https://orcid.org/0000-0002-5339-0722
Figures
Abstract
Non-native species have the potential to cause ecological and economic harm to coastal and estuarine ecosystems. Understanding which habitat types are most vulnerable to biological invasions, where invasions originate, and the vectors by which they arrive can help direct limited resources to prevent or mitigate ecological and socio-economic harm. Information about the occurrence of non-native species can help guide interventions at all stages of invasion, from first introduction, to naturalization and invasion. However, monitoring at relevant scales requires considerable investment of time, resources, and taxonomic expertise. Environmental DNA (eDNA) metabarcoding methods sample coastal ecosystems at broad spatial and temporal scales to augment established monitoring methods. We use COI mtDNA eDNA sampling to survey a diverse assemblage of species across distinct habitats in the Salish Sea in Washington State, USA, and classify each as non-native, native, or indeterminate in origin. The non-native species detected include both well-documented invaders and species not previously reported within the Salish Sea. We find a non-native assemblage dominated by shellfish and algae with native ranges in the temperate western Pacific, and find more-retentive estuarine habitats to be invaded at far higher levels than better-flushed rocky shores. Furthermore, we find an increase in invasion level with higher water temperatures in spring and summer across habitat types. This analysis contributes to a growing understanding of the biotic and abiotic factors that influence invasion level, and underscores the utility of eDNA surveys to monitor biological invasions and to better understand the factors that drive these invasions.
Citation: Duprey J, Gallego R, Klinger T, Kelly RP (2023) Environmental DNA reveals patterns of biological invasion in an inland sea. PLoS ONE 18(12): e0281525. https://doi.org/10.1371/journal.pone.0281525
Editor: Hideyuki Doi, University of Hyogo, JAPAN
Received: January 24, 2023; Accepted: November 28, 2023; Published: December 27, 2023
Copyright: © 2023 Duprey et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The minimal underlying dataset and complete analysis code are available at the following public Github repository: https://github.com/jdduprey/patterns_of_invasion. The version of code and data that appears in this publication has been cataloged on zenodo: https://zenodo.org/record/8436270 with the accompanying DOI: 10.5281/zenodo.8436270.
Funding: The authors received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Invasive marine species impose increasingly high economic and ecological costs [1, 2]. They threaten fisheries, aquaculture, and marine recreation [3–5], disrupt food-webs, alter habitat structure, and displace native species [6–8]. Climate change, coastal eutrophication, aquaculture, and maritime trade all facilitate the spread and establishment of non-native species [9–13].
Prevention and early detection of biological invasions leads to the most successful economic and ecological outcomes [14–16]. Towards this goal, the sampling of environmental DNA (eDNA) has shown promise when applied to the detection of non-native species [17]. Targeted quantitative Polymerase Chain Reaction (qPCR) methods have successfully revealed the presence of individual species of interest, including the terrestrial toad species Bufo japonicus formosus in Hokkaido, Japan and invasive crayfish in Baden-Württemberg, Germany [18, 19]. Data collected using qPCR has motivated management and legal action on invasive carp in the Great Lakes region of the United States [20]. Sampling and analysis of eDNA data can be used independently or to complement established monitoring methods such as visual surveys. For example, data from eDNA sampling was paired with conventional trap data to better quantify the invasion front of European green crab (Carcinus maenas) in the Salish Sea, Washington [21].
At the level of ecological communities, metabarcoding reflects a subset of species present. In the Celtic Sea and the Arctic Ocean, this technique has been used to detect multiple non-native species in water samples collected across a large spatial scale [22, 23]. Metabarcoding has further been employed to monitor the shifting ranges of harmful marine microalgae [24].
In invasive-species research, metabarcoding methods offer advantages of scalability and replicability, at comparatively low cost [25]. By using common "universal" primers such as the 313 COI fragment we use here [26], metabarcoding can sample species across the tree of life that are present at a particular location and point in time. Each metabarcoding assay will depict a sample of the underlying biodiversity [27], and although (as with any sampling method) the species detected are subject to various biases, these biases are driven by primer-template chemistry, and so are unlikely to systematically discriminate between non-native and native species. As such, amplicon sequencing is a powerful way to investigate a host of questions about ecological invasion, including those that require simultaneous data on native and non-native assemblage composition and species richness.
A foundational question in invasion ecology concerns the biotic and abiotic factors that make habitats vulnerable to invasion [28, 29]. Significant progress has been made towards answering this question via visual sampling. Surveys in Northern Europe and San Francisco Bay have shown that relative non-native richness (= non-native species richness / total species richness) tends to be higher in estuarine habitats of meso- and polyhaline salinties (~ 5–25 ppt) than in either fresh and oligohaline habitats (< 5 ppt), or more marine polyhaline and euhaline habitats (> 25 ppt) [30–33]. A global literature review of invertebrate invasion data [31] showed that, with few exceptions, temperate estuaries are more invaded than adjacent open coasts. Proposed explanations for this phenomenon include: (1) most ports occupy mesohaline and polyhaline areas of estuaries; (2) species adapted to mid-range salinities are more likely to survive the variable salinities and temperatures incurred during transport in ballast or as hull biofouling; (3) mid-salinity regions face propagule pressure from both salt-tolerant freshwater species and freshwater-tolerant estuarine species; and (4) mesohaline and polyhaline waters have lower native species richness than either fresh or euhaline waters [33–35].
Ecosystem-scale metabarcoding surveys provide a new lens through which to examine invasion ecology. In particular, this data source can contribute substantially to the debate over biotic resistance—the degree to which native species richness and abundance determines habitat invasibility. Ecologists have hypothesized that biotic resistance works via multiple mechanisms including competition, disease, and predation [36]. Resistance can result from a single key native species, or from the sum of resource utilization dynamics among the entire native assemblage [29]. Experimental, mesocosm, and observational studies have reached conflicting conclusions about the net effect of native species richness on habitat invasibility [37]. For example, the non-native zooplankton species Daphnia lumholtzi was more likely to be present in zooplankton mesocosms with higher native richness [38]. Conversely, in seagrass mesocosms, the biomass of non-native invertebrates was shown to have an inverse relationship with native invertebrate richness [39]. Observational field surveys have similarly reached mixed conclusions [40]. The net effect of native biotic resistance remains unclear, as does its relative importance compared to abiotic factors, the characteristics of potential invaders, and propagule pressure. In human-impacted estuarine ecosystems, aquaculture is an important source of propagule pressure [11, 41]. Past approaches to quantifying the relative effects of habitat invasibility—biotic resistance and abiotic factors—as opposed to propagule pressure have included probabilistic modeling [42, 43] and field manipulations [44].
