent a chromosomally polymorphic species (Adler et al. 2004). Our data on S. (N.) silvestre, although preliminary, seems to support this latter scenario. In fact, our sampled populations from various sites in Canada and USA seem to be conspecific with populations from northern Europe [identified as S. (N.) silvestre by P. Adler; data not shown]. Further studies will be necessary in order to establish the specific status of all S. (N.) silvestre populations.

The values mentioned above do not include species complexes. Much of the evidence for the existence of sibling species comes from the cytological work of K. H. Rothfels (e.g. Rothfels 1979) and subsequent generations of simuliid cytologists. Different lines of evidence suggest that rearrangements in sex chromosomes are a major evolutionary force, and such a mechanism may have played a prominent role in black fly speciation (Rothfels 1989). Although sibling species were not identified in the present study, we expect that a high degree of genetic divergence within a particular complex is indicative of cryptic diversity. The results confirm this prediction, as high values of genetic divergence were observed in taxa where sibling species are known: Helodon onychodactylus complex (max. 4.91%), Stegopterna mutata complex (max. 5.47%), Simulium (Simulium) arcticum complex (max. 6.5%), Simulium (Simulium) tuberosum complex (max. 5.44%), Simulium (Simulium) venustum complex (max. 5.31%). Maximum values of genetic divergence in these species complexes were well above those observed in S. (N.) silvestre, S. (S.) rostratum and P. travisi s.s. (i.e. 3.36–3.84%). Relatively deep divergences were observed in the branches corresponding to species complexes in the NJ tree, as in Simulium (Simulium) venustum complex, which subdivided into two subclades, suggesting that more than one species were represented. Whether or not these subclades correspond to cytologically recognized sibling species remains unclear and no obvious geographical pattern in the distribution of the genetic variability was observed in our sampling. The ability of this portion of the COI gene to detect cryptic diversity has been demonstrated in a number of invertebrate and vertebrate taxa (Hebert et al. 2004; Ball et al. 2005; Smith et al. 2006; Birky 2007; Clare et al. 2007), but this capacity has also been challenged by some authors (see Meyer & Paulay 2005; Meier et al. 2006). It appears that this particular attribute of COI varies across taxa. A possible example of the discovery of a previously unknown sibling species can be seen in the NJ tree for the Stegopterna mutata complex, which currently includes two nominal sibling species: St. mutata and St. diplomutata. Stegopterna mutata is a triploid (presumably autotriploid), parthenogenetic species whose populations consist only of females, whereas St. diplomutata is a diploid sexual species (Currie & Hunter 2003). In the absence of males, the two species can be distinguished only through examination of the polytene chromosomes of larvae. Stegopterna mutata and St. diplomutata both occur sympatrically in Algonquin Park, Ontario, arguably the most intensively studied area in the world for black flies. Although the presence of two cytological entities in Algonquin Park has long been recognized (e.g. Basrur & Rothfels 1959), the presence of three deeply divergent branches suggests that a third species might be represented in Algonquin Park. Intriguingly, Currie & Hunter (2003) suggested that St. mutata was the product of hybridization between St. diplomutata and a yet unidentified diploid species of Stegopterna. Whether one of the divergent branches of the St. mutata complex from Algonquin Park represents this unknown diploid species requires cytological confirmation. Nonetheless, DNA barcoding strongly suggests the presence of three species of Stegopterna in Algonquin Park where only two such species were known previously.

High genetic divergence values were found in Simulium (Nevermannia) craigi (max. 4.93%), Simulium (Nevermannia) quebecense (max. 5.79%) and Simulium (Simulium) murmanum (max. 4.58%). All of these entities are suspected to be complexes consisting of an indeterminate number of sibling species (Adler et al. 2004) and the high level of genetic divergence might be indicative of cryptic diversity. Another instance of the ability of COI to detect cryptic diversity is exemplified by P. travisi. This species is widely distributed throughout the mountains of western North America, ranging from Alaska and the Yukon Territories south to California, Arizona and New Mexico. Previous cytological work by Basrur (1962) and Adler et al. (2004) suggested that P. travisi may actually represent a complex of two sibling species, with isolated populations from the highlands of northern Colorado representing a separate entity. But because the Colorado population is isolated from the more widely distributed cytological form, and because chromosomal differences in allopatry cannot be evaluated, the Colorado form has until now been considered to be only a chromosomal variant (i.e. a ‘cytotype’) of P. travisi (Adler et al. 2004). However, the genetic divergence observed between the geographically isolated population of P. travisi in Colorado and those from elsewhere (respectively P. travisi 2 and P. travisi 1 in the species tree) ranged between 6.2% and 8.8%, suggesting that the Colorado population indeed constitutes a separate species. A similar situation was found in Prosimulium neomacropyga, where chromosomal differences are also known to exist between Colorado populations and their disjunct counterparts in Alaska and Yukon (respectively P. neomacropyga 2 and P. neomacropyga 1 in the species tree). Adler et al. (2004) considered these differences to represent polymorphisms of the same species. However, the level of genetic divergence between these allopatric populations was found to be between 5.4% and 7.9%, suggesting two separate entities (unpublished; J. Rivera & D. C. Currie, in prep.). Despite the genetic distinctiveness of Colorado populations of both P. travisi and P. neomacropyga, neither entity is morphology distinguishable from their more widely distributed sister species.

