An animal can be counted twice without ever appearing twice in the same photograph. If a giraffe seen years earlier is mistaken for a new individual, one living animal acquires two identities in the records.
That problem grew within a long-running photographic monitoring programme in Niger. A study by Mara Vukelić and colleagues, first published online in 2025, re-examined fifteen years of images and removed 145 duplicate identities. It also replaced a counting rule that could keep missing animals in the population indefinitely.
The correction reduced the estimated population relative to the older method. Yet the revised analysis still found strong growth, estimating 562 giraffes in 2018. An exaggerated count and a real conservation recovery had existed together.
A valuable archive that became harder to manage
The study drew on annual ground surveys conducted from 2005 to 2019. Teams travelled through the giraffes' range in vehicles, working with local knowledge to find groups and photograph individuals.
During the study period, these West African giraffes lived in an agricultural landscape shared with people and livestock. Finding them required coordination across that inhabited area. Survey routes and effort were not standardised in exactly the same way every year.
The resulting archive contained 7,237 photographs. Images had been organised into annual records, with people comparing coat patterns to identify animals. Across the full period, the original catalogue listed 979 identities.
As a population and its archive expand, the comparison task becomes more demanding. A photograph of a giraffe needs to be checked against a larger history of previous sightings. Differences in angle, lighting and the part of the body visible can make the same animal look unfamiliar.
The authors rechecked the archive with HotSpotter, software that suggests possible matches between natural patterns. A trained person then assessed those candidate matches. The programme assisted identification; it did not turn every photograph into an unquestionable automatic identity.
After correction, the catalogue contained 834 individuals: 434 females and 400 males encountered at some point during the fifteen-year record.
The extra identities were not extra animals
The team identified 145 duplicate identities, described in the paper as “ghost” individuals. These were additional records belonging to animals already represented in the catalogue.
Most of those errors appeared in the later years. Only 11 arose during the first six years, compared with 134 after 2010. That timing matters because the monitoring method became less reliable as the archive grew.
It would be wrong to subtract 145 directly from a single annual population count. The errors accumulated across a multi-year catalogue. Some affected different years, and the catalogue itself included animals that were not all alive simultaneously.
The distinction between a historical identity list and a current population is central to the paper. A fifteen-year record can contain newborns entering the population, adults dying and animals disappearing from the surveyed area. Its total measures how many identities appeared during that history, not how many giraffes stood on the landscape at its end.
Cleaning identities repaired one part of the problem. The next part concerned what to do with a giraffe that had a valid identity but was no longer seen.
A missing giraffe is an uncertain giraffe
The earlier counting procedure included animals observed during a survey and animals known from previous years whose deaths had not been confirmed.
That rule can make sense as an attempt to avoid declaring a missed animal dead. Its weakness is that carcasses are not always found. An animal can remain on the list year after year even after it has died or permanently left the monitored population.
The new analysis used capture–recapture methods. Here, “capture” meant an identifiable photographic encounter. Researchers used the pattern of sightings and absences across years to estimate both abundance and the probability of encountering animals.
Consider a giraffe photographed in three years, absent in the fourth and photographed again in the fifth. Its return establishes that it was missed in the fourth year. Many such encounter histories help quantify how often absence from a survey occurs without disappearance from the living population.
A giraffe never seen again is more ambiguous. These models estimate apparent survival, which cannot necessarily separate death from permanent emigration. That limit is particularly relevant in an open landscape.
The models also accommodated differences between years and sexes. Detection was high early in the study but became lower and more variable later. Males generally had lower encounter probabilities than females. Treating every annual count as equally complete would discard those differences.
The smaller estimate still showed recovery
The revised population estimate for 2018 was 562 giraffes, with a 95% confidence interval of about 524–600. The sex-specific estimates were 329 females and 233 males.
The analysis estimated an average population growth rate of approximately 12% across the study. That is an average derived from the population trajectory, not a claim that exactly 12% more giraffes appeared in every successive year.
The older and revised abundance estimates were relatively close during the early period. Between 2005 and 2010, their average absolute difference was seven giraffes. Between 2011 and 2018, the average gap widened to 82 animals.
Those differences combined the effects of duplicate identities and the earlier treatment of unobserved animals. The comparison changed both the cleaned records and the estimation method, so it does not isolate a single software improvement as the cause of the revised numbers.
Although the encounter data continued through 2019, the paper did not present a population estimate for that final year. In the time-varying model, some parameters at the end of the series could not be separated reliably. Stopping the abundance graph in 2018 reflected that statistical limitation.
None of these estimates should be presented as Niger's current population. They describe the years analysed in the paper.
What the survival results can establish
The study also fitted models that distinguished age classes. These generally indicated high apparent survival, especially among females, consistent with the observed population increase.
But an annual photograph archive does not observe every calf from the moment of birth. A calf that dies before its first photographic encounter may never enter a survival history. The youngest modelled category also combined animals under 18 months rather than representing a clean cohort followed continuously from birth to its first birthday.
A high survival estimate for that category should therefore not be reported as the proportion of all newborns that survived their first year.
The models quantify demographic patterns within the observed records. They did not experimentally test which local conservation action produced the recovery, or establish that sharing land with people was free of costs and conflict.
What the study does show is narrower and valuable: this monitored population grew substantially in a landscape used by people during the period studied. Recovery outside a conventional protected-area setting was observable, while the quality of its measurement still required improvement.
Keeping the field record useful
A 2019 study in Uganda likewise used repeated photographic identities and detection estimates to assess giraffe abundance. The settings and historical counting methods differed, but both studies demonstrated why photographs become more informative when organised into reliable individual histories.
In Niger, correcting the archive did not render the years of fieldwork wasted. Those photographs made the correction possible. Researchers could revisit identities, recover past encounters and estimate a trajectory more carefully than the original annual lists allowed.
The useful conservation result is therefore both biological and methodological. The population had grown, and the evidence for its size became more defensible when the record was challenged.
Maintaining that value requires consistent photographs, careful matching and explicit treatment of animals that go unseen. A monitoring programme must evolve as the population it follows becomes larger.
Sources
Mara Vukelić, Ella W. White, Julian Fennessy, Abdoul Razack Moussa Zabeirou, Kateřina Gašparová, Simon Valle, Michael B. Brown, Thomas Rabeil and Karolína Brandlová (2026; first published online 2025). Monitoring African Megafauna in an Anthropogenic Landscape: A 15-Year Case Study of the Vulnerable West African Giraffe. Animal Conservation, 29, 235–248.
Michael B. Brown, Douglas T. Bolger and Julian Fennessy (2019). All the eggs in one basket: A countrywide assessment of current and historical giraffe population distribution in Uganda. Global Ecology and Conservation, 19, e00612.
