vegdist: warn on using Morisita with 0/1 data, see issue #444

Author: jarioksa

R/vegdist.R

@@ -55,6 +55,9 @@
     if (method %in% c(7, 13, 15) && !isTRUE(all.equal(x, round(x))))
         warning("results may be meaningless with non-integer data in method ",
                 dQuote(inm))
+    if (method %in% 7 && round(max(x)) == 1) # morisita
+        warning("results may be meaningless with largest integer 1 in method ",
+                dQuote(inm))
     d <- .Call(do_vegdist, x, as.integer(method), PACKAGE = "vegan")
     d[d < ZAP] <- 0
     if (any(is.na(d)))

man/vegdist.Rd

@@ -226,7 +226,10 @@
   in the species present at site \eqn{j} that occur with only one
   individual in site \eqn{k} (Chao et al. 2005).
 
-  Morisita index can be used with genuine count data (integers) only. Its
+  Morisita index can be used with genuine count data (integers) only. It
+  is based on the idea of resampling without replacement and should not
+  be used with presence/absence data, but may give meaningless results
+  if compared sampling units (rows) have largest integer 1. Its
   Horn--Morisita variant is able to handle any abundance data.
 
   Mahalanobis distances are Euclidean distances of a matrix where