Reliability of qualitative behavior assessment (QBA) versus methods with predefined behavioral categories to evaluate maternal protective behavior in dairy cows

Abstract Our objective was to evaluate the reliability of five methods with predefined behavioral categories and qualitative behavior assessment (QBA) as potential indicators of maternal protective behavior of Holstein-Gyr crossbred dairy cows immediately after calving. Behaviors of 88 cows were recorded during the first handling of the cow-calf dyad (usually 30–60 min postpartum) for navel disinfection and identification. Methods with predefined behavioral categories were: maternal protective score (MPS), ranging from 1 (indifferent) to 6 (aggressive); displacement (DIS, from 1 to 5); agitation (AGI, 1–4); aggressiveness (AGG, 1–3); and attention (ATT, 1–3). For the QBA assessment, a visual analog scale was used to quantify 13 terms. This study involved two observer groups: Group 1 was 10 observers applying methods with predefined behavioral categories, whereas Group 2 was8 observers applying QBA. Each observer participated in three evaluation sessions, assessing the same 20 videos three times over 15 d. Intra- and inter-observer reliabilities were calculated by the Kappa coefficient (k) or Kendall coefficient (W) for predefined scores and QBA, respectively. For methods with predefined behavioral categories, intra-observer reliability (k) of 10 observers ranged from moderate to almost perfect (0.50 to 0.87, 0.50 to 0.75, 0.74–1.00, 0.50 to 0.86 and 0.57 to 0.86 for MPS, AGI, AGG, ATT, and DIS, respectively), whereas inter-observer reliability (k) of nine observers ranged from slight to perfect compared to a “gold” evaluator, with ATT score having a lower k (0.34 to 0.72), and AGG having a higher k (0.46–1.00). For QBA, intra-observer reliability ranged from strong to very strong for Principal Component (PC) PC1 (W = 0.71 to 0.89), PC2 (W = 0.74 to 0.90) and PC3 (W = 0.63 to 0.84). Inter-observer reliability ranged from strong to very strong for PC1 (W = 0.70; 0.70 and 0.73; within evaluations 1, 2 and 3, respectively), PC2 (W = 0.71; 072 and 0.77), and PC3 (W = 0.55; 0.54 and 0.73). In general, methods with predefined behavioral categories and the QBA were useful and reliable tools to evaluate maternal protective behavior of Hostein-Gyr crossbred dairy cows under commercial conditions.