Characterising and monitoring biological diversity to foster sustainable ecosystems is highly recommended as urban centres rapidly expand. However, much of New York City’s biodiversity remains undescribed, including in the historically degraded, but recovering Bronx River Estuary. In a pilot study to identify organisms and characterise biodiversity patterns there, 18S rRNA gene amplicons (V1–V3 region), obtained from river sediments and surface waters of Hunts Point Riverside and Soundview Parks, were sequenced. Across 48 environmental samples collected over three seasons in 2015 and 2016, following quality control and contaminant removal, 2,763 Amplicon Sequence Variants (ASVs) were identified from 1,918,463 sequences. Rarefaction analysis showed sufficient sampling depth, and community composition varied over time and by substrate at the study sites over the sampling period. Protists, plants, fungi and animals, including organisms of management concern, such as Eastern oysters (Crassostrea virginica), wildlife pathogens and groups related to Harmful Algal Blooms, were detected. The most common taxa identified in river sediments were annelid worms, nematodes and diatoms. In the water column, the most commonly observed organisms were diatoms, algae of the phylum Cryptophyceae, ciliates and dinoflagellates. The presented dataset demonstrates the reach of 18S rRNA metabarcoding for characterising biodiversity in an urban estuary.

eDNA, metabarcoding, New York City, next-generation sequencing, QIIME2, urban ecology

With the extensive modification of ecosystems by people and increasing urbanisation, the disrupted ecology of cities is garnering substantial research interest (Alberti 2008). Understanding how urban life forms are responding to habitat alteration, invasion by non-native species, pollution, human population growth, overexploitation, disease, climate change and/or interactions amongst these threats are key areas of urban ecology research. New York City is one of the world’s great metropolises, yet much of its biodiversity remains to be identified and described, especially the microfauna and microflora of its rivers and estuaries.

The once highly polluted, 23 mile-long (ca. 37 km), Bronx River is currently considered “impaired”, with pollutants including faecal coliforms, garbage, refuse, polychlorinated biphenyls (PCBs) and other toxins, coming from a combination of urban and stormwater runoff, Combined Sewer Overflow (CSO) outfalls, contaminated sediments, and other sources (BRA 2021; NYSDEC 2020). These waters have been impacted by invasive species, such as green (Carcinus maenas) and Asian shore (Hemigrapsus sanguineus) crabs. They are also affected by harmful algal blooms (HABs), including Gymnodinium dinoflagellates (Fuss and O’Neill 2015). Eukaryotic pathogens include Cryptosporidium parvum (an intestinal parasite that can affect humans), and the oyster pathogens Perkinsus marinus and MSX (Haplosporidium nelsoni). However, the recovering area now hosts several clean-up and revitalisation programmes, such as targeted restoration of American eels (Anguilla rostrata), river herring (Alosa pseudoharengus, A. aestivalis) and eastern oysters (Crassostrea virginica). In the Bronx River Estuary (Fig. 1), Hunts Point Riverside Park used to be an illegal garbage disposal site, but is now an integral part of the Bronx River Greenway (Kimmelman 2012). Soundview Park waters contain a successfully-restored oyster reef identified as a key research site (Grizzle et al. 2012).

Figure 1.

Sampling site map depicting greater New York City waterways. Inset: Detail map of sample area showing Soundview Park (SVP) and Hunts Point Riverside Park (HP) in bold outline. Right: site photos of the SVP and HP estuaries on the Bronx River. Map data 2019 (C) Google.

Although the first step in any ecological study is the correct identification of organisms in the focal system, there are numerous challenges in conducting biodiversity inventories (Bik et al. 2012; Bohmann et al. 2014; Taberlet et al. 2018; Deiner et al. 2021). Traditional surveys involving morphological classification and other techniques provide essential data, including abundance information. However, many taxa remain difficult to detect and describe due to a combination of their microscopic or cryptic natures, collection logistics, lack of taxonomic expertise and labour-intensive morphological assessments, leading to uncertain species identifications and biodiversity estimates. Rapid, effective and standardised approaches are needed to guide more detailed investments and cost-effectively complement morphological data for comprehensive management through improved biodiversity information (Bik et al. 2012; Bohmann et al. 2014; Rees et al. 2014; Taberlet et al. 2018; Deiner et al. 2021).

Recently developed, non-invasive and transformative environmental DNA (eDNA) metabarcoding technology provides volumes of data for biodiversity assessment via next-generation sequencing. DNA barcoding was the first worldwide effort to document life and identify species using genetic sequences from a standard DNA segment (Hebert et al. 2003). Approaches such as metabarcoding, a high-throughput extension of DNA barcoding, offer orders of magnitude more data than traditional morphological or DNA barcoding research. Multiple organisms can be identified simultaneously from genetic material extracted from environmental samples (e.g. water, air and sediment) by sequencing and analysing specific marker genes using primers that target their conserved flanking regions (Taberlet et al. 2018; Deiner et al. 2021).

