CONTRASTED ECOLOGICAL REPARTITION OF THE NORTHERN FOWL MITE ORNITHONYSSUS SYLVIARUM (MESOSTIGMATA: MACRONYSSIDAE) AND THE CHICKEN RED MITE DERMANYSSUS GALLINAE (MESOSTIGMATA: DERMANYSSIDAE)

The two hematophagous mite species, the Northern Fowl Mite Ornithonyssus sylviarum (Macronyssidae) and the Chicken Red Mite Dermanyssus gallinae (Dermanyssidae), are considered serious pests in fowl farms in the USA and in Europe respectively. However, neither of them seems to be restricted to either of these two continents according to sparse records in the literature. The aim here was to explore the respective ecological repartitions of these two species in France and to compare them with data from the USA. We thus analyzed hematophagous mesostigmatid mites collected in France in natura, in orchards agroecosystems, in pet bird and fowl farms and compared DNA sequences from some North American and French mites in both species. It is remarkable that O. sylviarum has been recurrently encountered in bird nests from France and yet been absent from French layer farms, whereas it is a serious pest in layer farms in the USA. On the other hand, O. sylviarum has been isolated from ornamental bird farms in France (canary, pheasant). This suggests either a strong impact of different farming practices between continents or a colonization in process in Europe. It remains to be explored whether the opposite applies to D. gallinae in the USA. Lastly, mites belonging to the special lineage D. gallinae L1 have been isolated from North American pigeons providing mt and nDNA sequences very close to French L1 isolates. This confirms the specific status of this cryptic entity.


INTRODUCTION
The two hematophagous mite species, the Northern Fowl Mite Ornithonyssus sylviarum (Canestrini and Fanzago, 1877) (Mesostigmata: Dermanyssoidea: Macronyssidae) and the Chicken Red Mite Dermanyssus gallinae (De Geer, 1778) (Mesostigmata:Dermanyssoidea:Dermanyssidae) are obligatory ectoparasites primarily parasitizing birds. They are serious pests in fowl farms in the USA and in Europe respectively. Neither of them seems to be restricted to either of these two continents according to sparse records in the literature.
Ornithonyssus sylviarum has been recorded in nests of various birds in several European countries (Ambros et al., 1992;Kristofík et al., 2007) as well as in some French layer farms (Bruneau et al., 2002), but is not considered a problem in layer farms in France. In contrast, it is is a well-known pest in permanent parasites (Moss, 1978). Likely due to this difference between the two economic pest species, although both are able to be starved over long periods, Kirkwood (1963) has shown that marked difference in the duration of their starvation ability separated them, since in identical conditions and in the absence of any host, D. gallinae lived for as long as 34 weeks, compared to only 3 weeks for O. sylviarum. Chen and Mullens (2008) have shown that the on-host survival duration in O. sylviarum may increase with respect to ambient temperature and humidity as well as to the mite's development stage, but nonetheless remaining far more inferior to the range noted in D. gallinae.
Most of the repartition data of these species are based upon morphological identifications. These appeared to be partly irrelevant concerning D. gallinae according to Moss et al. (1970). These authors have shown that non gallinae species of Dermanyssus had likely been previously misidentified and confused with D. gallinae in some North American records. And yet, Roy et al. (2009a, b), by using both morphological and molecular data, have shown that previous European records might have been incorrect as well, since non gallinae species revealed to be much more host specific than previously believed. This is also probably the key to the amazing and parentally transmitted behaviour differences noted by Nakamae et al. (1997) between two assumed D. gallinae populations.
As a result, one might wonder whether recorded distribution of populations in both species and for both continents are relevant or not. Second, for each species are mites parasitizing wild and domestic birds in each species are truly conspecific respectively?
Our aim is to establish the respective ecological segregation of these two species in France and to compare these findings with some data from the USA, at the strictly specific level and regardless of intraspecific population differentiation. Especially, we test whether O. sylviarum is present in the French wild avifauna and in French layer farms, as the presence of D. gallinae in French wild and domestic birds has already been shown by Roy et al. (2009b).
To this end, an inventory of hematophagous 208 mesostigmatid mites in birds at the specific level has been carried out in France and based on both morphological and DNA data. The obtained DNA sequences for each species are compared with sequences obtained from other countries, including the USA.

