Epidemiological Survey of Two Morphotypes of Demodex Folliculorum (Prostigmata: Demodicidade) in Young People from Southern Spain

Epidemiological survey of two morphotypes of Demodex folliculorum (Prostigmata: Demodicidade) in young people from southern Spain F.J. M´ arquez a , A.J. L´ opez-Montoya b , S. S´ anchez-Carri´ on c , I. Dimov d , M. de Rojas c,* a Department of Animal Biology, Vegetal Biology and Ecology, Universidad de Ja´ en, Spain b Department of Statistics and Operational Research, Universidad de Ja´ en, Spain c Department of Microbiology and Parasitology, Faculty of Pharmacy, University of Seville, Profesor García Gonz´ alez 2, 41012 Seville, Spain d Laboratory of Parasitology, Zoological Institute, Russian Academy of Science, Universitetskaya embankment 1, Saint Petersburg, 199034, Russia ARTICLE INFO Keywords: Demodex folliculorum Infection Eyelash Skin Age and gender structure Contact lenses ABSTRACT Different morphotypes of Demodex from humans have been described. Among them, molecular studies have made it possible to discern between the Demodex folliculorum and Demodex brevis. Further studies showed two morphotypes of D. folliculorum harboured two different habits (human skin and human eyelashes), both of them with finger-shaped terminal opisthosoma difficult to differentiate and that can be assigned to D. folliculorum, Thus, a complete morphometric study of the species, which in this study are referred to as Demodex folliculorum species complex, was carried out. From this morphometric and meristic study two distinct morphological forms (short and long) could be identified within the Demodex folliculorum species complex. These forms differ significantly in four out of the six biometric parameters we analysed: gnathosomal length and width, podosomal width, and opisthosomal length. Moreover, a comprehensive survey of the two morphotypes from different habitats (skin and eyelashes), was carried out in young people of Southern Spain. Therefore, an analysis of 104 asymptomatic students, which were contact lens wearers, and the presence of D. folliculorum was carried out. A statistical analysis based on Bayesian zero inflated Poisson GLM has been applied to our sample data. For the age group considered (18–24 years old), the overall D. folliculorum prevalence for skin face or eye infections (at least one of them) was 19.31 % (51 people), with a statistically significant higher prevalence in males men. Furthermore, there is a slight statistical correlation between the presence of Demodex in silicone hydrogel soft contact lens wearers. This study confirms: i) the existence of two morphotypes of D. folliculorum that appear segregated due to the parasitization microhabitats,ii) a higher prevalence of mites in men than in women, iii) the existence of a high number of statistically supported double infections (skin-eyelashes). It also provides epidemiological data on the prevalence of long and short forms of D. folliculorum in a healthy young population. * Corresponding author. E-mail addresses: [email protected] (F.J. M´ arquez), [email protected] (A.J. L´ opez-Montoya), [email protected] (S. S´ anchez-Carri´ on), [email protected] (M. de Rojas). Contents lists available at ScienceDirect Parasite Epidemiology and Control journal homepage: www.elsevier.com/locate/parepi https://doi.org/10.1016/j.parepi.2024.e00381 Received 5 March 2024; Received in revised form 13 September 2024; Accepted 24 September 2024 Parasite Epidemiology and Control 27 (2024) e00381 Available online 28 September 2024 2405-6731/© 2024 The Authors. Published by Elsevier Ltd on behalf of World Federation of Parasitologists. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). 1. Introduction Mites of the genus Demodex, show similar morphological characteristics due to the microhabitat where they develop: pilosebaceous follicles. Demodex species from humans are tiny mites (0.1–0.5 mm) with elongated appearance, four pairs of stumpy legs, and a long opisthosoma with annulations. This similar morphology could be due to convergent adaptative evolution among species of this genus (Thoemmes et al., 2014; Palopoli et al., 2015). After this permanent host association of D. folliculorum, studies on genome sequence and gene structure revealed an extensive genome reduction through relaxed selection and genetic drift, that coincided with an extreme reduction in the number of cells (Smith et al., 2022), representing an exceptional case among the Arthropods. This strategy could have determined the evolutionary history of Demodex species and the adaptation of different populations to specialized habitats. Demodex species are the most common ectoparasites of humans and currently two species have been described: