A gall mite, Aceria rhodiolae (Acari: Eriophyidae), altering the phytochemistry of a medicinal plant, Rhodiola rosea (Crassulaceae), in the Canadian Arctic

ABSTRACT The eriophyid mite Aceria rhodiolae (G. Canestrini) is known to induce galls on the flowers and leaves of roseroot, Rhodiola rosea L., in subarctic and alpine regions of Europe. After discovering galls on the inflorescences of roseroot in Nunavik (Québec), northeastern Canada, we examined the mites extracted from the galls and compared them with specimens of A. rhodiolae from Europe. Through morphological analyses, we demonstrate that the mites from galls in Nunavik are conspecific with A. rhodiolae from Europe. We then provide a detailed redescription of the mite species based on the morphology of adult females and males from Canada and Europe, using a combination of standard light microscopy, confocal microscopy and scanning electron microscopy. Because roseroot is well-known for its medicinal properties, we tested the hypothesis that roseroot galled by the mite had altered phytochemistry, by using salidroside and rosavins as indicators. Our results show a significant reduction of almost half in salidroside content (45.8%), but not in rosavins. Moreover, because the mite sometimes affects most or all of the inflorescence of R. rosea, it can considerably reduce the production of seeds. We also show that A. rhodiolae is widespread along the Ungava Bay (Nunavik), with 31.5% of 92 sites surveyed having at least a few to numerous plants galled. Given the importance of roseroot as a crop for Inuit communities and as medicinal products used by them and other Canadians, and also in view of the commonness of A. rhodiolae in the Canadian Arctic and its broad distribution in Europe, the impact of the mite and its relationship with roseroot should be examined further in Nunavik and elsewhere.


Introduction
Rhodiola rosea L. (Crassulaceae), also known as roseroot and goldenroot, is a highly valued medicinal plant species in the Old World, where it grows as a perennial plant in the Arctic and mountainous regions of Europe and Central Asia (Brown et al. 2002).
Records of traditional and modern uses include immunostimulation, as a remedy against fatigue, stress and memory problems, as well as for its mild antidepressant properties (Fida et al. 2014;Panossian and Wikman 2014). A recent study (Cayer et al. 2013) shows that extracts have significant anxiolytic activity in animal trials.
Although sometimes considered a subspecies (Rhodiola rosea ssp. rosea) somewhat distinct from Eurasian populations, R. rosea also grows in the eastern subarctic and a few mountain sites of lower latitudes in North America (Cuerrier et al. 2014a(Cuerrier et al. , 2014b. In Canada, R. rosea has a sporadic distribution along the shoreline of the eastern low Arctic and the Atlantic provinces (Small and Catling 2000;Aiken et al. 2007). Nunavik roseroot is genetically close to Scandinavian populations and contains the same phytochemical markers, namely salidroside, tyrosol, rosarin, rosavin and rosin, as seen in Eurasian samples. Yet, a new compound not found in Eurasia has been discovered in Nunavik, as well as differences in a number of analytes (Filion et al. 2008;Avula et al. 2009).
Several arthropod species are known to feed on roseroot. They include at least four species of aphids (Aphididae), two of which occur in North America (Blackman and Eastop 2006;Holman 2009); a caterpillar (Papillionidae) recorded from Asia, and from the Yukon where it feeds on Rhodiola integrifolia (previously a subspecies of R. rosea) (Layberry et al. 1998); two species of leaf-mining Diptera, an Agromyzidae (Europe; and the Yukon, on R. integrifolia) and a Syrphidae (Europe) (Griffiths 1976;Bland 1995;Schmid 2007); and a weevil and a bark beetle species (Curculionidae) boring the roots of R. rosea in Russia (Kuznetsova and Krivets 1981;Smetanin 2013). Most of these insects are host-specific to R. rosea or restricted to crassulaceous hosts. A phytophagous mite, Aceria rhodiolae (G. Canestrini, 1892) (Eriophyidae), is known to induce galls on the flowers, leaves and stems of R. rosea in Europe (e.g. Roivainen 1950;Boczek 1961;Buhr 1965). In Canada, no plant-feeding mite has yet been recorded from R. rosea. Clausen (1975) noted signs of deformations of roseroot in Canada, without providing any details of a diagnosis or the causative agent.
Eriophyid mites and their close relatives in the superfamily Eriophyoidea are strictly phytophagous, feeding and developing usually on a single or a few related plant species (de Lillo and Skoracka 2010). Although inconspicuous and minute (0.1-0.5 µm long), the feeding activity of many eriophyoid species results in a deformation of the plant host's tissues. The deformations, or galls, vary in shape from enclosed, pouch-like galls protruding from the leaves, to erinea (hair-like excrescences), leaf rolling, and various deformations of buds, inflorescences and even the bark (Keifer et al. 1982). There are over 4000 species of eriophyoids currently described worldwide, but the majority of species remain to be described, including in Canada (Lindquist et al. 1979;de Lillo and Skoracka 2010;Beaulieu and Knee 2014).
Herein, we report on the presence of an eriophyid mite galling the flowers of R. rosea in Nunavik, eastern Canada. First, we compare its morphology with European specimens of A. rhodiolae to test conspecificity, and provide a description of the mite, taking into account any possible differences between Canadian and European populations; and second, we test the hypothesis that the presence of A. rhodiolae has an impact on the medicinal phytochemistry of roseroot by comparing levels of salidroside and rosavins in galled and ungalled plants.

