library(renv)
renv::load()
renv::restore()
renv::status()Get diet table from fishbase
Summary
In this notebook we
- extract a diet table using fishbase
0. Setup
library(here)
library(DiagrammeR)
library(targets)
library(dplyr)
library(rfishbase)
library(riversea)1. Extract the diet table
First, we want to get the list of species present in our dataset.
data <- tar_read(sea_data_imputed)
species_list <- data$size |>
pull(species_valid) |>
unique()
head(species_list)[1] "Abramis brama" "Acipenser sturio"
[3] "Agonus cataphractus" "Alburnoides bipunctatus"
[5] "Alburnus alburnus" "Alosa alosa" We have 176 species in our data set. We want to retrive their diet for each of them.
We assume the following food web structure.
DiagrammeR::grViz(readLines(here("figures", "foodweb.dot")))Ellipses represent broad basal categories on which fishes feed on. Our goal here is to determine for each fish species on which of these categories it feeds, and whether it is piscivorous or not. To do so we scrap fishbase.
items <- rfishbase::fb_tbl("diet_items")
diet_species <- rfishbase::diet(species_list)
diet_category <- diet_species |>
left_join(items, by = "DietCode") |>
select(
Species,
SampleStage,
DietCode,
FoodI,
FoodII,
FoodIII,
ItemName,
Stage,
DietPercent
)
head(diet_category)# A tibble: 6 × 9
Species SampleStage DietCode FoodI FoodII FoodIII ItemName Stage DietPercent
<chr> <chr> <int> <chr> <chr> <chr> <chr> <chr> <dbl>
1 Abramis … adults 383 zoob… benth… amphip… amphipo… juv.… 16
2 Abramis … adults 383 zoob… benth… benth.… benth. … juv.… 4.76
3 Abramis … adults 383 plan… other… benthi… benthic… n.a.… 0.430
4 Abramis … adults 383 zoob… mollu… bivalv… bivalves juv.… 5.08
5 Abramis … adults 383 detr… detri… debris debris n.a.… 18.7
6 Abramis … adults 383 zoob… insec… insects insects larv… 40 We see that the food is described by the three columns FoodI, FoodII and FoodIII. The question is how we map our categories using those.
diet_category |>
pull(FoodI) |>
unique()[1] "zoobenthos" "plants" "detritus" "zooplankton" "others"
[6] "nekton" Here we see that only the “plants” and “others” category cannot be classified directly. Nekton correspond to other fishes so it implies piscivory.
We can go further in the classification for the ambiguous categories.
diet_category |>
filter(FoodI == "plants") |>
select(FoodII, FoodIII) |>
distinct()# A tibble: 7 × 2
FoodII FoodIII
<chr> <chr>
1 other plants benthic algae/weeds
2 other plants periphyton
3 other plants terrestrial plants
4 phytoplankton n.a./other phytoplankton
5 phytoplankton diatoms
6 phytoplankton green algae
7 phytoplankton dinoflagellates So using these subcategories we can categorize fish diet using our categories
- benthic algae > macrophytes;
- periphyton > biofilm;
- terrestrial plants > macrophytes;
- phytoplankton > phytoplantkon.
diet_category |>
filter(FoodI == "others") |>
select(FoodII, FoodIII) |>
distinct()# A tibble: 4 × 2
FoodII FoodIII
<chr> <chr>
1 others n.a./others
2 birds sea birds
3 birds n.a./other birds
4 herps toads/frogs These correspond to categories we do not consider in our framework, se we choose to ignore them.