Demodex folliculorum (Simon, 1842) and Demodex brevis (Akbulatova, 1963) (Palopoli et al., 2015; Palopoli et al., 2014). Jing et al. (2005)) redescribed by SEM D. brevis based on the absence of spurs on each leg, only five pairs of palpal claws at the end segment of palpals, spindle hypostoma, and cone-like terminus. The former species are preferably located in the follicular infundibulum of the face (nose, cheeks, forehead, temples, and chin are favoured locations) and in eyelash follicles and prove to be the most prevalent. In contrast, D. brevis is settled in deeper sebaceous and Meibomian glands and the prevalence is much lower and recorded rarely (Zhao et al., 2012; Litwin et al., 2017). Phylogenetic analysis of 18S rDNA (Thoemmes et al., 2014; Zhao et al., 2012), as well as mitochondrial 16S rDNA sequences (Zhao and Wu, 2012; Zhao et al., 2013a), reveals the existence of two clearly differentiated species that at the intraspecific level, are geographically structured. Furthermore, Zhao et al. (2013b), observed four morphotypes (excluding males and immature forms): long and short-bodied Demodex folliculorum with finger-like terminus and Demodex brevis with finger- or cone-like terminus. The results of sequence analysis of 16S rRNA gene demonstrated that the genetic distances, among the three phenotypes with finger-like terminus indicated an intraspecific variation (Fig. 1) (Zhao et al., 2013a; Zhao et al., 2013b; Hu et al., 2014). However, the genetic distance among the three phenotypes with finger-like terminus and the one with cone-like terminus suggested an interspecific variation. Thus, in this study the phenotypes with finger-like terminus are referred to as Demodex folliculorum species complex. It is therefore worth considering that the shape of the opisthosomal terminus is a major morphological feature to differentiate these two species. Numerous epidemiological surveys to determine the prevalence of D. folliculorum and D. brevis have been carried out in healthy people. Although the prevalence increases with the age of the host (Andrews, 1982; Liu et al., 2010), appears that most studies falling in the range of 10–20 % (Correa Fontt et al., 2020; Sevgen and Mor, 2019; Karaman et al., 2014; Cengiz et al., 2014; Cui and Wang, 2012; Vargas-Arzola et al., 2012; Yazar et al., 2008). These authors always isolated mites from the same microhabitat and they did not consider the relationships between populations from different locations (skin or eyelashes). In addition, mite identification is done solely based on biometric differences and therefore D. brevis could be misidentified as the short type of D. folliculorum as can be seen, for instance, in the photograph showed in Tilki et al. (2017)). Moreover, the pathogenic role of Demodex species in humans has been demonstrated in several studies on the association between both populations D. folliculorum Fig. 1. (a) D. folliculorum short bodied with finger-like terminus. Ventral female. Isolated from human skin. (b) D. folliculorum long-bodied with finger-like terminus. Ventral female. Isolated from human eyelashes. F.J. M´ arquez et al. Parasite Epidemiology and Control 27 (2024) e00381 2 and ophthalmological or dermatological diseases that should be treated individually. Among ocular pathologies, it is worth highlighting blepharitis (Liu et al., 2010; Kemal et al., 2005; Czepita et al., 2007; Zhang et al., 2018; Aumond and Bitton, 2020) along with chalazion (Liang et al., 2014; Yam et al., 2014), and corneal disease (Kheirkhah et al., 2007). Most authors in the past five years report that Demodex blepharitis accounts for ≥60 % of those with blepharitis (Rhee et al., 2023). Concerning the dermatological pathologies attributed to the presence of Demodex, rosacea stands out in order of importance (Bonnar et al., 1993; Moravvej et al., 2007; Türkmen and Türko˘ glu, 2019; Sarac et al., 2020; Yazısız et al., 2019; Kubanov et al., 2019; Forton, 2022), followed by perioral dermatitis (Searle et al., 2021). Sebaceous hyperplasia with oily or mixed skin seems to favor Demodex proliferation (Zhao et al., 2011). Thus, the objective of this work is, firstly, to know the prevalence values of D. folicullorum in healthy population of university students in southern Spain and to establish a relationship between the presence of the long and short finger-like terminus populations forms. Secondly, taking advantage of the exhaustive survey which incorporates contact lens wearers, the relationships between the presence of D. folliculorum and the use of contact lens will be analysed. 