ARTICLE

Regionally widespread parasitic water mites have relativelybroad host-species rangesJulia J. Mlynarek, Wayne Knee, Bruce P. Smith, and Mark R. Forbes

Abstract: Certain parasite species have free-living stages, so habitat range may influence host-species range. We tested whetherregional occurrence and habitat use of parasitic water mites were related to their host-species range. We collected 7445 ArrenurusDugès, 1834 mites from 7107 coenagrionid damselflies, representing 11 host species from 13 sites in southeastern Ontario andsouthwestern Quebec. Because larval water mites are difficult to identify morphologically to species, we chose to amplify thebarcode fragment of cytochrome oxidase subunit I to explore host-species ranges. Fifteen operational taxonomic units or cladeswere identified based on the amplification from 217 larval mites. The Arrenurus clades that were present in both bog and marshhabitats had a broader host-species range than clades found only in marshes (the comparison with one clade found only in bogslacked statistical power). As predicted, host-species range increased with the regional occurrence of an Arrenurus clade. Addi-tionally, the most commonly barcoded species also have high host-species ranges. This result could be because species withbroader host-species ranges are more common and were more likely to be sampled and barcoded (an explanation we favor), ordue to sampling bias. Although this is the first study exploring whether habitat range affects host-species range, furtherinvestigation is needed to tease apart which habitat factors influence host-species ranges the most.

Key words: host–parasite associations, Odonata, Coenagrionidae, Arrenuridae, host-species range.

Résumé : Certaines espèces de parasites présentent des stades libres, de sorte que la gamme d’habitats peut influencer la gammed’hôtes. Nous avons vérifié si la distribution régionale et l’utilisation des habitats d’hydrachnes parasites étaient reliées a leurgamme d’espèces hôtes. Nous avons prélevé 7445 hydrachnes Arrenurus Dugès, 1834 de 7107 demoiselles coenagrionidésreprésentant 11 espèces hôtes de 13 sites du sud-est de l’Ontario et du sud-ouest du Québec. Étant donné la difficulté que présentel’affectation de larves d’hydrachnes a des espèces précises sur la base de la morphologie, nous avons choisi d’amplifier lefragment code-barres de la sous-unité I de la cytochrome oxydase pour explorer les gammes d’hôtes. Quinze unités tax-onomiques opérationnelles ou clades ont été identifiés sur la base de l’amplification a partir de 217 larves d’hydrachnes. Lesclades d’Arrenurus présents dans les habitats tant de tourbière que de marais avaient de plus grandes gammes d’espèces hôtes queles clades présents uniquement dans les marais (l’efficacité statistique de la comparaison avec un clade présent seulement dansdes tourbières était faible). Comme prévu, la gamme d’espèces hôtes augmentait en fonction de la présence régionale d’un claded’Arrenurus. En outre, les espèces les plus fréquemment codées présentaient également de grandes gammes d’espèces hôtes. Celapourrait être dû au fait que les espèces présentant les plus larges gammes d’hôtes sont plus répandues et étaient les plussusceptibles d’être échantillonnées et codées (une explication que nous privilégions) ou a un biais d’échantillonnage. Bien qu’ils’agisse de la première étude a tenter de déterminer si la gamme d’habitats a une incidence sur la gamme d’hôtes, d’autrestravaux sont nécessaires pour départager les facteurs de l’habitat qui exercent la plus grande influence sur les gammes d’espèceshôtes. [Traduit par la Rédaction]

Mots-clés : associations hôte–parasite, odonates, coenagrionidés, arrénuridés, gamme d’hôtes.

IntroductionMany factors are thought to influence host specificity of para-

sites (Poulin 2007). Ecological factors are those that presumablyaffect encounter between the host and the parasite. These includespatial and temporal overlap between host and parasite and hostabundance (Vazquez et al. 2005). If the parasite is not present inthe same habitat or at the same time as a particular host species orwhen that species is abundant, then that parasite will not be ableto infect that host species effectively or at all (Poulin 2007). Otherbiological factors presumably include host defenses, parasite ex-ploitation, and transmission mechanisms that might be subject tophylogenetic constraints (Edwards and Vidrine 2006; Mouillot

et al. 2006) and that might influence local adaptation (Lajeunesseand Forbes 2002).

