A little more rain, a blocked connection between habitats, fewer patrols, expanding towns: each can affect giraffes, but they do not necessarily affect population size in the same way or on the same timetable.
A 2023 study by Monica L. Bond, Derek E. Lee and Maria Paniw brought those pressures into a single population model for Tanzania's Tarangire ecosystem. It followed simulated female giraffes through births, survival and movements over 50 years. The largest declines occurred when the model reduced survival to represent substantially weaker wildlife protection.
The study also asked a more hopeful question. Could an achievable improvement in survival compensate for other pressures? In the model, adding a small amount to survival probabilities was enough to offset the effects assigned to more frequent unusually wet seasons.
Those are conditional findings. The researchers did not watch a conservation intervention unfold for half a century, and the model did not establish how much money or how many patrols would deliver a given survival gain. Its value was to make the demographic consequences of competing assumptions explicit.
Building a population one animal at a time
The model drew on a long-running photographic study in an unfenced landscape containing a national park, a wildlife ranch, community conservation areas and village lands. That research programme had identified more than 3,100 giraffes of both sexes and different ages over eight years.
The simulated population was narrower: 1,000 females distributed among nine social communities. Each modelled giraffe had an age, community membership and, when old enough, a reproductive schedule. Every four-month step represented one of the ecosystem's three precipitation seasons.
A female could survive into the next step, produce a calf or, at the model's dispersal age, change social community. Random draws meant that identical starting conditions did not always produce identical outcomes. A population is made of finite individuals, so chance survival and reproduction matter even when average probabilities remain the same.
The model represented differences among social communities but did not simulate each animal's exact route through a mapped landscape. Community membership carried information about the environmental conditions and survival rates associated with that group.
Reproduction was deliberately simplified. Females began breeding at a fixed age and then followed a common interval between births. Male calves were excluded from the continuing female population. The model also reduced survival when a community exceeded an imposed size threshold, preventing unlimited growth.
These choices made a complex system manageable. They also define what the model can answer: changes in female abundance under its demographic rules, rather than every consequence of altered food, mating, maternal care or individual movement.
Checking the starting point
Before changing the future, the researchers checked whether their core model resembled patterns already observed.
Over a five-year comparison, simulated community abundances broadly overlapped estimates from the photographic records. An additional comparison with a longer-term aerial-survey trend gave a similar overall growth rate. Those checks supported using the model to explore scenarios, although reproducing past patterns does not verify every mechanism or guarantee a 50-year forecast.
The authors also changed survival, reproduction and dispersal parameters separately to see which most affected the outcome. Adult female survival had the strongest influence. That is biologically understandable: an adult that survives can contribute calves repeatedly, while an adult that dies removes those future reproductive opportunities as well.
An earlier demographic analysis, from the same research programme, had found large adult-survival differences across the wider management landscape. That study supplied part of the empirical basis for the later protection scenarios. It was observational evidence, however, with several environmental differences occurring together.
What “more rainfall” meant inside the model
The rainfall scenarios were more specific than a general prediction of a wetter climate. The researchers increased the frequency of above-average seasonal rainfall anomalies by 10% or 25%, sampling from rainfall conditions already represented in their recent historical data.
They did not increase every season's rain by those percentages. Nor did they simulate unlimited new rainfall extremes, changing vegetation communities or all the effects of warming.
Previously estimated associations between rainfall and survival were used to adjust the animals' chances of surviving the current season. The same rainfall effect was applied across social communities, even though the underlying field relationships could vary with location.
This is an important boundary around the result. A model can show what happens if a particular survival–rainfall relationship continues. It cannot by itself establish that future climate will preserve that relationship, or capture long-term changes in browse availability that it does not simulate.
Comparing the possible futures
The team tested 32 scenarios, with 500 runs of each. Some changed one factor; others combined pressures or added a proposed mitigation.
Under the control scenario, the mean final female abundance was 975 after 50 years, compared with the starting 1,000. Increasing the frequency of wetter-than-average seasons by 25% reduced the mean final abundance to 943.
When the model combined that rainfall scenario with improved survival in both principal protected areas, the mean final abundance rose to 1,162. These numbers come from the paper's scenario table. They are simulated females under stated assumptions, not counts of real giraffes predicted to be present on a particular future date.
The survival improvement was an addition of 0.01 to the predicted survival probabilities used by the model. Because its time step was four months, readers should not translate that directly into a one-percentage-point annual survival change. Repeated across many seasons, even a small change in an individual's probability of surviving can compound into a substantial population difference.
The weaker-protection scenarios imposed much larger survival reductions based on contrasts with less-protected areas. Their trajectories fell far more sharply than the rainfall or movement-barrier scenarios. Figure 5 shows why the authors treated maintained wildlife protection as central: populations subjected to the strongest survival reductions approached disappearance, while those assigned higher survival increased.
A modelled intervention is not an intervention trial
The terms “improved law enforcement” and “reduced law enforcement” are labels for scenarios in which survival probabilities changed. The model does not contain a tested conversion from patrol hours, funding or staffing to giraffe survival.
The observational evidence behind those labels makes them plausible scenarios. It does not prove that enforcement alone caused all the differences in survival used to construct them. Settlements, land use and other pressures also differed among the studied places.
Similarly, a scenario that blocked dispersal between the park and ranch represented a particular movement loss. It did not capture all possible infrastructure impacts, such as mortality while crossing a road or altered access to forage within a community's range.
The ranking of pressures therefore applies to the scenarios as constructed. It should not become a universal claim that climate or habitat connections matter little to giraffes. A different landscape, a different degree of habitat loss or a different climate mechanism could change that ranking.
Keeping the whole ecosystem in view
The model could also raise giraffe numbers by removing the predation pressure associated with lions. The authors did not treat predator removal as a desirable conservation strategy.
That distinction exposes a broader limitation of any single-species abundance target. An intervention can increase the modelled number of one animal while damaging other species and ecological relationships. Conservation decisions need to consider those wider outcomes, as well as whether an action is practical and acceptable to the people sharing the landscape.
The useful management inference is narrower: preventing avoidable losses of breeding females may give a long-lived population more capacity to persist through other pressures. The model provides a way to test that reasoning and identify which real-world measurements would improve it.
More direct estimates of how conservation actions affect survival, better information on movement among communities, and longer observations of rainfall and vegetation would make future scenarios more informative.
For now, the paper shows how small changes in individual survival can accumulate across decades. It turns an abstract argument about resilience into a set of explicit, testable demographic assumptions.
Sources
Monica L. Bond, Derek E. Lee and Maria Paniw (2023). “Extinction risks and mitigation for a megaherbivore, the giraffe, in a human-influenced landscape under climate change.” Global Change Biology 29:6693–6712. Full modelling study.
Derek E. Lee, Monica L. Bond, Bernard M. Kissui, Yustina A. Kiwango and Douglas T. Bolger (2016). “Spatial variation in giraffe demography: a test of 2 paradigms.” Journal of Mammalogy 97:1015–1025. Related survival evidence.
