Photography used by UCD researchers for non-invasive study of deer

Image Credit: Adam F Smith

Jana Joha provides insight into a recent paper published by UCD researchers in The Journal of Zoology, that shows how photography can be a useful and effective tool to measure the shape and size of physical traits of animals.

Researchers from UCD, Dr Simone Ciuti and PhD student Adam Francis Smith, along with Paolo Bongi, have developed a photogrammetry technique to acquire precise and accurate measurements of the complex shape of deer antlers. This method allows for the measurement of the size of objects from photographs. They used photographs of deer trophies (skulls with attached deer antlers) and free-living, wild deer in Phoenix Park, Dublin, to calibrate their cameras and validate their method. Through the use of photogrammetry, the researchers have overcome the challenge of capturing and handling animals in order to take measurements of physical traits. 

The physical traits of animals are often directly related to their innate behaviour and they can provide information on development, function as well as conferring an evolutionary significance. These traits allow researchers and scientists to understand the biology and behaviour of animals, which in turn allows for more effective conservation and management. Obtaining measurements from animals often involves their capture, which requires time-consuming and expensive sedation as well as physical restraint. This is both a laborious and difficult process and requires highly skilled researchers and veterinarians. Due to this, the number of traits being measured and recorded in animals is often small, making analysis difficult as there isn’t an accurate representation of the wider population.

Additionally, acquiring measurements after animals have been captured alters their behaviour, and this is especially true for large herbivores such as deer. In acute instances, it can even lead to the death of the animal upon release. Developing a method that can remotely and accurately collect trait measurements, while also avoiding the alteration of the animal’s behaviour would be incredibly advantageous for future research. 

Two camera systems were used to take the photographs for this research; a Nikon D3000 DSLR camera with a 200mm zoom lens and a ‘digiscope’ used to record distant images. The camera-lens was used to measure distances up to 100m while the digiscope was used to measure up to 200m. These cameras were first calibrated by taking photographs of a 25cm ruler placed on a flat vertical surface at different distances. Researcher Adam F. Smith explains that “the calibration is done by photographing a ruler every 5 metres up to 200m. In every picture, we can link the size of the ruler in centimetres to its size in pixels. Then, when we take a picture of any object and know the distance, we can calculate how many pixels is equal to 1cm”. This helped them to develop calibration models, which could calculate the sizes of deer antlers at long distances.

In order to validate the developed calibration models, the researchers took a random sample of photographs of the ruler and calculated the average size of the ruler in pixels. The calibration models were then used to predict the size of the ruler in centimetres based on the distance between the cameras and ruler. 

The researchers began testing their method on deer trophies, which were positioned on flat surfaces to imitate a free-living animal. The deer trophies were photographed at random distances on both cameras. Antler features were measured directly by hand, using a tape measure, and those measurements were used to estimate percentage errors. Smith notes that “the biggest challenge was finding enough trophy skulls from fallow and roe deer in Ireland. In UCD, we had a number of fallow deer skulls from The Phoenix Park. However, more samples would ensure more accuracy in our research. To solve this, I flew to Pisa, Italy - where our colleagues generously offered dozens of antlers from trophy animals to measure”.

The model was found to require correction to decrease the bias caused by the curved, tilted shape of deer antlers and by implementing a correction protocol the researchers were able to get a more accurate measurement of antler features. “Since an antler is not a "straight" object, some bias is introduced when it is photographed. From knowing the actual size of the antlers (from trophy skulls) which we measure by hand, and the estimate based on photogrammetry from our photographs, we can quantify this bias". He explained; “Effectively, the correction takes into account how much antlers curve based on their shape and size, and adds this value to our estimate, giving a corrected measurement of the antler.”

Further photogrammetric validation was carried out by photographing free-living, wild fallow deer in Phoenix Park, Dublin. The photogrammetric techniques and correction protocol developed by the team were used and the results showed that these methods were able to accurately measure antler features with minimal error. “Because we set up the trophy skulls and their antlers in the same way as a deer would be standing, we were confident that the method works regardless of the subject”. Smith further states that, “we were satisfied with the results because we saw how similar the photos and measurement methods were, but also because in less than one week we were able to measure the antlers of most of the male herd”.

Dr Simone Ciuti, Adam Francis Smith and Paolo Bongi used this new technique to show how photographs and photogrammetric techniques can be used to take relatively accurate and precise measurements of physical features of deer. Due to the success of this research, they believe that this method can be employed on other animals, especially ones that are difficult to capture and handle, such as crocodiles. Smith says that the first application of their technique will be “putting it to use in the Phoenix Park, where we can quickly measure the antlers of 100s of fallow deer bucks from a distance”, which “will be linked to individual rutting and mating behaviours, as well as health and body condition.” He believes that “it's also possible that researchers or wildlife managers can build upon these methods to answer their own questions, without having the stress and expense of catching animals.”.

The next step for the researchers is to improve accuracy and reduce human error as well as develop a way to automatically analyse photographs. Smith believes “reducing human bias from measurements is a good next step to improving the accuracy of the measurements”. Dr Ciuti also notes: “we are planning to deploy new software that would allow us to automatically measure the traits from pictures. This will reduce bias from human error and will speed up the process, allowing us to monitor significantly higher sample sizes from natural populations”.