2. Materials and methods In order to carry out the epidemiological study, the population was selected according to the following parameters: The exclusion criterion for people involved in this study was not to experience ocular or dermatological symptoms, not to be under corticosteroids or antibiotics therapy and an age not comprised between 18 and 24 years old. The studied population consisted of 264 young people 186 women and 78 men (Table 1). Mean age of the study group was 20.77 years (age range from 18 to 24 years old). On the other hand, in the study population, 104 students were users of soft silicone hydrogel soft contact lenses (75 women and 29 men). 3. Sample methods Methods used to collect Demodex mites from humans included the cellophane tape method (placing adhesive tape on the face to stick to the mites on forehead and cheeks) (Li and Wang, 1986) and collection of eyelashes (Kemal et al., 2005). In this case we take three pieces of adhesive tape (3 cm ×1 cm) and plucking eyelashes (3 eyelashes from each eyelid), for determining parasitism of skin face and eyelashes populations respectively. In accordance with the Helsinki II Declaration, before the start of the survey, all participants signed an informed consent, according to the model of Health Service of Andalusia (SAS). 3.1. Morphometric study From the set of collected mites from skin and eyelashes, 28 individuals from the skin (short body type with finger-like terminus) and 29 specimens from the eyelashes (long body type with finger-like terminus) of Demodex folliculorum adult females were selected for morphometric study of both populations. The specimens were cleared in lactic acid and mounted in Hoyer’s medium and carefully identified based on their morphological features (Jing et al., 2005). Demodex brevis was not detected. Following previous investigations [40, 5, 9] we compared seven biometric variables of individuals from different origins (skin or eyelashes): Gnathosomal length (Gn_L); gnathosomal width (Gn_W); podosomal length (Pd_L); podosomal width, (Pd_W); Table 1 Descriptive statistics of biometric variables by origin. From left to right: Variable name, Mean, standard deviation (SD), interquartile range (IQR), Range, Asymmetry (Skewness) and the tailedness of the distribution (Kurtosis). In plain text, data from skin population (N =28). In bold, data from eyelash population (N =29). Variable Mean SD IQR Range Skewness Kurtosis Gn_L 23.96 1.91 3.55 [20.50–27.30] 0.17 −1.16 21.26 1.90 2.77 [18.20–24.60] 0.04 ¡0.97 Gn_W 29.91 1.45 2.25 [27.30–32.50] 0.05 −0.75 25.87 2.18 3.18 [22.80–31.20] 0.74 0.21 Pd_L 87.88 12.79 20.34 [50.25–103.53] −1.00 1.02 80.90 11.92 19.85 [62.24–114.30] 0.49 0.57 Pd_W 64.69 7.09 13.70 [55.60–76.20] 0.33 −1.37 58.20 3.87 4.45 [47.90–65.20] ¡0.57 0.84 Op_L 151.38 17.01 19.75 [104.19–183.30] −0.87 1.26 254.92 31.90 55.12 [206.50–331.50] 0.33 ¡0.56 Op_W 55.98 6.26 9.27 [44.90–64.90] −0.50 −1.00 55.90 3.34 5.95 [51.00–63.80] 0.38 ¡0.74 Tot_L 263.23 23.88 23.51 [204.20–295.30] −1.31 1.50 357.09 39.07 61.22 [293.00–442.00] 0.19 ¡0.65 Op_L/Tot_L 0.58 0.71 0.84 [0.51–0.62] 0.66 0.84 0.71 0.82 0.90 [0.70–0.75] 1.74 0.86 F.J. M´ arquez et al. Parasite Epidemiology and Control 27 (2024) e00381 3 opisthosomal length (Op_L); opisthosomal width (Op_W); total length (Tot_L); ratio opisthosomal length/total length (Op_L/Tot_L). 3.2. Statistical methods applied to morphometric study Statistical methods applied to morphometrical and meristic study consisted in a descriptive study of the six biometric variables of individuals according to its origin. To do this, we made a summary table with the typical descriptive of measurements of the variables separated by groups, as well as graphically using Boxplots. We compare the mean lengths and widths of the measurements according to its origin using Barplots. With these types of graphs, we compare mean values of these variables, incorporating in turn a measure of dispersion such as the standard deviation. In order to finding for significant differences in the morphological features (Gn_L, Gn_W, Pd_L, Pd_W, Op_L and Op_W) of the Demodex between the two populations (skin and eyelash), we decided to use a Hotelling’s T2 test, a widely multivariate test used to deal with this kind of comparisons. Hotelling’s T2 test procedures is equivalent to a MANOVA which uses the Hotelling-Lawley trace in its estimation. This type of MANOVA accounts for non-independence of the response variables. With this analysis, we check whether there exist significant differences globally between the morphological features of the two populations of the Demodex. Once the global differences between morphological features have been found, a Games-Howell post-hoc multiple comparisons procedure were carried out to detect among which morphological features these differences occur. Games-Howell is the more applicable test for pairwise comparisons