Material and methods
Sampling Populations of R. rosea were surveyed by V.F., A.C. and Mariannick Archambault in August 2006 and 2007 along the coastline of Ungava Bay, in Nunavik (Québec, Canada; Figure 1). Flower heads showing signs of deformation were collected and stored in 60% ethanol for future examination in the laboratory at the University of Ottawa, and the Canadian National Collection of Insects, Arachnids and Nematodes (CNC) (Ottawa, Canada). Global positioning system (GPS) points (± 5 m) were recorded for the surveyed sites. Additional samples of galled R. rosea were taken from Base Island, Labrador, Newfoundland, Canada. See Figure 2B and the 'Material examined' section for more collecting details. Voucher specimens of R. rosea have been deposited in the Marie-Victorin herbarium (MT) (Jardin botanique de Montréal) and determined by A.C.

Mite specimen preparation and examination
Some mites found among the flower buds were slide-mounted in Hoyer's medium, and the rest were preserved in a vial with 95% alcohol. Mite specimens were studied by F.B. at 400× and 1000× magnification under a compound microscope (Leica DM5500) equipped with differential interference contrast (DIC), connected to a computer and a digital camera (Leica DFC420). Images and morphological measurements were taken via Leica Application Suite software 4.2 (Basic, Live Measurements, and Interactive Measurements modules). Morphological terminology follows that of Lindquist (1996) and measurements were made according to Amrine and Manson   (1996), as modified by , notably for the following characters: legs were measured from distal margin of tarsus to the proximal margin of trochanter; empodia were measured from distal apex to their junction with the margin of the tarsal segment. Other measurements that may need clarification are: ventral opisthosomal annuli were counted from the posterolateral corner of coxal field II; length of prodorsal shield includes the frontal lobe; the position of some leg setae (bv, l', l") were measured from the proximal margin of the segment bearing the seta; the number of microtubercles between setae sc are counted as the microtubercles on the first complete annulus (the first one is often incomplete) behind the prodorsal shield; the number of microtubercles between pairs of setae c2, d, e and f are counted ventrally. Internal genitalia were essentially described using the morphometrics proposed by Chetverikov et al. (2012Chetverikov et al. ( , 2013 based on DIC light microscopy, and further interpreted using Chetverikov (2014) for females, and Chetverikov (2015) for males. Because specimens from Italy (see 'Material examined') were in poor shape (body twisted or deformed, and some setae cut or only partly discernible), the length of leg setae and the distance between opisthosomal setae for those specimens were considered unreliable and were excluded. Measurements shown in the description are ranges (in μm). Adult female ventrolateral view (scanning electron micrograph) and (C) dorsolateral view (differential interference contrast light microscopy). Scale on (B) also applies to (C). The large size of female in (C) is in part due to some flattening of that specimen during the slide-mounting process. ch, cheliceral stylets; and other notations indicate setae of prodorsal shield and opisthosoma.
Specimens from Nunavik were compared with (1) descriptions of A. rhodiolae and other species associated with Crassulaceae in the literature, as well as with: (2) specimens borrowed from the Zoology Museum of the University of Padua (MZUP) (see 'Material examined' for specimen collection details); (3) specimens extracted from dried inflorescences of R. rosea borrowed from the Finnish Natural History Museum, University of Helsinski (MZH); and (4) specimens of Aceria destructor (Nalepa, 1891) collected from Sedum sp. (Crassulaceae). We also attempted to obtain specimens of A. rhodiolae previously collected by Jan Boczek in Poland (Boczek 1961) but without success (Mariusz Lewandowski pers. comm. August 2012). Mites were extracted from dried R. rosea (from MZH; and collected from Labrador, Canada) by removing and placing a part of the inflorescence in 95% alcohol. Mite specimens freed from plant tissues and floating in the alcohol were then picked up and slide-mounted.
