Background

Gasterophilus species are common obligate parasites in equines and are widely distributed worldwide. The genus Gasterophilus (Diptera: Oestridae) includes nine species [1]; of these, six are found in China: G. pecorum, G. intestinalis, G. nasalis, G. nigricornis, G. haemorrhoidalis and G. inermis [2,3,4]. Gasterophilus larvae parasitize the gastrointestinal tract of equids for 10–11 months and cause mucosal lesions, gastrointestinal ulcers, peritonitis, anemia and gastric rupture, which can be severely debilitating [5, 6]. The adults lay eggs directly on the hairs of host’s body, except G. pecorum, which lays its eggs on grass [1, 7]. In European and American countries where animal husbandry is highly developed, the most prevalent species are G. intestinalis and G. nasalis, with the other Gasterophilus species rarely observed [6,7,8,9,10,11,12,13]. In traditional Chinese pastoral areas (i.e. Inner Mongolia, Xinjiang and Qinghai) the distribution of Gasterophilus larvae is more complicated than in other regions. For instance, Xilin Gol Grassland of Inner Mongolia Autonomous Region is an important pastoral area in China with an abundance of grass and water. Six species of Gasterophilus were found in Inner Mongolia, the predominant species being G. intestinalis followed by G. nasalis, which is consistent with the prevalence of Gasterophilus in other countries and regions [14]. The diversity index of Gasterophilus in this region was found to be 1.31 [14]. The same Gasterophilus species have been found in Xinjiang and Inner Mongolia, but G. pecorum was the predominant species (96.17%) in Xinjiang, where the Gasterophilus species diversity index was very low (0.21) [14]. In Maduo County (MD) of Qinghai Province, only G. pecorum and G. nasalis have been documented, with the former being the predominant species (93.71%) and contributing to a low diversity index (0.23) [14].

Mitochondrial DNA markers have been widely used in studies of insect taxonomy, population genetics and evolution. More specifically, mitochondrial cytochrome c oxidase cox1 and cox2 genes have been used to examine inter- and intraspecific relationships in Diptera, Lepidoptera, Coleoptera, Hymenoptera and Hemiptera [15,16,17,18,19]. Previous investigations based on partial mitochondrial cox1 gene sequences revealed a high degree of genetic diversity that enabled differentiation of the Italian and Polish populations by PCR-restriction fragment length polymorphism [20]. In G. intestinalis from Daqing (DQ) and Yili (YL), China, genetic information is available for three mitochondrial genes, including cox1, NADH dehydrogenase subunit 5 gene (nad5) and 23S ribosomal RNA gene [21]. However, little is known about the genetic structure of geographically distinct Gasterophilus populations. In the present study, we explored the population genetic structure and haplotype distribution patterns in several populations of G. pecorum, G. intestinalis, G. nasalis and G. nigricornis in China based on analyses of mitochondrial cox1 and cox2 gene sequences.

Methods

Study sites and sample collection

From February 2014 to December 2015, third-stage Gasterophilus larvae were collected in Kalamaili Nature Reserve (KNR) of Xinjiang Uigur Autonomous Region, Duolun County (DL) of Inner Mongolia Autonomous Region and MD of Qinghai Province (Additional file 1: Figure S1). A total of 97 G. pecorum and 63 G. nasalis specimens were obtained from the three locations, 42 G. intestinalis specimens were obtained from KNR and DL and 45 G. nigricornis specimens were obtained from KNR and DL (Additional file 2: Table S1). Larvae were preserved in ethanol and identified based on morphology [1].

Among the sampling sites in this study, KNR of Xinjiang Uigur Autonomous Region is located in the desert subregion of northwestern China with an altitude of 600–1464 m, average annual temperature of 2.4 °C, average summer temperature of 22 °C, average annual precipitation of 159 mm and annual evaporation of 2090 mm [22]. DL is located at the southern end of Xilin Gol Grassland in Inner Mongolia and is part of the Eurasian steppe; it has an elevation of 1150–1800 m, average annual temperature of 1.6 °C, average summer temperature of 18.2 °C, average annual precipitation of 385 mm and average annual evaporation of 1449.4–1672.8 mm [23]. Xinjiang and Inner Mongolia are zoogeographical regions belonging to the same district in China [24, 25]. MD is located in southwest Qinghai Province northwest of Guoluo Tibetan Autonomous Prefecture and has a plateau continental climate, the altitude is 4500–5000 m, annual average temperature is -4.1 °C, average summer temperature is < 10 °C, average annual precipitation is 303.9 mm, and average annual evaporation is 1331.21 mm [26].

