Prevalence, Distribution, and Factors Associated with Vector-Borne Pathogen Infections in Pet Dogs from Different Geoclimatic Zones in Sri Lanka

Vector-borne pathogens (VBPs) cause signi ﬁ cant diseases in dogs in the tropics. In Sri Lanka, the scarce availability of previous studies on canine VBPs has hampered an accurate evaluation of their prevalence in pet dog populations. In this study, we collected demographic, clinical, and environmental data together with whole blood from 423 pet dogs from three geoclimatic zones in Sri Lanka. All blood samples were screened using a previously validated multiplex qPCR assay to detect the six most prevalent canine VBPs in tropical Asia. Multivariable logistic regression models were used to investigate environmental and host factors as predictors of VBP infections. Overall, 254 dogs (60.1 % , 95 % CI: 55.3 – 64.6 % ) were infected with one or more VBPs. Babesia gibsoni was the most prevalent VBP with 37.4 % (95 % CI: 32.7 – 42.2 % ) of dogs infected followed by Hepatozoon canis (21.04 % , 95 % CI: 17.25 – 25.24 % ), haemotropic mycoplasmas (10.2 % , 95 % CI: 7.5 – 13.4 % ), Babesia vogeli (5 % , 95 % CI: 3.2 – 7.5 % ), Ehrlichia canis (4.5 % , 95 % CI: 2.7 – 6.9 % ), and Anaplasma platys (3.8 % , 95 % CI: 2.12 – 6.1 % ). Predictors of VBP infections included tick infestation for H. canis ( p ¼ 0 : 05) and A. platys ( p ¼ 0 : 01), as well as age for B. gibsoni ( p ¼ 0 : 01) and H. canis ( p ¼ 0 : 05) infection. Local breed ( p ¼ 0 : 004), male dogs ( p ¼ 0 : 001) and ﬂ ea infestation ( p ¼ 0 : 04) were signi ﬁ cantly associated with haemotropic mycoplasma infections suggesting direct-blood exchange through ﬁ ghting and ﬂ eas as a possible means of transmission for these pathogens. Clinical results suggest that B. gibsoni and E. canis caused clinically signi ﬁ cant disease, especially in exotic breeds such as German shepherds and Rottweilers compared to the local breeds ( p < 0 : 001). Measures such as educating pet dog owners on the importance of being vigilant on ectoparasite infestation of their pets, preventing pet dogs from interacting with stray or community dogs, and the compliant use of effective ectoparasiticides will be crucial for effective control of VBPs in pet dogs in Sri Lanka.


Introduction
Vector-borne pathogens (VBPs), Babesia gibsoni, Babesia vogeli, Hepatozoon canis, Ehrlichia canis, Anaplasma platys, and potentially haemotropic mycoplasmas are canine bloodborne pathogens transmitted by tick-vectors that have been identified as ubiquitous across tropical Asia [1][2][3][4].Infection with these pathogens can result in significant clinical disease from the direct effects of the pathogen itself and the host immune response with potentially fatal consequences.Babesia gibsoni and E. canis can cause severe pathology, whereas subclinical infection predominates in B. vogeli, H. canis, and haemotropic mycoplasma infections.However, all these VBPs can cause significant disease in immunocompromised individuals and can result in complex pathologies when present as coinfections [5].
Additionally, many of these studies [7,8,10,14] utilised microscopic-based methods for diagnosing VBP, which are notorious for their low sensitivity [15].Only one study utilised molecular assays to detect the prevalence of E. canis in stray dogs residing within the Colombo district [11], and another utilised immunodiagnostic assays for the seroprevalence of Anaplasma spp.and E. canis in Kalutara district [13], Sri Lanka.To date, predictors for VBP infections and their clinical impact on dogs residing in Sri Lanka have not been assessed limiting the translational value of the previous studies.
Here, we conduct a comprehensive, island-wide, crosssectional study using high-throughput molecular diagnostics to assess the prevalence and predictors associated with VBP infection in Sri Lankan pet dogs.Such data will assist in identifying the clinical significance of VBP infection and facilitate better diagnosis, treatment, and control of these pathogens.

