Body size‐dependent effects on the distribution patterns of phoretic mite species assemblages on Rhynchophorus ferrugineus (Olivier, 1790)

Abstract Phoretic mites attach to different body parts of the red palm weevil (RPW), Rhynchophorus ferrugineus (Olivier, 1790), to disperse. However, the question of how the patterns of attachment sites are formed remains intriguing. Here, we examined RPW‐associated phoretic mites in four districts in Northern Portugal (macrohabitat), and investigated the patterns of mite distribution on six body parts of RPW (microhabitat). At the macrohabitat level, we detected seven phoretic mite taxa using the RPW host in each of the four studied districts, all documented for the first time in association with this invasive exotic species in Portugal. However, their relative abundance (species evenness) varied between districts, as did species diversity. All examined weevils carried mites, and the prevalence of the different taxa did not differ between districts or sex of weevils. Measured by mean abundance and degree of aggregation, Centrouropoda sp. proved to be the dominant taxon, while Acarus sp. and Curculanoetus rhynchophorus were considered common subordinate taxa and Uroovobella sp., Mesostigmata, Nenteria extremica and Dendrolaelaps sp. sparse taxa. At the microhabitat level, all taxa were present on all body parts of the RPW; the highest abundance was in a region encompassing the inner surface of the elytra and the membranous hind wings (subelytral space). Analysis of niche overlap revealed that the distribution patterns of phoretic mite taxa on the RPW were not randomly structured. In the subelytral space, interspecific coexistence of mites increased as a function of body size difference with the dominant Centrouropoda sp. We found that in the subelytral space the large dominant species Centrouropoda sp. displaced the larger species Uroobovella sp. and the similarly sized species Nenteria extremica, but coexisted with smaller taxa.


| INTRODUC TI ON
The red palm weevil (RPW), Rhynchophorus ferrugineus (Olivier, 1790) (Coleoptera: Curculionidae), is a serious pest of palms (Arecaceae) native to South Asia and Melanesia. Since the 1980s, RPW has become an invasive exotic insect in the Euro-Mediterranean region, the Near East, parts of North Africa and the Caribbean (EPPO, 2023). In Portugal, it was first detected in 2007 in the south, in the Algarve region (Fernandes, 2007), and has since spread rapidly throughout the five regions of mainland Portugal and the Madeira Archipelago (Boavida & Franca, 2008;Pete, 2010;Ramos et al., 2013). In the Azores Archipelago, as far as we know, there are no records of its presence. The insect causes economic and ecological impacts after feeding on and destroying the Canary Island date palm Phoenix canariensis (Chabaud, 1882), one of its preferred hosts, which is widely used as an ornamental plant in Portuguese coastal cities (Fernandes, 2016). The ability of RPW to grow and develop under a wide range of climatic conditions is mainly attributed to its cryptic feeding behaviour as a borer larva inside the host plant, protected from external temperature and humidity fluctuations (Faleiro, 2006;Murphy & Briscoe, 1999). In addition, the hidden feeding behaviour protects it from natural enemies and pesticides.
Other groups, such as the deutonymphs of Parasitidae or phoretic adult Mesostigmata, hold on with little more than pad-like ambulacra or by grasping setae with their chelicerae (Bajerlein et al., 2013;Seeman & Walter, 2023).
Phoretic mites can occur in groups of up to thousands of individuals on specific parts of the host's body and are thought to impose costs on the host in the form of reduced lifespan (Mazza et al., 2011) or reduced fertility and fecundity of the female (Hodgkin et al., 2010), at least under laboratory conditions. However, Hodgkin et al. (2010) also found that heavily infested female hosts produced the healthiest offspring, suggesting that the effects of phoretic mites on the host may be different for adults and larvae, or that the predominant phoretic mite species changes during host life. Phoresy has been considered a precursor to parasitism, as deutonymphs of Hemisarcoptes cooremani (Thomas, 1961) have been found to parasitise their dispersal host Chilocorus cacti (Linnaeus, 1767) by using its body fluids (Houck & Cohen, 1995). Nevertheless, none of the  Allam et al., 2013;Allam & Elbadawy, 2017;Allam & El-Bishlawi, 2010). Other phoretic mites are nematophagous (Kinn, 1984), which may reduce the efficacy of entomopathogenic nematodes used to treat RPW-infested palms in Portugal and other countries (DGAV, 2013). The study of RPW-associated phoretic mites can therefore help to identify the biodiversity components of the ecosystem in which this invasive exotic weevil has become established and to develop potential biological control measures against this palm pest.
Early studies on RPW-associated phoretic mites have documented that coexisting mite species form distribution patterns on RPW body regions. Porcelli et al. (2009) reported the preferential distribution of Centrouropoda almerodai and Uroobovella marginata on the body of RPW in Malta, attaching to the underside of the elytra and to the exposed surfaces of the sternum, pygidium, head and legs respectively. Slimane-Kharrat and Ouali (2019) found three mite species on RPW in Tunisia: C. almerodai on the underside of the elytra, U. marginata on the pygidium, thorax and head, and U. javae on the antenna, thorax and legs. However, how these patterns emerge is intriguing and not yet fully understood. It has been proposed that interspecific overlap in space utilisation (niche overlap) is the cause of interspecific interactions that shape species distribution patterns and lead to species coexistence (Holt, 2009). Niche overlap analyses have been used to infer the extent of such interactions by statistically comparing observed values with expected values generated by randomisation with null models representing assemblages in which no biological mechanisms regulate species coexistence (Gotelli & Graves, 1996). Rohde (1984) has defined the habitat for parasites on two levels: the microhabitat and the macrohabitat, and this can also apply to phoretic organisms. According to this author, the macrohabitat is the environment in which the host lives and which is controlled by physical and chemical parameters. The microhabitat is the host itself, that is, the confined environment in which the parasite lives and settles. Several studies have been conducted on parasites at these K E Y W O R D S interspecific interactions, macrohabitat, microhabitat, phoresy, red palm weevil

