The use of GPS technology has revolutionized how farmers monitor the behavior of their cattle when they are out in pastures. In this article, we show how data from digitanimal GPS collars can be used to identify changes in cattle behavior associated with calving. Through a study conducted on a commercial farm, we explore key indicators constructed from GPS data that can help identify when a cow has calved.
Unlike intensive production systems, such as dairy farming, where animals are confined, and calving occurs in controlled areas, extensive systems often scatter cattle across large tracts of land, making it very difficult to observe or monitor calving. However, calving in a cow’s life cycle is a critical moment. Detecting calving can enable early assistance in case of dystocia, and even when calving is normal, it can serve to protect the cow and calf from predators, adverse weather, and more. Currently, there are various commercial solutions based on sensors for remote calving detection. However, most of these technologies are designed for intensive livestock farming, where real-time data collection and transmission are much easier than in extensive systems.
GPS Collars as a Potential Solution
Today, GPS collars are becoming increasingly popular among farmers, presenting an opportunity to collect data on grazing livestock behavior. However, the collars often have low data temporal resolution due to battery limitations. Currently, most devices provide location data at intervals ranging from 30 minutes to two hours. At the School of Agricultural and Forestry Engineering (ETSIAM) at the University of Córdoba (UCO), we explored whether these low-resolution GPS data are sufficient to detect behavior changes associated with calving.
Study Methodology
The study was conducted on a dehesa estate situated in the municipality of Pedroche, Córdoba. Spanning approximately 200 hectares and divided into 14 enclosures, this estate is home to 75 cows (in addition to Iberian pigs) that graze. Following the typical practices of the region, rotational grazing of animals occurs across different enclosures throughout the year. A total of 57 cows were monitored, with 42 of them calving during the study period. Each cow was equipped with a smart collar featuring GPS technology. These collars allowed the collection of data on animal grazing behavior. The temporal resolution of the data was 30 minutes. The GPS tracking data obtained were processed using the statistical software R. Daily trajectories and daily spatial utilization of enclosures were calculated.
Behavior Indicators Around Calving
We considered two approaches to characterize animal trajectories and enclosure utilization behavior. The first approach involved calculating individual indicators (referring to a single cow) to determine what behavior changes around calving a single collar could detect. Individual indicators included the length of the trajectory and the home range, calculated as the minimum convex polygon. In the second approach, we calculated metrics involving the entire herd to assess changes in the behavior of an individual cow around calving in relation to her companions. Detecting these changes would require the use of multiple collars. The herd’s spatial indicators used to analyze cow behavior around calving were the maximum, mean, and minimum distances from each cow to the herd centroid.
Changes in Trajectory Length
There were no statistically significant differences in the daily trajectory length of the animals between the calving day and the surrounding days. However, we observed a marked decrease in the distance covered by cows in the days following calving compared to the days before. Before calving, cows covered an average of about 2.3 km per day, while after calving, this figure decreased to around 1.9 km.
Changes in Occupied Area
This indicator also observed differences between the pre- and post-calving periods, with the occupied area decreasing from 4.1 hectares before calving to 2.4 hectares after calving. However, again, there were no differences between the calving day and the surrounding days. It is noteworthy that the day before calving recorded a peak in the occupied area. This peak could have resulted from cows exploring the estate or enclosure in search of a suitable place to calve.
Distance to Herd Centroid: A Revealing Indicator
Significant differences were observed in the maximum, mean, and minimum distances of cows to the herd centroid during the 14 days surrounding calving (P < 0.001). On the calving day, we recorded the farthest distance to the centroid. In the graphs, cows moved away from the herd’s centroid two days before and after calving, but returned to similar distances as before calving on the third day as cows integrated with their calves. While this indicator appears useful for calving detection, herd dispersion during grazing may distort results, impacting centroid calculations (though GPS collar data can help detect subgroups).
The average distance to the herd centroid is more reliable than the maximum distance, which can show spikes caused by individual cow movements unrelated to calving, possibly resulting in false alarms. The minimum distance to the herd centroid also clearly reflects calving behavior, but if the calving cow approaches the herd at a specific time (or vice versa), this indicator would no longer be representative of the isolation behavior of the animal on the calving day.
Future Applications and Conclusions
In this study, we have identified measurable variables using GPS technology for detecting calving in cows. We identified daily behavioral changes related to calving using the studied metrics. However, improving GPS technology or using complementary methods for more frequent calving detection, such as hourly intervals, would maximize real-time monitoring benefits.
Towards a Calving Detection Model
Despite limitations, this study identifies patterns for a model to detect recent cow calving based on behavioral deviations. Extending this model to include calving success, like dystocia detection, can benefit commercial livestock management. Such implementation improves animal welfare and boosts economic gains through rapid intervention and higher survival rates.
Validation of variables and the importance of the number of devices
However, researchers must validate these variable changes in other farms and production systems to ensure proper model development. The distance to the herd centroid requires multiple devices placed on different animals to calculate and interpret group behavior effectively. Therefore, having multiple devices significantly increases the chances of detecting births effectively.
Towards more efficient animal management and care
In summary, the advances achieved in this study open new avenues for more efficient animal management and enhanced care. Implementing models based on the variables identified in this study could substantially improve early detection of births and/or associated issues. This approach not only enhances animal welfare but can also have a positive impact on the economic outcomes for livestock farmers.
💻 https://digitanimal.co.uk
📤 info@digitanimal.co.uk
📞 +44 75 1159 7693
