In UofTCoders/eeb430.2017.JuniorCoders: Team Junior Coders Final Report for EEB430 at UofT

library(tidyverse)
library(dplyr)
library(readxl)
library(broom)

Pred <- read_csv("../data/predator.csv.gz",na = c("","n/a","NA"),col_types = cols(`SD PP`=col_double()))
problems(Pred)

colnames(Pred) <- gsub(" ", "_", colnames(Pred))

Pred 

Pred$Latitude <- Pred$Latitude %>% 
    sub("º", "d",.)

library(sp)
Pred$Latitude <- as.numeric(char2dms(Pred$Latitude))

pred_abundance dataset

This dataset is a subset of the marine dataset. What does it include?: - Predator observations that are larger than 800 - Prey observations that are between 50 and 1000 - observations that are above 1000m depth

pred_abundance <- Pred %>% 
    group_by(Predator) %>% 
    tally() %>% 
    arrange(desc(n)) %>% 
    filter(n > 800)
pred_abundance

prey_abundance <- Pred %>% 
   group_by(Prey) %>% 
    tally() %>% 
    arrange(desc(n)) %>% 
    filter(n > 50 & n < 3000)
prey_abundance

pred_prey_dataset <- Pred %>%
    filter(Predator %in% pred_abundance$Predator) %>%
    filter(Prey %in% prey_abundance$Prey) %>%
    filter(Depth < 1000)# %>%
    #filter(Type_of_feeding_interaction == 'piscivorous') #| Type_of_feeding_interaction == 'planktivorous')

pred_prey_dataset

log predator mass vs log prey mass

As prey mass increases so does predator mass. When plotted this way, the data clusters into 4-5 blobs -- it's possible that there might be something underlying this pattern.

pred_prey_dataset %>% 
    ggplot(aes(x = Depth, y = Mean_PP)) +
    geom_point()


pred_prey_dataset %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass))) +
    geom_point() +
    geom_smooth(method = "glm")

Colour by specific habitat

Same plot, but now we colour data based on what habitat the predator-prey interaction is. It appears that habitat predicts the observed pattern... so what is it about habitat that underlies this pattern that as prey size increases so does predator size?

pred_prey_dataset %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Specific_habitat)) +
    geom_point() 

Colour by feeding interaction

Same plot, only we colour the data by the kind of feeding pattern the predator exhibits. It looks like this could also explain the data, although maybe not as strongly as habitat -- you see that predacious feeding spans the extremes, and picsivorous feeding does too. But these are really general feeding types... are feeding types found in more habitats than in others?

pred_prey_dataset %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Type_of_feeding_interaction)) +
    geom_point() 

Predacious data

predacious <- pred_prey_dataset %>% 
    filter(Type_of_feeding_interaction == 'predacious') %>% 
    select(Predator_mass, Prey_mass, Type_of_feeding_interaction, Prey, Specific_habitat) 

predacious %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Prey)) +
    geom_point()

# looking at the predacious data, there's two extremes:
# squid prey seem to make predator prey really big
# on the other hand, copepods and invertebrates appear to make predators really small
# an energy transfer issue? 

predacious %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Specific_habitat)) +
    geom_point() 

# this might just be a function of the ice zone having a lot of observations, but being in the ice zone or nearshore makes you small, while living in the shelf makes you large

Planktivorous data

planktivorous <- pred_prey_dataset %>% 
    filter(Type_of_feeding_interaction == 'planktivorous') %>% 
    select(Predator_mass, Prey_mass, Type_of_feeding_interaction, Prey, Specific_habitat)

planktivorous %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Prey)) +
    geom_point() 

planktivorous %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Specific_habitat)) +
    geom_point()

# all planktivorous species are found in the ice zone, and they're all really small
# this is probably driving that pattern against Bergmann's rule 

Piscivorous data

<<<<<<< HEAD:doc/Pred_farhan.Rmd

piscivorous <- pred_prey_dataset %>% 
    filter(Type_of_feeding_interaction == 'piscivorous') %>% 
    select(Predator_mass, Prey_mass, Type_of_feeding_interaction, Prey, Specific_habitat) 

piscivorous %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Prey)) +
    geom_point()

# You see some grouping here, it's a little more spaced out than what you saw for predacious, but there is definitely some grouping

piscivorous%>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Specific_habitat)) +
    geom_point()

# Specific habitat is also a really good indicator for prey size - predator size

Facet wrapping

pred_prey_dataset %>% 
    ggplot(aes(x = log(Prey_mass), y = log(Predator_mass), colour = Prey)) +
    facet_grid(Specific_habitat ~ Type_of_feeding_interaction) +
    geom_point() +
    geom_smooth(method = 'glm', colour = 'black')

Let's check out prey mass

fit1 <-  glm(Prey_mass ~ Specific_habitat, data = pred_prey_dataset, family = gaussian)
tidy(fit1)

2556d23e52845cb09df46608aef04fca71739f95:doc/Pred.Rmd

UofTCoders/eeb430.2017.JuniorCoders documentation built on May 28, 2019, 3:19 p.m.