A nonbiological factor that might affect parasite–host rangesmay be sampling effort. With increased collecting, a parasite spe-cies will likely be seen to have a greater host-species range than ifsampling was incidental (Poulin 1992). Sampling bias has beensuggested in many study systems ranging from mistletoes inAustralia (Grenfell and Burns 2009) to macroparasites of fishes inthe Mediterranean (Sasal et al. 1998). In other systems, some par-asites are often present in low numbers and might not be infect-ing many host species at particular sites, but still are present atmany sites on specific host species (Kennedy 2012). Therefore,spatial variables such as regional occurrence and (or) specificity ofhabitat use could potentially explain host-species ranges.

Received 3 April 2015. Accepted 24 June 2015.

J.J. Mlynarek and M.R. Forbes. Department of Biology, Carleton University, Nesbitt Building, 1125 Colonel By Drive, Ottawa, ON K1S 5B6, Canada.W. Knee. Canadian National Collection of Insects, Arachnids and Nematodes, Agriculture and Agri-Food Canada, K.W. Neatby Building, 960 CarlingAvenue, Ottawa, ON K1A 0C6, Canada.B.P. Smith. Department of Biology, Ithaca College, 953 Danby Road, Ithaca, NY 14850, USA.Corresponding author: Julia J. Mlynarek (e-mail: Julia.mlynarek@unb.ca).

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

1

Can. J. Zool. 93: 1–6 (2015) dx.doi.org/10.1139/cjz-2015-0077 Published at www.nrcresearchpress.com/cjz on 23 July 2015.

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Agr

icul

ture

and

Agr

i-fo

od C

anad

a on

09/

23/1

5Fo

r pe

rson

al u

se o

nly.

We studied the host–parasite associations between ArrenurusDugès, 1834 (Arrenuridae) water mites, the most common ecto-parasites of odonates (Corbet 1999), and their coenagrionid dam-selfly hosts. Arrenurus water mites are parasitic only during thelarval stage of their life cycle; the remainder of the time they areeither aquatic predators or have quiescent developmental stagesin the water column (Smith et al. 2010). Because Arrenurus mitespecies are only associated with a host during one of its life-history stages (i.e., the larva), they need to be adapted to theiraquatic habitat during their free-living stages, which make this amodel system to test for relationships between habitat and host-species ranges.

We tested for an association between water mite host-speciesrange and their general and specific habitat use by relating thehost-species range with the habitat preferences (marsh or bog) ofa given Arrenurus clade. We also tested whether the extent of theclade’s regional occurrence (number of sites where the water mitewas present, irrespective of wetland type) was associated withhost-species range. Not all odonates are widespread across a diver-sity of habitats, some are restricted to certain wetland types; thereis, however, also some overlap in odonate communities in bogsand marshes (Corbet 1999). We expect that there would be a pos-itive correlation between host-species range and habitat breadth,as well as between host-species range and regional occurrence.

Materials and methods

Sampling and identificationDamselfly specimens representing 14 coenagrionid species

were collected in 13 sites (6 bogs and 7 marshes) along a 400 kmlongitudinal transect in southeastern Ontario to southwesternQuebec (Table 1; see Mlynarek et al. 2014a). Collections were doneusing aerial sweep nets on a weekly basis for 8 weeks in June andJuly 2011. During each site visit, up to 30 individuals of each dam-selfly species were collected (following Mlynarek et al. 2014a). Allthe damselflies were stored individually in 95% ethanol. In thelaboratory, damselflies were reidentified to species. Each damsel-fly specimen was examined for water mites using a ZEISS SteREODiscovery V8 dissecting microscope. All mites were counted, col-lected, and stored in 95% ethanol.

A chosen subset of mites were sent to the Barcode of Life data-base (BOLD; available from http://www.boldsystems.org/, accessed10 December 2014) (Ratnasingham and Hebert 2007), CanadianCentre for DNA Barcoding (CCDB; available from http://www.ccdb.ca/, accessed 10 December 2014) at the Biodiversity Institute ofOntario for sequencing of the cytochrome oxidase subunit I (COI)using standard high-throughput methods (Ivanova et al. 2006).