Materials and Methods
2.1.Study Sites and Sampling.Sri Lanka, located between 5°55′ and 9°51′N latitude and 79°41′ and 81°53′E longitude, is an island nation with a tropical climate.The country can be divided into three climatic zones based on mean annual rainfall and geographical relief: the low-country dry zone, the low-country wet zone, and the mid-up country wet zone (Figure 1).The wet zone receives an average annual rainfall of over 1,750 mm, while the dry zone receives less than   Transboundary and Emerging Diseases 1,750 mm [16].Additionally, areas above 300m elevation are classified as the mid-up country, and those below 300 m are considered the low country [16].Dogs in Sri Lanka can be largely categorised into two groups as unowned (stray) or owned (pets).Stray dogs comprise the local dog breed and are exclusively free roaming.Owned dogs on the other hand are of local or exotic breeds such as Rottweilers and German Shepherds and their crosses and may be restricted within the confines of their owners' properties or allowed to roam unsupervised.Close contact and physical interactions between these stray and pet categories of dogs are common across the island.The average population density of 384 people per km 2 [17] in Sri Lanka was recorded to be proportional to the dog population density in several studies [18][19][20], meaning that a higher number of dogs resides in densely populated regions of the country.Accordingly, eight veterinary clinics/hospitals across the three geoclimatic zones (Figure 1) were selected based on veterinarians' willingness to participate in the study.Within each clinic, animals were randomly recruited, with exception of dogs requiring emergency attention.Where applicable, only one dog per owner was selected for inclusion in the study.
Sample size calculations were based on either estimating the prevalence of each VBP with 95% confidence or demonstrating freedom from the pathogen, whichever required the larger sample number, assuming a test sensitivity of 75% and specificity of 95% using the Shiny applet developed using the "epiR" package [21] and "Epitools" web platform (https:// epitools.ausvet.com.au/),respectively (Supplementary 1).

Collection of Data.
Data were collected during a 11month period from April 2020 to March 2021.Following owner consent, individual animal metadata including age, sex, neutering status, breed, history of vector-borne diseases as diagnosed by their veterinarian, as well as frequency of ectoparasitic treatment, "brand" (formulation), frequency of use, and duration from the last ectoparasiticide treatment given were collected from the owner of each participating dog to the best of their knowledge using a structured questionnaire (Supplementary 2).
Clinical manifestations such as poor body condition, presence of palpable splenomegaly and hepatomegaly, mucous membrane colour, body temperature, and palpable lymph node enlargement, were obtained through a physical examination by a veterinarian.Examination of the whole-body surface (including interdigital spaces) for ∼5 min was performed to identify tick, flea, and/or louse infestation.If at least one tick, flea, and/or louse was found, the pet was considered to have an active tick, flea, and/or louse infestation.The dog's body condition was assessed according to the WSAVA five scale BCS (body condition score) chart [22] and those with BCS <3 was categorised as having "low" BCS.
From each dog, individual whole blood samples (1-2 ml) were collected in a single sterile EDTA (ethylenediaminetetraacetic acid) tube through cephalic or lateral saphenous venepuncture.The collected blood samples were transported on ice and were stored at −20°C until DNA was extracted.

