Experiments in sheep-tick control require the comparison of tick counts on sheep pastured on an infested grazing or on a blanket dragged over the grazing. The former should include only attached females on axillae, forearms, head, neck and chest.
The distribution of tick counts on a group of sheep is in general not fitted by the normal distribution, nor, although it is positively skewed, by the Poisson. There is evidence that the distribution could be derived from a population distributed according to a negative binomial. Unless the data are more than ordinarily skewed (in which case, a ‘normalizing’ square-root transformation may be necessary), the best available method for comparing mean tick counts on sheep is the direct t-test using actual tick numbers. Within the range 0.3–103.7 ticks per sheep, the regression of standard deviations on means is linear and was calculated as 5=0.477x̄+1.246
For estimating the significance of small differences between dips or ground population densities, tick counts on groups of five sheep are inadequate. Groups of twenty or more sheep should be employed according to the magnitude of the difference required to be proved significant. For showing the trend, only, of tick activity throughout a season on a particular section of land weekly counts on ten sheep are adequate provided the same ten sheep are always used. The sources of the large variation in the tick counts of individual sheep are pointed out, with suggestions as to how this variation may be reduced.
In blanket dragging, counts of nymphs are best for estimating population densities. Drags may be. limited to 25 yd. A differential equation is given whereby the nymphs lost in the course of a drag are taken into account. This does not surmount the difficulty that uniformity of vegetation surface influences the efficiency of the blanket, which should therefore be used for comparisons only when the vegetation surfaces are of similar uniformity. A worn blanket picks up fewer ticks than a less worn blanket. The distribution of nymphal blanket counts is similar to that of female tick counts on sheep in that it is not in agreement with the Poisson law although positively skewed. It is in closer, though far from satisfactory, agreement with a ‘contagious’ distribution.
Because of the effect of changing meteorological conditions on tick activity, the densities of tick population (tick activities) on different plots must be compared by dragging the plots simultaneously. A virgin stretch of ground is necessary for each drag in each season. In such comparisons, the precision can more easily be increased by increasing the number of ‘occasions’ (days) of simultaneous draggings than by increasing the number of drags per ‘occasion’. Two, or three, drags per plot are sufficient provided the number of, ‘occasions’ is not less than 20. The significance of plot differences in density may be calculated from the analysis of variance of drags.
For plots less than 3 acres, the blanket method is easier; for large areas, especially with varied vegetation cover, tick counts on sheep are preferable.