Substrate regulation leads to differential responses of microbial ammonia-oxidizing communities to ocean warming

Distinctive temperature responses along a substrate gradient

Within the temperature range of ~14 to ~34 °C in our incubations, the observed AORs at the ambient substrate level (AOR_ambient, see Methods) varied over 3 orders of magnitude, from 0.5 to ~4000 nM d⁻¹, across a wide spectrum of ambient ammonium levels ranging from 14 nM to 96 μM (Fig. 1). Three different types of temperature response of AOR_ambient patterns at estuarine, shelf, and sea basin stations were observed: (I) a positive response with a T_opt of ≥34 °C (Fig. 1a, b); (II) a negative response, which has never been reported before, with a T_opt of ≤14 °C (Fig. 1d–f); and (III) a dome-shaped response with a T_opt of 20–29 °C (Fig. 1c, g–i).

The Type I pattern was observed at two of the three estuarine stations (JLR1 and JLR2, Fig. 1a, b) where ammonium concentrations were high (≥24 μM), and the AOR increased linearly as the temperature increased from 14 to 34 °C. In these cases, the T_opt was equal to or higher than the maximum experimental temperature of 34 °C (Fig. 1a, b). The Type II pattern was observed at the shelf stations (N1, M1, and M2), where NH₄⁺ concentrations ranged from 45 to 550 nM (Fig. 1d–f). In contrast to the Type I pattern, the T_opt of the Type II pattern was equal to or lower than the minimum experimental temperature of 14 °C, showing a continuously decreasing AOR as temperature increased. The Type III pattern was observed at station JLR3 (outer estuary), N2 (shelf), N3 and J1 (basin), for which the T_opt of the AOR varied from 20 to 29 °C, with rates decreasing toward both higher and lower temperatures (Fig. 1c, g–i). The NH₄⁺ concentrations of the Type III stations ranged from 14 to 5000 nM. Nevertheless, the highest T_opt values were observed at coastal sites with the highest ambient ammonium concentrations (Fig. 1).

Substrate regulates AOR and its thermal optimum temperature

For those stations with low ammonium concentrations, the AOR at in situ temperature increased when the substrate was enriched (AOR_enriched, additions of 2000 nM ¹⁵NH₄⁺) (Fig. 1f, i). Meanwhile, the T_opt of the AOR shifted significantly toward higher values (t test, p < 0.05; Fig. 1d, f, g, i). Although the resolution of the temperature interval set in our incubation experiments was not high enough to identify a precise T_opt for the AO community, the positive T_opt shift induced by ammonium enrichment was evident.

To further explore how the substrate-regulated T_opt of AOR in marine environments, we designed a Michaelis–Menten (M–M) thermal kinetics experiment for J1 (substrate deprived sea basin) and JLR4 (substrate-replete upstream estuary) stations with distinctive substrate concentrations (see details in Methods). The experimental results revealed that at a given temperature, the responses of AOR along with substrate addition can be fitted by the classic M–M curve (Fig. 2a, b; Supplementary Table 2), i.e. the rate increased as the NH₄⁺ concentration increased until the substrate became saturated. These M–M curves are temperature-dependent, with the maximum rate (V_max) and the half-saturation constant (K_m) increasing as the temperature increased until the saturation optimum temperature (T_opt-sat) was reached (Fig. 2c–f). Note that the T_opt-sat was defined as the optimum temperature of V_max at saturated substrate level (T_opt-sat, ~26 °C for station J1 and ~29 °C for JLR4 station; see Fig. 2c, d).

From the above results, we suggest that the T_opt-sat of a single-species derived from laboratory culture under a saturated substrate concentration¹⁸ may not properly represent its T_opt in the field, where ammonium is not always saturated. Some published biogeochemical models with a nitrification component have assumed that nitrification follows the Arrhenius relationship until temperature reaches the T_opt^23,24. However, the T_opt in these models is derived from pure laboratory cultures typically grown at saturating substrate concentrations (T_opt-sat), which may cause an overestimation of rates in the warming ocean at ambient unsaturated substrate concentrations.

