Assessing the impact of preventative measures to limit the spread of Toxoplasma gondii in wild carnivores of Madagascar

Abstract

Novel multihost pathogens can threaten endangered wildlife species, as well as humans and domestic animals. The zoonotic protozoan parasite Toxoplasma gondii is transmitted by members of Felidae and can infect a large number of animal species, including humans. This parasite can have significant health consequences for infected intermediate hosts and could further endanger wild carnivore populations of Madagascar. Building on an empirical characterization of the prevalence of the pathogen in local mammals, we used mathematical models of pathogen transmission in a multihost community to compare preventative measures that aim to limit the spread of this parasite in wild carnivores. Specifically, we examined the effect of hypothetical cat vaccination and population control campaigns on reducing the risk of infection by T. gondii in wild Eupleridae. Our model predicted that the prevalence of exposure to T. gondii in cats would be around 72% and that seroprevalence would reach 2% and 43% in rodents and wild carnivores, respectively. Reducing the rodent population in the landscape by half may only decrease the prevalence of T. gondii in carnivores by 10%. Similarly, cat vaccination and reducing the population of definitive hosts had limited impact on the prevalence of T. gondii in wild carnivorans of Madagascar. A significant reduction in prevalence would require extremely high vaccination, low turnover, or both in the cat population. Other potential control methods of T. gondii in endangered Eupleridae include targeted vaccination of wild animals but would require further investigation. Eliminating the threat entirely will be difficult because of the ubiquity of cats and the persistence of the parasite in the environment.

INTRODUCTION

Toxoplasmosis is caused by the protozoan parasite Toxoplasma gondii. Although its only definitive hosts are members of the felid family, T. gondii can infect diverse homeotherm species, including humans (Tenter et al., 2000). An estimated 30% of the global human population is infected by the parasite, which can cause serious health problems in vulnerable populations, including encephalitis, ocular disease, birth defects, or abortions. It can also result in major economic losses in the livestock industry (Flegr et al., 2014; Stelzer et al., 2019). Similarly, exposure to this parasite has resulted in acute or chronic diseases and even death in several wildlife species, including endangered animals (Dubey et al., 2020). The significant impact of this parasite on human, livestock, and wildlife health, its wide species and global geographic distribution, and its unique life cycle call for One Health approaches to understanding, preventing, and managing the disease (Aguirre et al., 2019; Innes et al., 2019).

Toxoplasma gondii has a complex life cycle that involves multiple hosts and includes sexual and asexual replication. Felids (domestic cats and their relatives) are the only definitive hosts in which the parasite can reproduce sexually to generate oocysts. Cats shed a large number of unsporulated oocysts in the environment through their feces. It is commonly assumed that oocyst shedding is transient; it is estimated to occur for only 1−2 weeks after the felid host's first infection (Dubey, 1995). Within 1−5 days, oocysts in the environment sporulate and become infectious (Dubey et al., 2011). Oocysts persist in moist soil for 6 months on average and up to 18 months, depending on the climate (Dumètre & Dardé, 2003). All warm‐blooded vertebrates can serve as intermediate hosts, which acquire infection through ingesting soil, water, or plants contaminated with oocysts or by consuming bradyzoites in other hosts tissues. Definitive hosts (any felid species) can become infected by consuming sporulated oocysts in the environment or by preying on intermediate and paratenic hosts. In rare cases, congenital transmission from infected mothers to offspring can occur.

The complexity of the T. gondii life cycle means that characterizing the determinants of the burden of this pathogen and exploring prospects for its control may benefit from the development of a mathematical model. Formalizing the process of transmission in the context of a multihost community opens the way to comparing and optimizing preventative measures and, specifically, the impact of these measures on hosts that are of conservation concern, such as the endangered carnivores of Madagascar. A few mathematical models have been built to investigate the transmission of T. gondii in definitive and intermediate hosts (Deng et al., 2020; González‐Parra et al., 2023), but none have tackled the conservation impacts of preventative measures.

