Molecular characterisation and morphological description of two new species of Hepatozoon Miller, 1908 (Apicomplexa: Adeleorina: Hepatozoidae) infecting leukocytes of African leopards Panthera pardus pardus (L.)

Background The African leopard Panthera pardus pardus (L.) is currently listed as a vulnerable species on the IUCN (International Union for the Conservation of Nature) red list of threatened species due to ongoing population declines. This implies that leopard-specific parasites are also vulnerable to extinction. Intracellular apicomplexan haemoparasites from the genus Hepatozoon Miller, 1908 have been widely reported from wild carnivores in Africa, including non-specific reports from leopards. This paper describes two new haemogregarines in captive and wild leopards from South Africa and provides a tabular summary of these species in relation to species of Hepatozoon reported from mammalian carnivores. Methods Blood was collected from nine captive and eight wild leopards at various localities throughout South Africa. Thin blood smears were Giemsa-stained and screened for intraleukocytic haemoparasites. Gamont stages were micrographed and morphometrically compared with existing literature pertaining to infections in felid hosts. Haemogregarine specific primer set 4558F and 2733R was used to target the 18S rRNA gene for molecular analysis. Resulting sequences were compared to each other and with other available representative mammalian carnivore Hepatozoon sequences from GenBank. Results Two species of Hepatozoon were found in captive and wild leopards. Of the 17 leopards screened, eight were infected with one or both morphologically and genetically distinct haemogregarines. When compared with other species of Hepatozoon reported from felids, the two species from this study were morphometrically and molecularly distinct. Species of Hepatozoon from this study were observed to exclusively parasitize a particular type of leukocyte, with Hepatozoon luiperdjie n. sp. infecting neutrophils and Hepatozoon ingwe n. sp. infecting lymphocytes. Phylogenetic analysis showed that these haemogregarines are genetically distinct, with Hepatozoon luiperdjie n. sp. and Hepatozoon ingwe n. sp. falling in well supported separate clades. Conclusions To our knowledge, this is the first morphometric and molecular description of Hepatozoon in captive and wild African leopards in South Africa. This study highlights the value of using both morphometric and molecular characteristics when describing species of Hepatozoon from felid hosts.

Background Members of the genus Hepatozoon Miller, 1908 are intracellular apicomplexan haemogregarines (Apicomplexa Levine, 1970: Adeleorina Léger, 1911: Hepatozoidae Wenyon, 1926) widely reported from amphibians, reptiles, birds and mammals, specifically including carnivores such as wild felids [1]. Feline hepatozoonosis was first reported in the early 1900's by Patton [2], who described an intraleukocytic parasite Leucocytozoon felis domestici Patton, 1908 from an domestic cat in India. Eighteen years later, Wenyon [3] distinguished a capsule encasing this parasite that was overlooked by Patton [2], and then reclassified this parasite as a species of Hepatozoon, apparently morphologically identical to those found in hyenas, dogs and jackals [2,3]. In the late 1990's Beaufils et al. [4] proposed that canid and felid Hepatozoon are separate species, due to their morphological differences. The majority of species of Hepatozoon detected in African carnivores were previously identified as Hepatozoon canis (James, 1905), until Levine [5] suggested that host species should be taken into account when determining the identity of these parasites. Peirce et al. [6] supported this by suggesting that researchers are wrong to assume that all species of Hepatozoon from African carnivore are synonyms of H. canis.
From a parasitological perspective, the leopard Panthera pardus (L.) is one of the least studied big cats. Known protozoan parasites in leopards include species of Toxoplasma (Nicolle & Manceaux, 1908), Sarcocystis Lankester, 1882, Hepatozoon, Giardia Künstler, 1882 and Isospora Schneider, 1881 [7]. Brocklesby & Vidler [8] were the first to report Hepatozoon-like organisms from a free-ranging African leopard P. pardus pardus (Linnaeus 1758) in Kenya. In the early 1970's, Keymer [9] reported Hepatozoon canis-like schizonts from the cardiac muscle of a leopard in Central Africa and four years later McCully et al. [10] reported Hepatozoon-like parasites from leopards in South Africa's KwaZulu-Natal Province. More recently, Pawar et al. [11] and Khoshnegah et al. [12] detected what they thought to be Hepatozoon felis (Patton, 1908) from two free-ranging Indian leopards Panthera pardus fusca (Meyer) as well as an unnamed species of Hepatozoon from a single free-ranging Persian leopard Panthera pardus ciscausica (Satunin).