Like this we can extract a diet table
diet_table <- tar_read(diet)
head(diet_table)# A tibble: 6 × 3
species stage prey_category
<chr> <chr> <chr>
1 Abramis brama adults detritus
2 Abramis brama adults fish
3 Abramis brama adults insect
4 Abramis brama adults crustacean
5 Abramis brama adults worm
6 Abramis brama adults zooplankton which for each species, and life stage, tells what it eats. We can also have the table a in “wide” format, where we have one column for each basal category.
diet_table_wide <- tar_read(diet_wide)
head(diet_table_wide)# A tibble: 6 × 13
species stage biofilm crustacean detritus echinoderm fish insect macrophyte
<chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 Abramis … recr… 0 1 0 0 0 1 0
2 Abramis … juv.… 0 1 1 0 0 1 1
3 Abramis … adul… 0 1 1 0 1 1 0
4 Acipense… recr… 0 1 0 0 1 1 0
5 Acipense… juv.… 0 1 0 0 1 0 0
6 Agonus c… adul… 0 1 0 0 0 0 0
# ℹ 4 more variables: mollusk <dbl>, phytoplankton <dbl>, worm <dbl>,
# zooplankton <dbl>2. Postprocess diet table
2.1. Add larvae stage
Next, we want to add a larvae stage for all species.
d_larvae <- diet_table_wide |>
group_by(species) |>
summarise(has_larvae = "larvae" %in% stage, .groups = "drop")
d_larvae |> count(has_larvae)# A tibble: 2 × 2
has_larvae n
<lgl> <int>
1 FALSE 157
2 TRUE 19So we see that for most species, they do not have a larvae stage. We can add them with a utility function. We assume that larvae only feed on zooplankton.
diet_with_larvae <- add_larvae(diet_table_wide)
diet_with_larvae |> filter(species == sample(species, 1))# A tibble: 2 × 13
species stage biofilm crustacean detritus echinoderm fish insect macrophyte
<chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 Barbatul… larv… 0 0 0 0 0 0 0
2 Barbatul… juv.… 0 1 0 0 0 1 0
# ℹ 4 more variables: mollusk <dbl>, phytoplankton <dbl>, worm <dbl>,
# zooplankton <dbl>2.2. Remove rare species
To focus on important species for the community we study we filter for very rare species. Let’s have a look.
sea_data <- tar_read(sea_data_imputed)
occurences <- sea_data$catch |>
group_by(species_valid) |>
summarise(occurence = n(), .groups = "drop") |>
arrange(occurence)occ_min <- 10
count_occ <- occurences |> count(occurence >= occ_min)
n_remaining <- pull(count_occ, n)[2]
count_occ# A tibble: 2 × 2
`occurence >= occ_min` n
<lgl> <int>
1 FALSE 69
2 TRUE 107We see that we have many species have a low occurence. For example, if you choose a threshold of 10, we have only 107 species that remain in the data set.
We created an utility function to remove rare species from the diet table.
diet_no_rare <- tar_read(diet_no_rare)
head(diet_no_rare)# A tibble: 6 × 13
species stage biofilm crustacean detritus echinoderm fish insect macrophyte
<chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 Abramis … recr… 0 1 0 0 0 1 0
