-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path01_Intro.qmd
37 lines (21 loc) · 9.24 KB
/
01_Intro.qmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
---
title: "Intro"
format: html
editor: visual
---
## **Modeling sea duck trends in the transboundary waters of the Salish Sea: a multi-scale approach to meet management planning needs.**
Waterbird species that overwinter in the Salish Sea have experienced significant changes in abundance in recent decades, the causes of which remain largely unresolved. Globally significant populations of sea ducks, including several vital sign indicators (Puget Sound Partnership, 2023), are among the species experiencing declines. Long-term monitoring programs have demonstrated their value for assessing broad-scale population trends of these species (e.g., Ethier et al., 2020), however, a seamless approach for generating and updating transboundary trends has yet to be realized, nor has an approach for assessing finer-scale trends at geographic extents relevant to conservation planners been developed.
We propose to close this information gap by applying spatially explicit hierarchical analytical techniques in a Bayesian framework to two parallel citizen science programs in Canada (BCCWS) and the U.S. (PSSS). Collectively, these programs have monitored overwintering waterbirds across 476 sampling locations in the Salish Sea since 2009 (15 years) using a nearly identical protocol. The statistical approach we propose will take advantage of the spatial relationships among count sites, allowing for more robust parameter estimates in places where data are sparse, and enabling trend outputs at finer spatial scales relevant to local management organizations (such as at the extent of marine management zones, Indigenous Marine Management Areas (IMMA) in Canada, and Indigenous Marine Stewardship Areas (IMSA) in the U.S.). Further, by making these results openly accessible on the NatureCounts platform, we will empower resource managers with regionally tailored, regularly updated trends and annual indices of abundance, which will ultimately expedite our understanding of drivers of change and provide baselines from which to assess our management actions.
## 1. Project Goals
The goals of our research is to use standardize, citizen-science monitoring data to (1) obtain scientifically credible measures of abundance trends of waterbirds in the transboundary waters of the Salish Seas at scales appropriate for resource management, (2) identify priority species for conservation, and (3) provide resource managers with openly accessible annual indices of abundance for model-based management planning. In turn, these modelling outputs can be used to assess environmental and human-induced mechanisms of waterbird changes and provide a foundation from which to tease apart whether local population fluctuations are a result of true changing abundance or shifts in species distributions over time (de Zwaan et al., 2024).
Based on previous research, we hypothesized that overwinter abundance trends in waterbirds will be influenced by guild-level characteristics (e.g., diet, migration distance, temperature-tolerance), and that these will vary spatially (Ethier et al., 2020; de Zwaan et al., 2024). Specifically, we predict that abundance declines will be more prevalent in higher tropic-level consumers because this guild is generally more responsive to shifts in the availability and quality of prey (e.g., Ainley & Hyrenbach, 2010; Ballance et al., 1997; Hyrenbach & Veit, 2003; Vilchis et al., 2014). Second, we predict that abundance declines will be more pronounced in migrants because these populations are able to shift their range in response to resource availability more readily than local breeders (e.g., Burger & Gochfeld, 1991; Marks & Redmond, 1994; Willie et al., 2020). Last, we predict negative abundance trends will be greater in cold-tolerant species, because this guild may be tracking climate conditions outside the bounds of the study area (i.e., cold water refugia to the north), compared to warm-tolerant species that may be stable or increasing throughout the study region as populations shift northward from previous core wintering habitats farther south (Hyrenbach & Veit 2003). Spatially, we anticipate that finer-scaled abundance trends to reveal patterns that are more homogenized (e.g., not significant) at national or international extents, even when finer-scale trends show significant spatial variation (e.g., Ethier & Nudds, 2015). We anticipate that some species will display north-south or east-west patterns in trends that are consistent with redistributions in response to climate change, and other species to display a patchwork of trends that are consistent with more localize drivers of change (Ethier et al., 2022).