DNA extraction, amplification and sequencing

Total genomic DNA was extracted using a standard phenol/chloroform protocol [27], with minor modifications. Extracted DNA was frozen in DNase-free water and stored at -20 °C. A 632 bp fragment of the mitochondrial cox1 gene was amplified with primers Gco1s (5'-CAG TTG GAA TAG ACG TTG ATA CTC-3') and Gco1an (5'-AGG AAG TTC AGA ATA GCA GTG TTC-3') [20], and a 688 bp fragment of the mitochondrial cox2 gene was amplified with primers TL2-J-3037 (5'-ATG GCA GAT TAG TGC AAT GG-3') and TK-N-3785 (5'-GTT TAA GAG ACC AGT ACT TG-3') [28]. PCR amplification of cox1 was performed according to a previously described protocol [20]. PCR conditions for cox2 were as follows: 94 °C for 2 min, 35 cycles of 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min, and 25 °C for 2 min. PCR products were visualized by using 1% agarose gel electrophoresis, purified with the BigDye XTerminator Purification kit (Applied Biosystems, Foster City, CA, USA) and sequenced on an ABI 3730XL Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Sequences obtained were deposited in the GenBank database under the accession numbers MG815141-MG815634.

Data analysis

Sequences generated in this study as well as published sequences of G. intestinalis (GenBank: GU265738-GU265748, GU299281-GU299283 and KR230402–KR230415) and G. nasalis (GenBank: GU265749-GU265758 and GU299284-GU299286) were used for data analysis. Raw sequences were proofread and edited using BioEdit v.7.0.9.0 [29]. Alignment was performed with the ClustalW algorithm in MEGA v.7.0 software [30], which was also used for basic sequence statistical analysis. Haplotype identification, haplotype diversity (Hd), and nucleotide diversity (π) were estimated using DnaSP v.5.0 [31]. The Kimura two-parameter method was used to calculate genetic distances among populations in MEGA. DnaSP v.5.0 was used to evaluate population pairwise fixation index (Fst) statistics and gene flow (Nm) values. Phylogenetic trees were constructed by the neighbor-joining (NJ) method using MEGA and were assessed with 1000 bootstrap replicates [32]. Haplotype networks were constructed using TCS v.1.21 [33] with a 95% parsimony criterion. The Likelihood Analysis with Metropolis Algorithm using Random Coalescence (LAMARC) programme [34] was used to estimate population size (θ), exponential growth rate (g) and migration rate based on the maximum-likelihood approach. The search strategy included three replications of 10 short initial chains and two long final chains. The initial chains were run with 500 samples and the sampling interval was set to 20. The final chains were performed with 10,000 samples with the same sampling interval. A burn-in of 1000 samples was used for each chain.

Results

Haplotype diversity

A total of 247 cox1 sequences were obtained for G. pecorum, G. intestinalis, G. nasalis and G. nigricornis, co-analyzed with 41 published sequences of G. intestinalis from China (DQ and YL), Italy and Poland, and of G. nasalis from Italy and Poland. An analysis of all cox1 sequences identified a total of 154 haplotypes. The haplotype distribution, Hd and π of each population based on the mitochondrial cox1 gene are shown in Additional file 2: Table S1. The values for Hd and π were as follows: G. pecorum (Hd = 0.976, π = 0.01581); G. intestinalis (Hd = 0.983, π = 0.01066); G. nasalis (Hd = 0.953, π = 0.01226); and G. nigricornis (Hd = 0.973, π = 0.01719). For each species, only two to five common haplotypes were observed within localities in China and there were no shared haplotypes across G. intestinalis and G. nasalis populations in Italy and Poland.

The 247 cox2 sequences of the four Gasterophilus species defined 127 haplotypes. The haplotype distribution, Hd and π of each regional population based on the mitochondrial cox2 gene are shown in Additional file 2: Table S2. The values for Hd and π were as follows: G. pecorum (Hd = 0.01093, π = 0.959); G. intestinalis (Hd = 0.00592, π = 0.951); G. nasalis (Hd = 0.01003, π = 0.961); and G. nigricornis (Hd = 0.01217, π = 0.918). Similar to cox1, only two to six common cox2 haplotypes were observed within populations of each species.

