Research paperWhat makes a good mother? Maternal behavioural traits important for piglet survival

- We found maternal care traits important for piglet survival.
- We developed qualitative scores for those important maternal care traits.
- High score for nest building resulted in higher piglet survival.
- High score for sow-piglet communication resulted in higher piglet survival.
- High score for carefulness towards piglets resulted in higher piglet survival.

The primary aim of our work was to find maternal behaviours important for piglet survival and to develop qualitative scores of those traits. Second, we studied the relationship between maternal behavioural scores, piglet mortality and the number of weaned piglets in sows of three different sow breeds (Norsvin Landrace (n = 12), Norsvin Duroc (n = 12) and crossbred Norsvin Landrace × Yorkshire (n = 14)). The following qualitative maternal behaviours were scored as follows: nest building activities prior to farrowing, sow communication (q_SC), and carefulness (q_SCR) on while sows were standing, moving and just before lying down. We also continuously recorded maternal care behaviors (nest building activities (c_NBA) and sow communication (c_SC; while standing/moving, before lying down) to test the relationship with the respective qualitative scores. There was a moderate positive correlation between the continuous measured c_NBA and the qualitative score for nest building q_NBA (r = 0.469) as well as between the qualitative score for communication, q_SC and the continuous, video-based measure of communication while standing, c_SC (r = 0.439), and the qualitative and quantitative scores similarly affected piglet survival. Since q_SC and q_SCR were highly correlated (r = 0.883), we tested the effect of those behavioural scores separately on production parameters (proportion of dead piglets and number of weaned piglets) using two models (model 1: q_NBA, q_SC; model 2: q_NBA, q_SCR), and compared their relative predictive accuracies using Akaike information criteria (AIC) and AIC weights. In models 1 and 2, piglet mortality decreased with higher q_NBA (P = 0.004; P < 0.001; respectively) due to less crushing (P < 0.001; P < 0.001, respectively) and, thus, more weaned piglets (P = 0.043; P = 0.035; respectively). Increases in both q_SC and q_SCR were associated with lower overall piglet mortality (P < 0.001; P < 0.001, respectively), fewer crushing incidences (P < 0.001; P = 0.002, respectively) and, therefore, more weaned piglets (P = 0.004; P = 0.030; respectively). Additionally, higher q_SC in model 1 was associated with a lower proportion of starved piglets (P = 0.002). Model 1 had better predictive accuracy than model 2 for all productive parameters. Our results demonstrated that our three defined maternal behaviour scores had a significant impact on piglet survival, and therefore we would like to proceed with testing of these scores in nucleus herds of Norsvin Landrace sows to further calculate heritabilities and potentially implement the most successful behavioural trait in the breeding program.