Environmental DNA has numerous advantages, offers high-throughput presence, absence and relative abundance data, and can improve representation of microscopic or cryptic taxa (Bik et al. 2012; Bohmann et al. 2014; Taberlet et al. 2018; Deiner et al. 2021). Environmental DNA metabarcoding is low-impact, efficient, cost-effective, rapid and replicable. The method has been effectively used in estuaries (Chariton et al. 2010, 2015; Leray and Knowlton 2015; Taberlet et al. 2018; Afzali et al. 2021; Carraro et al. 2021; García-Machado et al. 2021) and has found previously undetected or poorly characterised organisms, in particular bacterial, protist and invertebrate taxa (Bik et al. 2012; Bohmann et al. 2014; Goldberg et al. 2015; Taberlet et al. 2018; Deiner et al. 2021; Leese et al. 2021). Laboratory, computational and data storage limitations exist and reference data for taxonomic assignment of many groups are lacking, but the methods are continuously improving (Valentini et al. 2016; Taberlet et al. 2018; Deiner et al. 2021).

One advancement that facilitates biodiversity assessment and monitoring is state-of-the-art bioinformatics pipeline development to perform quality-control and large-volume data analysis (Taberlet et al. 2018), such as for the 18S rRNA amplicon datasets presented here. Some of the advantages offered by 18S rRNA metabarcoding are broad amplification across eukaryotic kingdoms, a rapidly growing reference database due to wide marker use, as well as conservation within and divergence amongst species genetic profiles (Leray and Knowlton 2016; Taberlet et al. 2018). This marker was selected here to provide a broad overview of estuarine eukaryotic biodiversity, including microorganisms, other algae and invertebrates, that would mirror our prior 16S rRNA metabarcoding work on prokaryotes (Naro-Maciel et al. 2020). Within the 18S rRNA gene, several markers for metabarcoding are being used. For protist taxa, the V4 and the V9 regions are utilised especially often (Stoeck et al. 2009; Dunthorn et al. 2012; de Vargas et al. 2015; Boenigk et al. 2018). Here, the V1–V3 region was targeted due to the high phylogenetic resolution available using hypervariable segments V2–V4, previously demonstrated in dinoflagellates (Ki 2012) and copepods (Wu et al. 2015). Initial checks against published databases and preliminary laboratory tests supported our choice of the V1–V3 region for common taxa or those of management concern, including Eastern oysters (Crassostrea virginica) and HAB-related taxa. Thus, the 18S rRNA dataset, presented here, was used to identify organisms, explore biodiversity patterns and establish a baseline for future work in the Bronx River Estuary.

This pilot study identified key organisms, explored biodiversity patterns and established a baseline for future work in the area, but the data must be interpreted with caution considering methodological issues. In total 48 environmental samples were successfully collected and sequenced for the 18S rRNA gene (n_water = 23; n_sediment = 25). Within these samples, protists, plants, fungi and animals encompassing 2,763 ASVs were recovered from a total of 1,918,463 post-quality-control sequences (Suppl. material 2: Table S2). Species accumulation curves of each sample reached an asymptote, indicating that the communities were surveyed with sufficient depth to detect robust differences in community structure and composition (Suppl. material 2: Fig. S1). At the study sites over the sampling period, community composition varied over time and by substrate (Fig. 2).

Figure 2.

18S rRNA community profiles of Amplicon Sequence Variants (ASVs) in sediment and water samples from Hunts Point Riverside and Soundview Parks shown at the level of phylum. Bar heights show relative abundance of sequences from each taxon.

Several organisms of known occurrence, including taxa of management concern, were detected. Commonly observed species identified in this survey included soft-shell clams (Mya arenaria) and blue mussels (Mytilus edulis) (Suppl. material 2: Table S1). Oyster DNA (Crassostrea virginica) was detected in both Soundview waters and sediments, but not in Hunts Point waters or sediment. Similarly, oysters have been observed at Soundview, but not at Hunts Point; the oyster parasite genus Perkinsus was detected only in the water at Soundview Park. The oyster pathogen MSX (Haplosporidium nelsoni) was not identified at any site, but the crustacean parasite Haliphthoros was found in Hunts Point sediments. Dinophyceae dinoflagellates (Gymnodinium, Heterocapsa, Karlodinium) and Raphidophyceae (Heterosigma akashiwo), which may cause Harmful Algal Blooms (Hara and Chihara 1987; Faust and Gulledge 2002; Millette et al. 2015; Lin et al. 2018), were sequenced from the Bronx River Estuary. Macroinvertebrate taxa considered to be indicators of estuary pollution (Pelletier et al. 2010; Smith et al. 2015) were not commonly found, except for the various small aquatic worms (Nematoda), some of which are consistent with poor water quality (Fig. 2). American eels (Anguilla rostrata) and herring (Alosa pseudoharengus and A. aestivalis), key organisms being restored and monitored in the Bronx River, were not detected.

