A national park can retain its boundaries while the population inside it changes almost beyond recognition. In Arusha National Park, Tanzania, researchers returning to a giraffe census first conducted in 1979–1980 found a population approximately half its former size. The loss was especially stark in the south: the earlier work had identified 175 giraffes there; the new surveys identified 15.
Yet the new population contained more calves per adult female than the old one. DNA also linked Arusha’s giraffes to animals elsewhere in northern Tanzania and Kenya. Neither observation cancelled the decline. Together, they made the conservation problem more revealing: a population can still reproduce, and retain traces of its former connections, while becoming much smaller in an increasingly isolated landscape.
Derek Lee, George Lohay and colleagues’ 2023 study supplied a missing second measurement after a 42-year interval. Its central achievement was to establish the scale and unevenness of the loss. Its central unanswered question was whether that loss had stopped.
Returning to recognizable animals
The historical study offered an unusually useful starting point. Its researchers had identified individual giraffes rather than simply counting whatever animals appeared during a single outing. That made it possible for the new team to repeat the essential approach: build a catalogue of recognizable animals, then examine how that catalogue grew through repeated encounters.
The new work comprised 12 surveys between June 2021 and February 2022, grouped into three sampling periods spanning the region’s rainfall seasons. Teams followed a fixed road network and changed the order and direction in which they covered it. They photographed the right side of encountered giraffes, recorded sex and age class, and matched the coat patterns using identification software.
This arrangement addressed two different sources of uncertainty. Repeated visits offered another chance to encounter giraffes missed previously. A consistent route, with its order varied, reduced the risk that different effort or habitual survey timing would manufacture differences between visits. Covering different seasons also mattered because animals need not remain equally easy to encounter as vegetation and resource use change.
The fieldwork identified 184 giraffes. That was a count of distinct recognized animals, not yet an estimate of every giraffe in the population. Some animals could remain unseen even after repeated searches. The researchers therefore needed a way to estimate the unseen portion—and, for the historical comparison, a way that did not confuse a change in analytical method with a change in abundance.
Why the paper gives two current estimates
For the comparison across decades, the team applied the earlier study’s method. It fitted a curve to the accumulation of newly identified individuals, estimating the eventual total that further sampling would approach. This gave 239 giraffes, with a 95% confidence interval of 212–266. The equivalent historical estimate was 471. Comparing those two estimates produced the reported decline of approximately 49%.
The researchers also used a more modern capture–recapture analysis that explicitly accounted for the probability of detecting animals. This estimated 208 giraffes, with a 95% confidence interval of 177–254. They regarded this as the better estimate of the population during the new study.
The distinction is consequential. The figure of 239 is the appropriate partner for the paper’s like-for-like historical calculation. The figure of 208 is the authors’ preferred contemporary estimate. They answer related but different analytical needs; neither should be mistaken for the 184 giraffes actually photographed. Their overlapping uncertainty intervals also show that they are not incompatible accounts of radically different populations.
| Measure | Earlier study | New study |
|---|---|---|
| Individually identified giraffes | 462 | 184 |
| Population estimate using the earlier method | 471 | 239 |
| Preferred new estimate accounting for detection | Not reported with this method | 208 |
This attention to observation is particularly important when comparing wildlife numbers. In a separate Tarangire survey experiment, Lee and Monica Bond found that aerial estimates were significantly lower than photographic capture–recapture estimates in three of four study areas after a seasonal adjustment. The experiment showed how the survey process itself can create large differences in apparent abundance. It did not provide a correction for Arusha, where the researchers instead preserved the earlier identification approach to make their historical comparison more meaningful. Our account of that experiment follows the visibility problem in detail.
A decline with a geography
The loss was not evenly distributed within Arusha. The earlier researchers had distinguished northern and southern subpopulations. The new team used the same division when assigning giraffes according to where most of their observations occurred.
The north still contained most of the identified animals: 168 in the new study, compared with 287 historically. In the south, the contrast was 15 against 175. Those are identified individuals rather than independently estimated totals for each region, but the contrast is substantial. A single park-wide percentage would conceal how much more severely the southern component had diminished.
