2018
DOI: 10.3389/fevo.2018.00149
|View full text |Cite
|
Sign up to set email alerts
|

Modelling Palaeoecological Time Series Using Generalised Additive Models

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

1
342
0

Year Published

2019
2019
2024
2024

Publication Types

Select...
8
1

Relationship

0
9

Authors

Journals

citations
Cited by 372 publications
(343 citation statements)
references
References 59 publications
1
342
0
Order By: Relevance
“…Generalized additive mixed models (GAMMs) were used to test temporal trends in the δ 15 N and Suess corrected δ 13 C values of skull bone samples between 1990 and 2017 years. By using GAMMs, it is possible to observe the tendency of stable isotope ratios to change over time without defining a particular curve order a priori, while also accounting for temporal correlation (Wood, 2017;Simpson, 2018). GAMMs were fitted using a Gaussian distribution with an identity link function in the Mixed GAM Computation Vehicle (mgcv version 1.8-28) package with the restricted maximum likelihood (REML) smoothing method (Wood, 2017) in R 3.6.0 (R Development Core Team, 2019).…”
Section: Discussionmentioning
confidence: 99%
“…Generalized additive mixed models (GAMMs) were used to test temporal trends in the δ 15 N and Suess corrected δ 13 C values of skull bone samples between 1990 and 2017 years. By using GAMMs, it is possible to observe the tendency of stable isotope ratios to change over time without defining a particular curve order a priori, while also accounting for temporal correlation (Wood, 2017;Simpson, 2018). GAMMs were fitted using a Gaussian distribution with an identity link function in the Mixed GAM Computation Vehicle (mgcv version 1.8-28) package with the restricted maximum likelihood (REML) smoothing method (Wood, 2017) in R 3.6.0 (R Development Core Team, 2019).…”
Section: Discussionmentioning
confidence: 99%
“…All statistical analyses were conducted in R, and levels were set to 0.05. All generalised α additive mixed-effects models ("GAMMs") were fit with restricted maximum likelihood ("REML") in the R package "mgcv" (Wood, 2011), as recommended by Simpson (2018).…”
Section: Statistical Analysesmentioning
confidence: 99%
“…In all the cases, fitted GAMs were estimated using maximum likelihood-based smoothness selection procedures, in particular the restricted maximum likelihood (REML). A continuous time first-order autoregressive process (CAR(1)) was chosen to account for the correlation between residuals [35]. To identify periods of transition, we estimated simultaneous confidence intervals from the posterior distribution of the model (under an empirical Bayesian formulation), and the first derivative of the fitted trend [35].…”
Section: Data Analysesmentioning
confidence: 99%
“…A continuous time first-order autoregressive process (CAR(1)) was chosen to account for the correlation between residuals [35]. To identify periods of transition, we estimated simultaneous confidence intervals from the posterior distribution of the model (under an empirical Bayesian formulation), and the first derivative of the fitted trend [35]. Periods of significant change are identified as those time points where the simultaneous confidence interval on the first derivative bounded away from zero [35].…”
Section: Data Analysesmentioning
confidence: 99%