The current issue I’m trying to solve is that of the “wetland category” in our layer. Out of my brood rearing birds, I have:

## [1] 140

Out of those 140 birds, you can see the following have this wetland category in them.

## [1] 81
Landcover id percentage
wetland 213720670_2022 1.6646043
wetland 213720674_2022 2.1272894
wetland 224780346_2022 11.0273304
wetland 213738411_2022 2.0209227
wetland 222780054_2022 2.0856645
wetland 224738015_2022 0.4794291
wetland 222779088_2022 1.2522138
wetland 225744046_2022 20.8747613
wetland 213738409_2022 0.1314237
wetland 225744076_2022 12.9549411
wetland 224739396_2022 1.9801681
wetland 225726229_2022 0.1754763
wetland 225705231_2022 0.4283307
wetland 216750042_2023 13.5714942
wetland 224738039_2023 0.2833411
wetland 224726249_2023 4.4371122
wetland 222758143_2023 57.4687535
wetland 225763224_2023 0.0586349
wetland 225744211_2023 8.4643166
wetland 224780410_2023 0.4959748
wetland 225736074_2023 0.7907082
wetland 225723152_2023 65.6121081
wetland 225771642_2023 0.5538367
wetland 225745134_2023 4.5996199
wetland 222758165_2023 1.3571580
wetland 213720127_2023 3.4083649
wetland 225791116_2023 5.4275627
wetland 224789214_2023 2.8907931
wetland 224780419_2023 0.0032172
wetland 225744678_2023 9.7939710
wetland 224789210_2023 1.0463972
wetland 224713205_2023 0.4733611
wetland 225763222_2023 26.7108745
wetland 225722520_2023 3.1136260
wetland 228718910_2023 0.8481434
wetland 222779129_2023 0.8052768
wetland 224788915_2023 2.5219702
wetland 225760227_2024 0.3642565
wetland 222724405_2024 2.0528304
wetland 213720139_2024 1.3967586
wetland 228775552_2024 0.5515172
wetland 228775608_2024 2.4343851
wetland 228775624_2024 5.8568331
wetland 224788915_2024 3.8308635
wetland 225743661_2024 0.2862123
wetland 228737351_2024 3.2663963
wetland 224779164_2024 0.7223657
wetland 225760459_2024 3.4158630
wetland 224780427_2024 1.0851117
wetland 222779129_2024 0.9164649
wetland 225739761_2024 0.7060638
wetland 228716718_2024 8.4831716
wetland 224728522_2024 0.6800668
wetland 228747555_2024 1.7367816
wetland 228737520_2024 0.8948796
wetland 228775690_2024 3.0123336
wetland 225744418_2024 1.3050207
wetland 225705711_2024 0.0966415
wetland 225763222_2024 49.3962874
wetland 228716711_2024 9.6244425
wetland 224728526_2024 2.6916223
wetland 228701162_2024 0.5589017
wetland 224728525_2024 8.3061379
wetland 225769676_2024 24.3566761
wetland 225722528_2024 6.3803370
wetland 222785799_2024 4.8745776
wetland 228747927_2024 8.9010325
wetland 228701835_2024 0.0380399
wetland 228716656_2024 3.0156050
wetland 228701834_2024 0.0302036
wetland 216789298_2025 0.0203516
wetland 222762439_2025 2.1022082
wetland 225743861_2025 1.3293247
wetland 228746654_2025 0.8830046
wetland 228748219_2025 8.5983151
wetland 211705291_2025 1.3734467
wetland 225769131_2025 7.8143026
wetland 225785391_2025 1.5831683
wetland 224780427_2025 0.8150489
wetland 249745503_2025 8.1989840
wetland 224727830_2025 0.1755579

This isn’t good, as this wetland category could be peatland, marsh, or swamp. To fix this, I first decided to read in the layer that Jake recommended. This layer is quite detailed, with over 300 specific classes (although from the map you can’t tell it’s that many).

I’m going to show you the methdology I’m using with two birds…the first one being a bird that had a wetland category replaced by this layer. That bird is 222779088_2022.

Let’s look at her current map:

You can see that she has about 20% wetland category (dark blue). She also has some forest and swamp.

Here are her current proportions of habitat. She has over 20% in that wetland category.

