Ok, we have the covariant derivative in the $k$ direction for a rank one tensors, contravariantly $$u^i{}_{;k}=u^i{,k}+\Gamma^i{}_{sk}u^s,$$ and covariantly $$u_{i;k}=u_{i,k}-\Gamma^s{}_{ik}u_s,$$ which indicate a the tensor varies in direction $k$.
Then for rank two tensor we have the possibilities $$A_{ij}\ ,\ A^i{}_j\ ,\ A^{ij}\ ,\ A_iB_j\ ,\ A^iB_j\ ,\ A^iB^j.$$
Then, for example $$A_{ij;k}=A_{ij,k}+\Gamma^s{}_{ik}A_{sj}+\Gamma^s{}_{jk}A_{is},$$ or $$(A_iB_j)_{;k}=(A_iB_j)_{,k}+\Gamma^s{}_{ik}A_sB_j+\Gamma^s{}_{jk}A_iB_s,$$ tell you how tensor $A_{ij}$ or $A_iB_j%$ vary in direction $k$.