Prior to the 21st century, classification of species of Hepatozoon was based on their life history, host identity and morphological characteristics. However, with recent advances in molecular techniques, phylogenetic analyses on the relationships between species have become possible [13][14][15][16]. Morphological characteristics used to distinguish between species of Hepatozoon include gamont and nucleus dimensions; position of the nucleus within the gamont; number and arrangement of vacuoles and staining properties [17,18], as well as characteristics of other developmental stages [19,20]. In Africa, domestic dogs and wild carnivores have been reported to have cases of asymptomatic hepatozoonosis, caused mostly by H. canis and H. felis [6,10,[21][22][23][24][25]. Although, H. canis and H. felis are not specific to either canids or felids, they have been reported infecting both carnivore families [26][27][28].
Due to constant improvement of molecular techniques, the number of studies on haemogregarines has systematically increased, with several studies relying solely on these methods to detect species of Hepatozoon in their hosts [11,[28][29][30][31][32]. Recent research has shown that a difference in p-distance of between 1-2% of the 18S rRNA gene is sufficient to distinguish between species of haemogregarines if supported by morphological data [16,18,33,34]. Other studies, such as Metzger et al. [35], incorporate both molecular and some degree of morphological investigations, showing the importance of utilizing a more holistic approach when distinguishing among species of Hepatozoon.
The objectives of the present study were to (i) investigate whether captive and wild leopards in South Africa are infected with species of Hepatozoon; (ii) identify any infections found using both molecular analysis of a fragment of the 18S rRNA gene and morphological characteristics of the gamont stage in peripheral blood; (iii) determine if any of the Hepatozoon spp. identified can be linked to hepatozoonosis based on clinical symptoms of the host. This study is the first report on the molecular and morphological characteristics of Hepatozoon species infecting captive and wild leopards in South Africa. van  For this study, leopards are classified as 'captive' if they were born or raised in captivity from an early age (< 1-month-old). Leopards are classified as 'wild' when born or caught at an adult stage in the wild or when held at rehabilitation centres for less than six months to be relocated. Age classes were allocated to all leopards sampled following Fattebert et al. [37], with cubs (< 1 year), subadults (1-3 years) and adults (> 3 years). Wild leopards were aged according to the nature of tooth wear and a combination of morphological cues as described by Stander [38] and Balme et al. [39] respectively. Captive leopards were aged according to the information provided by their ex situ managers.
Prior to blood collection, the dense pelt over the identified area of collection was shaved with a clipper and wiped with an alcohol-soaked cotton ball, which helped remove external skin contaminants and improved visualisation of the vein. Peripheral blood was usually collected from the jugular or the cephalic vein by venipuncture with the use of a sterile Vacutainer system, in BD Vacutainer ® (Franklin Lakes, USA) CAT (Clot Activator Tubes) and Vacutainer ® EDTA tubes for molecular analysis.
Small blood droplets (enough to provide three to four duplicate blood smears) were placed onto clean, prelabelled microscope slides to make thin blood smears. Blood smears were air-dried and subsequently fixed with absolute methanol for one minute. Once dry, blood smears were stored in slide boxes for further processing in the lab. A modified Giemsa (Fluka, Sigma-Aldrich, Steinheim, Germany) stain solution was prepared with distilled water (ratio of 9:1) in a 50 ml staining container. Air-dried blood smears were stained in the Giemsa solution for 20 min, rinsed with a slow stream of distilled water and again left to air dry.