To examine the interactive effects of temperature and substrate on AOR and T_opt, we adapted the Dual Arrhenius and Michaelis–Menten kinetics model (DAMM) developed by Davidson et al.^25,26. The temperature sensitivity of K_m (Q10_Km) and V_max (Q10_Vmax) (Fig. 2c–f) was applied. In this model, the rate was determined by four parameters, i.e. substrate concentration, temperature, Q10_Km, and Q10_Vmax (see Eq. (7) in Methods). By introducing the Q10_Vmax and Q10_Km values derived from station J1 and JLR4 into the DAMM model, we simulated how AOR responds to substrate and temperature synergistically (Fig. 3; Supplementary Fig. 2). Via the DAMM, we successfully predicted the AOR below the T_opt-sat under various ammonium levels (Fig. 3a, b). Moreover, we developed a T_opt model (see Eq. (9) in Methods) via the DAMM to see if we could predict the T_opt at various substrate levels. For the T_opt model, T_opt increased as substrate concentration increases under the criterion of Q10_Km > Q10_Vmax (Supplementary Fig. 2). This criterion was fulfilled in both J1 and JLR4 cases (Fig. 2c–f). We see the positive shift of T_opt due to the ammonium addition (up to ~100 nM at J1 station and up to ~10 μM at JLR4) can be closely predicted (Fig. 3c, d; Supplementary Fig. 2). Overall, the DAMM model successfully predicts the entire thermal response curve, including rates and T_opt, except when the manipulated temperatures exceed T_opt-sat (Fig. 3; Supplementary Fig. 2). AOR drops significantly when temperature is greater than the T_opt-sat, so heat-impaired biological enzyme activity^27,28 might result in deviations from the relationship between V_max (K_m) values and temperature from the Arrhenius law.

The substrate-dependent thermal optimum is attributable to the effect of temperature on biochemical kinetics and the structural stability of the enzymes. Increasing temperature promotes catalytic rate, thus, V_max increases due to increasing kinetic energy of reactants and rates of collision, as well as higher structural flexibility of enzymes^27,29. However, higher structural flexibility (lower stability) also results in active sites with a reduced ability for ligand recognition and binding, therefore, lower kinetic efficiency. Accordingly, one important physiological response of an organism to rising temperature will be a reduction in substrate affinity (Supplementary Table 2), and thus higher substrate demands (i.e. higher K_m value)^25,29,30,31. In other words, higher substrate levels help to compensate for enzyme structural stability losses and so promote growth rates at higher temperature. Note that some other microbes may respond differently to temperature, with Q10_Km ≤ Q10_Vmax for instance. This may lead to predictable yet unidirectional rate increases in response to warming (without substrate-regulated T_opt) until the T_opt-sat is reached, regardless of substrate changes.

Similarly, nutrient-dependent T_opt has been reported for phytoplankton growth in pure cultures previously. For instance, Thomas et al.³² indicated that the T_opt for growth of a marine diatom was a saturating function of major nutrient (nitrate and phosphate) concentration, and that the T_opt could decrease by 3–6 °C at low concentrations relative to that at saturated nutrient levels. In addition to studies of pure cultures, field studies have also suggested that organisms may tolerate higher temperature stresses when nutrients are more abundant. For example, kelp (Laminaria saccarina) with high nitrogen reserves have more capacity for thermal adaptation³³, while corals with symbionts limited by phosphate are more susceptible to heat-induced bleaching³⁴. Although these examples are functionally and taxonomically distant from AOA and ammonia-oxidizing bacteria (AOB), strong similarities in substrate/nutrient regulation characteristics may imply a similar mechanism of enzymatic thermal responses between chemoautotrophs and photoautotrophs.