DNA Extraction and Molecular
Screening.Extraction of DNA from canine whole-blood was performed using the DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol at the University of Peradeniya, Sri Lanka.The extracted DNA was stored at −20°C until shipped to the University of Melbourne, Australia.The extracted DNA was screened in duplicates using a previously developed TaqMan ® probe-based multiplex realtime PCR assay comprising two separate fourplex reactions for vector-borne bacteria and protozoa, respectively [23].Together these fourplex reactions are able to detect six common VBPs targeting a partial region of the 18S ribosomal RNA gene for the protozoans B. gibsoni, B. vogeli, H. canis, and a hypervariable region of the 16S rRNA gene for the bacteria E. canis, A. platys, and haemotropic mycoplasmas, viz.Mycoplasma haemocanis, Candidatus Mycoplasma haemominutum, and Candidatus Mycoplasma haematoparvum.The optimised PCR assays were run as 10 µl reactions using QuantiNova ® Probe PCR Master Mix (Qiagen, Hilden, Germany) with 1 µl of template DNA in 96-well skirted white plates (Qiagen, Hilden, Germany) using a QIAquant 96 5plex real-time PCR thermal cycler (Qiagen, Hilden, Germany).Mammalian mitochondrial DNA and equine herpes virus DNA were used as extraction and internal controls, respectively, along with a no template reaction control (NTC).Appropriate gBlocks™ synthetic double-stranded DNA sequences (Integrated DNA Technologies, USA) of the pathogens tested were used as positive controls.The samples that had achieved endpoint (plateau phase) on or before the last two cycles and had a Ct lower or equal to 35 cycles in both replicates were considered positive.
2.4.Data Analysis.Data were checked and cleaned for statistical analysis with Microsoft Excel ® for Microsoft 365 MSO (Version 2212 Build 16.0.15928.20196)and R version 4.2.0 [24] using "dplyr" [25] and "janitor" [26] packages.For descriptive and univariable analysis, dogs were categorised in to six age groups according to Harvey et al. [27] and two breed categories: local and exotics.Dogs exhibiting one or more signs from pale mucous membranes, reduced appetite/ anorexia, fever, lymphadenomegaly, splenomegaly and hepatomegaly were considered as having clinical signs of uncomplicated cases of VBP infections [28].The compliance to ectoparasiticides by owners of each study dog were based on whether they were being treated for ectoparasites, whether the frequency and duration of treatment were compatible with the product used, and whether the approximate duration since the last treatment adhered to the recommended frequency for the product used.The scoring protocol for each scenario is indicated in Supplementary 1.With this scoring, a score of 3 corresponds to a compliant owner, and those with minimal and moderate compliance received a score of 1 and 2, respectively.Owners with poor compliance with ectoparasite treatment and untreated dogs received a score of zero.
Univariable associations of VBP infection and host factors, i.e., age, breed, sex, neuter status, presence of ticks/fleas, and other related factors, i.e., geoclimatic zone, and clinical Transboundary and Emerging Diseases manifestations were assessed using binomial logistic regression with generalised linear model (GLM), Pearson χ 2 test or Fischer's exact test, based on the structure of data using R version 4.2.0 [24].The odds ratios were calculated with 95% confidence using "oddsratio" function of the "epitools" package [29].Variables used in fitting multivariable models (Supplementary 1) were compared in pairs using the Cremers V test using "rcompanion" package [30] in R to determine the degree of correlation before including them in models to omit collinearity.
Random effects logistic and ordinal regression analyses were employed to analyse the risk of infection with individual pathogen and with one or more infections, respectively.All selected variables (Supplementary 1) were included as fixed effects in the generalised linear mixed-effect models (GLMM) and sampling veterinary clinics were used as random effects using "lme4" package [31] in R. Ordered proportional odds logistic regression (with ordered logit model) was performed with "MASS" [32] package in R. The "sjPlot" [33] package in R was used to calculate odds ratios for the models and to visualise models where necessary.Selection of the models were performed by the backward stepwise elimination to find the best fit [34].The Brent-Wald test in "brant" package [35] in R was used to determine whether the proportional odds assumption holds for the ordered proportional odds logistic regression model [36], where the Brent-Wald test output is p ≥ 0:05 for the final selected model variables, the model was considered to hold the proportional odds assumption.The R code for the functions used in this study is available at https://github.com/ushata/Tick-borne-pathogens-SL-dogs.

Sample Descriptives.
A total of 423 blood samples were collected from pet dogs from the up-mid country wet zone (n = 307), low-country wet zone (n = 75), and low-country dry zone (n = 41) in Sri Lanka.More than half of the dogs were local breed dogs and their crosses (n = 265), while the rest were of exotic breeds (n = 147).Of the dogs presented, 45.2% were healthy and were presented to the veterinarian for routine vaccination, health checks, or neutering.Data on age group, sex, neutering status, breed, tick, flea, and louse infestation, and ectoparasiticide usage of the study cohort according to the geoclimatic zone are summarised in Supplementary 1.

Use of Ectoparasiticides.
Nearly half (n = 189, 44.7%) of the participating dog owners claimed to use ectoparasiticides or repellents to control ticks, fleas, and lice, while nearly 35% (n = 145) did not (Table 4).The remaining dog owners were unaware whether their pets had been administered treatment for ectoparasite control.Propoxur powder (30.8%, n = 60), afoxolaner tablets (18.97%, n = 37), and subcutaneous ivermectin injections (10.3%, n = 20) were the most common ectoparasiticides or repellents used by the study subjects (Table 4).Of those who claimed to treat their pets with ectoparasiticides, 55.3% (n = 169) were minimally compliant with the recommended treatment protocol for effective ectoparasite control for their preferred product, and 4.5% (n = 19) showed moderate degree of compliance.Only one of 189 dogs that received ectoparasiticides demonstrated recommended treatment compliance (Table 4).

