bglarkin/efficient_intercepts

Analysis of existing line-point-intercept data to learn whether fewer samples can return similar results

efficient_intercepts

At MPG Ranch, we count and measure plants every year to learn how the plant community is changing. In particular, we are interested in changes that result from our restoration and management of the land, and how these changes affect organisms that use plants for food or shelter.

Every five years, we survey plants across the ranch at 582 locations. At each location, we collect data from 200 points. Surveying 116,400 points requires a significant investment of time and money, and it also leads to a large human presence on the landscape, causing disturbance to wildlife. This investment and disturbance is worth it if need all the data to learn about changes in the plant community, but we've never tested whether we are collecting too much data. Can we answer our questions just as well with less data? The analysis and reporting included here investigate the efficiency of our survey methods and suggest alterations to our protocol.

In this analysis, data from the 2016 ranch wide survey are used. An interactive graphic of the 2016 data can be viewed on the MPG Ranch Livemap.

Our existing survey protocol will likely be revised as a result of this analysis.

This project was started on a Google Colaboratory Notebook, but was moved to speed up processing, allow production of a report in Markdown, and to improve readability for readers who are not familiar with Colaboratory functions. Users who wish to execute functions on the Colaboratory Notebook must have API keys to access data from the Google Cloud.

efficient_intercepts's Issues

Remove mapping tools

We won't be using maps in this report.

efficient_intercepts/raw_code_sandbox.R

Lines 49 to 62 in 78b8692

	# GG map, current version GitHub (slow loading)
	if(!requireNamespace("devtools")) install.packages("devtools")
	devtools::install_github("dkahle/ggmap", ref = "tidyup", force = TRUE)
	# mapping
	library("ggmap")
	# API keys
	# ————————————————————————————————————————
	# API keys are pulled from local resources and are not available in the hosted environment.
	# Users must have API keys for Google Big Query and Google Maps
	# Google Maps API (local file)
	mapKey <- fromJSON(file = "/Users/blarkin/Egnyte/Private/blarkin/ΩMiscellaneous/R_global/R_globalKeys.json")$mapKey
	register_google(key = mapKey)

Typo

Correct spelling of "vegetation"

efficient_intercepts/veg_survey_model.R

Line 22 in 2a8d6e5

#' the vegetaiton data, and the grid points used in the test were chosen to maximize contrasts.

Ground cover resampling code is incomplete

Complete the ground cover resampling code

• The code linked here is a good start, but needs to be modified and finished to continue the analysis
• This is a test as I have not done an issue or branch yet

Tasks

• create branch
• develop code to address the issue
• pull request
• merge branch to main
• close the issue

efficient_intercepts/raw_code_sandbox.R

Lines 268 to 281 in 0e8592a

	#### Ground cover ####
	# ——————————————————————————————————
	grcov_pull_df %>%
	filter(grid_point == 285) %>%
	mutate(detected = 1) %>%
	glimpse() %>%
	group_by(intercept_ground_code) %>%
	summarize(pct = sum(detected) / 2)
	## use sample_n https://dplyr.tidyverse.org/reference/sample.html
	samp_grcov <- grcov_pull_df[sample(nrow(grcov_pull_df), replace = TRUE, 100), ]

Revise loop function, replace with vectorized function

efficient_intercepts/raw_code_sandbox.R

Lines 168 to 186 in 04666e8

	# Create list objects for each grid point and transform objects to species-samples matrices
	for (i in 1:length(grid_points)) {
	# filter spe_mat_df to individual grid points and pivot to a species-samples matrix
	spe_mat_temp_df <-
	data.frame(
	spe_mat_df %>%
	filter(grid_point == grid_points[i]) %>%
	pivot_wider(names_from = key_plant_code, values_from = detected, values_fn = min, values_fill = 0) %>%
	arrange(transect_point) %>%
	select(-NV, -grid_point),
	row.names = 1
	)
	# store filtered data as list object
	spe_mat_list[[i]] <-
	assign(
	paste0("gp_", grid_points[i]),
	spe_mat_temp_df
	)
	}