Download presentation
Presentation is loading. Please wait.
Published byLaurence Ray Modified over 9 years ago
2
METs for evaluating experimental varieties.
3
Response variable: Grain yield lowmoderate A B extreme Basics of Genotype x Environment interaction Context: eg Drought stress in the target area Extent of GxE depends on range of E covered, choices of G included and type of response variable C
4
Response variable: Grain yield lowmoderate A B extreme G x E design problem: Context: Drought stress in the target area C Collect data to estimate responses, being Efficient Valid
5
The ‘classical’ Multi-Environment Trial for GxE 16 G (4 reps) (Randomised block design) 8 E Measure responses The variety diversity to explore. -How many? -Which ones? -Who decides? Sampling the variability in project domain -Which dimensions? How many? Who decides? Set by breeding objectives - multiple Design for efficiency, validity, practicality, legitimacy
6
GxE → OxC Genotype Climate Soil Management Farm resource endowment Market integration Gender, HH type Environment Yield Growth traits Disease resistance Profitability Acceptability Preferences X= ContextPerformance X = = Design principles? Genotype
7
Objectives determine design A.Objectives require average performance across environments GxE is part of the ‘noise’ Random selection of E to ‘represent’ target Number needed depends on σ 2 ge B. Objectives require detection, description and understanding of GxE Include hypotheses of GxC interactions Essential for designing efficient trials
8
Predictable Unpredictable Mappable Factors hypothesised to interact with G that you want to investigate -They have to vary within the study! Some can be manipulated or chosen for any plot: Planting date Fertilizer use Intercrop … Some can not: Soil type AEZ Farmer resource level Farmer gender Landscape niche Weather Pest pressure …
9
Can factor be manipulated ? Will you include it as an experimental factor? Definition of treatments Design choices Yes No Objectives of G x C hypotheses C factors to investigate Business as usual Yes No
10
Type of C factor Location Strata Random sampling Replication Design choices Predicatable Mappable Predicatable Non-mappable Unpredictable Do the trial Measure context variables
11
Data analysis Including Consistency across farms Unexplained variation = risk is small enough? New hypothese of G x C Carry on! Yes No
12
Do the trial with farmers because… 1.Assessing under realistic conditions and contexts 2.Measuring preferences 3.Sampling sufficient context variation 4.Making required large-N trials feasible 5.Participatory principles, empowerment and farmers rights
13
1m 0.5m
14
What might change when farmers are involved? Layout Researchers designFarmers and researchers design Farm 1 Farm 2 Farm 15 Experiment spread across many farms Layout may not be ‘neat’!
15
See the videos! https://www.youtube.com/watch?v=ItLyRW2L aAQ 4 on design of trials with farmers 1 on design of METs
16
Malawi example
17
What it might look like Alt gradient covers 4 AEZs About 20 Farmer groups - 2 main types About 20 farms per group of several types Each farm – bean field with 2-6 test plots Planted, managed, measured by farmers About 400 farms and 1600 plots = Large N?
18
Farms as Environments (Contexts) Deviation from straight line = GxE+residual GxE as risk for farmers Residual is simply more GxE
19
G x E G x C G x E Location Mega- environments (Mechanisms) Recommendation domains Maps G x C Location + situation VarietyCharacteristicsAdaptation Choices (Mechanisms)
20
Questions What works well and what needs to change in the way you do METs? What are the complexities and questions you face in designing and implementing METs?
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.