Discussion
This study reports the first comprehensive survey on the prevalence and predictors for canine VBPs in owned dogs in Sri Lanka.The study highlights the highly endemic nature of these VBPs, especially with respect to B. gibsoni and provides the first molecular evidence for the presence of canine haemotropic mycoplasmas and A. platys in Sri Lankan dogs.
In addition, this study attributes B. vogeli as the large Babesia species infecting dogs in Sri Lanka, which has to date, been likely misclassified as B. canis using light microscopy [7,14].This study revealed that over a third of owned dogs in Sri Lanka were positive for B. gibsoni compared to 15% detected by Weerathunga et al. [14] using light microscopy in Anuradhapura district.The prevalence of E. canis in owned dogs in Sri Lanka at 4.5% was significantly lower than the previous 17% reported for 'apparently healthy' stray dogs in the country using nested PCR [11] and 24.5% using microscopy by Weerathunga et al. [14].This discrepancy might be explained due to differences in study cohorts (owned versus stray dogs) in the former and the low specificity (high-false positivity rate) of microscopic identification of E. canis morulae in the latter.Identification of Ehrlichia morulae through light microscopy is challenging as they are transient in the blood and low in number [38] and can be confused with cells granules [39], contributing towards false positive results.With the exception of H. canis, the prevalence of the Rhipicephalus linneai-transmitted pathogens, B. vogeli, E. canis, and A. platys in this study were significantly lower than those reported in the surveys of stray dogs in neighbouring India [1,2].The prevalence of H. canis and haemotropic mycoplasmas in owned dogs in Sri Lanka are comparable to those reported in stray dogs in India [1].
Owing to COVID-19 related interprovincial travel restrictions, samples were sourced from only eight veterinary clinics/hospitals, with the majority sourced from a single geoclimatic zone, which may have introduced sampling and selection bias.However, we collected beyond the minimum number of samples required for the study and detected all six anticipated VBPs, even in pet dogs that received veterinary care.Therefore, despite the limitations, this study provides robust evidence of the burden of VBP in dogs in     Transboundary and Emerging Diseases Sri Lanka and potential predictors of infection.Historical information, such as ectoparasiticide usage and previous VBP pathogen infection, may be influenced by recall and behavioural biases from pet owners and inclusion of dogs seen in veterinary clinics/hospitals in this prevented the assessment of VBP infection rates in dogs not regularly receiving veterinary care.The multivariable analysis indicated local breeds and male dogs with observable flea infestation as a predictor for haemotropic mycoplasma infection compared with female dogs and exotic breeds without flea infestation.A higher prevalence of haemotropic mycoplasmas has been demonstrated to associated with fighting dogs [40][41][42] and those housed in kennels [42][43][44], where dog-to-dog contact is frequent.Although not significant, there was a positive correlation between dogs presenting with wounds, predominantly caused by dog bites, and haemotropic mycoplasma infection, further indicating dog fighting as a likely mode of transmission for this group of pathogens.This may indicate that males, and local dog breed that are less likely to be confined at home by their owner, may be at greater risk of engaging in dog fights and acquiring haemotropic mycoplasma infection.In addition to fighting, higher blood concentrations of androgenic hormones in entire male can also contribute to poorer immunity, which may increase their susceptibility to haemotropic mycoplasma infection [45,46].
The transmission of haemotropic mycoplasmas was initially proposed to be by R. linnaei [47], but without convincing evidence.Fleas, on the other hand, are known vectors for Mycoplasma haemofelis [48], a closely related feline-specific species to Mycoplasma haemocanis [49,50].Recently, transmission of haemotropic mycoplasma has been demonstrated in a population of dogs on ectoparasiticides and in the absence of arthropod vectors, strongly suggesting non-vectorial transmission for these pathogens [42].Interestingly, the canine haemotropic mycoplasmas were more likely to occur in dogs residing in the up-mid-country wet zone in contrast to all five vector-borne pathogens that were significantly more common in dogs in the low-country wet and dry zones.The up-mid-country wet zone experiences mean annual temperatures between 10°C and 25°C [51] and relative humidity between 55% and 90%, compared to higher mean annual temperatures (25°C-30°C) experienced in the low country wet and dry zones.According to Silverman et al. [52], the life span of 90% of unfed Ctenocephalides felis adults on average is between 8-22 days in temperature and humidity ranges of up-mid-country wet zone as opposed to 2-8 days in that corresponding to low country wet and dry zones [52].Furthermore, 61% (95% CI: 55-66%) of the dogs in the up-mid country in our study were infested with fleas compared to 40% (95% CI: 30-51%) in the low country.Nevertheless, in a previous study conducted in south-east Asia, neither flea (mainly Ctenocephalides felis) infestation was correlated with haemotropic mycoplasmas infection nor were these pathogens found in fleas collected from infected dogs [53].
