4.5. Considerations regarding the experimental set-up
This paper presented results from a large screening experiment where 148 genotypes comprising 61 % wild, 7 % feral and 32% cultivated, were subjected to modest drought (restricted-water) and ample-water regimes (see Appendix Table A. 3.). As such, for the feasibility of the experiment, we included maximally four replicates per genotype per treatment because this was the maximum manageable number, allowing for the identification of the largest differences within the material. Damage and mortality of some plants caused variation in the real number of replicates across genotypes (Appendix Table A.8.). Consequently, the mixed-effects model that we applied could not estimate genotype effects very precisely but, rather, it put the genotype effects closer to the mean effect (an effect called shrinkage). It is therefore important to note that in our analyses, individual genotypes acted mostly as a replication at the genotypic level to test cultivation status, genetic group and location effects on the responses.
Despite the close relationship between vegetative growth and yield capacity of coffee plants (Cilas et al. , 2006), one should note that our study focused on responses of comparatively juvenile plants and we did not include effects of ontogenetic changes on responses yet certain ontogenetic changes may affect performance in later life stages. For example, as mentioned above in the discussion about growth-tolerance trade-offs, relatively fast growth in young plants under dry conditions, could be maladaptive later in life as it can result in larger more water demanding phenotype. To assess how drought affects trees over a larger time of their life, more mature trees (of about five years) need to be considered