Nevertheless, the higher thermal optimum of AO in the estuarine system (e.g., JRL1, JLR2, and JLR3) than in the offshore environment (e.g., N3 and J1) can be explained by a substrate-regulated T_opt. Note that field AOR represents explicitly the collective activity of the AO community composed of AOA and AOB, which may have distinctive thermal tolerances and affinities for substrate. Therefore, community structure very likely plays a role in modulating the thermal response patterns of community AOR in the field environments, in addition to substrate concentration.

Rate proportion and community thermal optimum

To further examine to what extent the community structure (proportions of AOA and AOB) might shape the thermal response patterns of community AOR observed in the field, we added allylthiourea (ATU) to inhibit the activity of AOB for rate discrimination (see Methods; Supplementary Discussions). Results showed that the inhibitory efficiency of AOR was gradually reduced with increasing offshore distance (Fig. 4). That is, from the estuary (JLR4, JLR1, JLR2, and JLR3) to the shelf (N1 and N2) and the sea basin (N3), the relative contribution of AOB to the community AOR dropped from as high as ~100% in the upper estuary down to ~70% in the shelf transition zone, and near 0% in the basin. Meanwhile, the AOA/AOB gene copies data (see Supplementary Methods) from estuary to sea basin (Fig. 4) also clearly show that the abundance of AOA relative to AOB increased exponentially with increasing offshore distance. A similar offshore pattern of community distribution was also observed in other regions, such as from the Pearl River estuary to the South China Sea³⁵, and from the freshwater region of the Chesapeake Bay to the coastal and open ocean water column³⁶. This pattern suggests that AOB strongly prefer substrate-replete niches, and vice versa for AOA^20,37, agreeing well with our M–M experimental data that the substrate saturation condition for AOB-dominated water at JLR4 was several orders of magnitude higher than that for AOA-dominated water from J1 (Supplementary Table 2). The K_m values of AOB in JLR4 varied from 7 to 55 μM in accordance with varying temperatures from 10 to 37 °C, while the K_m values of AOA in J1 varied from 13 to 44 nM over a similar temperature range (Supplementary Table 2). Results were supportive of previous pure culture and field studies which showed the minimum ammonium demand for AOB is >1 μM and K_m values range from 28 to 4000 μM^38,39,40,41, while minimum ammonium demand and K_m value for AOA are <10 and 133 nM⁴², respectively.

The low K_m for AOA indicates that the ambient substrate concentration can easily reach saturation state for AOA-dominated basin regions (see K_m values in Supplementary Table 2), thus, leading to a dome-shape thermal response pattern with the T_opt near to T_opt-sat likely around 26 °C (Figs. 1h, i, 3c), consistent with a report for an AOA isolate in lab culture¹⁸. Meanwhile, a T_opt-sat of ~30 °C (Figs. 1a, b, 3d) falls within the range of 29–35 °C reported for AOBs in pure culture^43,44. Since the AOB-dominated JLR4 case showed a very wide T_opt as substrate varied, the negative response patterns at temperatures between 10 and 30 °C can be seen when the substrate was diluted below the several μM level, mimicking the seaward advection of a water mass and subsequent substrate dilution. Accordingly, we speculate that the three types of AOR thermal response patterns observed in the field were a result of the combined thermal response patterns of AOB and AOA along the substrate gradient. Although this type of negative response pattern has never been reported before in the field, we observed it at shelf stations and in our dilution experiment. This suggests that when AOBs dominate the community AOR in the shelf transition zone (Fig. 4), community AOR may respond negatively to temperature rise due to a low T_opt under substrate stress. In fact, comparison of the community AOR between ambient conditions (20 nM ¹⁵NH₄⁺) and after significant tracer addition (2000 nM ¹⁵NH₄⁺), we can clearly see the magnitude of rate stimulation at in situ temperature was higher at the shelf station dominated by AOB (M2 station, Fig. 1f) than at the basin station (J1 station, Fig. 1i) where the AOR was dominated by AOA. Such distinct rate stimulations by substrate additions implicitly indicate the degree of substrate saturation was lower on the shelf, where AOB contributed more to the AOR.

