Seasonal phoresy as an overwintering strategy of a phytophagous mite

Migration by attachment to insects is common among mites that live in temporary habitats. However, because plants provide relatively stable habitats, phytophagous mites are generally not dependent on other animals for dispersal, so whether these mites can consistently be phoretic on insects through a particular life stage remains unclear and controversial. Here, we describe an obligate phoresy of a wholly phytophagous mite, Aceria pallida, in which the mites accompanied the psyllid Bactericera gobica to its winter hibernation sites, thus successfully escaping unfavourable winter conditions, and returned to reach the buds of their host plant early the following spring. This finding provides evidence of a new overwintering strategy that has contributed to the evolutionary success of these tiny phytophagous mites.

The gall mite Aceria pallida Keifer (Acari: Eriophyidae) and the psyllid Bactericera gobica Loginova (Hemiptera: Psyllidae) are the most important pests of the wolfberry 30 , Lycium barbarum L. (Solanaceae), which is widely cultivated in northwestern China and is of great importance in Chinese traditional medicine 31,32 . Preliminary observations found no mites in their common hibernation sites on the host plant, but many mite galls were found in spring buds and were closely related to the eggs laid by psyllids. We hypothesized that these psyllids likely contribute to the breakout and prevalence of the mites because no galls occurred if the buds were isolated from adult psyllids in early spring. Therefore, we postulated that mite phoresis on psyllids is not accidental.
Thus, we investigated this mite phoresis on psyllids over the course of 2 years, focusing on the attachment period in the late fall and the detachment period in the early spring, and the adaptive phoretic structure was also identified. The primary hibernation sites of the mites were examined, and the artificial induction of phoresy and detachment was studied.

Results
Seasonal phoresy of A. pallida. The life history of A. pallida is shown in Fig. 1. In winter, the activity of the phoretic mites on the psyllids almost completely ceases, but the psyllids may seek a comfortable site to protect themselves from harsh conditions (Fig. 1a). Thus, these insects represent a favourable environment for phoretic mites. As the temperature increases in early spring, the psyllids immediately feed on the buds of the host plant, and their activities (feeding, mating and laying eggs) provide the mites sufficient time to dismount (Fig. 1b). Afterward, the mites reproduce and live inside the galls of the host plant (Fig. 1c), and their dispersal could be by wind and other means in spring and summer. In late autumn, the mites emerge from the galls and temporarily hibernate under the abdomen of the inactive psyllid nymphs (Fig. 1d) and then transfer to the adults during or after eclosion. This strategy constitutes a well-developed, obligate seasonal phoresy that allows these eriophyoid mites to survive winter by attaching to the psyllids and to be transported to the host plant in the early spring (Fig. 1e).
The annual prevalence and phoretic rate of the mites on the psyllids (2 years of data) are shown in Fig. 2a, and both decreased from March (87%, 16.35 ± 2.05) to April (34%, 1.1 ± 0.32) during the spring detachment period (Fig. 1b,e). No phoretic mites were found on the adult psyllids, except for sporadic mounting on the body surface and wings, during the vigorous L. barbarum growth period from May to September. Nevertheless, as the temperature decreased and the leaves fell, the prevalence and phoretic rate increased rapidly from October (75%, 8.98 ± 2.10) to November (93%, 15.9 ± 4.47) and reached a steady state in December (93%, 26.7 ± 3.04) (Fig. 1a,e).
The primary hibernation site of A. pallida is not on the host plant but as a phoretic on the psyllids. No A. pallida gall mites were found surviving in the typical hibernation sites (bark, branches and buds) 13,33,34 , but live deutogynes of another free-living eriophyoid mite, the wolfberry rust mite Aculops lycii Kuang, was observed, although none were present on the psyllids. Moreover, the galls caused by A. pallida were not found on the plants that were isolated from the psyllids, but the mite infection rate was 100% when 10 adult psyllids were released on each plant (Fig. 3a). Furthermore, the number of galls caused by detached mites was significantly correlated with the number of psyllid eggs (P < 0.001, Fig. 3b); more eggs imply that the adult psyllids spend more time on these leaves and provide more time for the mites to detach. The entire worm-like body of A. pallida is structurally adapted for phoresy. There were two structures on the body of B. gobica suitable for mite attachment (Fig. 4a). Part I was the space under both metapedes coxae on the metathorax, and Part II was adjacent to the rostrum. These two structures include some spaces, which are surrounded by inter-segmental membranes, in which the vermiform mites can conceal themselves. More than 97% of the mites were phoretic on Part I of the psyllids (Fig. 2b), with up to 72 phoretic mites on female psyllids and 67 on male psyllids, and A. pallida can also attach to other psyllids with similar structures,  such as E. robinae (phoretic rate: 7.46 ± 1.81, phoretic probability: 73%). Additionally, the mites can also occasionally be found on other arthropods (ants, aphids, ladybugs, stinkbugs, and beetles), but none have structures similar to those of the psyllids to facilitate phoresis by the mites.