External morphology was further studied using scanning electron microscopy (SEM) (Philips XL30), at the Microscopy Centre at Agriculture and Agri-Food Canada (AAFC), Science & Technology Branch (Ottawa). Specimens already stored in 95% alcohol were transferred with a pipette into a microporous capsule (30-μm pores) partly immersed in 100% alcohol in a small Petri dish, for 15 min; they were subsequently transferred to another Petri dish with 100% alcohol for another 15 min to ensure that specimens were effectively submerged in undiluted 100% alcohol before undergoing critical-point drying. Specimens were then mounted on SEM specimen stubs using a small paint brush with a few remaining hairs, and sputter-coated with gold before examination under SEM.
Additional imaging of the female prodorsal region and the male internal genitalia was obtained using a confocal light scanning microscope (CLSM) (Zeiss LSM 510) at AAFC, under 40× magnification (1.4/oil apochromatic objective), and with the following settings (Chetverikov 2012): excitation wavelength 405 nm, emission wavelength range 420-750 nm. Images were acquired at a resolution of 1024 × 1024 pixels and an electronic zoom of 3×, using ZEN lite 2012 software. In some cases, two or more images taken at different depths of a specimen were merged into a single image using Helicon Focus 5.3.14 (© Helicon Soft Ltd., 2000Ltd., -2013. Illustrations (line drawings) of morphological features were prepared using Adobe Illustrator version 15.0.0 (Adobe Systems Inc.). Selected digital photos were first imported into Adobe Illustrator, and lines were traced over the structures of interest. Other images, including from light microscopy, SEM and CLSM, were modified using Photoshop CS5 version 12.0 (Adobe Systems Inc.) to prepare plates and to improve clarity of features.

Material examined
All specimens of A. rhodiola from roseroot, R. rosea. Canada: along shore of Ungava Bay, Nunavik, Quebec: 9 adult females, 1 August 2007, coll. V. Filion; Base Island, Labrador, Newfoundland: 7 adult females, 2 adult males, 4 August 2012, 56.633°N, 61.586°W, coll. A. Cuerrier; Italy: Verona: 4 adult females and 1 male previously preserved in vials labelled as 'Phytoptus rhodiolae Can.' themselves stored within a jar numbered CXXXIX (and # 608 in Valle 1955), coll. (probably) G. Canestrini over 100 years ago (MZUP); Russia: Pechengsky District (formerly Petsamo, Finland, and indicated as such on the herbarium sample): 11 adult females and 1 male extracted from a single flower specimen of R. rosea (number: MZH 115326) mounted on a herbarium sheet, collected by I. Frosius (the surname is difficult to read and may be misspelled here), probably in the 1920s (as per communication with Juhani Terhivuo, MZH). Comparative material: 8 adult females labelled as 'Phytoptus destructor (Nal.)', stored in jar # CXXXV (# 594 in Valle 1955), coll. from Sedum sp. at an unknown locality, (probably) by G. Canestrini over 100 years ago (MZUP). Kept at the CNC: slide-mounted and 95% alcohol-preserved specimens of A. rhodiolae and two dried, galled R. rosea plants, collected from Canada; slide-mounted A. rhodiolae specimens that were extracted from R. rosea flowers, and a single dried, galled flower, from Russia; and three specimens each of A. rhodiolae and A. destructor obtained from MZUP. Other material was returned to MZUP (slides, vials) and MZH (herbarium sheet).