Biological material
A detailed list of the analyzed French samples is available in Appendix 1. Additionally, a few samples collected in other countries, including the USA, are detailed in Appendix 2. Mites were sampled between 2005 and 2009. Two types of samples are considered here:   1) Material sampled in the bird's immediate environment (nest or litter). This mainly concerns wild avifauna.
2) Some mites have been directly sampled as such: mite aggregates or individuals found in the host environment (crevices, under dried droppings. . . ) or on host. This mainly concerns birds which are reared or kept by humans.
In order to get an overview of the different bird ecologies, samples have been collected in a diversity of environments, both in wild avifauna and in domestic birds. Within each of these two categories, several different types may be distinguished. The wild avifauna has been sampled in crop agroecosystems or not, and some on urban systems. Domestic birds' samples come from industrial layers, industrial broilers, amateur fowl, pet bird facilities, and even one isolate collected in game bird breeding facilities. Finally, a diversity of geographical origins within France has been sampled, in order to avoid artifacts due to endemism.
Additionally, mites have been sampled in several other countries throughout Europe, as well as in the USA, Brazil, Australia (see Appendix 2).

Nests/litter analysis
Mites were isolated from nests following the method described in de Lillo (2001), slightly modified as described in (Roy et al., 2009b). This method is based on the immersion of the nest or litter sample in a fixed quantity of water and subsequent filtration using stacked sieves. Next, the obtained filtrate is reimmersed into water and observed using a dissecting microscope. Compared to the Berlese funnel method used in Burtt et al. (1991), de Lillo's method allows detection of both living and dead mites. All detected hematophagous Mesostigmata were isolated, since their heterogenous coloration due to blood ingestion and digestion makes them highly recognizable using stereoscopic magnifying glasses. They then were digested by proteinase K, then mounted into Hoyer's medium and identified using a photonic miscroscope. In some individuals, DNA contained into the digestion product was purified and used in molecular analyses (see below).
Mites are not counted in the present study. A simple occurrence assessment is used per sample (presence/absence of each species).
A Chi-square test was used to assess the association between species (D. gallinae, O. sylviarum) and their ecological distribution assuming as the null hypothesis that ecological distribution of hematophagous Mesostigmata species is homogenous. The distribution between the observed occurence of D. gallinae and O. sylviarum in Mesostigmata containing nest/litter and their theoretical distribution was also tested under the same null hypothesis.

DNA analyses
DNA sequencing of some gene portions (16S rRNA, COI coding gene, Tropomyosin) was performed in some isolates following Roy et al. (2009a, b) and Roy (2009) in order to check the specific identity and/or assess conspecificity. One to five individuals were sequenced per isolate. Phylogenetic analyses of obtained partial sequences of the mitochondrial coding gene for cytochrome oxidase I were run using both maximum-likelihood and maximumparsimony methods. Maximum-likelihood (ML) analysis has been performed using the online software PHYML (Guindon et al., 2005) with the substitution model GTR. Branch supports were estimated by approximate Likelihood Ratio Tests (aLRT) following Anisimova and Gascuel (2006). Maximumparsimony (MP) analysis has been performed using PAUP* 4.0b10 (Swofford, 2001): a heuristic analysis has been run with TBR branch swapping and 10,000 random additions saving all most parsimonious trees. Heuristic searches in TNT 1.1 (Goloboff et al., 2008, program available with the sponsorship of the Willi Hennig Society from http://www.cladistics.org/tnt.html) were used to obtain relative Bremer (Goloboff and Farris, 2001) and bootstrap support values. TNT searches recovered the same topology and tree length as PAUP, but calculation of support values is much more efficient in TNT.

French samples
Overall, 729 French samples were analyzed, of which 361 contained at least one hematophagous Mesostigmata individual (see Table 1 Table 1). Within D. gallinae, the special lineage L1 (Roy et al., 2009a) has been shown strongly isolated on a reproductive point of view (Roy, 2009). Therefore, it is noted D. gallinae L1 all along present paper, and non L1 gallinae mites are labelled D. gallinae s. str. Concerning the specific identification of Ornithonyssus, only two species have been isolated in present samples. O. sylviarum individuals provided rather clear morphological characteristics sensu Micherdziński (1980), similar to the North American isolate OSBM's. Ornithonyssus sp. 1 was very similar to, but slightly differed from O. bacoti (Hirst, 1913), according to Micherdziński (1980) and compared with individuals of O. bacoti from a lab strain cultured in the French Museum of Natural History (O. Bain). It did not key to another of the Ornithonyssus species sensu Micherdziński (1980).

