Diversity and Interactions between Picobiine Mites and Starlings

Simple Summary This study investigates the complex interactions between quill mites of the family Syringophilidae and starlings. We identified five species of quill mites infesting 24 species of starlings, uncovering intricate host–parasite dynamics across the Palaearctic, Afrotropical, Oriental, and Oceanian zoogeographical regions. A detailed statistical analysis of the Picobiinae–Sturnidae network revealed low connectivity but high modularity, indicating specific relationships between the mites and their starling hosts. The network structure demonstrated four distinct modules, highlighting the specialised and compartmentalised characteristics of these associations. Furthermore, the distribution of Picobia mites was found to align with the phylogeny of their starling hosts, with particular mites targeting specific starling clades. The social and breeding behaviours of starlings were observed to contribute to the high prevalence of these mites. This comprehensive network analysis provides new insights into the ecological dynamics of host–parasite interactions. Abstract The subfamily Picobiinae (Acariformes: Syringophilidae) comprises obligate and permanent parasites of birds found exclusively in the quills of contour feathers. We studied associations of picobiine mites with birds of the family Sturnidae (Aves: Passeriformes) across the Palaearctic, Afrotropical, Oriental, and Oceanian zoogeographical regions. Among the 414 examined bird individuals belonging to 44 species (35.2% of all sturnids), 103 individuals from 24 species (54.5% of examined species) were parasitised by quill mites. The diversity of mites was represented by five species, including one newly described, Picobia malayi Patan and Skoracki sp. n. Statistical analysis of the Picobiinae–Sturnidae bipartite network demonstrated a low connectance value (Con = 0.20) and high modularity, with significant differences in the H2′ specialisation index compared to null model values. The network structure, characterised by four distinct modules, highlighted the specificity and limited host range of the Picobiinae–Sturnidae associations. The distribution of Picobia species among starlings was congruent with the phylogeny of their hosts, with different mites parasitising specific clades of starlings. Additionally, the findings suggest that the social and breeding behaviours of starlings influence quite a high prevalence. Finally, our studies support the validity of museum collections to study these parasitic interactions.


Introduction
The subfamily Picobiinae, established by Johnston and Kethley in 1973, comprises obligate and permanent parasites of birds, exclusively found in the quills of contour feathers covering the head, neck, breast, abdomen, and back of their avian hosts [1,2].An exception within this subfamily is Calamincola lobatus Casto, 1977, a unique species belonging to a monotypic genus, which inhabits the quills of wing feathers of the Groovebilled Ani, Crotophaga sulcirostris Swainson (Cuculiformes: Cuculiudae) [3].Within the quills, picobiines feed on soft tissue fluid by piercing the quill wall with their long needlelike chelicerae [4,5].The entire life cycle of these mites occurs within the quills, from larva through two nymphal stages (protonymph and deutonymph) to adults.The dispersal stage involves adult and fertilised females, which seek new feathers to infest, either on the same host individual or through vertical transmission from parent to chick [4,6,7].
Since the early 2000s, research has significantly expanded our understanding of picobiine diversity, with numerous new species and genera being described.Comprehensive reviews of picobiine fauna have been conducted across various zoogeographic regions, covering all except Antarctica (see references in [8]).Currently, the subfamily Picobiinae comprises 116 species grouped into 12 genera [8,9,15].Due to the vast diversity of these mites, their wide host range, and the many avian hosts yet to be studied, current research represents only the beginning of uncovering the full extent of picobiine fauna.
The largest avian order, Passeriformes, comprises more than 6200 extant species grouped into over 1350 genera and 140 families [16].The research presented here continues large-scale studies focusing on quill mites across particular families within the order Passeriformes.
The family Sturnidae (Starlings) includes approximately 125 species grouped into 36 genera [17,18].Starlings are medium-sized birds that adapt to various habitats, including dry savannas, scrublands, farmlands, grasslands, urban areas, and forests in temperate and tropical climates [17].These birds are found in the Palaearctic, Afrotropical, Oriental, and Oceanian zoogeographical regions [16].Starlings are social birds with often nomadic habits, assembling in huge flocks.Some species breed in colonies, which can be very large (even tens of thousands of pairs).They mostly breed in cavities, but some species also place their nests in the open.Their evolutionary success can be attributed to their high degree of dispersal.Cooperative breeding is also found in many sturnid species [16].All these traits make starlings very interesting as hosts of highly specific quill mites, including picobiines.Unfortunately, our understanding of quill mites parasitising this avian family remains incomplete.To date, three species of the genus Picobia Heller, 1880, have been described from this host group [19][20][21].
The current study has the following objectives: first, it describes a new species of the genus Picobia, and second, it aims to detail the diversity, interactions, and level of specialisation between syringophilid ectoparasites and their starling hosts on a worldwide scale.

