Heylen et al. Parasites Vectors          (2021) 14:576  
https://doi.org/10.1186/s13071-021-05014-8 Parasites & Vectors
RESEARCH Open Access
A community approach of pathogens 
and their arthropod vectors (ticks and fleas) 
in dogs of African Sub-Sahara
Dieter Heylen1,2,3*, Michael Day4^, Bettina Schunack5, Josephus Fourie6,7, Michel Labuschange8, 
Sherry Johnson9, Samuel Maina Githigia10, Foluke Adedayo Akande11, Jahashi Saidi Nzalawahe12, 
Dickson Stuart Tayebwa13, Ortwin Aschenborn14, Mary Marcondes15 and Maxime Madder16,17 
Abstract 
Background: Arthropod-borne pathogens and their vectors are present throughout Africa. They have been well-
studied in livestock of sub-Saharan Africa, but poorly in companion animals. Given the socio-economic importance 
of companion animals, the African Small Companion Animal Network (AFSCAN), as part of the WSAVA Foundation, 
initiated a standardized multi-country surveillance study.
Methods: Macro-geographic variation in ectoparasite (ticks and fleas) and pathogen communities in dogs was 
assessed through molecular screening of approximately 100 infested dogs in each of six countries (Ghana, Kenya, 
Nigeria, Tanzania, Uganda and Namibia), both in rural and urban settings. The most important intrinsic and extrinsic 
risk factors within the subpopulation of infested dogs were evaluated.
Results: Despite the large macro-geographic variation in the dogs screened, there was no consistent difference 
between East and West Africa in terms of the diversity and numbers of ticks. The highest and lowest numbers of ticks 
were found in Nigeria and Namibia, respectively. Most often, there was a higher diversity of ticks in rural habitats than 
in urban habitats, although the highest diversity was observed in an urban Uganda setting. With the exception of 
Namibia, more fleas were collected in rural areas. We identified tick species (including Haemaphysalis spinulosa) as 
well as zoonotic pathogens (Coxiella burnetti, Trypanosoma spp.) that are not classically associated with companion 
animals. Rhipicephalus sanguineus was the most abundant tick, with a preference for urban areas. Exophilic ticks, such 
as Haemaphysalis spp., were more often found in rural areas. Several multi-host ticks occurred in urban areas. For R. 
sanguineus, housing conditions and additional pets were relevant factors in terms of infestation, while for a rural tick 
species (Haemaphysalis elliptica), free-roaming dogs were more often infested. Tick occurrence was associated to the 
use of endoparasiticide, but not to the use of ectoparasiticide. The most prevalent tick-borne pathogen was Hepato-
zoon canis followed by Ehrlichia canis. High levels of co-parasitism were observed in all countries and habitats.
Conclusions: As dogs share a common environment with people, they have the potential to extend the network of 
pathogen transmission to humans. Our study will help epidemiologists to provide recommendations for surveillance 
and prevention of pathogens in dogs and humans.
*Correspondence:  dheylen@itg.be
^Michael Day—in memoriam
1 Eco-Epidemiology Group, Department of Biomedical Sciences, Institute 
of Tropical Medicine, Antwerp, Belgium
Full list of author information is available at the end of the article
© The Author(s) 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which 
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the 
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or 
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line 
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory 
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this 
licence, visit http://c reati vecom mons.o rg/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco 
mmons. org/p ubli cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Heylen et al. Parasites Vectors          (2021) 14:576 Page 2 of 20
Keywords: Dog, Sub-Sahara Africa, Ticks, Fleas, Vector-borne pathogens, Ixodes, Haemaphysalis, Rhipicephalus, 
Amblyomma, Coxiella burnetii
Background measures to control (zoonotic) diseases, such as the 
More than half (around) 55% of the world population establishment of proper treatment strategies and preven-
currently live in urban areas, with estimates for 2050 ris- tion, relies on the elucidation of pathogen and reservoir 
ing to 68% and close to 90% of this rapid increase taking hosts in a given area [3]. Given the importance of small 
place in Asia and Africa [1, 2]. In sub-Saharan Africa, companion animals, we initiated a multi-country sur-
expansion of farmland and urbanization—partly due veillance study on flea and tick communities and their 
to economic growth—are two of of the most drastic pathogens in sub-Saharan Africa. As part of the African 
and widespread manifestations of human-driven envi- Small Companion Animal Network (AFSCAN), which 
ronmental change. Human-mediated environmental focuses on enhancing companion animal veterinary 
alterations, such as deforestation and establisments of care across Africa through the creation of a sustain-
settlements in natural ecosystems, give rise to increased able veterinary network for Africa, we have attempted 
risk of exposure to vector-borne pathogens and new to identify the vector-borne pathogens and determine 
opportunities for novel vector-borne transmission cycles and ectoparasite status of dogs in both rural and urban 
[3]. Green infrastructures are needed for livestock, but areas in six African AFSCAN countries: Ghana, Kenya, 
they are promoted in urban areas as solutions for a wide Nigeria, Tanzania, Uganda and Namibia. Based on the 
range of services, including water management, air qual- collection of biological samples (ectoparasites, blood and 
ity, recreation services related to biodiversity, among oth- serum) we attempted to answer the following questions: 
ers [4, 5]. However, green (open) spaces near crowded (1) To which extent do ectoparasite and their pathogen 
areas are associated with increased human exposure risk communities vary macro-geographically? (2) Are parasite 
to wildlife-associated parasites and pathogens [see 6] communities in urbanized areas similar to those of rural 
and, although reversing exposure to wildlife itself, there is areas? (3) Which additional extrinsic risk factors and host 
increased exposure through domestic animals [7]. characteristics (age, sex, health status, anti-parasite treat-
Dogs and cats are implicated in the changing epidemi- ment) are related to ectoparasite infestations and patho-
ology of pathogens of public health concern [3, 7]. The gen prevalence (within the group of ectoparasite-infested 
role of companion animals in vector-borne diseases in animals)?
sub-Saharan Africa has not been addressed by the One 
Health organization in a standardized macro-geographic Methods
way. While veterinarians in these countries work hard, Study design and site
their numbers are few, they often work in geographi- This was a multi-site field survey to establish the current 
cally isolated areas, with limited resources, and they community of the most important dog (Canis familiaris) 
might be hampered by limitations in training. The need ectoparasite species (ticks and fleas) and vector-borne 
has emerged to properly understand urban and rural pathogens. Approximately 100 ectoparasite-infested dogs 
zoonotic disease risk areas. The socio-economic value per country (Ghana, Kenya, Nigeria, Tanzania, Uganda 
of pets has increased in importance in recent decades, and Namibia) were screened and sampled in urban and 
especially for sedentary household and farmers. Fleas rural habitats (Fig. 1). In general, urban areas are defined 
and ticks are the most common ectoparasites of dogs, as cities with a large population (> 20,000 inhabitants) 
and both vector a wide array of vector-borne patho- and an extensive housing infrastructure (mainly offices, 
gens that cause diseases such as borreliosis, bartonel- markets), an elaborate network of transportation and 
losis, ehrlichiosis, rickettsiosis and anaplasmosis. They access to piped water, modern housing and electricity. 
are both a nuisance and a substantial threat to canines, Rural areas, on the other hand, are sparsely populated, 
both directly and indirectly, through the pathogens they the infrastructure and housing are poor and in most 
transmit. While the former pertains to clinical signs of cases there is no piped water, no electricity and a poor 
physical damage, such as wounds and rashes due to bites, public transportion infrastructure. The main activities in 
the latter often relates to tick-borne diseases in wild and rural areas are associated with agriculture, and dog own-
domestic dogs [7–9]. ers rarely  give veterinary care to dogs. All sampling and 
Local tick and flea abundances depend in a multi- treating of dogs in the study occurred during the rainy 
faceted way on the presence of multiple hosts in a suit- season, except for Namibia (for timing of sampling, see 
able ectoparasite habitat. Implementation of effective Additional file 1: Fig. S1).
H eylen et al. Parasites Vectors          (2021) 14:576 Page 3 of 20
Fig. 1 Overview of the sampling locations in the six African countries (Ghana, Kenya, Nigeria, Tanzania, Uganda and Namibia). Locations given in 
blue and red indicate rural and urban habitats, respectively
Two regions were sampled in Ghana: the Greater animals. In Uganda, the two urban areas sampled were 
Accra region (GAR) in the south, and Akumadan in the located around the capital city of Kampala and Wakiso 
Ashanti region of the country. In GAR, samples were district, and the two rural areas were located in Mbarara 
obtained from veterinary clinics in Accra (East Legon, municipality in the western region of Uganda and Iganga 
Haatso-Atomic, Dome, Madina) and Tema metropo- municipality in the eastern region. Wakiso district encir-
lis. In Akumadan, the settlements from which the pets cles Kampala city, forming peri-urban parts of the capi-
were brought were Nkwakwaa, Asempanaye, Asuo- tal city. Most of the dogs in Wakiso and Kampala lived 
suo and Afrancho; these are rural and agricultural areas in fenced houses for purposes of either companionship 
with moist semi-deciduous forest and thick vegetation or security. In comparison, the rural area of Mbarara dis-
cover and undergrowth. In Kenya, the urban areas sam- trict is located in the cattle corridor and dogs there have 
pled were Nairobi (capital city) and Mombasa (coastal a wide area of farm land to roam as they herd and guard 
urban town); the dogs sampled here were well kept and cattle, with possibility of interacting with wildlife, while 
had access to housing, good welfare facilities and veteri- the rural area of Iganga is an agricultural base for pro-
nary care. The rural area (Narok) sampled in Kenya was duction of cereals including millet, maize, beans, ground-
a pastoral area where dogs accompany livestock to graz- nuts and rice, and dogs guard homes and often follow 
ing areas and have the potential to interact with wild the owners to the farmed areas and gardens. In Nigeria, 
Heylen et al. Parasites Vectors          (2021) 14:576 Page 4 of 20
one urban and one rural setting was sampled from each respectively. For each dog the cumulative number of 
of the northcentral and southwest geopolitical division ticks was determined (the scores of all body areas were 
of Nigeria. The rural areas were areas occupied mainly summed), and for the statistical analyses, three catego-
by farmers with no amenities. In Tanzania, the urban ries were created: the ‘No’ (cumulative number of ticks: 
areas selected were in the region of Morogoro, and the 0), ‘Light infestation’ (1–3), ‘Moderate infestation’ (4–10 
rural areas were located between Morogoro and Dar es ticks) and ‘Severe infestation’ (> 11) categories. For fleas, 
Salaam, the capital. The urban and rural areas selected an estimate for the entire animal was obtained as follows: 
in Namibia were located between Walvisbaai and Het- ‘No’ (0 fleas), ‘Light infestation’ (1–10 fleas), ‘Moderate 
tiesbaai, in the eastern part of the country bordering the infestation’ (11–50 fleas) and ‘Severe infestation’ (> 50 
Atlantic Ocean, with the urban sites situated more in the fleas). Up to 30 ticks were collected from each animal and 
center of the cities whereas the rural areas were more placed into a plastic tube with screwcap that contained 
inland. 70% ethanol. As many fleas as possible were collected 
In order to incentivize owners to provide animals from the animal and placed into the same collection jar 
for sampling, sampling was performed in parallel with as the ticks (i.e. one jar per animal).
