It sounds like you have a few different questions. For your title question, i.e., was occupancy (present/absence accounting for detectability) the same across nesting beaches and can occasion level covariates explain any variation?

Have a look at these digestible sources for help:

Your second question ask more detail about nesting success specifically. I'm sure there are a number of ways to measure and nuance this, but the paper that Safa links above seems like a fair place to start.

Take some time to refine your questions and explore how you might go about answering those questions.

Best,

Caleb

P.S. Be sure to understand the assumptions of the statistical tests you decide to use and if your data can/do reasonably meet these assumptions.

If you have multiple locations and years then I think you might want to look at multilevel generalised linear models (also known as generalised linear mixed models).

Observations from the same location in different years will be correlated so you can get spurious inferences by ignoring that structure. The particular model depends on the outcome you are interested - for example a count variable would lead to something like a multilevel Poisson model and binary data a multilevel logistic regression.

If you are new to these models it is worth consulting with someone locally. They have a strong methods group in psychology at ASU who use these methods if there is no one in your area.

There are a number of factors that might influence nesting success. How do you measure nesting success? and how much within beach variability do you have? Then factors such as beach location, beach material, beach dimension, weather during nesting, predation, etc. How do you quantify protected beaches (protected from what), human population (is this beach use by people?). Then there may be interactions. Heavily utilized beaches my deter predation. And in what way are the beaches used? And do these factors tend to be constant over time at each beach or do they vary? Some factors are probably similar from year to year, others such as weather may not. For factors that are stable over time, you are then faced with multiple measures so sample size through time is reduced. Necessary sample size will depend upon the size of effects.

Determine correlations between factors and select factors based on likely cause. Rule out factors that previous studies found to have small effects but include them in your discussion and bibliography. You should look at repeated measures ANOVA once your data have meaningful structure. Variation of the dependent variable reduces the effectiveness of least squares regression

This is an interesting study. Best wishes and good luck.

I would use some Mixed effect model that control for site-to-site variability. In R, lme4 is one package you might try. Despite a straight forward idea, they are not very simple tho. I'd suggest doing some reading to make sure you specify appropriate effects.