Host Sampling
The mite material used in this paper was collected from dry bird skins primarily housed in the ornithological collection of the Bavarian State Collection of Zoology (SNSB-ZSM) in Munich, Germany.Additionally, the Eurasian Starling (Sturnus vulgaris) was examined at the Icelandic Institute of Natural History (IINH) in Reykjavik, Iceland, and the Superb Starling (Lamprotornis superbus) was examined in the Ornithology Section of the National Museum of Kenya (NMK) in Nairobi, Kenya.Approximately 10 contour feathers were removed from each bird specimen, specifically from the area near the cloaca.

Mites Preparation, Identification, and Depository
All collected feathers were examined under a stereomicroscope and opened using fine forceps.Infested feathers were placed in tubes with Nesbitt's solution at room temperature for 3 days, following the protocol introduced by Walter and Krantz [22] and Skoracki [11].Subsequently, mites were mounted on microscope slides in Hoyer's medium [22].Identifications and drawings of mite specimens were conducted using an Olympus BX51 light microscope (Olympus Corp., Tokyo, Japan) with differential interference contrast (DIC) optics.Figures were prepared using a camera lucida attachment.In the descriptions, all measurements are provided in micrometers; the dimension ranges of the paratypes are provided in parentheses, following the data from the holotype.The idiosomal setation follows Grandjean [23] as adapted for Prostigmata by Kethley [24].The nomenclature of leg chaetotaxy follows that proposed by Grandjean [25].The morphological terminology follows Kethley [1] and Skoracki et al. [8].The abbreviations PF and NPF are used for the physogastric and non-physogastric forms of the females, respectively.
Mite specimens are curated at the following repositories, abbreviated as AMU for the Department of Animal Morphology, Adam Mickiewicz University in Pozna ń, Poland, and SNSB-ZSM for the Section Arthropoda Varia, Bavarian State Collection for Zoology, Munich, Germany.
The list with full data of the collected mite material is presented in Supplementary Materials.

Bird Systematics and Zoogeographical Regions
The scientific names and systematics of the birds follow Clements et al. [18] and Winkler et al. [17].Zoogeographical regions follow Holt et al. [26] and Ficetola et al. [27].

Statistical Analyses
To describe patterns within the studied host-parasite ecological two-way web, the "bipartite" package for R was used [28].This approach allows for a quantitative description of the ecological connections between parasites and their hosts [29].In our matrix, parasite prevalence was utilised as a quantitative index.Descriptive statistics were computed utilising Quantitative Parasitology v.3.0 on the Web [30][31][32].
First, we calculated connectance (connection ratio) in a bipartite network.This measure determines the ratio of the number of actual connections in the netDwork to the maximum possible number of connections.It indicates network density and is expressed as a value between 0 and 1, where 1 means that every possible element from one part of the network (parasites) is connected to every element from the other part of the network (hosts) and 0 means that there are no connections.Next, we calculated the C.score, a measure describing the tendency towards non-co-occurrence of species pairs.A high value of this index indicates a tendency for species to not co-occur, suggesting competition or other forms of mutual avoidance.Conversely, a low C.score indicates that species tend to co-occur more often than expected based on random distribution.To determine whether parasites are generalists or specialists, we used the H 2 ′ measure.This statistic ranges from 0 to 1, where 0 indicates that all species are generalists (interacting with many hosts), and 1 indicates that all species are specialists (each species only interacts with one host).The null.t.test was employed to check whether the observed H 2 ′ values significantly deviated from random values [33].We also calculated nestedness, which is a measure of order in the network.If parasites with a low number of interactions with hosts share these interactions with other hosts with a larger number of interactions, the network is considered nested.Nestedness temperature measures how much the actual nestedness of a network differs from a perfectly nested network.For a perfectly nested network, the temperature would be 0. A higher temperature indicates less perfect nesting.As a measure of the functional diversity of parasites, we used d ′ and d.The d ′ is normalised to the maximum possible diversity for a particular species in the network, ranging from 0 to 1, with 1 indicating maximum functional diversity.If d ′ for a species is 1, it means that it has maximum functional diversity, indicating that its interactions with other species are very diverse and unique compared to other species in the network.The d is a raw measure of functional diversity based on functional distances between species.High d values suggest a greater diversity of parasites in their interactions with hosts.Finally, we calculated modularity as a probability value that measures how well the modular structure fits the data.A value close to 1 indicates that the modular model has a relatively good fit for the observed data, suggesting that modularity is a significant feature of this network.

Systematics
In our study, we identified five mite species of the subfamily Picobiinae parasitising 24 starling species (Table 1).Below, we present a list of the mite species along with their hosts and distribution, including new records.At the end of this section, we have constructed a key to all picobiine species associated with birds of the family Sturnidae.Male.Not found.