T A X O N O M Y C L A S S I F I C A T I O N
Biodiversity ecology, Entomology, Invasion ecology two levels (Castro & Santos, 2013;Gobbin et al., 2021). In the RPWphoretic mite association, the macrohabitat of the mite includes the environment where the RPW, the host, lives (as a geographical distribution) and its microhabitat consist of the preferred regions on the host's body. To our knowledge, no studies combining macrohabitat and microhabitat analyses of RPW-associated phoretic mites have been conducted. In the present study, we documented for the first time the diversity of RPW-associated phoretic mite species in Portugal, particularly in the districts of Viana de Castelo, Braga, Porto and Aveiro in Northern Portugal (macrohabitat), and their distribution in different parts of the weevil's body (microhabitat). The generated dataset of mite abundance per body part was then analysed for niche overlap using null models to reveal possible mechanisms of coexistence of mixed mite taxa in this multi-symbiont host. The traps were inspected weekly; the captured weevils were transported to the laboratory in plastic containers and fed with apple slices. The specimens were then stored in the refrigerator at 8°C until dissection.

| Identification and distribution patterns of RPW-associated mites
The cold-anaesthetised weevils were separated by sex and dissected individually under a stereomicroscope. Phoretic mites were counted on six body parts of the weevils, that is, neck, head-antenna, thorax, legs, ventral surface (abdominal sterna) and subelytral space (membranous hind wings + inner elytra surface), then removed with a camel hair brush and preserved in 70% ethanol.
For identification, selected specimens of each observed mite morphotype were mounted on microscopic slides in lactic acid (90% solution in water; Helle & Sabelis, 1985) and examined with an Axiophot microscope (Carl Zeiss) connected to a computer running Leica Application Suite X (LAS X) image processing software.

