Results

The complete randomization of pigs across treatments resulted in initial BW not differing among DON treatments (p > 0.100; data not shown). The phase 1 and phase 2 growth performance data are presented in Table 3. Phase 1 ADG and ADFI decreased with increasing DON concentration by up to 52% and 44%, respectively (Linear P < 0.001). As such, G:F during phase 1 decreased by up to 13% as DON concentration increased (Linear P = 0.037). During Phase 2 (d 14–24), ADFI decreased by up to 39% as DON increased (Linear P < 0.001). Growth continued to be decreased by DON in a linear (P < 0.001) and quadratic (P = 0.034) manner by up to 22% decline across the titration treatments. Despite continued reduced feed intakes, G:F during Phase 2 increased with increasing DON concentration by up to 23% (Linear P < 0.001).

 | DON mg/kgf |  | P-valueg
 | 0.2 | 0.9 | 1.7 | 2.4 | 3.1 | 4.0 | 5.1 | SEM | Linear | Quadratic
Phase 1, d 0–14 |  |  |  |  |  |  |  |  |  | 
ADG, kg | 0.509a | 0.469ab | 0.259de | 0.404bc | 0.354 cd | 0.284de | 0.245e | 0.025 | <0.001 | 0.150
ADFI, kg | 0.638a | 0.572ab | 0.413 cd | 0.489bc | 0.471bc | 0.364d | 0.355d | 0.250 | <0.001 | 0.103
Gain:Feed | 0.795ab | 0.818a | 0.652c | 0.829a | 0.748abc | 0.774ab | 0.688bc | 0.028 | 0.037 | 0.688
Phase 2, d 14–24 |  |  |  |  |  |  |  |  |  | 
ADG, kg | 0.697a | 0.714a | 0.664ab | 0.736a | 0.644ab | 0.535b | 0.547b | 0.032 | <0.001 | 0.034
ADFI, kg | 1.127a | 1.048ab | 0.927bc | 1.047ab | 0.888bcd | 0.747 cd | 0.688d | 0.046 | <0.001 | 0.196
Gain:Feed | 0.619b | 0.684ab | 0.715a | 0.704a | 0.674ab | 0.716a | 0.760a | 0.021 | <0.001 | 0.619

A small number of pens were removed for Gain:Feed following prespecified outlier diagnostics outlined above: 1 pen from 5.1 mg/kg treatment and 2 pens from 3.1 mg/kg. The 1.7 mg/kg treatment mean did not follow the otherwise consistent dose-response pattern and influenced regression model fit; therefore, regression analyses were conducted with and without this treatment level, with models excluding it used for primary interpretation. A second mycotoxin screen was conducted on the 1.7 mg/kg treatment, and the results were the same, confirming accuracy of the analysis. Treatment means for all seven dietary DON concentrations are presented for transparency (Fig. 1). Over the entire 24 d study period, compared to the 0.2 mg/kg diet, increasing dietary DON concentrations reduced BW by 1.22 kg for each 1 mg/kg increase in dietary DON (Linear P < 0.001; _R_2 = 0.50; Fig. 1A). Similarly, overall ADG declined linearly with increasing DON (Linear P < 0.001; Fig. 1B), corresponding to a reduction of approximately 0.04 kg/d per 1 mg/kg DON increase (_R_2 = 0.62). Average daily feed intake exhibited the strongest correlation with DON concentration (_R_2 = 0.69), decreasing by 0.07 kg/d per mg/kg DON increase (Linear P < 0.001; Fig. 1C). Overall Gain:Feed was not linearly related to DON concentration (P = 0.472; Fig. 1D), however a quadratic tendency was observed (P = 0.088; Fig. 1D), and regression analysis demonstrated minimal correlation (_R_2 = 0.01).

Figure 1: Dose-response effects of corn derived dietary deoxynivalenol (DON) on nursery pig growth performance. Panels depict DON’s impact on A) end body weight (BW) at d 24, and overall (d 0–24) average daily gain (ADG; B), C) average daily feed intake (ADFI), and D) Gain:Feed. No quadratic responses were observed for BW, ADG, or ADFI, P > 0.10. The 1.7 mg/kg treatment mean is displayed for transparency but was excluded from regression model fitting following pen-level outlier diagnostics.

One objective of this study was to determine whether a biological breakpoint existed within the tested DON range using segmented regression models. Linear, two-segment, and linear-plateau regression models were used to compare model fit. Model comparison indicated that linear models provided the best overall fit based on AIC and BIC for the effect of DON (mg/d) intake on ADG). Gain:Feed was not associated with dietary DON concentration (_R_2 = 0.01); therefore, breakpoint model comparisons were not conducted for this parameter. Due to the strong linear relationship of this dataset, no clearly identifiable break point was detected.