Construction of linear mixed model (using R) I would like to use Lineal Mixed model to see if the treatments I applied to some soil changed significantly their CO2 fluxes.
I have 2 temperature (t1, t2) and 3 inundation (w0,w1,w2), resulting in 6 treatments (table 1). For each of my treatment I have ten replicates. To increase my number of replicates, I sampled 5 times each of my pot (Repeated measurements with Date of sampling). Here is my experimental design
I have used Lineal Mixed model to answer my question, but my model seems to not meet the assumption of no collinearity - I have strips lines (see last figure). Is it due to the construction of my model? or should I use another test? -I have repeated measurements).
Here are some more explanation about how I constructed my model and checked for assumption:
How I constructed my Lineal Mixed model:
1. I load my dataframe , where PID reflect the repeated measurement. I enter PID, Temperature and Flooding  as factorial variables and Fluxes has numerical

2. I construct my model with temperature and Flooding as fixed effect, and as random error, I put temperature and flooding (because they are crossed) and PID, because I have repeated measurement. My code looks like:
TemperatureFlooding.model = lmer (Fluxes ~ Temperature*Flooding + (1|PID) + (1|Temperature) + (1|Flooding), data = BETADecompositioncsv )

My results are: 

3. likelihood ration test: I want to see if  inundation, temperature and temperature*inundation have a significant influence on CO2 emission. I use the likelihood ratio test for that. As illustration my code looks like that:
Temperature.modelNULL = lmer (Fluxes ~ Temperature + (1|PID) + (1|Temperature) + (1|Flooding),data = BETADecompositioncsv, REML=FALSE)

TemperatureFlooding.modelFULL = lmer (Fluxes ~ Temperature*Flooding + (1|PID) + (1|Temperature) + (1|Flooding),
                        data = BETADecompositioncsv, REML=FALSE)

anova(Flooding.modelNULL,TemperatureFlooding.modelFULL)

--> For all I found a p value <0.05
3. I check the first assumption of the model: collinearity
For that I do a residual plot with the following code:
plot(fitted(TemperatureFlooding.model),residuals(TemperatureFlooding.model))

and I get that: 

--> my model seems to not meet the assumption of no collinearity - I have strips lines (see top figure). Is it due to the construction of my model? or should I use another test? -I have repeated measurements.

 A: First of all, this is not a generalized linear mixed model. It is simply a linear mixed model.
Second, notice from your output that your model has a singular fit. Often this indicates that the model is over-fitted with respect to the random effects. In other words the data do not support the random effects structure that you have specified. You are fitting random intercepts for Temperature and flooding, yet you have only 2 and 3 levels of these. The software assumes that the random effects are normally distributed, so it is trying to estimate variances for two normally distributed variables where there is only 2 and 3 observations for them. You should not do this.
You are also specifying both of these variables as fixed effects. They should be one, or the other, not both. We have already established that there are insufficient levels of each of them to warrant fitting them as random effects, and you also appear to be interested in the fixed effects of each, and their interaction. It is hardly surprising that these data do not support such a model. A good place to start would be with this model:
lmer(Fluxes ~ Temperature*Flooding + (1|PID) ...)

If this produces sensible output, then, subject to your a priori hypothesis, you may wish to allow the fixed effects for Temperature, Flooding and their interaction to vary by PID by estimating random slopes.
Regarding collinearity, unless the variables are perfectly correlated, or almost so, this will not be a problem. Some correlation between covariates is quite normal, and to be expected. Having said that, as per @whuber's comments, your plot has nothing to do with collinearity.