Movement between the two parts of the park had also changed. About 10% of the observed population crossed between them during the new study, compared with approximately 3% in the earlier work. Historically, the recorded movers were adult males. The recent observations included adult females and calves as well as adult males.
More movement within a park does not demonstrate restored movement beyond its boundaries. Nor does it reveal why animals changed their space use. The finding does show why a conservation assessment cannot assume that the internal distribution described four decades earlier still holds. The places animals occupy, and the kinds of animals moving between them, can change alongside their total number.
More calves did not mean the population had recovered
The age structure might initially appear encouraging. Adults made up 58% of the newly observed population, compared with 76% historically. There were 0.78 calves per adult female, against 0.33 in the earlier study. The adult sex ratio also shifted from approximately equal numbers of males and females to fewer males per female.
These are descriptions of population composition. A calf-to-female ratio is not the same as an annual birth rate, and it does not measure the fraction of calves that will survive to adulthood. It can change because there are more calves, fewer adult females, or a combination of changes. Likewise, a larger proportion of young animals can coexist with a smaller total population.
The distinction matters directly here. Interpreting the calf ratio alone as recovery would disregard the evidence that abundance had fallen substantially. The study establishes that the recent population had a different structure. Following identified animals through subsequent years would be needed to determine whether its births, survival and movements now support growth.
DNA remembered a wider landscape
The researchers added another kind of historical evidence by comparing mitochondrial DNA from 22 Arusha giraffes with sequences from other populations. The complete comparison included 218 sequences. Mitochondrial DNA is inherited through the maternal line, so shared sequence variants can reveal connections among maternal lineages.
Arusha contained three such variants, or haplotypes. Two were shared with giraffes in the Tarangire ecosystem. Another was shared with sampled giraffes in northern Tanzania and southeastern Kenya. The network in the paper therefore placed Arusha’s animals within a wider history of female-mediated connections across the landscape.
That result needs its own time qualification. Shared maternal variants are evidence of historical connectivity; they do not show that females are successfully crossing today’s roads, settlements or agricultural land. The study did not date each connection or track a present-day crossing between these populations. Mitochondrial sequences also cannot establish the full extent of present inbreeding or capture genetic exchange through males. The authors explicitly called for nuclear genetic markers to investigate the population’s current genetic condition.
The genetic evidence strengthens the case for considering Arusha in its landscape setting. It supplies no shortcut around measuring whether usable connections still exist.
The cause and timing remain open
Arusha is unfenced, and the authors describe increasing isolation associated with urban development and agricultural expansion outside it. They discuss several possible explanations for the uneven decline, including differences in habitat conditions, external land use and illegal hunting. They do not establish which mechanism caused Arusha’s loss, or quantify their separate contributions.
The southern population’s poorer historical calf survival also complicates a simple story in which one recent external pressure explains everything. Conditions may already have differed between the two parts of the park at the first census. The new study identifies where the loss was greatest; determining why requires evidence beyond the two population snapshots.
The 42-year gap leaves a second major uncertainty. A decline between two endpoints can arise from a steady reduction, an earlier collapse followed by stability, or several periods of loss and recovery. These surveys cannot distinguish those histories. Calling the result an ongoing annual decline would go beyond what was measured.
That makes regular follow-up unusually valuable. Further photographic surveys could establish the current direction of change and track the survival of known animals. Work on movement and nuclear DNA could then distinguish an accessible landscape from one whose former connections survive mainly in inherited variation. Investigation of the southern area could test the explanations for its much greater loss.
Arusha’s giraffes were still visible, reproducing and using different parts of the park when researchers returned. The census showed why those observations were insufficient reassurance. A population’s continued presence is easier to notice than the animals missing from it.
Sources
Lee, Derek E.; Lohay, George G.; Madeli, James; Cavener, Douglas R.; and Bond, Monica L. (2023). “Masai giraffe population change over 40 years in Arusha National Park.” African Journal of Ecology, 61, 345–353. Paper and DOI. Open full paper.
Lee, Derek E., and Bond, Monica L. (2016). “Precision, accuracy, and costs of survey methods for giraffe Giraffa camelopardalis.” Journal of Mammalogy, 97, 940–948. Paper and DOI.