providing free rabies vaccination and free ectopara-
site prevention concomitant therapy. Field assessment Blood collection and processing
screening efforts involved obtaining owner consent and Whole blood samples were collected using a syringe and 
gathering animal details, performing physical examina- the appropriate needle. Blood was used for the prepara-
tions, scoring of tick predilection sites and specimen tion of a Whatman® FTA® card (GE Healthcare, Chi-
sampling (ectoparasites, blood). Laboratory assessments cago, IL, USA) on which blood was preserved for DNA 
involved the analysis of samples collected in the field to analyses. Serum was used for the IDEXX’s 4Dx Plus 
identify the ectoparasite species collected (ticks, fleas) kits (IDEXX Laboratories, Inc., Westbrook, ME, USA) 
and the identification of vector-borne parasites. All screening tool for Ehrlichia canis/ewingii, Anaplasma 
sampling equipment, data forms, and IDEXX tests (see phagocytophilum/platys, Borrelia burgdorferi and Diro-
section  Blood collection and processing) were cen- filaria immitis following the manufacturer’s manual. 
trally supplied. Molecular analyses (FTA cards [see sec- Serum was obtained by centrifuging the collected blood 
tion  Blood collection and processing], identification of after clotting in plain collection tubes; the serum was 
vector species and pathogen screening) were performed stored frozen at −  20  °C in plastic screw-cap tubes for 
centrally in the same laboratory (Clinomics, Bloemfon- future research.
tein, South Africa). IDEXX kits were locally interpreted.
Vector‑borne pathogen identifications
Inclusion criteria Blood samples were shipped using FTA card technol-
In urban areas, each investigator established a link with a ogy. The cards were punched (diameter of punches: 3 × 
veterinary practice, and privately owned dogs visiting the 5  mm) and subjected to DNA isolation procedures. Per 
veterinarian were sample during the visit. In rural areas, dog, five ticks (or < 5 if fewer ticks were found) and 5 fleas 
where most dogs were likely to be free-roaming and/or (or < 5 if fewer fleas were found) were randomly sampled 
community-owned dogs, sampling was performed based from the jar for molecular identification using multiplex 
around the rabies vaccination/ectoparasite control provi- PCR. Ticks and fleas collected from each animal were 
sion. No sampling was performed at animal shelters. No pooled separately for DNA isolation. These samples were 
restriction to breed or age was made. For each registered homogenized by bead-bashing before being process-
dog, sex, age, weight, body condition score (5-point- ing using the MagMAX™ DNA Multi-Sample Ultra Kit 
scale: very thin [1], underweight, ideal, overweight, obese (Thermo Fisher Scientific, Waltham, MA, USA) accord-
[5]), housing (free-roaming, yard, indoor), parasiticide ing to the manufacturer’s specifications and eluted with 
treatment (ectoparasiticide and deworming drugs) and 100 µL elution buffer.
presence of other companion animals were recorded Blood samples and ectoparasites were subsequently 
using a standardized data capture form (see Additional screened using PCR techniques for the presence of fol-
file 2: Capture form). lowing tick-borne pathogens: Babesia rossi, Babesia 
canis, E. canis, Ehrlichia chaffeensis, A. platys, A. phago-
Ectoparasite burden assessment and collection cytophilium, Rickettsia conorii, Rickettsia africae, Cox-
Seven different body areas of each dog were system- iella burnetti and Hepatozoan canis. Blood samples were 
atically screened for ticks, with a burden score assigned additionally screened for a mosquito-borne (Dirofilaria 
individually to each body area  : 0 indicating the absence immitis) and a tsetse fly-borne (Trypanosoma spp.) 
of ticks and 1, 2 and > 3 indicating 1, 2 and > 3 ticks, pathogen.
H eylen et al. Parasites Vectors          (2021) 14:576 Page 5 of 20
Table 1 Overview of the targets and their respective DNA templates used in multiplex qPCR assay screenings
Target Canine blood Tick Flea Limit of detection (copies/ References
PCR)
Babesia rossi X X 5 [24]
Babesia canis X X 5 [24]
Ehrlichia canis X X 5 [25]
Ehrlichia chaffeensis X X 5 [26]
Anaplasma platys X X 16 [27]
Anaplasma phagocytophilum X X 9 [28]
Rickettsia conorii X X 8 [29]
Rickettsia africae X X 8 [29]
Coxiella burnetti X X 8 [30]
Hepatozoon canis X X 5 In-houseb
Dirofilaria immitis X 8 [31]
Trypanosoma spp.a X 5 [32, 33]
Bartonella henselae X 5 [34]
Mycoplasma haemofelis X 16 [35]
Babesia felis X 8 In-houseb
Dipylidium caninum X 8 [36]
a viz. Trypanosoma vivax, T. congolense, T. evansi and T. brucei
b Hydrolysis probe was designed in-house (Clinomics, Bloemfontein, South Africa)
Table 2 Overview of primers and probes used for the in-house qPCR-screenings of three pathogenic agents
Target In-house forward primer In-house reverse primer In-house hydrolysis probe
Hepatozoon canis GGC AGTG AC GGT TAA CGG GGG GCA CCA GACT TG CCC TCCA ATTG CCG GAG AGG GAGC CT GAG AAA CGG 
Dirofilaria immitis CTTT GG AAT ATG TGT TTT TTTG GA GAG CCCT C
Babesia felis AAG AAG CTC GTA GTT GAA TTTC TG CC GAG AAG CCG AAG CAAC AC AAAT CC AG TGC GTT TTCC GAC TGG CT TGGCA 
For all PCRs, the final forward and an reverse primer concentrations were 400 nM. The final probe concentration was 200 nM
The isolated DNA (5  µl) served as template in 15-µl Fisher Scientific) to determine which samples had detect-
hydrolysis probe-based multiplex quantitative PCR able levels of target DNA. All qPCR runs included a 
(qPCR) assays using the Luna® Universal Probe qPCR DNA-negative extraction control, a host-negative control 
Master Mix (New England Biolabs Inc., Ipswich, MA, which indicated that the assays did not detect host DNA, 
USA) to detect the target of interest according to the host a no-template control and a positive control.
(canine blood) and sample type (ticks, fleas). A universal Primers and probes (H. canis, D. immitis and Babesia 
thermal cycling program was used for all multiplexes, felis) were designed using Geneious (http:// www.g enei 
excluding the Dipylidium caninum plex which had an ous. com/) and validated in silico using sequence data 
extended elongation time due to the longer expected available on GenBank (Table 2).
amplicon size. The targets and their respective DNA tem-
plates are shown in Table 1. Vector identification
All extracted DNA samples were subjected to a first- The PCRs were performed using the Q5® Hot Start High-
round screening using a positive extraction control to Fidelity 2× Master Mix (New England Biolabs, Inc.) in 
assess DNA isolation and an internal amplification con- 10-µl reaction volumes containing 2 µl of DNA isolated 
trol to assess template-derived inhibition of the PCR. In from tick/flea using primers 5′-AAA GAT GAC CAAA CT 
cases where neither the internal amplification control nor TGA TCA TTTA GAGG-3 and 5′-TCG ATG AAG AAC 
any other targets were detected, the samples were diluted GCA GCC AGCT-3′ at a final concentration of 500  nM 
1:1 using 10% Chelex Resin (Bio-Rad laboratories, Her- each which amplifies the internal transcribed spacer 
cules, CA, USA). The results were obtained and analyzed 1 (ITS1) region of the ticks and fleas. Thermal cycling 
using QuantStudio™ Real-Time PCR software (Thermo entailed a polymerase activation step at 98 °C for 2 min; 
Heylen et al. Parasites Vectors          (2021) 14:576 Page 6 of 20
then 30 cycles of 98 °C for 10 s, 65 °C for 20 s and 72 °C in statistical power for each of the tests; and (ii) only 
for 75  s; with a final extension step at 72  °C for 5  min. those variables that were highly significant (P < 0.01) were 
The PCR products were sequenced and analyzed using considered to be the main result in the Discussion sec-
the Basic Local Alignment Search Tool (BLAST) for tion and Abstract. A variable that explained part of the 
identification. variation, though in a less significant way (P < 0.05), was 
For ticks which could not be identified using this left in the models to correct for its confounding effect 
sequenced region, primers which amplified the 16S ribo- and to provide further inspiration for future studies. For 
somal mitochondrial DNA (mtDNA) of the ticks were all analyses, a stepwise backward selection procedure was 
used [3]. The reactions were performed using the Plati- used to select the best model. At each step we excluded 
num™ SuperFi II PCR Master Mix (Thermo Fisher Sci- the fixed factor with the highest non-significant P-value 
entific) in 10-µl reactions containing 2 µl of DNA isolated (P > 0.05), re-ran the model and examined the P-values 
from ticks with the primers at 500  nM final concentra- of the fixed factors in the reduced model. Model reduc-
tion each. Thermal cycling entailed a polymerase activa- tion continued until only significant factors (P < 0.05) and 
tion step at 98 °C for 2 min; then 30 cycles 98 °C for 10 s, their lower order interaction terms were left [11]. For the 
60  °C for 45 s and 72  °C for 30 s; with a final extension statistical comparison of the parasite community, Fisher’s 
step at 72 °C for 5 min. exact tests were executed whereby the species distribu-
tion (in the population of parasitized individuals) was 
Statistical analysis compared between habitat types (urban vs rural) and 
For the tick- and flea-infested subpopulations, the pro- countries. In addition, the Shannon diversity index was 
portions of infested dogs per ectoparasite taxon and their computed [12]. All prevalence estimates are reported as 
infestation intensities were compared between countries the mean ± standard error (SE). All data management 
(Ghana, Kenya, Nigeria, Tanzania, Uganda and Namibia) and statistical analyses were performed in SAS v 9.3 (SAS 
and urbanization level (urban vs rural), as well as the pro- Institute, Cary, NC, USA).