| Statistical methods
All analyses were conducted using the open-source R environment (R Developmental Core Team, 2021). Prevalence (the proportion of weevils that harboured mites), mean intensity (the mean number of mites per infected weevil), mean abundance (the mean number of mites per weevil) and their respective confidence intervals (at a 95% confidence level) were calculated as in Rózsa et al. (2000).
The prevalence of mite species in each district and between female and male weevils was compared using Fisher's exact test, with the exact p-value given. Kruskal-Wallis chi-squared in conjunction with Dunn's post-hoc test (Bonferroni correction applied) at p < .05 as significance level were performed to detect differences between districts in the mean intensity of each mite species (Rózsa et al., 2000). Poulin's index of discrepancy (D) (Poulin, 1993 included abundance-based Hill diversity numbers ( q D) of the three q orders, that is, q = 0 (species richness), q = 1 (Shannon diversity or effective number of frequent species in the assemblage) and q = 2 (Simpson diversity or effective number of highly frequent species in the assemblage) were performed using the R package iNEXT (Chao et al., 2014(Chao et al., , 2020Hsieh et al., 2016). Plotting of the Lorenz curve and calculation of the Gini index were performed with the R package ineq (Zeileis & Kleiber, 2014). While the Lorenz curve is a graphical representation of the inequality between the size of individuals in a community (species evenness), the Gini index measures the extent to which the size of individuals within a community deviates from a perfectly equal distribution and ranges from 0 (complete equality) to 1 (complete inequality; Patil & Taillie, 1979). To visualise the distribution of mites on the host and to evaluate the uncertainty of the hierarchical cluster analysis (method 'ward.D', 'Euclidean' distance and bootstrap replications set to 10,000), a resource utilisation matrix consisting of the number of individual occurrences of each taxa (species abundance) in the different body parts of RPW (microhabitats) was analysed using the R-packages bipartite (Dormann, 2022) and pvclust (Suzuki & Shimodaira, 2006) respectively. The analysis of niche overlap was performed with the R package EcoSimR 1.00 (Gotelli et al., 2015). First, the aforementioned 7 × 6 resource utilisation matrix was used to determine the Pianka's index of niche overlap, which represents the average mean spatial overlap of all possible species pairs in the community and ranges from 0 (no overlap) to 1 (complete overlap; Pianka, 1974). Subsequently, the observed Pianka's index was statistically compared (p < .05) with expected values from simulated 7 × 6 resource utilisation matrices generated using the Monte Carlo randomisation algorithms RA2 and RA3 for null model analysis (Gotelli & Graves, 1996). RA2 assumes random equal utilisation of space (i.e. relaxed niche breadth), while RA3 assumes that each body part has an equal probability of harbouring a species (i.e. retaining niche breadth; Lawlor, 1980). The Pearson coefficient (r) with p < .05 was used for all correlation analyses. Species diversity in the four RPW-associated mite assemblages (i.e. Viana do Castelo, Braga, Porto, Aveiro) was compared using rarefaction/extrapolation curves of Hill numbers q D (Chao et al., 2014(Chao et al., , 2020, with exponential value q = 0 for richness, q = 1 for Shannon's diversity and q = 2 for Simpson's diversity (Figure 2a). We found a steep increase up to seven accumulated taxa in all districts; however, Aveiro needed more individuals to reach the seven taxa compared to Viana do Castelo, Braga and Porto. Shannon's diversity, which gives more weight to common taxa in the assemblage, was well recorded in all districts, being significantly higher in Viana do Castelo and decreasing in Porto, Aveiro and Braga with nonoverlapping confidence intervals. Simpson's diversity, which counts the effective number of highly common taxa in the assemblage, was also well recorded in all districts; Viana do Castelo and Porto had similarly high values with overlapping 95% confidence intervals on the rarefaction/extrapolation curves, while Aveiro had a lower value and Braga the lowest value. The four mite assemblages were then compared in terms of species evenness-a measure of the relative abundance of species within the community-using the Lorenz curve method and calculating the Gini index ( Figure 2b). In combination with species richness, these tools can be used to assess species diversity. Accordingly, species evenness decreased in the order Viana do Castelo > Porto > Aveiro > Braga (Figure 2b), confirming that the assemblage Viana do Castelo had the most even distribution and thus the greatest diversity of RPW-associated mite taxa.

| Macrohabitat level-Diversity and ecology of RPW-associated phoretic mites in Northen Portugal
The ecology of the four assemblages-in terms of descriptive statistics, such as prevalence, mean intensity, mean abundance and Poulin's index of discrepancy for each RPW-associated mite taxa-is summarised in Table   Centrouropoda sp. was clearly the dominant taxon with the highest MA values and the lowest D values, followed by Acarus sp. and C. rhynchophorus (the smallest mites of the seven taxa) with medium to high MA and D values, which were considered common subordinate taxa. The remaining four taxa, that is, Uroobovella sp., N. extremica, Mesostigmata and Dendrolaelaps sp., were each found in very low numbers and highly aggregated in the four RPW-associated mite assemblages and were then considered sparse taxa.

