This post is a running log of miscellaneous coding and TidyTuesday exercises!

13 February 2023

We are doing a TidyTuesday exercise as a group activity for the Roberts Lab meeting today. We are analyzing the FeederWatch dataset with the goal that each person makes 1-2 fun plots.

I was interested in 1) how bird abundance in Washington State varies by vegetation type and 2) how bird abundance responds to pet (dog and cat) abndance.

Code

Loading data for Tidy Tuesday lab meeting practice.

Setup

Load packages.

library(tidyverse)
library(RColorBrewer)

Load data frame for FeederWatch

# Read in the data manually

feederwatch <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2023/2023-01-10/PFW_2021_public.csv')
site_data <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2023/2023-01-10/PFW_count_site_data_public_2021.csv')

Examine dataset

Look at feederwatch dataset.

str(feederwatch)

Look at site dataset.

str(site_data)

How does bird abundance relate to habitat type in Washington State?

Plot: Distribution plot for bird abundance per unit effort for each habitat type

How does bird abundance relate to pet abundance?

Plot: Correlation plot/regression plot of total birds sighted across abundance of cats and dogs

Preparing data

Next, prepare the data so that I can answer the above questions.

We will need one dataframe that has site descriptions and bird counts for each site at each time point.

Manipulate feeder bird count data

Calculate total birds sighted at each location over all time points normalized to the effort of observation (total birds normalized = birds / effort).

feeder_df<-feederwatch %>%
  select(loc_id, Month, Day, Year, subnational1_code, how_many, effort_hrs_atleast)%>%
  arrange(loc_id)%>%
  mutate(norm_count=how_many/effort_hrs_atleast)%>%
  group_by(loc_id, subnational1_code)%>%
  summarise(total_birds_norm=sum(how_many))

head(feeder_df)

Now manipulate the site data to be able to pull out cat and dog presence and types of trees.

site_df<-site_data%>%
  select(loc_id, evgr_trees_atleast, dcid_trees_atleast, fru_trees_atleast, cacti_atleast, cats, dogs)%>%
  gather(key="vegetation_type", value="tree_abundance", -loc_id, -cats, -dogs)%>% #gather 
  group_by(loc_id)%>%
  mutate(mean_cats=mean(cats, na.rm=TRUE))%>% #calculate mean dogs and cat abundance 
  group_by(loc_id)%>%
  mutate(mean_dogs=mean(dogs, na.rm=TRUE))%>%
  select(!cats)%>%
  select(!dogs)%>%
  unique()

Note that some sites have multiple habitat types. We will keep these for now.

Now merge the two data frames together and subset by WA.

df<-left_join(feeder_df, site_df)

df<-df%>%
  rename("cats"="mean_cats", "dogs"="mean_dogs")%>%
  gather(key="animal", value="animal_abundance", cats:dogs)%>%
  mutate(vegetation_type=if_else(vegetation_type=="dcid_trees_atleast", "deciduous", 
                                 if_else(vegetation_type=="evgr_trees_atleast", "evergreen",
                                         if_else(vegetation_type=="fru_trees_atleast", "fruit", "NA"))))


df_WA<-df%>%
  filter(subnational1_code=="US-WA")

Plots

Plot effect of tree type and abundance on bird sightings.

plot1<-df_WA%>%
  filter(!vegetation_type=="NA")%>%
  filter(!vegetation_type=="cacti_atleast")%>%
  
  ggplot(aes(x=tree_abundance, y=total_birds_norm, group=vegetation_type, colour=vegetation_type))+
  #facet_wrap(~vegetation_type, ncol=3)+
  geom_point(aes(group=vegetation_type), alpha=0.2, position=position_jitterdodge(0.2))+
  stat_smooth(method="lm", show.legend=FALSE)+
  scale_color_brewer(palette="Dark2")+
  scale_fill_brewer(palette="Dark2")+
  theme_classic()+
  ylab("Birds (normalized to effort)")+
  xlab("Tree Abundance")+
  ggtitle("More birds with more trees, regardless of tree type!")+
  theme(
    legend.position="right",
    legend.title=element_blank()
  );plot1

ggsave("trees_birds.png", plot1, width=6, height=6)

Plot effects of pet abundance on bird sightings.

plot2<-df%>%
  ggplot(aes(x=animal_abundance, y=total_birds_norm, group=animal, colour=animal))+
  geom_smooth(aes(fill=animal))+
  scale_color_brewer(palette="Dark2")+
  scale_fill_brewer(palette="Dark2")+
  theme_classic()+
  ylab("Birds (normalized to effort)")+
  xlab("Animal Presence")+
  ggtitle("Birds are scarce when pets are sometimes present!")+
  theme(
    legend.position="right",
    legend.title=element_blank()
  );plot2 

ggsave("pets_birds.png", plot2, width=6, height=6)

The full script can be found on GitHub here

Plot Results

Bird abundance increases with increased vegetation, but there is no difference by type of trees.

Bird abundance is lowest when pets are sometimes present.