DNA was extracted from 285 randomly selected larval watermites. This random selection was stratified as follows: if present,10 mites from every host species at every site were collected. Eachmite was chosen randomly from a different host individual froma given species if there were enough infected individuals of thatspecies. Eleven damselfly species had thoracic mites and eightdamselfly species had both thoracic and abdominal mites. Inthose latter species, five thoracic and five abdominal water miteswere collected because thoracic and abdominal mites are oftendifferent species (Mitchell 1964).

Contiguous sequences were edited and assembled from chro-matograms using SEQUENCHER version 4.7 (Gene Codes, AnnArbor, Michigan, USA). COI sequences were aligned manually inMESQUITE version 2.75 (Maddison and Maddison 2011) accordingto the translated amino acid sequence. Sequences are deposited inGenBank (KJ709142–KJ709343; Supplementary Table S11). Homol-ogous sequences from four Arrenurus water mite individuals col-lected from sphagnum sprite (Nehalennia gracilis Morse, 1895) andsedge sprite (Nehalennia irene (Hagen, 1861)) damselflies (Mlynareket al. 2014b) and four marsh bluet (Enallagma ebrium (Hagen, 1861))damselflies (Mlynarek et al. 2013) were included in the COI align-ment, and seven water mite species were selected from GenBankto serve as outgroups (AB530314, JX838402, JX836526, JX835088,JN018105, JN018109, AB530311). Pairwise distances were calculatedusing neighbour-joining analysis with the uncorrected (“p”) model inPAUP* version 4.0b10 (Swofford 2003).

Phylogenetic analysis of the COI data set was performed usingmaximum-likelihood inference in PhyML version 3.0 (Guindonand Gascuel 2003). The best-fit model of molecular evolutionwas determined to be GTR+I+G using MrModeltest version 2.3(Nylander 2004). Branch support was estimated using nonpara-metric bootstrap with 1000 replicates.

Habitat and regional occurrenceWe collected at two different site types (bogs and marshes) to

increase the number of different hosts species that were included,as this would allow breadth in host-species range values. Both ofthese wetland types are lentic water bodies with different wateracidity, which could have an effect on the survival of immatureand adult water mites. If a water mite clade is present at both sitetypes, then we can expect that its free-living aquatic stages canwithstand a wider range of habitats than those only occurring atone site type (e.g., a bog specialist may only be able to survive in anacidic environment).

Regional occurrence of a water mite species was simply a countof the number of sites a water mite clade or “species” was col-

1Supplementary Table S1 is available with the article through the journal Web site at http://nrcresearchpress.com/doi/suppl/10.1139/cjz-2015-0077.

Table 1. List of study sites in southeastern Ontario and southwestern Quebec to determinewater mite (genus Arrenurus) host-species ranges.

Site Site type Latitude Longitude

Johnville, Quebec Bog 45°20=38.67==N 71°44=18.17==WMarlington, Quebec Bog 45°02=44.39==N 72°12=16.98==WLarge Teafield, Quebec Bog 45°08=04.02==W 74°13=00.47==WNewington, Ontario Bog 45°07=21.68==N 74°57=55.27==WHebert, Ontario Bog 44°29=54.69==N 76°24=53.66==WWestport, Ontario Bog 44°41=57.69==N 76°22=46.55==WLac aux Cerises, Quebec Marsh 45°16=51.00==N 72°10=09.34==WCooper, Ontario Marsh 45°06=59.04==N 74°30=51.59==WPointe Leblanc, Quebec Marsh 45°02=12.18==N 74°27=56.70==WStony, Ontario Marsh 45°17=33.03==N 75°48=25.29==WJack, Ontario Marsh 44°32=01.82==W 76°22=35.31==WBarb, Ontario Marsh 44°31=27.54==N 76°22=25.89==WOsprey, Ontario Marsh 44°30=25.73==N 76°24=38.08==W

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

2 Can. J. Zool. Vol. 93, 2015

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Agr

icul

ture

and

Agr

i-fo

od C

anad

a on

09/

23/1

5Fo

r pe

rson

al u

se o

nly.

lected at within our sampling season. Because we collected at13 sites, the maximum regional occurrence a species can have is13. Regional occurrence does not differentiate between marshesand bogs; it accounts for all sites.

