The formula for the Moran coefficient (MC) is
$$MC = \frac{n}{K}\frac{\sum_i \sum_j w_{ij} (y_i - \overline{y})(y_j - \overline{y})}{\sum_i (y_i - \overline{y})^2}$$
where \(n\) is the number of observations and \(K\) is the sum of all values in the spatial connectivity matrix \(W\), i.e., the sum of all row-sums: \(K = \sum_i \sum_j w_{ij}\).
If any observations with no neighbors are found (i.e. any(Matrix::rowSums(w) == 0)) they will be dropped automatically and a message will print stating how many were dropped. (The alternative would be for those observations to have a spatial lage of zero, but zero is not a neutral value.)