Giant anteaters (Myrmecophaga tridactyla, Linnaeus 1758) are the largest anteater species and one of the most charismatic Neotropical mammals (Diniz and Brito, 2012). This insectivorous mammal has a wide geographical distribution ranging from Belize and Guatemala to the north, throughout South America, excluding the Andes region, to Northern Argentina (Wetzel, 1985). Despite this wide distribution, the overall population trend for the species is decreasing (Miranda et al., 2014). For instance, the species is considered the most threatened mammal of Central America, and in several parts of its range it is in danger of extinction due to habitat loss, road kill, human-wildlife conflict and sometimes hunting; therefore, this mammal is listed as Vulnerable by the IUCN SSC (Miranda et al., 2014). Several efforts have been made to study the species’ ecology (Shaw et al., 1987; Miranda, 2004; Medri and Mourão, 2005; Miranda et al., 2006; Macedo et al., 2010; Rojano-Bolaño et al., 2015; Di Blanco et al., 2015, 2017). Most of these initiatives have taken place in South America. Furthermore there are several long-term initiatives dedicated to the species (e.g. Anteaters & Highways Project - http://www.giantanteater.org/; Proyeto Hormiguero - https://www.cunaguaro.co/anteater-project; Iberá Project - http://www.proyectoibera.org/en/english/especiesamenazadas_osohormiguero.htm; Instituto de Pesquisa e Conservação de Tamanduás no Brasil - http://www.tamandua.org/). However, there are still important gaps in the knowledge about the basic life history and population biology of this species. This lack of data can be challenging when trying to define conservation strategies and prioritize conservation actions in an effort to maximize effective resource use.

Population viability analysis (PVA) is an important tool used in conservation biology to assess the viability of populations and rank alternative management scenarios (Reed et al., 2002; Radchuk et al., 2016). In addition, PVA is useful for testing the relative importance of model parameters through sensitivity analysis and identifying knowledge gaps (Tuberville et al., 2012). A well-delimited sensitivity analysis is a requisite to identifying parameters that are likely to have a strong effect on the system (Pe’er et al., 2013). This analysis is a good choice to test the impact of uncertainties in the species’ life history parameters and to analyze possible outcomes on the species’ demographics of these uncertainties (Lacy, 2019).

There are basic criteria required to run a PVA (Morrison et al., 2016) and the quality of data used in the models can vary widely depending on the populations involved (Lacy, 2019). This will greatly influence the validity of the model (Lacy, 2019). PVA models have already been built for giant anteaters. Diniz and Brito (2013) based their work on data and a baseline model previously developed in a Ph.D thesis by Miranda (2004), who studied a wild population in Emas National Park, Central Brazil. However, the growth rate of the population modeled was negative even before considering threats. Another model was built to evaluate a re-introduced population in the Iberá Reserve, Argentina, in a diploma thesis by Zamboni (2016), and parameters where based on data collected from that population.

The aim of this study is to design a robust baseline model for giant anteaters where each parameter is clearly explained and tested. For this reason, a sensitivity analysis was also performed to evaluate the relative impact of each parameter on model results to guide future research efforts and better understand potential threats. This baseline model reflects the biological potential of giant anteaters in the absence of significant threat and can be adapted as a conservation tool for any giant anteater project and conservation action plans in the future. Finally, to illustrate the potential use of the baseline model we tested the impact of road kill on a population in the Cerrado of Mato Grosso do Sul State, Brazil.

The baseline population model was built using data based on life history literature, field work conducted by the Anteaters & Highways Project in the Brazilian Cerrado, researchers’ expertise (A.D. and A.B.), and information on the captive population provided by the giant anteater international studbook (Schappert, 2019). We also consulted existent PVAs and used them for comparative purposes. Vortex version 10.3.5.0 (Lacy and Pollak, 2017) was used to generate our PVA models (baseline and case study). This software is an individual-based simulation model that simulates stochastic demographic, genetic and environmental processes from birth to death, throughout each year and for each individual following the parameters and probabilities given by the user.

The Cerrado biome is known as the Brazilian savanna. It is the second largest biome in South America and has been considered as one of the world's hotspots of biodiversity (Mittermeier et al., 1999). This is one of the regions of Brazil that most suffered recent losses of natural vegetation, mainly due to human activities such as agriculture and infrastructure development (Beuchle et al., 2015). Our baseline model was designed to reflect giant anteater life history in this biome, as some key parameters of this baseline were based on field data from Cerrado study sites of Anteaters & Highways project. No harvest rates and no increase in mortality due to road kill, disease or fire were included, as the baseline model was designed to represent a healthy population in the modified landscape of the Cerrado under no significant threat.

Attempts to determine the relative contribution of different life history parameters to population growth rates are an important part of population biology of any species (Caswell and Hastings, 1980). Our baseline model represents the biological potential of giant anteaters based on the parameters previously described focusing in the Cerrado biome. Although, we understand that this model does not represent a realistic simulation of giant anteater populations over its distribution range, it does provide the basis upon which future models can be developed. Researchers that have specific data on threats, for example, or on immigration and dispersal of individuals between populations may use the baseline model to construct more complex models; this is an important feature since data scarcity is probably a major limitation in assessing population viability (Radchuk et al., 2016).