| Microhabitat level-Distribution patterns of RPW-associated mites and spatial overlap between co-occurring taxa
Mites were directly separated from six body parts of RPW during identification, namely the head-antenna, neck, legs, thorax, abdominal sterna and subelytral space (membranous hind wings + inner elytra surface) and quantified (Figure 4a). A two-sided plot of mite Potential spatial overlap between co-occurring mite taxa was investigated by calculating Pianka's index and by using null model approaches to suggest potential species interactions for more direct testing (Gotelli & Graves, 1996). We used the abundance of each taxon in the different body parts of RPW as a spatial resource. The niche plot of spatial resource utilisation showed the subelytral space F I G U R E 2 Alpha diversity of phoretic deutonymphs associated with the red palm weevil in Northern Portugal's districts of Viana do Castelo, Braga, Porto and Aveiro. (a) Sample size-based rarefaction and extrapolation diversity curves for red palm weevil-associated phoretic mite assemblages in Northern Portugal's districts of Viana do Castelo (magenta), Braga (green), Porto (cyan) and Aveiro (red), based on Hill's order (q) with q = 0 (species richness), q = 1 (exponential of Shannon entropy index) and q = 2 (inverse of Simpson index; Chao et al., 2020). The solid line is the rarefaction curve and the dotted line is the extrapolation curve. The intersection between the solid and dotted lines represents the observed values. The extrapolation goes up to 50,000 individuals to exceed the actual sample size. The shaded areas represent 95% confidence intervals obtained by the bootstrap method based on 100 replicates. (b) Lorenz curves and Gini index comparing the evenness of the relative abundance of red palm weevilassociated phoretic mite assemblages in the Northern Portugal's districts of Viana do Castelo, Braga, Porto and Aveiro. TA B L E 1 Summary of descriptive statistics of RPW-associated phoretic mites in four districts of Northern Portugal.  Figure 5a). The observed Pianka's index was also significantly higher than the expected Pianka averages of 10,000 simulated null assemblages (p < .05), regardless of whether the RA2 (relaxed niche breadth) or RA3 (retaining niche breadth) random algorithm was used for their constructions (Table 2; Figure 5b).
The latter suggests that the observed distribution patterns of RPWassociated mite taxa were not randomly structured, but most likely occurred due to biological mechanisms that force the species to coexist.

| Interspecific coexistence of phoretic mites in the subelytral space increases as a function of body size difference with the dominant taxon
The interaction between pairs of taxa in the subelytral space was investigated using the Pearson correlation of their respective abundances in a network plot. Significant correlations were found be- [r(236) = .15, p = .017] (Figure 6a). Interestingly, the significant correlations between the dominant taxon Centrouropoda sp. and the common subordinate taxa C. rhynchophorus and Acarus sp., which were also the taxa with the smallest mites, were positive, whereas they were negative between Centrouropoda sp. and the sparse taxa Urobovella sp. and N. extremica, whose mites did not differ much in size compared to the dominant mites. A Pearson correlation coefficient was then calculated to assess the linear relationship between the difference in body size (Δ body size) and the difference in relative abundance in the subelytra (Δ relative abundance) for pairs of taxa formed between the dominant taxon Centrouropoda sp. and each of the other six taxa. The results showed a strong negative correlation between the two variables, r(4) = −.85, p = .033 (Figure 6b).