While previous marine and aquatic metabarcoding surveys have cataloged invasion fronts and the distributions of non-native species, none has systematically investigated invasion levels across a range of coastal habitats. Here, we describe variation in invasion levels in the southern Salish Sea (Washington State, USA) determined using eDNA sequences from discrete water samples collected from eight nearshore sites.
Methods
Water sampling
To characterize eDNA assemblages we used data originally collected for the analysis reported in Gallego et al. (2020). Water samples were collected in the intertidal zone of state and county parks along shores of Hood Canal and San Juan Island, Washington, USA (Fig 1). In Washington state no permits are necessary for small-scale collection of sea water. Abiotic conditions at these sites fall roughly along a gradient of salinity, temperature, and wave energy, with lower salinity, higher temperature, and lower wave energy at the Hood Canal sites compared with the San Juan Island sites [45]. Substrate type differs across these sites as well: the three southernmost sites along the Hood Canal consist of mudflat habitat, while the two northern sites on the Hood Canal are cobble beaches, and the San Juan Island sites are primarily rocky bench [45].
Sampling sites are coded by habitat type: red = more-retentive estuarine mudflat, purple = more-retentive estuarine cobble beach, blue = rocky bench.
Three one-liter replicate bottle samples were collected at each of the eight sites monthly between March 2017 and August 2018. Samples were filtered through 0.45 μm cellulose and these filters were preserved in Longmire’s buffer prior to DNA extraction [46]. Water temperature, salinity, and dissolved oxygen were measured during each sampling event with a multiprobe (Hannah Instruments, USA) and a salinity refractometer.
Sequencing and bioinformatics
The DNA was purified with a phenol-chloroform-isoamyl alcohol extraction following procedures in Renshaw et al. (2015) [46]. Extracted DNA served as a template for polymerase chain reaction (PCR), amplifying 313 base pairs of the COI gene region [26]. PCR was conducted using protocols from Kelly et al. (2018) [47]. Three subsamples of the template DNA served as technical process replicates. These technical replicates were amplified separately and sequenced separately to assess the variability of the PCR process itself. Secondary indexing tags were introduced with a two-step PCR protocol in order to avoid index amplification bias [48]. We sequenced multiplexed sampling events using MiSeq v2-500 and v3-600 sequencing kits via manufacturer specifications. Each sequencing run included three samples of terrestrial species DNA not present in the study region–red kangaroo Macropus rufus or ostrich Struthio camelus–as a positive control to avoid misassignment of sequences to samples, and to control for “index-hopping” [45] “tag-jumping” [49].
Sequencing quality control, denoising into Amplicon Sequence Variants (ASVs) and taxonomical assignment were conducted using custom Bash and R scripts. A Github repository containing these scripts, and access to the corresponding FASTA sequence data is linked in the supplementary material (S1 File).
First, we executed Bash scripts to run the open-source programs Cutadapt and DADA2 for primer-trimming and removal of PCR artifacts [50, 51]. DADA2 was used to estimate the ASV composition of each sample. Second, we conducted further quality control of the results by discarding replicates with low coverage containing PCR anomalies and spurious ASVs. Full details and rationale can be found in Gallego et al. (2020).
We used two bioinformatic tools to assign taxonomy to ASVs. Insect v1.1 was run to assign taxa via informatic sequence classification trees [52]. Additional ASVs were assigned to taxa using a custom COI database with anacapa and Bowtie2 [53, 54]. For the purpose of quantifying the species richness of the eDNA sampled, we retained only those ASVs identified to the species level. We performed a secondary BLASTn search on all ASVs assigned to the species level. Any ASV with a percent sequence identity (pident) < 95% with its best-match reference sequence was considered unclassified. A species was considered present in a sample if an ASV assigned to that species appeared in at least 1 out of 3 replicates for a given site and date. See Gallego et al (2020) for full bioinformatic details.
Identification of native, cryptogenic, and non-native species
The retained ASVs with species-level assignments were referenced against published literature and peer-reviewed online databases to classify each species as either native to the Salish Sea, non-native to the Salish Sea, or cryptogenic. If World Register of Marine Species (WORMS), Algaebase, Biodiversity of the Central Coast, or the National Estuarine and Marine Exotic Species Information System (NEMESIS) listed a species as having a distribution containing the Salish Sea; then the species was classified as native [55–58]. If no peer-reviewed source contained the species distribution, or if there were fewer than three published detections of the species, then the species was classified as cryptogenic.
Species were classified as non-native if multiple peer-reviewed sources listed the species as non-native to the Salish Sea or non-native to other temperate estuaries (n = 17). Or, if there was no such record, the species was classified as non-native if all Salish Sea native species within the same genus were present in the NCBI COI database, and the observed ASV both differed from those in the database and matched the non-native unambiguously (n = 2). Where peer-reviewed literature contained contradictory information regarding native distribution, species were classified as non-native if at least three distinct sources supported non-native status in the Salish Sea.
In order to exclude cases of ASVs from poorly cataloged native taxa being incorrectly assigned to non-native sister taxa; ASVs with percent sequence identity < 98% assigned to non-native taxa were included as non-native only if multiple peer-reviewed sources listed the species as non-native to the Salish Sea (n = 2). See supplementary materials for complete BLAST results and classification flowchart (S1 Fig in S1 File).
Testing for invasion-associated variables
To determine which abiotic and biotic factors best predicted non-native species richness, we modeled non-native species richness as a Poisson variable, with the expected value being a function of salinity, temperature, and native species richness; we used R and the rstanarm package [59, 60] to compare model fits using subsets of these candidate predictor variables and chose the best-fit model using leave-one-out cross-validation, which maximizes the out-of-sample predictive value among candidate models [61]. This pool of predictive variables was selected because the invasion ecology literature suggests non-native species richness varies across ranges of salinity, temperature, and native species richness, and because we were able to measure these factors in the field. Consequently, we tested models with combinations of these three factors.
