Constant Flux of Spatial Niche Partitioning through High-Resolution Sampling of Magnetotactic Bacteria

Figure 1 shows box plots of the total population of the three MTB morphotypes in the two aquaria over time (data in Table 1). Between days 7 and 37, the total quantities of MTB spirilla and MTB rods decreased, whereas the total quantity of MTB cocci slightly increased. After day 67 and until day 150, numbers of all counted MTB morphotypes increased, with the exception of MTB rods in aquarium A. A rotation of aquarium A at day 44 was done to test the influence of the starting conditions relative to magnetic north.

Figure 2 shows the horizontal distribution of the three dominant MTB morphotypes in each aquarium over time. The plots contour the number of bacteria at each site relative to the average value for the morphotype at the given day of measurement separately for each aquarium. Initially (day 7), MTB cocci and MTB rods concentrated toward the north, which might be due to mixing and pouring the sediment into the aquaria in the ambient, northward-directed field. On day 37, MTB cocci began to grow at the corners of the aquaria. Population densities remained stable for MTB cocci from day 7 through day 120 in aquarium B and initially for MTB rods in both aquaria over days 7 to 67. This is supported by a Pearson correlation analysis where significance is indicated at probability levels of 0.05 and 0.01 (Table 2). Interestingly, when MTB cocci had higher correlation values between successive measurements, there was no correlation in the same period for MTB spirilla and vice versa.

In most other cases, it appeared that the relative number of MTB morphotypes did not increase or decrease at the same places, but rather maxima (above-average numbers in red in Fig. 2) and minima (below-average numbers in green in Fig. 2) of populations varied over time. On the other hand, the relative variability (single standard deviation/mean) tended to stabilize after day 37 for MTB cocci and remained fairly constant for MTB rods and MTB spirilla, although the relative variability for the latter could be argued to have slightly increased over time (Table 1). In general, the data suggest that the MTB populations observed were relatively copiotrophic: changes in a population tended to boom or bust heterogeneously in space despite changes in absolute numbers. It is ambiguous how to interpret MTB populations at a given site relative to an average when the relative variability exceeds 1.0. Bootstrap analysis suggests that the uncertainty about the relative variability remains stable to within 20% until the number decreases from 36 to 20 sites. This should be interpreted with caution, as the size of the population generally increased for all MTB morphotypes except for MTB rods in aquarium A (see below), so the populations are not at a steady state.

Toward the end of the experiment, MTB spirilla generally inhabited the perimeter of the aquaria, while MTB cocci populated more toward the center. The finding that the two morphotypes consistently preferred to live in distinctly separate places was one of the most important observations of our study. One problem with the way in which the changes in MTB were visualized in Fig. 2 is that the color scale is optimized for relatively high cell counts. One sees a different view when doubly normalizing the data (see Fig. S3 in the supplemental material), which is exaggerated when cell numbers are low, as for MTB rods, yet has the advantage of weighting each morphotype and time equally.

We collected sediment from a natural pond situated in a forest near Landshut, Germany, 80 km northeast of Munich (48°35′14.98″N, 12°04′43.71″E) that hosts abundant and well-characterized MTB (21, 36). Grain size analyses of five sediment samples showed log-normal distributions with maxima at 15 to 23 μm, whose dominant magnetic contribution comes from noninteracting single-domain particles, probably in the form of intact magnetosome chains (37). On 14 July 2014 (considered day 1, the start of the experiment), we took 4 liters of sediment from the sediment-water interface to a depth of 10 cm, thoroughly homogenized it, and then poured it into two single-piece (seamless) glass aquaria 30 cm long, 20 cm wide, and 15 cm high. The seamless glass aquaria were gas impermeable, and thus, the only point at which oxygen could diffuse into the system was from dissolved O₂ in the water above the sediment in the aquaria. This preserved the oxic-anoxic redox gradient that naturally occurs in these sediments at a relatively shallow sediment depth (see Fig. S1 in the supplemental material). The sediment-water interface in the aquaria was at a 5-cm height, with the rest being filled with pond water up to 4.5 cm below the top of the aquarium. Constant water height was maintained throughout the course of the experiment.

The two replicate aquaria (aquaria A and B) were placed with their long axes parallel to the ambient geomagnetic field declination in a room where the temperature remained within 20°C to 26°C. Plastic boards with small holes covered the aquaria to reduce evaporation. As the aquaria were not air tight, their covers inhibited contamination from aerosols in the laboratory. However, the environment was not exceptionally dissimilar to the natural state of the shallow pond, which also received a constant influx of aerosols from the suburban Munich area. A total of 36 sites in each aquarium, spaced in rows every 4 cm in the long-axis direction and 3 cm in the other, were sampled at roughly 30-day intervals beginning on day 7 by using a pipette to extract 200 μl of sediment at each site from the upper centimeter of the sediment column, where the MTB were most concentrated (18). A mark indicating the 1-cm level on the pipette was visually placed at the sediment-water interface to within an accuracy of ±1.5 mm. Each site was defined as a 1- by 1-cm² area divided into two rows of three points whose position from one sampling to the next was systematically shifted to avoid disturbance: first in the upper left-hand corner, then in the upper right-hand corner, then between the two upper corners, then in the lower left-hand corner, then in the lower right-hand corner, and finally between the two lower corners, always the same for each site. On day 44, aquarium A was rotated 180° in the long-axis direction.

