Interactions between ticks and their tortoise hosts are common ecological phenomena (Loehr et al. 2006; Okanga and Rebelo 2006; Tavassoli et al. 2007) and parasites are sometimes responsible in shaping community structures (Begon et al. 2006). Shaping of community structure may be mediated via ticks' potential as reservoirs for pathogens that transmit tick-borne infectious diseases (Ehlers et al. 2016; Wang et al. 2016; MacDonald et al. 2018). While interactions between ticks and cattle, as well as free-ranging mammalian and avian hosts, have been widely reported (e.g., Anderson and Magnarelli 1984; Fyumagwa and Hoare 2005; Horak et al. 2006a, 2007; Guglielmone and Nava 2017; Di Lecce et al. 2018; Ishak et al. 2018), little is known about tick–chelonian interactions in East Africa and especially in and around human-modified habitats.

Some studies have shown that rate of tick infestation in tortoises depend on characteristics of habitats (Stein et al. 2008), seasonality (Fielden and Rechav 1994), body condition (Loehr et al. 2006; Ehlers et al. 2016; Segura et al. 2019), degree of interactions, and the ability of ticks to infest hosts (Ehlers et al. 2016). Furthermore, tick burden on tortoises appears to be associated with individual tortoise traits (Fielden and Rechav 1994), sex (Okanga and Rebelo 2006; MacDonald et al. 2018), and age class (Okanga and Rebelo 2006; MacDonald et al. 2018; Segura et al. 2019).

The pancake tortoise, Malacochersus tornieri (Siebenrock 1903), represents a model species of tortoise with disjunct populations and restricted distribution along the north and southeastern parts of Kenya (Malonza 2003); northern, central, and southern Tanzania (Klemens and Moll 1995: Mwaya et al. 2018, 2019); and to a limited extent, the northern Zambian province of Nakonde (Chansa and Wagner 2006). Although the information on tick infestation on natural populations of M. tornieri is scanty, and first reported by Loveridge and Williams (1957), some information exists on other species of African testunids, especially in southern Africa. For example, Durden et al. (2002) reported a new tick species, Amblyomma geochelone, being hosted by a critically endangered ploughshare tortoise, Astrochelys (Geochelone) yniphora, in Madagascar. In South Africa, Stygmochelys pardalis has been reported to be parasitized by Amblyomma marmoreum while Bell's hinge-back tortoise (Kinixys belliana) hosted mostly Amblyomma nuttalli, and Naqualand speckled padloppers (Chersobius [Homopus] signatus) was observed hosting Ornithodoros campactus and Ornithodoros savignyi (Horak et al. 2006b, 2017). A recent report of tick infestation on a terrestrial tortoise in East Africa is that of Okanga and Rebelo (2006) on introduced Aldabran giant tortoises (Aldabrachelys gigantea) kept in a semicaptive environment in Kenya. This species was found to host 3 different tick species of the genus Amblyomma; the most prevalent was Amblyomma sparsum, followed by A. nuttalli and Amblyomma hebraeum.

So far, there is no systematic research about tick parasitism on M. tornieri in the wild. Because of the species' limited ranging behavior and distribution (Moll and Klemens 1996), these tortoises may probably be relatively less prone to tick parasitism. Furthermore, being an obligate crevice-dwelling tortoise, they might present a more challenging host environment for ticks, mainly due to possible tick dislodgement from the tortoise body as it crawls against the walls of tapering rock crevices. We currently have limited information on the interaction between ticks and M. tornieri and on potential ecological and conservation concern. To our knowledge, nothing is currently known on how tick infestation on natural populations of M. tornieri is related to sex, age class, and site of attachment on tortoise's body. Also, the magnitude of tick infestation in M. tornieri between undisturbed (inside national parks) and disturbed (grazing and farming areas) habitats where humans, their livestock, and M. tornieri come into contact remains unevaluated.

This study, therefore, aimed to assess tick infestation on M. tornieri within Tarangire National Park (TNP) and degraded areas outside of the park. Specifically, we tested if 1) tick infestation in M. tornieri is higher in the adjacent degraded grazing and agricultural lands (outside the park) than within the park; 2) tick burden on M. tornieri varies with seasonality, between sites of attachment on the exterior body, age class, and sex; and 3) tick infestation varies with body condition in M. tornieri. This study is important considering the declining rate of M. tornieri (Mwaya et al. 2019) as a result of illegal collection for the pet trade and human-driven perturbations such as habitat loss and fragmentation. By comparing tick infestation in M. tornieri within pristine habitats and adjacent degraded lands, we can predict potential effects of habitat degradation from tick prevalence on M. tornieri. Highlighting the interactions between ticks and pancake tortoises can also be important to inform future conservation decisions related to translocation and reintroduction programs (e.g., Jørgensen 2014). For instance, the interaction between the endangered tuatara (Sphenedon punctatus) and a tick (Amblyomma sphenodonti) was used in planning the translocation of tuatara from Middle Island and reintroducing the species in Tiritiri Matangi Island of New Zealand (Moir et al. 2012).