Haplotype networks

Gasterophilus pecorum

The minimum spanning network calculated with TCS software using cox1 haplotypes of G. pecorum had three sub-networks (Fig. 1a). Of the 48 cox1 haplotypes, two (GpH4 and GpH19) were shared between KNR and DL and three (GpH1, GpH11 and GpH24) were shared between KNR and MD, while no common haplotypes were detected between DL and MD. The sub-networks formed a star-like structure that was derived from haplotype GpH4. Most unique haplotypes in the DL population were distributed around GpH4, whereas those unique to the MD population mostly surrounded GPH11. Haplotypes in the KNR population were more evenly distributed throughout the network. Individuals of GpH37 and GpH48 constituting a single sub-network were all from MD. In general, the cox2 haplotypes formed two sub-networks with GpH17 and GpH43 as the central haplotypes (Fig. 1b). The haplotype distribution of each geographical population was similar to that of cox1. Individuals in the GpH43-centric sub-network were the same as those in the cox1 haplotype sub-network comprising only MD haplotypes. Of the 45 cox2 haplotypes, three (GpH2, GpH5 and GpH17) were shared between KNR and DL, four (GpH2, GpH3, GpH4 and GpH10) were shared between KNR and MD, and GpH2 was shared by all three populations.

Fig. 1
figure 1

Haplotype networks for G. pecorum. a Haplotype network for G. pecorum based on the cox1 gene. b Haplotype network for G. pecorum based on the cox2 gene

Full size image

Gasterophilus intestinalis

The minimum spanning network calculated with TCS software using cox1 haplotypes of G. intestinalis had a single network and no sub-networks. Of the 48 cox1 haplotypes, three (GiH1, GiH4 and GiH13) were shared by KNR and DL, and one each was shared by KNR and DQ (GiH13), DL and DQ (GiH26), and DL and YL (GiH20). The network structure was dominated by a single haplotype (GiH4) (Fig. 2a). Haplotypes from Italy and Poland were all private haplotypes (i.e. found in only one population) and were less directly linked to those from China localities. The cox2 haplotype network had a star-like structure, with the most abundant haplotype (GiH6) in the center and distributed across all geographical regions (KNR and DL) (Fig. 2b). The second most frequent haplotype GiH9 was also shared by KNR and DL.

Fig. 2
figure 2

Haplotype networks for G. intestinalis. a Haplotype network for G. intestinalis based on the cox1 gene. b Haplotype network for G. intestinalis based on the cox2 gene

Full size image

Gasterophilus nasalis

The minimum spanning network calculated with TCS software using cox1 haplotypes of G. nasalis had two sub-networks. Four (GnH2, GnH6, GnH7 and GnH17) of the 33 haplotypes were shared between KNR and DL, and two (GnH6 and GnH17) were shared among KNR, DL and MD (Fig. 3a). All haplotypes from Italy and Poland were private. Haplotypes in Poland constituted a separate sub-network, while those in Italy were less directly linked to haplotypes from localities in China. The cox2 haplotype network was dominated by GnH16, which was represented by a single individual (Fig. 3b). Haplotypes GnH5, GnH6 and GnH11 were shared between KNR and MD, while GnH27 was common to DL and MD. The MD haplotypes were mostly focused on branches centered on GnH5, there was no similar phenomenon in the cox1 haplotype network. Haplotypes in KNR and DL were randomly distributed throughout the network.

Fig. 3
figure 3

Haplotype networks for G. nasalis. a Haplotype network for G. nasalis based on the cox1 gene. b Haplotype network for G. nasalis based on the cox2 gene

Full size image

Gasterophilus nigricornis

The minimum spanning network calculated with TCS software using cox1 haplotypes of G. nigricornis had two sub-networks centered around GniH6 and GniH26 (Fig. 4a). Most GL haplotypes were distributed around GniH26. The cox2 haplotype network comprised a single network centered around haplotype GniH16 (Fig. 4b). Five of the 22 haplotypes differed by more than eight mutations from haplotype GniH16, and most DL haplotypes were more closely related to the ancestral one.