Screening of blood smears
Stained smears were examined under the 100× oil immersion objective of a Nikon Eclipse E800 compound microscope (Nikon, Amsterdam, The Netherlands) and digital images of any infections detected were captured with an attached Nikon DS-Fi1 digital camera and accompanying software. Haemoparasites were identified through comparison of morphometric data to previous studies on species of Hepatozoon from carnivores [12,19]. Parasitaemia was calculated per 100 host cells, with ~ 500 host cells (ten fields of 50 host cells) examined per blood smear. Photomicrographs of blood smears were calibrated according to the guidelines stipulated by the ImageJ Image Processing and Analysis software [40]. Measurements of parasites and leopard blood cells were taken with the ImageJ version 1.47 software program (Wayne Rasband National Industries of Health, USA) (http://image j.nih.gov/ij). All measurements are in micrometres and are given as the range followed by the mean ± standard deviation (SD) in parentheses.

Molecular analysis
Blood samples collected directly into Vacutainer ® EDTA tubes were thawed and used for molecular protocols. DNA was extracted with the KAPA Blood PCR Kit B (Kapa Biosystems, Cape Town, South Africa) according to the protocol provided by the manufacturer. DNA for the dried blood collected from the erythristic leopard was extracted by using the Kapa Express DNA extraction kit (Kapa Biosystems, Cape Town, South Africa), following the manufacturer's protocols.
Haemogregarine-specific primers 4558F (5ʹ-GCT AAT ACA TGA GCA AAA TCT CAA-3ʹ) and 2733R (5ʹ-CGG AAT TAA CCA GAC AAA T-3ʹ) [41] were used for the detection of Hepatozoon species through PCR (polymerase chain reaction) amplification of the 18S rRNA gene. Fragments of between 995 and 1002 nucleotide (nt) were amplified using the primer set as mentioned above. PCRs were performed with 1.25 µl (10 µM) of each primer, 12.5 µl Kapa Blood Mix B, 7.5 µl molecular grade nuclease-free water (Thermo Fisher Scientific, Vilnius, Lithuania) and 2.5 µl whole blood to make up a final volume of 25 µl per sample. The PCR was undertaken in a Bio-Rad C1000 TouchTM Thermal Cycler PCR machine (Bio-Rad, Hemel Hempstead, UK), under the following conditions. Initial denaturation step of 5 min at 95 °C, followed by 35 cycles of denaturation for 30 s at 95 °C, annealing for 30 s at 50 °C and extension for 1 min at 72 °C. This was followed by a final extension of 7 min at 72 °C, and products were held at 4 °C. Resulting amplicons were visualised on a 1% agarose gel stained with gel red and using a Bio-Rad Gel-DocTM XR+ imaging system (Bio-Rad, Hemel Hempstead, UK) under ultraviolet light.
All positive, purified PCR products were sent for sequencing to Inqaba Biotechnical Industries (Pty) Ltd. (IBSA) (Pretoria, South Africa), a commercial sequencing company, for sequencing in both directions. Resultant sequences species identity was verified against previously published sequences using the Basic Local Alignment Search Tool (BLAST) [42]. Haemogregarine species identity was determined by establishing the closest BLAST match (97-100% to existing sequences available on the GenBank database). All sequences matching Hepatozoon spp. were considered positive and, since they were identical within each new species, only one representative sequence of each was included in further analysis.
