Book•
Mixed Effects Models and Extensions in Ecology with R
11 Apr 2011-
TL;DR: In this paper, the authors apply additive mixed modelling on phyoplankton time series data and show that the additive model can be used to estimate the age distribution of small cetaceans.
Abstract: Limitations of linear regression applied on ecological data. - Things are not always linear additive modelling. - Dealing with hetergeneity. - Mixed modelling for nested data. - Violation of independence - temporal data. - Violation of independence spatial data. - Generalised linear modelling and generalised additive modelling. - Generalised estimation equations. - GLMM and GAMM. - Estimating trends for Antarctic birds in relation to climate change. - Large-scale impacts of land-use change in a Scottish farming catchment. - Negative binomial GAM and GAMM to analyse amphibian road killings. - Additive mixed modelling applied on deep-sea plagic bioluminescent organisms. - Additive mixed modelling applied on phyoplankton time series data. - Mixed modelling applied on American Fouldbrood affecting honey bees larvae. - Three-way nested data for age determination techniques applied to small cetaceans. - GLMM applied on the spatial distribution of koalas in a fragmented landscape. - GEE and GLMM applied on binomial Badger activity data.
Citations
More filters
TL;DR: A protocol for data exploration is provided; current tools to detect outliers, heterogeneity of variance, collinearity, dependence of observations, problems with interactions, double zeros in multivariate analysis, zero inflation in generalized linear modelling, and the correct type of relationships between dependent and independent variables are discussed; and advice on how to address these problems when they arise is provided.
Abstract: Summary 1. While teaching statistics to ecologists, the lead authors of this paper have noticed common statistical problems. If a random sample of their work (including scientific papers) produced before doing these courses were selected, half would probably contain violations of the underlying assumptions of the statistical techniques employed. 2. Some violations have little impact on the results or ecological conclusions; yet others increase type I or type II errors, potentially resulting in wrong ecological conclusions. Most of these violations can be avoided by applying better data exploration. These problems are especially troublesome in applied ecology, where management and policy decisions are often at stake. 3. Here, we provide a protocol for data exploration; discuss current tools to detect outliers, heterogeneity of variance, collinearity, dependence of observations, problems with interactions, double zeros in multivariate analysis, zero inflation in generalized linear modelling, and the correct type of relationships between dependent and independent variables; and provide advice on how to address these problems when they arise. We also address misconceptions about normality, and provide advice on data transformations. 4. Data exploration avoids type I and type II errors, among other problems, thereby reducing the chance of making wrong ecological conclusions and poor recommendations. It is therefore essential for good quality management and policy based on statistical analyses.
5,894 citations
Cites background or methods from "Mixed Effects Models and Extensions..."
...Zuur et al. (2009a) showed how the use of a data transformation resulted in different conclusions about long-term trends compared to an appropriate analysis using untransformed data; hence it may be best to avoid transforming response variables....
[...]
...These include using lagged response variables as covariates (Brockwell & Davis 2002), mixed effects modelling (Pinheiro & Bates 2000), imposing a residual correlation structure using generalized least squares (Zuur et al. 2009a) or allowing regression parameters to change over time (Harvey 1989)....
[...]
...This approach is more difficult if there is no clear sequence to the observations (e.g. multiple observations on the same object), but in this case one can include a dependence structure using random effects (Pinheiro & Bates 2000; Fitzmaurice et al. 2004; Brown & Prescott 2006; Zuur et al. 2009a)....
[...]
...More serious examples of violation can be found in Zuur et al. (2009a)....
[...]
...In Zuur et al. (2009a) a large number of data sets were analysed in which dependence among observations played a role....
[...]
TL;DR: The glmmTMB package fits many types of GLMMs and extensions, including models with continuously distributed responses, but here the authors focus on count responses and its ability to estimate the Conway-Maxwell-Poisson distribution parameterized by the mean is unique.
Abstract: Count data can be analyzed using generalized linear mixed models when observations are correlated in ways that require random effects However, count data are often zero-inflated, containing more zeros than would be expected from the typical error distributions We present a new package, glmmTMB, and compare it to other R packages that fit zero-inflated mixed models The glmmTMB package fits many types of GLMMs and extensions, including models with continuously distributed responses, but here we focus on count responses glmmTMB is faster than glmmADMB, MCMCglmm, and brms, and more flexible than INLA and mgcv for zero-inflated modeling One unique feature of glmmTMB (among packages that fit zero-inflated mixed models) is its ability to estimate the Conway-Maxwell-Poisson distribution parameterized by the mean Overall, its most appealing features for new users may be the combination of speed, flexibility, and its interface’s similarity to lme4
4,497 citations
Cites result from "Mixed Effects Models and Extensions..."