Discussion
Phoresy among arthropods as a means of migration from degraded habitats is ubiquitous in nature 16,18,29 , but phoresy as an overwintering strategy in phytophagous arthropods seems to be much rarer. In this study, we observed that wholly phytophagous gall mites, the deutogynes of A. pallida, have evolved a suite of complementary behavioural and structural characters in response to the seasonal challenge of overwintering and locating suitable habitats on host plants in early spring. Because they share the same host plant and degree of host specificity as the psyllid B. gobica 35 , the mites can dismount when the carrier insects arrive at the host plant and then reproduce on it.
We suggest that this novel overwintering strategy may have evolved from the shelter-seeking behaviour of the phytophagous mites under cold and dry environmental conditions 10,26 . Mites that live in shelters (gall-inducing and refuge-seeking species) may be more likely to be phoretic on arthropods than those that are free-living. Based on our observation that no rust mites, A. lycii, were found on B. gobica during the hibernation period, we deduced that of the rust mite A. lycii and the gall mite A. pallida, two eriophyoid species on the same host plant L. barbarum, only the gall-forming mite is phoretic on psyllids as an overwintering strategy. This could be explained by the different lifestyle and morphological characteristics of the species: free-living eriophyoid mites may have a relatively solid prodorsal shield and thicker tergites for resistance to winter conditions than the shelter-seeking mites 28 . Furthermore, the harsh winter conditions, the presence of a suitable carrier on the same host plant, the life history synchronization with the host, and particularly, the structural adaptations of A. pallida (such as its worm-like body and the appropriate space under both metapedes coxae of B. gobica) are likely the essential factors that promoted the evolution of this relationship.
In relation to the phoretic host (B. gobica) or other psyllids (such as E. robinae), the worm-like body of the deutogyne of the eriophyoid mite (A. pallida), with its accentuated development of microtubercles but shortened body size, constitute a special structural adaptation for phoresy, and we suggest that the structure and stability of these morphological adaptations are not inferior to those of other phoretic mites. In addition, the accentuated development of the microtubercles in the eriophyoid mites, which facilitate forward movement in confined spaces, is analogous to the function of the chaetae of annelid worms 28 . The deutogynes of the eriophyoid mites generally exhibit reduced or suppressed microtuberculation, which seems to conserve the body fluids of hibernating deutogynes, rendering their cuticles more resistant to water loss 28 . The wolfberry rust mite, A. lycii, which overwinters on the host plant, has similar structural features (electronic supplementary material, Fig. S1). However, the microtuberculation of the A. pallida deutogynes are longer than those of the protogynes, so we suggest that the carrier insect may provide a relatively humid environment for this mite during hibernation, and the need for microtuberculation to achieve phoresis may have driven the development of this structural difference.
The phoretic association between the mites and insects 16,20,36 and the association between the gall-forming arthropods and the host plant [37][38][39][40][41] have been considered to be model systems for studies of coevolution, but the phoresy of the gall mite may indicate a more advanced and complicated association between the mite, the carrier insect and the host plant. Because gall-forming arthropods are well known for their ability to manipulate host-plant morphology and physiology 37,39 , they could affect the performance of other herbivores. For example, the gall mite Aceria cladophthirus Nalepa increases the susceptibility of its host plant to spider mite Tetranychus urticae Koch 42 , and the performance of a butterfly Neuroterus saltatorius Edwards decreases with increasing gall wasp Erynnis propertius Scudder & Burgess density 43 . Furthermore, the galls caused by the midge Rabdophaga salicisbrassicoides Packard increase the abundance of aphids and their attendant ants 44 . The gall mite A. pallida can undoubtedly benefit from phoresy, but whether the mites benefit their carrier psyllids in direct or indirect ways during the growing season and whether they are clearly associated throughout the year require further study. Our results help to settle the debate regarding the existence of phoresy in phytophagous mites 5,12,14,28 because the seasonal phoresy of A. pallida, based on the definition and classification [15][16][17] , is a typical obligate phoresy. Although many phytophagous mites have some adaptations for dispersal by other carriers in the growing season 5,14,24 , they are better adapted to expand their habitats than to overcome harsh conditions. Michalska and Skoracka 5 have summarized the different dispersal modes of eriophyoid mites (wind, carriers, ambulation, and rain), but almost all of these dispersal mechanisms occur in the growing season, and none of the dispersal modes by carriers can be considered a type of phoresy. Thus, we deduced that the dispersal in the growing season should be understood as a normal population dynamic event, while phoresy as a migration strategy in phytophagous mites represents a seasonal displacement of a population 45 .
Another interesting finding is that the number of phoretic mites on B. gobica females was higher than on males following a harsh winter, but we suggest that the mites cannot discriminate the between sexes or even among species (such as E. robinae) of psyllid hosts. Instead, the adult female psyllids always have a larger body size 35 and a longer lifespan 46 that may provide more space for passengers and enable the mites to survive for a longer time despite difficult winter conditions. Furthermore, we suggest that female psyllids should be better hosts for phoresy because the adult females will spend more time on the buds of the host plant in early spring when laying their eggs.