METHODS

Study Area Description. — The study was conducted within Tarangire National Park (TNP) and outside areas adjacent to the park. The study sites within TNP included rock-strewn parts of Balloon, Matete, Tarangire Tanzania Telecommunication Company Limited (TTCL), Poachers Hide, Poachers Observation Point, and Mawe Mbiti (Fig. 1). Most of these areas are characterized by rocky terrain dominated by Vachellia (Acacia)–Commiphora deciduous bushland with underneath herbaceous vegetation (Moll and Klemens 1996; Mwaya et al. 2018). Sites inside TNP served as an ecological benchmark (Sinclair et al. 2002) because of its pristine habitats. The sites outside TNP were located in the adjacent degraded grazing and agricultural lands, which comprised north of Vilima Vitatu, Vilima Vitatu north–south, Vilima Vitatu south, Besi, and Sino of the Babati district (Fig. 1). Other sites outside TNP included Jenjeluse of the Chemba district, and Kolo-Kwalua and Kolo-Kinyasi, located in the Kondoa district (Fig. 1). All study sites fall within the semiarid climatic belt of Maasai steppe in the northern and central Tanzania between 3°40′S and 5°35′S and 35°45′E and 37°00′E, and at elevations of 200–1600 m. Precipitation is highly variable and unpredictable between years (Baird and Leslie 2013). The rain season usually occurs between November and May and a dry season from June through October (Mwaya et al. 2018).

View Figure

• Download as Powerpoint
• Download Figure

Data Collection. — Microhabitats of M. tornieri were identified, and systematic searches in rock crevices were conducted by a team of 3 experienced people using flashlights to investigate inside dark crevices. Upon encounter, each tortoise was hand-captured and aged using a carapace length (CL) criterion established for different age classes (Mwaya et al. 2018). Adult tortoises were sexed using sexually dimorphic tail morphology (Klemens and Moll 1995; Spawls et al. 2002) and usually had CL ≥ 90 mm; juveniles and subadults were not sexed (Mwaya et al. 2018). Each captured tortoise was visually examined for tick infestation on 3 main sites of attachment: the shell, mostly on the carapace; the exposed front parts; and the exposed rear parts as shown in Fig. 2. The shell (carapace and plastron) was examined first, followed by exposed front parts, comprising the head, neck region, arms, armpits, and shoulder. Then we examined the exposed rear parts of the tortoise, including legs, groin, and tail. The number of ticks observed in carapace, exposed front parts, and exposed rear parts was recorded separately for comparative analysis. Tick surveys on M. tornieri were repeated at 2- to 3-mo intervals from September 2010 to December 2012. Since M. tornieri individuals may have been encountered more than once, the recorded sample size includes repeated encounters. Each M. tornieri captured for the first time was marked by a combination of notches on the rear marginal scutes according to Cagle (1939), and a tortoise found with ticks was photographed before being released safely back to its original crevice. These photographs were then used for tick identification as our permit did not allow the collection of sample specimens.

View Figure

• Download as Powerpoint
• Download Figure

In addition, we used a separate data set of 155 records of M. tornieri collected by the authors in 2013 whereby the number of ticks, CL (± 1 mm using a Vernier caliper), and body weight (± 1 g using a digital weighing balance) for each pancake tortoise encountered inside and outside the TNP was recorded. The body weights and CLs were then used to ascertain body condition index (BCI) of M. tornieri in response to tick burden. Body condition was used as an indicator for health status of M. tornieri. We used these data to assess if tick infestation on M. tornieri varied BCI.

Data Analysis. — We used online tick guidebooks (Latif and Walker 2016; Madder et al. 2017) to help identify ticks to the genus level. The photographs were further shared with experts in the field.