...This example previously appeared in similar forms (Zuur et al., 2009) and (Bolker et al., 2013); the data were originally published by (Roulin and Bersier, 2007)....
[...]
...This example previously appeared in Zuur et al. (2009) and Bolker et al. (2013) and was originally published by Roulin and Bersier (2007)....
[...]
...This example previously appeared in similar forms (Zuur et al., 2009) and (Bolker et al....
[...]
United Nations Environment Programme1, American Museum of Natural History2, Imperial College London3, Swansea University4, University College London5, National University of Cordoba6, Tel Aviv University7, Max Planck Society8, University of Oldenburg9, Microsoft10, University of Oxford11, University of Wisconsin–Eau Claire12
TL;DR: A terrestrial assemblage database of unprecedented geographic and taxonomic coverage is analysed to quantify local biodiversity responses to land use and related changes and shows that in the worst-affected habitats, pressures reduce within-sample species richness by an average of 76.5%, total abundance by 39.5% and rarefaction-based richness by 40.3%.
Abstract: Human activities, especially conversion and degradation of habitats, are causing global biodiversity declines. How local ecological assemblages are responding is less clear--a concern given their importance for many ecosystem functions and services. We analysed a terrestrial assemblage database of unprecedented geographic and taxonomic coverage to quantify local biodiversity responses to land use and related changes. Here we show that in the worst-affected habitats, these pressures reduce within-sample species richness by an average of 76.5%, total abundance by 39.5% and rarefaction-based richness by 40.3%. We estimate that, globally, these pressures have already slightly reduced average within-sample richness (by 13.6%), total abundance (10.7%) and rarefaction-based richness (8.1%), with changes showing marked spatial variation. Rapid further losses are predicted under a business-as-usual land-use scenario; within-sample richness is projected to fall by a further 3.4% globally by 2100, with losses concentrated in biodiverse but economically poor countries. Strong mitigation can deliver much more positive biodiversity changes (up to a 1.9% average increase) that are less strongly related to countries' socioeconomic status.
2,532 citations
TL;DR: The temporal dynamics of the composition of vaginal bacterial communities in 32 reproductive-age women over a 16-week period revealed the dynamics of five major classes of bacterial communities and showed that some communities change markedly over short time periods, whereas others are relatively stable.
Abstract: Elucidating the factors that impinge on the stability of bacterial communities in the vagina may help in predicting the risk of diseases that affect women’s health. Here, we describe the temporal dynamics of the composition of vaginal bacterial communities in 32 reproductive-age women over a 16-week period. The analysis revealed the dynamics of five major classes of bacterial communities and showed that some communities change markedly over short time periods, whereas others are relatively stable. Modeling community stability using new quantitative measures indicates that deviation from stability correlates with time in the menstrual cycle, bacterial community composition, and sexual activity. The women studied are healthy; thus, it appears that neither variation in community composition per se nor higher levels of observed diversity (co-dominance) are necessarily indicative of dysbiosis.
1,114 citations
TL;DR: It is found that thermal tolerance breadths generally increase with latitude, and do so at a greater rate in the Northern Hemisphere, and differences between terrestrial and marine ectotherms in how thermal physiology varies with latitude may relate to the degree of temperature variability experienced on land and in the ocean.
Abstract: A tenet of macroecology is that physiological processes of organisms are linked to large-scale geographical patterns in environmental conditions. Species at higher latitudes experience greater seasonal temperature variation and are consequently predicted to withstand greater temperature extremes. We tested for relationships between breadths of thermal tolerance in ectothermic animals and the latitude of specimen location using all available data, while accounting for habitat, hemisphere, methodological differences and taxonomic affinity. We found that thermal tolerance breadths generally increase with latitude, and do so at a greater rate in the Northern Hemisphere. In terrestrial ectotherms, upper thermal limits vary little while lower thermal limits decrease with latitude. By contrast, marine species display a coherent poleward decrease in both upper and lower thermal limits. Our findings provide comprehensive global support for hypotheses generated from studies at smaller taxonomic subsets and geographical scales. Our results further indicate differences between terrestrial and marine ectotherms in how thermal physiology varies with latitude that may relate to the degree of temperature variability experienced on land and in the ocean.
1,026 citations