Global change and future nitrification

Sea surface temperature (SST) will inevitably increase in most of the ocean as atmospheric greenhouse gas concentrations continue rising. According to the 2013 Intergovernmental Panel on Climate Change Representative Concentration Pathways (RCP) 8.5 scenario, mean global sea surface temperature (SST) will increase by up to 4 °C by the year 2100⁴⁵, equivalent to a rise of 0.04 °C/y. The seasonal temperature (1 m depth) ranges from 13 to 32 °C for the Jiulong River estuary⁴⁶ (Supplementary Fig. 1; Supplementary Table 3). The average SST rise over the period of 1960–2010 in the studied area was 0.02 °C/y⁴⁷, about half of the predicted maximum increase rate of the global mean. On the other hand, hydrography of the northern South China Sea is influenced by regional climate and local hydrodynamics, including the western boundary current intrusion, internal waves and monsoon winds⁴⁸. The water temperature of the upper 200 m in the northern South China Sea ranges from 13 to 31 °C at different depths^48,49 and the mean temperature rise for the entire 200 m water column in the studied area was reported to be 0.09 °C/y from 1975 to 2005⁵⁰. Generally speaking, the rise of water temperature in our study region is similar to or even larger than the global mean, suggesting this global warming rate can be applied in our study area (see Supplementary Discussion).

Using the thermal responses of kinetic parameters, we try to evaluate warming effects on the competition between AOB and AOA. The specific affinity (α), the ratio of V_max/K_m, can be used to represent a microbe’s ability to scavenge substrate from dilute environments. Thus, microorganisms with higher α values are superior competitors when substrate is limiting. We found the specific affinity was higher for AOA-dominated J1 than for AOB-dominated JLR4 (Supplementary Table 2; Fig. 5a). Moreover, α values of AOB vary in a narrower range relative to those of AOA as temperature changes. The less variable α may reflect that AOB cope better with variable temperature in their habitat (Supplementary Table 2; Fig. 5a). In contrast, AOA are more sensitive to temperature change. As aforementioned in the DAMM model, the criterion to have substrate-dependent T_opt is Q10_Km > Q10_Vmax, which also results in a reduction in α during warming for both AOA and AOB. Yet, the relative reduction in AOA specific affinity as temperature increases is more significant (Fig. 5a), suggesting AOAs are more competitive in low temperature environments relative to AOBs, and so may not be favored in a warming ocean. On the other hand, the specific affinity of AOBs is insensitive to temperature change, suggesting their adaptation to nearshore environments with greater temperature fluctuation. The seaward gradient in temperature fluctuations and ammonium concentrations determine the nitrifier community, thus, thermal response pattern of community AOR observed in the field.

To predict the trends of AOR in different geographic spaces in warming ocean, we compile the available marine AOR data to examine the AOR changes empirically. If we assume the biogeographic distribution of AO community remains unchanged and consider solely the warming effect on AOR relative to the onsite temperature, we found the thermal responses of AOR in nearshore and offshore are quite different (Fig. 5b). More specifically, the higher T_opt of these AO communities in nearshore regimes allows ocean warming to promote coastal AOR when the temperature change increment is <10 °C. The AOR then drops with further temperature increases, likely due to the impaired enzyme activity as discussed above. On the other hand, community AOR decreases linearly with warming (Fig. 5b) in offshore waters, mainly because current ambient temperatures are close to T_opt-sat for AOAs and/or higher than T_opt of AOB in low substrate state.

According to the mean global sea surface temperature increase of 4 °C by 2100, AOR might increase by 0.4–30% in nearshore systems, with a mean increase of 13% (Fig. 5b). However, at contemporary substrate levels the community AOR in oligotrophic environments would decline by 13–33%, with a mean decrease of 27% in response to the 4 °C warming in ocean temperature (Fig. 5b).

