# Use of track counts and camera traps to estimate the abundance of roe deer in North-Eastern Italy: are they effective methods?

> Counting roe deer in Italian forests usually means dozens of hunters driving the woods. In six game reserves, two trackers counting fresh tracks on transects with CyberTracker produced estimates that tracked the hunters’ counts closely, at about 28% less cost than camera traps. The authors also show how to tell roe deer footprints from wild boar and red deer.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2018-romani-giannone-use-track-counts-camera/

## Citation and access

- **Authors:** Toni Romani, Carmelinda Giannone, Emiliano Mori and Stefano Filacorda
- **Year published:** 2018
- **Journal or source:** Mammal Research
- **Full citation:** Romani, T., Giannone, C., Mori, E. and Filacorda, S. 2018. Use of track counts and camera traps to estimate the abundance of roe deer in North-Eastern Italy: are they effective methods? Mammal Research 63(4): 477–484.
- **DOI:** 10.1007/s13364-018-0386-9
- **Access status:** Publisher link only
- **Where to find it:** https://doi.org/10.1007/s13364-018-0386-9

## Tags

- **Species:** Mammals > Red deer; Mammals > Roe deer; Mammals > Wild boar
- **Region:** Europe > Italy
- **Tracking:** Spoor age estimation; Track counts; Track measurements; Tracking; Tracks; Tracks and signs identification
- **Behaviour and ecology:** Behaviour; Daily travel distance
- **Methods and technology:** Body mass–day range scaling; Camera traps; CyberTracker; CyberTracker software; Formozov–Malyshev–Pereleshin (FMP) formula; Pearson correlation; Population estimation; Smartphones; Survey costs
- **Data type:** Existing data; New field data
- **Observer qualification:** Reported but not validated
- **Discipline:** Wildlife tracking
- **Year published:** 2018

## Detailed keywords

- **Species:** Studied: European roe deer (Capreolus capreolus capreolus). Other ungulates in the study area whose footprints had to be told apart from roe deer: wild boar (Sus scrofa) and red deer (Cervus elaphus).
- **Location and habitat:** Six forest game reserves in Friuli Venezia-Giulia region, North-Eastern Italy. Site 1, Moggio Udinese (Udine province, 800 to 1200 m): mixed conifer and deciduous forest, high grasslands used for herding. Site 2, Vito d’Asio (Pordenone province, 700 to 1050 m): mainly deciduous forest and forest glades. Site 3, Trasaghis (Udine province, 200 to 600 m): riparian forest near the Tagliamento river, with corn, wheat and soybean fields. Sites 4 and 5, Drenchia and Stregna (Udine province, 400 to 900 m): beech, hazel and ash forest with cultivated valleys and pastures. Site 6, Doberdò del Lago (Gorizia province, 30 to 160 m): limestone karst plateau with scrubland and oak woods in a human-dominated landscape. The authors describe the soils as mud and clay.
- **Topics:** population density estimation; comparing count methods; drive counts and vantage point counts by hunters; camera trapping without individual recognition (Rowcliffe et al. 2008 method, often called the random encounter model); track counts on transects; telling roe deer footprints apart from wild boar and red deer; counting only recent tracks; daily travel distance and animal speed; cost-effectiveness of surveys; wildlife management and selective culling.
- **Statistics or models:** Pearson correlation (Student’s t distribution, n minus 2 degrees of freedom) in R 3.3.2; Rowcliffe et al. (2008) camera-trap density formula; Formozov–Malyshev–Pereleshin (FMP) track-count formula; body mass–day range scaling rules (Carbone et al. 2005; Keeping 2014); cost analysis. Inconsistencies in the paper: (1) the text gives p = 0.004 (camera traps) and p = 0.003 (track counts), but Figure 3 gives p = 0.04 and p = 0.03; with r = 0.85 and six sites, the figure’s values are the more plausible; (2) the text names “Boskon-guard” camera traps, while Table 4 lists the Scout Guard SG-560V, and 24 cameras at 150 euros each would be 3600 euros, not the 3360 euros shown; (3) Table 1 gives Site 2 as 4.84 ha and Site 3 as 5.65 ha, far smaller than the other sites (perhaps a decimal or units error); (4) the text refers to speeds “in Table 1”, but Table 1 holds only habitat data; (5) the daily travel distance used in the FMP formula (2.31 km) matches speed 2 only (0.210 km/h over 11 hours), although the text says the mean of both speeds was used; (6) Figure 3’s caption and axes say “drive counts”, while the text compares against counts by hunters (drive counts plus vantage point counts). Transect lengths and the camera and track-count density figures for each site are not given in the text; they appear only as points in Figure 3.
- **Data type:** New field data: camera trapping (24 cameras, 1 December 2014 to 31 March 2015) and track counts (four transects per reserve), December 2014 to March 2015. Existing data: counts by hunters shared by Friuli Venezia-Giulia Region, for 2012 to 2013, 2013 to 2014 and 2014 to 2015 (drive counts in March and April).

## Synopsis

### 1. What they did

Wildlife managers in Italy usually count roe deer with drive counts and vantage point counts carried out by hunters. These are thought to work best in woodland, but they need many people: in this study, from 14 to 94 volunteers per site. The authors asked whether two methods needing far fewer people, camera trapping and track counts, would give similar density estimates.

They worked in six forest game reserves in Friuli Venezia-Giulia, North-Eastern Italy, from December 2014 to March 2015:

- Counts by hunters. Drive counts and vantage point counts on randomly chosen areas of 25 to 40 ha, at least twice, in March and April, when roe deer are solitary and territorial and so spread more evenly. The region shared these results, along with counts from the two previous years.