Fig. 4
figure 4

Haplotype networks for G. nigricornis. a Haplotype network for G. nigricornis based on the cox1 gene. b Haplotype network for G. nigricornis based on the cox2 gene

Full size image

Phylogenetic analysis

Gasterophilus pecorum

The phylogenetic tree constructed with the NJ method based on the cox1 gene included 48 haplotypes divided into seven clades (Fig. 5a). Clade 1 included 73.68% of DL and 50% of KNR samples with seven and nine private haplotypes, respectively. Clades 3 and 7 comprised 77.42% of MD samples, and all haplotypes except GpH41 and GpH44 were private. An NJ tree constructed from cox2 sequences had five clades (Fig. 6a). DL haplotypes, including nine that were private, were mainly distributed in Clades 2 and 3, accounting for 84.21% of individuals in the DL population. Clades 1 and 5 contained the remaining 13 private haplotypes of MD except for GpH35, and included 87.10% of individuals from MD. KNR haplotypes were mainly found in Clades 1 and 2, the former included seven private haplotypes and 40% of individuals, while the latter contained six private haplotypes and 50% of individuals.

Fig. 5
figure 5

Neighbor-joining tree based on the cox1 gene for G. pecorum, G. intestinalis, G. nasalis and G. nigricornis. a Gasterophilus pecorum. b Gasterophilus intestinalis. c Gasterophilus nasalis. d Gasterophilus nigricornis

Full size image
Fig. 6
figure 6

Neighbor-joining tree based on the COII gene for G. pecorum, G. intestinalis, G. nasalis and G. nigricornis. a Gasterophilus pecorum. b Gasterophilus intestinalis. c Gasterophilus nasalis. d Gasterophilus nigricornis

Full size image

Gasterophilus intestinalis

The NJ tree constructed based on the cox1 gene included 44 haplotypes that were divided into four clades (Fig. 5b). Haplotypes from Italy and Poland were independently clustered and all were distributed in Clade 2, with the populations from the four geographical regions of China forming the three remaining clades. In the NJ tree constructed based on cox1 haplotypes, KNR and DL populations did not show any obvious clustering (Fig. 6b).

Gasterophilus nasalis

The NJ tree constructed based on cox1 haplotypes had three clades (Fig. 5c). Three of the Chinese populations (KNR, DL and MD) formed Clade 1, and the Italian and Polish populations formed Clades 2 and 3, respectively. The NJ tree constructed based on the cox2 gene had 31 haplotypes divided into three clades (Fig. 6c). The KNR population was mainly distributed in Clade 1, which included 10 private haplotypes and 66.67% of KNR individuals. All four private haplotypes of MD were distributed in clade 3, which included 77.78% of MD individuals. The DL population was distributed in three clades with no obvious aggregation.

Gasterophilus nigricornis

The NJ tree constructed from cox1 haplotypes was divided into three clades (Fig. 5d). The DL population was mainly distributed in Clade 1, which contained eight private haplotypes and 56.25% of DL individuals, whereas the KNR population was evenly distributed. The NJ tree constructed based on cox1 haplotypes was also divided into three clades (Fig. 6d). The private haplotypes of DL were all distributed in Clade 1, which included 68.75% of DL individuals. The haplogroup clustering of the KNR population was similar to that of cox1.

Genetic structure and gene flow

Gasterophilus pecorum

Mean genetic distances between different populations were calculated based on cox1 and cox2 sequences. Distances based on cox1 gene sequences among DL, KNR and MD populations were 0.008 (KNR and DL), 0.023 (KNR and MD) and 0.024 (DL and MD), whereas distances based on cox2 gene sequences were 0.006 (KNR and DL), 0.015 (KNR and MD) and 0.017 (DL and MD). These results demonstrate that the genetic distance was smallest between KNR and DL populations and larger than for other populations. Fst values for cox1 sequences were 0.03245 (KNR and DL), 0.19067 (KNR and MD) and 0.22841 (DL and MD), while the corresponding Nm values were 7.45416, 1.06116 and 0.84452, respectively. Fst values for cox2 sequences were 0.04216 (KNR and DL), 0.17652 (KNR and MD) and 0.17893 (DL and MD), and Nm values were 5.67979, 1.16627 and 1.14719, respectively. These results indicate that there was a low degree of genetic differentiation (Fst < 0.05, Nm > 4) between KNR and DL populations, and moderate genetic differentiation (0.15 < Fst < 0.25) and low level of gene flow (Nm < 4) between MD and the other two populations.