In natural habitats, rodent (or other intermediate hosts’) control cannot be used to limit T. gondii (Kijlstra et al., 2008). This leaves 2 possibilities for limiting exposure for wild carnivores: cat vaccination, a strategy used for example in the parallel situation of dog vaccination for protection of wildlife from rabies (Cleaveland et al., 2006), and slowing the turnover of the cat population, which will limit T. gondii spread because infected cats only shed oocysts for a limited time after their first infection and usually do not shed further for the duration of their life (but see Zhu et al, 2022). Although no vaccines are currently commercially available to prevent oocyst shedding in cats and protect against T. gondii in intermediate hosts, several candidates have shown promising results (Innes et al., 2019).

Our focal, vulnerable population in Madagascar is a unique and endemic family (Eupleridae) of 7 monospecific genera that has evolved in isolation for at least 20 million years (Veron, 2020; Yoder et al., 2003). Felids (Felis sp.) were introduced with successive human colonization of the island, starting as early as the 17th century, and have since colonized the entire island (Farris et al., 2015; Rasambainarivo et al., 2017; Sauther et al., 2020).

In Madagascar, serological studies suggest that a large proportion of humans and wild carnivores are exposed to T. gondii (Lelong et al., 1995; Pomerantz et al., 2016; Rasambainarivo et al., 2018). For example, serological studies indicate that 92% of fosa (Cryptoprocta ferox), 48% of ring tailed vontsira (Galidia elegans), and 27% of broad striped vontsira (Galidictis fasciata) in protected areas and 82% of cats in villages neighboring protected areas are exposed to T. gondii (Pomerantz et al., 2016; Rasambainarivo et al., 2018). Although its effects on wildlife at the population level are unknown, T. gondii causes disease, reduced fitness, and death in several species of captive wild animals from Madagascar, including Eupleridae (Browning et al., 2021; Corpa et al., 2013; Spencer et al., 2004). Therefore, it is crucial to evaluate strategies for limiting the spread of this parasite in wildlife populations that may be further endangered by this introduced pathogen.

We modeled the dynamics of T. gondii in a multihost population with a range of empirical data on host serological status to ground our analysis and assessed the effects of preventative measures, including a hypothetical cat vaccination and cat population control campaigns, on reducing the risk of infection by T. gondii in wild carnivores of Madagascar.

METHODS

Modeling

To study the dynamics of infection by toxoplasmosis in the environment and a community of interacting domestic (cats), peridomestic (rodents), and wild (carnivores) animals, we built a deterministic compartmental model adapted for the distribution of multiple hosts present in Madagascar that considers the direct and indirect life cycles of T. gondii (Figure 1).

FIGURE 1.

Model of Toxoplasma gondii transmission among domestic cats, rodents, the environment, and wild carnivores.

Our model included susceptible (S _c), infectious (I _c), and recovered and resistant (R _c) components that describe the state of individuals in a total population (N _c) of cats. The cat population grew following a logistic growth model with an intrinsic growth rate: r _c = b _c − m _c, where b _c and m _c are the natural birth rate and mortality rate of cats, respectively. We assumed that N _c is in equilibrium so its value was the same as the carrying capacity (K _c = N _c) (Table 1).

TABLE 1.

Parameters in the model of the transmission dynamics of Toxoplasma gondii among domestic cats, rodents, the environment, and wild carnivores.