We used a viable cell counting technique to quantify the MTB morphotypes (16), which allows one to microscopically quantify each MTB morphotype without using more sophisticated methods such as quantitative PCR (qPCR) (29), etc. High-throughput 16S rRNA gene sequencing (Illumina) of three replicate samples from the pond sediment identified uncultivated taxa corresponding to Magnetococcus and the rod-shaped organisms “Candidatus Magnetobacterium bavaricum” and “Candidatus Magnetobacterium casensis” (38), similar to MTB recovered in other freshwater sediments from Germany, China, and Russia, etc. (18, 25, 39, 40). The sediment was transferred to centrifuge tubes, 4,800 μl distilled water was then added to 200 μl of sediment (5 ml total), and the mixture was homogenized by shaking. Although toxic on longer time scales, the use of distilled water is an advantage when observing MTB under a microscope in that it activates their propulsion. From day 120, the MTB population became so large at certain locations that we had to dilute 200 μl sediment with 7.25 times more water for a total slurry of 35 ml. From the centrifuge tubes, 10 μl of the sediment-water slurry was placed onto a glass slide, and 10 μl of distilled water was added beside the drop (Fig. 5). An O-ring was placed on the slide, and a coverslip was placed on the O-ring to inhibit evaporation. The slide was placed beneath an optical microscope in the center of Helmholtz coils (Petersen Instruments magnetodrome). Directing a horizontal magnetic field oriented toward the drop of distilled water induced the MTB to swim out of the sediment and into the clear water, where they were identified and counted via light microscopy (model CKX41SF; Olympus) at the air-water interface. The strength of the magnetic field produced by the coils was 0.6 mT, about 12 times the strength of the magnetic field in Munich.

The viable cell counting technique allowed us to distinguish three general morphotypes, cocci (defined as round MTB with diameters of ca. 1 to 4 μm), spirilla (generally 2 to 3 μm long and slightly concave, which included vibrios, as the two morphotypes are difficult to distinguish), and rod shaped (“Candidatus Magnetobacterium bavaricum” and/or “Candidatus Magnetobacterium casensis,” here grouped and referred to as MTB rods), which are easily distinguished by their rod shapes and dark appearance (see Fig. S2 in the supplemental material) (41, 42). Several different coccus and/or spirillum species exist in the sediment, but they cannot be distinguished optically and are thus grouped solely by morphotype. Figure 6 shows the number of MTB counted from 10 different samples over time. From these data, we allowed the MTB to swim in an induced magnetic field for 20 min before the number of MTB in each sample was counted. The number of counted bacteria was a minimum number since we cannot be sure that all of the MTB migrated out of the sediment.

Graphs of MTB distributions were made with Surfer 12 (Golden Software), selecting the kriging method to contour the data (43). Plots in Fig. 2 were made by calculating the average number of MTB of each morphotype from the 36 points in each aquarium and then subtracting the number of MTB morphotypes at each sampling position by the average number of MTB of that morphotype in the same aquarium for the day in question (e.g., the numbers are relative to the average, with red indicating values that are higher than average and green indicating values that are lower). Fig. S3 in the supplemental material shows the same data except doubly normalized so that the minimum counts are 0 and the maximum counts are 1. To test the effect of the initial conditions on the MTB population, aquarium A was rotated 180° on day 44. The maps of aquarium A are shown in the same position relative to the start of the experiment (Fig. 2). It took 6 days to count the MTB at the 72 sites (36 sites and 2 replicate aquaria), 3 days for each aquarium, with the number of MTB in aquarium A being counted first and the number in aquarium B being counted second; “day” in our study is according to the first day when counting began for aquarium A.

Magnetotactic bacteria are microaerophilic and sensitive to O₂; thus, they prefer to live near the oxic-anoxic interface in sediments where O₂ levels are low (44). For this reason, O₂ profiles were measured at nine locations in each aquarium by using a Unisense OX50 oxygen microsensor (tip diameter of 50 μm) with a 0.3-μmol/liter detection limit on day 120. The sensor was fixed on a computer-driven micromanipulator mounted onto a heavy laboratory stand. Three seconds were required to reach stable readings. Three north-south transects were spaced at 6-cm intervals, 8 cm apart, on the east-west axis. Measurements were made for every 1-mm depth, starting 5 mm above the sediment-water interface and ending 10 mm below it. Oxygen concentrations in water ranged from 100 to 150 μmol/liter, while O₂ disappeared at a depth of 3 mm in the sediment (see Fig. S2 in the supplemental material).