Statistical analysesTo determine if there was potential for a sampling bias, we

tested whether there was a relationship between the number ofhost species and the number of mites of a given clade that werebarcoded, determined using a Spearman’s correlation (rS). Follow-ing this analysis, we tested for an effect of habitat use (marsh only,bog only, both bogs and marshes) on host-species range using aKruskal–Wallis H test, which was most appropriate because of thesmall sample sizes. If there was a significant difference, we thenperformed a Mann–Whitney U test to determine between whichpairs of comparisons were significantly different. Finally, a secondSpearman’s correlation was performed to determine the relation-ship between regional occurrence of a water mite clade and thenumber of recorded host species. All three statistical tests wereperformed in JMP version 11 (SAS Institute, Inc. 2013).

Results

Arrenurus sampling and identificationA total of 7445 Arrenurus water mites were collected from

7107 damselflies across 13 sites. Eleven of 14 damselfly species(taiga bluet, Coenagrion resolutum (Hagen in Selys, 1876); azurebluet, Enallagma aspersum (Hagen, 1861); boreal bluet, Enallagmaboreale Selys, 1875; common bluet, Enallagma cyathigerum (Char-pentier, 1840); E. ebrium; Hagen’s bluet, Enallagma hageni (Walsh,1863); orange bluet, Enallagma signatum (Hagen, 1861); fragile fork-tail, Ischnura posita (Hagen, 1861); eastern forktail, Ischnura verticalis(Say, 1839); N. gracilis; N. irene) were infected by at least one Arrenu-rus water mite individual from 1 of the 13 sites.

A 683 base pairs (bp) fragment of COI was successfully amplifiedfrom 217 Arrenurus specimens. A total of 327 characters were par-simony informative and 356 were uninformative. The maximumlikelihood performed on the COI data set resulted in a well-supported evolutionary hypothesis (tree length (TL) = 1355, consis-tency index (CI) = 0.4, retention index (RI) = 0.93; Fig. 1) with mostnodes having moderate to high support. The ingroup was dividedinto 15 well-supported molecular operational taxonomic units(OTU) or clades (Fig. 1). Mean (±SD) interclade divergence was10.7% ± 5.8% and mean (±SD) intraclade divergence was 0.7% ± 0.5%(cf. Mlynarek et al. 2014b).

Habitat and regional occurrenceThere was a strong positive relationship between the number of

host species and number of mites of a given clade barcoded (Spear-man’s correlation, rS = 0.78, p < 0.001; Fig. 2). Water mites thatwere sampled and barcoded more often had a higher number ofhost species. Water mite clades collected at both bogs andmarshes had a broader host-species range (Kruskal–Wallis H test,H = 8.23, df = 2, p = 0.016; Fig. 3). From the Mann–Whitney U tests,there was no difference between host-species range and numbersof mites found only in bogs or only in marshes (U = 0.35, df = 1,p = 0.72) or between host-species range of mites only in bogs andhost-species range for mites in both bogs and marshes (U = 1.49,df = 1, p = 0.14). This latter result is likely due to the small samplesize, as there was only one mite clade that was found only in bogs.There was a significant difference between host-species range ofmites found only in marshes and host-species range of mitesfound in both bogs and marshes (U = 2.54, df = 1, p = 0.011), with thelatter having a significantly greater host-species range.

There was a strong positive relationship between regional oc-currence of a water mite clade and its host-species range (Spear-man’s correlation, rS = 0.84, p < 0.001; Fig. 4). Water mite cladesthat were collected at more sites had broader host-species ranges,although there were cases of clades with small sample sizes being

found on a surprisingly high number of host species for thatsample size (Table 2). Similar exceptions were found with respectto regional occurrence: for example, repeated instances of a claderepresented by two individuals collected at two sites (Table 2).

