I am trying to make a model that reflects the following experimental design (split split design): A tree is divided up into quadrants and in each quadrant, 3 leaves are measured for 20 trees. The models that I have made have really high degrees of freedom for the denominator (i.e: 687.04) and the same degrees of freedom appeared for other interactions, which should not be the case since they are different levels. I am not sure of how to adjust the nested and random factors so that they properly reflect the experimental design. Quadrant is nested within tree and date is nested within quadrant

Injection_lm_a<-lmer(data=CIT_Block, Fv.Fm~Treatment*Date/Quadrant+(1|Tree)+(1|Quadrant))
anova(Injection_lm_a)

Injection_lm_b<-lmer(data=CIT_Block, Fv.Fm~Treatment*Date+(1|Tree)+(1|Quadrant))
anova(Injection_lm_b)

Injection_lm_c<-lmer(data=CIT_Block, Fv.Fm~Treatment+DateTreatmentQuadrant+(1|Treatment/SampleNumber/Cuadrant))
anova(Injection_lm_c)

Injection_lm_d<-lmer(Fv.Fm ~ Treatment*Cuadrant + (1 | Tree/Quadrant) + (1 | Date), data = CIT_Block)
anova(Injection_lm_d)