Parameter	Value
Birth rate of cats (b c)	0.3/52
Mortality rate of cats (m c)	0.3/52
Transmission rate from contaminated environment to cats (βc)	0.54/52
Probability of infection when predator consumes an infected rodent (g)	1.0
Number of rodents killed·cat−1·week−1 (predation rate) (a c)	104/52
Number of rodents killed·carnivore−1·week−1 (predation rate) (a r)	104/52
Birth rate of rodents (b r)	16/52
Mortality rate of rodents (m r)	2/52
Transmission rate from contaminated environment to rodents (βr)	0.4/52
Birth rate of wild carnivores (b f)	1/52
Mortality rate of wild carnivores (m f)	1/52
Transmission rate from contaminated environment to wild carnivores (βf)	0.54/52
Recovery rate (number recovered/week) (γ)	0.5
Contamination rate of environment by one infected cat (λ)	1/16
Decontamination rate (d)	1/26

Cats may acquire T. gondii directly from contact with the contaminated environment at a rate β_c or indirectly from consumption of infected prey (intermediate hosts). We assumed that cats preyed on rodents or other intermediate hosts at a rate a _c and with a probability (g) of acquiring the infection when consuming an infected intermediate host. We further assumed that recovered cats maintain life‐long immunity (Dubey, 1995).

The environment becomes contaminated through the shedding of oocysts by cats, and infected cats contaminate the environment at a rate λ, and infected areas decontaminate at a rate d. The estimated persistence of the parasite in moist soil varies from 46 days to 18 months depending on temperature and humidity. Using available values of T. gondii infectivity from multiple lines of research (Dumètre & Dardé, 2003), we averaged the survival of the parasite in moist soil at 180 days (26 weeks); thus, the decontamination (d) rate is 0.03.

The rodent population (N _r) grew following a logistic growth rate with an intrinsic growth rate r _r and a carrying capacity K _r. Rodents became infected through contact with the environment at a rate β_r. We also assumed that rodents develop life‐long infections when acquiring parasites from a contaminated environment. Because of the low probability of repeated vertical transmission in rats (Rattus norvegicus) (Dubey et al., 1997), we assumed that infected rodents gave birth to susceptible individuals.

Infections of wild carnivores by T. gondii do not influence the definitive and intermediate host transmission cycle of T. gondii because they are dead‐end hosts (Dubey, 1995). We further assumed that wild carnivores develop life‐long infections when acquiring parasites from a contaminated environment or consumption of infected rodents. Therefore, we used a susceptible‐infected system of ordinary differential equations to model the spread of infection in wild carnivores.

Based on the limited evidence of vertical transmission in wild carnivores, we assumed that infected and recovered individuals from all species gave birth to susceptible individuals.

The set of equations that describes the spread of infection in this community are as follows.

In cats,

$$\frac{dN\mathrm{c}}{dt}=\left(b\mathrm{c}-m\mathrm{c}\right)\times N\mathrm{c}\times \left(1-\frac{N\mathrm{c}}{K\mathrm{c}}\right),$$

$$\begin{array}{ccc}\hfill \frac{dS\mathrm{c}}{dt}& =& b\mathrm{c}\times \left(1-v\mathrm{c}\right)\times N\mathrm{c}-\beta \mathrm{c}\times E\times S\mathrm{c}-m\mathrm{c}\times N\mathrm{c}\hfill \\ & & -g\times a\mathrm{c}\times S\mathrm{c}\times \frac{I_{\mathrm{r}}}{N_{\mathrm{r}}},\hfill \end{array}$$ (1)

$$\frac{dI\mathrm{c}}{dt}=\beta \mathrm{c}\times E\times S\mathrm{c}+g\times a\mathrm{c}\times S\mathrm{c}\times \frac{I\mathrm{r}}{N\mathrm{r}}-m\mathrm{c}\times I\mathrm{c}-\gamma \times I\mathrm{c},$$

and

$$dR\mathrm{c}=\gamma \times I\mathrm{c}+b\mathrm{c}\times v\mathrm{c}\times N\mathrm{c}-m\mathrm{c}\times R\mathrm{c}.$$

