Species accumulation curves ("exact") with a single inventoried species: confusion between sites and richness?

[luroy]

Dear all,
I’ve just started using the “specaccum” function with the “exact” method to plot species accumulation curve, so I apologize in advance if my following comment is due to my unfamiliarity with this method.
I want to make species accumulation curves for several forests that I have inventoried with 8 quadrats.
In one of these forests, I recorded only one species, which is present in three of the eight quadrats.
However, when I use the "specaccum" function for this forest as follows:
specaccum(data,method="exact")
I get the following result and error message:

no actual accumulation since only one site provided
Species Accumulation Curve
Accumulation method: exact
Call: specaccum(comm = data, method = "exact") 
Sites     1
Richness  3
sd       NA

The "data" object is a matrix with the quadrats (sites) as rownames and the species names as colnames.

<style> </style>

	SpeciesA	SpeciesB	SpeciesC	SpeciesD	SpeciesE	SpeciesF
X1	0	3	0	0	0	0
X2	0	0	0	0	0	0
X3	0	0	0	0	0	0
X4	0	0	0	0	0	0
X5	0	2	0	0	0	0
X6	0	12	0	0	0	0
X7	0	0	0	0	0	0
X8	0	0	0	0	0	0

Therefore, I don't understand why the error message says that only one site is provided, nor why the richness is three. There seems to be a kind of inversion between sites and richness, while the community matrix seems correct.

Thank you for your attention to this matter,

Léa

[jarioksa]

Confirmed.

There are numerous reproducible ways of getting this:

## only one non-zero species
x <- matrix(0, 1000, 1000) # matrix of zeros
x[c(125, 618), 413] <- 1 # add two occurrences for species 413, still 999998 zeros
specaccum(x)
## take only one species from a data set
specaccum(dune[,2])
## one sampling unit should be OK with the message
specaccum(dune[2,])

I think this happens because of an implicit assumption that a vector (column or row) can only come from a single sampling unit. I don't remember why I assumed so: the function was released in 2003 (vegan 1.6-0) with this assumption which pre-dates using version control systems. Similar assumption was made in diversity indices as well, and it may just been carried over here without much thinking.

I'll see what to do.

[jarioksa]

The branch one-species-accumulation changes the behaviour, and seems to work with data frames. However, this will change the behaviour of input vectors: earlier they were regarded as data for one sampling unit and several species, but now they are regarded as data for one species and several sampling units. This also happens if input data are a matrix which drops dimensions when one row is selected (unless drop = FALSE was used). However, data frames do not drop dimensions and there distinction is maintained between rows and columns.

There is a thin possibility that this breaks some other package that expected the old behaviour. I don't expect so, but better check before merging.

[luroy]

Dear @jarioksa
Thanks for responding so quickly!
I'm sorry, I'm not sure that I understand the implication of your comment. I ran the code of the specaccum R. function you modified and tried again. But I get the same result and error message. I don't know if other implementations to fix the bug are necessary. Let me know if I can help.

[jarioksa]

Try with the reproducible examples in my earlier messages. They work with me. If they don't work with me, you probably did not use the updated function from the github.

[luroy]

@jarioksa, I had accidentally forgotten to detach the vegan package from R., so it was the old function that was running... I'm really sorry for the inconvenience. Thank you very much for your help and for your reactivity and availability !!!

[jarioksa]

Fixed with 6a5f9b3