The mares were distributed among routinely-managed groups of approximately 30 pregnant mares, grouped based on anticipated foaling dates.
As gathers could have affected foaling date, we used a linear mixed effects model to test for differences between foaling dates before and after gathers (but before contraception management).
All horses are considered to age a year on the first of August, regardless of their actual foaling date.
We monitored pregnant mares 2 3 times per week to ensure accurate estimation of foaling date.
We included these terms in the mixed effects models described below to determine if weather affected foaling date.
After controlling for weather and mare age, mean foaling date did not differ before and after gathers (P = 0.63).
Although the removal of offspring can induce estrous cycling in ungulate species [62], it is unlikely that the removal of foals has influenced foaling date among PZP-treated mares on Shackleford Banks.
To test for differences in foaling date with PZP treatment, while controlling for pseudo-replication and unequal variances, we used a linear mixed effects model.
Among mares treated with PZP, those receiving a higher total number of applications foaled later in the season than did those receiving fewer applications, with each additional application associated with a 0.5 month delay in foaling date (Linear Mixed Effects Model: estimate = 0.55, SE = 0.21, t = 2.61, r2 = 0.65, P = 0.01, see Fig. 2A).
In addition, each consecutive PZP application was associated with a 0.8 month delay in foaling date, on average (Linear Mixed Effects Model: estimate = 0.83, SE = 0.23, t = 3.64, r2 = 0.65, P<0.0008, see Fig. 2B).
To determine whether the number of PZP applications (total or consecutive) or the number of years between pregnancies (total or consecutive) influenced foaling date, we ran separate linear mixed effects models using these as predictor variables.