Gasterophilus intestinalis

Genetic distances among different populations based on cox1 sequences are shown in Table 1. Mean genetic distances among Chinese, Italian and Polish populations ranged between 0.018–0.023. The distances among populations in China did not differ significantly (0.009–0.012), whereas the distance between KNR and DL populations based on cox2 sequences was 0.006. Fst and Nm values for cox1 sequences among the six populations are shown in Table 2. Considerable genetic differentiation (Fst > 0.25) and low level of gene flow (Nm < 4) occur among Italian, Polish and Chinese populations, but restricted by the very small sample size of Italian population, this result remains to be verified. Genetic differentiation was low among KNR, DL and DQ populations (Fst < 0.05, Nm > 4) and moderate among the remaining populations in China (0.05 < Fst < 0.15), which also exhibited a low degree of gene flow (Nm < 4). Fst and Nm values for cox2 sequences between KNR and DL were 0.01847 and 13.28546, respectively, indicating a low level of genetic differentiation (Fst < 0.05) and frequent gene flow (Nm > 4).

Table 1 Mean genetic distance between G. intestinalis populations in distinct areas based on mitochondrial cytochrome c oxidase subunit 1 gene
Full size table
Table 2 Pairwise fixation index (Fst: below diagonal) and gene flow (Nm: above diagonal) values for six G. intestinalis populations
Full size table

Gasterophilus nasalis

Mean genetic distances among populations based on cox1 sequences are shown in Table 3. Distances based on cox2 sequences were 0.009 (KNR and DL), 0.015 (KNR and MD) and 0.013 (DL and MD), and those based on cox1 sequences ranged between 0.009–0.023, with the closest genetic distance observed between KNR and DL populations. The sequence dataset for cox2 also revealed a close genetic distance between KNR and DL and showed that MD was distant from the other populations. Fst and Nm values for the cox1 gene among the four populations are shown in Table 4. Apart from the low degree of genetic differentiation (Fst < 0.05) and frequent gene flow (Nm > 4) between KNR and DL, the other populations showed high genetic differentiation (Fst > 0.25) and low gene flow (Nm < 4).

Table 3 Mean genetic distance between G. nasalis populations in distinct areas based on mitochondrial cytochrome c oxidase subunit 1 gene
Full size table
Table 4 Pairwise fixation index (Fst: below diagonal) and gene flow (Nm: above diagonal) values for five G. nasalis populations
Full size table

Gasterophilus nigricornis

The mean genetic distances between KNR and DL populations calculated based on cox1 and cox2 sequences were 0.019 and 0.013, respectively. Fst and Nm values for the cox1 gene between KNR and DL were 0.07371 and 3.14167, respectively, while those for the cox2 gene were 0.08215 and 2.79321, respectively. These results showed that there was little genetic differentiation (Fst > 0.05) and low level of gene flow (Nm < 4) between KNR and DL.

Demographic history and migration

Populations are often far from equilibrium, and not all deviate in the same direction. The neutrality of all populations was estimated using Tajima’s D and Fu’s Fs (Additional file 2: Table S3). The results of the neutrality test indicated a recent rapid expansion in two populations (KNR and DL) of G. pecorum, three populations (KNR, DL and DQ) of G. intestinalis, and in the KNR population of G. nasalis, whereas the other populations had remained relatively stable.

The LAMARC analysis (Additional file 2: Table S4) revealed that population size (θ) and exponential growth rate (g) were relatively high for the DL population followed by the KNR population of G. pecorum, while the low θ value and negative g value of the MD population indicated a declining trend. Migration was high from DL to KNR and from KNR to MD, whereas the emigration and immigration rates of MD and DL, respectively, were low. Among G. intestinalis populations, DQ had the highest θ and g values followed by KNR. High migration was observed among Chinese populations, including from DL to KNR, DL to YL, DL to DQ and YL to DQ. In general, KNR populations had low emigration and high immigration rates, while the opposite was true for DL. In European populations, migration was more frequent from Poland to Italy than in the other direction. The relationship between the Italian and Chinese populations was more complex than that between the Polish and Chinese populations, but the results may be biased due to the limited number of samples from Italy. Among G. nasalis, the highest g and θ values were observed in the MD and Italy populations, respectively. In China, the highest rates of migration were from DL to KNR and MD. The migration rates from Italy to Poland were even higher, while those between the Chinese and European populations were generally low. The θ values of G. nigricornis in KNR and Dowered were similar, whereas g values were higher in DL than in KNR. Migration rates between the two populations were similarly high, with migration from KNR to DL being slightly higher.

