ข้อมูลจากบทคัดย่อ
ยังไม่มีสรุปภาษาไทยสำหรับระเบียนนี้
ด้านล่างเป็นบทคัดย่อต้นฉบับภาษาอังกฤษจากข้อมูลบรรณานุกรม โปรดตรวจบทความต้นฉบับก่อนอ้างอิง
================================================================================READMECode repository accompanying the article:"Biogeographic constraints structure ecological interaction outcomes acrossisland ecosystems"================================================================================ 1. TITLE--------------------------------------------------------------------------------Biogeographic constraints structure ecological interaction outcomes acrossisland ecosystems Short title: Biogeography shapes ecological interactions on islands 2. AUTHORS--------------------------------------------------------------------------------* authors for correspondence. Mauro Galetti * Andrés F. Ramírez-Mejía * Daiane C. Carreira Fernando Gonçalves Mayara Guimarães Beltrão Jeferson Vizentin-Bugoni Haruko Ando Seheno Andriantsaralaza Cláudia Baider Julia Ball Jedediah F. Brodie Rafael Silveira Bueno María Guerrero Campos Juliana Cordeiro Bo Dalsgaard Sara Saging Isabel Donoso Donald R. Drake Nik Fadzly N. Rosely Patricia dos Santos Ferreira Noemí Matías-Ferrer F. B. Vincent Florens Francisco E. Fontúrbel Sara M. E. Gabrielson Juan Martínez Gómez Lucas Gonçalves da Silva Patricia Guedes Christopher Kaiser-Bunbury Seokmin Kim Laís Lautenschlager Shawn J. Leroux Alba Costa Hannah Madden Lucas P. Martins Kim R. McConkey Félix M. Medina Jayasilan Mohd-Azlan Raquel Muñoz-Gallego Eva K. Nóbrega Manuel Nogales Renske E. Onstein José Ignacio Orellana Ana Filipa Palmeirim Marco A. Pizo Roberto Resendes Gonzalo Rivas-Torres Ricardo Rocha Katerina Sam Chanpen Saralamba Elise Sivault Tuanjit Sritongchuay Fabio Tarazona-Tubens Maria da Anunciação Ventura Maximilian Vollstädt Kenta Watanabe Xue Xiao Pedro Jordano Anna Traveset Correspondence: mauro.galetti@unesp.br and andresfeliper.mejia@gmail.com 3. JOURNAL OF PUBLICATION--------------------------------------------------------------------------------Science Advances Article DOI: [DOI WHEN AVAILABLE]Zenodo DOI (this repo): [10.5281/zenodo.23848102]Raw data (Dryad): [10.5061/dryad.8931zcs5x] 4. BRIEF INTRODUCTION------------------------------------------------------------------------------- Ecological interactions regulate key ecosystem processes, yet their large-scale drivers remain poorly understood. Although biogeography explains patterns of species richness, its ability to predict the outcomes of ecological interactions has rarely been tested. This study used standardized experiments with artificial (plasticine) fruits deployed on 43 island systems worldwide (27 countries) to identify the drivers of attacks on fruit models. The standardized assay provides a global baseline for testing how biogeography and environmental gradients shape ecological interactions on islands, and for anticipating the effects of defaunation, biological invasions and environmental change. This repository contains the code, Stan models and fitted Bayesian models used to wrangle the data, validate the statistical models through a generative simulation, fit the hierarchical Bayesian models, and produce all results reported in the article and in its Supporting Information (SI). Software: R 4.5.1; Stan (via the R package 'cmdstanr' 0.5.3). The full list of R packages and versions is printed at the end of supporting_information_S3.pdf (section S16, 'Computational environment'). 5. FILE-BY-FILE EXPLANATION-------------------------------------------------------------------------------- 5.0 Suggested order of execution-------------------------------- 1. supporting_information_S2.R -> cleans the raw data and creates the two .rds data files (data_structure_simulation.rds and data.rds). 2. supporting_information_S1.R -> generative simulation (model validation). 3. supporting_information_S3.R -> formal analysis (fits the models or loads the fitted .rds objects, then produces all results, tables and figures). The .qmd files contain the same code as the .R scripts, interleaved with text,equations and figures, and render to the SI documents. NOTE: the fitted-model chunks in the .R / .qmd files are set to eval = FALSE(models are slow to sample) and the fitted objects are instead read from the.rds files provided here. To refit a model, run the corresponding'cmdstan_model()' / '$sample()' chunk against the matching .stan file. Files NOT included in this repository but required by the scripts: - ilhas2025.xlsx (raw experimental data; deposited in Dryad, DOI 10.5061/dryad.8931zcs5x). 