Is there a way to perform quadratcount() in spatstat without ignoring the marks?

Question

Imagine I have a point pattern distribution in a square and all points have marks (ie plant trait data). I want to divide this square in fe 4 quadrats. Now I am interested in the spatial distribution of the traits, ie I need to now which trait occurs in which quadrat. Only counting the number of points per quadrats does not help in this case, i need to know the points + marks. However, quadratcount ignores the marks. The split-function does not really help as one would need to combine the results per quadrat afterwards, that is not really practical.

# let's try the ant-dataset and count ants in 2x2 quadrats

quadratcount(ants, 2,2)

# raw number of points
           x
y           [-25,389) [389,803]
  [334,717]        22        20
  [-49,334)        27        28

# let's count the abundance of different groups of marks per quadrat

quadratcount(split(ants), 2,2)

Cataglyphis:
           x
y           [-25,389) [389,803]
  [334,717]         5         3
  [-49,334)         9        12

Messor:
           x
y           [-25,389) [389,803]
  [334,717]        17        17
  [-49,334)        18        16

[UPDATE]

I would need a dafaframe with the coordinates corresponding to each mark, grouped by quadrats. With extracted x/y values including the marks. I can do it manually, but it is not very elegant. It might look like this:

$q1[(-25, 389), -49, 334)]
    x   y marks
    x   y          V3
29 275 325      Messor
36   2 267      Messor
39 124 293      Messor
40 197 207      Messor
41 218 297      Messor
42 236 241      Messor
43 298 271      Messor
44 345 239      Messor
45 388 287      Messor
50 289 188      Messor
51 331 145      Messor
66 236 205      Messor
76 326 303 Cataglyphis
80  -7 233 Cataglyphis
83 233 256 Cataglyphis
84 364 239 Cataglyphis
89 324 101 Cataglyphis

$q2[(-25, 389), (334,717)] 
x   y   marks
...

$q3[(389,803), (334,717)]
x   y   marks
...

$q4[(389,803), (-49,334)]
x   y   marks
...

Wondering if someone understand my problem, looking forward to any ideas!

Answer 1

It's unclear exactly what you want. The quadratcount function counts the number of points falling in each sub-region. If you want to retain the actual point coordinates, then quadratcount is not the appropriate function. You can use methods for split and cut to sub-divide the point pattern while retaining the coordinates.

I'll make two guesses.

1. It seems you probably want something like this (demonstrated for the ants data):

B <- quadrats(ants, 2, 2)
Y <- split(ants, B)
Z <- lapply(Y, as.data.frame)

where B is a tessellation object (class tess ) specifying the four tiles; Y is a list containing the four subsets of the point pattern ants within these four tiles; and Z is a list of data frames with columns x, y, marks .

2. Responding to your comment that

The split-function does not really help as one would need to combine the results per quadrat afterwards, that is not really practical.

If you just want to know, for each tile, the number of points of each type that fall in the tile, you could do

simplify2array(quadratcount(split(ants, 2, 2)))

which produces a 3D array containing these numbers.