In rodents,

$$\frac{d{N}_{\mathrm{r}}}{dt}=\left(b\mathrm{r}-m\mathrm{r}\right)\times N\mathrm{r}\times \left(1-\frac{N\mathrm{r}}{K\mathrm{r}}\right),$$

$$\begin{array}{ccc}\hfill \frac{d{S}_{\mathrm{r}}}{dt}& =& b_{\mathrm{r}}\times N_{\mathrm{r}}-\beta \mathrm{r}\times E\times S\mathrm{r}-m\mathrm{r}\times S\mathrm{r}-g\times a\mathrm{c}\times N\mathrm{c}\hfill \\ & & \times \frac{S\mathrm{r}}{N\mathrm{r}}-g\times a\mathrm{f}\times N\mathrm{f}\times \frac{S\mathrm{r}}{N\mathrm{r}},\hfill \end{array}$$ (2)

and

$$\begin{array}{ccc}\hfill \frac{dI\mathrm{r}}{dt}& =& \beta \mathrm{r}\times E\times S\mathrm{r}-m\mathrm{r}\times I\mathrm{r}-g\times a\mathrm{c}\times N\mathrm{c}\times \frac{I\mathrm{r}}{N\mathrm{r}}-g\times a\mathrm{f}\hfill \\ & & \times N\mathrm{f}\times \frac{I\mathrm{r}}{N\mathrm{r}}.\hfill \end{array}$$

In wild carnivores,

$$dN\mathrm{f}/dt=\left(b\mathrm{f}-m\mathrm{f}\right)\times N\mathrm{f}\times \left(1-\frac{N\mathrm{f}}{K\mathrm{f}}\right),$$

$$\begin{array}{ccc}\hfill \frac{dS\mathrm{f}}{dt}& =& b\mathrm{f}\times N\mathrm{f}-\beta \mathrm{f}\times E\times S\mathrm{f}-m\mathrm{f}\times S\mathrm{f}-g\times a\mathrm{f}\hfill \\ & & \times S\mathrm{f}\times \frac{I\mathrm{r}}{N\mathrm{r}},\hfill \end{array}$$ (3)

and

$$\frac{dS\mathrm{f}}{dt}=\beta \mathrm{f}\times E\times S\mathrm{f}-m\mathrm{f}\times I\mathrm{f}+g\times a\mathrm{f}\times S\mathrm{f}\times \frac{I\mathrm{r}}{N\mathrm{r}}.$$

In the environment

$$dE/dt=\lambda \times I_{\mathrm{c}}\times \left(1-M\times E\right)-d\times E.$$ (4)

Simulations were run for 10 years (520 weeks) with a timestep of 1 week.

Baseline model parameters (Table 1) were obtained from the literature (Afonso et al., 2006; Lélu et al., 2010; Marinović et al., 2020). To perform the numerical simulations, we considered cat life expectancy in rural Madagascar was 3 years. Therefore, we assumed a baseline constant population with equal birth and mortality rates (bc = mc = 0.3/52). This baseline was modulated under exploration of different control methods. We initially assumed that the diets of cats in rural Madagascar were unsupplemented and consisted mainly of wild caught prey, similar to that of wild animals (i.e., each cat and wild carnivore preyed on 2 rodents/week, which is the average predation rate of rural cats in other environments [Loyd et al., 2013]). We further assumed that in our setting, based on temperature and humidity in Madagascar, T. gondii oocysts would remain viable and infectious for approximately 6 months (26 weeks).

Population sizes were set initially at 100 cats, and populations included 1 infectious individual actively shedding oocysts, 100 wild carnivores, and 1500 rodents (Table 2). For all 3 populations, increments and decrements to population size occurred only by birth and death. We assumed that there was no immigration or emigration for model simplification. Altering these values did not alter outcomes.

TABLE 2.

Initial values of parameters in the model of the transmission dynamics of Toxoplasma gondii among domestic cats, rodents, the environment, and wild carnivores.

