We could be witnessing the sixth mass extinction at an alarming rate worldwide. It’s marked by the rapid loss of species due to human activities like habitat destruction, pollution and climate change. Unlike previous mass extinctions, which were caused by natural events, this one is driven by human impact – like growing populations, pollution, invasive plant species and human-wildlife conflict.
Large mammals are especially at risk, in Africa as elsewhere. For instance, nearly 60% of wild herbivores – such as elephants and hippos – are already threatened with extinction.
Effective conservation and recovery strategies are needed. To develop them, you need to know how the population of a certain animal is doing and, if it is in decline, what’s causing it.
One tool that’s useful here is a model, using biology, maths, statistics and computer software.
The problem is that there aren’t enough of these realistic, effective models for large mammals. There’s a shortage of appropriate data and the models are complex to build.
I was part of a team that developed a model to help fill that void. It’s the first to account for how large mammal populations interact with each other and their environment while also incorporating their detailed biology. It draws on valuable existing data and can be adapted for various wildlife species.
We tested the model on populations of east Africa’s topi (a large antelope). From the results we’re able to deduce that the drivers of the topi’s massive population decline were habitat loss, poaching and killing by predators.
Knowing what’s driving population declines is extremely valuable. Large mammals play a critical role in ecosystems. Changes to their populations will also affect many other species and could cause the extinction of connected species.
How the model works
Our model combines different types of data, like total population size from aerial surveys and ground vehicle counts, with predicted data on population figures. This allows us to estimate and track population trends that can’t be captured by just one data type. It considers factors like animal age, sex, gestation length, weaning period, calves per birth per year, birth rates, survival, and environmental influences like rainfall and temperature.
Essentially, the model starts with educated guesses, then updates these guesses as it processes more observed data.
The model can tell what causes a decline in two ways.
First, it finds out which factors (such as rainfall) have a strong negative impact on things like birth rates, survival or recruitment, and shows exactly how they affect each other.
Second, it lets us use simulations to see how changing one of these factors, while keeping others unchanged, changes the population by influencing its key characteristics (such as birth rate).
Testing the model on topi
We tested our model on the topi population found in Kenya, Tanzania and other African countries. We chose the topi because it’s a large herbivore in decline.
The topi is an elegant antelope weighing between 91kg and 147kg, with a long face and uniquely twisted horns. One of the largest remaining topi populations in east Africa occurs in the Greater Mara-Serengeti Ecosystem, which straddles the border between Kenya and Tanzania.
Kenya’s Directorate of Resource Surveys and Remote Sensing has, since 1977, monitored numbers and distribution of topi, and other large wild herbivores and livestock, using aerial surveys in the country’s rangelands, covering 88% of Kenya.
Based on this data, we can see that topi numbers have declined persistently and strikingly (by 84.5%) in Kenya’s Masai Mara ecosystem between 1977 and 2022, even those in protected conservation areas.
This decline indicates a high risk of extinction if the trend persists. This is a serious concern, since other antelope species, such as the roan, have gone extinct in the Mara in recent decades.
But the causes haven’t been fully established.
We ran the aerial and ground survey data into the model in a computer on a monthly interval. This approach allows the model to capture patterns in trends and dynamics on a monthly scale. It allows us to see the distribution of births per month, the timing of births, the degree to which multiple females in a population give birth around the same time, the proportion of females in a population that give birth, the total number of individuals of each age and sex in each month, and the proportion of young that survive to adulthood.
The model starts with initial guesses based on existing knowledge, and refines the guesses as it processes more actual data.
It produces results that match the observed patterns of population decline, seasonality of births and how many animals survive to become juveniles or to adulthood.
Based on these findings, we see that the decline in the topi population is driven by a combination of low adult female numbers, low newborn survival and low recruitment into the adult class because most young (over 95%) die before they become adults.
Based on the model, we attribute these changes to impacts from environmental changes, human activities and predation. For instance, since adult animals are the least sensitive to climatic changes, this suggests other factors – such as habitat loss or deterioration, poaching or high predation rates – are likely contributing to the decline.
The new model enhances our understanding of large herbivore population dynamics besides confirming existing knowledge.
By combining different kinds of data from different sources, the model helps estimate and track important population details that one type of data alone can’t show. For example, for the first time data is captured that can track the total number of topi of each age and sex in each month, how many adult female topi are ready to conceive and the various stages of pregnancy. This method also estimates changes in the total topi population by age and sex in all four zones of the Mara, even in zones without direct ground age and sex data.
Refining and enhancing the model
The team is now extending the model to include more features (like the influence of livestock numbers), make it user friendly, apply it to more wildlife species and assess the effectiveness of ongoing and planned management actions.
Improving our understanding of the drivers of large mammal losses will ensure that the right conservation actions are taken. It’ll also ensure resources aren’t wasted because solutions could include investing in major infrastructure, changing wildlife conservation and livestock production policies, changing law enforcement and rehabilitation of wildlife habitats – all of which are costly.