Discussion

In this study, analyses of the mitochondrial cox1 and cox2 genes of G. pecorum, G. intestinalis, G. nasalis and G. nigricornis revealed that besides cox1 gene haplotypes in the DL population of G. nasalis, the other populations of each species showed high Hd, indicating a high degree of cox1 and cox2 gene polymorphism in Gasterophilus species, all of which showed a high degree of concordance with fewer common haplotypes and fewer samples sharing the same haplotype. The high diversity of G. pecorum has been previously reported [35]. Compared to the mitochondrial cox1 gene, the cox2 sequences of G. pecorum and G. nasalis showed higher Hd and π and more private haplotypes. The haplotype networks and phylogenetic trees revealed that haplotypes of DL and MD populations of G. pecorum had distinct distributions that were similar to the trends of G. nasalis. The Italian, Polish and Chinese populations of G. intestinalis clustered into three clades, all Chinese isolates clustered together without geographical separation. Compared to the G. nigricornis KNR population, MD showed a certain degree of aggregation.

Host migration affects the genetic structure of parasite populations [36, 37]. The geographical distribution of species is closely related to biological and non-biological factors in the ecological niche, including exogenous factors such as geological events, changes in climate and food distribution as well as intrinsic species characteristics such as behavioral sensitivity, food web specificity and diffusibility. These factors act either directly or indirectly on both the host and parasite, with more profound effects on the latter [38]. The life-cycle of Gasterophilus spp. consists of the obligate parasitic larval stage on the horse and the free-living adult stage, and is influenced by host movement and environment. The three sampling sites in China selected for this study represent different types of grassland with distinct climates. Only G. pecorum and G. nasalis were collected in MD, our analysis revealed a moderate degree of genetic differentiation and a lack of gene flow between MD and populations in other regions. In contrast, more frequent genetic flow was observed between the KNR and DL populations of all four species. In general, the four Gasterophilus species showed a consistent population genetic structure in the KNR and DL populations, whereas two of the species in MD differed from their counterparts in KNR and DL. This implies that even if there is host exchange as a part of livestock husbandry trade, the natural conditions of the Qinghai-Tibet Plateau characterized by an alpine climate and high altitude differ from those of the other two regions located on natural geographical barriers such as Kunlun and Qilian Mountains, which can lead to genetic differentiation while preventing gene exchange among populations. Xinjiang and Inner Mongolia constitute a single zoogeographical region (Mongo-Xinjiang region) [24, 25]; consistent with this, our results suggest that Gasterophilus populations in KNR of Xinjiang and DL of Inner Mongolia share similar a genetic structure and species type.

Gasterophilus pecorum, G. intestinalis and G. nasalis from DL showed a consistently high migration rate, suggesting that DL is the origin of Gasterophilus flies in Mongolia-Xinjiang and Qinghai-Tibet regions of China. The Xinjiang Uygur Autonomous Region includes KNR at the heart of the Eurasian continent. Historically, the Silk Road of China passed through this area, which saw frequent trade of animals such as horses that were important means of transportation [39]. The generally higher migration rates of Gasterophilus flies in KNR indicate that Gasterophilus spp. in the area spread from multiple locations, likely reflecting the frequent movement of their hosts. Although we sampled a limited number of sites in this study, our results nonetheless demonstrate that Gasterophilus spp. migrated westward from the grasslands of eastern China. The neutrality test results showed that G. pecorum and G. intestinalis underwent significant population expansion and were the predominant species in KNR and DL, respectively.

Conclusions

We investigated the genetic structure and inter-population relationships of four common Gasterophilus species of different geographical populations based on analyses of mitochondrial cox1 and cox2 genes. Based on these results, we propose dispersal patterns of these species in China’s three major pastoral areas.