5.1 R scripts (.R) - analysis code----------------------------------These scripts are the code-only versions of the three Quarto documents (thecode chunks extracted from the .qmd files). supporting_information_S1.R Generative simulation and model validation. Simulates a synthetic dataset that reproduces the structure of the real data (43 islands, grids, plants, countries, island types and the distances among islands), using a simplified version of the causal model (DAG). Parameters are drawn for plant, grid, country and island effects (the latter with a Gaussian-process spatial correlation among islands), and for the effects of island type, altitude, isolation, native vegetation and bush cover on fruit removal. Fruit removal is simulated with a binomial distribution (15 fruits per bush). Five Stan models are then fitted to the simulated data (one per causal effect: altitude, island type, island isolation, bush cover and native vegetation) to confirm that each model recovers the "true" parameter values. Requires data_structure_simulation.rds and the helper functions_mod_diagnostics.r. supporting_information_S2.R Data wrangling. Reads the raw spreadsheet (ilhas2025.xlsx), corrects labels and minor errors (e.g., fruits assigned to the wrong plant), summarizes the proportion of fruits attacked by each consumer type, lists the realms, countries and ecoregions sampled, and assigns numeric indices to the categorical variables. supporting_information_S3.R Formal analysis. Loads data.rds, standardizes predictors, prepares the data list for Stan, and, for each of the seven causal questions (latitude, island isolation, island size, sampling elevation, human footprint, native vegetation, bush cover), fits (or loads) five models: total frugivory, bird frugivory, lizard frugivory, putative seed dispersal (bird + lizard) and putative seed predation (rodent + arthropod). For every model it produces trace plots, Rhat/ESS diagnostics and posterior predictive checks, and extracts the posterior distributions. It then computes the results presented in the article: island-level and marginalized probabilities of attack by island type and biogeographic realm, contrasts among island types and realms, slope parameters, regression plots, counterfactual (causal-effect) contrasts, the estimated proportion of unidentified attacks per realm, and a sensitivity analysis comparing Bayesian imputation with complete-case analysis for bush cover. functions_mod_diagnostics.R helper function 'mod_diagnostics()' used to plot Rhat / effective sample size diagnostics). 5.2 Quarto documents (.qmd) - sources of the Supporting Information-------------------------------------------------------------------Each .qmd file renders to a .docx/.pdf Supporting Information document. supporting_information_S1.qmd (Supporting Information S1) (i) Detailed protocol for the fruit-removal experiment (artificial red plasticine fruits, grid and bush design, 72-h exposure, identification of consumers from bite marks, macroecological and local variables recorded); (ii) description of the data-generating process, including the DAG of hypothesized causal relationships; and (iii) the generative simulation with the five validation models (altitude, island type, island isolation, bush cover, native vegetation), parameter-recovery plots and conclusion. Code counterpart: supporting_information_S1.R. supporting_information_S2.qmd (Supporting Information S2) Data wrangling: loading the raw data, percentage of missing values, percentage of fruit attacked by each consumer type, table of realms, countries and ecoregions, map of sampled islands, island distance matrix, data cleaning and aggregation per plant, indexing of categorical variables, and saving of the .rds files. Code counterpart: supporting_information_S2.R. supporting_information_S3.qmd (Supporting Information S3) Formal analysis. Sections: S1 introduction; S2 R packages and citations; S3 data wrangling for modelling (missing values, bush-cover imputation indices, standardization); S4 custom functions (average_effects, cond_effects, slope_pars); S5 mathematical notation and adjustment sets from the DAG; S6-S12 models for the effect of latitude, island isolation, island size, island (sampling) altitude, human footprint, native vegetation and bush cover (each with total frugivory, bird, lizard, seed dispersal and seed predation models: notation, Stan code, sampling setup, MCMC diagnostics and posterior extraction); S13 results (effects of realm and island type, contrasts, slope parameters, regression plots and causal effects); S14 estimation of the proportion of unidentified attacks across realms; S15 imputation sensitivity analysis; S16 computational environment. Code counterpart: supporting_information_S3.R. 5.4 Data objects (.rds) - processed data----------------------------------------data.rds Analysis-ready data created by supporting_information_S2.R. A list with: 'codes' (numeric indices to country, island, grid, realm, ecoregion, biome and island-type labels), 'dist_islands' (symmetric matrix of distances in km among the 43 islands), and 'data' (one row per plant: 15 fruits deployed, total fruits attacked and number attacked by
เหตุผลที่อยู่ในฐานติดตาม
ระเบียนนี้ได้รับ Impact Signal 82/100 จากความใหม่ แหล่งเผยแพร่ ความร่วมมือ และสัญญาณในข้อมูลบรรณานุกรม คะแนนนี้ใช้จัดลำดับการติดตาม ไม่ใช่การตัดสินคุณภาพงานวิจัย
ประเด็นที่เกี่ยวข้อง: Ecology and Vegetation Dynamics Studies · Animal Ecology and Behavior Studies · Plant and animal studies
บทบาทของนักวิจัยและสถาบันไทย
Kim R. McConkey · Chanpen Saralamba · National Science and Technology Development Agency · National Center for Genetic Engineering and Biotechnology · Mahidol University
ข้อจำกัดของข้อมูล
หน้านี้เป็นระเบียนบรรณานุกรมและข้อมูลจากบทคัดย่อ ยังไม่ใช่บทวิเคราะห์ฉบับเต็มหรือการประเมินคุณภาพงานวิจัย ควรตรวจสอบ DOI และเอกสารต้นฉบับก่อนนำไปอ้างอิง