Parameter	Value
Susceptible cats (S c)	99
Infectious cats (I c)	1
Recovered cats (R c)	0
Proportion of Environment contaminated (E)	0
Susceptible rodents (S r)	1500
Infected rodents (I r)	0
Susceptible euplerids (S f)	100
Infected euplerids (I f)	0

Model‐derived estimates were compared with observed seroprevalence data from domestic cats, wild carnivores, and rodents obtained from cross‐sectional surveys conducted in Madagascar (Pomerantz et al., 2016; Rasambainarivo et al., 2018, Rasambainarivo et al., unpublished data)

Sensitivity analyses

To explore which parameters had the greatest impact on the prevalence of infection in carnivores, we performed local sensitivity analyses (Figure 2). While keeping all other parameters constant, we modified the value of each model parameter across their range published in the literature and measured the change in model‐predicted prevalence of infection of carnivores. Unspecified parameter ranges were assigned by dividing (lower bound) and multiplying (upper bound) the parameter estimate by 2.

FIGURE 2.

Change in seroprevalence of Toxoplasma in wild carnivores relative to disease parameters (order of parameters, greatest impact at top; vertical line, parameters at their respective base values; length of bars, uncertainty associated with each parameter; orange, values in the upper range; green, values in the lower range).

Preventative measures

We evaluated the effect of a cat vaccination campaign on the seroprevalence of endemic carnivores against T. gondii. A hypothetical vaccine was assumed to provide life‐long immunity against infection in all vaccinated cats. We further assumed that a proportion of cats ($v$ _c) can be vaccinated before they can get infected and vaccination completely prevents shedding of T. gondii oocysts from vaccinated cats (Mateus‐Pinilla et al., 2002)

Sterilization campaigns through trap neuter release or trap vasectomy–hysterectomy release are methods commonly used to limit the population of cats and may influence pathogen spread through a reduction in births, lower cat population turnover, and thus decrease the replenishment of susceptible cats. To assess the effects of a cat spay and neuter program on the infection of wild carnivores, we performed our analyses while varying the birth rate of cats.

RESULTS

Our model predicted that at equilibrium, the prevalence of exposure to T. gondii in cats would be around 72% and that in rodents and wild carnivores, seroprevalence would reach 2% and 43%, respectively. The predicted values were in agreement with the seroprevalence observed in carnivores, cats, and rodents from multiple protected areas of Madagascar (Pomerantz et al., 2016; Rasambainarivo et al., 2018; F.R., personal observation).

Our sensitivity analysis indicated that uncertainty as to transmission to rodent populations alongside rodent population size was the core determinant of our predicted prevalence within carnivores; transmission rate from the environment to carnivores and carnivore populations was of secondary importance (Figure 2). However, this range of parameter uncertainty still yielded predictions within the range observed in natural populations (Pomerantz et al., 2016; Rasambainarivo et al., 2018; F.R., personal observation).

We modeled the prevalence of infection in wild carnivores as a function of the proportion of susceptible cats that were vaccinated and the longevity of cats. Our results showed that even with high vaccination efforts, a highly efficacious vaccine, and low cat population turnover, eliminating T. gondii from wild carnivores was highly unlikely. In the absence of vaccination, when cats lived <4 years on average, >40% of carnivores were infected by T. gondii (Figure 3). Extending the longevity of cats and thus reducing the felid population turnover near protected areas had only a moderate effect on the prevalence of exposure to T. gondii in wild carnivores (reduction of <20%). Extremely high cat vaccination coverage (>90%) was necessary to eliminate T. gondii from the wild carnivore population.

FIGURE 3.

Predicted seroprevalence of Toxoplasma gondii in wild carnivores as a function of the percentage of cats vaccinated and their longevity.