To quantify the variation of non-native species richness and level of invasion across sampling sites and sampling months, we used the vegan package in R to perform permutational multivariate analyses of variance (PERMANOVA) using Euclidean pairwise distances, with site and month as nested grouping factors [62].
Results
In total, sequencing yielded 50.8 million reads across 86 unique sampling events. Denoising to ASVs and removal of replicates with low coverage retained 45.0 million of these reads made up of 4,848 unique ASVs. Of these ASVs, 1,364 could be annotated to a taxonomic level of family, genus, or species, representing 22.8 million reads. 324 ASVs were assigned to the species level and had a percent identity with the matching reference sequence of > 95%. These 324 ASVs represented 11.0 million retained reads (Table 1).
Composition
After the quality-control process we were able to assign ASVs to 272 species across the 8 sampling sites. Of these 272 species, we identified 251 as native or cryptogenic, and 21 as non-native (Table 2). Seven phyla were represented among these non-native species: Rhodophyta, Ochrophyta, Mollusca, Arthropoda, Chordata, Cnidaria, and Annelida. Non-native species included 9 algal species, 5 bivalve species, 3 copepod species, 1 ascidian species, 1 hydrozoan species, 1 polychaete species, and 1 amphipod species.
Two non-native bivalve species—the Manilla clam (Ruditapes philippinarum) and the Pacific oyster (Crassostrea gigas)—are commonly cultivated in commercial aquaculture operations in Hood Canal. These aquaculture species were detected in 25 of all 86 and 13 of all 86 samples respectively. We note that many species present in the environment are not represented in the data; any given metabarcoding locus amplifies a cross-section of species present, rather than a complete set. Accordingly, species detected here are a subset of those present, with detection bias driven largely by template-primer binding affinity. Hence, nondetection of a species cannot be used to infer that species’ absence from the environment without further information about the amplification efficiency of the species with the primer set in question.
Relative and absolute species richness
Absolute native species richness varied significantly between sites, but not between months (PERMANOVA: site R2 = 0.47, p < 0.001; month R2 = 0.08, p = 0.23). Absolute non-native species richness varied significantly between both sites and months. (PERMANOVA: site R2 = 0.54, p < 0.001; month R2 = 0.14, p = 0.002) The majority of non-native species (18 of 21) were detected only at the more-retentive estuarine sites in Hood Canal (Twanoh, Potlatch, Lilliwaup, Triton Cove, Salisbury). By contrast, only 3 of 21 species were detected at the more marine sites (Friday Harbor, Cattle Point, Lime Kiln) on San Juan Island (Fig 2). Of these, the hydrozoan Bougainvillia mucus was the only species detected exclusively at the San Juan Island sites. Two non-native species were shared between Hood Canal and San Juan Island sites: the clam Nuttallia olivacea and the harmful alga Pseudochattonella farcimen, which was the most frequently detected non-native species.
For each site, if a species was detected at least once during the 18-month sampling period it is displayed as a unique detection.
Relative non-native richness was calculated as non-native species richness / total species richness. Relative non-native richness varied significantly between both sites and months (PERMANOVA: site R2 = 0.56, p < 0.001; month R2 = 0.17, p < 0.001). At the San Juan Island sites, relative non-native richness was negligible across the months sampled. Relative non-native richness was greater at the Hood Canal sites, peaking at 0.25 (8 / 32) at the Lilliwaup site in August 2017. We found substantial variation in relative non-native richness at each of the Hood Canal sites across sampling events (Fig 3). Relative non-native richness was higher during the warmer, more productive spring and summer months (May-September), and lower during the cooler months (October-March) Figs 4 and 5.
Sites are ordered from the southernmost more-retentive estuarine sites on the left, to the northernmost less-retentive rocky bench sites on the right.
Numbers in black are the mean absolute non-native species richness at a site if it was sampled in both 2017 and 2018, or simply the absolute non-native species richness if it was only sampled in 2017.
Each point represents a sampling event. Sampling sites are differentiated by color.
The mean richness of non-native species was low at the San Juan Island Sites (Friday Harbor = 0, Cattle Point = 1, Lime Kiln = 1) compared to the Hood Canal sites (Twanoh = 4, Potlatch = 4, Lilliwaup = 4, Triton Cove = 5, Salisbury = 2). The mean richness of native species was greater at the San Juan Island sites (Friday Harbor = 48, Cattle Point = 58, Lime Kiln = 62) than it was at the Hood Canal sites (Twanoh = 32, Potlatch = 33, Lilliwaup = 30, Triton Cove = 34, Salisbury = 38) (Fig 6).
Error bars correspond to +/- 1 standard deviation. The color gradient reflects the rough gradient from warmer, more-retentive estuarine mudflats (red) to cooler, less-retentive rocky bench (blue).
Invasion level across biotic and abiotic conditions
Our sampling captured a broad range of water temperatures (min = 7.17 C, max = 22.60 C) (Fig 5), salinity (min = 10.00 ppt, max = 30.31 ppt) and native species richness (min = 13, max = 78). Candidate Poisson regression models of non-native species richness included combinations of salinity, temperature, and native species richness as predictive variables. The best-fit model included just temperature and native species richness (Table 3 and Fig 7). Within the selected model, higher temperatures were associated with greater non-native species richness (0.122 mean slope estimate, [0.101–0.145] 10% - 90% credibility interval), while greater native species richness was associated with lower non-native species richness (-0.011, [-0.018 –-0.005]) (Table 4 & Fig 7). Given these parameter estimates, we would expect to observe, on average, one additional non-native species with an increase in water temperature from 15 C to 17.5 C–consistent with anticipated near-term changes in temperate regions [63]–under conditions similar to those sampled here. This would represent, at a minimum, an increase in non-native species richness of 10% relative to the Hood Canal communities–and a greater relative increase at sites with a lower baseline richness of non-native species.
Predictions are plotted as a function of temperature, the dominant effect. Data is subdivided into “Higher Native Richness” (n = 37) and “Lower Native Richness” (n = 39), by the 50th percentile among all sampling events.