portion of pathogen-infected ectoparasite batches. We 
emphasize that for each country, urban and rural set- Results
tings are different (see descriptions above), meaning that Ectoparasites
a generalized continent-wise comparison ‘urban versus Of all infested dogs examined (N = 584), 95.4% had ticks 
rural’ has little epidemiological relevance. For the flea- and 51.9% fleas (47.3% were co-infested with both ticks 
and tick-infested dogs (overall number: 584), the above- and fleas). In total, 13 tick and three flea taxa were identi-
mentioned proportions and intensities were included as fied based on the ectoparasite’s DNA. Higher ectopara-
response variables in models with the following explana- site diversity was found in rural areas compared to urban 
tory variables: the individual’s intrinsic (age, sex, body areas (see Shannon index, Table 3), with the exception of 
condition, deworming drugs and ectoparasiticide) and Uganda and Namibia. The highest and lowest diversity of 
extrinsic risk factors (pet density and housing condi- ectoparasites was found in rural Ghana (Shannon index: 
tions). For the proportion of pathogen-infected hosts (i.e. 1.60; 10 different taxa identified in 42 infested individu-
based on blood and serum screenings) the pooled sam- als) and urban Nigeria (Shannon index: 0.44; 4 taxa in 
ple (i.e. all sampled individuals, irrespective of the type 51 individuals), respectively. Ectoparasite communities 
of ectoparasite they were infested with) was considered (fleas and tick species; Additional file 1: Table S1) signifi-
(overall number: 601), assuming that past infections not cantly differed among each country (Fisher’s exact tests; 
necessarily relate to current ectoparasite status. For this for all pair-wise comparisons among countries P < 0.001). 
purpose, generalized estimation equation models (GEE) In the following sections we report on the dogs’ geo-
were fitted to the data [see 10], taking into account the graphic occurrence and extrinsic and intrinsic exposure 
statistical dependence of observations in the same areas. risk factors, all of which are related to ectoparasite preva-
The residuals for burden categories and pathogen pro- lence and infestation intensity.
portions were assumed to follow a binomial distribution 
(logit-link, in ordinal and logistic regression, respec- Tick infestation
tively). Because of the limited amount of independent Prevalence
data as well as the high number of tests on the same set Within the subpopulation of ectoparasite-infested dogs, 
of plots, the following model restrictions were imposed ticks were more often found than fleas (Table 3). Conse-
on models that included extrinsic and intrinsic risk fac- quently, the among-country variation in tick prevalence 
tors: (i) no interaction terms among the main explana- was low and contrasts between urban and rural areas 
tory variables were fitted as adding these to the model were small (all P-values > 0.05). Literally all examined 
would lead to (almost) saturated models and reductions dogs of Tanzania, Nigeria and Namibia were infested with 
H eylen et al. Parasites Vectors          (2021) 14:576 Page 7 of 20
Table 3 Tick and flea prevalence and intensity in infested dogs of six African countries
Ticks and Fleas Overall Tanzania (%) Kenya (%) Uganda (%) Nigeria (%) Ghana (%) Namibia (%)
prevalence 
(%) Rural Urban P Rural Urban P Rural Urban P Rural Urban P Rural Urban P Rural Urban P
Ticks
 Rhipicephalus sanguineus 67.5 71.1 91.5 ns 27.6 61.5 ** 2.3 47.5 ** 94.7 96.1 ns 78.4 85.7 * 90.3 83.3 ns
 R. appendiculatus 0.4 0.0 0.0 0.0 0.0 2.3 2.5 0.0 0.0 0.0 0.0 0.0 0.0
 R. simus 2.0 0.0 2.1 9.2 0.0 0.0 0.0 1.8 0.0 0.0 0.0 0.0 5.6
 R. microplus 0.2 2.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 R. senegalensis 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.0 0.0 0.0 0.0 0.0
 Rhipicephalus spp. 6.7 26.7 6.4 ** 0.0 0.0 2.3 17.5 * 3.5 0.0 2.7 0.0 9.7 19.4 ns
 Haemaphysalis elliptica 6.5 0.0 0.0 18.4 3.9 * 11.4 10.0 ns 8.8 7.8 5.4 0.0 0.0 0.0
 H. leachi 0.6 0.0 0.0 0.0 0.0 0.0 0.0 5.3 0.0 0.0 0.0 0.0 0.0
 H. spinulosa 1.5 2.2 2.1 0.0 0.0 11.4 2.5 ns 0.0 0.0 0.0 0.0 0.0 0.0
 Haemaphysalis spp. 17.3 0.0 0.0 56.6 26.9 * 54.6 22.5 ** 3.5 0.0 18.9 2.0 * 0.0 0.0
 Amblyomma variegatum 0.4 0.0 0.0 0.0 0.0 0.0 0.0 1.8 0.0 2.7 0.0 0.0 0.0
 Amblyomma spp. 0.9 2.2 0.0 0.0 0.0 2.3 0.0 0.0 0.0 8.1 0.0 0.0 0.0
 Ixodes sp. 0.6 2.2 2.1 0.0 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0
 Tick 95.5 100.0 100.0 ns 94.7 88.5 ns 81.8 95.0 ns 100.0 100.0 ns 97.3 87.8 ns 100.0 100.0 ns
 Intensity 6.8 21.2 ns 22.5 13.0 ns 12.2 19.2 ns 28.1 49.0 * 8.1 46.3 * 19.4 4.7 ns
 Shannon index 0.9 0.5 * 1.2 0.8 *** 1.2 1.4 *** 0.8 0.4 ns 1.2 0.1 ** 0.3 0.7 ns
Fleas
 Ctenocephalides felis 53.7 71.1 51.0 ns 75.4 48.2 ns 85.1 41.7 *** 52.5 5.9 *** 53.3 45.8 ns 12.5 28.6 ns
 Echidnophaga gallinacea 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 3.3 0.0 0.0 0.0
 Echidnophaga sp. 3.7 0.0 2.0 8.7 0.0 0.0 0.0 0.0 0.0 6.7 2.1 0.0 20.0 ***
 Flea 55.6 71.1 51.0 * 76.8 48.2 ** 85.1 41.7 *** 52.5 5.9 *** 56.7 47.9 * 12.5 45.7 **
 Intensity 6.45 31.8 * 13.6 40.0 * 2.3 5.0 0.0 0.0 9.5 14.3 ns 0.0 0.0
 Shannon index 0.00 0.2 ns 0.3 0.0 ns 0.0 0.0 0.0 0.0 0.5 0.2 ns 0.0 0.7 ns
Co-infestation 47.3 71.7 50.9 ** 71.1 37.0 ** 68.0 36.0 ** 52.6 5.9 * 54.8 35.7 ** 12.5 45.5 ***
Shannon i ndexa 1.25 1.0 ns 1.5 1.1 ** 1.3 1.6 *** 1.2 0.4 *** 1.6 0.8 ** 0.5 1.3 *
Investigated dogs (N) 584 46 53 76 27 50 50 57 51 42 56 32 44
For each dog, a single extraction was made of a pooled set of ticks and/or fleas that was subsequently screened for the presence of DNA belonging to a particular tick and flea species. Next, the percentage of extracts (i.e. 
dogs) containing DNA of a specific taxon was derived, within the population of infested dogs. For statistical outcomes on pairwise macrogeographic differences, see Fig. 2
***P < 0.001, **P < 0.01, *P < 0.05, ns (not-significant) P > 0.05
Habitat differences (rural vs urban) are investigated for countries with a presence of at least 10% in one of its habitats
a As a measure of species diversity, a Shannon diversity index and accompanying significance level of Fisher’s exact test are provided
Heylen et al. Parasites Vectors          (2021) 14:576 Page 8 of 20
Fig. 2 Macro-geographic variation in ectoparasite prevalence. Percentages within the population of infested dogs parasitized with the most 
common tick (black and gray shading) and flea (red and blue shading) taxa (overall prevalence per taxon > 5%; see Table 3). For each taxon, the 
same letters above columns indicate that the the contrast between countries is not statistically different from zero
at least one tick (prevalence 100%), which differed from level.  Rhipicephalus sanguineus was by far the most 
other locations that showed a higher proportion of flea- dominant in the pool of identified ticks (64.5%). Signifi-
infested-only dogs. Prevalences of co-infested individu- cant proportions of Haemaphysalis elliptica within the 
als (fleas and ticks) significantly varied among locations tick community were found in Kenya, Uganda, Nige-
(range: 5.9–71.7%; χ2 = 33.85; df = 5; P < 0.001; Table  3), ria and rural Ghana. In Uganda, nine Haemaphysalis 
with the number of co-infested animals overall generally spinulosa-infested dogs were sampled (rural: 13.2%; 
being higher in rural areas  (Logitrural-urban = 0.97 ± 0.18; urban: 9.8% of identifications). A large proportion of 
Z = 29.70; P < 0.001), except  in Ghana (no difference; ticks could only be identified at genus level, which was 
P = 0.059) and Namibia where the rural prevalence was especially the case for Haemaphysalis spp. collected 
lower (Rural < Urban:  Logitrural-urban = −  1.76 ± 0.61; in rural areas. Less than 1% of the tick-infested dogs 
Z = 8.24; P < 0.004). We refer to Fig. 2 for pairwise-com- sampled contained DNA that belonged to the genera 
parisons between countries. Amblyomma and Ixodes. Co-infestations, i.e. more than 
one species feeding on the same host individual, were 
Community observed in 8.9% of tick-infested dogs, and was domi-
With respect to the tick’s community (based on all nated by the co-occurrences with R. sanguineus (Addi-
DNA identifications obtained from the pool of ticks of tional file 1: Table S2).
each dog; see Fig. 3 and Additional file 1: Table S1), the 
Shannon diversity index was higher in rural areas, with 
the exception of Namibia (rural: 0.32; urban: 0.66) and Infestation intensity
Uganda (rural: 1.21; urban: 1.40). Differences were found The following insights were obtained regarding tick 
to be statistically significant in all countries (Table 3) but intensity (the ectoparasite loads in tick-infested indi-
Namibia and Nigeria (Fisher exact test, P = 0.05). Also, viduals). First, the among-country variation was high 
the between-country variation in tick communities was (χ2 = 82.18; df = 5; P < 0.001), with the proportion of 
high, as all countries significantly differed from each the subpopulation having intermediate to high loads 
other (Fisher exact test pair-wise country comparisons, ranging from 4.7 (Namibia) to 49.0% (Nigeria). In 
P-values < 0.001). Nigeria, the cumulative L ogitrural-urban = 0.77 ± 0.38 
The most prevalent genus in almost all areas was (Z = 3.97; P = 0.046) and  in Ghana the cumulative 
Rhipicephalus (73.5% of all identifications; Fig.  3), but  Logitrural-urban = 1.01 ± 0.45 (Z = 5.11; P = 0.02). Con-
Haemaphysalis was the genus most often found in trasts in infestation intensities (rural vs urban) for all 
rural Kenya (67.1%) and Uganda (76.3%). Nine differ- other countries did not significantly differ from zero.
ent tick taxa were successfully identified at the species 
H eylen et al. Parasites Vectors          (2021) 14:576 Page 9 of 20
Fig. 3 Graphical overview of the tick communities found in urban and rural areas of the six African countries participating in the AFSCAN project 
(see Additional file 1: Table S1 for raw data). Numbers represent the PCR signals allocated to a tick taxon in the infested dogs. Per dog, DNA was 
extracted from a pooled set of ticks prior to carrying out the PCR analysis. 
