Comment on “Graphene oxide regulates the bacterial community and exhibits property changes in soil” by J. Du, X. Hu and Q. Zhou, RSC Advances, 2015, 5, 27009

Abstract

Human society is dependent on the provision of ecosystem goods and services mediated by soil organisms. It is important, therefore, to determine whether new and emerging materials influence the structure and function of soil biological communities. A recent study concerning the fate of the emerging nanomaterial, graphene oxide, in soil concluded that it promotes bacterial diversity. Here we demonstrate that due to issues with their experimental design and data analysis, their claim is unjustified.

---

Graphene exhibits a range of desirable properties that herald its widespread use in electronics and other industries. Consequently, its environmental release is predicted to increase along with the need to understand its ecological effects. Recently, Du et al.¹ investigated the fate of pristine graphene oxide (PGO) in soil. They concluded that the addition of PGO to soil promotes bacterial diversity; however, due to issues with their experimental design and data analysis, we believe this claim to be unjustified.

To appropriate any scientific observation it is critical to ensure proper replication of experiment treatments. While Du et al.¹ replicated (n = 3) their control and PGO-amended soils (PGOS) for chemical analyses, they pooled the replicates within treatments before extracting DNA and sequencing 16S rRNA gene amplicons to characterise bacterial diversity. Their data preclude, therefore, the necessary statistical analyses to determine whether PGO-amendment of soil significantly influenced bacterial diversity.

An additional problem with their interpretation of the data, results from the fact that the number of sequences obtained per sample was insufficient to exhaustively account for all taxa present. While this is normal for diverse communities such as soil bacteria, it is important when dealing with incomplete inventories to use an equal number of sequences per sample to calculate and compare diversity, otherwise it is not apparent whether samples with larger numbers of taxa are genuinely more diverse or were simply characterised in more depth. Du et al.¹ obtained 15 651 and 17 882 sequences for the control and PGOS samples, respectively. As they did not normalise the number of sequences prior to comparing bacterial diversity, it is not surprising that they observed slightly more taxa in the more deeply sequenced PGOS sample. For example, in the control and PGOS samples, respectively, there were 691 ± 3 vs. 702 ± 5 observed operational taxonomic units (OTUs) and 713 ± 6 vs. 731 ± 9 predicted OTUs (Chao1). With diversity estimates this close, it is likely that a different conclusion could have been drawn if the number of sequences per sample had been normalised.

On a less technical note, comparable nanomaterials, such as carbon nanotubes and fullerenes are predicted to accumulate in soil at a rate of ca. 1–100 ng per kg per year.² We question, therefore, the ecological relevance of the dose of 5 g PGO per kg soil applied in the Du et al.¹ study. While using this high dose facilitated extraction and chemical characterisation of graphene oxide particles from soil post-incubation, we argue that from a biological perspective it renders any observable effects irrelevant and unhelpful in the context of environmental impact assessment. Indeed, the authors themselves referred to the concentration of 5 g kg⁻¹ as “very high”. Lastly, Du et al.¹ did not include a treatment in which soils were amended with a ‘non-nano’ carbon allotrope, such as graphite. These non-nano control treatments allow for the assessment of any nano-specific effects,³ although identification of an appropriate bulk compound can sometimes be difficult. However, we contend that this is important considering that one of the aims of their study was to “assess the risks of nanomaterials”.¹

While the effort by Du et al.¹ to improve understanding of the fate of graphene in soil is laudable, we believe that their experimental design and analysis renders their study inconclusive and their claims unjustified.

References

1. J. Du, X. Hu and Q. Zhou, RSC Adv., 2015, 5(34), 27009–27017.
2. F. Gottschalk, T. Sonderer, R. W. Scholz and B. Nowack, Environ. Sci. Technol., 2009, 43(24), 9216–9222.
3. E. J. Petersen, Control experiments to avoid artifacts and misinterpretations in nanoecotoxicology testing (NIST.SP.1200-11), National Institute of Standards and Technology (NIST), Gaithersburg, Maryland, 2015, p. 10.

---