p. 692

Lambs were present on HREC from November through mid-May, and only the months of this period were used in analyses involving lambs. To avoid confounding patterns of ewe and lamb killing, only months when lambs were absent (May-October) were used in analyses involving ewes. Removal of coyotes, which could take place during any month of the year, was regressed against only months when lambs were present.

For example, removing 3 coyotes in 1 month might not decrease the number of lambs and ewes killed until the following month, or the following year. Therefore, we used a cross-correlation analysis (Diggle 1990) to detect correlations at all monthly intervals from 0 to 24 months.

p. 693

Annual lamb and ewe kills and kill-rate were not correlated with the number of coyotes removed.

p. 694

Annual mean number of coyotes removed was 10.8 (SE=1.6, n=13). Annual mean number of missing sheep was 177 (SE=8.9, n=8), which was an average of 11.8% (SE=0.5, n=8) of the HREC flock. Annual number of missing sheep was not correlated with annual number of kills.

Neither monthly number of lambs killed nor lamb kill rate was correlated with number of coyotes removed. Similarly, neither monthly number of ewes killed nor ewe kill rate was correlated with number of coyotes removed.

The 12-month lag of the number of coyotes removed had the highest correlation coefficient for number of lambs killed and lamb kill rate.

p. 696

If only certain coyotes kill sheep, increasing numbers of coyotes removed will have no effect on the number of sheep killed unless the problem coyote is removed. Thus, the lack of relation between the number of lambs killed and number of coyotes removed may be because most of the coyotes removed were not killing sheep.

However, preventative effects were low; the highest negative coefficient of determination was 0.08, indicating that only about 8.4% of the variability in kill rates can be explained by the removal of coyotes.

There was no indication that removal of coyotes reduced predation the following year (i.e., lagged intervals >12 months), suggesting that coyote density was not being reduced by removal efforts.

p. 697

Thus, in areas where coyotes are naïve with respect to control efforts, removal of the problem territorial animals may be more successful. At HREC or other pasture operations with consistent coyote control, typical removal efforts may differentially remove young coyotes and only occasionally remove problem coyotes, resulting in weak negative correlations between total coyotes removed and subsequent sheep kills.

p. 698

Results of the analyses presented here suggest the need for more selective targeting of coyotes…. A method that is selective for sheep-killing coyotes is the 1080 Livestock Protection Collar, which is limited in availability by restrictions necessary for its safe use.