Differential Diagnosis
Picobia malayi sp.n. is morphologically similar to Picobia indonesiana Skoracki and Glowska, 2008.In females of both species, the hypostomal apex is with a pair of bluntended projections; each lateral branch of the peritremes has a similar number (seven or eight) of chambers; the propodonotal shield is divided into three sclerites (two lateral and unpaired medial); setae h2 are distinctly longer than f1; the bases of setae ag2 are situated postero-lateral to ag1.This new species is easily distinguishable from P. indonesiana by the following features: in females of P. malayi, each medial branch of the peritremes has seven chambers; the stylophore is 295-310 long; setae h1 are 5.5 times longer than f1.In females of P. indonesiana, each medial branch of the peritremes has four or five chambers; the stylophore is 205 long; setae h1 are about twice as long as f1.

Differential Diagnosis
Picobia malayi sp.n. is morphologically similar to Picobia indonesiana Skoracki and Glowska, 2008.In females of both species, the hypostomal apex is with a pair of bluntended projections; each lateral branch of the peritremes has a similar number (seven or eight) of chambers; the propodonotal shield is divided into three sclerites (two lateral and unpaired medial); setae h2 are distinctly longer than f1; the bases of setae ag2 are situated postero-lateral to ag1.This new species is easily distinguishable from P. indonesiana by the following features: in females of P. malayi, each medial branch of the peritremes has seven chambers; the stylophore is 295-310 long; setae h1 are 5.5 times longer than f1.In females of P. indonesiana, each medial branch of the peritremes has four or five chambers; the stylophore is 205 long; setae h1 are about twice as long as f1.considerably, ranging from 5.9% to 100% for particular avian species, and the confidence intervals were wide, ranging from 0.3 to 100.The examined avian material included 11 uninfested host species (N = 59) (Table 3).

Discussion
In our studies, we found that, among the representatives of the subfamily Picobiinae, starlings are parasitized exclusively by those of the genus Picobia.This genus belongs to the Picobia-generic group and is the most species-rich genus in the family Picobiinae, with 45 described species [9,15].Most species of this genus are associated with birds of the order Passeriformes (41 species), while a smaller number of Picobia species have been recorded from non-passeriform birds, such as Piciformes (three species) and Bucerotiformes (one species).Due to the numerous associations of the genus Picobia with passerine hosts, discovering representatives of this genus on starlings is not unexpected.More interestingly, the distribution of different Picobia species parasitising birds of the family Sturnidae shows a highly congruent pattern with the phylogeny of their hosts [37][38][39][40][41][42][43].
The clade of jungle starlings from the Indo-Pacific Islands region is a deep branch in sturnid phylogeny [16].The recent distribution of these ancestral lineages indicates that the evolutionary origin of Sturnidae is possibly in Southeast Asia, perhaps towards the Southwestern Pacific [44].Infested members of this host lineage are parasitised by Picobia malayi (on Gracula, Strepsocitta, and Sarcops) and P. indonesiana (on Enodes, Mino, and species of diverse island radiation of genus Aplonis).The main group of starlings, which is widespread in savanna and woodland regions in the Old World, is divided into crown radiations of Eurasian Savanna Starlings and African Starlings.All members of the Eurasian clade in our dataset, e.g., Sturnus, Spodiopsar, and Acridotheres, but also Creatophora (whose native range is in East and South Africa), are parasitised by Picobia sturni.Finally,

