GAM phenophase estimation

Knowledge graph centered on GAM phenophase estimation with 71 nodes and 195 connections. Top connected: Rocky Mountain Biological Laboratory, Two-form Bumble Bee, Bumble Bees, Buff-tailed Bumble Bee, Yellow-fronted Bumble Bee.

Method synopsis

Using Generalized Additive Models to estimate emergence, peak, and senescence timing from sparse time-series abundance data. Models fit abundance curves and extract phenological metrics from first and last occurrences relative to peak abundance.

Synthesized from method descriptions across 3 papers using this protocol.

Procedure as described in the canonical source

Steps below were extracted from the paper that introduces this protocol — Bee phenology is predicted by climatic variation and functional traits (2020), Ecology Letters. Implementations in other papers (listed below) may differ.

1. Estimate phenophases using GAMs
   Fit Generalized Additive Models with day-of-year as explanatory variable and abundance as response using cubic spline smoothing basis with Gaussian distribution family. Set dimension of smoothing basis to 5 when there were ≥5 observations, and 5 for ≥5 observations. Determine peak timing by calculating predicted date of maximum abundance, emergence as first occurrence of 5% of maximum, senescence as last occurrence of 5% of maximum
   Quantities: Models fit for 67 bee species, phenophase estimates for 519 emergence dates, 438 peak dates, 584 senescence datesDuration: Computational analysis performed post-data collectionConditions: Statistical modeling environment
   Equipment: R software, mgcv R package

Typical Equipment

• R software
• mgcv R package

Output Measurements

• emergence date
• peak abundance date
• senescence date

Papers Using This Protocol (3)

Species Studied With This Protocol

Places Where Used