Oxic-anoxic regime shifts mediated by feedbacks between biogeochemical processes and microbial community dynamics

Abstract

Although regime shifts are known from various ecosystems, the involvement of microbial communities is poorly understood. Here we show that gradual environmental changes induced by, for example, eutrophication or global warming can induce major oxic-anoxic regime shifts. We first investigate a mathematical model describing interactions between microbial communities and biogeochemical oxidation-reduction reactions. In response to gradual changes in oxygen influx, this model abruptly transitions between an oxic state dominated by cyanobacteria and an anoxic state with sulfate-reducing bacteria and phototrophic sulfur bacteria. The model predictions are consistent with observations from a seasonally stratified lake, which shows hysteresis in the transition between oxic and anoxic states with similar changes in microbial community composition. Our results suggest that hysteresis loops and tipping points are a common feature of oxic-anoxic transitions, causing rapid drops in oxygen levels that are not easily reversed, at scales ranging from small ponds to global oceanic anoxic events.The role of microbial communities in regime shifts is poorly understood. Here, the authors use a mathematical model and field data from a seasonally stratified lake to show that gradual environmental changes can induce oxic-anoxic regime shifts mediated by microbial community dynamics and redox processes.

Fig. 1

Schematic diagram of the microbial ecosystem model. The model consists of three bacterial functional groups (CB cyanobacteria, PB phototrophic sulfur bacteria, SB sulfate-reducing bacteria) and four chemical substrates (P phosphorus, O oxygen, S_R reduced sulfur, S_O oxidized sulfur). Arrows denote the consumption (blue arrows) and production (magenta arrows) of chemicals by the microbial populations. Orange lines represent growth inhibition of the microbial populations. Green arrows indicate abiotic oxidation of reduced sulfur to oxidized sulfur

Fig. 2

Illustration of the two alternative stable states. a, b When the model starts with a low initial population density of cyanobacteria (CB), it develops towards an anoxic ecosystem with, a high abundances of phototrophic sulfur bacteria (PB) and sulfate-reducing bacteria (SB) and, b a high concentration of reduced sulfur (S _R) but low oxygen concentration (O). c, d When the model starts with a high initial population density of cyanobacteria, it develops towards an oxic ecosystem with, c high abundances of cyanobacteria and, d high concentrations of oxygen and oxidized sulfur (S _O). Parameter values are given in Supplementary Table 1, with P _b = 9.5 μM, α _O = 8 × 10⁻⁴ h⁻¹. Initial conditions: a, b N _PB = N _SB = 1 × 10⁷ cells L⁻¹, N _CB = 5 × 10¹ cells L⁻¹, S _O = 300 μM, S _R = 300 μM, O = 10 μM, P = 10 μM; c, d N _PB = N _SB = 1 × 10² cells L⁻¹, N _CB = 1 × 10⁸ cells L⁻¹, S _O = 500 μM, S _R = 50 μM, O = 300 μM, P = 4 μM

Fig. 3

Regime shifts between oxic and anoxic states. a Cyanobacterial population density and, b oxygen concentration predicted at steady state, as function of the oxygen diffusivity. Blue lines indicate the oxic state and red lines the anoxic state. In a the blue and red arrows indicate the basins of attraction of the oxic and anoxic state, respectively, and the dashed orange line is the separatrix between these two basins of attraction. In b T ₁ and T ₂ indicate the two tipping points of the system and black arrows illustrate the hysteresis loop. Parameter values are given in Supplementary Table 1, with P _b = 9.5 μM. The initial cyanobacterial density varies, while the other initial conditions are set at N _PB = N _SB = 1 × 10⁸  cells L⁻¹, S _O = 250 μM, S _R = 350 μM, O = 150 μM, P = 9.5 μM

Fig. 4

Two-parameter bifurcation diagram. The diagram illustrates how the model predictions vary with phosphorus availability and oxygen diffusivity. Blue region=oxic state with cyanobacteria. Pink region=alternative stable states, with either cyanobacteria in the oxic state or sulfate-reducing bacteria in the anoxic state. Gray region = alternative stable states, with either cyanobacteria in the oxic state or coexistence of sulfate-reducing bacteria and phototrophic sulfur bacteria in the anoxic state. Green region=anoxic state with sulfate-reducing bacteria. Orange region=anoxic state with coexistence of sulfate-reducing bacteria and phototrophic sulfur bacteria. Parameter values are given in Supplementary Table 1

Fig. 5

Data from the seasonally stratified Lake Vechten. Contour plots of seasonal changes in, a temperature; b dissolved oxygen; c sulfate; d sulfide. The plots are based on 189 water samples (black dots) taken from March 2013 to early April 2014. e Seasonal changes of the strength of stratification (black line, quantified by the squared buoyancy frequency N ² at the thermocline) and oxygen saturation just below the thermocline (blue line, measured at 6 m depth). f Seasonal changes in relative abundances of cyanobacteria (green line), phototrophic sulfur bacteria (magenta line) and sulfate-reducing bacteria (light blue line) in the metalimnion (at 6 m depth), based on 16S rRNA sequence data

Fig. 6

Evidence for regime shifts between oxic and anoxic states in Lake Vechten. a Hysteresis loop of oxygen saturation in the hypolimnion (7 m depth) plotted against the inverse of the stratification strength (1/N ², where N ² is the squared buoyancy frequency). The inverse of the stratification strength provides a simple proxy of oxygen diffusivity across the thermocline (see Methods). Data points are from March 2013 to March 2014; red arrows indicate the direction of time. b Co-occurrence network of bacteria in the metalimnion, based on 16S rRNA sequence data. Green lines represent positive interactions (co-occurrence), whereas magenta lines represent negative interactions (mutual exclusion)