The software package Geneious R11 (http://www.genei ous.com [43]) was used to assemble and edit resultant sequence fragments. Sequences were aligned using the Clustal W 2.1 alignment tool [44] implemented within Geneious R11. A model test was performed using jMod-elTest 2·1·7 [45], to determine the most suitable nucleotide substitution model, according to the Bayesian information criterion (BIC). The model with the best BIC score was the General Time Reversible [46]  Although eight sequences were obtained from infected leopards, only a single representative of each species of Hepatozoon amplified in the present study was used for phylogenetic analyses. Phylogenetic analyses consisted of two datasets, the first alignment a large dataset (n = 297) including all representative H. felis sequences from GenBank (Additional file 1: Figure S1), and the second alignment based on the results of the first included 57 representative sequences ( Table 2). Bayesian inference (BI) was used to infer phylogenetic relationships. The BI analysis was performed using MrBayes 3.2.2 [47] implemented from within Geneious R11. To assess posterior probability support the Markov Chain Monte Carlo (MCMC) algorithm was run for one million generation for the first larger dataset (297 sequences), and 10 million generations for the second the smaller dataset (57 sequences), sampling every 100 generations and using the default parameters. The first 25% of the trees were discarded as 'burn-in' with no 'burn-in' samples being retained. Results were visualised in Trace, to assess convergence and the ‛burn-inʼ period. Furthermore, uncorrected p-distances for the sequences used were also calculated in PAUP (Phylogenetic Analysis Using Parsimony) version 4.0a152 (Additional file 2: Table S1).   [48], details of the new species have been submitted to ZooBank. The Life Science Identifier (LSID) of the article is urn:lsid:zoobank.org:pub:9E65A924-729F-43A5-AE1F-8001204B6A6A. The LSID for the new name Hepatozoon luiperdjie n. sp. is urn:lsid:zoobank. org:act:2293B0B3-3B91-4BDC-8246-50215B80F8D5. Etymology: The species epithet is derived from the Afrikaans language diminutive name for the host P. p. pardus, which in Afrikaans is referred to as "luiperdjie".

Remarks
Prior to this study, Hepatozoon spp. reported from wild carnivores were only H. felis, H. canis and mostly as an unidentified species of Hepatozoon [6, 10-12, 22, 24, 25, 29, 49-63], except by Hodžić et al. [20] who recently described Hepatozoon silvestris Hodžić, Alić, Prašović, Otranto, Baneth & Duscher, 2017 in an European wildcat from eastern Europe. Infection with a morphologically and genetically distinct Hepatozoon sp. was confirmed by our study in wild African leopards in various areas throughout South Africa, both in male and female hosts. The haemogregarine described here appears to develop only gamont stages in the peripheral blood of P. p. pardus. Therefore, with no division stages detected, it was placed within the genus Hepatozoon.
Hepatozoon luiperdjie n. sp. was on average longer than H. canis from Cerdocyon thous (L.) in Brazil [51], and longer, but with a similar average width to H. felis from Felis catus (L.) in Israel [19], therefore within the relative morphometrical range of both species (Table 3). This haemogregarine was morphometrically most similar to an unnamed species of Hepatozoon detected in Lynx rufus (Schreber) from the USA [57]. It also differed from the other species of Hepatozoon detected in the monocytes of leopards during our study. The most striking feature of H. luiperdjie n. sp. is the densely chromatisized, acentric nucleus, relatively smaller than that of H. felis [19,20]. Hepatozoon luiperdjie n. sp. also seemed to exclusively infect the neutrophils of the host. The lifecycle of this parasite remains to be determined.

Etymology:
The species epithet is derived from that of the Zulu language name for the host P. p. pardus, which in Zulu is referred to as "ingwe". Noun in apposition.

Remarks
This haemogregarine appears to develop only gamont stages in the peripheral blood of P. p. pardus, and with no division stages detected it was placed within Hepatozoon. Hepatozoon ingwe n. sp. measured within the same range size as H. luiperdjie n. sp. described above. However, H. ingwe n. sp. seemed to exclusively infect the lymphocytes of the host, unlike the gamonts of H. luiperdjie n. sp. Hepatozoon ingwe n. sp. was morphometrically similar to an unnamed Hepatozoon detected in the neutrophils of P. p. ciscaucasica from Iran [12] (Table 3). This haemogregarine also measured longer and wider than H. felis from domestic cats in Israel [19], widely considered a redescription of H. felis (Table 3). Hepatozoon ingwe n. sp. was significantly longer (P = 0.0064) and somewhat wider (P = 0.3023), with a comparatively larger surface area than that of H. luiperdjie n. sp. (P = 0.0593) ( Table 3). Characteristic features of H. ingwe n. sp. include the pale blue staining cytoplasm containing bright pink staining granules, prominent vacuoles at the posterior, and elongated nuclei similar to that of H. canis. The life-cycle of this parasite remains to be determined.