A systematic review of PVA studies found that details of the baseline scenario were often lacking (Pe’er et al., 2013). Morrison et al. (2016) re-ran 90 PVA baseline models from the literature, 40% of which failed to be reproducible and therefore not available for future extended use. This paper provides a very detailed baseline model to avoid these restrictions and ensure that this study is repeatable and can be reproduced by other giant anteater projects.

Sensitivity analysis is a tool used to evaluate the robustness of a model to variations in parameter values (Tuberville et al., 2012). Our empirical evaluations of different models in sensitivity testing demonstrate that the more robust the model is to variations in a particular parameter, the less sensitive the model results are to the input values of that parameter. Our results show that the percentage of females breeding and mortality rates, particularly those of younger age classes, should be research priorities to increase the robustness of the baseline model. Age of first and last reproduction of males should not be a research priority for PVA analysis as they have very little influence on the overall model results for this polygynous species. While mortality rates are probably specific to particular conditions in each study site, the percentage of females breeding may be more related to female reproductive health (Amendolagine et al., 2018), which may or may not be associated with study site. Our results suggest that mortality rates, percentage of female breeding, and age at first reproduction for females should be research priorities in future studies (Fig. 1 and Supplementary material 2). This knowledge – key life parameters – will only be accurately achieved with long-term studies.

Bearing in mind that our baseline is reflecting a giant anteater scenario in the Brazilian Cerrado biome we applied an estimated annual mortality rate of 8% up to age 11 years, adding an additional 5% each subsequent year. Since this is a sensitive parameter this value must receive attention when conducting giant anteater PVAs for specific populations and in other biomes. Mortality rates may be lower in protected areas or higher in areas or populations under high anthropogenic threat.

Baseline values should be carefully analyzed when preparing and running PVA simulations (Pe’er et al., 2013); parameters need to make biological sense. These decisions made by researchers are essential to avoid incomprehensive and unreliable application of parameters (Morrison et al., 2016). This would avoid errors, such as the one that occurred in past giant anteater PVAs (Diniz and Brito, 2013 based on Miranda, 2004) when they erroneously estimated the sex ratio at birth. This parameter (Fig. 1 and Supplementary material 2) can greatly influence model results due the polygynous mating system. Male-biased sex ratios used in previous models traced back to the Ph.D thesis of Miranda (2004), where he used a male-biased sex ratio in the model developed due to the capture of more males than females. There is evidence in the literature that males are captured and collared more often than females (Shaw et al., 1987; Medri and Mourão, 2005; Bertassoni, 2017). However, in the Anteaters & Highways Project we captured 1M:1 F and we believe there is no a priori evidence to suggest a skewed sex ratio at birth in the wild. One potential explanation for a male-biased sex ratio in captures is that researchers usually only collar adult individuals and usually refrain from capturing females with pups on their back to prevent injury of the pup. This could decrease the number of females captured at any given time. Another possibility is that males move more than females and therefore are more prone to be captured (Bertassoni, 2017). In captivity the sex ratio at birth is not male biased (Schappert, I., 2019). Finally, sex ratio at birth cannot be inferred by adult sex ratios. A bias in the sex ratio of the adult population are much more likely due to differential mortality rates rather than skewed sex ratio at birth and are likely to be threat based and population specific.

Giant anteaters are impacted by road presence throughout their distribution (Freitas et al., 2014; Ribeiro et al., 2017; Ascensão et al., 2017). Our case study of the impact of road kill on wild giant anteaters applied our baseline model with real data collected during three years on the impact of vehicle collisions. Model results based on these observed road kill rates suggest that mortality due to road kill may not necessarily lead to the extinction of local populations of giant anteaters but may reduce the stochastic growth rate by half, reducing the population’s resilience and ability to withstand or recover from the other anthropogenic threats. It could actually be more than half as the population is slower to grow out of smaller, more vulnerable population sizes. Our case study model does not consider potential additional impacts of future habitat loss, pesticides, dog attacks, disease outbreaks or fire, but it does suggest that road kill may make local extinctions more likely if other threats occur. This example also clearly demonstrates the importance of gathering detailed data on threats. The impact of road kill was very different depending upon the sex ratio of the animals killed. The loss of females due to vehicle collisions can be more impacting and can even lead to local extinction, whereas road kill of males has little effect at the rates tested. This is standard for a polygynous species, in which females are the limiting sex on reproduction and population growth. Therefore, information on sex ratio and age class of road-killed giant anteaters is key to interpreting the influence of this threat on the population.

PVA models are living models, in the sense that they must be continuously updated as better or new data are compiled or available. The model proposed here has limitations; for example, we did not incorporate density-dependency into our PVA due to lack of data, although such processes often have effects on population dynamics and extinction risk if relevant for the species (Pe’er et al., 2013; Morrison et al., 2016). In addition, the field data used in this model for key parameters are only based on a few years of observations, which require caution when interpreting the model (Lacy, 2019)

The incorporation of the sensitivity analyses using PVA techniques can guide conservation efforts for different species. Here, we provide a complete and detailed presentation of PVA input parameters for the giant anteater, which can be used for conservation planning for the species. As more data about the species become available from long-term projects this model should be updated. The model should also be adapted to local realities, where threats not considered here will need to be included. We hope that long-term projects studying giant anteaters can generate extensive population-level data to accurately estimate demographic parameters, following research recommendations outlined here, to create more robust PVA models in the future.

None.