Phytochemical analysis
The rhizomes of healthy (n = 94) and deformed (n = 60) plants were compared (by V.F. and A.S.) for salidroside and rosavins content. Each measure was made in triplicates and resulting values were averaged before statistical analysis. The rhizomes were used because their concentration of salidroside is higher than in seeds, and much higher than in leaves or stems (Filion et al. 2008). We use the same common laboratory extraction method as described in Filion et al. (2008), using 90% ethanol. Again, the high-performance liquid chromatography with diode-array detection analysis was used to decipher the variation of phytochemical contents among samples (see Filion et al. 2008;Avula et al. 2009). An unpaired t-test with Welch's correction (due to populations with possible unequal variances) was used to evaluate differences.

Diagnosis and similar species
Adults can be distinguished from other Aceria species by the following combination of characters. Only one female form is known. Prodorsal shield 37-44 long, including apically a small frontal lobe, pointed or narrowly rounded; Setae sc 27-35 apart; first pair of submedian lines reaching posterior third of shield, where they curve inwards and then outwards (curved portion sometimes interrupted into short lines); submedian I usually branching posteriorly before its curving, roughly forming a broad, reversed 'Y' facing sc tubercles; many short ridges or nodules scattered posterolaterad submedian I, and more densely scattered laterad submedian II. Empodium with four pairs of rays. Coxal plates with conspicuous rounded ridge(s) surrounding medially setae 1a tubercles, and diagonal ridges laterad tubercles of setae 1b. Prosternal apodeme strong, slightly broadened posteriorly. Epigynial coverflap 12-15 long × 23-27 wide with 8-12 longitudinal ridges; a pair of small, rounded lateral flaps flanking the coverflap. Opisthosoma covered by pointed microtubercles throughout, slightly larger dorsally than ventrally. Setae c2 36-51, d 44-68, e 15-21, f 26-40.
Aceria destructor may be a close relative, in part based on similarity in prodorsal shield pattern, particularly submedian line I and the ridge laterally branching from it ( Figure 3C, 'a') (Nalepa 1891;however, illustration in Farkas (1965) differs). Aceria rhodiolae differs from A. destructor by (Table 2): generally shorter sc setae; fewer opisthosomal annuli, and accordingly, by setae d-f being inserted on different annuli number; a transversally narrower genital coverflap that bears fewer longitudinal ridges, with a few that are usually broken or abbreviated; posteromedian region of prodorsal shield, between sc tubercles, smooth or with few rather indistinct lineae (  (1891); not in Farkas (1965)].
Aceria stinsonis (Keifer 1939) is moderately similar to A. rhodiolae. Based on comparison with description in the literature (Keifer 1939), A. rhodiolae may be primarily differentiated from A. stinsonis by: its longer sc; prodorsal shield with submedian I branched posteriorly; and coxisternal region ornamented with several nodules, and ridge(s) mesad seta 1a and laterad 1b (only a few scattered nodules for A. stinsonis; see also Table 2).

