Help on simple comparison of means I have two groups (control and experimental). I want to test whether increased irrigation will affect clutch size of lizards. In order to do this, I need to have separate plots within the irrigated areas and within the control areas, in order to fence off the lizards. I am hoping to do a power analysis and determine how large my sample size needs to be. For instance, I could fence off four individuals in five different irrigated areas and fence off four individual in five different control areas....
(4 individuals in 5 plots for irrigated and 4 individuals in 5 plots for control)=40 individuals
My first question is, what type of analysis would I carry out? ANOVA? Or is this a nested design? Can I compare the mean of all lizards in the irrigated vs control? And also compare within irrigated and within control? 
My second question is, once I know which test, how do I determine sample size needed to see a true effect?
I'm probably over thinking this! :) 
 A: *

*This feels like an ANOVA to me. A single factor ANOVA. You could do
two-way ANOVA, but I'm not sure what your other factor is. Treatment
is a factor. But location could be the other? i.e. the 10 blocks you
are dividing up the area into. Then you would have to be able to
randomly assign treatment within the blocks that you divide.

*You could just do 20 irrigated and 20 control, but not sure that's
what you want. If not this is like a randomized block design.

*As of right now you do not have a nested design. If you imagine your
setup as just 20 irrigated and 20 control, and you took the other
factor (besides treatment) to be the individual lizards, then a
nested design would be if you then asked...Does clutch size increase
between male and female lizards consistently differ? Then you could
have, $$E[Y] = SEX + SEX/LIZARD + TREATMENT$$

*Yes you can compare the individual means. Given the one-way ANOVA
setting, if we can reject $H_0$ then we know that at least one of the
factor level means is different. Ideally, we’d like to know which of
the factor level means is different and in what way they are
different. For this you can do pairwise comparisons with confidence
intervals.

*In a two-way ANOVA setting you have factors A and B, and then you have the factor levels, a and b. So we have ab total treatments under study and if all of these are included in the study we have a complete factorial study where a two-way anova with 5 and 4 levels for the two factors gives 5*4 = 20 total levels and we need a minimum of 20 observations.


I may come back and edit this more but I'd like to hear from some other people too.
