Let's assign indices to the leaves of the Merkle tree: from left to right we number the leaves $0$, $1$, $\ldots$, $2^n-1$, or 0000, 0001, ..., 1111 in binary.
When we now walk from the root to a leaf, at each node 0 means taking the left branch and 1 means taking the right branch (when viewed from above).
Initially we have an empty tree. If we start adding elements, we need to insert them at some index (= at some leaf).
If we choose the the indices incrementally, we obtain a dense or chronological Merkle tree. There are only leaves on the left, all the leaves on the right are empty. This also means that there is a large empty subtree on the right.
If we choose the indices randomly, we obtain a sparse Merkle tree. Empty and non-empty leaves are randomly distributed across the tree.