Comparative Spatial Paleoecology: Assessing Niche Competition between Eocene North American Multituberculates and Rodents Regarding Forest Resources to Elucidate the Cause of Multituberculate Extinction

Multituberculate extinction is often cited as a classic case of competitive exclusion, coinciding with the first rodent arrivals in the late Paleocene. Analyzing 124 North American multituberculate last occurrence records during the Eocene from 56 to 34 million years ago, this study aimed to differentiate Eocene multituberculate and coeval rodent floral associations through geographic spatial analysis to understand niche overlap between the two groups.  If competitive exclusion with rodents was a factor in multituberculate extinction, both multituberculates and rodents would be predicted to share similar forest habitat preferences and have competed for similar ecological niches regarding their forest associations. Using spatial analysis, this study found that Eocene rodents and multituberculates did not overlap in their forest associations. The findings indicate that multituberculates were unique in inhabiting a specific type of ancient forest habitat, favoring forests composed of Metasequoia, Glyptostrobus, and Alnus, and thus thrived in wetter northern temperate forest communities during the Eocene. Metasequoia and Glyptostrobus declined significantly in North America during the later Cenozoic, coinciding with multituberculate decline and extinction as the global climate shifted toward colder and drier climates around the Eocene-Oligocene boundary. In contrast, the success of rodents is attributed to their much broader forest affinity. These preferences align with the widespread distribution of rodents today, contributing to their modern success. The absence of any similar reconstructed forest habitat preferences between rodents and multituberculates suggests that changing forest structure, rather than competitive exclusion, drove multituberculate extinction.

Methods
How was this dataset collected?

The dataset consists of latitude and longitude data for Eocene North American fossil sites. It includes multituberculates (Neoplagiaulacidae and Neolitomus) and six rodent families (Ischyromyidae, Cylindrodontidae, Sciuravidae, Eutypomyidae, Eomyidae, and Protoptychidae) as defined in McKenna and Bell (1997). Data sources included primary literature, the Paleobiology Database, and various museum collections. Each site’s location was verified and adjusted using Google Earth to the nearest outcrop of rock. Data on fossil plants were also collected and included in this study. Where available, each occurrence was referenced with a PaleobioDB number. The PaleobioDB database is licensed under a CC0 International License. Specific references for each occurrence are stored in the MAMMAL_REF and PLANT_REF subfolders.

How has this dataset been processed?

To analyze the spatial relationships between mammal and plant fossil groups, four ranked parameters were calculated for each comparison. The necessary data, including point and polygon shapes, is located in the DATA_ANALYSIS/GIS_FILES folder. This folder contains a file named Project-Multituberculates.qgs, which loads all relevant data, including the outlines of the world's coastlines and a 2x2 grid used for model counts. Centroid and polygon shape files of the point data were also processed. Details on how the four parameters were calculated are available in the README.txt file. For calculating the EDRF parameter, Python scripts and necessary data are provided in the EDRF-Program&Data folder within the DATA_ANALYSIS.zip folder.

Null model values for each parameter are determined using scripts and data in the NULL_MODEL.zip folder, with methods detailed in the Methods.txt file, including instructions for running the scripts in the STOCHASTIC and PROBABILISTIC subfolders.

The GRID_MODEL.zip folder includes tallies and modeled occurrences for the Chi-Square test, with results reported in a separate file. Tallies for each grid were made using the mammal and plant data in the GIS_FILES folder.

Tables and appendices summarizing the conclusions from the data analysis are also included.