Concerning the specific identification within
Dermanyssus, all mites from domestic birds have been successfully identified at the specific level. As for samples from wild avifauna, some mites were too damaged for morphological identification, and/or DNA characterization was not yet fully de-  veloped at the moment of sample exploration (before or at the beginning of the study of Roy et al. 2009a), thus 51 samples contained Dermanyssus mites not identified at the specific level.
Only 12 samples (ie 3.3% hematophagous Mesostigmata positive samples) allowed for detection of two different species in a single nest, eight of which grouped together species of different families (Macronyssidae + Dermanyssidae or Macronyssidae + Laleapidae). The four remaining cases (1.1% hematophagous Mesostigmata positive samples) grouped together D. gallinae either with D. apodis (3 cases) or with D. carpathicus (1 case) (see Appendix 1).
As in Roy et al. (2009b), D. gallinae s. str. appeared to be the most generalist species in France, not only from the host point of view but also with regard to the environment (see Figure 2). Note that it was present in free-range as well as in caged layer farms, and in layer as well as in broiler farms (see Appendix 1). The special lineage D. gallinae L1 (Roy et al., 2009a, b) was present not only in wild but also in farm environments, almost exclusively on pigeons and never on hens. Other Dermanyssus species have been found only in wild avifauna. O. sylviarum has been found in almost all situations, but never in poultry farms in France (see Appendix 1), which is significant given the number of farms sampled (n=132). Moreover, only D. gallinae s. str. has been isolated from other European farm samples under test (see Appendix 2), but the number of samples (n=8) is insufficient to estimate the pres-ence/absence of O. sylviarum.
Lastly, the percentage of samples containing hematophagous Mesostigmata is presented in Table  2. The presence of mites in direct samples (collection of mites directly in the bird's environment or on the host) was not significant as such with regards to a prevalence estimate. Most farms had been visited as a result of farmers' complaints. As for the wild avifauna, the exact number of wild birds examined is not known. Consequently, only nests or litters are considered here. At the generic level, Dermanyssus appeared present more often than Ornithonyssus. But most of the Dermanyssus positive nest samples contained non gallinae species. On the contrary, when considering solely D. gallinae and O. sylviarum, the latter appeared slightly more often than the former (P < 0.001). It is important to keep in mind that these data are mainly composed of wild bird nests. Overall, both D. gallinae and O. sylviarum have been regularly encountered in the wild avifauna, and in distant points in France (see Figure 1).

Other samples
In addition to the French samples, 24 different mite samples of non French origin are listed in Appendix 2. Hematophagous mesostigmatid mites from layer farms collected in other European countries, as well as from Australia and from Brazil, all belong to D. gallinae s. str. In contrast, none of the three North American gallinae-like isolates belong to D. gallinae s. str. The only isolate from a North American  (Roy et al., 2009a;Roy, 2009) and have shown COI being a powerful factor in species delineation within Dermanyssus while not as powerful for interspecific relationships' resolution. Moreover, on the morphological basis, mites in this isolate key to D. gallinae, except that setae j1 and s1 are off the dorsal shield, sensu Moss (1978). These characters make it closer to D. gallinoides, however, leg chaetotaxy is typical of D. gallinae. As a result, the single-molecule analysis is considered here to be a sufficient in order to estimate that isolate JOW is strongly isolated from D. gallinae s. str. as well as L1 on the reproductive point of view. From this point, here will be referred to this entity as D. sp. JOW.
Two other gallinae-like isolates have been sampled in some pigeons in the USA (JOW_N33 and JOW_123). Both revealed to belong to the special lineage L1 of D. gallinae (Roy et al., 2009a, b). This special lineage has been recorded almost exclusively on pigeons at several locations in France and proves here to be also present on North American pigeons. COI (accession numbers FN646512-21) and Tropomyosin (accession numbers FN646507-8) sequences obtained from these two isolates are extremely close to the L1 French isolates haplotypes (1-2% divergence in COI, 0-1% in Tropomyosin intron n between American and French isolates, as well as between French isolates).
Finally, the DNA sequencing of some mitochondrial regions showed that isolates of O. sylviarum from French bird nests under test in the present study were conspecific to the North American poultry farm isolate OSBM. Based on an rRNA 16S region isolated following Roy et al. (2009a), isolates sampled in France (EMBL accession numbers: FN599080, FN646506) and isolate OSBM (accession number: FN599081) are diverging by 2-3% from each other, and by 1-2% from an isolate sampled from African Gold Breasted Starlings (Cosmopsarus regius) (accession number AY185362) (Schrenzel et al., 2003). Based on the COI coding region, OSBM (accession number FN599077) and isolates sampled in France (accession numbers FM179677, FN432541, FN432544, FN432545, FN432547, FN432548, FN599074) were diverging by 2-3% from each other. Within these French isolates, individuals sampled in tit nests located in the South of France from orchards were diverging by 1% from the two isolates sampled in nests of Montagu's Harrier, Circus pygargus (Linnaeus, 1758) in the Western North of France (FS5 and FS6, accession numbers FN432544, FN432545, FN432547, FN432548) and from individuals sampled in a canary cage in the east-central France (Rh1, accession number FN646526). FS5, FS6 and Rh1 provided exactly the same haplotype. In contrast, 14-15% divergence separates all these North American and Branch length are much shorter (more than 3 times shorter) between each other than between their common clade and their common ancestor with O. bacoti (Figure 3). All that strongly suggests that variations between included O. sylviarum samples do be intraspecific variations, not interspecific ones.