- Camera traps. 24 cameras placed at random along and near the transects, checked every 15 days, with no bait. Because roe deer cannot be told apart as individuals, density was calculated with the Rowcliffe et al. (2008) method, which uses how often animals pass a camera, how fast they move, and the camera’s detection zone. Speed came from published studies and from body mass scaling rules, using an average winter weight of 19.75 kg for roe deer in the study sites and 11 hours of activity a day.

- Track counts. Four transects per reserve, placed at random and away from animal paths and trails (which would inflate the counts), and not running parallel to fences, roads, rivers or corridors. Experienced trackers counted only recent roe deer tracks and recorded them with CyberTracker software on a smartphone. Density was calculated with the Formozov–Malyshev–Pereleshin (FMP) formula, which uses the number of tracks crossing the transect, the transect length and the daily distance roe deer travel (2.31 km).

They compared the methods with Pearson correlations and worked out the cost of camera trapping and of track counts, including equipment, human time at 10 euros an hour, and travel.

### 2. What they found

Telling roe deer footprints apart

- Roe deer footprints could be told apart from those of wild boar and red deer by size and shape. No young roe deer were present between December and March, and all footprints found were from adults (3 to 5 cm long).

- Footprints of intermediate size were separated from wild boar (even young boar) by the tips of the hooves, which are more pointed in roe deer, and by the dewclaws, which sit wide to the sides in wild boar and in line with the hooves in deer.

Density estimates

- Counts by hunters gave densities of 3.58 to 20.44 roe deer per 100 ha in 2014 to 2015, the lowest at Site 1 (Moggio Udinese) and the highest at Site 6 (Doberdò del Lago). Densities at each site were similar in the two previous years.

- Camera-trap estimates were strongly correlated with counts by hunters (r = 0.94), and so were track-count estimates (r = 0.85). Camera-trap and track-count estimates were also strongly correlated with each other (r = 0.94).

- Track counts sometimes gave lower densities than counts by hunters (Figure 3).

Costs

- Both camera trapping and track counts needed only two field workers.

- Camera trapping cost 12,880 euros: cameras, 808 hours of work and 4800 km of travel.

- Track counts cost 9334 euros: CyberTracker programming, one smartphone, four forensic rulers, 768 hours of work and 4800 km of travel. That is about 28% cheaper than camera trapping.

### 3. What they concluded

- In these forests, camera trapping and track counts both appear to be as effective as counts by hunters for estimating roe deer density, and track counts were the cheapest.

- Track counts may give lower estimates than drive counts, and this difference should be considered when choosing a method.

- Track counts seem reliable on sandy, muddy or damp ground, like that of their study sites, but cannot be used where soil is too hard, dry or rocky. Camera traps need suitable places to set them, which can be hard in open areas or near settlements. Some authors consider the Rowcliffe camera method unreliable.

- How long tracks last depends on habitat and weather, and how much deer move affects how many tracks they leave; both can vary between sites.

- The sample was small (six sites), and data from different sites were pooled, so local differences in habitat, weather and deer movement may have influenced the results. The cost analysis did not allow for reusing cameras in later surveys, which would make the two methods closer in cost.

- The best method always depends on the study area. The authors suggest that, in their study area, both methods could be reliable and easy alternatives to counts by hunters.

### 4. How tracking was or was not used

Track counts were one of the two methods tested. Trackers walked randomly placed transects and counted recent roe deer tracks crossing them. They had to identify roe deer footprints and separate them from wild boar and red deer by size, the shape of the hoof tips and the position of the dewclaws, allowing for how footprints change with gait and ground. The paper illustrates the front and hind footprints of all three species, drawn by the first author. Only recent tracks were counted, which implies judging the age of tracks, although the paper does not say how this was done. Track data were recorded with CyberTracker on a smartphone. No trailing is reported.

### 5. Tracker or observer certification

Reported but not validated. The paper says the field workers were “experienced trackers” who knew the study site and species well, and “specialized trackers”, but gives no certification or test of their skill. The first two authors are with Kokulandela.org, CyberTracker Italia Association, but the paper does not say whether the trackers held CyberTracker certificates.

### 6. Other technologies, skills or innovations

- CyberTracker software (programmed in 16 hours) on a smartphone, recording georeferenced track counts that were then downloaded to a computer. The authors found it user-friendly and not time-consuming.

- 24 camera traps, used without bait, with the Rowcliffe et al. (2008) density method, which does not need individual animals to be recognized.

- Body mass scaling rules to estimate roe deer speed and daily travel distance.

- Forensic rulers for measuring footprints.

### 7. Implications for conservation (from the paper)

- Monitoring population density is a vital part of conservation and management, especially for species of conservation concern and species that come into conflict with people, such as crop raiders.

- Monitoring needs reliable methods, and in times of economic crisis the cheapest effective method should be found and encouraged.

- Track counts may give wildlife managers a cost-effective alternative to counts by hunters, which need many volunteers. The study sites were not suitable for pellet counts because of open areas, ecotourism and settlements.

### 8. Implications for tracking (Original Wisdom's observations)

_These are Original Wisdom's observations, not the authors'._

- Track counts here matched camera traps closely, at lower cost and with only two people in the field, which suggests a practical role for skilled trackers in routine deer monitoring.

- The method depends on trackers telling similar hoofed animals apart and counting only fresh tracks, yet their skill was not tested or reported. Certifying the trackers would make the method easier to trust and to repeat elsewhere.

## Cite this summary

Lawrence, K. 2026. Summary of: Use of track counts and camera traps to estimate the abundance of roe deer in North-Eastern Italy: are they effective methods? (Romani and Giannone et al., 2018). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2018-romani-giannone-use-track-counts-camera/