Ecological correlations significantly higher loads   (Logitfree-yard = −  0.86 ± 0.27; 
Analyses were carried out for those country–taxa combi- Z = 9.93; P = 0.0016).  The lowest R. sanguineus prevalence 
nations for which at minimum  two countries had a 10% and load were observed in dogs that were kept indoors. 
prevalence in a given habitat type (see Table 3). As each The housing conditions did not show any associations in 
taxon contains a different set of countries, comparisons the other tick taxa, except for H. elliptica, with the free-
of outcomes between taxa should not be generalized. roaming dogs being far more often and heavily infested 
As mentioned in the Methods section, due to the many than the dogs kept indoors  (Logitindoor-free = − 1.91 ± 0.69; 
tests performed on the same datasets, only factors that Z = 7.63; P = 0.0058). The influence of intrinsic charac-
significantly explain the variation at α-level = 0.01 are teristics (age, body condition and sex) varied among tick 
discussed in the following sections; however, covariates taxa, but were found to be less important than the effect 
with P < 0.05 remained in the model for their potential of  deworming (see below). In Rhipicephalus spp. (i.e. in 
confounding effects, as well as a source of inspiration for the cluster of unidentified ticks belonging to the genus 
future research. Rhipicephalus), female dogs were more often infested 
In addition to the previously mentioned differences ( Logitfemale-male = 0.76 ± 0.27; Z = 7.81; P = 0.005) and body 
between countries and level of urbanization, the follow- condition showed a positive association with tick preva-
ing covariates explained part of the variation in infesta- lence and loads (Logit = 0.60 ± 0.21; Z = 8.29; P = 0.004). 
tions. For R. sanguineus loads and prevalence, housing In contrast, tick loads in R. sanguineus were negatively 
conditions mattered (χ2 ≥ 48.00;  P < 0.0001) in that the correlated with body condition (Logit = −  0.40 ± 0.14; 
highest loads were observed in yard dogs. Yard dogs Z = 8.27; P = 0.004). With regard to deworming drugs, 
were more often infested  (Logitfree-yard = −  0.71 ± 0.36; for Rhipicephalus sp. and H. elliptica, dogs that never 
Z = 9.91; P = 0.048)  than free-roaming ones, and carried received deworming drugs were less likely to be infested 
Heylen et al. Parasites Vectors          (2021) 14:576 Page 10 of 20
(see Table 4 for contrasts). In contrast, for Haemophysa- deworming drugs tended to have more fleas than dogs 
lis spp. (i.e. in the cluster of unidentified ticks belonging that were treated more recently (< 6 months; see Table 4 
to the genus Haemophysalis), the group that received a for contrasts). No additional effects of extrinsic risk fac-
drug > 6 months previously had significantly more ticks tors were found.
than the recently treated ones (within 6  months), but 
also more than the untreated dogs. For R. sanguineus, Vector‑borne pathogens
no association with deworming drugs were found. With DNA of pathogenic agents was detected in the collected 
regard to other extrinsic exposure risk factors, in addition blood samples (7 pathogen taxa), ticks (9 taxa) and fleas 
to habitat and macro-geographic sources of variation, in (3 taxa); serum antibodies against four pathogen genera 
environments with more dogs around we found slightly were also detected. For all biological samples, countries 
fewer H. elliptica-infested dogs (Logit = −  0.21 ± 0.08; strongly differed in terms of pathogen distributions and 
Z = 6.95; P = 0.008), but more Haemophysalis sp.-infested prevalence. In contrast to the findings for ectoparasites, 
ones (Logit = 0.16 ± 0.05; Z = 10.51; P = 0.0012). habitat differences in the vector-borne pathogens were 
less consistent and obvious.
Fleas
Prevalence Host blood
Substantial variation was observed for fleas among Hepatozoon canis was the most prevalent pathogen found 
countries (χ2 > 56.12; df = 5; P < 0.003), but  it should be in the blood samples of 601 dogs (overall 58.6%), but its 
emphasized that—due to the dominance of ticks—flea prevalence varied greatly  among country–habitat com-2
prevalence outcomes are to be interpreted with care and binations (range: 8.82–98%; χ  = 159.55; df = 5; P < 0.001). 
only for the subpopulation of ectoparasite-infested dogs. This pathogen was found to be consistently more preva-2
In particular, dogs from rural areas had significantly lent in rural areas  than in urban areas (χ  = 33.83; df = 1; 
more fleas than dogs from urban areas (χ2 = 29.32; df = 1; P < 0.001). This was also the case for Ehrlichia canis and 
P < 0.001), and this pattern was consistent across almost A. platys  in  Nigeria. Figure  4 and Table  5 provide an 
all countries (range L ogit  = 1.01 [Tanzania], overview of prevalence data and contrasts between coun-rural-urban
2.13 [Uganda]; Z = 6.09–18.42; P = 0.013 to < 0.0001). tries and/or habitats, respectively.
Namibia was an exception to the rule, with more fleas No further analyses were performed on pathogens that 
observed in urban dogs ( Logitrural-urban = −  2.05 ± 0.68; 
were not detected (B. canis, E. chaffeensis, A. phagocyt-
Z = 9.13; P = 0.0025). Further macro-geographical con- ophilum, R. conorii) or occurred in very low numbers (B. 
trasts at the  taxon level are shown in Table 3 and Fig. 2. rossi [3.8%], C. burnetti [0.5%], D. immitis [2.8%], Trypa-
By far, Ctenocephalides felis was the most prevalent flea nosoma spp. [0.2%]). Of all dogs that were infected with 
species (overall prevalence: 53.7%). Echidnophaga gal- at least one pathogen (N = 434), 30.9% were co-infected 
linacea-infested individuals were only found in Ghana with at least two pathogenic agents, with H. canis × E. 
(3.3%). Unidentified Echidnophaga spp. were collected canis (10.1%; Additional file 1: Table S4) being the most 
from a large proportion of the dogs (20%) tested in urban prevalent combination. Especially in Ghana and rural 
Namibia. In those dogs with flea co-infestations (3.5% of Nigeria, co-infections were commonly seen (> 40%). Sev-
infested dogs), C. felis × Echidnophaga sp. was the most eral individuals (5.1%) had  three to four pathogens in 
common combination observed (3.1% overall), with the their blood. In terms of pathogen community  distribu-
highest occurrence in the rural Kenya setting (9.4%) tions, most country communities greatly differed from 
(Additional file 1: Table S3). each other (Fisher exact tests, all P-values < 0.014),  with 
the exception of the pairwise comparisons Namibia 
versus Nigeria (P = 0.21) and Ghana versus Tanzania 
Infestation intensity (P = 0.093). Surprisingly, within each country no signifi-
The among-country variation was high (χ2 = 23.31; cant habitat differences (urban vs rural) in pathogen dis-
df = 5; P < 0.001), with the proportion of dogs with inter- tribution were found (all P-values > 0.05).
mediate to high loads ranging from 0% (Nigeria and Seroconversion was detected most often in response to 
Namibia) to 40.0% (urban Kenya). In Kenya, intensities Anaplasma spp. (13.1%) and Ehrlichia spp. (26.3%). For 
were significantly higher in urbanized areas (cumulative most of the six countries studied, seroprevalences were 
 Logitrural-urban = 1.72 ± 0.71; Z = 6.00; P = 0.014). higher than their respective pathogen genera traced in 
the host blood (Fig. 4; Table 5).
Ecological correlations
None of the intrinsic risk factors were associated with 
C. felis flea loads. Dogs that were never treated with 
H eylen et al. Parasites Vectors          (2021) 14:576 Page 11 of 20
Table 4 Ecological models for ectoparasite prevalence and loads in infested dogs
Covariate R. sanguineus : all countries Rhipicephalus sp.: Ta., Ug., Na. H. elliptica: Ke., Ug. Haemaphysalis. sp.: Ke., Ug., Gh. C. felis:  all countries
Prevalencea Loadb Prevalence Load Prevalence Load Prevalence Load Prevalence Load
Housing F/Y: − 0.71 ± 0.36 − 0.86 ± 0.27** F/Y: 0.43 ± 0.29  0.13 ± 0.35 
c onditionsc I/Y: − 0.96 ± 0.41* − 1.80 ± 0.30** I/F:-1.89 ± 0.65** − 1.91 ± 0.69**
Age (months)
Sex
 Female vs male 0.72 ± 0.27** 0.76 ± 0.27**
Body condition − 0.37 ± 0.18* − 0.40 ± 0.14** 0.57 ± 0.21** 0.60 ± 0.21** 0.35 ± 0.17*
Dewormingd
  < 1 month 2.08 ± 0.42*** 2.15 ± 0.44** 0.56 ± 0.52 1.95 ± 0.66** − 0.29 ± 0.33 − 0.30 ± 0.33 − 0.87 ± 0.35* − 1.24 ± 0.39**
 1–6 months 2.11 ± 0.45*** 2.17 ± 0.42** 3.57 ± 0.84*** 4.27 ± 0.76** − 0.62 ± 0.33 − 0.62 ± 0.32 − 1.19 ± 0.39** − 1.28 ± 0.42**
  > 6 months 1.46 ± 0.53** 1.42 ± 0.53** − 0.85 ± 0.77 − 0.18 ± 0.83 1.03 ± 0.31** 1.22 ± 0.28** − 0.48 ± 0.40 − 0.01 ± 0.39 
Number of dogs − 0.13 ± 0.05* − 0.13 ± 0.05* − 0.21 ± 0.08** − 0.42 ± 0.12* 0.16 ± 0.05** 0.14 ± 0.04*
around
Parameter estimates (± empirical standard error) from  eneralized estimation equations (GEEs) that model the tick and flea species’ prevalence (levels: 0, 1) and infestation loads (levels: absent, low, intermediate, high) 
for the population of ectoparasite-infested dogs. For a given taxon, only multiple countries with a prevalence of at least 10% for at least one of the habitat types were included (see Table 3). Country and habitat (rural vs 
urban) contrasts have been omitted from the table, but were included in all analyses. In none of the analyses did ‘ectoparasiticide treatment’ significantly explain tick variation (P > 0.05)
Ta(nzania), Ug(anda), Na(mibia), Ke(nya), Ni(geria), Gh(ana)
***P < 0.001, **P < 0.01, *P < 0.05, ’ ’  P > 0.05
a Prevalence indicates: model estimates reflect the probability that ectoparasite has level ‘1’; bLoad indicates model estimates that reflect the probabilities of tick load levels having higher ordered values, i.e. positive signs 
indicate higher loads with continuous explanatory variables, or higher than the reference category (when contrast is tested among the levels of categorical explanatory variables)
c Housing conditions contrasts include: F/Y (Free-roaming vs Yard); I/Y (Indoor vs Yard); I/F (Indoor vs Free-roaming); dContrast with group of dogs that have never been treated with a deworming drug 
Heylen et al. Parasites Vectors          (2021) 14:576 Page 12 of 20
Fig. 4 Macro-geographic variation in pathogen (sero-) prevalence in the blood samples collected from dog. Percentages of dogs infected with 
vector-borne pathogens based on DNA screening (gray shading), and seroprevalence against two taxa (yellow and green shading). For each 
pathogen, the same letters above columns indicate that the the contrast between countries is not statistically different from zero
Ticks Fleas
Also in ticks (537 pools screened), H. canis (60.5%) was In the 261 flea pools, M. haemofelis (overall preva-
the most prevalent pathogen. The absence of C. burnetti lence: 14.2%) was found in Kenya and Uganda (range: 
in the blood contrasts with the results from analysis of 10.0–30.8%) and in urban Tanzania, Ghana and Namibia 
the tick pools (15.1%). The prevalence of A. platys (10.6%) (11.8–16%). The prevalence of D. caninum was > 20% in 
was similar to that in the blood, while E. canis (overall Uganda and rural Nigeria, it was regularly observed in 
5.4%) was less often found. In addition to the highly sig- urban Namibia (11.8%), but it was not common in other 
nificant variation among countries (χ2 > 25.03; df = 5; all regions (< 6%). Bartonella henselae-infected fleas were 
P-values < 0.001), in Nigeria, three pathogens were sig- not collected, except in the urban Ghana setting (8.3%). 