Differential diagnoses
Each species of haemogregarine described here infected a particular type of leukocyte: H. luiperdjie n. sp. were exclusively found in neutrophils and H. inwe n. sp. in lymphocytes. The parasitaemia of H. ingwe n. sp. (30.8%), and H. luiperdjie n. sp. (11.0%), was higher than the range of 0.1-4.0% reported by other studies on domestic cats in Israel [64], an ocelot Leopardus pardalis (L.) in Brazil [35], a Tsushima leopard cat Prionailurus bengalensis (Kerr) in Thailand [54], Iriomote cats P. b. iriomotensis (Imaizumi) in Japan [52] and a Tsushima leopard cat P. bengalensis in Japan. The parasitaemia of H. ingwe n. sp. was also higher than that of H. luiperdjie n. sp., possibly indicating that leopard host immune system may be better at suppressing infections by H. luiperdjie n. sp. in neutrophils, than H. ingwe n. sp. in lymphocytes. Additionally, there seemed to be a clear association between sex and parasitaemia in this study, with average parasitaemia in females (H. luiperdjie n. sp. parasitized 18.2% and H. ingwe n. sp. parasitized 35% of host cells) higher than that in males (H. luiperdjie n. sp. parasitized 5.99% and H. ingwe n. sp. parasitized 20% of host cells). This may be a noticeable health concern, since it is still unclear whether feline hepatozoonosis can be transferred within the uterus. However, this trend in parasitaemia needs further investigation over a larger variety and number of felid hosts.
In addition to infecting different types of host cells, the two new species showed clear morphological distinctions of peripheral blood gamont stages on a morphometric basis and differences in staining properties. The gamonts of H. luiperdjie n. sp. are shorter, thinner and with a smaller surface area than those of H. ingwe n. sp. Furthermore, the staining properties of their cytoplasm, the marked presence of vacuoles in H. ingwe n. sp. and the morphometric and staining differences in their nuclei. In terms of nucleus dimensions, the nucleus of H. luiperdjie n. sp. was significantly shorter (P < 0.0001), broader (P < 0.0001) and larger than that of H. ingwe n. sp. (P = 0.0025). Both species described here had a larger surface area compared to the unnamed species of Hepatozoon parasitizing neutrophils of Leopardus pardalis from Brazil [35] and an unnamed species of Hepatozoon parasitising neutrophils of P. p. ciscaucasica from Iran [12] (Table 3). While both new species of Hepatozoon had a smaller surface area as compared to an unnamed species of Hepatozoon from domestic cats [65] and H. canis from Cerdocyon thous from Brazil, respectively [61] ( Table 3). Data on the morphometrics of large numbers of gamonts are scarce, with some studies measuring only a few gamonts [19,20], or even only a single gamont [35]. Our study focused on screening live hosts, while most other studies analysed necropsied or biopsied samples [19,20,66], or did not focus on morphological descriptions of the gamont stage [10,11,22,24,27,29,31,52,53,55,58,64,[67][68][69]. Several of these studies reported only on molecular detection due to the general low parasitaemia of these haemogregarines [66], as confirmed by the absence of gamonts in the peripheral blood smear of CF5 in our study.