Table contains details of the credibility intervals for the posterior distributions of the model’s coefficients.
Discussion
The results of our survey demonstrate varying levels of invasion between habitat types in the Salish Sea. The more marine San Juan Island habitats show negligible levels of invasion. The mesohaline estuarine mudflats of Hood Canal show greater levels of invasion, driven primarily by the number of non-native species of algae and bivalves. The transitional habitat at Salisbury at the mouth of the Hood Canal has an intermediate invasion level. These results are consistent with known dynamics in invasion ecology—that more-retentive estuarine habitats exhibit higher levels of invasion than adjacent open coasts and that increasing temperatures make estuarine habitats more vulnerable to invasion [31, 40, 64–66]. Our results are also consistent with expectations of the biotic resistance hypothesis—that habitats with greater native species richness may resist invasion [36, 66].
Non-native species detected
Our survey contributes new information relevant to the monitoring of non-native and invasive species in the Salish Sea, and detects non-native species previously unreported within the Salish Sea. Gracilaria vermiculophylla is a red alga whose native range encompasses the western North Pacific. It is a known invader of temperate estuaries around the world and has been documented in the eastern Pacific from Mexico to Alaska [67]. The ecological impacts of Gracilaria vermiculophylla are mixed. It has been demonstrated to out-compete native algae, but also to increase habitat complexity [68]. It has previously been identified in the Salish Sea by both visual and molecular methods. Possible vectors of introduction include hull fouling of commercial vessels, transport associated with the aquaculture oyster Crassostrea gigas, and transport in ballast water [67, 69]. Gracilaria vermiculophylla is known to tolerate mesohaline conditions, which is consistent with our findings where it was present in 34 of 86 samples, exclusively in Hood Canal.
Caulacanthus okamurae is a red alga whose native range encompasses the western Pacific. It is a known invader of the eastern Pacific, detected in Baja California, Mexico in 1944 and in Prince William Sound, Alaska in 1989. It has been documented on San Juan Island and the nearby Strait of Georgia [58]. We detected C. okamurae at all 5 Hood Canal sites, and in 23 of 64 samples. We did not detect it on San Juan Island. In California, C. okamurae has increased total turf cover and overall algal and invertebrate diversity in intertidal systems. It was found to displace macroinvertebrates while favoring copepods, ostracods and other algae including Ulva spp., Chondracanthus spp., and Gelidium spp. [58, 70].
Botrylloides violaceus is a colonial ascidian native to the western Pacific. It is a well-documented invader of the eastern Pacific and was detected by visual survey in the southern Salish Sea in 1998 [71]. B. violaceus outcompetes both native and non-native fouling species including native ascidians and mussels [72]. It fouls aquaculture equipment and impacts recruitment of aquaculture species [73].
Species captured by our sampling but previously unreported in the Salish Sea include Pyropia haitanensis, a red alga native to the western Pacific where it is widely cultivated. It makes up the majority of all Chinese cultivation of Pyropia spp. [74]. The taxonomy of the family of seaweeds that includes Pyropia spp., Porphyra spp., and Neoporphyra spp. remains an area of active research. There are no prior visual records or herbarium specimens of P. haitanensis or its historical synonyms in the eastern Pacific or the Salish Sea. A metabarcoding survey of seaweed communities in the northern Gulf of Mexico detected P. haitanensis using an 23S rDNA assay, distinct from COI. Little is known about the ecological impacts of non-native Pyropia spp. We also detected Callithamnion corymbosum. There are no prior visual records or herbarium specimens of C. corymbosum in the Salish Sea. C. corymbosum is a red alga native to the temperate Atlantic. C. corymbosum was documented by visual survey at Willapa Bay on the outer coast of Washington State in 2008 [75]. There is no published literature on the impacts of C. corymbosum on native communities.
Predictors of invasion level
Temperature.
We observed an increase in invasion level with increases in temperature across the range of habitat types and conditions represented in our samples. Previous mesocosm and observational studies generally have found non-native fouling species to be more tolerant of high temperatures than their native counterparts. Field studies have found increased recruitment of non-native ascidians—including B. violaceus—to be positively correlated with higher water temperatures, and native recruitment to be negatively correlated with higher temperatures [12, 65]. In Long Island Sound, USA the invasive alga G. vermiculophylla was found to have higher growth rates at higher temperatures relative to the native Gracilaria tikvahiae [12]. Among the Poisson regression models we tested, the model including temperature and native species richness was found to be the best predictor of non-native species richness. Temperature was the dominant factor (Fig 7). This result provides additional evidence, at an ecosystem scale, for the positive response of non-native species, relative to native species, to higher temperature conditions.
More-retentive estuary vs. better-flushed coast.
Globally, temperate estuaries experience higher levels of invasion than adjacent better-flushed coastal habitats [31–33]. Ours is the first survey to compare levels of invasion in a more-retentive estuary with better-flushed rocky habitat using eDNA methods. Our findings—higher levels of invasion at the more-retentive estuarine Hood Canal sites, intermediate levels at the transitional habitat at Salisbury, and negligible invasion levels at the better-flushed San Juan Island sites—are consistent with the findings of conventional surveys. Multiple mechanisms proposed by previous investigators may account for these results.
Oceanographic circulation likely plays a role. Limited circulation in estuaries may increase the likelihood that newly introduced species will be retained long enough to successfully reproduce [32]. Our results are consistent with this hypothesis: invasion levels are highest in the more-retentive environment of Hood Canal and lowest in the less-retentive environment of San Juan Island, where water circulation is stronger [76].
Aquaculture.
Our findings may be driven by the history of aquaculture in our study region. It is probable that the Hood Canal sites experienced greater non-native propagule pressure in the form of historical transport of shellfish for aquaculture, along with the non-native algae and parasites accompanying these shellfish. Dabob Bay is an inlet in the Hood Canal situated between the Salisbury and Triton Cove sites (Fig 1). Between 1919 and 1935 Dabob Bay was a destination for unregulated shipments of Crassostrea gigas oyster spat on discarded shells originating from Japan [77, 78]. The escape and establishment of non-native species from aquaculture operations is a well-documented phenomena globally [41, 79, 80]. The high number of Manilla clam and Pacific oyster detections in our data, as well as the high proportion of species native to the temperate western Pacific, are consistent with this mechanism. Of all species detected, 15 of 21 have native ranges in Japan and the western Pacific. Among the algae, 7 of 9 have native ranges including Japan. Mytilicola orientalis—native to the western Pacific—is a copepod parasite of Crassostrea gigas.