nificantly more prevalent in rural habitats than in urban Co-infection (9.4%) was less common in the infected flea 
habitats (Table 6). pools (N = 64) than in the tick pools, although it should 
An overview of distributions across countries and habi- be noted that the odds of detecting co-infected indi-
tat types is shown in Fig. 5 and Table 6. In the infected viduals were lower due to the low number of pathogenic 
pools of ticks (N = 369), the most prevalent co-infection agents that were screened for (Additional file 1: Table S6). 
was C. burnetti × H. canis (12.2% of pools with at least Of 15 pairwise comparisons on flea-borne pathogen 
one pathogen; Additional file 1: Table S5). In rural Nige- community  distributions, 11 did show differences (all 
ria and Kenya, > 50% of infected pools had > 1 pathogenic P-values > 0.098); significant habitat differences were not 
agent. Comparison of the pathogen distributions among found in any  country (see Table 6 for Shannon indexes).
countries based on PCR  analyses showed that Namibia 
did not differ from the four other countries (pairwise Ecological correlations
comparisons with Uganda, Tanzania, Nigeria, Ghana: all In all analyses we corrected for macro-geographic spatial 
P-values > 0.11), while all other pairwise country compar- variation (at the country level), but allowed for habitat-
isons did (P < 0.047). Habitat differences (urban vs rural) related contrasts in ectoparasite loads to drive potential 
within each country were found in Nigeria and Ghana, associations (i.e. urbanisation level  was not included 
with higher pathogen diversities in their rural areas (see in the models). Prevalence of a small number of patho-
Table 6 for Shannon indexes). gens was high enough to fit reliable models (tick-borne 
pathogens: A. platys, H. canis, E. canis and C. burnetti; 
flea-borne pathogens: M. haemofelis, D. caninum; sero-
prevalence: Anaplasma spp. and Ehrlichia spp.) and this 
H eylen et al. Parasites Vectors          (2021) 14:576 Page 13 of 20
Table 5 Pathogen prevalence in the blood of dogs from six African countries
Pathogen prevalence Overall Tanzania (%) Kenya (%) Uganda (%) Nigeria (%) Ghana (%) Namibia (%)
Rural Urban P Rural Urban P Rural Urban P Rural Urban P Rural Urban P Rural Urban P
Pathogens in blood
 B. canis 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 B. rossi 3.8 0.0 1.9 12.0 3.9 8.0 6.0 3.5 3.9 0.0 1.8 0.0 0.0
 E. chaffeensis 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 A. platys 14.1 13.6 21.2 ns 4.0 0.0 0.0 6.0 33.3 7.8 ** 27.5 29.8 ns 19.3 7.4 ns
 A. phagocytophilum 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 R. conorii 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 R. africae 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 C. burnetti 0.5 0.0 0.0 2.7 0.0 0.0 0.0 1.8 0.0 0.0 0.0 0.0 0.0
 H. canis 58.6 77.3 67.3 ns 85.3 53.9 ** 98.0 86.0 ns 56.1 25.5 ** 67.5 45.6 * 29.0 8.8 *
 E. canis 18.5 27.3 19.2 ns 13.3 19.2 ns 0.0 4.0 31.6 15.7 * 25.0 28.1 ns 22.5 19.1 ns
 D. immitisa 2.8 4.6 5.8 1.3 0.0 0.0 0.0 0.0 0.0 7.5 14.0 ns 0.0 0.0
 Trypanosoma spp.a 0.2 0.0 0.0 0.0 0.0 0.0 0.0 1.8 0.0 0.0 0.0 0.0 0.0
 Individuals (N) 601 44 52 75 26 50 50 57 51 40 57 31 68
 Shannon index 1.0 1.1 ns 1.0 0.8 ns 0.3 0.6 ns 1.3 1.2 ** 1.2 1.4 ns 1.1 1.0 ns
Sero‑positive
  Anaplasma spp. 13.1 20.5 21.2 ns 9.7 7.7 4.0 24.0 ** 4.0 20.5 *** 0.0 3.9 23.3 7.9 **
  Borrelia spp. 0.2 0.0 0.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
  Ehrlichia spp. 26.3 31.8 28.9 ns 22.2 15.4 ns 10.0 4.0 ns 54.0 32.0 * 35.0 21.2 ns 40.0 25.4 ns
   Heartwormb 0.3 0.0 0.0 0.0 3.9 0.0 0.0 0.0 0.0 0.0 1.9 0.0 0.0
  Individuals (N) 579 44 52 72 26 50 50 50 50 40 52 30 63
DNA-based (qPCR) pathogen prevalence in the blood of dogs from urban and rural areas of 6 African countries. Prevalence of 2 additional pathogens—not transmitted by fleas or ticks (D. immitis and Trypanosoma 
spp.)—are reported as well. In addition, sero-prevalence of four pathogen genera are given, for which the IDEXX test was used. For statistical outcomes on pairwise macro-geographic differences, see Fig. 4. As a measure 
of species diversity, a Shannon index has been provided
***P < 0.001, **P < 0.01, *P < 0.05, ns (not-significant) P > 0.05
a Non-tick-or flea-borne pathogens
b Positive in IDEXX test, but all negative in qPCR’s
Heylen et al. Parasites Vectors          (2021) 14:576 Page 14 of 20
Table 6 Pathogen prevalence in flea and tick pools collected from infested dogs
Tick and flea pools Overall Tanzania (%) Kenya (%) Uganda (%) Nigeria (%) Ghana (%) Namibia (%)
Rural Urban P Rural Urban P Rural Urban P Rural Urban P Rural Urban P Rural Urban P
Ticks
 B. canis 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 B. rossi 2.4 0.0 0.0 6.9 0.0 0.0 6.5 1.8 3.9 5.1 0.0 0.0 0.0
 B. felis 0.6 4.4 0.0 0.0 0.0 0.0 0.0 1.8 0.0 0.0 0.0 0.0 0.0
 E. chaffeensis 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 A. platys 10.6 20.0 26.0 ns 0.0 0.0 0.0 6.5 29.8 3.9 ** 15.4 13.3 ns 0.0 2.6
 A. phagocytophilum 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
 R. conorii 3.7 2.2 0.0 18.1 0.0 ** 5.3 6.5 0.0 0.0 0.0 0.0 0.0 2.6
 R. africae 2.1 4.4 0.0 1.4 0.0 7.9 0.0 1.8 0.0 7.7 2.2 0.0 0.0
 C. burnetti 15.1 11.1 2.0 ns 63.9 0.0 *** 15.8 10.9 ns 29.8 2.0 ** 0.0 0.0 0.0 0.0
 H. canis 60.5 75.6 76.0 ns 80.6 62.5 ns 50.0 41.3 ns 91.2 58.8 *** 56.4 46.7 ns 29.0 20.5 ns
 D. immitis 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.2 0.0 0.0
 E. canis 5.4 2.2 8.0 2.8 0.0 0.0 0.0 12.3 13.7 ns 10.3 6.7 ns 3.2 0.0
 Tick pools (N) 537 45 50 72 24 38 46 57 51 39 45 31 39
 Shannon index 1.2 0.9 ns 1.2 0.0 ** 1.0 1.3 ns 1.3 0.9 ** 1.2 1.0 ns 0.3 0.6 ns
Fleas
 B. henselae 0.8 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 8.3 0.0 0.0
 M. haemofelis 14.2 3.7 16.0 ns 22.6 30.8 ns 12.5 10.0 ns 13.6 0.0 5.9 12.5 ns 0.0 11.8
 D. caninum 11.9 3.7 4.0 3.8 0.0 27.5 25.0 ns 31.8 0.0 5.9 4.2 0.0 11.8
 Flea pools (N) 261 27 25 53 13 40 20 22 1 17 24 2 17
 Shannon index 0.7 0.5 ns 0.4 0.0 ns 0.6 0.6 ns 0.6 0.0 0.7 1.0 ns 0.0 0.7
Pathogen prevalence in 261 flea pools and 537 tick pools collected from infested dog individuals in urban and rural areas of six African countries. No pair-wise comparisons have been performed for countries in which 
less than three dog individuals have been sampled in one of its habitats. For statistical outcomes on macrogeographic contrasts, see Fig. 5
***P < 0.001, **P < 0.01, ns (not-significant) P > 0.05
H eylen et al. Parasites Vectors          (2021) 14:576 Page 15 of 20
Fig. 5 Macro-geographic variation in pathogen prevalence in ectoparasites isolated from dogs. Percentages of pools of ticks collected from dogs 
that were infected with one of the common tick-borne (gray shading) and flea-borne (red and blue) pathogens (overall prevalence > 5%; see 
Table  6). For each pathogen, the same letters above columns indicate that the the contrast between countries is not statistically different from zero
for a selected number of countries (i.e. countries where (0.011 ± 0.004; Z = 9.42; P = 0.0022; Additional file  1: 
prevalences < 10% were excluded). Therefore, compari- Table  S7). For all abovementioned pathogens found in 
sons of ecological correlations between pathogens should the blood, deworming drug-treated individuals had lower 
not be over-interpreted, as different sets of countries are prevalence than the individuals that had never been 
used for each pathogen. Given the high number of tests treated.