DISCUSSION
Among the eight species of hematophagous Mesostigmata recorded in the present study, the Dermanyssus species have been encountered more frequently than the Ornithonyssus species. Nevertheless, O. sylviarum is the only species of Ornithonyssus which has been repeatedly isolated, whereas each of the five species of Dermanyssus have been isolated rather frequently. The laelapid Myonyssus is a genus known to parasitize shrews and its presence is likely to be fortuitous (a single individual in a single nest).  It is obvious that O. sylviarum and D. gallinae are both present in the wild avifauna in France, and in some non-industrial bird breeding facilities (pet birds, game birds), but the former is absent to date from industrial fowl farms in France, especially from layer farms.
This contradicts the report of Bruneau et al. (2002). And yet, arguments advanced by these authors for mite identification are not discriminating at the specific level, nor even at the generic level, but are characters allowing a family-level discrimination.
Phylogenetic topologies based on mtDNA sequences do not highlight important distances between populations from layer farms (American isolate OSBM) and from the wild avifauna (French and African samples). Present data do not show that populations from farms and from the wild avi-fauna are different. However, this does not either mean that these populations are closely related and what's the more that transfers are at all possible. A close relationship between poultry and wild populations of D. gallinae is also established in Roy et al. (2009b), but no common COI haplotype has been isolated between wild and domestic birds, although mites collected in some very distant farms share the same haplotypes. In both cases, much larger sample batches along with analyses using population genetic tools are needed in order to assess the real closeness of populations parasitizing wild and domestic birds.
In any case, it is worth noting that O. sylviarum is present in environments where synthetic molecules targeting arthropoda are largely applied such as conventional orchards and wheat fields (see Appendix 1). O. sylviarum has shown being able to develop resistance to inhibitors of acetylcholinesterase as well as to pyrethinoids in the USA (Mullens et al., 2004). Populations in this species are regularly exposed to organophosphates (Ravap EC ® ) or carbamates (sevin-80s ® ) in North American poultry farms, not only by spraying, but also in some cases by digging (Mullens et al., 2009;Rubinoff I., pers. comm.), due to the typical parasitic habits of O. sylviarum described above. And yet, it is also exposed to various pesticides in French pet birds breeding facilities, including organophosphates and carbamates (M. Serre, Association Ornithologique Rhodanienne, pers. comm.), which may have resulted in a selection of resistant populations in France too. This might explain its presence in conventionally controlled crop farm agroecosystems. European layer farms are regularly treated with organophosphates and carbamates too, as well as other synthetic pesticides (such as amitraze, for instance). In France, in particular, an organophosphate-based pesticide (phoxim, Bye-Mite ® /Bye-Pou ® ) is allowed for usage not only between, but also during flocks since 2007. And many products composed of various molecules, including organophosphates (e.g. dichlorvos-based products) or carbamates (e.g. Carbyl ® ), have long been used illegally during flocks, as a consequence of the lack of allowed products for this period. All these prod-ucts are sprayed in the farm buildings, which does not differ so much from North American use. As a result, it is very unlikely that pesticide use might be the cause of the disbalance in the red mites' distribution between North American and European layer farms.
It is not impossible that O. sylviarum is currently in the process of colonizing European farms, since a few cases of infestations in layer farms have been recently reported in Northern Europe (J. Chirico, pers. comm.). And some pet bird as well as pheasant breeders provided accounts suggesting that O. sylviarum is in the process of invading their facilities (red mites staying on host since a few years). Some population genetics would be needed in order to determine whether populations found in farms in Europe come from wild birds or correspond to different populations, which might have potentially been transferred by human activities.