To assess the difference in tick infestation on M. tornieri with respect to location (inside vs. adjacent degraded grazing and agricultural lands outside the park), age class (adults vs. juveniles), sex (males vs. females), seasonality (dry vs. wet), and site of attachment on tortoise exterior body parts (carapace, exposed front parts, or exposed rear parts), a generalized linear mixed model (GLMM) was used. Binomial distribution was used to fit our zero-inflated data with tick presence or absence on the tortoise body being a response variable. This approach of using a binomial distribution on zero-inflated data was similarly used by Njovu et al. (2019).

Since the crevices were surveyed more than once, there were multiple encounters of M. tornieri and individual crevices; therefore, our GLMM included identification of the crevice where M. tornieri was found and identification of the individual M. tornieri (as noted by notches in carapace) as random effects. As sexual identification of M. tornieri was restricted to adults, we ran 2 GLMMs, one for age class (named as age model) and the other for sex (sex model). In both GLMMs, we included location, season, and site of attachment as standard explanatory variables. Both models included interaction terms of age class and sex and the other standard explanatory variables (location, season, and site of attachment) to assess their cumulative effect in the presence of ticks on the exterior body of M. tornieri. However, we excluded the interaction between age class and season in the age model due to model overfitting. Thus, we used the age model to determine the variation in tick presence on M. tornieri body parts in response to location, seasons, and site of attachment as the model included all individuals assessed, and we did not use the model for sex, as it excluded young M. tornieri. To run the GLMM we used the glmer function from the lme4 package (Bates et al. 2015), and stepwise elimination of model terms based on p-values from type II Wald χ²which was run by using the Anova function from the car package (Fox and Weisberg 2019).

Furthermore, we assessed tick burden in relation to body condition. We first calculated BCI by dividing body weight by CL of each M. tornieri. As tick burden was a count data, a Poisson generalized linear model (P-GLM) was used for subsequent analysis. However, we used a negative binomial generalized linear model (NB-GLM) after P-GLM showed significant overdispersion (z = 2.818, p = 0.002) when subjected under the dispersiontest function of the AER package (Eddelbuettel 2009). In our NB-GLM, the number of ticks was treated as a response variable while CL, body weight, and BCI were explanatory variables. We fitted our NB-GLM by using the glm.nb function of the MASS package (Brian et al. 2019), and backward elimination of model terms based on p-values from a type II Wald χ² test. All statistical analyses were done in R. 3.6.1 software (R Core Team 2019).

RESULTS

Variation of Tick Burden on M. tornieri. — A total of 1217 ticks were encountered in 143 M. tornieri from all surveyed sites. For parasitized tortoises, tick burden ranged from 1 to 219 ticks per tortoise. All ticks identified belong to the family Ixodidae under the single genus Amblyomma. The most likely species, according to expert opinion, was A. nuttalli.

Tick presence varied significantly between locations (χ² = 10.125, df = 1, p = 0.001) with the odds ratio of having ticks being 178% higher in the adjacent degraded grazing and farming lands (outside the park) than within TNP (Fig. 3a). Furthermore, tick burden on M. tornieri varied significantly between dry and wet seasons (χ² = 9.433, df = 1, p = 0.002) whereby higher parasitism was observed during wet season (3.130 ± 0.609 ticks/turtle SE) than during dry season (0.333 ± 0.071 ticks/turtle SE; Fig. 3b). Additionally, tick burden significantly differed between sites of attachment—carapace, exposed front parts, and exposed rear parts (χ² = 10.833, df = 2, p < 0.001)—with tick presence being higher by an average of 2.789 ± 0.353 ticks/turtle SE in exposed front parts and 4.026 ± 0.383 ticks/turtle SE in exposed rear parts than in carapace (Fig. 3c).

View Figure

• Download as Powerpoint
• Download Figure

Tick Infestation on Pancake Tortoise Within Age and Sex Groups. — Further results show that tick presence is not statistically significant (χ² = 3.027, df = 1, p = 0.082) between adults (1210 ticks on 138 tortoises) and juveniles (7 ticks on 5 tortoises). Additionally, tick presence was neither cumulatively affected by age class and location (χ² = 0.425, df = 1, p = 0.514) nor age class and part of attachment (χ² = 0.123, df = 2, p = 0.940). Across sex groups, tick presence between male tortoises (684 ticks on 78 pancake tortoises) and females (526 ticks on 60 pancake tortoises) was not statistically significant (χ² = 0.323, df = 1, p = 0.570). In addition, we did not observe significant interaction between sex and location (χ² = 0.426, df = 1, p = 0.514), sex and season (χ² = 0.005, df = 1, p = 0.943), or sex and site of attachment (χ² = 5.127, df = 2, p = 0.077) in tick presence on M. tornieri.

