Added 30-08-2024: it turns out that a forty-years old paper by Murray Bell contains all the ingredients to answer the original question a sequence $\langle B_n:2\le n<\omega\rangle$ of subalgebras of $\mathcal{P}(\omega)/\mathit{fin}$ such that $B_n$ is $\sigma$-$n$-linked but not $\sigma$-$(n+1)$-linked (for all $n$). This can be converted into a sequence as desired in the question. Details in the paper on ArXiV.org.