Rare species advantage in Antarctic Lakes

TL;DR

While ecology and evolutionary sciences have long debated whether diversity is maintained selectively, this work measures selection over a $10^4$-fold range of abundance in naturally coevolving communities and implicate rare species advantage.

Abstract

The maintenance of diversity in complex ecological communities despite unpredictable dynamics and competitive exclusion is thought to require continual influx of new species or competitive advantages that accrue as species become rare. We examine isolated planktonic microbial communities under permanent ice cover in Antarctic lakes, recording prokaryotic abundance across 9 communities, 11 years, 30~m of depth, and thousands of species in the McMurdo LTER. We quantify rare species advantage by modeling community dynamics under frequency-dependent selection. We find persistent diversity and pervasive negative frequency dependence with limited immigration and turnover. While ecology and evolutionary sciences have long debated whether diversity is maintained selectively, we measure selection over a $10^4$-fold range of abundance in naturally coevolving communities and implicate rare species advantage.

Figures (4)

• Figure 1: McMurdo Dry Valley lakes harbor distinct microbial communities stratified by depth. (A) MDV lakes Bonney and Fryxell are located in the Taylor Valley of Southern Victoria Land in East Antarctica. (B) 16S ribosomal gene sampling spanned 15-30 m of depth in three lakes over 11 summers. Depth profiles of temperature, salinity and chlorophyll fluorescence correspond to stable stratification into epilimnion, chemocline and hypolimnion with two photosynthetic zones. (C) Species composition clusters by lake and depth stratum on principle coordinate axes (Bray-Curtis PCoA). Time courses of samples pooled by depth stratum (solid lines) remained within clusters, showing greater homogeneity across time than across communities (inset: recentered time series with 2013 value at the origin).
• Figure 2: Diversity and taxonomic compostition of communities persists over time. (A) Phylum-level community composition persists across time. (B) Rarefied species richness varies by community over a 3.1-fold range (Tukey box plots of species richness measured at different times). (C) Richness is not correlated with time.
• Figure 3: All communities exhibit negative frequency dependence. (A) We infer the effective frequency-dependent selection in each community by fitting time series of species abundance to a simple model of frequency-dependent selection by maximum likelihood. (B) Frequency dependence is negative and retains its shape across lake and depth. A common form of frequency dependence $s_k$ shared by all communities scaled by a depth-dependent rate constant $\delta_i$ (black lines) explains 79.8% of the variance in the inferred selection coefficients.
• Figure 4: Saturation in species-time curves indicates limited immigration and turnover. During the last third of time series, novel ribotypes are observed at rates at most 8.5%/year and at most 1.4%/year attain frequencies within the observation window. The long-term rate of immigration to observable frequency is therefore bounded to at most 1 to 9 novel ribotypes/year across communities.