In the USA, no mite belonging to D. gallinae s. str. has been isolated in this study. The two samples from North American wild avifauna containing mites belonging to D. gallinae clearly belong to the special lineage L1. This confirms the specific status of this cryptic entity as already suggested in Roy et al. (2009a) and Roy (2009). Indeed, cryptic species are species which do not possess any morphological divergence with their closest sister(s) and are usually the result of a recent speciation event. Establishing the specific status of cryptic entities is not so easy. Overall, reproductive isolation corroborations need the congruence of more than one line of evidence, for instance morphology and DNA sequences. In the case of cryptic species, morphological characters do not help. Consequently, it was necessary to explore some other characters (DeSalle et al., 2005), such as life histories or ecological associations. In the case of lineage L1, molecular assessment has been checked (Roy, 2009) by comparing topologies based on mt-and nDNA and including isolates of several different origins in France. Isolates under test in this study were almost solely pigeons. In the present study, the close association of L1 with pigeons is confirmed. This characteristic, along with specific COI (mtDNA) and Tropomyosin (nDNA) haplotypes is maintained throughout Eu-rope and America. This is a considerable insight in favour of a complete reproductive isolation between this lineage and other D. gallinae entities, which appears to be more a cryptic species complex.
As for mite samples from layer farms collected in the USA, no D. gallinae s. str. has been isolated, but the number of samples is far too small to establish the presence/absence of this species in North American layer farms. J. Owen and B. Mullens, although having performed recurrent experiments in layer farms (Mullens et al., 2001(Mullens et al., , 2009Mullens, 2004, Owen et al., 2009, . . . ), only sampled mites belonging to O. sylviarum with one exception (J.O. and B.M. pers. comm.), which provided mites belonging to a non gallinae species (D. sp. JOW). Their studies did not target D. gallinae, which should be searched for by different means, due to its nidicolous habits. As a result, it is difficult to establish the presence/absence of D. gallinae in North American layer farms, due to the lack of sampling data in such environments. Nevertheless, the isolation of a non gallinae species in a North American layer farm is very interesting. Roy et al. (2009a, b) have shown that previous European records based on morphological identification alone were at least partly erroneous in the morphologically plastic species composing the gallinae group sensu Moss (1968Moss ( , 1978. And yet, Moss et al. (1970) have highlighted the likely frequent misidentifications which might have led to consider D. gallinae more often present than it was in the USA. These authors have shown that non gallinae species in the genus Dermanyssus have likely been repeatedly confused with D. gallinae. Consequently, given that no study involving morphological along with molecular analyses in the North American Dermanyssus acarofauna have been performed, the ecological distribution of D. gallinae in this continent is to be considered unkown. D. gallinae s. str. is the only species of Dermanyssus recorded in layer farms in other parts of the world, including South America according to present data. Moreover, in France, non gallinae species of Dermanyssus appear strictly restricted to the wild avifauna. Is D. sp. JOW a second Dermanyssus species adapted to layer farm environments? Or was its presence in a small free range hen house a fortuitous one? It is worth noting that D. gallinae s. str. is a relatively recent problem in Brazilian layers, since O. bursa was the predominant ectoparasite around 1939 and D. gallinae was established being more common in 1997 (Tucci et al., 2008). The causes of this colonization in process in Brazilian farms remain unclear. It is not known whether D. gallinae s. str. was present in the wild avifauna in Brazil before.
Many questions remain unsolved here. It would be especially very useful to further explore the "dermanyssofauna" in North American layer farms as well as in the wild avifauna, to check whether D. sp. JOW is recurrently present or not, and whether D. gallinae s. str. is also present or not.

CONCLUSION
It is remarkable that O. sylviarum is so commonly encountered in French bird nests under test, yet absent from French layer farms, whereas it is a serious pest in layer farms in the USA. This suggests that different farming practices between both continents might explain the paradox of the omnipresence of both species in wild avifauna and a selective presence in French layer farms. Nevertheless, this difference does not seem to be attributable to pesticides commonly used in these farms. In any case, it remains to be checked whether the opposite applies to D. gallinae s. str. in the USA. It will be also interesting to look for additional samples of D. sp. JOW and characterize it using some other DNA sequences, in order to check it specific status and establish whether it is recurrently present in North American layers.
In any case, it is not excluded that populations under test in either O. sylviarum (France and USA) or D. gallinae (Europe, Brazil, Australia), although respectively conspecific, are genetically different and do not possess the same ability to parasitize hens in farms. Some population genetics in both species are needed in order to assess it. And more generally further surveys of ectoparasites in both wild nests and domestic fowl farms are lacking in the USA and are desperately needed in order to answer such important questions.