Molecular analyses
Amplicons were derived from H. luiperdjie n. sp. and H. ingwe n. sp. from the blood of leopards and the details of all sequences used for analysis are presented in Table 2. Although eight sequences were obtained from infected leopards, only a single sequence per species of Hepatozoon was used in the phylogenetic analysis (Fig. 2), as sequences obtained from samples with single species infections were identical for the respective new species. Sequences obtained from leopards with mixed Hepatozoon infections were not included in the phylogenetic analysis, as sequences contained a double chromatogram peak or two separate bases called at the same position (heterozygous positions) from the two new species of Hepatozoon amplified. Based on the uncorrected p-distance for the 18S rRNA gene between H. ingwe n. sp. and H. felis [amplified from the Asiatic lion, Panthera leo persica (L.) (GenBank: HQ829440) and Indian leopard, Panthera pardus fusca (L.) (GenBank: HQ829444)] interspecific divergence was 1.0% (Additional file 2: Table S1). Hepatozoon luiperdjie n. sp. and H. felis amplified from domestic cats, F. catus (GenBank: AY620232 and AY628681) and from a Bengal tiger, P. tigris tigris (L.) (GenBank: HQ829445) had an interspecific divergence of 1.0% (Additional file 2: Table S1). The interspecific divergence between H. luiperdjie n. sp. and H. ingwe n. sp. was 1.0%.
For the BI phylogenetic analyses, first a large dataset was used comprising 297 sequences (Additional file 1: Figure S1). This analysis included all the available 18S rDNA sequence data of H. felis downloaded from Gen-Bank. The genera Hemolivia, Hepatozoon (parasitising amphibian, reptilian and rodent hosts), Karyolysus, Haemogregarina and Dactylosoma formed separate and well-supported clades at the base of the phylogeny (Additional file 1: Figure S1). Species of Karyolysus and several species most likely incorrectly identified as species of Hepatozoon, formed a sister group to a large clade comprising species of Hepatozoon from large mammals.  [19].
The second BI phylogenetic analysis was based on 57 18S rDNA sequences (Fig. 2). Species of Hepatozoon isolated from large mammal hosts formed a large wellsupported clade (Clades A-E, Fig. 2). Clades A and B (monophyletic group), and Clades C, D and E, formed a polytomy of four distinct groups. In Clade A, H. ingwe n. sp. was shown as a sister taxon to a well-supported monophyletic cluster of H. felis in big cats from India, namely the Asiatic lion (GenBank: HQ829440) and Indian leopard (GenBank: HQ829444) (Clade A in Fig. 2). Furthermore, H. silvestris (GenBank: KX757032), isolated form the European wild cat Felis silvestris silvestris Schreber, and H. apri (GenBank: LC314791) isolated from the Japenese boar, Sus scrofa leucomystax Temminck formed a polytomy at the base of clade A. In clade B, H. luiperdjie n. sp. clustered with sequences isolated from H. felis in domestic cats from Spain (Gen-Bank: AY620232; AY628681), a bengal tiger, P. tigris tigris (GenBank: HQ829445) from India, and H. felis amplified from a tick Rhipicephalus sanguineus (Latreille) that was collected from an Asiatic lion Panthera leo persica (Gen-Bank: KY056823). Clade C was a monophyltic clade of H. van As et al. Parasites Vectors (2020) 13:222 canis and species of Hepatozoon isolated from various canid hosts. In clade D, H. martis isolated from the European badger Meles meles (L.) and the European pine marten Martes martes (L.), formed a sister group to H. ursi isolated from the Japanese black bear Ursus thibetanus japonicus Schlegel and the sloth bear Melursus ursinus (Shaw). Clade E comprised H. americanum, isolated from domestic dogs, Canis familiaris L. and an unnamed species of Hepatozoon in crab-eating foxes Dusicyon thous azarae (L.).   [22] and East et al. [25]. Prevalence of Hepatozoon varies between hosts and regions. The overall prevalence of hepatozoonosis recorded in this study was 56%, which is similar to that reported in Iriomote cats Prionailurus bengalensis iriomotensis from Japan (56.7%) [55], spotted hyenas Crocuta crocuta (Erxleben) from Zambia (56%) [28], and captive Asiatic lions P. l. persica from India (55.56%) [11]. Prevalence found during our study was slightly higher than the prevalence of unknown species of Hepatozoon reported from Indian leopards P. p. fusca (50%) [11] and much higher than that of African wild dogs Lycaon pictus (Temminck) in South Africa (0.7%) [31]. The prevalence of H. luiperdjie n. sp. (50%) was similar to that of an unknown species of Hepatozoon reported from Indian leopards P. p. fusca in India (50%) [11], and it was lower than the prevalence of H. ingwe n. sp. (44%). The prevalence of both new species described in this study was higher in males than females, with 63% of males infected with H. luiperdjie n. sp. and 50% of males infected with H. ingwe n. sp., and 38% of females infected with H. luiperdjie n. sp. and H. ingwe n. sp., respectively.