when variances were nonhomogeneous, as in our case. Hotelling’s T2 test was fitted using HotellingsT2Test function of the ‘DescTools’ package (Vehtari et al., 2021), and its Games-Howell post-hoc multiple comparisons were conducted using games_howell_test function of the ‘rstatix’ package (Kassambara, 2024), via R software (R Core Team, 2021). We use a support vector machine (SVM) as a predictive method for differentiating morphotypes of Demodex through the biometric analysis of their anatomy using values of morphological parameters. The purpose of a SVM is to construct optimal decision boundaries among classes in a decision plane. A decision plane separates a set of objects having different class memberships. To achieve this goal, we used plot3d and rgl.planes functions of the R packages ‘e1071’ (Meyer et al., 2023) and ‘rgl’ (Murdoch et al., 2022). 4. Statistical methods applied to epidemiological study With the aim of determining how infection type and use of contact lenses affect to the number of Demodex, we construct a regression model. Generalized linear model (GLM) was used to conduct this regression analysis. Specifically, a zero-inflated Poisson GLM is used to model count data that has an excess of zero counts. Further, theory suggests that a separate process from the count values generates the excess zeros and that the excess zeros can be modelled independently. Zero-inflated models are used when the zeros observations have two different origins, these are, the “sampling zeros” and the “structural zeros”. The sampling zeros are assumed that were occurred by chance, that is, the subjects who are exposed to the outcome but may or may not register the event of interest during the study period. The structural zeros occurred when the subjects always produce zero counts. Therefore, a Bayesian zero-inflated Poisson GLM was fitted to determine the relationship among the total number of Demodex as dependent variable (counting), considering the infection type (categorized as 0 =no infection, 1 =eyelashes infection, 2 =skin infection, 3 =mixed infection (eyelashes and skin) and contact lenses (categorized as a dichotomous variable) as explanatory variables. Obviously, the explanatory variables were defined as fixed effects. Bayesian zero-inflated Poisson GLM was fitted using “brms” package (Bürkner, 2017) via R software (Ripley, 2001). The function brm performs a Markov Chain Monte Carlo (MCMC) estimation to approximate the posterior distributions of the model parameters. Four chains with 5000 iterations each (warmup =2500), for the first simulations the treedepth was set to 20, and default priors of the “brms” package were used. The R^2 value was also estimated with bayes_R2 function of “brms” package. It is obviously that we are not going to find multicollinearity problems in the model, however, we check the multicollinearity via variation inflation factors (VIFs) test using the check_collinearity function from the “performance” package in R (Lüdecke et al., 2021). Multicollinearity in the explanatory variables of the regression model is insignificant when VIF values were <5 (James et al., 2013). Potential scale reduction factor (Rhat) for the stationary of the posterior distribution was reached when Rhat <1.01 (Vehtari et al., 2021). The significance of the parameters was hold whether their 95 % credible interval of the posterior distribution of the model parameters did not overlap zero. We plotted the uncertainty intervals - high-density intervals for the effects using the plot model function in the “sjPlot” package (Lüdecke et al., 2021). Goodness of fit test was accomplished using “DHARMa” package (Hartig, 2024) which implements a method to standardize residuals for GLM (Dunn and Smyth, 1996; Gelman and Hill, 2007). This method based on the residuals allow for visual analysis just like residuals of linear models that would display standardized residuals uniformly distributed. We also implemented a Bayesian goodness of fit test to validate the candidate model using pp_check function from “brms” package. This method compares our observed data with the simulated data from the posterior predictive distribution. The R code to performance the statistical analysis is in Supplementary Material. Fisher’s exact test was applied in order to evaluate the associations between sex (man and woman) and the presence or absence of the mite infection (non-infected and infected). We used the fisher.test function from the package “stats” in R to carry out this test. To do this, we decided to performance Fisher’s exact test separately in three different categories of infection, one for cases of only infection was detected in the eyelashes, and another for cases of only infection was detected