DiscussionOur main objectives were to determine whether Arrenurus

clades collected only from marshes or from bogs had greater hostspecificity (i.e., narrower host-species range) than those mitesfound in both habitat types, as well as to explore how regionaloccurrence of a water mite clade related to its number of hostspecies exploited. Both variables are correlated with host-speciesrange: as the regional occurrence of a water mite species in-creased so did its host-species range. Furthermore, water mitesthat were present in both marshes and bogs had broader host-species ranges than if they were only present at marshes.

There are issues with this type of correlational study in thatcertain clades or species may be regionally less well representedin samples than they actually are in nature. We expect a positiverelationship between the number of mites barcoded from a givenclade and the regional occurrence or host-species range, becausewith more mites, there is a greater likelihood for them to besampled from different sites (habitats) or different host species.However, this problem is not easily decoupled from the fact thatmites with broader host-species ranges or broader regional occur-rences will be sampled more and randomly chosen for barcodingmore often. In brief, the sample size for a given clade or speciescould reflect something biological: that is, how well representedin nature that clade is. We think that this is the case for our databecause mites were chosen for barcoding using stratified randomsampling based on host species. That is, we ensured that mitesfrom all host species were represented to increase the chances ofdocumenting all or most host–parasite relations. We also foundthat clades represented by even small samples can still have rela-tively broad host-species ranges for that sample size, using thisapproach (Table 2). We could not test explicitly whether there wasa sampling bias, given that we constrained our sampling, for thereasons discussed above.

Our findings provide some support for the argument that if aparasite species is able to survive at more sites during its free-living stages, then it could come into contact with and parasitizemore potential host species. However, our sample sizes were rel-atively small, as there was only one clade found exclusively atbogs. As a group, Arrenurus mites are able to live in a variety offreshwater habitats and attack a broad variety damselfly anddragonfly hosts (Smith et al. 2010).

Even though this study encompasses a small regional area, it isone of the first studies exploring the drivers of host-species rangein Arrenurus water mites. The region used for our study is notablefor its high diversity of Arrenurus species, and it would be expectedthat many of the species found there would spill over into otherecozones (I. Smith, personal communication). Arrenurus watermites are interesting because of their natural history; they have tobe able to survive in a given environment during free-living stagesof their life cycle (Smith et al. 2010). Being both a parasite of aerialinsects (as a larva) and a free-living aquatic organism (during therest of its life cycle), they must be adapted to a diversity of condi-tions that change dramatically during the progression of their lifecycle.

Arrenurus may be selected to be generalist parasites becausedensity of single host species in the environment constrains thelikelihood of encountering a suitable host (Wolinska and King2009). The majority of the Arrenurus species in this study infectseveral host species spanning at least two coenagrionid damselflygenera, which overlap within freshwater wetlands. Being gener-alists makes them different from Unionicola Haldeman, 1842 mitesinfecting mussels (Edwards and Vidrine 2006). Unionicola mites

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

Mlynarek et al. 3

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Agr

icul

ture

and

Agr

i-fo

od C

anad

a on

09/

23/1

5Fo

r pe

rson

al u

se o

nly.

Fig. 1. Maximum-likelihood evolutionary tree of 683 base pair (bp) fragment of cytochrome oxidase subunit I (COI) from 224 water mites:217 ingroup and 7 outgroup specimens (tree length (TL) = 1355, consistency index (CI) = 0.4, retention index (RI) = 0.928). Scale bar =0.1 substitutions per site. Bootstrap percentages are shown above the nodes.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

4 Can. J. Zool. Vol. 93, 2015

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Agr

icul

ture

and

Agr

i-fo

od C

anad

a on

09/

23/1

5Fo

r pe

rson

al u

se o

nly.

tend to be specific to one or two host species within the samegenus (Edwards and Vidrine 2006). Like Unionicola, four clades(OTUs 2, 3, 6, and 13) were collected from only one host species.However, in each of these clades, few specimens were collected:

two of OTU 13, seven of OTU 6, one of OTU 3, and three of OTU 2.These species either have narrow host-species ranges or there wassome sampling error. In our case, we did not know which speciesof water mites we were sampling as larvae, i.e., we did not know apriori whether the water mites were host specific. However, wesampled host damselflies in a stratified and standardized way, sowe can suggest that the numbers of water mites collected andbarcoded can be considered a proxy of presence or absence acrossa range of sites and habitats.

