At the intermediate harvest, significantly different from the controls, plants from the low MWCNT treatment had no measurable N2 fixation potential . At the final harvest, N2 fixation potentials were significantly lower than the control levels for both the low and medium CB treatments as well as for the low MWCNT treatment . Within the CB treatments without including the control, the final N2 fixation potential increased linearly with CB concentration , indicating that the CB inhibition of N2 fixation potential was mitigated at higher CB concentrations. The whole-plant N2 fixation potential was calculated as the product of the dry nodule biomass per plant and the N2 fixation potential normalized to dry nodule biomass.The result represents both the total amount of nitrogenase that has been synthesized and the specific nitrogenase activity and therefore provides an overall assessment of N2 fixation capacity on a whole plant basis. At the intermediate harvest, across all CNM treatments except for the medium GNP treatment, the whole plant N2 fixation potential appeared diminished relative to the control by 30% or more . The inhibition appeared to be 90% or more for the low and medium CB treatments and was 100% for the low MWCNT treatment . At the final harvest, all CNM treatments appeared to reduce the whole-plant N2 fixation potential by 46% or more relative to the control . Whole-plant N2 fixation potentials in both the low and medium CB treatments and in the low MWCNT treatment were less than 10% of the control level ; in the medium and high MWCNT treatments,macetas por mayor values were less than 28% of the control level .

Within the CB treatments, without including the control, the final whole-plant N2 fixation potential appeared to increase linearly with CB concentration , although the differences between the low and medium CB concentrations did not appear to be significant . Kapustka and Wilson36 previously reported a compensatory relationship between the specific nitrogenase activity and dry nodule biomass per plant: the reduction in soybean N2 fixation potential was offset by an increase in total dry nodule biomass, resulting in a higher final yield. However, we did not observe such a compensatory mechanism here as both the final total dry nodule biomass and the final N2 fixation potential appeared reduced in most CNM treatments relative to the control . This compounded each effect, resulting in relatively greater diminishment in the final whole-plant N2 fixation potential . TEM analyses were performed at the final harvest to examine the potential ultra structural changes and CNM accumulation inside soybean root nodules. The low CNM concentration treatments were prioritized, since they appeared more impactful than higher CNM concentration treatments. For control nodules , infected nodule cells were densely packed with symbiosomes, inside which bacteroids were present and surrounded by the symbiosome membrane.Electron-translucent granules inside bacteroids were likely poly that had accumulated as carbon reserves during active photosynthesis, for fueling N2 fixation under low carbon availability periods.In the low CB and low MWCNT treatments, nodules appeared comparatively empty and had atypical large vacuoles.The bacteroid density inside nodules, determined by quantitative analysis of nodule TEM images , was significantly lower in the low CB and low MWCNT treatments than in the control .

This lack of bacteroids corroborates the findings herein that nodules in the low CB and low MWCNT treatments had significantly lower final dry biomass per nodule and final N2 fixation potential than the controls . Densely packed bacteroids were evident in the low GNP treatment . The bacteroid width measured from nodule TEM images of all three low CNM treatments differed significantly from that of the control . Further, black particles with sizes comparable to that of dry CB powder were observed in both the symbiosome and nodule cell cytoplasm , indicating uptake of CB into nodules. The accumulation was also apparent for GNPs : both single and aggregated structures with morphologies similar to that of GNP dry powder were observed inside nodules. The CNMs accumulated inside nodules were not likely to interfere with the N2 fixation potential measurement . Additionally, in all three CNM treatments, putative starch granules, indicative of low N2 fixation activity,appeared inside nodule cells . Where they occurred, reductions in nodule count, dry nodule biomass, and N2 fixation potential in the CNM treatments would indicate that CNMs had negatively impacted many steps during root nodule development.The possible explanations include toxicity to either the soybean plant or N2 fixing bacteria, or both. The low MWCNT treatment stunted soybean stem and leaf growth, and reduced the final dry root biomass. Plants from this treatment also formed fewer nodules with the least dry biomass and the lowest N2 fixation potential. As such, toxicity to the plant could have inhibited nodulation and N2 fixation potential in the low MWCNT treatment.

Previously, CeO2 nanoparticles caused soybean leaf oxidative stress and damage; these effects corresponded with root nodule N2 fixation potential diminishment, with the explanation that the plant preferentially invested energy aboveground rather than to nodules below ground.Phytotoxicity was reported to reduce nodule count in the Medicago truncatula–Sinorhizobium meliloti symbioses exposed to nanoparticle-containing bio solids; multiple nodulation-related genes of M. truncatula were down regulated in the nanoparticle treatment relative to those in either the control or the treatment with bulk/ionic forms of the metals.Alternatively, Bradyrhizobium japonicum could have been negatively affected by CNMs in the soil.The toxicity might manifest as killing, thus lowering the abundance of infecting bacteria. There could have also been altered bacterial metabolism with impeded signal communication with soybean plant roots. CNMs might have damaged bacterial cell membranes, hence rendering bacteria incapable of attaching to root hairs to begin the infection process.Some defective bacteria might enter an infection thread but would not proliferate normally; they might also become trapped, thus disabling the infection thread.Empty nodules could form if stressed B. japonicum did not penetrate nodule cells to form symbiosomes and differentiate into bacteroids, even though early nodule establishment was successful. Additionally, CNMs could have sorbed onto B. japonicum, thus allowing for CNM-coated bacteria to become engulfed in the infection thread. In such cases, B. japonicum might have carried CNMs into nodule cells, during which B. japonicum might have been continuously exposed to CNMs. The accumulation of starch granules in CNM treated nodules suggests that bacteroids could be defective, that is, either not fully differentiated with normal N2 fixation capability or having defective metabolism.As a result of the low N2 fixation efficiency, soybean plants may have responded by reducing carbon supply to these nodules and further suppressing N2 fixation.Besides possibly being transported by B. japonicum, CNMs could also have been taken up by nodule cells directly. The pore size of plant cell wall is commonly estimated to be 5–20 nm,nft hydroponic and hence nanomaterials larger than 20 nm would be excluded.However, new pores with larger diameters might form on plants to facilitate nanomaterial uptake.It has been reported that the pore size can be as large as 50 nm, as evidenced by the observation of gold nanoparticles inside tobacco plants.In the present study, CB nanoparticles had a size of 36.6 ± 8.3 nm , which suggests CB could cross plant cell walls. MWCNTs could also be taken up when oriented perpendicularly to the cell surface since the outer diameter was 18.8 ± 4.1 nm . Previously, both individual and aggregated MWCNTs were found to be piercing wheat roots and entering the root cell cytoplasm in a hydroponic study,while direct MWCNT uptake from soil into a root cell of red clover was observed by TEM.The apparent accumulation of GNPs inside nodules is also interesting, as GNPs have a thickness of 8–12 nm but a larger diameter . It could be possible that GNPs might have folded into sheets of smaller diameters, or there might be additional internalization pathways for GNPs. In a previous hydroponic study, graphene oxide was demonstrated to accumulate inside root cells of Arabidopsis plants.