Taxonomic remarks
The specimens that we borrowed from the Canestrini collection had been preserved in vials (containing an alcohol-based fluid) labelled as from Verona, Italy, the presumed type locality for A. rhodiolae (Canestrini 1892: 'place of origin, Veronese'; note, however, that Amrine and Stasny (1994) mentioned 'woods near Trentino, Italy' as the type locality). Unfortunately, the specimens that Canestrini used to describe 'Phytoptus' rhodiolae (syntypes) are probably lost, because no slides labelled as P. rhodiolae were retrieved from his collection (now hosted by the MZUP; Paola Nicolosi pers. comm.).
Canestrini's description (1892) of A. rhodiolae is the only one that includes morphological illustrations. His illustrations of the ventral habitus and prodorsal shield partly agree with our observations; they show 58 ventral opisthosomal annuli (61-68 for our specimens), 12 longitudinal ridges on the coverflap (8-12 on our specimens), and sc setae about two-thirds the length of the dorsal shield (his text says sc at least as long as the shield, which is more concordant with our observations). The text mentions 'approximately 60 annuli', which is, again, near the lower end of the range of our count of dorsal annuli (66-75). More importantly, his illustration of the prodorsal shield shows differences (no nodules or ridges other than the main median, admedian and submedian lines, and the two pairs of submedian lines show different paths) from the specimens we examined, and only three pairs of rays can be seen on all the four empodia illustrated (the text also mentions three pairs of rays only). This may in part be due to the suboptimal microscope qualities and standards of taxonomic descriptions of that period. Roivainen (1950) mentioned that A. rhodiolae from Sweden had more opisthosomal annuli than specimens studied by Canestrini (1892), with 70-75 opisthosomal annuli, which is more consistent with our results. The few other descriptions of A. rhodiolae (Nalepa 1898(Nalepa , 1911Liro and Roivainen 1951;Farkas 1965) appear as a subset of, or equivalent to Canestrini's description (e.g. they all mention '60 annuli', and three-rayed featherclaws), with no additional information. We consider that three-rayed empodia is probably a mistake made by Canestrini, and which was duplicated by other authors.

Galls
Galls were observed primarily on the female fruiting inflorescences of R. rosea (= Sedum rosea (L.) Scop., = Sedum rhodiola DC.) and occasionally on the upper leaves surrounding the infructescence (Figure 8). Galled tissues turned fleshy, wrinkled, and whitish or yellowish green ( Figure 8B,C). The galled flowers were patchily distributed within inflorescences ( Figure 8C,D), and sometimes comprised most or all of the inflorescence, giving a cauliflower-like appearance ( Figure 8B).

Local distribution
Our survey has found galled R. rosea at many sites scattered along the shore of Ungava Bay, from the extreme northwest point (near Quaqtaq: 61.047°N, 69.634°W) to near the northeast extreme of the surveyed area (south of Killiniq: 60.359°N, 64.850°W) and as far south as Kuujjuaq (58.148°N, 68.336°W) ( Figure 1B). From 92 sites studied in Nunavik, 29 (31.5%) had at least a few (often numerous) galled individuals of R. rosea, and the remaining 63 had apparently no infested individuals. In addition, we have observed galled R. rosea in Labrador, at multiple sites on Base island (56.633°N, 61.586°W) and in the vicinity of Saglek Fjord (58.51°N, 63.25°W), Nain (56.54°N, 61.70°W) and Rigolet (54.18°N, 58.44°W).