performed on the same dataset and the higher likelihood 
for Type I-errors, only results with P < 0.01 will be dis- Flea‑borne pathogens
cussed (see "Statistical analysis"). Pathogen presence was not explained by any of the dog’s 
intrinsic and extrinsic exposure risk factors (at a signifi-
Tick‑borne pathogens in blood and ticks cance level α = 0.05; see Additional file  1: Table  S8 for 
The presence of pathogens in the ticks was explained by tendencies).
its presence in the dog’s blood (range Logit estimates: 
1.41 ± 0.24–4.35 ± 0.33; all P < 0.01; Table  7). Additional Pathogen–tick associations
variation was explained by the tick loads in H. canis (R. An explicit analysis of those pools of ticks in which the 
sanguineus: 0.33 ± 0.11; Z = 8.74; P = 0.0031) and C. DNA of only a single species was detected (Table  8) 
burnetti (H. leachi: 0.44 ± 0.15; Z = 8.66; P = 0.0032 and again showed a higher occurrence of C. burnetti in the 
Haemaphysalis sp.: 0.47 ± 0.15; Z = 9.97; P = 0.0016). genus Haemaphysalis (H. elliptica: 55%; Haemaphysa-
Dogs in better body condition had less H. canis infected lis sp.: 40%) compared to the genus Rhipicephalus (R. 
ticks (−  0.47 ± 0.17; Z = 7.79; P = 0.0053), but sur- sanguineus: 7.5%; Rhipicephalus  sp. 10%). Also, in R. 
prisingly this association was not found for the blood conorii, Haemaphysalis ticks (10.0–17.3%) were more 
infections. Blood prevalence of A. platys (0.42 ± 0.12; often infected than Rhipicephalus ticks (12.7–20.0%). 
Z = 13.89; P < 0.001) and E. canis (0.25 ± 0.10; Z = 6.26; The opposite was true for A. platys, which was more 
P = 0.012), as well as seroprevalences for Anaplasma spp. often found in the genus Rhipicephalus (R. sanguineus: 
(0.77 ± 0.09; Z = 81.10; P < 0.001), were positively associ- 12.7%; Rhipicephalus sp.: 20.0%) than in the genus 
ated with R. sanguineus tick loads. Furthermore, sero- Haemaphysalis (H. elliptica: 0.0%; Haemaphysalis sp.: 
prevalence of Anaplasma spp. was positively associated 1.3%). Hepatozoon canis, on the other hand, was found 
with Rhipicephalus sp. (0.91 ± 0.16; Z = 33.07; P < 0.001) in reasonable numbers (> 45%) in the most prevalent tick 
and H. leachi (0.42 ± 0.13; Z = 11.56; P < 0.001). Ehrlichia taxa (R. sanguineus, Rhipicephalus sp., H. elliptica and 
spp. seroprevalences tended to be higher in older dogs 
Heylen et al. Parasites Vectors          (2021) 14:576 Page 16 of 20
Table 7 Ecological models for tick-borne pathogens in host blood and ticks from infested dogs
Covariate A. Platys: Ta, Na, Ni, Gh H. canis: all countries E. canis: Ta, Na, Ke, Ni, Gh C. burnetti (tick 
only): Ke, Ug, Ni
Blood Tick Blood Tick Blood Tick
Age (months) − 0.38 ± 0.16*
Body condition − 0.47 ± 0.17**
Tick loads
 R. sanguineus 0.42 ± 0.12*** 0.33 ± 0.11** 0.25 ± 0.10**
 Rhipicephalus sp.
 H. leachi − 0.48 ± 0.25* 0.44 ± 0.15**
 Haemaphysalis spp. 0.42 ± 0.18* 0.47 ± 0.15**
Deworminga
  < 1 month − 1.55 ± 0.44*** − 1.26 ± 0.31*** − 0.74 ± 0.36***
 1–6 months − 1.58 ± 0.38*** − 1.25 ± 0.32*** − 0.36 ± 0.29***
  > 6 months − 1.10 ± 0.43** − 1.00 ± 0.38** − 0.68 ± 0.40**
Pathogen in blood  tissueb
 Yes–No 4.00 ± 0.30*** 1.41 ± 0.24*** 4.35 ± 0.33*** 3.02 ± 1.12**
Parameter estimates (± empirical standard error) from the logistic regressions (GEEs) that model the pathogen prevalence (levels: 0, 1) in host blood and the ticks. 
Only countries for which at least one area had a prevalence > 10% were included. The main assumption here is that pathogen in the blood is driven by vector presence 
(proxy: ticks found on dogs) and the dog’s physiology; therefore, extrinsic characteristics that correct for tick presence (urban vs rural; housing conditions; dogs 
around) are not included. We assume macro-geographic variation in pathogen (wildlife) reservoirs at the  country level; therefore ‘country’ remained in all of the 
models.
Sex of dog, dogs in the environment and ectoparasiticide treatment did not significantly explain any of the variation, and therefore are not shown in the table
***P < 0.001, **P < 0.01, *P < 0.05 
a Contrasts with group of dogs that have never been treated with a deworming drug
b Only included in the analyses on pathogens in feeding ticks
Table 8 Pathogen-tick associations 
R. sanguineus (1) R. R. simus Rhipicephalus H. elliptica (3) H. spinulosa Haemaphysalis A. variegatum Ixodes spp.
appendiculatus spp. (2) spp. (4)
A. phagocytophi- 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
lum
B. canis 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
B. rossi 0.6 0.0 0.0 0.0 5.0 0.0 9.3 0.0 50.0
D. immitis 0.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
E. chaffeensis 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
E. canis 7.8 0.0 0.0 0.0 0.0 0.0 0.0 0.0 50.0
H. canis 63.0 a,2,3,4 100.0 80.0 46.7 c, 1 75.0 b, 1 20.0 65.3 a, 1 100.0 100.0
A. platys 12.7 a 0.0 0.0 20.0 a 0.0 b 0.0 1.3 b 0.0 0.0
R. africae 0.6 0.0 0.0 0.0 0.0 0.0 1.3 100.0 0.0
R. conorii 0.3 a, 3,4 0.0 0.0 3.3 a 10.0 b,1 0.0 17.3 b, 1 0.0 0.0
C. burnetti 7.5 a, 3,4 50.0 0.0 10.0 a 55.0 b, 1 0.0 40.0 b, 1 0.0 0.0
 Tick samples 332 2 5 30 20 5 75 1 2
Only the dogs in which a single tick taxon was observed (based on the extractions of the set of pooled ticks) were included in the analysis. Statistical analyses on 
the occurrence of pathogens (H. canis, A. platys, R. conorii and C. burnetti) were done on tick taxa with ≥ 20 individuals (R. sanguineus, Rhipicephalus spp., H. elliptica, 
Haemaphysalis spp.). Africa-wide comparison: within a row, same letters behind prevalences indicate no significant difference. Country-corrected comparisons: 
numbers (see column headings for tick reference numbers) refer to the tick species from which the prevalence differs (P < 0.05). For this latter analysis, in the following 
groups of countries a sufficient number of dogs was sampled to allow for pairwise statistical comparisons. Tanzania, Namibia, Uganda: Rhipicephalus spp. vs R. 
sanguineus; Ghana, Kenya, Uganda: Haemaphysalis spp. vs R. sanguineus; Kenya, Uganda: H. elliptica vs (R. sanguineus and Haemaphysalis spp.); Uganda: R. sanguineus 
vs (Rhipicephalus spp., Haemaphysalis spp., H. elliptica)
Africa-wide comparison: same lowercase letters indicate no significant difference. Country-corrected comparisons: for pathogens in each tick species’ column,  
pathogens followed by a number is linked with a significant difference (P < 0.05) with one of the other tick species: (R. sanguineus (1), Rhipicephalus spp. (2), H. elliptica 
(3), Haemaphysalis sp. (4)
For the following groups of countries a sufficient number of dogs were sampled to allow for pairwise statistical comparisons between tick taxa: Tanzania, 
Namibia, Uganda: Rhipicephalus sp. vs R. sanguineus; Ghana, Kenya, Uganda: Haemaphysalis spp. vs R. sanguineus; Kenya, Uganda: H. elliptica vs (R. sanguineus and 
Haemaphysalis sp.); Uganda: R. sanguineus vs (Rhipicephalus spp., Haemaphysalis sp., H. elliptica)
H eylen et al. Parasites Vectors          (2021) 14:576 Page 17 of 20
Haemaphysalis sp.). Only tick taxa with ≥ 20 individuals life-cycle and preferences far more complicated than 
were considered in these enumerations. those of R. sanguineus [15]. In our extensive survey, few 
dogs were found to be infested with  R. microplus.