## # A tibble: 5 × 2
##   Landcover percentage
##   <chr>          <dbl>
## 1 forest         4.06 
## 2 grassland      0.107
## 3 shrubland      0.405
## 4 swamp         71.1  
## 5 wetland       24.4

Now, we can overlay the new habitat layer with only those pixels that were assigned wetland…and see what the new layer would assign them.

Here’s part of the code you should see. As there’s over 300 classes in the new layer, I had to make a quick ruleset. I based this off of words. For my birds, these categories worked and I wasn’t left with any Other values.

  #   grepl("swamp", ClassName, ignore.case = TRUE) ~ "24", # swamp
  #   grepl("marsh", ClassName, ignore.case = TRUE) ~ "23", # marsh
  #   grepl("forest|woodland|tree|pine", ClassName, ignore.case = TRUE) ~ "2", # forest
  #   grepl("shrubland", ClassName, ignore.case = TRUE) ~ "7", # shrub
  #   grepl("grassland|prairie|meadow", ClassName, ignore.case = TRUE) ~ "9", # grassland
  #   grepl("fen|bog", ClassName, ignore.case = TRUE) ~ "22",# peatland
  #   grepl("agriculture", ClassName, ignore.case = TRUE) ~ "15", #cropland
  #   grepl("developed", ClassName, ignore.case = TRUE) ~ "17", # urban
  #   grepl("rock|bluff|beach|cliff", ClassName, ignore.case = TRUE) ~ "13", # barren
  #   grepl("water", ClassName, ignore.case = TRUE) ~ "18", # water
  #   TRUE ~ "Other" # Catches anything that doesn't match the above
  # )) %>% 
  # mutate(row_id = row_number()) %>%
  # select(-ID, -weight)

Here’s the bird’s old proportions, and the new ones, updated with the layer. We can see that forest and swamp increased. The new layer also added some peatland and marsh. The new layer didn’t completely agree that all those wetland pixels should truly be wetland.

## # A tibble: 7 × 3
##   Landcover updated_prop old_prop
##   <chr>            <dbl>    <dbl>
## 1 forest           7.24     4.06 
## 2 grassland        0.113    0.107
## 3 marsh            7.08    NA    
## 4 peatland         3.67    NA    
## 5 shrubland        0.529    0.405
## 6 swamp           81.4     71.1  
## 7 wetland         NA       24.4

What does her updated map look like? You can toggle back and forth the habitat_crop (the original raster), and habitat_final_map, the updated raster with no wetland values to better see what was replaced.

Now, unfortuately, the following layer does not extend past southern Ontario and Quebec. From the above example, we tend to see that the pixels become similar to the pixels around it.

For individuals that still had the wetland category left (about 18), I took the ones with less than 10% of that wetland category, and just make those pixels the value of the pixel closest to them.

Here’s an example with bird: 225722520_2023

## # A tibble: 4 × 2
##   Landcover percentage
##   <chr>          <dbl>
## 1 peatland      69.5  
## 2 swamp          0.863
## 3 water         28.0  
## 4 wetland        1.58

This bird has a bit of the wetland category left. We can graph her and see that she’s in Northern Ontario. Her wetland’s seem to be on the edges of bodies of water.

Since the percentage is quite small, I’m going to take those wetland pixels and replace them with the class of the pixel nearest in the greatest proportion.

So, her percentages used to look like this:

## # A tibble: 4 × 2
##   Landcover percentage
##   <chr>          <dbl>
## 1 peatland      69.5  
## 2 swamp          0.863
## 3 water         28.0  
## 4 wetland        1.58

And now they look like this. It looks like a few of the pixels became water, and a few became peatland.

## # A tibble: 3 × 2
##   Landcover percentage
##   <chr>          <dbl>
## 1 peatland      70.3  
## 2 swamp          0.863
## 3 water         28.9

So, at this point, I’m left with about four birds that have really high percentages of wetland. I’ll graph them all so you can see.

Bird: 222758143_2023, 57.1% wetland

Bird: 225723152_2023, 65.6% wetland

Bird: 225763222_2023, 26.7% wetland

Bird: 225763222_2024, 49.4% wetland

So, as you can see, a lot of these wetlands are just surrounded by forest. I did some previous analysis looking at if swamps, peatland, or marshes are more likely to be surrounded by forest but there wasn’t a clear answer.

We need to figure out what to do with these birds. Do we manually guess?