| DISCUSS ION
Here we document a high species richness of phoretic deutonymphs of Uropodina and Astigmata associated with RPW in four districts of Northern Portugal-as part of the first study carried out on these organisms in Portugal-and propose body size effects as determinants of coexistence or exclusion among mite taxa during the formation of distribution patterns on RPW. A total of seven mite taxa were found, but only two of them (i.e. N. extremica and C. rhynchophorus) could be identified to the species level based on the original description of the life form of deutonymphs (Fain, 1974;Kontschán et al., 2014). In both cases, all described morphological details were found identical in the observed specimens. N. extremica has already been described in association with RPW (Dilipkumar et al., 2015). However, the first known association of C. rhynchophorus was with Rhynchophorus phoenicis (Fabricius, 1801) (Coleoptera: Curculionidae) (Fain, 1974) and was later also found with RPW in the United Arab Emirates (Al-Deeb et al., 2011). The other five taxa did not show sufficient similarity with the original descriptions and illustrations of the previously described species. In the case of mites identified as Mesostigmata, we suspected that they might be Uropodina deutonymphs, especially because of the long, thin chelicerae with which they apparently root about in the soil/substrate in search of nematodes (Koehler, 1997).
However, other diagnostic characters of the infraorder Uropodina, such as the size of the setae and the position of coxae I in relation to the tritosternal base, remain to be verified on adult specimens, which we have not been able to observe. The palp-apotele (or claws) appeared subdistal and on the inner side as in the Mesostigmata.
Ongoing work in our laboratory attempts to classify and name these mites to species level, including any potentially new species.
An interesting observation of this study was that no member of the mites that are phoretic as adults (i.e. Diplogyniidae, Ascidae, Laelapidae, Macrochelidae, Melicharidae) was found. Other phoretic groups, such as Parasitidae and the rare phoretics Cheyletidae and Oribatida, were also not found. Although we cannot say for sure, some mites, especially external mites and those that do not have strong means of attachment, may have been lost in the traps, during transport of the weevils to the laboratory or during storage in the refrigerator until dissection.
The seven species were present at each of the four sites.
The other six taxa were less abundant and showed a high degree of intraspecific aggregation, especially the sparse taxa Uroobovella sp., N. extremica, Mesostigmata and Dendrolaelaps sp. This behaviour might be important to avoid exclusion by the dominant taxon, as coexistence of ecologically closely related species is possible if interspecific interference is not more important than intraspecific competition (den Boer, 1986;Ives, 1991). Nevertheless, aggregated distributions are also characteristic of parasites (Goater et al., 2014), and given the documented cases of transition from commensalism to antagonism in phoretic mites (Houck & Cohen, 1995;Seeman & Walter, 2023), there is a possibility that these taxa have begun to exploit the RPW host, which needs further experimental verification.
We found the greatest amount of phoretic mites on the subelytral space of RPW. It has previously been suggested that this body region is preferred by phoretic mites because it can protect from fluctuating temperatures and ambient humidity but also to avoid being stripped as the weevil moves through dense palm fibres (Al-Deeb et al., 2011;Dilipkumar et al., 2015). These authors speculated that the attachment of mites to body regions other than the subelytra might depend on the available space at the time of attachment.
In our case, we observed mixed groups of the different taxa attaching to all body parts examined, especially in the subelytral space.
Deciphering the stabilising mechanisms of the coexistence of mixed mite taxa on RPW could provide new insights into how species diversity is maintained in a multi-symbiont host. The use of null models helped us to determine a nonrandom structure of mite attachment pattern on RPW. For example, niche overlap among RPW-associated TA B L E 2 Estimates observed and expected values of Pianka's niche overlap index. Expected values were determined using the RA2 and RA3 random algorithms to generate 10,000 null assemblages. Significant p-value <.05 is printed in bold. Future work will investigate how the seasonal dynamics of individual taxa affect interactions between them and the host to gain further insight into the potential ecological and evolutionary consequences of their association with RPW.

F I G U R E 5
Null model analysis of phoretic niche overlap in red palm weevil-associated mite taxa. (a) Niche plot and (b) histogram plot of resource utilisation (in our case red palm weevil body parts) by co-occurring phoretic mite taxa determined by the RA2 and RA3 random algorithms, each generating 10,000 null assemblages, implemented by the R package EcoSimR (Gotelli et al., 2015). The niche plot is a visualisation of the mite taxa × body part utilisation matrix for the original data matrix (red) and the simulated (RA2 and RA3) data matrices (blue); the area of each circle is proportional to the utilisation of a body part by a mite taxon. The histogram shows the simulated Pianka values (blue bars) for each RA2 or RA3 algorithm, a vertical red line corresponding to the calculated Pianka metric for the original data, long dashed lines indicating the one-sided 95% confidence limits and short dashed lines indicating the two-sided 95% confidence limits. writing -original draft (lead); writing -review and editing (lead).

ACK N O WLE D G E M ENTS
The authors thank the editor and two anonymous reviewers for use-

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets containing information on sampling location, date of sampling, abundance data and distribution patterns of phoretic deutonymphs associated with the red palm weevil in Northern Portugal, used and/or analysed in the current study, are available from the Dryad at https://doi.org/10.5061/dryad.0cfxp nw6n (Matos et al., 2023).
F I G U R E 6 Correlation analyses of mite taxa abundance in the subelytral space. (a) Correlation network of mite taxa abundance in the subelytra of the 238 weevils examined. Each bubble node corresponds to one mite taxon. The size of the bubbles is proportional to the size of the mite taxon, that is, Urb, Uroobovella sp. > Cnt, Centrouropoda sp. > Nnt, Nenteria extremica > Mss, Mesostigmata > Dnd, Dendrolaelaps sp. > C.r., Curculanoetus rhynchophorus > Acr, Acarus sp. The green edges represent positive Pearson correlation coefficients (r > 0), while the red edges represent negative Pearson correlation coefficients (r < 0). The colour intensity and the width of the edges are proportional to r; the value of r is only shown for significant correlations (p < .05).
(b) Correlation between the difference in relative abundance (the percentage composition of mites of a given taxon to the total number of all taxa in the population; Δ relative abundance) in the subelytral space and the difference in body size (Δ body size) of the dominant taxon (Cnt) with each of the other six mite taxa (i.e. Urb, Nnt, Mss, Dnd, C.r., Acr). The shaded area along the regression line corresponds to the 95% confidence intervals. The inset shows the values of r and p. Significance p < .05.