in the skin and for both cases of infection (eyelashes and skin). The strength of the associations was checked via Cramer’s V test using assocstats function of the “vcd” package (Meyer et al., 2023). When the values of Cramer’s V test (ES) are equal or less than 0.2 (ES ≤0.2), the association is considered as weak, if 0.2 <ES ≤ 0.6, the association is moderate, and if ES >0.6 the associations is strong. F.J. M´ arquez et al. Parasite Epidemiology and Control 27 (2024) e00381 4 5. Results 5.1. Morphometric study Adult females of D. folliculorum and D. brevis can be separated by morphometrical features as: total length, the position of the vulva at the level of the coxae IV in D. folliculorum and behind them in D. brevis. Moreover, D. folliculorum presents seven pairs of palpal claws and shows tiny spurs on each leg while in D. brevis only five pairs of palpal can be located and spurs are not showed on their legs. Furthermore, D. brevis shows a pointed shape of the end of the opisthosoma, while in D. folicullorum is rounded. In this study, no morphological features were detected between D. folliculorum species complex. Thus, a complete biometric study was carried out in specimens from skin and eyelashes. The total length range of the individuals analysed varies between those coming from the skin (204.20 to 295.30 μ m), versus those coming from the eyelashes (293.00 to 442.00 μ m). The mean in the total length is greater in the case of the eyelashes (average: 263.23 μ m, SD: 23.88) compared to the skin (average: 357.09 μ m, SD: 39.07) (Table 1). Means values and its standard deviations for the six variables considered are represented in Fig. 2. The statistics relied with the distribution of both populations are showed in Fig. 3. A Hotelling’s T2 test was performed to determine the effect of population (skin and eyelash) on morphological features (Gn_L, Gn_W, Pd_L, Pd_W, Op_L and Op_W). The Hotelling’s T2 test was detected globally significant difference among the types of population on morphological variables (Hotelling’s T2 =73.648, pvalue <2.2e-16). According to Table 2, there exists significance differences in the completely pairwise comparisons but Op_W variable, with a p-value =0.953. Six biometric patterns as predictor values were used to construct support vectors in the first phase. First training phase and SVM results for two biometric patterns dependent variable to be classified: Demodex morphotypes (skin or eyelashes). Correlation matrix shows high correlation of Gn_W with Gn_L and Po_L features, and that determine that in the second phase of the study Gn_W were excluded of analysis. In order of importance, we selected as predictors the following variables: Gn_L, Pd_W and Op_L (Table 2). Demodex from skin (blue) and from eyelashes (red) have been represented in a point cloud (3D scatterplot) using the support vector machine technique. We have found a perfect classification of both samples through the maximum margin hyperplane (Fig. 4). The corresponding classifier is Clas = − 14.9201 +0.009214008*Gn_L +0.015592936* Pd_W -0.082217298 * Op_L. When the value of Clas >0, the concerned Demodex mites comes from the eyelashes, while Clas <0 corresponds to a skin Demodex (Fig. 3). 5.2. Epidemiological study With regard to the epidemiological data on demodectic infection, the following can be highlighted. The overall D. folliculorum prevalence for skin face or eye infections (at least one of them) was 19.31 % (51 people) (Table 1). When different data groups were considered, prevalence for skin face was 15.53 % (365 specimens isolated) and 16,28 % for eyelashes (153 specimens isolated). From these two populations of Demodex those specimens in better conditions were chosen for morphometrical Fig. 2. Barplots showing the means values with its standard deviations for the six variables considered. F.J. M´ arquez et al. Parasite Epidemiology and Control 27 (2024) e00381 5 studies (28 and 30 respectively). The mean of D. folliculorum count for each population was 0.31 mites per eyelash and 0.96 mites per cm 2 respectively (infection index was calculated as in Rivera et al., 2013 (Rivera et al., 2013)). When sex distribution was analysed, for each population, 7.19 % of infected people were men and 8.33 % women if D. folliculorum eyelashes population were considered. Similar results show the infection of skin population (7.95 % for men and 8.33 % for women) (Table 3). The effect of contact lens use determined higher percentages of infection in both men and women (Table 4, Fig. 5), going from an infection that affects 19.6 % of non-user men to 63.5 % of users. In the case of women, something similar occurs, such that the Fig. 3. Boxplots represent the values of the six variables