AcknowledgementsWe thank A. Morrill for help with the collecting and reading a

draft of the manuscript, and Nature Conservancy, L’Associationdu Marais-de-la-Rivière-aux-Cerises (LAMRAC), Queen’s UniversityBiological Station (QUBS), and the National Capital Commission(NCC) for collecting permits. This research was funded by a Natu-ral Sciences and Engineering Research Council of Canada (NSERC)Discovery Grant awarded to M.R.F. and an NSERC Canada Gradu-ate Scholarship (CGS) granted to J.J.M.

ReferencesCorbet, P.S. 1999. Dragonflies: behavior and ecology of Odonata. Comstock Pub-

lishing Associates, Ithaca, N.Y.Edwards, D.D., and Vidrine, M.F. 2006. Host specificity among Unionicola spp.

(Acari: Unionicolidae) parasitizing freshwater mussels. J. Parasitol. 92(5):977–983. doi:10.1645/GE-3565.1. PMID:17152937.

Grenfell, M., and Burns, K.C. 2009. Sampling effects and host ranges in Australianmistletoes. Biotropica, 41(6): 656–658. doi:10.1111/j.1744-7429.2009.00586.x.

Guindon, S., and Gascuel, O. 2003. A simple, fast, and accurate algorithm toestimate large phylogenies by maximum likelihood. Syst. Biol. 52(5): 696–704. doi:10.1080/10635150390235520. PMID:14530136.

Ivanova, N.V., Dewaard, J.R., and Hebert, P.D.N. 2006. An inexpensive,automation-friendly protocol for recovering high-quality DNA. Mol. Ecol.Notes, 6(4): 998–1002. doi:10.1111/j.1471-8286.2006.01428.x.

Kennedy, C.R. 2012. Can a specialist parasite species of a widespread and com-mon host species be rare? The case of Spinitectus inermis (Nematoda: Cystidi-colidae) in eels Anguilla anguilla. Folia Parasitol. 59(2): 131–138. doi:10.14411/fp.2012.018. PMID:22779113.

Lajeunesse, M., and Forbes, M. 2002. Host range and local parasite adaptation.Proc. R. Soc. B Biol. Sci. 269(1492): 703–710. doi:10.1098/rspb.2001.1943.

Maddison, W.P., and Maddison, D.R. 2011. Mesquite: a modular system for evo-lutionary analysis. Version 2.75 [computer program]. Available from http://mesquiteproject.org [accessed 15 December 2014].

Mitchell, R. 1964. Study of sympatry in water mite genus Arrenurus (family Arre-nuridae). Ecology, 45(3): 546–558. doi:10.2307/1936107.

Mlynarek, J.J., Knee, W., and Forbes, M.R. 2013. Relative geographic range ofsibling species of host damselflies does not reliably predict differential par-asitism by water mites. BMC Ecol. 13: 50. doi:10.1186/1472-6785-13-50. PMID:24351055.

Fig. 2. Correlation of sampling bias between the number ofcoenagrionid host species collected and the number of individualsof each Arrenurus water mite collected from 13 sites in southeasternOntario and southwestern Quebec. Numbers represent the Arrenurusoperational taxonomic units (OTU).

Fig. 3. Boxplot representing the habitat specialization of Arrenuruswater mites and their host-species ranges. Bog–Marsh represents theArrenurus water mites collected at both sites.

Fig. 4. Correlation of Arrenurus water mite host-species range versusArrenurus water mite regional occurrence from 13 sites insoutheastern Ontario and southwestern Quebec. Numbers representthe Arrenurus operational taxonomic units (OTU).

Table 2. Ecological and host-range parameters of the 15 Arrenuruswater mite operational taxonomic units (OTUs) that were collectedduring the reproductive season of 2011 from 13 sites in southeasternOntario and southwestern Quebec.