The two new haemogregarines had dissimilar effects on their respective host cells. Hepatozoon luiperdjie n. sp. caused enlargement of neutrophil cells and their nuclei and H. ingwe n. sp. reduced the size of lymphocytes and condensed their nuclei. Although Baneth et al. [19] reported on a species of Hepatozoon that infects both neutrophils and lymphocytes, our study showed that co-infecting species of Hepatozoon can inhabit different types of leukocytes, with different effects and morphological characteristics of their gamont stages.
It is evident based on morphological and molecular data that H. luiperdjie n. sp. and H. ingwe n. sp. are distinct species. These species are also distantly related to H. felis (GenBank: AY628681) based on 18S rDNA sequence comparisons, isolated by Criado-Fornelio et al. [50] from domestic cats from Spain. The H. felis isolates from Spain [50] are widely regarded as the representative H. felis isolates to be used for comparison [25,55,60]. Thus, based on the phylogenetic relationships and comparisons of H. felis and H. felis-like species of Hepatozoon, the identity of the Hepatozoon infecting large carnivores, currently identified as H. felis by Pawar et al. [11] is questioned. Therefore, based on these genetic distinctions, as well as morphological characteristics and effect on host cells, the two haemogregarines described here were deemed to be two different species and new to science. The phylogenetic results from our study showed H. luiperdjie n. sp. and H. ingwe n. sp. as distinct species as compared to the currently recognised species of Hepatozoon infecting large mammal hosts, i.e. H. americanum, H. apri, H. canis, H. felis, H. martis, H. silvestris and H. ursi. We therefore suggest that these haemogregarines may need to be re-classified based on morphological, morphometric and molecular analysis.
Prior to our study, only H. felis, H. canis and several unknown species of Hepatozoon have been reported from African carnivores [6,8,10,22,24,28,31,57,64,65], but this study confirmed a mixed population of two genetically distinct haemogregarines from both captive and wild leopards, across males and females and from leopards representative from three core populations as identified by Daly et al. [36]. The topology of our BI tree confirmed the suggestion of Hodžić et al. [20], that H. felis should be viewed as a species complex.

Conclusions
As shown in this study, morphology and the effect on host cells are important parameters that should be taken into account when identifying species of Hepatozoon. By using different techniques of identification, a better understanding of the parasite and its relation to its host may become possible. Thus, Hepatozoon species identified as either H. felis or H. canis based on the host parasitized should be re-evaluated using both morphological and molecular characteristics, as well as the type of and effect on the host cells. The value of the results from our study, in addition to describing two new haemogregarine species, is that we present results obtained from live animals in the wild. It is important to identify these parasites to species level in order to better understand potential zoonotic effects on different host species, and to further investigate the possible transfer of haemogregarines from wild to domestic animals. In addition, this paper provides valuable criteria to be considered when describing Hepatozoon infections from wild carnivores. Possible future work on these haemogregarines should include in depth investigations on the life cycles and vectors of these species.

Supplementary information
Supplementary information accompanies this paper at https ://doi. org/10.1186/s1307 1-020-3933-6.  Table S1. Estimates of divergence using partial 18S rDNA sequences from the haemogregarines species used in the current study. Distance matrix showing ranges as percentage for the genetic p-distances between the sequences. Alignment length 1924 nt.