Tick Burden in Relation to Body Condition. — We found tick burden (in both adults and juveniles) varied significantly with CL (χ² = 13.938, df = 1, p = 0.0002) and BCI (χ² = 6.449, df = 1, p = 0.011). Carapace length had a positive effect on tick burden, increasing it by an average of 38% ± 0.01% SE per millimeter increase in CL (Fig. 4a) while BCI had a negative effect on tick burden on M. tornieri, reducing it by an average of 93% ± 0.35% SE per 1-unit increase in BCI (Fig. 4b). On the other hand, we did not find a significant influence of body weight on a number of ticks found attached on M. tornieri (χ² = 0.0781, df = 1, p = 0.780).

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

Locations, Site of Attachment, and Seasonality of Tick Burden. — Malacochersus tornieri hosted the ectoparasitic tick A. nuttalli in the Tarangire–Manyara ecosystem. The first record of a sole A. nuttalli infecting a type specimen of M. tornieri was provided by Loveridge and Williams (1957). In recent times, in East Africa A. nuttalii has been reported from the Aldabran giant tortoise (Aldabrachelys gigantea) kept in a semicaptive environment in Kenya (Okanga and Rebello 2006). Other tortoises that have been reported to be parasitized by A. nuttalli include the South African chelonians in the genus Kinixys, namely Kinixys zombensis and K. belliana (Horak et al. 2006b, 2017).

Although A. nuttalli appears to parasitize several species within the family Testudinidae, there is generally some degree of tick–host specificity. The Namaqualand speckled padloper (Chersobius [Homopus] signatus), for example, is parasitized mostly by O. compactus and O. savignyi (Loehr et al. 2006) and the majority of leopard tortoises (Stygmochelys pardalis) have been observed to host A. marmoreum (Horak et al. 2006a, 2017). The 2 species of tortoises (C.[Homopus] signatus and S. pardalis) are found sympatrically in South Africa. While it is most likely that M tornieri hosts mostly A. nuttalli, other testunid tick species may also parasitize M. tornieri. This is because of the extended geographical range of the species that spans from the Somali–Maasai biome dominated by Vachellia (Acacia) woodland in Kenya, through most part of Tanzania, to the Miombo woodland dominated by Brachystegia spp. in the Zambesian biome (Briggs and Van Zandbergen 2016) found in northern Zambia. A need, therefore, exists to extend similar research beyond the Tarangire–Manyara ecosystem.

Furthermore, our findings show a significantly higher tick burden on M. tornieri in degraded grazing and farming lands than in pristine habitats within TNP. This pattern suggests that habitat quality plays an essential role in tick prevalence (Ehlers et al. 2016). The observed pattern, in the present study, is unlikely to be attributed by the presence of livestock and domestic dogs (Canis lupus familiaris) in the pancake tortoises habitats outside protected areas, since adult ticks from livestock and dogs do not have an affinity for, and therefore will not, infect chelonians (Latif and Walker 2016; Madder et al. 2017). Instead, low habitat quality, such as reduced herbaceous diversity outside the park, due to anthropogenic activities, could be stressful to the tortoises because of a likely inadequate nutritional base. This may, consequently, compromise tortoise body immunity. This argument is consistent with a report from Ehlers et al. (2016) who found that fewer ticks infested Malagasy tortoises (Astrochelys radiata and Pyxis arachnoides) inside Tsimanampetsotsa National Park than the adjacent degraded grazing and agricultural areas in Madagascar. Another study conducted in Kenya found that tick burden on Aldabran tortoises was approximately 4 times higher outside the Haller Park than in areas within the park (Okanga and Rebelo 2006). In addition, it is also possible that pancake tortoises outside the protected area have extended ranging behavior needed for foraging because herbaceous food source is degraded and can be scarce as opposed to less ranging behavior of the tortoises inside the park. The influence of ranging behavior on tick parasitism has been elucidated by Robbins et al. (1998) for Testudo graeca nikolskii.

We also found that tick infestation varied between sites of attachment on the exterior body of M. tornieri, with most ticks found attached on exposed front parts, comprising the head, shoulder, neck and arms, and exposed rear parts including legs, groin, and tail while few ticks were observed on the carapace. According to Loehr et al. (2006), as observed in South African Namaqualand speckled padlopers, tick attachments on the exposed front and exposed rear parts have 3 advantages: First, they are relatively soft and easily pierceable by tick mouthparts. Second, they are highly vascularized with blood vessels, from which ticks are easily rewarded with their blood meal. Third, these sites provide desirable refuge areas offering both optimal microclimate and somewhat formidable protection against harsh outside environmental conditions.