Our results also suggest that the projected gradual increase in atmospheric deposition of anthropogenic nitrogen (including ammonium)^7,51,52 may aid in the thermal adaption of ammonia-oxidizing microorganisms in ammonium-depleted offshore environments. However, most of the NH₄⁺ from the atmosphere will be utilized first by phytoplankton in the surface layer above the nitracline, due to a competitive advantage of phytoplankton toward NH₄⁺ relative to nitrifers^17,53. If this ammonium was to be supplied directly to the niche of AOA, the offshore AOR may response positively to warming before reaching the T_opt-sat. Note that AOAs prefer low temperature and cope with low substrate, so the peak AOR in open ocean generally appears near the nitracline¹⁷ where the vertical temperature gradient is also large. Therefore, the warming effect may also have a differential influence across the vertical scale.

In contrast, in the shallow nearshore dominated by AOBs additional inputs of anthropogenic NH₄⁺ cannot further promote T_opt-sat to alter the empirical trend. Thus, a positive temperature response in nearshore regimes is more certain. According to our observational data (Fig. 4), although AOB gene copies are much less, AOBs appear to be the major AOR contributor in the mid-to-lower estuary, coastal seas and even in shelf zones unless ammonium is down to approximately hundreds of nM level. Thus, we speculate the niche space of AOBs may expand in the future in the land-ocean transition zone, because of their greater thermal adaptability and continuously increasing anthropogenic nitrogen inputs from continents.

The observed differential warming effects on eutrophic nearshore systems and oligotrophic offshore regions would have significant implications for climate feedbacks of the marine nitrogen cycle over a wide trophic range. For example, AOR increases in the coastal ocean would subsequently promote N₂O emissions. In fact, field and laboratory culture studies showed that AOB have a greater potential than AOA to generate N₂O. The N₂O yield from AOB is approximately two times higher than that of AOA cultures⁵⁴. Thus, in the nearshore where AOB-dominated AOR, N₂O production may be further enhanced. This would serve as a positive feedback for global warming, while the suppression of AOR in offshore regions creates a negative feedback. On the other hand, during nitrogen recycling in turbid coastal and estuary systems, increasing AO would potentially enhance NO₂⁻/NO₃⁻ production to fuel denitrification in micro-niches⁵⁵, which is another important pathway for N₂O production. Thus, warming and excessive nitrogen input would further exacerbate the emissions of N₂O from coastal seas.

In addition to climate feedbacks, the impact of global warming on oceanic ammonia oxidation may also change the distribution of nitrogen species. The suppression of AO by ocean acidification⁸ was also reported to further exacerbate the inhibition of AO caused by warming. At the base of the euphotic zone in the stratified open ocean, this dual suppression of AO will substantially reduce the amount of NH₄⁺ converted back to NO₃⁻ by nitrification, ultimately favoring smaller primary producers that are more competitive for NH₄⁺ rather than NO₃⁻. Meanwhile, NO₃⁻-supported primary producers, such as large diatoms, would be at a competitive disadvantage. This might be disadvantageous for carbon export to the deep sea, and thus constitute a positive feedback to global warming.

Of course, multiple environmental factors such as acidification, stratification, deoxygenation, and UV light are changing simultaneously. To better predict how nitrification responds to interactive global change forcing, multiple factors should be included in experiments in the future. Moreover, both field and laboratory AO responses of the two steps of nitrification and associated N₂O yield should also be investigated at various levels of ammonium to assess their impacts on global warming. Meanwhile, genomic and proteomic information is urgently needed to determine the physiological mechanisms of the species-specific temperature responses of marine AO organisms. Besides nitrification, to improve model predictive ability and have in-depth understanding of the biogeochemical role of nitrogen in microbial ecosystems, the temperature sensitivity of kinetic parameters of associated nitrogen processes needs to be investigated. Moreover, it is important to carefully examine the extrapolation from short-term experiments, such as our hourly manipulations, to long-term microbial responses to environmental change (decades). Analogous long-term experiments (thousands of generation) for marine nitrifiers are needed to explore the possibility of evolutionary adaptation to simultaneous changes in ammonium supplies and temperature in the future ocean.

Source: Ecology - nature.com