Unexpected ‘relationships’ among species were identified in two instances. The first is represented by the sister-species relationship between Prosimulium frohnei and the Colorado population of P. travisi (= P. travisi 2), as opposed to between the latter population and P. travisi 1. Although closely related, P. frohnei and P. travisi (both within the Prosimulium hirtipes species group) are easily distinguishable morphologically. A similar situation was found between Prosimulium formosum and the more widespread P. travisi sibling (= P. travisi 1). In this case, P. formosum grouped with P. travisi 1, clustering specifically within populations from southern Alberta (data not shown in Fig. 1). In this instance, very little genetic divergence was observed between this pair of species. This situation possibly indicates transfer of mtDNA elements through hybridization between these two species. Both species are included in the P. hirtipes species group but are easily distinguished morphologically. This presumed hybridization might compromise the ability of COI to discriminate among species pairs. Further studies, including additional members of the P. hirtipes species group, will be needed to uncover potential limitations of the COI gene for recovering species identity in this speciose lineage. Incomplete sampling is known to compromise the performance of COI in other organisms (Meyer & Paulay 2005) and more samples are needed to more fully understand the limitations, if any, of the barcoding gene for black fly identification.

Supraspecific relationships

The DNA barcode tree profile exhibited some degree of concordance with current concepts in black fly phylogeny. Members of the subfamily Parasimuliinae (represented in this study by Parasimulium crosskeyi), are considered to be the most plesiomorphic lineage of simuliid (Adler et al. 2004). Exemplars of this divergent lineage were placed as the sister-group of the simuliine genus Cnephia (represented in this study by C. dacotensis), which together formed the sister group of all other Simuliidae. Within this latter lineage, members of the tribe Prosimuliini (Twinnia, Gymnopais, Helodon and Prosimulium) formed a monophyletic lineage, except for the inclusion of Simulium (Eusimulium) bracteatum (a member of the tribe Simuliini). Although not all genera within the Prosimuliini were resolved as monophyletic, monophyly was evident at the subgeneric level in Helodon (subgenera Parahelodon and Helodon) and at the species-group level in Prosimulium (in the hirtipes–macropyga–magnum species groups). The remaining simuliine taxa clustered in different parts of the tree, with Greniera, Stegopterna, Metacnephia, Simulium (Boreosimulium), Simulium (Aspathia), and Simulium (Simulium) all forming monophyletic lineages. In contrast, Simulium (Hellichiella) and Simulium (Nevermannia), although exhibiting a certain degree of cohesion, each formed two different groups in different parts of the tree. This apparent unordered clustering pattern has also been observed in other taxa (e.g. Ball et al. 2005; Cywinska et al. 2006). This lack of cohesion among members of putative monophyletic lineages does not invalidate the effectiveness of the NJ algorithm, as the goal of DNA barcoding is to discriminate species based on sequence divergence rather than to reconstruct deep phylogenetic relationships.

Population structure

The COI gene has long been used as a molecular marker for studying phenomena at the population level. Thus, the portion of the COI gene proposed for DNA barcoding (Hebert et al. 2003a) has potential for revealing population structure in species sampled throughout their entire range. In P. travisi, for example, interpretation of phylogeographical patterns in the context of the Wisconsinan glaciation provided evidence of glacial refugia and postglacial migratory routes (J. Rivera and D. C. Currie, in preparation). These patterns were highly congruent with those exhibited by similarly distributed cordilleran animal and plant taxa (e.g. Wheeler & Guries 1982; Thorgaard 1983; Soltis et al. 1997; Comes & Kadereit 1998; Byun et al. 1999; Demboski et al. 1999; Brunsfeld et al. 2001; Nice et al. 2005; Albach et al. 2006; Maroja et al. 2007, to name only a few). The historical biogeography of Nearctic black flies has been little studied, owing largely to a dearth of fossils. We anticipate that the barcoding gene has the potential to shed important new light on the historical biogeography of black flies and other soft-bodied organisms, as already documented for the Ephemeroptera (Ball et al. 2005) and rotifers (Gómez 2005).