Anthropogenic disturbance.
Anthropogenic disturbances are known to facilitate biological invasions in marine environments. Such disruptions may make habitat less suitable for native species while also creating space and freeing-up resources for non-native species. Shoreline armoring may facilitate the spread of non-native seaweeds [81]. Musculista senhousia is more successful in disturbed eelgrass beds than undisturbed beds [82]. Additionally, the disturbance caused by the introduction of a single fouling species can facilitate the establishment of additional non-native species and result in an invasional cycle. Such a cycle has been documented in the Gulf of Maine facilitated by the Japanese alga Codium fragile spp. [83, 84]. Hood Canal has a far greater percentage of shoreline armoring than San Juan Island. Moreover, the historical aquaculture trade may have served as the catalyst for an invasional cycle in addition to creating propagule pressure. In sum, the relatively high frequency and intensity of shoreline hardening and aquaculture in the Hood Canal could contribute to the higher level of invasion recorded there.
Biotic resistance.
Consistent with numerous previous investigations [35, 85], we found lower native species richness at soft-bottom mesohaline and polyhaline habitats and greater native species richness at more marine rocky bench habitats. Our best-fit model of non-native species richness includes both temperature and native species richness. Notably, this model outperforms the model that includes temperature alone. This finding aligns with previous investigations in which high native species diversity has been correlated to lower levels of invasion, but is eclipsed by other factors—often traits of the invaders, or abiotic filters [29, 86]. An observational study specific to fouling communities by Stachowicz & Byrnes (2006) suggested that non-native richness in these communities is negatively correlated with native richness only if space is limited and native species that facilitate non-native species settlement are rare [66].
Detection and annotation with eDNA
As with other methods of non-native species monitoring, eDNA surveys are subject to false positives, false negatives, and biases [87]. eDNA shedding and decay rates vary across species, life-stages, and environmental conditions [88]. Primer bias may lead to inconsistent probabilities of detection across taxa [89]. For tracking ongoing biological invasions, open-source sequence data offer an advantage over visual data in that they can be amended as genetic reference databases improve. However, the degree to which eDNA assemblages reflect the underlying community depends on the quality of these databases [90]. We cannot use metabarcoding to identify species that are not in these databases.
Due to primer bias, the absence of a species at a sampling site cannot be inferred from the absence of an ASV in a water sample. For example, the invasive western Pacific alga Sargassum muticum can be visually identified at many of our sampling sites, but S. muticum DNA was never identified in the water samples. However, given the diversity of life histories and morphologies represented among both native and non-native species in our study region, there is no reason to assume a systematic bias towards detection of either native or non-native species. Therefore, inferences about invasion level and non-native species assemblages are as likely to reflect underlying ecological processes as inferences drawn from visual survey data.
Conclusion
Our results show that within the Salish Sea, more-retentive estuarine habitats with mid-range salinities exhibited higher levels of biological invasion than less-retentive, less-estuarine habitats, where invasion levels were comparatively low. Across habitats and seasons, higher water temperatures and lower native species richness were associated with higher non-native species richness. The significant positive relationship observed between temperature and invasion level suggests that habitats may be most vulnerable to invasion during seasonal windows in the spring and summer, and more importantly, suggests that habitats in the Salish Sea may become increasingly vulnerable to biological invasion as anthropogenic climate change expands seasonal windows and causes waters to warm.
The species-level data generated by this survey may help guide ongoing monitoring efforts, especially for species that were previously undetected in the Salish Sea, or are difficult to differentiate from native relatives by visual means, such as the algae Gelidiophycus freshwateri and its native cousins, or Pyropia haitanensis and native Pyropia species. Our data provide a rough “when and where” for future monitoring and control efforts. Using eDNA to estimate invasion level provides additional information to help guide management decisions, including those regarding the utility of prevention versus containment measures. eDNA surveys can consistently sample both non-native and native species at an ecosystem scale, offering a promising new stream of data for illuminating regional invasions, and for investigating foundational questions in invasion ecology.
References
- 1. Diagne C, Leroy B, Vaissière A-C, Gozlan RE, Roiz D, Jarić I, et al. High and rising economic costs of biological invasions worldwide. Nature. 2021;592:571–6. pmid:33790468
- 2. Molnar JL, Gamboa RL, Revenga C, Spalding MD. Assessing the global threat of invasive species to marine biodiversity. Front Ecol Environ. 2008;6:485–92.
- 3. Alemu I JB, Schuhmann P, Agard J. Mixed preferences for lionfish encounters on reefs in Tobago: Results from a choice experiment. Ecol Econ. 2019;164:106368.
- 4. Shiganova TA. Invasion of the Black Sea by the ctenophore Mnemiopsis leidyi and recent changes in pelagic community structure. Fish Oceanogr. 1998;7:305–10.
- 5. Whitlow WL. Changes in survivorship, behavior, and morphology in native soft-shell clams induced by invasive green crab predators. Mar Ecol. 2010;31:418–30.
- 6. Calizza E, Rossi L, Careddu G, Sporta Caputi S, Costantini ML. A novel approach to quantifying trophic interaction strengths and impact of invasive species in food webs. Biol Invasions. 2021;23:2093–107.
- 7. Irigoyen AJ, Trobbiani G, Sgarlatta MP, Raffo MP. Effects of the alien algae Undaria pinnatifida (Phaeophyceae, Laminariales) on the diversity and abundance of benthic macrofauna in Golfo Nuevo (Patagonia, Argentina): potential implications for local food webs. Biol Invasions. 2011;13:1521–32.
- 8. Sullaway GH, Edwards MS. Impacts of the non-native alga Sargassum horneri on benthic community production in a California kelp forest. Mar Ecol Prog Ser. 2020;637:45–57.
- 9. Bailey SA. An overview of thirty years of research on ballast water as a vector for aquatic invasive species to freshwater and marine environments. Aquat Ecosyst Health Manag. 2015;18:261–8.
