How many rhinos are there?

(This article was originally published in The Horn, spring 2014. Author: Richard Emslie, Scientific Officer of the IUCN SSC African Rhino Specialist Group)

Just as an investor needs a full set of company accounts to make informed investment decisions, rhino managers need estimates of rhino numbers (and age/sex structures, mortality data, inter-calving intervals etc.), to assess population size and performance and decide how best to manage the population to ensure good breeding performance is maintained.

People often think counting rhinos is easy; you just fly a plane around a reserve and count the rhino you see. In reality, counting rhinos is tricky, as rhinos do not only occur in open flat areas without trees. Often a proportion of rhinos will remain undetected; either hidden by vegetation or terrain. Some animals may be missed if they are directly under the plane or if observers fail to see them.

Despite white rhinos being big and preferring open areas, if one flies straight, aerial transects experience shows that in hillier bushveld areas you can miss a significant proportion of rhino (as much as 40% in Hluhluwe-iMfolozi Park). If uncorrected, such counts simply provide a biased minimum number that underestimates true population size. This is why other methods are usually used to estimate rhino numbers.

Some counting methods use a distance-sampling approach where you not only count animals along a transect; but if on the ground you measure the perpendicular distance of the animals to the line (see figure 1). If in the air, you record rhino sightings according to which band they are seen in as you go further from the plane, to the outer edge of the transect area. In general the assumption is you count most (or all) of the rhinos near the line but the further you go, the fewer will be counted. In effect, distance sampling makes allowance for animals being missed further out along the transect and uses all sightings to create an estimate of the true number of animals, with confidence levels.

AfRSG image 1

Black rhinos are even worse to count from the air as they often move after being flown over. They also prefer areas with thicker bush. With transect counts, you could fly once over an area and record a minimum number that is completely different to the number actually on the ground.

To be useful, managers need accurate estimates of the true numbers of animals. Ideally one uses counting methods, which if repeated give pretty similar results (precise) and estimate true numbers (unbiased). Count accuracy is a function of two things – bias and precision. In some years, visibility conditions or observers may be better, meaning it is easier to see rhino, which can further complicate interpretation of population estimates.

Generally due to cost, one cannot afford to repeat counts; but if a counting method is repeated and produces very variable results (i.e. is imprecise) then it is pretty useless, as managers can’t be sure if differences between counts over time reflect reality or could be a sampling error.

For this reason, if black rhinos are counted from the air (which they have to be in vast areas such as Kruger National Park) then they are counted using helicopters that intensively fly in a criss-cross pattern around blocks (demarcated by GPS). If observers don’t see a rhino on the first fly past, they have a good chance of seeing it on the second or third fly past as it may be moving and easier to spot.

It is prohibitively expensive to count 100% of a huge park using helicopter block counts, so scientists usually count a random but still significant selection of blocks (or undertake a number of distance sampling line transects on the ground) and use statistical analytical methods to covert the raw data into an estimate of the number of rhinos, plus a measure of the confidence in the estimate. This is usually shown by scientists as a ‘confidence level’.

It is not possible to reliably detect small changes in rhino numbers year-on-year using such methods, and this coupled with the high survey costs is why counts are usually undertaken every two to three years in the big reserves. There is generally a trade-off between precision (tighter confidence levels) and the amount of sampling (cost). Although intensive block counts and distance-sampling surveys both seek to estimate true numbers of animals; block counts are better as they are usually much more precise.

However, for most smaller reserves, individual identification methods provide the most reliable, accurate and precise rhino population estimates. In many reserves, management may know exactly how many rhinos they have. ID-based monitoring methods start to become impractical in either vast areas or those with more than approximately 400 rhinos.

When observing a rhino, you might see one or more notches cut in the outside of the ear. This animal is individually recognisable. Each rhino (or a sample of the rhino in the reserve) that are ear-notched will have different ear notch patterns, allowing field rangers on patrol with binoculars to easily identify them as specific individuals. Figure 2 illustrates the Zimbabwe ear-notch system. Modern digital cameras can also help distinguish between rhinos that have not been notched (e.g. based on horn configuration).

If rhinos are seen regularly, you will know how many there are in a reserve without the need for statistics. You also get an early warning from your monitoring if one is missing. However, even if rhinos are seen less frequently and/or not all have ear-notches, then Rhino Bayesian Mark Recapture software can be used to analyse sightings and re-sightings of rhinos to produce population estimates with confidence levels.

Mark-recapture statistics essentially use the sighting, re-sighting data to calculate the probabilities of there being different numbers of rhinos that you haven’t seen yet. It adds the best estimate of the number not seen, to the known minimum number you have seen, to get a total population estimate. The method also allows you to derive confidence levels around your population estimate.

AfRSG image2

Most rhino population estimates today are based on ID methods and provided there is ample ground coverage these can be pretty accurate. However, a significant proportion of rhinos live in large populations in vast areas, and here at best, estimates have a certain degree of uncertainty.

Thus, when you see estimates for total numbers of rhinos in Africa this represents a best estimate of the number. However, be aware that there is a degree of uncertainty around this number, given the confidence levels around many individual population estimates (especially of the largest populations). In general, rhino population estimates are more accurate than many other species; and the actual rhino numbers are probably within +/-5-10% of the total estimate. Additional information such as inter-calving intervals, mortalities, and proportions of adult females with calves also help assist with interpretation of population trends and performance.

Just imagine how hard it is to count the world’s smaller species!