You might look at the Cochran–Mantel–Haenszel test.  The data are arranged in a 3-dimensional contingency table.

Examples:

http://www.biostathandbook.com/cmh.html

http://rcompanion.org/handbook/H_06.html

Another approach is to use regression models that are appropriate for nominal data or proportion data.

These include log-linear models, logistic regression, and beta regression.

These are particularly useful if you need to add other terms to your model or need a more complex model than can be handled with traditional tests for nominal data.

When using these, be sure to think about what post-hoc testing is available in the software you are using.

In all cases, it is important to understand the uses, limitations, and assumptions of these models. Here are a few simple examples.

Log-linear model:

http://rcompanion.org/handbook/H_08.html

Models for proportion: logistic, beta, and offset Poisson

http://rcompanion.org/handbook/J_02.html

Thanks for the advice! I am doing non-linear regression (using a negative binomial distribution) with my data to explore some questions too. I wholeheartedly agree that you need to understand the post-hocs and which are available with your software. Thankfully, the packages/libraries on R-CRAN are more than helpful and their associated documents detail their limitations well enough. However, just for the time, I'd like to figure out if any of my sites have prevalences that are significantly different than the others. That is, " Ho: prevalence at site A= site B = site C = Site D." I want to be sure that my methods for this are good enough or if I should be looking to compare all of the prevalences (for each month) from each site at once...like how you might do if you were doing a pairwise ANOVA. I know there isn't a wrong or right answer, but there is always a better one! 

I just saw your first post :) That is so very helpful and sorry I had missed it before I posted a response. My internet is wonky at times and not everything always loads. My apologies. 

I would start with a generalized linear model with a binomial family. 

http://www.statmethods.net/advstats/glm.html

Hey, Mackenzie! I am doing a GAM instead of a GLM, as not all the variables respond in a linear nature. I have already started that process and looked into quite a few techniques and have done some modeling already. However, I am writing the descriptive section of my results at the moment, so what I  need at this point is to state which months and sites have greater observations of infested fish and to determine if they were truly significant from others. I have already done these statistical comparisons but started to question if they were, in fact, the most accurate methods. As for the modeling component, my plan was to pair my model with a CART analysis( or something similar) to get an idea of the relationships between infections and the variables I collected data for. I have a feeling I'm making this out to be much harder than it actually is :) 

Sometimes the ideal statistical approach is not obvious or straightforward. At first glance, an ANOVA comes to mind. You could see if some months have a higher proportion of infected fish than others. But you also apparently want to compare sites to each other as well. Mixed models maybe. Regression if you have other possible predictor variables besides location. Rate ratios also come to mind but I'd have think some more about the best approach. There may be a preferred method used in your field which you can apply. . .. see what other studies have reported using in the literature.

I share the most up-to-date statistical methods for parasitologists.

In addition, the authors have a page for statistical analysis: Quantitative Parasitology.

http://www2.univet.hu/qpweb/qp10/index.php

I suggest reading the instructions to import data very well, because only numbers are allowed.

http://www2.univet.hu/qpweb/qp10/QPweb_notes.html

I also attached the article.