In contrast, there were no ticks reported to be piercing through the shell's sutures of terrestrial tortoises such as C. (Homopus) signatus (Loehr et al. 2006) and Aldabrachelys gigantea (Okanga and Rebelo 2006). Peculiarly, ticks in pancake tortoises pierced the carapace from weak strategic areas (Mautner et al. 2017), along the sutures that bridge adjoining scutes, especially on sloping marginals and on healed scars of the traumatized shell. Along all these areas on the carapace, ticks are protected from being brushed away as a result of wedging behavior of the tortoise against floor and ceiling of rock crevices. This observation presents an interesting example of evolutionary trade-off: although shell pliability affords exploitation of narrow rock crevices as antipredator escape strategy, this pliability can be regarded as handicap because it exposes the tortoise to an easier tick infestation on the weak carapace.

Furthermore, we found a significant difference in tick burden on M. tornieri between dry and wet seasons whereby higher tick parasitism was recorded during the wet season than the dry season. According to Walker et al. (2001), ticks are less tolerant of the harsh temperature during the dry season. Therefore, ticks pass dry season in diapause and synchronize their life cycle so that the reproductive season coincides with the beginning of the wet season (Loehr et al. 2006).

Age and Sex Patterns of Tick Infestation. — There were neither significant difference in tick burden between juvenile (sexually immature) and adult M. tornieri nor between males and females. A report from Okanga and Rebelo (2006) supports these findings, that tortoise body size and age had little correlation with tick load on Aldabrachelys gigantea. In contrast with other testunids, Robbins et al. (2001) reported a higher tick load on female Rhinoclemmys areolata than on males. Also, Robbins et al. (1998) found that males of the Russian spur-thighed tortoise, T. graeca nikolskii, harbored more ticks than females. All these may be explained by body size sex dimorphism in both R. areolata and T. graeca nikolskii.

Body size sex dimorphism in the pancake tortoises is not very obvious (Mwaya et al. 2018). While Moll and Klemens (1996) reported females of pancake tortoises being morphologically larger than males, Kabigumila (2002) and Malonza (2003) found no body size sex dimorphism. Thus, the possibility of lack of body size sex dimorphism in our study population may explain why there was no difference in tick load between sexes. With regard to lack of differences in tick load between adults and juveniles, this can be a result of a small sample size of juveniles. Juveniles are relatively difficult to find in the wild (Klemens and Moll 1995).

Body Condition and Body Size in Relation to Tick Burden. — Tick burden on M. tornieri was negatively correlated with BCI irrespective of the location of pancake tortoise. This suggests that healthier M. tornieri are predicted to host fewer ticks than the less healthy individuals. This is probably because ticks need to expend more energy penetrating the skin of the healthier individuals or because individuals with more ticks are affected such that their BCI is ultimately lowered. Our findings are consistent with the observation from Segura et al. (2019), who found that body condition of Testudo graeca is negatively related to tick abundance in the Maamora Forest in Morocco. However, Loehr et al. (2006) found no measurable effects of tick infestation on body condition of C. (Homopus) signatus.

On the contrary, we found that M. tornieri with larger CL, irrespective of the age and sex groups, are correlated with a higher number of ticks than the ones with smaller CL. This pattern can be explained by the fact that the large CL correlates with surface area of exposed soft parts of tortoise exterior which, consequently, provides more surface area for more ticks to attach as opposed to individuals with reduced CL.

Management Implication and Limitations. — This study shows that the incidences and severity of tick infestation on M. tornieri are higher in degraded grazing and farming lands outside TNP than areas within the park. Prevalence of tick parasitism in human-encroached habitats could potentially upsurge the prevalence of tick-borne diseases to humans and domestic animals if measures to address the situation are not taken into consideration. Considering the fact that pancake tortoises have been popular in an international live animal trade (Luiijf 1997; Goh and O'Riordan 2007), the transmission of tick-borne diseases abroad seems a likely consequence if no preventive measures are taken (Burridge and Allan 2000; Walker et al. 2001; Burridge and Simmons 2003; Wang et al. 2016).

In this study, we did not incorporate biotic and abiotic characteristics when examining the dynamics of tick infestation on M. tornieri across the 2 habitats. Further study on home range and feeding breadth of M. tornieri is recommended to fully determine the cumulative effects of spatial–temporal variation on tick infestation and its consequences, taking into consideration the levels of disturbance and abiotic–biotic interactions.