- 10. Chan FT, Stanislawczyk K, Sneekes AC, Dvoretsky A, Gollasch S, Minchin D, et al. Climate change opens new frontiers for marine species in the Arctic: Current trends and future invasion risks. Glob Change Biol. 2019;25:25–38. pmid:30295388
- 11. Naylor R, Williams S, Strong D. Aquaculture—A Gateway for Exotic Species. Science. 2001;294:1655–6.
- 12. Stachowicz J, Terwin JR, Whitlatch RB, Osman RW. Linking climate change and biological invasions: Ocean warming facilitates nonindigenous species invasions. Proc Natl Acad Sci. 2002;99:15497–500. pmid:12422019
- 13. van Tussenbroek BI, van Katwijk MM, Bouma TJ, van der Heide T, Govers LL, Leuven RSEW. Non-native seagrass Halophila stipulacea forms dense mats under eutrophic conditions in the Caribbean. J Sea Res. 2016;115:1–5.
- 14. Davidson AD, Hewitt CL, Kashian DR. Understanding Acceptable Level of Risk: Incorporating the Economic Cost of Under-Managing Invasive Species. PLOS ONE. 2015;10:e0141958. pmid:26536244
- 15. Vander Zanden MJ, Hansen GJA, Higgins SN, Kornis MS. A pound of prevention, plus a pound of cure: Early detection and eradication of invasive species in the Laurentian Great Lakes. J Gt Lakes Res. 2010;36:199–205.
- 16. Wimbush J, Frischer ME, Zarzynski JW, Nierzwicki-Bauer SA. Eradication of colonizing populations of zebra mussels (Dreissena polymorpha) by early detection and SCUBA removal: Lake George, NY. Aquat Conserv Mar Freshw Ecosyst. 2009;19:703–13.
- 17. Woodell JD, Neiman M, Levri EP. Matching a snail’s pace: successful use of environmental DNA techniques to detect early stages of invasion by the destructive New Zealand mud snail. Biol Invasions. 2021;23:3263–74. pmid:34093071
- 18. Chucholl F, Fiolka F, Segelbacher G, Epp LS. eDNA Detection of Native and Invasive Crayfish Species Allows for Year-Round Monitoring and Large-Scale Screening of Lotic Systems. Front Environ Sci [Internet]. 2021 [cited 2022 May 25];9. Available from: https://www.frontiersin.org/article/10.3389/fenvs.2021.639380
- 19. Mizumoto H, Kishida O, Takai K, Matsuura N, Araki H. Utilizing environmental DNA for wide-range distributions of reproductive area of an invasive terrestrial toad in Ishikari river basin in Japan. Biol Invasions. 2022;24:1199–211.
- 20. Jerde CL. Can we manage fisheries with the inherent uncertainty from eDNA? J Fish Biol. 2021;98:341–53. pmid:31769024
- 21. Keller AG, Grason EW, McDonald PS, Ramón-Laca A, Kelly RP. Tracking an invasion front with environmental DNA. Ecol Appl. 2022;n/a:e2561. pmid:35128750
- 22. Holman LE, de Bruyn M, Creer S, Carvalho G, Robidart J, Rius M. Detection of introduced and resident marine species using environmental DNA metabarcoding of sediment and water. Sci Rep. 2019;9:11559. pmid:31399606
- 23. Lacoursière-Roussel A, Howland K, Normandeau E, Grey EK, Archambault P, Deiner K, et al. eDNA metabarcoding as a new surveillance approach for coastal Arctic biodiversity. Ecol Evol. 2018;8:7763–77. pmid:30250661
- 24. Jacobs-Palmer E, Gallego R, Cribari K, Keller AG, Kelly RP. Environmental DNA Metabarcoding for Simultaneous Monitoring and Ecological Assessment of Many Harmful Algae. Front Ecol Evol [Internet]. 2021 [cited 2022 Apr 1];9. Available from: https://www.frontiersin.org/article/10.3389/fevo.2021.612107
- 25. Andres KJ, Lambert TD, Lodge DM, Andrés J, Jackson JR. Combining sampling gear to optimally inventory species highlights the efficiency of eDNA metabarcoding. Environ DNA. 2023;5:146–57.
- 26. Leray M, Yang JY, Meyer CP, Mills SC, Agudelo N, Ranwez V, et al. A new versatile primer set targeting a short fragment of the mitochondrial COI region for metabarcoding metazoan diversity: application for characterizing coral reef fish gut contents. Front Zool. 2013;10:34. pmid:23767809
- 27. Kelly RP, Closek CJ, O’Donnell JL, Kralj JE, Shelton AO, Samhouri JF. Genetic and Manual Survey Methods Yield Different and Complementary Views of an Ecosystem. Front Mar Sci [Internet]. 2017 [cited 2022 Jun 8];3. Available from: https://www.frontiersin.org/article/10.3389/fmars.2016.00283
- 28. Heger T, Pahl AT, Botta-Dukát Z, Gherardi F, Hoppe C, Hoste I, et al. Conceptual Frameworks and Methods for Advancing Invasion Ecology. AMBIO. 2013;42:527–40. pmid:23532717
- 29.
Olyarnik SV, Bracken MES, Byrnes JE, Hughes AR, Hultgren KM, Stachowicz JJ. Ecological Factors Affecting Community Invasibility. In: Rilov G, Crooks JA, editors. Biol Invasions Mar Ecosyst [Internet]. Berlin, Heidelberg: Springer Berlin Heidelberg; 2009 [cited 2022 Jan 13]. p. 215–38. Available from: http://link.springer.com/10.1007/978-3-540-79236-9_12
- 30. Paavola M, Olenin S, Leppäkoski E. Are invasive species most successful in habitats of low native species richness across European brackish water seas? Estuar Coast Shelf Sci. 2005;64:738–50.
- 31.
Preisler RK, Wasson K, Wolff WJ, Tyrrell MC. Invasions of Estuaries vs the Adjacent Open Coast: A Global Perspective. In: Rilov G, Crooks JA, editors. Biol Invasions Mar Ecosyst Ecol Manag Geogr Perspect [Internet]. Berlin, Heidelberg: Springer; 2009 [cited 2022 Jan 14]. p. 587–617. Available from: https://doi.org/10.1007/978-3-540-79236-9_33
- 32. Wasson K, Fenn K, Pearse JS. Habitat Differences in Marine Invasions of Central California. Biol Invasions. 2005;7:935–48.