Discussion
Our search through the literature and databases (e.g. Zoological Records; Amrine and Stasny 1994; J. Amrine and E. de Lillo unpubl. database of world eriophyoid species, pers. comm.) indicate that A. rhodiolae is the only current valid eriophyoid species recorded from the plant genus Rhodiola. An Eriophyes sp. (considering current concepts, this mite could actually belong to a genus other than Eriophyes, such as Aceria) was reported from reddish deformed flower heads of 'red orpine' (Clementsia) in North America, probably USA (Felt 1940). The red orpine mentioned may actually be Rhodiola rhodantha A. Gray (H. Jacobsen) (=Clementsia rhodantha), a species thriving in the western (mountainous) USA. Rhodiola rosea was reported as the host plant of Phytoptus eucricotes Nalepa, by Canestrini (1892: 706), but this was a mistake that he corrected in the same publication (1892: 721; P. eucricotes lives on Lycium europaeum L.). Rhodiola is a relatively small plant genus (with 90 species), and used to be considered within the larger, more broadly defined genus Sedum (stonecrops), which comprises 420 species (Stevens 2001 onwards). Two eriophyids are recorded from Sedum, in Europe: Aceria destructor and Cecidophyes glaber (Nalepa), both causing deformation of the buds and flowers of Sedum reflexum L. and other Sedum spp. (Alta and Docters van Leeuwen 1946;Petanović and Stanković 1999). Considering the entire family Crassulaceae as potential host plants, only two additional eriophyoid species are known, both collected from the leaves of sand lettuce [Dudleya caespitosa (Haw.) Britt. and Rose] in California: Aceria stinsonis and Aculus cotyledonis (Keifer 1939). This represents a total of five eriophyids associated with Crassulaceae worldwidea rather humble tally considering that approximately 1400 species from 34 genera belong to that plant family (Stevens 2001 onwards).

Gall types and plant organs affected
Previous publications on A. rhodiolae are concordant with our observations, and describe the galls it induces as fleshy, wart-like outgrowths, with inflorescences deformed into a fleshy, frizzy ball-like mass that are reminiscent of small cauliflowers (Ross and Hedicke 1927;Liro and Roivainen 1951;Buhr 1965). Galled tissues are reported as yellowish, reddish or violet (Löw 1881;Kari 1936;Moesz 1938;Liro and Roivaninen 1951). The stems of R. rosea can also be galled by A. rhodiolae (Ross and Hedicke 1927;Wahlgren 1948;Buhr 1965), but more rarely than flowers and leaves (Kari 1936). Interestingly, the original description by Canestrini (1892) and the text of Boczek (1961) mention only leaves as the plant organ affected by galling, not the flowers. However, even Nalepa (1893Nalepa ( , 1898Nalepa ( , 1911, soon after the publication of Canestrini's description, mentioned both leaves and flowers as being galled. Given that galled inflorescences of R. rosea are conspicuous, it is difficult to conceive that both Canestrini and Boczek overlooked the galled inflorescences. Boczek (1961) made his observations on 8 August 1957, at which time healthy flowers would have already bloomed into fruits (R. rosea blooms in late June to early July in Nunavik and Newfoundland; Cuerrier and Hermanutz 2012). So, it may be that flowers were not severely infested in the sites that he visited, leaving him overlooking galled inflorescences. It is also possible that the differential galling of plant organs is affected by the local climate, soil, or variations in physiology of the populations of R. rosea or of the mite. Boczek (1961) mentioned that A. rhodiolae induces galls especially along the margins of leaves. The illustration by Liro (1941), which, before our study, constituted the only adequate visual representation of galling by A. rhodiolae, somewhat agrees with Boczek's point, showing most of the leaf galled tissues in clusters, near or along the leaf margins, although galls often cover a large portion of the leaf.

Geographic distribution
Observations of Roivainen (1950) and Liro and Roivainen (1951) indicate that in northern Sweden and Finland, and the nearby region of Russia, roseroot plants are commonly galled by A. rhodiolae, and that high infestation rates (up to 60-100%) can be seen locally (Liro and Roivainen 1951), which is even higher than what we observed in Nunavik.
Aceria rhodiolae now appears Holarctic, with the Canadian Arctic as the only or main records west of the Atlantic Ocean, and numerous records east of the Atlantic, in Italy (Canestrini 1892(Canestrini , 1894; Germany (Nalepa 1893(Nalepa , 1898(Nalepa , 1911; Tatra mountains overlapping Poland (Boczek 1961) and Slovakia (Magas-Tátra mountains ;Szépligeti 1890;Baudyš 1938;Moesz 1938); Austria (Dürrenstein mountain, near Lunz; Löw 1881); Norway (Leatherdale 1959); northern parts of Sweden (Julin 1936;Roivainen 1950) and Finland (Enontekiö (Lapland); Liro 1941), and the nearby region of Russia (formerly Finland, Petsamo region; Kari 1936;Liro and Roivaninen 1951; this study). The species therefore appears relatively widespread in Europe (Buhr 1965), andWahlgren (1948) also mentions the species as present in Iceland, Greenland, Scotland and the Swiss Alps (although without providing references for such records). It is probable that the distribution of the mite largely follows that of its host plant, in subarctic and alpine regions, and therefore may occur as far east as the mountains of Central Asia (for instance, Tien-Shan and Himalaya; Small and Catling 2000), and as far west and south as where R. rosea occurs in western Nunavut and in North Carolina (Roan Mountain) in North America, respectively. However, the range of R. rosea is shrinking in the south and it is considered endangered or threatened in southeastern USA (Plant Industry Division 1998;Nongame and Natural Heritage Program 1999).

