Interpreting output of dredge I ran a GLM for the first time in R and am not sure how to interpret the result. 
This is the input:
global.modelAcar <- lm(Acar ~ logNutrientsc*logNDSc*logNNNc, data = dat)
summary(global.modelAcar)
options(na.action=na.fail)
MAcar <- dredge(global.modelAcar)
MAcar

and this is the first few lines of the result:
Global model call: lm(formula = Acar ~ logNutrientsc * logNDSc * logNNNc, data = dat)
---
Model selection table 
        (Int)     lND      lNN      lgN  ... lND:lNN:lgN df  logLik AICc delta weight
2   2.159e-17 -0.2590                    ...              3 -26.445 59.6  0.00  0.214
3   1.682e-17         -0.25420           ...              3 -26.497 59.7  0.10  0.203
1   7.778e-18                            ...              2 -27.799 59.9  0.36  0.179
5   1.220e-17                  -0.16580  ...              3 -27.256 61.2  1.62  0.095
6   2.249e-17 -0.2295          -0.09269  ...              4 -26.283 61.7  2.17  0.072

Question: I understand that model 2 is the best model and shows lND to have a negative effect on diversity. The second best model shows lNN to have a negative effect. No value means no effect. AIC values show that these model are not very informative. Is this interpretation correct or am I missing something? 
 A: The function MuMIn::dredge simply returns a list of models with every possible combination of predictor variable. As for your results, allow me to disagree with what you said:

I understand that model 2 is the best model and shows lND to have a negative effect on diversity. 

that's partially right, 1ND indeed has a negative effect on diversity, but from the delta (=delta AIC) you cannot distinguish model 2 from 3, 1, and 5 since (using the common thumb rule) they have dAIC < 2 

No value means no effect. 

That's not correct. dredge returns a list with every possible combination of variables, if a variable doesn't have a value, it means it was not included in the model. For example, model 3 only has 1NN, besides intercept obviously.

AIC values show that these model are not very informative. Is this interpretation correct or am I missing something?

Since the first four models have similar support (notice also that their Akaike weight, wich varies from 0 to 1, are not relatively high) I strongly suggest that you use model averaging, take a look at MuMIn::model.avg and also read Chapter 4 of Burnham & Anderson (2002). I hope this are clearly enough, but feel free to ask again
