Aggregation Phenomena in Cyanobacterial, Stromatolite Analogues Alexander Petroff Tanja Bosak Biqing Liang Daniel Rothman Min Sub Sim May 16, 2008 Abstract Earth's biological history is largely the story of microbial evolution. If one is to understand the evolution of such fundamental processes as photosynthesis, it is imperative to recognize and interpret microbial fossils. To these ends, we examine the forces that shape a modern cyanobacterial mat from the hot springs of Yellowstone that is thought to grow in a manner similar to an ancient structure called a conical stromatolite. Here we show that the initial stages in the growth of this mat is analogous to crystal-nucleation with long-range interactions. This result provides intuition for the forces that shape modern mats and, by analogy, conical stromatolites. Furthermore, the importance of long-range interactions suggests that many current models of stromatolite morphogenesis, which rely exclusively on local forcing, are not applicable to the early stages of growth. To understand the evolution of life on does a conical morphology of a microbial mat early Earth, one must distinguish and inter- imply about the types of microbes that grew pret biotic and abiotic processes preserved it? in the rock record. Although there is often In order to understand why a microbial chemical evidence that signals the presence mat might develop a conical morphology, we of life during the formation of a geological study a population of modern, cone-forming feature, many of the oldest presumed fossils cyanobacterial mats from the hot springs of are only identified by a distinctive morphol- Yellowstone National Park. These mats grow ogy [8]. In such cases, how does one inter- in springs that are supersaturated in silica pret a presumed fossil? One example that is or calcium carbonate; as a result cones are abundant throughout Archean and Protero- mineralized in-situ. As these mats grow, zoic formations is the conical stromatolite. they produce laminated mineral-precipitates These laminated, lithified sedimentary struc- which passively preserve the mat's morphol- tures have long been thought to have formed ogy. Since these mats grow in a manner simi- when minerals precipitated within a micro- lar to how some ancient stromatolites are be- bial mat. The conical morphology of these lieved to have formed, modern cone-forming deposits is thought to reflect the shape the mats offer a unique opportunity to test hy- microbial mat that grew it [1][4][5]. Assum- potheses about stromatolite morphogenesis ing that these stromatolites are fossils, what [9]. In this paper, we examine the earliest 1 stages of the stromatolite morphogenesis: the aggregation of bacteria. It has been noted that the photosyn- thetic rate of bacteria within Yellowstone stromatolite analogues is substantially dimin- ished by the presence of oxygen. However, the negative effects of oxygen are somewhat mitigated when the bacteria aggregate into clumps. It has been proposed that the clumps foster other microbial communities that con- sume the oxygen within the clump, thus maintaining a low-oxygen environment [10]. This observation suggests that the formation of clumps could be a response to oxygen. It has never been shown that oxygen sensitivity causes the formation of clumps. To test the hypothesis that clumps form Figure 1: A clump of cyanobacteria under in response to oxygen, we observe the the 50x magnification using chlorophyll epifluo- behavior of small (∼200µm) samples of bac- rescence. The scale bar is 0.02 cm. (A) teria in response to aerobic (5% CO2, 21% A clump in oxygenated water at the begin- O2, 74% N2) and anaerobic (5% CO2, 5% H2, ning of an experiment. (B) The same clump 90% N2) environments under a microscope. one hour later. Any filament that leaves the On the time scale of tens of minutes, samples clump is observed to quickly retract back into in oxygenated water contracted slightly. Any it. (C) A sample placed in anoxic medium filament that began to move away from the at the beginning of an experiment. (D) The clump quickly retracted (Figure 1 (A) and same sample one hour later. At the end of the (B)). However, samples in anoxic medium experiment all of the filaments that left the became more disperse within tens of min- clump continued to move away from regions utes (Figure 1 (C) and (D)). We therefore of high density. conclude that aggregation is a behavioral re- sponse to oxygen. To further explore the role of oxygen macroscopic clumps. in the formation of clumps, we followed the The observation that bacteria transi- evolution of a mat as photosynthesis gradu- tion from a disordered state to an ordered ally raised the concentration of dissolved oxy- state a critical oxygen concentration suggests gen. Clumps were only observed in mats the formation of bacterial aggregates is simi- when the concentration of dissolved oxygen