diff --git a/R/bias-helpers.R b/R/bias-helpers.R
index 585abfe..4f5c587 100644
--- a/R/bias-helpers.R
+++ b/R/bias-helpers.R
@@ -21,6 +21,13 @@ NULL
 #' @describeIn cooccurrence Adjacency matrix of the co-occurrence network
 #' @export
 cooccurrence_matrix <- function(mat, all = TRUE) {
+    # To compute the number of times each pair of taxa cooccur, we leverage the
+    # `compute_ratios()` function. Doing so requires that the samples (rows of
+    # `mat`) and the taxa (cols of `mat`) have names.
+    if (is.null(rownames(mat)))
+        rownames(mat) <- paste0("S", seq(nrow(mat)))
+    if (is.null(colnames(mat)))
+        colnames(mat) <- paste0("T", seq(ncol(mat)))
     ratios <- mat %>%
         tibble::as_tibble(rownames = "Sample") %>%
         tidyr::gather("Taxon", "Value", -Sample) %>%
diff --git a/tests/testthat/test-center.R b/tests/testthat/test-center.R
index b125d5d..0f40e9d 100644
--- a/tests/testthat/test-center.R
+++ b/tests/testthat/test-center.R
@@ -117,6 +117,14 @@ test_that("sampling rows or using `weights` gives an equal estimate", {
         center(d2, weights = w.rows, method = "rss"))
 })
 
+test_that("the center is correctly computed when sample and taxa names are missing", {
+    d2.no_names <- d2
+    rownames(d2.no_names) <- colnames(d2.no_names) <- NULL
+    c1 <- center(d2)
+    names(c1) <- NULL
+    c2 <- center(d2.no_names)
+    expect_equal(c1, c2)
+})
 
 # Additional tests to add
 # - Check various in scale and out scale combinations