Discussion In comparison, most Haemaphysalis spp. ticks are from 
The objective of the study was to determine the most wildlife and considered to be exophilic [16]. As a conse-
important vector-borne pathogens and ectoparasites of quence, several identified (H. leachi, H. elliptica, H. spinu-
dogs in six sub-Saharan African countries. We handled losa) and undefined Haemaphysalis ticks were observed 
the data collection using a rigorous pre-defined proto- more often on dogs from rural areas than on those from 
col and did a meta-analysis on standardized data. The urban areas. In H. leachi and H. elliptica (previously con-
occurrence of ticks, fleas and pathogens in both vector sidered as H. leachi as well [17]), adults parasitize domes-
and host were investigated, focusing on strong contrasts tic and wild carnivores, while the immature stages feed 
between broad country-specific urban categories (see on rodents. For those countries where there were suffi-
"Methods" section). Several surveys on ticks and their cient data to allow statistical analysis on the latter species 
pathogens have been conducted in Africa, but these (Kenya and Uganda), free-roaming dogs—often going 
mainly consider production animals in agricultural areas into wildlife habitats—indeed did show higher preva-
[13]. This study shows that, despite the significant socio- lences. Haemaphysalis spinulosa adults appear to feed 
economic value of companion animals to humans, they on various small- and medium-sized carnivores, as well 
represent an indirect risk for zoonotic vector-borne dis- as hedgehogs [16]. A very high proportion of individuals 
eases in a One Health perspective by hosting pathogenic belonging to the genus Haemaphysalis could not be iden-
agents and their vectors. As dogs are in close contact to tified, but genetic clustering revealed a large group taxon 
wildlife and to production animals (e.g. cattle), they are at that most likely involves a new or genetically uncharac-
the interface of several vertebrate communities, although terized species (manuscript in preparation).
with different functionalities to humans. Twelve vector-borne pathogens were detected in host 
More than 70% of the dogs sampled were infected tissue and vectors. For the most common tick-borne 
with at least one vector-borne pathogen in the blood, pathogens, we found a strong correlation between their 
and infection rates were even higher in the vectors col- presence in host blood and their presence in the ticks, 
lected from these infected dogs. Even in the absence of without any indication of a causative correlation or vec-
any thorough knowledge on local host abundances and tor competence. When the distributions of the patho-
diversity, several of the associations and geographi- gens with respect to tick species are considered, the 
cal patterns turn out to be the consequence of the tick most striking contrast is the high prevalence of C. bur-
vectors’ biology. Rhipicephalus sanguineus, commonly netti in the genus Haemaphysalis in which all members 
known as the brown dog tick [14], is found worldwide in of the genus are known to be vector-competent for this 
warmer climates and is a monotropic (dogs) three-host pathogen [18]. Surprisingly, only few dog blood samples 
tick. It is an endophilic tick that completes its entire life- tested positive for this pathogen despite individual dogs 
cycle indoors, such as inside kennels. It typically shows carrying infected ticks. Ticks could have been infected 
negative geotropism after feeding, a behavior in which in the nymphal stage, when feeding on reservoir com-
the ticks climb to the walls of borrows or man-made shel- petent hosts. The pathogen was also not observed in 
ters and hide in cracks and crevices to oviposit or molt to ticks from Ghana and Namibia. The occurrence of H. 
the next stage. Therefore, it is not surprising that in most canis was higher when dogs were more heavily infested 
countries (with the exception of Namibia, where sam- (with both Rhipicephalus spp. and Haemaphysalis spp.), 
pling happened during the dry season) the tick was more but the pathogen did not show a strong preference for a 
often found in urban areas than rural areas, and found particular tick species, indicating that both genera could 
more often on dogs that were restricted to their yards equally contribute to H. canis’ transmission—via inges-
(i.e. closer to outdoor man-made structures, like ken- tion of infected ticks. Ehrlichia canis and A. platys preva-
nels). However, not all Rhipicephalus ticks are endophilic, lences in the blood were positively correlated with R. 
which is likely the reason why the group of unidentified sanguineus, as were the Anaplasma sp. and—to a lower 
ticks belonging to this genus has a less outspoken pref- extent— Ehrlichia sp. seroprevalences, indicating the 
erence for rural or urban areas (Table  3). Adults of R. central role of R. sanguineus ticks in pathogen exposure 
appendiculatus, R. microplus, R. simus and R. senega- in domestic dogs across almost the complete African 
lensis are all linked to cattle and wildlife such as buffalo continent. Rickettsia conorii was absent in host blood and 
and large antelope. Furthermore, with the exception of R. mainly found in the East African countries. We empha-
microplus, all of these ticks are considered to be three- size that prevalences (both in ectoparasites and vector-
host ticks with di- or telotropic behavior, making their borne pathogens) are likely affected by parasite-induced 
Heylen et al. Parasites Vectors          (2021) 14:576 Page 18 of 20
host mortality, which especially in highly virulent vector- host and pathogen communities. Explanatory analyses of 
borne pathogens (e.g. B. rossi often found in the genus correlative data at this large scale are very often affected 
Haemaphysalis [19, 20]) could strongly influence our by biases and confounders (e.g. effects of wildlife popula-
interpretation of the presented cross-sectional data. Bor- tion density) that cannot be controlled for due to a lack 
relia seroprevalence was close to zero, likely because of of information. Furthermore, as the main objective was 
the low prevalence of Ixodes species, the main vectors of to determine the most important ectoparasites and vec-
Borrelia burgdorferi sensu lato [21]. tor-borne pathogens, the focus of this study was on the 
By far the most abundant flea species found on dogs subpopulation of infested individuals, which is why out-
was C. felis, carrying M. haemofelis and D. caninum. comes have to be interpreted in terms of occurrence and 
Interestingly, associations with deworming drugs were diversity, rather than true prevalence. With regard to the 
found for fleas and ticks. Dogs that were recently treated vector-borne pathogens, we consider bias should be less 
carried the most ticks, whereas the longer in the past strong, as pathogen occurrence in the body is the out-
the treatment (or no treatment at all), the lower the tick come of past infestations (which not necessarily corre-
counts. For fleas and Haemaphysalis sp. we observed pat- lates with the infestation levels at capture). But here also, 
terns in the opposite direction: untreated dogs had over- since tissue tropism may heavily differ among pathogens 
all more ectoparasites than treated individuals. Although and show (unknown) temporal patterns herein, blood 
these trends are contradictory, we could hypothesize screening does not allow every pathogen to be identi-
that dog owners would more likely treat their dog when fied with equal probability. Nevertheless, to the best of 
ectoparasites are observed. Knowing fleas and ticks our knowledge, this work forms the most extensive and 
belonging to the genus Haemaphysalis are more difficult standardized study in sub-Saharan countries so far, giv-
to observe visually because of their small size, the likeli- ing an overview of important vectors and vector-borne 
hood of a dog being treated might directly be related to pathogens; as such, the information could serve as base-
the size of the ectoparasites. line data for future research and interventions.
Additional screening for mosquito-, tsetse- and We also found tick species and pathogens that are not 
sand fly-borne pathogens resulted in a low number of classically associated with companion animals but which 
observed cases. Dirofilaria immitis seroprevalence was still have the potential to transmit zoonotic disease-caus-
very low, despite its much higher prevalence in dog ing pathogens in dogs. High levels of co-infestation and 
blood. The discrepancy between antigen level and PCR co-infections were observed, adding to the zoonotic risk, 
results may be explained in a number of ways. First, the given the high potential of bridging opportunities to cat-
molecular assay employed in this study was able to detect tle and humans via vectors and/or immuno-modulations 
samples with very low levels of microfilaremia that could and atypical virulence patterns (due to co-parasitism). 
not be detected by the less sensitive serological test. Posi- Furthermore, we found multi-host ticks in urban areas, 
tive PCR signals, indicative of D. immitis DNA, could which have the potential to extend the network of patho-
have amplified DNA from other filarioids commonly gen transmission to humans.
found in dogs. A subset of the presumed D. immitis-pos-
itive DNA samples was post-hoc subjected to sequenc- Conclusions
ing, which revealed Acanthocheilonema reconditum and This standardized surveillance underscores the impor-
Dirofilaria repens (using BLAST and NCBI database). tance of ectoparasites and their pathogens in dogs of 
Secondly, blocked antigen may be a consequence of the sub-Saharan Africa, with co-parasitism being the rule 
presence of immune complexes, which lowers sensitivity rather than the exception. Future research needs to 
of the antigen detection. Protocols such as heat treatment include wildlife host surveys, tick densities in the off-
might disrupt these complexes and allow detection of the host environment, detailed habitat characteristics and 
antigen. Also, if a dog has an infection that was estab- specific resources that may support dense popula-
lished < 4 months previously, immature D. immitis may tions of wildlife hosts. Furthermore, species-specific 
be present (identified by PCR) but the antigen test will be responses to space characteristics [23] in least-cost 
negative regardless of how it is performed. Finally, it has path analyses that make use of habitat connectivity will 
been shown that in dogs infected with only one or two substantially increase our understanding of how spatial 
adult females, or only male worms, antigen is unlikely elements could affect local vector-borne pathogen risk. 
going to be detected [22]. Integration of this knowledge with a good understand-
A few additional considerations should be noted ing of current complexities in socio-economic and cli-
regarding the outcomes of this study. Many of the corre- mate changes will enable policymakers and scientists to 
lations may be the results of macro-geographic and habi- develop and provide prevention strategies.
tat differences, which come with differences in vector, 
H eylen et al. Parasites Vectors          (2021) 14:576 Page 19 of 20
Abbreviations (Pty) Ltd., Bloemfontein, South Africa. 8 Clinomics, Bloemfontein, South Africa. 
BLAST: Basic local alignment search tool; DNA: Deoxyribonucleic acid; GEE: 9 School of Veterinary Medicine, College of Basic and Applied Sciences (CBAS), 
Generalized estimation equation (model); mtDNA: Mitochondrial DNA; qPCR: University of Ghana, Accra, Ghana. 10 Department of Veterinary Pathology, 
Quantitative real-time polymerase chain reaction. Microbiology and Parasitology, University of Nairobi, Nairobi, Kenya. 11 Depart-
ment of Veterinary Parasitology and Entomology, College of Veterinary 
Medicine, Federal University of Agriculture, Abeokuta, Nigeria. 12Supplementary Information  Sokoine University of Agriculture, Morogoro, Tanzania. 13 Research Center for Tropical 
The online version contains supplementary material available at https://d oi. Diseases and Vector Control, College of Veterinary Medicine, Animal Resources 
org/ 10.1 186/ s13071- 021- 05014-8. and Biosecurity, Makerere University, Kampala, Uganda. 14 School of Veterinary 
Medicine, University of Namibia, Neudamm, Namibia, South Africa. 15 São 
Paulo State University, São Paulo, Brazil. 16 University of Pretoria, Pretoria, South 
Additional file 1: Fig. S1. Overview of sampling times and average Africa. 17 Clinglobal, Tamarin, Mauritius. 
seasonal variation in precipitation and temperature. Table S1. Distribu-
tion of PCR signals allocated to an ectoparasite taxon (identification at Received: 19 June 2021   Accepted: 12 September 2021
genus level and more precise) in the infested dogs of urban and rural 
areas. Table S2. Distribution of co-infested dogs within the subpopulation 
of tick-infested dogs. Table S3. Co-infestations by different flea species 
(identification at genus level and lower). Table S4. Co-infections in dog 
blood. Table S5. Co-infections in dog ticks. Table S6. Co-infections in dog References
fleas. Table S7. Correlations with sero-prevalences. Table S8. Correlations  1. Zipperer WC, Pickett STA. Urban ecology: patterns of population growth 
with flea-borne pathogens. and ecological effects. Chichester: Wiley; 2012.