Current limitations for DNA barcoding black flies

Unveiling the genetic identity of morphologically identical sibling species is of paramount importance in black fly taxonomy. Several logistic and methodological limitations must be overcome in order to realize the full potential of DNA barcoding for accurately discriminating among recently diverged species. Black fly larvae are routinely collected into Carnoy's fixative (one part glacial acetic acid:three parts 95% ethanol), to preserve their polytene chromosomes for cytological analysis. This has, until now, been the only means to establish sibling species identity. Unfortunately, Carnoy's fixative causes the DNA to fragment, rendering it unsuitable for molecular analysis (Koch et al. 1998). In order to obtain material suitable for both cytogenetic and molecular analyses, we routinely collected black fly larvae in both Carnoy's fixative and 95% ethanol; however, this approach is problematic because more than one sibling species is often present at a particular collecting site. Accordingly, matching a chromosomally verified sibling species with a corresponding COI sequence can be difficult or even impossible unless both analyses (molecular and cytological) are performed on the same individual. In other words, it is often impossible to cytologically confirm the identity of specimens used for DNA analysis. Thus, we recommend that in the future, a single individual must be sectioned into three parts soon after collection and while still alive: the abdomen, where the salivary glands are located, must be preserved in Carnoy's fixative for cytological analysis; the thorax must be preserved in ethanol for molecular analysis; and the head, where most taxonomic characters needed for morphological identification are found, should be preserved in ethanol and retained as a voucher. Unfortunately, this protocol proves to be tedious and time-consuming, demanding coordination between at least two collectors to increase precision and collecting efficiency. A more palatable alternative would be to extract DNA directly from Carnoy's fixative-preserved specimens. The denaturing properties of acids notwithstanding (Koch et al. 1998), DNA has been isolated successfully from mosquitoes (Collins et al. 1987) and black flies fixed in Carnoy's solution (Pramual et al. 2005). Standardized DNA-extracting protocols for specimens preserved in Carnoy's fixative will certainly help to overcome the methodological constraints described above. Once these protocols are developed, field collections of specimens will demand no additional time or equipment than is used historically in the study of black flies. The dissection of larvae into three parts will still be required, although in the comfort of the laboratory, where correspondence between chromosomes and DNA for each individual can be more easily assured.

Conclusions

The COI barcoding gene correctly distinguished nearly 100% of the morphologically distinct species (which constituted c. 87% of the sampled taxa), thus demonstrating its utility to discriminate among morphologically recognized black fly species. Barcoding also revealed high levels of genetic divergence in known or suspected lineages of sibling species (i.e. the remaining 13% of the sampled taxa), suggesting that it might prove useful for distinguishing cryptic diversity. In fact, populations of ‘Prosimulium travisi’ and ‘Prosimulium neomacropyga’ from highland sites in Colorado were found to be specifically distinct from more widely distributed typical populations (as previously suggested by cytological evidence), necessitating nomenclatural changes to the current taxonomy. Nonetheless, additional molecular techniques (including the use of microsatellite data) together with COI and perhaps other markers may be necessary to overcome difficulties associated with discriminating recently diverged sibling species. This is a clear example of how DNA barcoding can generate taxonomic hypotheses that can be tested using other, more ‘classical’, methodologies (morphology, cytology, behaviour, natural history, etc.). Such an integrative approach might be necessary to address species recognition problems in the Prosimulium hirtipes species group and other problematic lineages of black flies.

As a mitochondrial gene, it is not surprising that COI carries phylogeographical signal, and this attribute increases the value of the barcoding gene as a tool for biogeographical and population-based studies. As COI databases become better populated with sequences from throughout the entire range of a species, it may actually become possible to establish the provenance of particular samples. This could have an impact on conservation efforts by enhancing the enforcement of laws protecting endangered populations of plants and animals.

Acknowledgements

We wish to thank Alina Cywinska for her assistance with the laboratory work, Peter Adler (Clemson University) for his helpful advise and for providing specimens of Prosimulium travisi from Colorado. Deborah Finn (Oregon State University) also provided critical material of Prosimulium neomacropyga from Colorado. Brad Hubley (Royal Ontario Museum, Department of Natural History), Gerald F. Shields and Judith Pickens (Carrol College, Montana) and Karen Needham (University of British Columbia, Department of Zoology) provided valuable technical and logistic support during field work. The staff at the Laboratory of Molecular Systematics of the Royal Ontario Museum, Amy Lathrop and Kristen Choffe, provided extensive advice and expertise during laboratory work. We are also grateful to Mateus Pepinelli and Will Shim for their assistance in different phases of this study, as well as to Wendy Lu, who provided technical support during the preparation of this manuscript. This research was funded by a Natural Sciences and Engineering Research Council of Canada Discovery Grant to D.C.C. and through funding to the Canadian Barcode of Life Network from Genome Canada (through the Ontario Genomics Institute), NSERC and other sponsors listed at http://www.BOLNET.ca.

Conflict of interest statement

The authors have no conflict of interest to declare and note that the funders of this research had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.