The data collected were disease severity of a leaf disease of tea in two seasons. The trial was carried out in 3 locations. Each location had test cultivars as well as 2-3 reference cultivars. However, all the test and reference cultivars are not repeated in each location. The composition is different from place to place.
Analysis depends upon objectives. As I understand, your data contains mixed effects while the cultivars are not repeated. If you collected data over different time periods, it can be solved using liner mixed models under repeated measure design. For appropriate answer, complete detail of your study objective and methodology is required.
Thank you. The objectives are to classify the cultivars in to different resistant/susceptible classess (highly resistant, resistant, moderately resistant &susceptible) and to indentify cultivar x environmental interactions. The data were collected for two seasons May-Sep in 2013 & 2014 at 3 locations.
Mixed model data analysis would fit with your objectives. It works with missing data points where the treatments are not replicated in each location.
Depending on your data you can use or not variance analyses to compare your cultivars with a general linear model under repeated measure (as mentioned). So which kind of data do you have?
dead you rate disease severity with scale (for ex a 0 to 5 scale) or do you do a metric measure for the disease ( for ex: percentage of diseased leaves or leaf surface with symptoms).
Anyway, you are stuck with the mixed model that our colleagues mentioned earlier. However, you should know in advance what is fixed and what is random in your design (objectives). for example, is your sites are fixed, and or your cultivars are random. Seasons are usually by nature (statisticians) random.
For such classification, non replicated data can work. Last year I found slow rusters of wheat using heatmaps because i had have multivariate data. But you have only disease severity then for such objectives cluster analysis can work better I guess. Moreover, for cultivar*environmental studies use regression analysis. But again if you have non-replicated data for this objective too the response surface methodology (RSM) will work. Good luck.
Thank you all for your comments and suggestions. I will try .......
You can organize data according to all parameters that have assessed and others like climatic or soil analysis. You can evaluate the possibility of analysis with nonparametric statistics (pricipal components or other) and conduct a more comprehensive analysis.
Everything will depend on the amount of information you have available.
- Badges
- Science topic
- Similar topics
- Agricultural Science
- Agronomy
- Crop Science
- Crop