Additional file 2. Capture form.  2. United Nations Department of Economic and Social Affairs. 68% of the 
world population projected to live in urban areas by 2050, says UN. UN; 
2018. https:// www. un. org/ devel opment/ desa/ en/ news/ popul ation/ 
Acknowledgements 2018- revis ion- of- world- urban izati on- prosp ects.h tml. Accessed 7 Sept 
We thank Sherry Johnson (Ghana), Samuel Maina Githigia (Kenya), Foluke Ade- 2021.
dayo Akande (Nigeria), Jahashi Saidi Nzalawahe (Tanzania), Tayebwa Dickson  3. Colella V, Nguyen VL, Tan D, Lu N, Fang F, Yin ZJ, et al. Zoonotic vector-
(Uganda) and Ortwin Aschenborn (Namibia) for their technical assistance. borne pathogens and ectoparasites of dogs and cats in Eastern and 
Southeast Asia. Emerg Infect Dis. 2020;26:1221–33.
Authors’ contributions  4. Perini K, Sabbion P. Urban sustainability and river restoration: green and 
DH analyzed the data and, together with MM, wrote the first draft. BS, MMar blue infrastructure. Hoboken: Wiley; 2016.
and JF gave substantial scientific input on the final manuscript version. SJ,  5. Hansen R, Pauleit S. From multifunctionality to multiple ecosystem 
SMG, FAA, JSN, TD and OA provided local technical assistance. ML performed services? A conceptual framework for multifunctionality in green infra-
and analyzed laboratory tests. MD initiated and supervised the study. All structure planning for urban areas. Ambio. 2014;43:516–29.
authors read and approved the final manuscript.  6. Braks MAH, Van Wieren SE, Takken W, Sprong H, editors. Ecology and pre-
vention of Lyme borreliosis. Wageningen: Wageningen Academic; 2016.
Funding  7. Day MJ. One Health: the importance of companion animal vector-borne 
DH is funded by the Marie Sklodowska-Curie Actions (EU-Horizon 2020, Indi- diseases. Parasit Vectors. 2011;4:49.
vidual Global Fellowship, project no. 799609). The study was funded by Bayer  8. Matjila PT, Leisewitz AL, Jongejan F, Bertschinger HJ, Penzhorn BL. 
Animal Health GmbH, an Elanco Animal Health company, within the frame- Molecular detection of Babesia rossi and Hepatozoon sp. in African wild 
work of the African Small Companion Animal Network (AFSCAN) program of dogs (Lycaon pictus) in South Africa. Vet Parasitol. 2008;157:123–7.
the World Small Animal Veterinary Association (WASAVA) and supported by  9. Matjila PT, Leisewitz AL, Jongejan F, Penzhorn BL. Molecular detection of 
Idexx Laboratories and Clinvet International (Pty) Ltd. The funders had no role tick-borne protozoal and ehrlichial infections in domestic dogs in South 
in the study design, data collection, interpretation and analysis, decision to Africa. Vet Parasitol. 2008;155:152–7.
publish, or preparation of the manuscript.  10. Molenberghs G, Verbeke G. Models for discrete longitudinal data. New 
York: Springer; 2005.
Availability of data and materials  11. Steyerberg EW. Clinical prediction models: a practical approach to devel-
The datasets used and/or analyzed during the current study are available from opment, validation, and updating. New York: Springer; 2009.
the corresponding author on reasonable request.  12. Montagna PA. Using SAS to manage biological species data and calculate 
diversity indices. SCSUG. 2014:1–5.
 13. Adakal H, Biguezoton A, Zoungrana S, Courtin F, De Clercq EM, Madder M. 
Declarations Alarming spread of the Asian cattle tick Rhipicephalus microplus in West 
Africa-another three countries are affected: Burkina Faso, Mali and Togo. 
Ethics approval and consent to participate Exp Appl Acarol. 2013;61:383–6.
The study was carried out according to the national animal welfare  14. Dantas-Torres F. Biology and ecology of the brown dog tick, Rhipicephalus 
regulations. sanguineus. Parasit Vectors. 2010;3:26.
 15. Walker JB, Keirans JE, Horak IG. The genus Rhipicephalus (Acari, Ixodidae): 
Consent for publication a guide to the brown ticks of the world. 1st ed. New York: Cambridge 
Not applicable. University Press; 2000.
 16. Sylla M, Ndiaye M, Souris M, Gonzalez JP. Ticks (Acari: Ixodida) of the 
Competing interests genus Haemaphysalis Koch, 1844 in Senegal: a review of host associa-
The authors declare that they have no competing interests. tions, chorology, and identification. Acarologia. 2018;58:928–45.
 17. Apanaskevich DA, Horak IG, Camicas JL. Redescription of Haemaphysalis 
Author details (Rhipistoma) elliptica (Koch, 1844), an old taxon of the Haemaphysa-
1 Eco-Epidemiology Group, Department of Biomedical Sciences, Institute lis (Rhipistoma) leachi group from East and southern Africa, and of 
of Tropical Medicine, Antwerp, Belgium. 2 Interuniversity Institute for Biosta- Haemaphysalis (Rhipistoma) leachi (Audouin, 1826) (Ixodida, Ixodidae). 
tistics and Statistical Bioinformatics, Hasselt University, Diepenbeek, Belgium. Onderstepoort J Vet Res. 2007;74:181–208.
3 Department of Ecology and Evolutionary Biology, Princeton University,  18. Knobel DL, Maina AN, Cutler SJ, Ogola E, Feikin DR, Junghae M, et al. 
Princeton, NJ, USA. 4 School of Veterinary and Life Sciences, Murdoch Uni- Coxiella burnetii in humans, domestic ruminants, and ticks in rural western 
versity, Murdoch, WA, Australia. 5 Bayer Animal Health, Elanco Animal Health Kenya. Am J Trop Med Hyg. 2013;88:513–8.
Inc., Leverkusen, Germany. 6 Clinvet LLC, Waverly, USA. 7 ClinVet International 
Heylen et al. Parasites Vectors          (2021) 14:576 Page 20 of 20
 19. Penzhorn BL. Why is Southern African canine babesiosis so virulent? An  30. de Bruin A, de Groot A, de Heer L, Bok J, Wielinga PR, Hamans M, et al. 
evolutionary perspective. Parasit Vectors. 2011;4:51. Detection of Coxiella burnetii in complex matrices by using multiplex 
 20. Penzhorn BL. Don’t let sleeping dogs lie: unravelling the identity and quantitative PCR during a major Q Fever outbreak in The Netherlands. 
taxonomy of Babesia canis, Babesia rossi and Babesia vogeli. Parasit Vectors. Appl Environ Microbiol. 2011;77:6516–23.
2020;13:184.  31. Rojas A, Rojas D, Montenegro VM, Baneth G. Detection of Dirofilaria 
 21. Eisen L. Vector competence study with hard ticks and Borrelia burgdorferi immitis and other arthropod-borne filarioids by an HRM real-time qPCR, 
sensu lato spirochetes: a review. Ticks Tick-Borne Dis. 2020;11:3. blood-concentrating techniques and a serological assay in dogs from 
 22. Little S, Saleh M, Wohltjen M, Nagamori Y. Prime detection of Dirofilaria Costa Rica. Parasit Vectors. 2015;8:170.
immitis: understanding the influence of blocked antigen on heartworm  32. Taylor TK, Boyle DB, Bingham J. Development of a TaqMan PCR assay for 
test performance. Parasit Vectors. 2018;11:1. the detection of Trypanosoma evansi, the agent of surra. Vet Parasitol. 
 23. Heylen D, Lasters R, Adriaensen F, Fonville M, Sprong H, Matthysen E. 2008;153:255–64.
Ticks and tick-borne diseases in the city: role of landscape connectivity  33. Silbermayr K, Li FY, Soudre A, Muller S, Solkner J. A novel qPCR assay for 
and green space characteristics in a metropolitan area. Sci Total Environ. the detection of African animal trypanosomosis in trypanotolerant and 
2019;670:941–9. trypanosusceptible cattle breeds. Plos Negl Trop Dis. 2013;7:8.
 24. Troskie M, de Villiers L, Leisewitz A, Oosthuizen MC, Quan M. Develop-  34. Bouhsira E, Franc M, Boulouis HJ, Jacquiet P, Raymond-Letron I, Lienard E. 
ment and validation of a multiplex, real-time PCR assay for Babesia rossi Assessment of persistence of Bartonella henselae in Ctenocephalides felis. 
and Babesia vogeli. Ticks Tick-Borne Dis. 2019;10:421–32. Appl Environ Microbiol. 2013;79:7439–44.
 25. Thomson K, Yaaran T, Belshaw A, Curson L, Tisi L, Maurice S, et al. A new  35. Tasker S, Peters IR, Mumford AD, Day MJ, Gruffydd-Jones TJ, Day S, et al. 
TaqMan method for the reliable diagnosis of Ehrlichia spp. in canine Investigation of human haemotropic Mycoplasma infections using a 
whole blood. Parasit Vectors. 2018;11(1):350. https:// doi. org/ 10. 1186/ novel generic haemoplasma qPCR assay on blood samples and blood 
s13071-0 18- 2914-5. smears. J Med Microbiol. 2010;59:1285–92.
 26. Loftis AD, Massung RF, Levin ML. Quantitative real-time PCR assay for  36. Labuschagne M, Beugnet F, Rehbein S, Guillot J, Fourie J, Crafford D. 
detection of Ehrlichia chaffeensis. J Clin Microbiol. 2003;41:3870–2. Analysis of Dipylidium caninum tapeworms from dogs and cats, or their 
 27. da Silva CB, Pires MS, Vilela JAR, Peckle M, da Costa RL, Vitari GLV, et al. A respective fleas Part 1. Molecular characterization of Dipylidium caninum: 
new quantitative PCR method for the detection of Anaplasma platys in genetic analysis supporting two distinct species adapted to dogs and 
dogs based on the citrate synthase gene. J Vet Diagn Invest. 2016;28:5. cats. Parasit. 2018;25:30. https:// doi. org/ 10.1 051/p aras ite/ 201802 8.
 28. Courtney JW, Kostelnik LM, Zeidner NS, Massung RF. Multiplex real-time 
PCR for detection of Anaplasma phagocytophilum and Borrelia burgdorferi. 
J Clin Microbiol. 2004;42:3164–8. Publisher’s Note
 29. Renvoise A, Rolain JM, Socolovschi C, Raoult D. Widespread use of Springer Nature remains neutral with regard to jurisdictional claims in pub-
real-time PCR for rickettsial diagnosis. FEMS Immunol Med Microbiol. lished maps and institutional affiliations.
2012;64:126–9.
Ready to submit your research ?  Choose BMC and benefit from: 
• fast, convenient online submission
•  thorough peer review by experienced rese archers in your field
•   rapid publication on acceptance
•  support for research data, including large and complex data types
•  gold Open Access which fosters wider collaboration and increased citations 
•  maximum visibility for your research: over 100M website views per year 
 
At   BMC, research is always in progress.
Learn more biomedcentral.com/submissions