## 2. Overview of Methods
Our research will deploy cutting-edge statistical approaches to assess transboundary, spatially explicit abundance trends of coastal waterbirds at scales that are appropriate for conservation practitioners. Specifically, the spatial scales and extents (i.e., boundaries) of the analysis will be determined in collaboration with resource practitioners on either side of the international boarder (see NETWROKING for details). Our analysis will use spatially varying coefficient models (SVCs, Gelfand et al. 2003) to account for relationships between variables that are not uniform across large spatial areas. This modelling approach was first applied to continent wide bird abundance data to assess winter bird population trends using discrete aerial units (Meehan et al. 2019) and an intrinsic conditional autoregressive model (iCAR; Besag 1974). The modelling framework was later adapted (Meehan et al. 2024) to incorporate continuous space using a triangulated model mesh and stochastic partial differential equation (SPDE; Lindgren et al. 2022). The benefits of a continuous-space (SPDE) versus a discrete-space (iCAR) models are (1) finer resolution of trends, (2) a better understanding of the range of spatial correlation, and (3) a reduction in boundary effects associated with discrete-space analyses. However, many management units (such as geopolitical boundaries) are divided by discrete polygons, making the iCAR approach appropriate in many instances. We will therefore develop workflows which allows for either an iCAR or SPDE SCV approach to assess and compare estimates of annual relative abundance as well as long-term trends of coastal waterbirds in the Salish Sea.
The basic statistical unit for the analysis will be the mean count of each species on a survey route across months within the species-specific survey window (see Ethier et al. 2020). Our models will describe the maximum counts of each species within this window for either a grid (discreate space) or site (continuous space) encompassing unique combinations of routes k during a year t. The log-linear predictor of the expected count per geographic unit will take the form:
log(µ~it~) = α~i~+τ~i~Year~itk~+κ~k~+y~it~
Parameters include a grid- or site-specific random intercept α~i~ and a grid- or site-specific random slope coefficients for the year effect τ~i~, both with spatial structure (iCAR or SPDE), which allows for information on relative abundance and year effects, respectively, to be shared across neighbours. Year will be scaled such that the maximum year is 0. Differences in relative abundance among routes κ will be modelled with an independent and identically distributed (idd) random effect. To derive an annual index of abundance per spatial unit, we will include a random effect per geographic unit-year y with an idd, and combine these effects with α and τ. Models will be fit using a Bayesian framework with Integrated Nested Laplace Approximation (INLA) using the R-INLA package (Rue and Martino 2009) for R statistical computing (version 4.3.1; R Core Team 2021). Following model analysis and validation, posterior medians and 95% credible intervals (CI) will be computed per geographic stratum for each species and guilds such that our *a priori* hypotheses can be assessed and priority species for conservation identified. Guild specific trends will include an addition random idd effect for species.
All model outputs will be displayed on the NaureCounts web portal and will be accessible through the *naturecounts* R package. The outputs from this analysis will therefore provide resource managers with openly accessible annual indices of abundance for model-based management planning.
## 3. Using this Technical Guide
In this technical guide we detailing the analytical methods used to calculate broad- and fine-scale trends and annual indices of abundance for all species regularly monitored by the BCCWS and PSSS. Specifically, this guide will provide step-by-step instructions on (1) data access, (2) data wrangling and processing, (3) setting up the analysis for various spatial scales (continuous and discrete space), and (4) running the analysis and generating output tables and maps. This guide assumes that you have a basic understanding of R. We recommend that you become familiar with [‘R for Data Science’](http://r4ds.had.co.nz/) by Garrett Grolemund and Hadley Wickham, which covers how to import, visualize, and summarize data in R using the [tidyverse](https://www.tidyverse.org/) collection of R packages.
## 4. Acknowledgement
This project was financially supported by the SeaDoc Society, a program of the Karen C. Drayer Wildlife Health Center, School of Veterinary Medicine, University of California, Davis.