Morisita index error in vegdist

[amc228]

I am trying to calculate a Morisita index using vegdist, using a species abundance matrix that looks like so:

sp1	sp2	sp3	sp4
0	4	0	1
0	3	2	1
1	3	1	6

When I run distm<-vegdist(df, method='morisita') I get the following error:

"Warning message: In vegdist(x, method = "morisita", na.rm = TRUE) : missing values in results "

The output looks like so:

1	2	3	4	5	6
NaN

summary(distm) says that all values in the distance matrix generated are NAs:

Min. 1st Qu.	Median	Mean 3rd Qu.	Max.	NA's
0	0	0	0	123873

Checks so far:

• all values in the species abundance matrix are integers (as I understand the morisita index requires this)
• there are no 'NAs' or negative values in the species abundance matrix. There are no 'missing values'.
• changing na.rm to FALSE produces the same result
• all other indicies work - it is only morisita that produces this error

Happy to share the real data if anyone is interested - I'm really confused as to what could be causing this error.

[jarioksa]

Yes, this can happen. I have known this since implementing the function in 2001, and I thought I had warned about this in the help page, but there seems to be no warning. I'll add a warning.

If you have (several) species with only one occurrence, the correction term (λ) with x (x-1) becomes zero and you end up with a division with zero, and that's NaN. It is just written in the formula so it is the destiny, and you cannot use Morisita index for such data.

[jarioksa]

A reproducible way of triggering this behaviour is

library(vegan)
data(BCI)
m <- as.matrix(BCI)
vegdist(pmax(m-24, 0), "morisita")

[jarioksa]

I cannot reproduce the problem with your data:

library(vegan)
m <- matrix(c(0,0,1,4,3,3,0,2,1,1,1,6), 3, 4)
m
#     [,1] [,2] [,3] [,4]
#[1,]    0    4    0    1
#[2,]    0    3    2    1
#[3,]    1    3    1    6
vegdist(m, "morisita") ## this will be OK:
#          1         2
#2 0.0000000          
#3 0.2941176 0.1326531

I know Morisita will fail sometimes (just look at its formula: it must fail sometimes), but it works with your example.