We suggest that the phoresy of other phytophagous mite species may have evolved similarly to that reported here for A. pallida, and because these phoretic phytophagous mites may have the potential to be seriously harmful pests, their control and quarantine should be achieved by eliminating their carrier insects. Furthermore, when investigating the phoresy of other phytophagous mites, more attention should be paid to the insects and other arthropods that share the same habitat during the hibernation period. Finally, the gall-inducing mites on deciduous plants, especially the species in the Aceria genus that share the same habitat as the psyllids, may have greater potential for phoresy and require further study.
The adult psyllids were collected from the host plant using an aspirator (11 cm in height and 3 cm in diameter) and transferred to the lab. The carriers (B. gobica) were fastened onto stubs with double-sided tape, and a fine, bent micropin was used to slit the two large hind coxae from the midline; the sexes of the psyllids and the phoretic mites on different parts of the carriers were recorded under a Leica M205C stereomicroscope (Leica Microsystems, Wetzlar, Germany). The phoretic rate was expressed as the number of mites attached per psyllid, and the phoretic probability was expressed as the ratio of the number of psyllids that carried mites to the total number of psyllids.
The verification of the primary overwintering sites. There were two eriophyoid species (the gall mite, A. pallida, and the rust mite, A. lycii) on the same host plant (L. barbarum), and we verified the presence of the deutogyne mites of these two species in their general hibernation sites (bark, n = 500; branches, n = 800; buds, n = 1200) 13,33,34  Dismount induction. On 8 March 2014, the psyllids were collected from the experimental field and then 2 adult psyllids (1 female and 1 male) were released onto one seedling 2 days later (n = 50). Each seedling was isolated with a transparent plastic pipe (15-cm high, 6-cm diameter with the top sealed with a 60-mesh net) and reared in climate-controlled boxes (25 °C, 30-60% RH, 16: 8 h L: D). The galls on the leaves caused by A. pallida were examined 5 days later.
Morphological observations and measurements. The specimens of the gall mite A. pallida and its carrier B. gobica and the rust mite A. lycii used for scanning electron microscope (SEM) and morphological measurements were collected at the experimental field. The protogynes (non-phoretic stage) of A. pallida were collected from the galls of the host plant L. barbarum, and the protogynes of A. lycii were collected from the infected leaves on 28 July 2013. The deutogynes (phoretic stage) of A. pallida were collected on 08 January 2014 from the insect carrier B. gobica, and the deutogynes of A. lycii were collected on 12 March 2014 from their hibernation sites (bark, branches and buds). All of the morphological observations and measurements were conducted at the Institute of Medicinal Plant Development in Beijing, China.
Slides were mounted using Keifer's F-medium and modified Berlese medium 47 , and the specimens were measured based on the methods outlined by de Lillo 48 . The specimens were examined with a Leica DM2500 (Leica Microsystems, Wetzlar, Germany) research microscope with phase contrast, and photographs of the slide-mounted mites were taken using the same microscope connected to a computer using Leica LAS image analysis software. The terminology used to describe the eriophyoid morphology and classification follows that of Lindquist 25 . For the SEM studies, the live eriophyoid specimens were prepared following the methods used for fresh eriophyoid mites, and the adult psyllids were routinely processed based on the "acrolein method", according to Alberti and Nuzzaci 49 . The prepared samples were fastened onto stubs with double-sided tape and then coated with gold for 180 s at 30 mA in a JFC-1600 ion sputter (JEOL). They were then observed and photographed in a JSM-6510LV SEM (JEOL), and the SEM images of the phoretic mites were colorized in Photoshop CS5 (Adobe Systems). Four morphological characters (proterosoma length, hysterosoma length, body width and microtubercle length), which represent the main differences between the protogynes and the deutogynes 50 , were analysed, although there may be some other characteristics that differ slightly 50 . The length of the microtubercle (n = 200) was determined based on SEM, and the others characteristics were measured from slide images (n = 15 per character). The differences between the protogynes and deutogynes of A. pallida were evaluated statistically using a t-test to compare means in SPSS 20.0 software (IBM, Chicago, IL), and the values were reported as the mean ± SE.
Additionally, to confirm whether the mites could be phoretic on other psyllid species with similar structures or other arthropods with different structures during the loading period (October~November), the psyllid Euphalerus robinae Shinji was collected from its host plant Gleditsia japonica Miq. near the experimental site on 10 October 2014, and then 60 adult psyllids were released into a 60-mesh net over a branch with phoretic mites. Then, the phoretic rate of A. pallida on E. robinae was investigated 3 days later. On 15 October 2014, other arthropods (ants: Camponotus sp.; aphids: Aphis sp.; ladybugs: Propylaea japonica Thunb.; stinkbugs: Adelphocoris fasciaticollis Reuter; and beetles: Lema decempunctata Gebler; n = 30 per species) that shared the same habitat with A. pallida were collected from L. barbarum at the experimental site. The numbers of phoretic A. pallida on E. robinae or other arthropods were examined under a Leica M205C stereomicroscope.