2 Abramis … larv… 0 0 0 0 0 0 0
3 Abramis … juv.… 0 1 1 0 0 1 1
4 Abramis … adul… 0 1 1 0 1 1 0
5 Agonus c… larv… 0 0 0 0 0 0 0
6 Agonus c… adul… 0 1 0 0 0 0 0
# ℹ 4 more variables: mollusk <dbl>, phytoplankton <dbl>, worm <dbl>,
# zooplankton <dbl>We can check if we have species with missing diet.
diet_no_rare |>
filter(is.na(stage)) |>
pull(species)[1] "Engraulis maeoticus" "Raja microocellata" "Symphodus bailloni" 3. Life stage lengths
library(rfishbase)
species_list <- diet_no_rare |>
pull(species) |>
unique()
d_maturity <- maturity(species_list) |>
select(Species, Lm) |>
filter(!is.na(Lm)) |>
group_by(Species) |>
summarise(
Lm_avg = mean(Lm),
.groups = "drop"
)
head(d_maturity)# A tibble: 6 × 2
Species Lm_avg
<chr> <dbl>
1 Abramis brama 28.8
2 Alburnus alburnus 9.90
3 Alosa alosa 42.5
4 Alosa fallax 28.9
5 Ammodytes tobianus 13
6 Anguilla anguilla 50.8 species_missing <- setdiff(species_list, d_maturity$Species)
maturity("Barbus barbus")# A tibble: 4 × 41
Species SpecCode autoctr StockCode MaturityRefNo Sex AgeMatMin AgeMatMin2
<chr> <int> <int> <int> <int> <chr> <dbl> <dbl>
1 Barbus ba… 4472 2212 4670 6258 fema… 4 5
2 Barbus ba… 4472 2213 4670 6258 male 3 4
3 Barbus ba… 4472 7184 4670 59043 fema… 3 7
4 Barbus ba… 4472 5919 4670 59043 male 2 5
# ℹ 33 more variables: AgeMatRef <int>, tm <dbl>, Number <int>, r2 <dbl>,
# SE_tm <dbl>, SD_tm <dbl>, LCL_tm <dbl>, UCL_tm <dbl>, LengthMatMin <dbl>,
# LengthMatMin2 <dbl>, Type1 <chr>, LengthMatRef <int>, Lm <dbl>,
# LCL_Lm <dbl>, UCL_Lm <dbl>, LmaxLm <dbl>, LmaxLmType <chr>,
# LmaxLmRef <int>, LmaxStudy <dbl>, LmaxStudyType <chr>, SE_Lm <dbl>,
# SD_Lm <dbl>, Locality <chr>, C_Code <chr>, E_CODE <int>, Comment <chr>,
# Entered <int>, DateEntered <dttm>, Modified <int>, DateModified <dttm>, …size_extrema <- tar_read(size_extrema) |> rename(species = species_valid)
maturity_length <- tar_read(maturity_length)
sizes <- inner_join(maturity_length, size_extrema, by = "species")
head(sizes)# A tibble: 6 × 4
species maturity_length length_min length_max
<chr> <dbl> <dbl> <dbl>
1 Abramis brama 28.8 1.8 56
2 Alburnus alburnus 9.90 1.2 15.6
3 Alosa alosa 42.5 4.2 36.5
4 Alosa fallax 28.9 2.8 40
5 Ammodytes tobianus 13 3.2 27
6 Anguilla anguilla 50.8 2.84 96 sizes |> filter(maturity_length > length_max | maturity_length < length_min)# A tibble: 8 × 4
species maturity_length length_min length_max
<chr> <dbl> <dbl> <dbl>
1 Alosa alosa 42.5 4.2 36.5
2 Chelidonichthys cuculus 26.1 7 23
3 Clupea harengus 22.5 2.7 22.5
4 Diplodus sargus 18.3 3.5 13.9
5 Lophius piscatorius 54.9 13 27
6 Pollachius pollachius 39.4 3.8 35
7 Raja microocellata 71.7 13 71
8 Sparus aurata 30.8 2.5 26.5Maybe only juveniles are captured?
diet_full <- tar_read(diet_size)
head(diet_full)# A tibble: 6 × 15
species length_min length_max stage biofilm crustacean detritus echinoderm
<chr> <dbl> <dbl> <chr> <dbl> <dbl> <dbl> <dbl>
1 Abramis br… 0 2.00 larv… 0 0 0 0
2 Abramis br… 2.00 2.00 juv.… 0 1 1 0
3 Abramis br… 2.00 Inf adul… 0 1 1 0
4 Agonus cat… 0 2.00 larv… 0 0 0 0
5 Agonus cat… 2.00 Inf comb… 0 1 0 0
6 Alburnus a… 0 2.00 larv… 0 0 0 0
# ℹ 7 more variables: fish <dbl>, insect <dbl>, macrophyte <dbl>,
# mollusk <dbl>, phytoplankton <dbl>, worm <dbl>, zooplankton <dbl>