according to origin (skin and eyelash): The quartiles of each distribution are represented by the top and bottom of the boxes, the median is the line across the box, the mean value is represented by the cross inside the box. Upper and lower limits are represented by the vertical solid lines. Table 2 Pairwise comparisons between populations using Games-Howell post hoc test. Variables Comparisons Estimate Low CI 95 % Upper CI 95 % p-adj Gn_L Eyelash-skin 2.70 1.69 3.72 1.77e-6 Gn_W Eyelash-skin 4.04 3.05 5.02 7.87e-11 Pd_L Eyelash-skin 6.98 0.41 13.5 0.038 Pd_W Eyelash-skin 6.49 3.42 9.56 1.11e-4 Op_L Eyelash-skin −104 −117 −89.9 1.11e-12 Op_W Eyelash-skin 0.08 −2.62 2.78 0.953 F.J. M´ arquez et al. Parasite Epidemiology and Control 27 (2024) e00381 6 Fig. 4. Three-dimensional feature space with the two Demodex classes defined by variables Gnathosomal length (Gn_L), Podosomal width (Pd_W and Opistosomal length (Op_L). Red dots indicate eyelash samples, while blue dots represent skin mites. The boundary decision hyperplane is shown in grey. The scales are in μ m. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 3 Prevalence values considering all people surveyed. Total number of mites collected and Infection Index (as number of mites per lash or number of mites per cm 2 , depending on the sampling site considered). All Volunteers Considered D. folliculorum from eyelashes D. folliculorum from skin Mixed infection Men (78) No. Infected people 19 21 17 Prevalence (%) 24.36 26.92 21.79 No. mites 77 173 – Infection Index 0.02 0.07 – Women (186) No. Infected people 22 22 15 Prevalence (%) 11.83 11.83 8.10 No. mites 76 192 – Infection Index 0.02 0.08 – Total (264) No. Infected people 41 43 32 Prevalence (%) 15.53 16.30 12.12 No. mites 153 365 – Infection Index 0.05 0.15 – Table 4 Prevalence values considering the use of contact lenses: number of cases (in bold) and percentage of prevalence (in brackets). Use of contact lens No (%) Yes: (%) Total: (%) Men N56 (71.79) 22 (28.20) 78 (100) Only eye inf. 2 (3.57) 1 (4.54) 3 (3.84) Only skin inf. 2 (3.57) 3 (13.63) 5 (6.41) Mixed inf. 7 (12.50) 10 (45.45) 17 (21.79) Total inf. 11 (19.64) 14 (63.63) 25 (32.05) Women N140 (75.26) 46 (24.73) 186 (100) Only eye inf. 4 (2.85) 2 (4.34) 6 (3.22) Only skin inf. 3 (2.14) 2 (4.34) 5 (2.68) Mixed inf. 8 (5.71) 7 (15.21) 15 (8.06) Total inf. 15 (10.71) 11 (23.91) 26 (13.97) F.J. M´ arquez et al. Parasite Epidemiology and Control 27 (2024) e00381 7 percentage of infection ranges from 10.7 % to 23.9 %. Fisher’s exact test for associations between sex (man and woman) and the existence of demodicosis (uninfected versus infected) for both cases of infection (eyelashes and skin) show a p-value =0.0011, indicating significant differences due to sex. The value obtained for Cramer’s V (0.21) indicates a moderately significant strength of association between Demodex infection and the sex of the individual. If we consider exclusively the infection in the eyelashes (Table 2, Fig. 2), we obtain a value of p =0.0049 in the Fisher test, which indicates a significant difference when considering sex. In this case, the Cramer’s V value is 0.18, indicating a weak significant strength of association. When we consider skin infection, the value obtained for the Fisher test is 0.0008, which indicates significant differences according the sex, although in this case the strength of association is moderately significant (Cramer’s V of 0.22). Therefore, we can say that the proportion of infected is not the same depending on whether the potential host is a man or a woman, assuming a significance level of 5 %. With regard to the number of specimens of Demodex folliculorum, ranged from 1 to 8 mites in eyes whereas the number of mites from skin ranged from 2 to 16, being the most infected 6 people with 12 mites. Concerning mixed infections, 32 people showed infections in skin face and eyelashes simultaneously (Table 1). Table 5 and Fig. 6 show the results of the Bayesian zero-inflated Poisson GLM fitted. Zero-inflated model split the process in two parts; the first is a summary of the full model and can be interpreted as a standard Poisson model. The second part (the zero component) predicts whether or not the outcome is a certain zero (the logistic part of the model). A common way of interpreting logistic regression models is to exponentiate the coefficients, which places the coefficients in an odds-ratio scale. With zero-inflated models, the logistic part of the model predicts non-occurrence of the outcome. In this regard, the zi-intercept of the model can be interpreted as follows: the baseline odds ratio of being among the never infected is exp. (0.02) =1.02. We found significant effects in all the categories for the two response variables considered in the model. People with infection in eyes, skin and both concerning people without any infection stage, increased significantly the total abundance of Demodex with coefficients and credible intervals ranged at values: 9.84 (6.01, 19.14), 10.51 (6.69, 19.82) and 10,96 (7.15, 20.28) respectively, (see Table 5 and Fig. 6). Fig. 5. Boxplot of men (M) and woman (W), with Demodex folliculorum species complex positive infection (eyelashes, skin or mixed infection), considering the use of contact lenses (Y =yes, N =no). Red points represent the median value. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 5 Parameter estimates for the Bayesian zero-inflated Poisson GLM. From left to right: Predictor variable, estimated coefficient, estimate error and the lower and upper credible 95 % intervals (CI). Non-overlapping CIs with zero imply a significant (p-value <0.05) difference between groups. ZI- intercept refers to the intercept of the zero-inflated model part. No infection and no contact lenses are the baseline categories for both categorical explicative variables. Predictor Estimate Est. Error. lower-95 % CI upper-95 % CI Intercept −8.54 3.65 −17.87 −4.71 ZI-intercept 0.02 0.02 0.00 0.07 Infect_Eyelashes 9.84 3.66 6.01 19.14 Infect_Skin 10.51 3.65 6.69 19.82 Infect_Both 10.96 3.65 7.15 20.28 Use of contact lens_Yes 0.18 0.09 0.01 0.36 F.J. M´ arquez et al. Parasite Epidemiology and Control 27 (2024) e00381 8 It is noteworthy that the average age of contact lens users in Spain is around 30 years, a figure almost equal to the average age of contact lens users worldwide (31 years). The overall percentage of silicone hydrogel soft contact lens fittings is 37 %. Women constitute the majority of contact lens users in Spain, representing around 59 %, a value slightly lower than that of the rest of the world (67 %) (Santodomingo et al., 2012). According to the model estimate, people who wear contact lenses significantly increased the total abundance of Demodex in relation to those who do not wear them, with a statistically significant parameter: 0.18 (0.01, 0.36) but with a minimal influence in the development of ocular infection (Fig. 6). Bayesian R 2 for zero-inflated Poisson GLM model explained 87.42 % of the variance of the number of Demodex, and VIFs values were all <3 for the explanatory variables in the model. The graphical goodness-of-fit of the model according to the “DHARMa” package and using the pp_check function of the “brms” package, can be found in the Supplementary Material, along with the R code for the entire statistical analysis. 6. Discussion 6.1. Morphometrical aspects The molecular identification of four phenotypes of human Demodex mites demonstrated long- and short-bodied D. folliculorum with finger-like terminus and D. brevis with finger or cone-like terminus (Zhao et al., 2013a). The molecular data permit to reclassification D. brevis with finger-like terminus, formerly classified as D. brevis, to be D. folliculorum and it might be a morphological variant of D. folliculorum classical form. We found significant differences in four of six parameters considered with the exception of podosomal length (Pd_L) and opisthosomal width (Op_W) (Table 2). The SVM allows delimiting the maximum margin hyperplane that provides a perfect classification of both populations (eyelash and skin) (Fig. 4). In the reviewed literature, epidemiological papers have only considered one of both habitats and have not been reported correlations between both populations. Demodex species from humans show a reduced transfer capacity among host and low dispersal rates (Palopoli et al., 2015). In this context it may be possible to hypotetized the segregation of two morphotypes in D. folicullorum as response to an adaptation on characteristics of two specialized microhabitats (hair follicles and sebaceous glands) that determine the difference registered in the size of each mite morphotype. Based on the morphometric results obtained in this work, and the statistical analysis applied, we can conclude that the two morphotypes of D. folliculorum, long and short body with finger-like terminal isolated from different habitats (eyelashes and skin), can be differentiated by applying a simple equation. Fig. 6. Influence of factors on the number of Demodex folliculorum species complex in young people. Estimated coefficients (and 95 % high-density intervals) of the covariates for the explanatory variables. 95 % high-density intervals overlapping 0 (solid vertical line) indicate a non-significant coefficient. No infection and no contact lenses are the baseline categories for both categorical explicative variables. F.J. M´ arquez et al. Parasite Epidemiology and Control 27 (2024) e00381 9