Honing in on your theoretical knowledge of linear algebra, real analysis (proofs of calculus concepts, heavily recommended), and even probabilistic/measure theory (not required) will be more than enough preparation for Statistics Graduate school. Also learning common statistical coding languages such as R and Python are useful, as well as some comp sci classes (objected-oriented programming, data structures, algorithms are what I recommend). Although called a "Statistics" PhD, you won't need too much Statistical knowledge (from undergrad) to succeed in it.
Also order of importance when it comes time to apply:
LoR's (Letter of Recs), GPA, GRE, GRE subject
My biggest regrets are:
-not having more, advanced theoretical knowledge in math
-not taking GRE subject
Actual answer to question:
Elements to Statistical Learning,
Hoff--First Course in Bayesian Statistical Methods,
Introduction to Statistical Learning with Applications in R (easiest read out of them all)