DISCUSSION

Our results provide a first characterization as to which parameter uncertainties most contribute to uncertainty in T. gondii prevalence in definitive and intermediate hosts. They indicated that further expanding data on rodent acquisition of infection and rodent population dynamics are critical directions for future research. Many small rodent populations have erratic dynamics characterized by multiannual population cycles or irregular outbreaks, which can have major ecological and economic impacts and important implications for the transmission dynamics of infectious diseases. Our results also highlighted the importance of oocyst survival time (inverse of d). Documented reductions in oocyst survival associated with temperature and desiccation (Dumètre & Dardé, 2003; Lélu et al., 2012) suggest that global warming may affect prevalence of T. gondii in intermediate hosts, including endemic carnivores (Bouchard et al., 2022; Pilfold et al., 2021). Richer characterization of T. gondii prevalence across a range of climatic environments would open the way to bounding expectations for this outcome.

Our analyses revealed the difficulty of controlling the parasite in a multihost environment. Across most scenarios, T. gondii persisted in wild carnivores, although a combination of interventions resulted in a lower seroprevalence. Although the development of vaccines against T. gondii is an exciting development in this landscape, sustained vaccination efforts are required given the rapid turnover in cat populations. Our results suggest that vaccination of cats could reduce the risks of infection in carnivores if a large proportion of cats were vaccinated (>90%). In practice, this would mean vaccinating all newborn kittens in the first few weeks of their lives. The vaccination scenario presented here is a best‐case scenario in which cats live in a closed population; cats are vaccinated before they become infectious, often at a young age; and the vaccine provides life‐long immunity. In an open population with high turnover, such as in feral or free‐ranging cat populations, the probability of attaining such rates of vaccination is unlikely.

Alternatively, vaccination of the endangered carnivores may provide a valuable option to protect wildlife against the disease, as demonstrated by the analogous scenario of rabies in African wild dogs (Lycaon pictus) (Hofmeyr et al., 2004; Prager et al., 2011). Nevertheless, the vaccination of endangered wildlife populations has historically faced significant controversy, and the application of an off‐label vaccine on wild populations would require careful examination (Gilbert et al., 2020; Woodroffe, 2001).

At the time of writing, no vaccine is approved and commercially available to prevent the infection and the shedding of oocysts from cats or the disease in wildlife. Currently, only one live attenuated vaccine, Toxovax, is commercially available to reduce the damage caused to the sheep industry by congenital toxoplasmosis in Europe and New Zealand. Another live attenuated strain vaccine derived from the strain T‐263 is also being considered and reportedly blocks the excretion of oocysts in cats fed live bradyzoites (Freyre et al., 1993; Zhang et al., 2022). Although live attenuated vaccines seem to be highly effective, inducing host cellular and humoral immunity against T. gondii without causing disease, they also have drawbacks, including short shelf life, safety issues for handling personnel, and the possibility of reversion to virulence phenotypes.

Both reducing births and increasing survival will slow cat population turnover, assuming there is some upper bound on cat populations. The devil is likely in the details. It has been suggested that even low levels of demographic connectivity can significantly reduce the effectiveness of management interventions structured around sterilization (Miller et al., 2014). In addition, this strategy assumes that infected cats develop life‐long immunity and do not shed oocysts on reinfection, a widely held assumption that is questioned. Recent studies suggest that reshedding occurs under certain conditions, particularly when cats are infected with heterologous strains (Zhu et al., 2022; Zulpo et al., 2018). This further emphasizes the challenging nature of eliminating this threat and highlights crucial areas for further exploration, including the T. gondii genotypes circulating in Madagascar.

We used theoretical models and empirical data to explore the T. gondii life cycle in Madagascar and to investigate paths to reducing impact on endangered carnivore species. Based on our results, we suggested control methods that could reduce exposure of wild carnivores and highlighted the enormous challenge in eliminating this threat entirely. Because the parasite is expected to persist, there is an urgent need to enhance understanding of its distribution and impact on endangered wildlife.

Rasambainarivo, F. T. , Randrianarisoa, S. , Rasolofoniaina, O. A. , Rice, B. L. , & Metcalf, C. J. E. (2024). Assessing the impact of preventative measures to limit the spread of Toxoplasma gondii in wild carnivores of Madagascar. Conservation Biology, 38, e14300. 10.1111/cobi.14300