Discussion
In our studies, we found that, among the representatives of the subfamily Picobiinae, starlings are parasitized exclusively by those of the genus Picobia.This genus belongs to the Picobia-generic group and is the most species-rich genus in the family Picobiinae, with 45 described species [9,15].Most species of this genus are associated with birds of the order Passeriformes (41 species), while a smaller number of Picobia species have been recorded from non-passeriform birds, such as Piciformes (three species) and Bucerotiformes (one species).Due to the numerous associations of the genus Picobia with passerine hosts, discovering representatives of this genus on starlings is not unexpected.More interestingly, the distribution of different Picobia species parasitising birds of the family Sturnidae shows a highly congruent pattern with the phylogeny of their hosts [37][38][39][40][41][42][43].
The clade of jungle starlings from the Indo-Pacific Islands region is a deep branch in sturnid phylogeny [16].The recent distribution of these ancestral lineages indicates that the evolutionary origin of Sturnidae is possibly in Southeast Asia, perhaps towards the Southwestern Pacific [44].Infested members of this host lineage are parasitised by Picobia malayi (on Gracula, Strepsocitta, and Sarcops) and P. indonesiana (on Enodes, Mino, and species of diverse island radiation of genus Aplonis).The main group of starlings, which is widespread in savanna and woodland regions in the Old World, is divided into crown radiations of Eurasian Savanna Starlings and African Starlings.All members of the Eurasian clade in our dataset, e.g., Sturnus, Spodiopsar, and Acridotheres, but also Creatophora (whose native range is in East and South Africa), are parasitised by Picobia sturni.Finally, the lineage of African Starlings, sister to Eurasian Savannah Starlings, was split into two distinct lineages [16].The clade of African Red-winged Starlings of genus Onychognathus is parasitised by Picobia wisniewskii, while African Glossy Starlings (represented by Lamprotornis, Poeoptera in our data) are parasitised by Picobia lapmrotornis.It can be assumed that representatives of individual Picobia species were already present in the ancestors of specific clades of Sturnidae.
Currently, most picobiine mites are treated as monoxenous parasites (56% of all species), inhabiting only one host species.Oligoxenous, or mesostenoxenous species, which live on hosts within a single genus or family, comprise 18% and 22% of the picobiine species, respectively.Only a small fraction (4%) is metastenoxenous, associating with hosts across multiple families but restricted to a single avian order [8].To date, we do not have records of mite species switching between representatives of different bird orders.The mite species associated with starlings can be classified as narrow oligoxenous parasites, such as P. wisniewskii, which is found on birds of the genus Onychognathus, and mesostenoxenous parasites, which include all other mite species, i.e., P. indonesiana, P. malayi, P. sturni, and P. lamprotornis, inhabiting birds from different genera.However, all of these species are specific to a particular phylogenetic clade of Sturnidae.
The Picobiinae-Sturnidae bipartite network (Figure 3) exhibited a low connectance value (Con = 0.25).This means that only 25% of all possible connections between picobiines and starlings are observed.This result is also confirmed by a C-score = 1, indicating that picobiines associated with this group of birds exhibit complete non-co-occurrence.Consequently, it is not surprising that the H 2 ′ specialisation index is also 1, showing that picobiines are highly specialised and do not interact with each other.The temperature of nestedness, 54.49, indicates the moderate internal organisation of the network.The normalised specialisation level (d ′ ) (maximum functional diversity) in all cases was 1, meaning that each of these parasite species has unique ways of interacting with hosts in the network.However, the non-normalised value of this index, "d", was highest for Picobia wisniewskii (d = 1.86) and Picobia lamprotornis (d = 1.79), indicating that these species have the greatest diversity in host interactions compared to other species within this taxon (P.malayi d = 1.51,P. sturni d = 1.44, and P. indonesiana d = 0.97).As a result, we observed very high modularity in our network, identifying five distinct modules (Figure 4), each comprising hosts ranging from two to eight.
The present studies on the prevalence of infestation show a relatively high rate, with few species where prevalence is below 20% (Table 1).This rate can be correlated with the highly social behaviour of the family Sturnidae, cooperative breeding, and the fact that most species are cavity breeders [16].It is worth noting that many of the studies of quill mites were conducted on museum skins in various collections.In such circumstances, whole quill mite populations remain enclosed in the calamus cavity of the feather quill, where they can be investigated even after many decades.Thus, one could ask-is it possible to consider such encapsulated population studies comparable with field studies performed on wild host populations?In this study, conducted entirely in museum avian collections, we found a prevalence of Sturnus vulgaris of 21.2% (all seasons pooled, different countries in the Palaearctic, see Table 3).Skoracki et al. [36] found in a study of S. vulgaris conducted in Poland (during the spring season) that the prevalence was 28.98%.These values do not differ significantly from each other (Fisher exact test, two-tailed p = 0.29), suggesting that population studies of quill mites performed in museum collections can be comparable to investigations in the field.

Conclusions
This study reveals the highly specialised and phylogenetically congruent nature of the interactions between quill mites of the subfamily Picobiinae and their starling hosts.We identified five species of quill mites parasitising 24 starling species, highlighting the narrow host specificity and significant specialisation within the Picobiinae-Sturnidae bipartite

Figure 3 .
Figure 3. Bipartite network graph of interactions between Picobiine mites (left) and their Starling hosts (right).

Figure 4 .
Figure 4. Modules of the Picobiine mites-Starlings communities.The intensity of the colour of each square indicates the strength of the interaction between particular parasite species (horizontal axis) and their host species (vertical axis).

Figure 4 .
Figure 4. Modules of the Picobiine mites-Starlings communities.The intensity of the colour of each square indicates the strength of the interaction between particular parasite species (horizontal axis) and their host species (vertical axis).

Table 2 .
Host species infested by quill mites with the index of prevalence (IP) and 95% confidence interval (CI, Sterne's method).