Arrenurus OTU NRegionaloccurrence Habitat

Host-speciesrange

1 18 5 B–M 22 3 1 B 13 1 1 M 14 24 7 B–M 35 4 2 M 26 7 3 M 17 5 3 B–M 28 57 11 B–M 79 8 4 B–M 410 36 9 B–M 411 2 2 M 212 2 1 M 113 23 7 B–M 214 2 2 B–M 215 25 8 B–M 4

Note: N is the number of mites; “habitat” is mite presence at either bog (B),marsh (M), or both (B–M).

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

Mlynarek et al. 5

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Agr

icul

ture

and

Agr

i-fo

od C

anad

a on

09/

23/1

5Fo

r pe

rson

al u

se o

nly.

Mlynarek, J.J., Knee, W., and Forbes, M.R. 2014a. Explaining susceptibilityand resistance to a multi-host parasite. Evol. Biol. 41(1): 115–122. doi:10.1007/s11692-013-9251-6.

Mlynarek, J.J., Knee, W., and Forbes, M.R. 2014b. Host phenology, geographicrange size and regional occurrence explain interspecific variation indamselfly–water mite associations. Ecography, 38(7): 670–680. doi:10.1111/ecog.00997.

Mouillot, D., Krasnov, B.R., Shenbrot, G.I., Gaston, K.J., and Poulin, R. 2006.Conservatism of host specificity in parasites. Ecography, 29(4): 596–602. doi:10.1111/j.0906-7590.2006.04507.x.

Nylander, J.A.A. 2004. MrModeltest. Version 2 [computer program]. Programdistributed by the author. Evolutionary Biology Centre, Uppsala University,Uppsala, Sweden.

Poulin, R. 1992. Determinants of host-specificity in parasites of freshwaterfishes. Int. J. Parasitol. 22(6): 753–758. doi:10.1016/0020-7519(92)90124-4. PMID:1428509.

Poulin, R. 2007. Evolutionary ecology of parasites. Princeton University Press,Princeton, N.J.

Ratnasingham, S., and Hebert, P.D.N. 2007. BOLD: The Barcode of Life Data

System (www.barcodinglife.org). Mol. Ecol. Notes, 7(3): 355–364. doi:10.1111/j.1471-8286.2007.01678.x.

SAS Institute, Inc. 2013. JMP. Version 11 [computer program]. SAS Institute, Inc.,Cary, N.C.

Sasal, P., Desdevises, Y., and Morand, S. 1998. Host-specialization and speciesdiversity in fish parasites: phylogenetic conservatism? Ecography, 21(6): 639–643. doi:10.1111/j.1600-0587.1998.tb00557.x.

Smith, I., Cook, D., and Smith, B. 2010. Water mites (Hydrachnidiae) and otherarachnids In Ecology and classification of North American freshwater inver-tebrates. 3rd ed. Edited by J. Thorp and A. Covich. Academic Press, London, UK.pp. 485–586.

Swofford, D.L. 2003. PAUP*: phylogenetic analysis using parsimony (*and othermethods). Version 4.0b10 [computer program]. Sinauer Associates, Inc., Sun-derland, Mass.

Vazquez, D.P., Poulin, R., Krasnov, B.R., and Shenbrot, G.I. 2005. Species abun-dance and the distribution of specialization in host–parasite interaction net-works. J. Anim. Ecol. 74(5): 946–955. doi:10.1111/j.1365-2656.2005.00992.x.

Wolinska, J., and King, K.C. 2009. Environment can alter selection in host–parasite interactions. Trends Parasitol. 25(5): 236–244. doi:10.1016/j.pt.2009.02.004. PMID:19356982.

Pagination not final (cite DOI) / Pagination provisoire (citer le DOI)

6 Can. J. Zool. Vol. 93, 2015

Published by NRC Research Press

Can

. J. Z

ool.

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Agr

icul

ture

and

Agr

i-fo

od C

anad

a on

09/

23/1

5Fo

r pe

rson

al u

se o

nly.