Drastic population fluctuations explain the rapid extinction of the passenger pigeon

Abstract

To assess the role of human disturbances in species' extinction requires an understanding of the species population history before human impact. The passenger pigeon was once the most abundant bird in the world, with a population size estimated at 3-5 billion in the 1800s; its abrupt extinction in 1914 raises the question of how such an abundant bird could have been driven to extinction in mere decades. Although human exploitation is often blamed, the role of natural population dynamics in the passenger pigeon's extinction remains unexplored. Applying high-throughput sequencing technologies to obtain sequences from most of the genome, we calculated that the passenger pigeon's effective population size throughout the last million years was persistently about 1/10,000 of the 1800's estimated number of individuals, a ratio 1,000-times lower than typically found. This result suggests that the passenger pigeon was not always super abundant but experienced dramatic population fluctuations, resembling those of an "outbreak" species. Ecological niche models supported inference of drastic changes in the extent of its breeding range over the last glacial-interglacial cycle. An estimate of acorn-based carrying capacity during the past 21,000 y showed great year-to-year variations. Based on our results, we hypothesize that ecological conditions that dramatically reduced population size under natural conditions could have interacted with human exploitation in causing the passenger pigeon's rapid demise. Our study illustrates that even species as abundant as the passenger pigeon can be vulnerable to human threats if they are subject to dramatic population fluctuations, and provides a new perspective on the greatest human-caused extinction in recorded history.

Keywords: ancient DNA; genome sequences; toe pad.

Fig. 1.

Long-term average effective population size (N_e) estimates of the passenger pigeon. N_e estimates were constructed using G-PhoCS based on different locus sampling settings. Each histogram and bar describes the mean value and 95% credible interval of N_e based on a certain model and dataset. The sampling settings include l-kbp loci separated from each other by 10, 20, 50, and 100 kbp, indicated by 1k/10k, 1k/20k, 1k/50k, and 1k/100k, respectively, and 0.1-, 0.2-, 1-, and 2-kbp loci separated from each other by 50 kbp, indicated by 0.1k/50k, 0.5k/50k, 1k/50k and 2k/50k, respectively (see SI Appendix for details). The N_e based on the sampling setting of 1k/50k (indicated by an asterisk) is reported in the main text.

Fig. 2.

Demographic history of passenger pigeons. PSMC analyses were applied to individual diploid genomes of three passenger pigeons, BMNH794, BMNH1149, and BMNH3993. Uniform false-negative rate correction was applied to BMNH1149 and BMNH3993 with correction rates of 40% and 60%, indicated by “BMNH1149 [0.4]” and “BMNH3993 [0.6]”, respectively (Methods and SI Appendix). “g” indicates generation time in years, and “μ” indicates genomic substitution rate.

Fig. 3.

Predicted breeding ranges of the passenger pigeon. Predicted breeding ranges at current day (A), the LGM (B), and the LIG (C). Predicted breeding ranges are based on ENMs built with 19^th century occurrences of breeding passenger pigeons (triangles in A as training data and circles as testing data) and the seven bioclimatic variables that each contributed at least 5% to the current-day model. The green areas delineate potential breeding ranges for the passenger pigeon. Increasing shades of green represent suitable areas for the passenger pigeon using thresholds of increasingly higher omission rates (0%, 1%, and 5%).

Fig. 4.

Historical oak coverage and passenger pigeon carrying capacity from 21,000 y before present (YBP) to present day. (A) Historical oak coverage was converted from fossil-pollen records for northern and eastern North America (SI Appendix, Fig. S11) (47). (B) The carrying capacities of passenger pigeons are annual pigeon abundances that could be sustained by acorn production. The median acorn production of red oaks (Quercus rubra, ●) and white oaks (Quercus alba, ○) collected from multiple sites and years (SI Appendix, Table S8) (30, 31) were used to calculate carrying capacity.