Models of functional ecospace diversification within life-habit frameworks (functional-trait spaces) are increasingly used across community ecology, functional ecology, and paleoecology. In general, these models can be represented by four basic processes, three that have driven causes and one that occurs through a passive process. The driven models include redundancy (caused by forms of functional canalization), partitioning (specialization), and expansion (divergent novelty), but they also share important dynamical similarities with the passive neutral model. In this second of two companion articles, Monte Carlo simulations of these models are used to illustrate their basic statistical dynamics across a range of data structures and implementations. Ecospace frameworks with greater numbers of characters (functional traits) and ordered (multistate) character types provide more distinct dynamics and greater ability to distinguish the models, but the general dynamics tend to be congruent across all implementations. Classification-tree methods are proposed as a powerful means to select among multiple candidate models when using multivariate data sets. Well-preserved Late Ordovician (type Cincinnatian) samples from the Kope and Waynesville formations are used to illustrate how these models can be inferred in empirical applications. Initial simulations overestimate the ecological disparity of actual assemblages, confirming that actual life habits are highly constrained. Modifications incorporating more realistic assumptions (such as weighting potential life habits according to actual frequencies and adding a parameter controlling the strength of each model’s rules) provide better correspondence to actual assemblages. Samples from both formations are best fit by partitioning (and to lesser extent redundancy) models, consistent with a role for local processes. When aggregated as an entire formation, the Kope Formation pool remains best fit by the partitioning model, whereas the entire Waynesville pool is better fit by the redundancy model, implying greater beta diversity within this unit. The ‘ecospace’ package is provided to implement the simulations and to calculate their dynamics using the R statistical language.
Life-habit/functional-trait codings for the Kope and Waynesville Formation species pool
KWTraits.csv is a comma-separated value (.csv) format file listing the aggregate species pool for the Kope and Waynesville Formation used in empirical analyses. (The file is also included as a data file within the 'ecospace' R package.) The first three columns list taxonomic information. The remaining columns list ecospace character states (functional traits). See supplementary appendix A and Novack-Gottshall (2007) for information on characters and states. See text for explanation of how multistate characters were rescaled.
K&WTraits.csv
Two-model model-selection support data files for Kope and Waynesville Formation samples, stratigraphic section, member, and formation aggregates
File is in comma-separated value (.csv) format. The first five columns describe the Paleobiology Database collection identification number, scale (hand sample, stratigraphic section, etc.) of the sample, and stratigraphic/section names. Columns 6–14 list sample size (S, species richness) and values for eight disparity statistics (with NA designating when a statistic could not be calculated, because there were fewer than four unique life habits in the sample); see text for descriptions and abbreviations of statistics. The last column identifies which model has the best support among those candidates considered. The remaining columns list the classification-tree support each sample has for each candidate model considered. emp2-modelfits.csv lists model support using the classification tree trained on the 50% and 100%-strength training data sets. emp3-modelfits.csv lists model support for the tree trained on 50%, 90%, and 100% training data.
emp2-modelfits.csv
Three-model model-selection support data files for Kope and Waynesville Formation samples, stratigraphic section, member, and formation aggregates
File is in comma-separated value (.csv) format. The first five columns describe the Paleobiology Database collection identification number, scale (hand sample, stratigraphic section, etc.) of the sample, and stratigraphic/section names. Columns 6–14 list sample size (S, species richness) and values for eight disparity statistics (with NA designating when a statistic could not be calculated, because there were fewer than four unique life habits in the sample); see text for descriptions and abbreviations of statistics. The last column identifies which model has the best support among those candidates considered. The remaining columns list the classification-tree support each sample has for each candidate model considered. emp3-modelfits.csv lists model support for the tree trained on 50%, 90%, and 100% training data.
emp3-modelfits.csv
Five-model model-selection support data files for Kope and Waynesville Formation samples, stratigraphic section, member, and formation aggregates
File is in comma-separated value (.csv) format. The first five columns describe the Paleobiology Database collection identification number, scale (hand sample, stratigraphic section, etc.) of the sample, and stratigraphic/section names. Columns 6–14 list sample size (S, species richness) and values for eight disparity statistics (with NA designating when a statistic could not be calculated, because there were fewer than four unique life habits in the sample); see text for descriptions and abbreviations of statistics. The last column identifies which model has the best support among those candidates considered. The remaining columns list the classification-tree support each sample has for each candidate model considered. emp5-modelfits.csv lists model support for the tree trained on 50%, 75%, 90%, 95%, and 100% training data.
emp5-modelfits.csv
Supplementary Appendices 1-4 for manuscript
Appendix 1 gives an example of how life-habit character states were inferred and coded. Appendix 2 describes technical details on classification tree methods and confusion matrices. Appendix 3-4 give further details for the other Supplementary data files on Data Dryad.
EcomodelsII_Appendices.docx
Supplementary Figure 6
Comparing statistical dynamics for different ecospace framework structures: varying number of characters, (A) 5 characters, (B) 15 characters, and (C) 25 characters. Each framework had mixed character types, in identical proportions (40% binary, 20% three-state factor, 20% five-state factor, and, 20% five-state ordered numeric character types). 5 "seed" species were chosen at random to begin each simulation. Other simulation details and graphical interpretation are the same as is Figure 2. Trends in total variance were excluded because the inclusion of factors prevented their calculation. The dynamics are generally similar, although larger frameworks allow modestly more powerful model selection using classification-tree methods (83%, 85%, and 86% of training models, respectively, classified correctly using classification-tree methods). See Supplementary Appendix 2 for additional details.
Novack-GottshallBSuppFigure6_2col.tif
Supplementary Figure 7
Comparing statistical dynamics for different ecospace framework structures: varying character types, (A) factor, (B) ordered factor, (C) ordered numeric, and (D) binary. Each framework had 15 characters, four states per character (except for binary, which had two binary states per character), and five seed species. Trends in total variance were excluded in parts A and B because the inclusion of factors prevented their calculation. Other simulation details and graphical interpretation are the same as in Supplementary Figure 6. Dynamics are generally similar, but frameworks built with ordered factors performed substantially better (94% of trained models classified correctly) than the others (78% for unordered factors, 79% for ordered numerics, and 81% for binaries). See Supplementary Appendix 2 for additional details.
Novack-GottshallBSuppFigure7_2col.tif
Supplementary Figure 8
Comparing statistical dynamics for different ecospace framework structures: varying number of "seed" species chosen at random at start of simulation, (A) 3 species, (B) 5 species, and (C) 10 species. Each framework had 15 mixed character types, such that part B is identical to Supplementary Figure 6B. Trends are only plotted starting at 5 species for comparative purposes, which explains idiosyncratic behaviors at low sample sizes (i.e., missing trend lines in part A and overlapping models in part C). Other simulation details and graphical interpretation are the same as is Supplementary Figure 6. Note that the functional-diversity statistics could not be calculated for the redundancy model in part A because their calculation requires a minimum of four unique life habits; however, all statistics (except V) were included as potential predictor variables in the classification tree algorithm. Dynamics are generally similar across models, but simulations with fewer seed species provide the most powerful model selection using classification-tree methods (85% of models classified correctly for both 3-seed and 5-seed simulations). Starting with larger numbers of seed species impedes enacting distinct model rules, and results in 75% of models classified correctly. See Supplementary Appendix 2 for additional details.
Novack-GottshallBSuppFigure8_2col.tif
Supplementary Figure 9
Performance of classification tree used on validation samples as a function of sample size. Please see README for additional details.
Novack-GottshallBSuppFigure9_2col.tif