in lar to a phase transition. Specifically, we ex- the medium surpassed a certain value around plore an analogy between clumping and crys- 120 nM cm−3. This observation suggest that tal nucleation. The formation of a clump re- there is a critical concentration of dissolved quires an interface between the oxygenated oxygen at which point bacteria cease to be water and the low-oxygen environment within repulsed by one-another and begin to form the clump. Our observations are consistent with the hypothesis that when the surface- 2 area-to-volume ratio of the clump is too high potential that the microbial mat extremizes, microbes leave the surface faster than they the basic, short-range interactions that shape reproduce within it. Thus, small clumps dis- a mat are consistent with a Landau-Ginzberg solve while large clumps grow. Crystal nucle- equation. ation is similar. This process occurs when the Z t K concentration of a solute approaches a criti- −βH = d2x ρ2 + uρ4 + (rρ)2: (1) cal value where fluctuations in the solute's 2 2 local concentration become large. Some of This equation captures the basic phe- these fluctuations concentrate enough mate- nomenology of the system: when t reaches a rial in a small enough volume such that there critical value near zero the equilibrium den- is a distinct interface between the solute crys- sity transitions from zero to a finite value tal and solvent. Surface tension imparts an around − t . The model becomes more in- energy to the crystal-solvent interface. The 4u teresting when one recognizes that there is resulting crystal is only stable if the change also a long-range interaction due to the long in the Gibbs Free Energy is sufficiently large distance that oxygen can diffuse between pho- to balance the surface energy. Since the sur- tosynthetic clumps of oxygen-sensitive bacte- face energy scales with the surface area while ria. To model this behavior one must include the change in the Gibbs Free Energy scales a term of the form with the volume, these energies are balanced at a critical radius: small crystals redissolve Z −βH = dxdy ρ(x)g(jx − yj)ρ(y) (2) and large crystals grow [6]. We therefore sus- L pect that the increased reproductive rate in a microaerophilic environment is the analogy of where g is a positive function. the change in the Gibbs Free Energy while the The process of phase-separation was adverse effects of living on a clump's surface first described theoretically by Lifshitz, Sly- give rise to an effect similar to surface tension. ozov, and Wagner in the limit that nucleat- Thus far, the analysis has only drawn a pos- ing crystals do not directly interact and the sible analogy between phase separation and total amount of solute is constant. Since clump formation. This analogy is only useful the change in the Gibbs Free Energy scales if the predictions for the growth of crystals with the amount of super-saturated material during phase separation can give some intu- which is constantly falling out of solution to ition for the the evolution of a microbial mat. form crystals, the the length-scale where the At low concentrations of oxygen, bac- surface energy balances the Gibbs Free En- teria glide away from one another and the ergy increases in time. As the critical radius system relaxes to a low-density state. If, increases, the smallest crystals redissolve and however, oxygen levels are raised, there is recrystallize on larger crystals in a process some beneficial, short-range interaction that known as Ostwald ripening. In the limit of in- attracts the microbes into clumps with finite finite time there is only one, gigantic crystal. density. This process is analogous to crystal The theory predicts that the critical radius nucleation where bacterial density ρ is the grows with the cube-root of time [6]. If there order-parameter and the dimensionless oxy- is also a long-range repulsive force, the pic- gen concentration t is the control-parameter. ture changes slightly. The long-range interac- Although there is no obvious thermodynamic tion prevents the formation of crystals larger than some size that depends on the nature 3 of the interaction. In this case, rather than forming a single crystal, the system develops a multitude of small crystals with a charac- teristic size. Interactions between neighbor- ing crystals also impose a characteristic spac- ing between the crystals. In two dimensions, the crystals are typically arranged in a reg- ular hexagonal lattice; which maximizes the density of crystals subject to the constraint Figure 2: Clumps that grow on a thin biofilm that there is a fixed lattice spacing. This re- typically form a roughly hexagonal lattice. sult has been shown for systems with a va- The scale bar is one centimeter. riety of long-range interactions including the the Coulomb force [3] and more complected forces such as those that describe Langmuir monolayers [7].