Suboptimal reproductive performance is a major contributor to economic losses in dairy farming. Studies conducted in Europe and North America have consistently linked poor reproductive outcomes with reduced cow lifespan and significant financial losses in dairy systems [1], [2], [3]. One of the first steps toward achieving a successful pregnancy in cows is the timely and accurate detection of estrus, a behavioural indicator that precedes ovulation [4]. Effective estrus detection is crucial for reproductive management on dairy farms, as it is associated with improved insemination rates, optimized calving intervals, and increased pregnancy outcomes [5]. However, despite its critical role in reproductive success, estrus detection remains a challenging and labor-intensive task. The variable nature of estrus behaviour, combined with increasing herd sizes and labor constraints on modern dairy farms, often makes consistent and accurate detection difficult to achieve through visual observation alone.

Over the past several decades, various estrus detection tools have been developed and integrated into dairy farm operations, primarily relying on identifying behavioural changes associated with estrus [4], [6]. For example, one of the most well-established indicators of estrus behaviour is standing to be mounted, which can be monitored cost-effectively by applying chalk to the cow's tail base before the expected estrus period [5], [7]. The chalk is removed by friction when the cow is mounted by other cows, providing a visual indicator of estrus activity [5], [7]. In addition to mounting behaviour, secondary indicators, such as increased physical activity [8], can be monitored using automated activity monitors (AAM). These wearable sensors detect locomotion changes linked to estrus, reducing reliance on direct visual observation [4]. In fact, estrus detection aids have significantly contributed to improving reproductive performance on dairy farms by offering valuable support to dairy producers, and by enabling the possibility of targeted reproductive management protocols [9].

Despite its advancements, one significant challenge when using estrus detection tools is the occurrence of false positive alerts. Studies have shown that the rate of false positive estrus alerts can reach up to 29% on dairy farms [10], [11], [12], [13]. These inaccuracies may arise from behaviours such as cows licking off tail chalk [14] or the misclassification of activity changes when cows become lame [15]. One potential solution could be the integration of multiple estrus-related behaviours (e.g., mounting behaviour and increased physical activity) into detection systems. In fact, studies have suggested that combining tools with visual observation of estrus can improve the accuracy of true estrus detection [16], [17], [11]. However, this approach does not alleviate the time demands associated with visual observation. Integrating cost-effective visual markers, such as TC, with an AAM could improve the accuracy of estrus detection while minimizing the need for continuous visual monitoring on dairy farms. Although previous studies have compared the use of TC and AAM for estrus detection [18], [11], [13], limited research has evaluated their combined use in the same animal [18], particularly during spontaneous estrus, which may better represent the underlying reproductive physiology of the modern lactating dairy cow.

The primary objective of this cohort study was to assess how two estrus detection tools (AAM and TC), used separately and in combination, were associated with true estrus events in spontaneous episodes of estrus from lactating Holstein cows. Our secondary objectives were to explore additional associations with true estrus, considering different factors such as estrus characteristics (peak and duration), milk production and peri-estrus ovarian structures dynamics.