- 33. Wolff W. Exotic invaders of the meso-oligohaline zone of estuaries in The Netherlands: Why are there so many? Helgoländer Meeresunters. 1998;52:393–400.
- 34. Nehring S. Four arguments why so many alien species settle into estuaries, with special reference to the German river Elbe. Helgol Mar Res. 2006;60:127.
- 35. Remane A. Die Brackwasserfauna. Verseichnis Beroffentlichungen Goldsteins. 1934;36:34–74.
- 36. Jeschke J, Debille S, Lortie C. Biotic resistance and island susceptibility hypotheses. 2018. p. 60–70.
- 37. Jeschke J, Aparicio LG, Haider S, Heger T, Lortie C, Pyšek P, et al. Support for major hypotheses in invasion biology is uneven and declining. NeoBiota. 2012;14:1–20.
- 38. Lennon JT, Smith VH, Dzialowski AR. Invasibility of Plankton Food Webs along a Trophic State Gradient. Oikos. 2003;103:191–203.
- 39. France KE, Duffy JE. Consumer Diversity Mediates Invasion Dynamics at Multiple Trophic Levels. Oikos. 2006;113:515–29.
- 40. Stachowicz J, Fried H, Osman RW, Whitlatch RB. Biodiversity, Invasion Resistance, and Marine Ecosystem Function: Reconciling Pattern and Process. Ecology. 2002;83:2575–90.
- 41. Mckindsey CW, Landry T, O’beirn FX, Davies IM. BIVALVE AQUACULTURE AND EXOTIC SPECIES: A REVIEW OF ECOLOGICAL CONSIDERATIONS AND MANAGEMENT ISSUES. J Shellfish Res. 2007;26:281–94.
- 42. Leung B, Mandrak NE. The risk of establishment of aquatic invasive species: joining invasibility and propagule pressure. Proc R Soc B Biol Sci. 2007;274:2603–9. pmid:17711834
- 43. Smyth ERB, Drake DAR. The role of propagule pressure and environmental factors on the establishment of a large invasive cyprinid: black carp in the Laurentian Great Lakes basin. Can J Fish Aquat Sci. 2022;79:6–20.
- 44. Holle BV, Simberloff D. Ecological Resistance to Biological Invasion Overwhelmed by Propagule Pressure. Ecology. 2005;86:3212–8.
- 45. Gallego R, Jacobs-Palmer E, Cribari K, Kelly RP. Environmental DNA metabarcoding reveals winners and losers of global change in coastal waters. Proc R Soc B Biol Sci. 2020;287:20202424. pmid:33290686
- 46. Renshaw MA, Olds BP, Jerde CL, McVeigh MM, Lodge DM. The room temperature preservation of filtered environmental DNA samples and assimilation into a phenol–chloroform–isoamyl alcohol DNA extraction. Mol Ecol Resour. 2015;15:168–76.
- 47. Kelly RP, Gallego R, Jacobs-Palmer E. The effect of tides on nearshore environmental DNA. PeerJ. 2018;6:e4521. pmid:29576982
- 48. O’Donnell JL, Kelly RP, Lowell NC, Port JA. Indexed PCR Primers Induce Template-Specific Bias in Large-Scale DNA Sequencing Studies. PLOS ONE. 2016;11:e0148698. pmid:26950069
- 49. Schnell IB, Bohmann K, Gilbert MTP. Tag jumps illuminated—reducing sequence-to-sample misidentifications in metabarcoding studies. Mol Ecol Resour. 2015;15:1289–303. pmid:25740652
- 50. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods. 2016;13:581–3. pmid:27214047
- 51. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal. 2011;17:10.
- 52. Wilkinson SP, Davy SK, Bunce M, Stat M. Taxonomic identification of environmental DNA with informatic sequence classification trees. [Internet]. PeerJ Preprints; 2018 Mar. Available from: https://peerj.com/preprints/26812v1
- 53. Curd EE, Gold Z, Kandlikar GS, Gomer J, Ogden M, O’Connell T, et al. Anacapa Toolkit: An environmental DNA toolkit for processing multilocus metabarcode datasets. Methods Ecol Evol. 2019;10:1469–75.
- 54. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat Methods. 2012;9:357–9. pmid:22388286
- 55.
Starzomski Brian, Proudfoot Beatrice, Brietzke Chanda, Cruickshank Ian, Miskelly James, Reynolds John, et al. Biodiversity of the Central Coast [Internet]. University of Victoria; 2022. Available from: https://www.centralcoastbiodiversity.org/
- 56. Horton T, Kroh A, Ahyong S, Bailly N, Bieler R, Boyko CB, et al. World Register of Marine Species (WoRMS) [Internet]. WoRMS Editorial Board; 2022 [cited 2022 Jun 2]. Available from: https://www.marinespecies.org
- 57.
Guiry M.D., Guiry G.M. AlgaeBase. World-wide electronic publication. National University of Ireland, Galway; 2022.
- 58. Fofonoff PW, Ruiz GM, Stevens B, Simkanin C, Carlton JT. National Exotic Marine and Estuarine Species Information System [Internet]. 2018. Available from: https://invasions.si.edu/nemesis/
- 59.
R Core Team (2021). R: A language and environment for statistical computing [Internet]. Vienna: R Foundation for Statistical Computing; Available from: https://www.R-project.org/
- 60. Goodrich B, Gabry J, Ali I, Brilleman S. rstanarm: Bayesian applied regression modeling via Stan [Internet]. Available from: https://mc-stan.org/rstanarm/
- 61. Vehtari A, Gelman A, Gabry J. Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Stat Comput. 2017;27:1413–32.
- 62. Oksanen J, Simpson G, Blanchet F, Kindt R, Legendre P, Minchin P, et al. vegan: Community Ecology Package [Internet]. 2022. Available from: ttps://CRAN.R-project.org/package = vegan
- 63. Meier HEM, Dieterich C, Eilola K, Gröger M, Höglund A, Radtke H, et al. Future projections of record-breaking sea surface temperature and cyanobacteria bloom events in the Baltic Sea. Ambio. 2019;48:1362–76. pmid:31506843
- 64. Lee H, Thompson B, Lowe S. Estuarine and scalar patterns of invasion in the soft-bottom benthic communities of the San Francisco Estuary. Biol Invasions. 2003;5:85–102.