In summary, CB, MWCNTs, and GNPs negatively affected soybean nodulation and N2 fixation potential, particularly at lower concentrations. Nanomaterials appeared to accumulate inside nodules in the low CB and low GNP treatments, suggesting direct CNM interference with N2 fixing symbioses. To test the role of CNM agglomeration in the observed inverse dose–response relationships, we studied CNM agglomeration and sedimentation in aqueous soil extracts . We used two CNM concentrations for CB, MWCNTs, and GNPs, to investigate the effect of CNM concentration on CNM colloidal stability, and thus on CNM bio-availability. At the higher CNM concentration , all CNMs agglomerated rapidly after CNMs were mixed into the soil extract, with the hydrodynamic diameter increasing to several micrometers; less than 22% of the nanomaterials remained suspended after 2 h . After 12 h, more than 90% of CNMs at 300 mg L−1 settled out as large agglomerates . At the lower CNM concentration , negligible agglomeration and sedimentation were observed for all CNMs in the soil extract in the first 2 h, although over the following 56 days there was gradual formation of smaller aggregates that slowly settled out . Across CNMs, at the 10 mg L−1 concentration, CB was significantly more stable in the soil extract than the other two CNMs, according to normalized nanomaterial suspension absorbances and hydrodynamic diameters . After 7 d, CB was more well dispersed, with a smaller average hydrodynamic diameter as compared to MWCNTs and GNPs in the soil extract suspension . This was also supported by the zeta potential and electrophoretic mobility measurements, in which 10 mg L−1 CB had a more negative ζ potential and EPM than either the 10 mg L−1 MWCNTs or the 10 mg L−1 GNPs . Environmental scanning electron microscopy was performed to observe the agglomerate morphologies of 10 and 300 mg L−1 CNMs in the soil extract upon deposition onto clean quartz sand . At the lower concentration , CB, MWCNTs, and GNPs appeared dispersed, with their surfaces apparently covered by a thin layer of organic matter from the soil extract . By contrast, at the higher CNM concentration , large agglomerates, with sizes of several micrometers, were observed . Specifically, CB formed loose appearing agglomerates composed of smaller aggregates, and MWCNTs were entangled together resulting in porous agglomerates, but GNPs appeared to stack together and were embedded in polymeric material, forming more compact agglomerates. The magnitude of CNM effects on soil microorganisms and plants depends partly on CNM bio-availability, which in turn is affected by CNM dispersal in soil pore water.Soil pore water is often rich in dissolved organic matter,which can improve CNM dispersal in the aqueous phase.However, the cationic content of soil pore water would shrink the electrical double layer and thus diminish electrostatic repulsion. As van der Waals forces increase relative to repulsive forces, CNMs agglomerate.Naturally occurring colloids in soil pore water could also affect CNM dispersal by heteroaggregation.To explain the observed inverse dose–response relationships , we hypothesized that, as CNM concentration increased, nanomaterials agglomerated into larger structures with relatively low bio-availability in soil. Such agglomeration-induced decreases in CNM bio-availability led to decreased CNM effects on plants and root symbioses at higher CNM concentrations. Through the agglomeration study, we confirmed that CNM agglomeration increased with increasing CNM concentration in soil water extracts. The differing CNM colloidal stability across the two CNM concentrations may be attributed to soil physicochemical properties , in particular, ionic strength.In deionized water , electrostatic repulsion dominated between CNMs with diminished van der Waals attraction, resulting in relatively stable CNM suspensions.For each CNM type, CNM hydrodynamic diameters in deionized water immediately following sonication were similar across CNM concentrations, averaging 204 ± 8 nm , 333 ± 27 nm , and 242 ± 9 nm across 10 and 300 mg L−1 . However, in the soil extract , when the van der Waals forces were stronger than repulsive force due to electrical double layer shrinking, CNMs agglomerated; the degree of agglomeration increased with CNM concentration.Additionally, at the higher CNM concentration , the collisions between particles increased,and the effect of dissolved organic matter on improving CNM dispersal was decreased.Collectively, the effect of CNM concentration on CNM agglomeration was exacerbated in the soil extract. It is worth noting that CNM agglomeration dynamics could be more complex in situ in soils, particularly in the rhizosphere due to root uptake, rhizospheric exudation, and microbial processes.For example, plant roots can release many lowmolecular-mass compounds that can affect soil pH and CNM agglomeration.