We identified that the risk of infection of B. gibsoni increases with age.Possible explanations for such an outcome are multifactorial.The multimodal nature of transmission of B. gibsoni through its tick vector, direct infection through infected blood exchange (e.g., during dog fights [54] and blood transfusion), and transplacental transmission [55] can lead to frequent infection of dogs by this pathogen, with the former factors being cumulative with age.Furthermore, treatment of B. gibsoni usually fails in the complete elimination of the pathogen [56], even when the recommended azithromycin-atovaquone combination [57] is used, causing most dogs to remain subclinically infected for life.Given the study population's characteristics in Sri Lanka, dog fighting is also likely to play an essential role in transmitting this pathogen.However, transplacental transmission appears to be absent or negligible, given  Transboundary and Emerging Diseases 13 Conversely, for A. platys and H. canis, known to be transmitted by R. linnaei bites and ingestion, respectively, tick infestation was identified as a significant predictor of infection.An active tick infestation can increase the likelihood of tick ingestion and in turn, H. canis infection.In addition, the risk of H. canis infection increased with animal age.Additionally, louse infestation appears to be a predictor for A. platys infection.Owing to low prevalence of louse infestations these results are of limited value, although A. platys DNA was previously detected in lice from pups in Australia [58].
Treatment of H. canis is likely missed in most instances as the majority of infections are subclinical and, therefore, owners are less likely to seek veterinary treatment.Even if infected dogs are presented for treatment, it is necessary to administer an intensive treatment regime of fortnightly imidocarb dipropionate until elimination of the pathogen is achieved, which most pet owners are less likely to comply with, causing a persistent infection, which is likely to accumulate on a population level over time [59].For E. canis and B. vogeli, no significant predictors of infection were identified in our models.Even though the real-time PCR assay is reported to have over 95% diagnostic sensitivity [23], noncirculating, subclinical infections of VBPs, as well as latent or chronic (pancytopenic) phases of E. canis infections, could lead to under-reporting of the true prevalence of these pathogens [60].
Most veterinary clinics in Sri Lanka are not equipped with resources to perform haematological, biochemical, or PCR-based tests to diagnose VBP infections [61], with veterinarians relying on clinical presentation rather than proactively investigating the presence of VBPs in dogs exhibiting subclinical signs.This is noteworthy as more than half of the VBP infections in our study were subclinical or clinically unremarkable.Subclinical infections can act as a "ticking time bomb," for example in cases of latent ehrlichiosis [62] or in instances where VBPs can reactivate to cause acute disease during times of natural or iatrogenic immunosuppression [55,63,64].In addition, subclinically infected dogs can act as a continual source of infection for other dogs.Therefore, where possible, best practice recommendations should be followed by Sri Lankan veterinarians to screen and treat for VBPs, in particular B. gibsoni, in dogs presenting with a history of roaming, fighting and/or tick exposure, regardless of whether overt clinical signs are present.
Even though no breed-based differences were identified for VBP infection, exotic breeds were more likely to display clinical signs of VBP infection on physical examination compared to local dogs.There is no concrete evidence to explain this disparity, but available data suggest that host immunological and genetic factors may be responsible for this difference [65,66].In this study, pale mucosae indicating anaemia was the most significant clinical manifestations of B. gibsoni infection, while reduced or loss of appetite and presence of peripheral lymphadenomegaly were significant to a lesser extent.For E. canis, peripheral lymphadenopathy was associated with infection   [5].Indeed, there was an association between clinical signs and multiple VBP infections (p ¼ 0:054).Year-round tick control using an ectoparasiticide product that both repels and kills ectoparasites prior to feeding is the mainstay of preventing VBP infections in dogs in the tropics [70,71].The vast majority of owners used topical short-acting preparations that required a high frequency of administration to be effective (e.g., propoxur-based powders, pyrethroid-, amitraz-based shampoos), or that lacked proven ectoparasiticidal efficacy (e.g., herbal shampoos) [72].In addition, propoxur-resistant ticks have been identified in Sri Lankan cattle [73], but its efficacy against local R. linnaei populations remains unknown.Almost a fifth of dog owners used systemically acting products that require ectoparasite feeding to occur prior to kill (e.g., isoxazolines).Only a minority of pet owners used fipronil-based products that have proven efficacy for the prevention of VBPs in the tropics [71].Regardless of ectoparasiticide product of choice, almost all the owners did not follow the recommended treatment schedule for ectoparasiticides.In contrast, over 60% of the dog owners of highincome nations such as the United States of America are aware of the veterinary recommendations for the prevention of ticks and fleas on their pets [74].
These factors have likely contributed to the high prevalence of VBP infection observed in pet dogs in Sri Lanka, even in animals receiving ectoparasiticide treatment.Client choice of ectoparasiticide treatment is not only dependent on affordability, but also on product availabiity given that many effective commercial ectoparasiticides are not registered in Sri Lanka [61].These results highlight the need for veterinarians to follow best practice guidelines when advising clients on both ectoparasiticide choice and the importance of strictly following labelled ectoparasiticide treatment frequency for the prevention of VBP in Sri Lanka.