- 65. Sorte CJB, Williams SL, Zerebecki RA. Ocean warming increases threat of invasive species in a marine fouling community. Ecology. 2010;91:2198–204. pmid:20836440
- 66. Stachowicz J, Byrnes J. Species diversity, invasion success, and ecosystem functioning: disentangling the influence of resource competition, facilitation, and extrinsic factors. Mar Ecol Prog Ser. 2006;311:251–62.
- 67. Krueger-Hadfield S. Everywhere you look, everywhere you go, there’s an estuary invaded by the red seaweed Gracilaria vermiculophylla (Ohmi) Papenfuss, 1967. BioInvasions Rec. 2018;7:343–55.
- 68. Weinberger F, Buchholz B, Karez R, Wahl M. The invasive red alga Gracilaria vermiculophylla in the Baltic Sea: adaptation to brackish water may compensate for light limitation. Aquat Biol. 2008;3:251–64.
- 69. Saunders GW. Routine DNA barcoding of Canadian Gracilariales (Rhodophyta) reveals the invasive species Gracilaria vermiculophylla in British Columbia. Mol Ecol Resour. 2009;9 Suppl s1:140–50. pmid:21564973
- 70. Smith JR, Vogt SC, Creedon F, Lucas BJ, Eernisse DJ. The non-native turf-forming alga Caulacanthus ustulatus displaces space-occupants but increases diversity. Biol Invasions. 2014;16:2195–208.
- 71. Cohen A, Mills C, Berry H, Wonham M, Bingham B, Bookheim B, et al. A Rapid Assessment Survey of Non-indigenous Species in the Shallow Waters of Puget Sound [Internet]. 1998. Available from: https://www.dnr.wa.gov/publications/aqr_nrsh_cohen1998.pdf
- 72. Rajbanshi R, Pederson J. Competition among invading ascidians and a native mussel. J Exp Mar Biol Ecol. 2007;342:163–5.
- 73. Carman M. Tunicate faunas of two North Atlantic-New England islands: Martha’s Vineyard, Massachusetts and Block Island, Rhode Island. Aquat Invasions. 2009;4:65–70.
- 74. Wang W, Wu L, Xu K, Xu Y, Ji D, Chen C, et al. The cultivation of Pyropia haitanensis has important impacts on the seawater microbial community. J Appl Phycol. 2020;32:2561–73.
- 75. Hansen G. Willapa Bay: A True Hotspot for Seaweed Introductions on the U.S. West Coast. Ecosystem Services of Algae in the Estuaries of the West Coast; 2008.
- 76.
MacCready P, Siedlecki S, McCabe R, Banas N. LiveOcean [Internet]. University of Washington; 2021. Available from: https://faculty.washington.edu/pmacc/LO/LiveOcean.html
- 77. Nims C. Oyster Farming in Washington, Part 1 [Internet]. 2020 [cited 2022 Jun 3]. Available from: https://www.historylink.org/file/21070
- 78. White J, Ruesink JL, Trimble AC. The Nearly Forgotten Oyster: Ostrea lurida Carpenter 1864 (Olympia Oyster) History and Management in Washington State. J Shellfish Res. 2009;28:43–9.
- 79. Haupt TM, Griffiths CL, Robinson TB, Tonin AFG. Oysters as vectors of marine aliens, with notes on four introduced species associated with oyster farming in South Africa. Afr Zool. 2010;45:52–62.
- 80. Ju R-T, Li X, Jiang J-J, Wu J, Liu J, Strong DR, et al. Emerging risks of non-native species escapes from aquaculture: Call for policy improvements in China and other developing countries. J Appl Ecol. 2020;57:85–90.
- 81. Bulleri F, Airoldi L. Artificial marine structures facilitate the spread of a non-indigenous green alga, Codium fragile ssp. tomentosoides, in the north Adriatic Sea. J Appl Ecol. 2005;42:1063–72.
- 82. Allen B, Williams S. Native eelgrass Zostera marina controls growth and reproduction of an invasive mussel through food limitation. Mar Ecol Prog Ser. 2003;254:57–67.
- 83. Grosholz ED. Recent biological invasion may hasten invasional meltdown by accelerating historical introductions. Proc Natl Acad Sci. 2005;102:1088–91. pmid:15657121
- 84. Levin PS, Coyer JA, Petrik R, Good TP. Community-Wide Effects of Nonindigenous Species on Temperate Rocky Reefs. Ecology. 2002;83:3182–93.
- 85. Dethier MN, Schoch GC. The consequences of scale: assessing the distribution of benthic populations in a complex estuarine fjord. Estuar Coast Shelf Sci. 2005;62:253–70.
- 86. Fleming J, Wersal R, Madsen J, Dibble E. Weak non-linear influences of biotic and abiotic factors on invasive macrophyte occurrence. Aquat Invasions. 2021;16:349–64.
- 87. Cristescu M, Hebert P. Uses and Misuses of Environmental DNA in Biodiversity Science and Conservation. Annu Rev Ecol Evol Syst. 2018;49.
- 88. Andruszkiewicz Allan E, Zhang WG, Lavery A C, F. Govindarajan A. Environmental DNA shedding and decay rates from diverse animal forms and thermal regimes. Environ DNA. 2021;3:492–514.
- 89. Elbrecht V, Leese F. Validation and Development of COI Metabarcoding Primers for Freshwater Macroinvertebrate Bioassessment. Front Environ Sci [Internet]. 2017 [cited 2022 Jun 3];5. Available from: https://www.frontiersin.org/article/10.3389/fenvs.2017.00011
- 90. Stoeckle MY, Das Mishu M, Charlop-Powers Z. Improved Environmental DNA Reference Library Detects Overlooked Marine Fishes in New Jersey, United States. Front Mar Sci [Internet]. 2020 [cited 2022 Jun 10];7. Available from: https://www.frontiersin.org/article/10.3389/fmars.2020.00226