RESUMEN
A total of 720 barrows (line 200â ×â 400, DNA genetics) were used in two 42-d nursery trials (initially 6.20â ±â 0.12 kg and 5.63â ±â 0.16 kg, respectively) to evaluate strategies for allotting pigs to pens in randomized controlled trials. At placement, the population was split into three cohorts with similar average weight and standard deviation and randomly assigned to one of the three allotment strategies. Strategy 1 (random) utilized a simple randomization strategy with each pig randomized to pens independent of all other pigs. Strategy 2 (body weight [BW] distribution) sorted each pig within the cohort into one of the five BW groups. One pig from each weight group was then randomly assigned to a pen such that distribution of BW within pen was uniform across pens. Strategy 3 (BW grouping) sorted pigs within the cohort into 3 BW categories: light, medium, and heavy. Within each BW category, pigs were randomized to pen to create pens of pigs from each BW category. Within each experiment, there were 72 pens with five pigs per pen and 24 pens per allotment strategy. For all strategies, once pigs were allotted to pens, pens were allotted to one of the two treatments for a concurrent trial. In experiment 1, environmental enrichment using ropes tied near the pan of the feeder was compared to a control with no enrichment. In experiment 2, treatment diets consisted of basal levels of Zn and Cu from the trace mineral premix for the duration of the study (110 and 17 mg/kg, respectively; control), or diets (supplemented control) with carbadox (50 g/ton; Mecadox, Phibro Animal Health, Teaneck, NJ) fed in phase 1 (days 0 to 22) and 2 (days 22 to 43), pharmacological levels of Zn and Cu (2,414 mg/kg Zn from ZnO; 168 mg/kg Cu from CuSO4) fed in phase 1, and only pharmacological levels of Cu (168 mg/kg Cu from CuSO4) fed in phase 2. These treatment designs were used to determine the impact on coefficient of variation (CV) and to estimate the number of replications required to find significant treatment differences based on allotment strategy. There were no meaningful allotment strategyâ ×â treatment interactions for either study. For between-pen CV, pigs allotted using BW distribution and BW grouping strategies had the lowest CV at allotment and final weight in both trials. For overall average daily gain in experiments 1 and 2 in experiment 2, the BW distribution strategy required the fewest replications to detect differences in performance. However, there is no meaningful difference between allotment strategies in replications required to detect significant differences for gain:feed ratio.
Decreasing variation between experimental units increases the likelihood of finding a statistically significant difference if one exists. Assignment of animals to experimental units (pens) may contribute to that variation. Therefore, the purpose of this trial was to investigate the effect that different methods of allotting pigs to pens (experimental unit) have on variation and in turn, the number of replications required to detect a significant difference of a given amount between treatments. The random strategy assigned pigs to pens in a completely random fashion. The body weight (BW) distribution strategy ordered pigs from lightest to heaviest and created five groups based on BW. Each pen was randomly assigned one pig from each of the five groups. The BW grouping strategy again ordered pigs from lightest to heaviest but split pigs into three groups based on BW and each pen was randomly assigned pigs from only one BW group such that there were pens of light pigs, pens of medium pigs, and pens of heavy pigs. Ultimately, the best allotment strategy depends on the parameter of interest. For final BW and overall ADG, the BW grouping method required the fewest pens to detect statistically significant differences.