Impact on phytochemistry
Salidroside is a phenylethanol derivative with antidepressant properties (Brown et al. 2002;Kurkin 2003;Tolonen et al. 2004;Panossian and Wikman 2014). Therefore, the floral tissue deformation by eriophyid mites not only has a negative influence on the reproductive capability of R. rosea plants through the sterilization of flowers, but also significantly decreases the medicinal quality of the plants.
This may be the first time that a mite is shown to have an adverse effect on the phytochemistry of a medicinal plant. However, such findings are not surprising, given the wide range of morphological, biochemical and physiological effects of both galling and non-galling eriophyoid mites observed so far on various host plants (see reviews by Kielkiewicz 2010a, 2010b). The distortion of plant tissues is itself linked to multiple processes affecting the directly injured epidermal cells, as well as neighbouring cells, the entire leaf and sometimes neighbouring leaves. In addition to having plant growth regulatory effects involved in the formation of galls (Royalty and Perring 1996;de Lillo and Monfreda 2004), the salivary compounds injected by the mites can also affect other hormones or enzymes involved in broader processes, including the indirect defence of plants against herbivores (Petanović and Kielkiewicz 2010a;Samsone et al. 2012). Eriophyoids, including galling and vagrant (i.e. non-galling) forms, may also substantially compromise the leaf gas exchange and photosynthesis of not only the leaf that is fed on, but also those of ungalled, neighbouring leaves (Royalty and Perring 1989;Larson 1998;Samsone et al. 2012;Patankar et al. 2013).
Several eriophyoid species are known to transmit plant viruses, some of which have a considerable impact on the plant host (e.g. Navia et al. 2013), and others enhance the establishment of fungal pathogens, and vice versa, thereby increasing disease severity (Gamliel-Atinsky et al. 2010). It is therefore not impossible that some of the changes in a plant's physiology and chemistry (e.g. overall drop in salidroside levels) may be in part due to or exacerbated by pathogens that are transmitted by eriophyoid mites.

Deutogynes and overwintering
It is unknown whether A. rhodiolae has an additional female form (deutogyne). A few Aceria species associated with herbaceous perennial plants have both protogyne and deutogyne females, such as Aceria anthocoptes (Nalepa) on Canada thistle (Petanović et al. 1997) and Aceria chondrillae (G. Canestrini) on skeleton weed (Krantz and Ehrensing 1990). Adult females of A. rhodiolae, regardless of the possible existence of deutogyne females (which are typically the overwintering life stage of deuterogynous species of eriophyoids), presumably overwinter within galled tissues, or more strategically, near the stem base or on rhizomes near the soil surface (Krantz and Ehrensing 1990;McClay et al. 1999) so as to easily colonize, in the spring, new shoots of R. rosea growing from the same rootstocks as the previous year's host.
(AAFC) for their help during CLSM; Jim Amrine (West Virginia University) and Enrico de Lillo for sharing their knowledge and information on eriophyoids via their database; Wayne Knee (AAFC) for his help during SEM; and King Wan Wu and Wayne Knee (AAFC) for slide-mounting specimens of A. rhodiolae and A. destructor.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This paper benefited financially from Nunavik Bioscience (Makivik Corporation).