Proc npar1way

Initially, a bivariate analysis of BCS and predictors was performed. As some of the data groups had few observations, it was decided, prior to analysis, to merge compatible groups to increase the statistical power. A graphical representation of the results was made using proc univariate (SAS Inst.). The statistical significance of the bivariate analysis was performed using proc npar1way (SAS Inst.). An expected clear non-normality of the scores was observed, and based on this finding, the Kruskal–Wallis test was used to test for significance.
The data were analyzed with a multinominal logistic regression model with ordered data (ordinal scale), using the cumulative logit link (proc genmod, SAS Inst.). The training level, profession of the data collector, and age were continuous variables. The type of horse, discipline/intended use, sex, and site of collection were fixed effects. Data collector was included as a repeated factor due to an assumption that differences between collectors could be expected.
After an initial multinominal logistic regression analysis, “profession of data collector” was excluded, as it showed a very high level of correlation with “site of collection/type of consent”. The remaining factors were retained in the model even if they were not significant due to confounding effects between factors.
An additional analysis was made where BCS categories were merged into three groups: ideal (BCS 5,6), above ideal (BCS > 6), and below ideal (BCS < 5). The rationale behind this was that the ordinal scale of BCS may not reflect the same mechanisms in horses scoring below the ideal BCS vs. horses scoring above the ideal BCS. Mechanisms that increase the risk of being below ideal are likely different from the mechanisms that decrease the risk of being above ideal.
At first, simple contingency tables were made with p-values based on chi-square tests (Fisher’s exact test was used if there were fewer than five observations in a cell). A multinomial logistic regression model (proc glimmix, SAS Institute) was performed, with ideal BCS as the reference value, and below and above ideal BCS as categories. The modelling approach followed the same procedure as the ordinal analyses. Data collector was introduced as a random variable.
For explanatory variables with more than two levels and a p-value below 0.10 for the overall effect of the variable, further analyses were performed to investigate if there were groups that differed significantly from the rest of the groups. This was done stepwise, so the group with the largest difference from the average BCS was tested against the remaining groups. If this was significant, the procedure was repeated, comparing the next group to the remaining groups. This procedure continued until the p-value was non-significant.

For all data, ANOVA and two-way Student’s t-test was used to determine statistical significance. Analysis of the ratio of isoprenylogue species was done using Wilcoxon-Mann-Whitney test (PROC NPAR1WAY, SAS software, version 9.4). Outliers were calculated and excluded from statistical analyses. Significance is indicated by *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001.

GraphPad Prism V9 Software (San Diego, CA, USA) was used for most of the statistical analyses. The data from the cytotoxicity assays and transcriptional analysis of Am069, Am240, and Am392 were analyzed by two-way ANOVA followed by a Dunnett’s test for multiple comparisons. Raw numbers from the cytotoxicity assays were used in the statistical analysis. The data from RT-qPCR experiments quantitating A. marginale relative to tick cell numbers were log2 transformed, and significant differences between groups were determined using two-way ANOVA followed by a Tukey’s test for multiple comparisons.
Intracellular growth of bacterial colonies was measured as the total area of threshold anti-Msp2 fluorescence divided by the total area of threshold phalloidin. The effects of treatment, incubation period, and the interaction term were analyzed using a generalized linear model (PROC GLIMMIX; SAS version 9.4, SAS Institute Inc., Cary, NC, USA). The data were well fit by the beta distribution, and comparisons of interest were made by including an LSMEANS statement for the interaction term sliced by the incubation periods and significance values adjusted by the step-down Bonferroni (Holm) procedure. The effects of treatments and time on colony sizes were compared by visual inspection of empirical cumulative distribution functions (PROC NPAR1WAY, EDF; SAS).

Fresh krill were not available for the experimental laboratory trials needed for this study, so preserved animals were used in these trials. Determining if and how the preservation treatment changed the morphology of krill were important for accurate interpretation of the results. When Germany first initiated its krill research program in the Southern Ocean in 1976, researchers compared fresh, frozen, ethanol-preserved, and formalin-preserved krill samples to evaluate potential effects of preservation on morphological properties. They observed shrinkage in samples preserved in ethanol, but no statistically significant differences were detected in formalin fixed or frozen animals (Volker Siegel, pers comm.). Because these factors are fundamental for the reliability of our results, and because these historical results never were published, this result required corroboration.
During a survey conducted aboard the Norwegian fishing vessel Juvel (Emerald Fisheries ASA) off the South Orkney Islands in late January to early February 2012, krill were collected fresh from the catch using a Macroplankton trawl (see [18] for additional descriptions of the trawl). The trawl was lowered from the sea surface to 200 m depth and hauled at 2.5–3 knots. Sex and maturity stages of krill were determined using the classification methods outlined by [19] . Total body length was measured (±1 mm) from the anterior margin of the eye to the tip of telson, excluding the setae, according to the “Discovery method” used in [20] . Carapace width was measured using a caliper at its widest cross section point. A total of 30 krill including juveniles, sub adults, adults with gravid females at stage FIIID were preserved individually in borax-buffered formalin (4%), and body length and carapace width were measured again after 2 and 10 months.
Comparisons of any temporal change in the morphology measurements was made using an analysis of variance test (Proc NPAR1WAY, SAS Institute Inc., Box 8000, Cary, N.C., U.S.A.) with the 0.05 level accepted as indicating statistical significance.