1. Conducting Experimental Trials Gary Palmer

2. Scientific Method  Formulation of Hypothesis  Planning an experiment to objectively test the hypothesis Treatments Treatments Experimental design Experimental design Replication Replication  Careful observation and collecting of data from the experiment  Interpretation of the experimental results

3. Experiment Treatments Prowl Prowl Orthene Orthene Ridomil Ridomil

4. Experiment TrteatmentRateNozzles MH 2 gal Fine MH + Prime+ 1.0 gal + 0.5 gal Coarse MH + Prime+ 1.5 gal + 0.5 gal Coarse MH + Butralin 1.0 gal + 0.5 gal Coarse MH + Butralin 1.5 gal + 0.5 gal Coarse

5. Experiment TrteatmentRateNozzles MH 2 gal Fine MH Coarse MH + Prime+ 1.0 gal + 0.5 gal Coarse MH + Prime+ 1.5 gal + 0.5 gal Coarse MH + Butralin 1.0 gal + 0.5 gal Coarse MH + Butralin 1.5 gal + 0.5 gal Coarse

6. Effects of Application of MH by Coarse vs Fine Nozzles on Yield of Burley Tobacco

8. How many reps are needed? A. <3 B. 3 C. 4 D. 5 E. 6 F. >6 G. It depends

9. Why Replicate?  Treatment is repeated two or more times In a randomized pattern In a randomized pattern  Provides an estimate of experimental error  Provides a more precise measurement of treatment effect

10. How many reps are needed?  Number required Depends on magnitude of detectable differences desired Depends on magnitude of detectable differences desired 100 lb/a, 200 lb/a, 600 lb/a, ?100 lb/a, 200 lb/a, 600 lb/a, ? Inherent variability of subject matter Inherent variability of subject matter CV = Coefficient of VariationCV = Coefficient of Variation Expressed in %Expressed in %

11. Inherent variability Coefficient of Variation (CV)  Small Grain (6-8%)  Tobacco yields (16-18%)  Black shank (50-70%)  Johnson grass (30-40%)

12. Inherent Variability  Actual CV%  Tobacco Location MH Residue Clayton 44.4Clayton 44.4 Reidsville29.9Reidsville29.9  Wheat Location Stand Vigor  Clayton 6.3 55.5 Reidsville 5.3 64.7

13. Burley Tobacco No-till Variety Trial Taylor Co. – Adam Pike Farm

14. Burley Tobacco Yield Harrison Co. – Ricci Rowland Farm

15. How can we increase precision?  The precision of an experiment can always be increased by additional replications  Degree of improvement falls off rapidly as replication increase Example: To double the precision with which two means can be separated in a trail with four replications, reps would have to increase to 16. Example: To double the precision with which two means can be separated in a trail with four replications, reps would have to increase to 16.

16. Arrangement of Replications Confounded with fields/soil type Field 1 Field 3 Field 2 Wet Spot Rep 1 Rep 3 Rep 2 Rep 4

17. Arrangement of Replications Confounded with soil type Rep 1 Rep 2 Rep 3 Rep 4Rep 5 Rep 6 Rep 7 Rep 8

18. Arrangement of Treatments Randomized complete Block Rep 2 Rep 3 Rep 4Rep 5 Rep 6 Rep 7 1 234567 2451376

19. Arrangement of Treatments Balanced Incomplete Block Design 1 23 456 1 1 4 82 5 5 1 Blk 1 Blk 2 789 Blk 3 4736 Blk 6Blk 5Blk 4 9 9 Blk 11Blk 10 Blk 8Blk 7 Blk12 Blk 9 7 7 2 2 3 3 6 6 4 5 REP 1 8 REP 2 REP 3 189 REP 4

20. Arrangement of Treatments Unbalanced Incomplete Block Design 3567 57 2 174 4 6 1 1 2 2 7 136 3 3452456Blk 6 Blk 2 Blk 3 Blk 4 Blk 5 Blk 1 Blk 7

21. Replication Confounded with Time  Not enough growth chambers available?

22. Simulated Data TreatmentLb/a A3300 B3200 C3100 D2900 E2800 F2700

23. Herbicide Treatment TreatmentMean Check1800 Prowl2300 Command2200 Spartan2250 Prowl+Command2335 Prowl+Spartan2295 Command+Spartan2400

24. Treatment Rep 1 Rep 2 Mean Check110025001800 Prowl220024002300 Command230021002200 Spartan230022002250 Prowl+Command200026702335 Prowl+Spartan160028002200 Command+Spartan280020002400

25. The Completely Randomized Design   The CRD is the simplest of all designs.   Replications of treatments are assigned completely at random to independent experimental subjects.   Adjacent subjects could potentially have the same treatment.

26. The Completely Randomized Design   Sample layout: Different colors represent different treatments. There are 4 (1-4) treatments with 4 replications (A-D) each

27. Experimental Design  Randomized Complete Block Design RepABCD 4133 1442 2324 3211

28. Experimental Design  Sampling Treatment1234 SamplesABCDABCDABCDABCD

29. Experimental Designs  Factorial One or more factors in one experiment One or more factors in one experiment  Study factors that might make cured tobacco darker Nitrogen (200, 300, 400 lb/a) Nitrogen (200, 300, 400 lb/a) Phosphorus (Low, soil test, high) Phosphorus (Low, soil test, high) Potassium (Low, soil test, high) Potassium (Low, soil test, high) Harvest date (3, 4, 5 weeks) Harvest date (3, 4, 5 weeks)

30. Experimental Designs  Factorial One or more factors in one experiment One or more factors in one experiment  Study factors that might make tobacco darker Nitrogen (200, 300, 400 lb/a) Nitrogen (200, 300, 400 lb/a) Phosphorus (Low, soil test, high) Phosphorus (Low, soil test, high) Potassium (Low, soil test, high) Potassium (Low, soil test, high) Harvest date (3, 4, 5 weeks) Harvest date (3, 4, 5 weeks)  3x3x3x3x4=324 plots If 4 rows by 30 ft = 3.12 acres If 4 rows by 30 ft = 3.12 acres

31. Interpretation  What if plant response to nitrogen changes as potassium rates change? How do you interpret? How do you interpret?  What if plant response to nitrogen changes as harvest date increase and potassium rates decrease? How do you interpret? How do you interpret?

32. Split-Plot Design  Main plot Ex. Nitrogen Rate Ex. Nitrogen Rate Least precision Least precision  Split Plot Ex. Harvest Date Ex. Harvest Date Most precision Most precision

33. Split-Plot Design Nitrogen Rate Harvest days after topping Reps 20050203040 D 40020405030 40030504020C 20020405030 40030502040B 20050403020 40040205030A 20020403050

34. Yield of Burley Tobacco Varieties with Different Fertility – UK Woodford Co. Farm Nitrogen was applied in the ammonium nitrate form.

35. Experimental Unit Size  In general, variability decreases as plot size increase.  However, once a certain size is reached the increase in precision falls off rapidly.  Little gain in field plots beyond 0.1 acre  Precision is good at 0.01 to 0.02 for most crop (4 rows by 30-35 ft in tobacco)  Individuals may be used as an experimental unit Trees, cows, goats, etc. Trees, cows, goats, etc.

36. Experimental Unit Size  Covariance Removing variability of some independent variable Removing variability of some independent variable Difference in beginning weight of cows in testDifference in beginning weight of cows in test Differences in plant standsDifferences in plant stands