In davan690/beech-publication-wr: Small mammal dynamics in NZ beech forests

Introduction {#intro}

Worldwide, but particularly on islands, introduced predators have had significant impacts on native populations [@gurevitch2004], and in the worst cases global extinctions [@doherty2017; @towns2011]. On the Island of Guam, the invasion of the Brown Tree Snake (Boiga irregularis) was identified as the primary driver of multiple local extinction events [@fritts1998] and foxes continue to predate heavily on native marsupials throughout Australia, impacting the geo-spatial range of many native marsupial species [@kinnear2016]. In New Zealand, mammalian predators contribute to a disproportionately large group of invaders that impact native populations nationally [@saunders2001] and have destructive impacts on native birds in New Zealand’s (NZ) remaining/remnant native forests. In these systems direct predation is commonly proposed as the reason for native species decline [@fritts1998; @kinnear2016]. These contemporary ecological communities usually contain at least one of the following four mammalian predators [@king2005]; Stoats (Mustela erminea); Brushtail possums (Trichosurus vulpecula); Ship rats (Rattus rattus referred to as rats); House mice (Mus musculus; referred to as mice).

# #data file
# source("./R/ecosystem-simulation/sim-raw-data.R", echo = FALSE)
# libraries needed
# source("./R/r-packages-needed.R", echo = FALSE)
# source("./R/theme_raw_fig3s.r", echo = FALSE)
# source("./R/davidson_2019_theme.r", echo = FALSE)
require(ggthemes)
require(tidyverse)
require(lubridate)

# sorting out fonss
# marh 2019
require(extrafont)
# this just simply reduces long name to "Times" as in code below
windowsFonts(Times = windowsFont("TT Times New Roman"))
# add font to database if needed
# font_import()
# sorting fonts
# windowsFonts()
# loadfonts(device = "win")
# windowsFonts(Times = windowsFont("TT Times New Roman"))

# data function loaded
# this function simulated data
#based on two inputs below
source("./code/01_sim_code_for_data_generation_function.R")

#variable data
no.stoats <- c(25,15,25,15,10,160,200,200,25,15,25,20)
stoats <- c(12,10,12,5,7,75,100,100,10,3,5,4)

#simulated dataset function
sim.dat <- sim_code_for_data_generation(no.stoats = no.stoats, stoats = stoats)

Stoats are regarded as the top predator when they are present in NZ beech forests [@king1983], since their deliberate introduction in the late nineteenth century [@king2017]. Stoat control is commonly undertaken in these systems to protect native species that are vulnerable to mammalian predation [@white2006], in particular, hole-nesting species like mohua (Mohoua ochrocephala; @odonnell2017). However, the primary food source for stoats in NZ forests are rodents [@jones2011], and consequently there is a concern that reducing stoat populations to protect native species may allow rodent populations to increase (e.g @rayner2007). An increase in the number of rats and mice would offset the benefits of stoat control because rodents are known to consume the eggs, chicks and even adult birds [@russell2015; @towns2006; @latham2017], directly compete with native species for food resources such as flowers and seeds [@mcqueen2008] and predate on invertebrates [@ruscoe2012]. In this paper we address the question: does stoat control lead to increased abundance of rodents, particularly the most common rodent, mice, in NZ beech forests?

Studies elsewhere in the world have shown that removing or reducing the abundance of a top predator often leads to an increase in the numbers of predators at lower trophic levels (termed mesopredator release; for a review see @prugh2009), which in turn, can lead to suprising outcomes[@caut2009] and often negative outcomes for native species (for examples see @rayner2007; @robles2002). While mesopredator release has been widely documented elsewhere, it is unclear if stoat control in NZ forests will cause rodent populations to increase in the wild. Rodent populations in NZ forests are predicted to respond strongly to variation in food supply [@choquenot2000; @ruscoe2001; @blackwell2001; @blackwell2003; @ruscoe2005], primarily seed availability (Figure \@ref(fig:figure-one-plot1)).

# plotting function
#plotting dataset (sim.dat)
source("./Code/02_code_for_fig1_function.R", echo = FALSE)

#uses the function above to generate plot
result.plot <- output_for_fig_1_func(sim.dat = sim.dat)
result.plot

# Figure caption
cap1.stes <- c("Expected changes in rodent populations in New Zealand beech forests (bottom panel) in response to changes in seed availability (top panel) during non-mast and mast years (grey blocks). In the bottom panel, differences between the hollow and filled symbols represent the proposed differences between areas with and without stoat control respectively. The arrows labelled A-D represent four of the five proposed outcomes of stoat removal that we tested; A) during non-mast years when little seed is available we expect that mice populations will be similar, B) at the peak of mouse abundance (during winter and spring in mast years) when we would expect larger mouse populations in areas where stoats are absent (hollow symbols), C) during increased seed availability (generally between Summer and Autumn), we expected areas with stoat control to also have larger mouse populations; D) when mouse populations are declining from peak abundance, stoats would additionally reduce mouse abundance through predation in addition to other density dependent processes.")

This is particularly pronounced in beech forests (spp. nothofagus) throughout New Zealand, where, between years beech seed production is highly variable. Little seed is produced in most years (Figure \@ref(fig:figure-one-plot1)a: non-mast years) with occasional years of high seed production (Figure \@ref(fig:figure-one-plot1)a: mast years; grey boxes). Mouse populations are low in the non-mast years, due to low food availability [@choquenot2000; @king1983]. In mast years, when seed becomes abundant, mouse populations can increase quickly following a predictable seasonal cycle. Seed begins to fall and accumulate on the forest floor in late summer (February) allowing mouse populations to increase, with mouse populations typically remaining high through winter (August) and into the following spring (November). Beech seed that is not consumed by mice and other seed predators germinates in spring to early summer, meaning this food resource disappears and mouse populations begin to decline in the subsequent seasons. If the following year is a non-mast year with little seed available, mouse populations fall to low levels.

Previous studies have investigated the likely response of mouse populations to stoat control by modelling the outcome of interactions between stoats, mice and seed availability. @blackwell2001 made four predictions regarding the likely effects of stoat predation on mouse dynamics (see Figure \@ref(fig:figure-one-plot1) arrows) with a subsequent field study [@blackwell2003] concluding that stoat predation should have minimal effects on the population dynamics of mice. The authors identified three different phases in the eruption cycle where stoats could have an effect (Figure \@ref(fig:figure-one-plot1)) and indicated that stoat control had little detectable effect on mouse populations during the peak, decline and low phases of the beech eruption cycle. Subsequent modelling work reached similar conclusions [@tompkins2006; @tompkins2013] but identified that the response of mice to stoat control should depend on interactions with rats. Specifically, [@tompkins2013] concluded that, where rats were present, stoat control alone should allow rats to increase, which would have a suppressive effect on mouse populations through either predation or competition. In contrast, when both stoats and rats were controlled, mouse populations could increase to higher levels than in the absence of control (see Figure \@ref(fig:figure-one-plot1)).

Our aim was to test the predictions outlined above and in both @blackwell2003 and @tompkins2006 using data from our independent, large-scale field study. Specifically, during our study we measured the abundance of rodents on trapping grids over six years in beech forest in two adjacent valleys, one with intensive stoat trapping and one without. In each valley we also manipulated rat densities by including trapping grids where rats were removed and compared these to grids without rat removal. This allowed us to examine if predicted responses of mouse populations to stoat control (see Prediction A-D) was influenced by interactions with rats (Prediction E).

davan690/beech-publication-wr documentation built on March 29, 2020, 11:09 a.m.