In terms of alpha diversity, Soundview Park sediments were significantly higher in the observed number of ASVs compared with all other sites (Kruskal-Wallis, p = 0.05, Fig. 3A). While Soundview water also trended towards higher biodiversity compared with Hunts Point water, the difference was non-significant. However, none of the sites differed significantly in Shannon Diversity (p > 0.05, Fig. 3B). Sediment communities between Hunts Point and Soundview were differentiated by the presence of several key taxa missing or less proportionally abundant at Hunts Point. For example, Soundview sediments had higher proportions of arthropod DNA detected than those at Hunts Point (Figs 2 and 3). In agreement with our results on overall alpha diversity metrics, Soundview Park sediments were more taxonomically diverse when compared with Hunts Point. Water samples from both sites were not apparently different in taxonomic composition (Figs 2–4). However, there were clear differences between the taxonomic make-up of sediments and water column samples, driven mostly by the more frequent detection of annelid worms and nematodes in sediments and larger proportions of diatoms, dinoflagellates and Protalveolata in water samples (Fig. 2). The community turnover (i.e. beta diversity) of eDNA from water samples was significantly different from that of sediment (r² = 0.169, P = 0.001; Fig. 4). Water samples were homogeneous amongst sites (r² = 0.069, P = 0.834). In contrast, sediment samples from Soundview Park were distinct from those at Hunts Point (r² = 0.245, P = 0.001; Fig. 4).

Figure 3.

Alpha diversity comparison between sediment and water samples from Hunts Point Riverside and Soundview Parks computed using A) Observed ASVs and B) Shannon Diversity. Pairwise comparisons are indicated as follows: * = p < 0.05, ** = p < 0.001, ns = non-significant.

Figure 4.

Community comparisons amongst substrates (sediment and water) and sites (Hunts Point Riverside and Soundview Parks), based on Amplicon Sequence Variants (ASVs) and using Non-metric multidimensional scaling analysis (NMDS).

Future metabarcoding work in the area would benefit from lessons learned during and resources developed since this pilot study. The high quality, comprehensive protocols now available to standardise and ensure eDNA metabarcoding excellence should be carefully followed (Taberlet et al. 2018; Minamoto et al. 2021). To better incorporate unicellular organisms and viruses, specialised methods, including use of filters with finer pore size, should be employed. While state-of-the art bioinformatics work conservatively identified and removed errors and contaminants, ongoing research would additionally benefit from stringent laboratory checks. These include the use of extraction blanks in a PCR-free lab, positive controls to assess amplification efficiency and negative controls to identify contaminants directly. Technical replicates address contamination, errors including rare taxa detection, and PCR amplification and sequencing variation. Further, although the V1 – V3 segment did capture known organisms and those of management interest, reference data bases for the increasingly used V4 or V9 regions may be more complete, thus resulting in more identifications. Finally, organismal abundance does not necessarily correlate with sequence abundance given amplification biases and errors such as primer-template fidelity and suboptimal annealing temperatures. Thus even inferences about relative abundance should be interpreted with caution (Fonseca 2018; Taberlet et al. 2018). Comparing metabarcoding results to conventional survey data will continue to be essential for ground-truthing and optimising both methods (Fediajevaite et al. 2021).

In conclusion, the 18S rRNA V1 – V3 dataset, presented here, complements our prior study, “16S rRNA Amplicon Sequencing of Urban Prokaryotic Communities in the South Bronx River Estuary”. Future work will comparatively analyse information from these two genetic regions and new data from Cytochrome Oxidase I, the standard locus for animal barcoding (Hebert et al. 2003). Despite its advantages, the 18S rRNA marker alone is insufficient to fully characterise biodiversity and a suite of markers would provide a more complete profile (Leray and Knowlton 2016; Taberlet et al. 2018) to further describe life in a complex urban estuary.

All 18S rRNA amplicon gene sequences from this study are posted on the NCBI Sequence Read Archive (SRA) under BioProject PRJNA606795 accession numbers SAMN19729835–SAMN19729882 (Table 1). All DNA extracts are stored at the American Museum of Natural History. Bioinformatics and statistical scripts are available as a supplement to this article (Suppl. material 1: Documents S1, S2).

The authors declare no competing interests.

We are grateful for the New York University Research Challenge Fund and NYU Liberal Studies New Faculty Scholarship and Creative Production Awards (to ENM) and to private donors through Experiment.com (to IW) for funding the research. Site access was provided by NY/NJ Baykeeper and the New York City Department of Parks and Recreation (Natural Resources Group). Special thanks are extended to student assistants Christian Bojorquez, NaVonna Truner, Sean Thomas, Jennifer Servis, Patrick Shea, Vanessa Van Deusen and Seth Wollney, as well as to Dr .Brendan Reid.