Which 2-sample univariate t-test to use? X1 is wing length, X2 is tail length for 45 male and 45 female bugs.
Which 2-sample univariate t-test should I use? 
My thought was to use Hotelling's T-square?
But Hotelling's is multi-variate not univariate. Now, I'm not sure...  
Any ideas?
 A: From my point of view, when there are two explanatory variables and both have just two levels, we have the famous two-by-two contingency table. Fisher’s exact test can take such a matrix as its sole argument. Alternatively you can use Pearson’s chi-squared test. 
If your null hypothesis is not the 25:25:25:25 distribution across the four categories (i.e. say it's 9:3:3:1), you'll have to calculate the expected frequencies explicitly.
Then perform the chi-squared test (in R) like that:
chisq.test(observed,p=c(9,3,3,1),rescale.p=TRUE)
# rescale.p is needed because the probabilities do not sum to 1.0

A: While your questions is not clear (which means do you want to compare?) you can consult the wiki: Comparing Means to decide what to do.
A: As others have said, you need to clarify your question. 
However, I'm guessing that you want to determine if wing length or tail length differ between male and female bugs. In this case I would just do a couple of two sample t-tests. So for wing length you would have the following hypothesis:
H_0: wing length does not depend on gender
H_1: wing length differs by gender.

You would have something similar for tail length.