Conclusions
This study comprehensively reports the prevalence and associated factors for VBP infections in Sri Lankan pet dogs and identifies key gaps in their control.To prevent VBP infections, veterinarians should recommend appropriate and effective ectoparasiticides according to their labelled recommendations.Sri Lankan veterinarians should \provide dog owners with information on practices that would minimise the transmission risk for B. gibsoni and haemotropic mycoplasmas and be cautious when using canine blood for transfusions.Screening for VBP infections is recommended in such instances, preferably using molecular diagnostics.

12 Transboundary and Emerging Diseases TABLE 4 :
Route and frequency of administration (A), preparations of ectoparasiticides (B), and the overall compliance of ectoparasiticide usage by the pet owners (C) within this study.

TABLE 1 :
Prevalence of single and mixed vector-borne pathogen infections detected in dogs in this study.

TABLE 2 :
Single and mixed vector-borne pathogen infection in dogs in Sri Lanka according to geoclimatic zone, age group, sex, neutering status, breed group, tick, flea,
Bold values signify the statistically significant predictors.

TABLE 5 :
Summary of univariable statistical associations of vector-borne pathogen (VBP) and clinical manifestations in study dogs.

TABLE 6 :
Odds ratios and coefficient estimates of the generalised linear mixed models and ordered logistic regression model for vector-borne pathogen (VBP) infection in Sri Lankan pet dogs.

TABLE 6 :
[68,69]ed.Variance of the intercepts of the random effects of dog clinics.Transboundary and Emerging Diseases whereas pale mucosae and poor body condition were associated to a lesser extent.Peripheral lymphadenomegaly resulting from increased proliferative activities in the lymph nodes by E. canis antigens[67]is a well-known clinical sign during acute canine ehrlichiosis.Severe bone marrow suppression and haemolysis are potentially fatal manifestations in dogs with E. canis and B. gibsoni, which are likely complicated by host immune responses[68,69].Infections with B. vogeli, H. canis, A. platys, and haemotropic mycoplasmas did not demonstrate clinically overt disease.Noninfectious causes (e.g., heart-, hepatic-or kidney-disease, neoplasia, etc.) exacerbate clinicopathological manifestations of VBPs, but so can the presence of coinfections with other VBPs † Standard error.‡ Confidence interval with 95% confidence.§ Significance considered at 0.05.¶ Interclass correlation coefficient.α