A Framework for Aquatic Invasive Species Surveillance Site Selection and Prioritization in the US waters of the Laurentian Great Lakes

Abstract

Risk-based prioritization for early detection monitoring is of utmost importance to prevent and mitigate invasive species impacts. The Great Lakes Water Quality Agreement, a binational commitment between the United States and Canada to restore and protect the waters of the Laurentian Great Lakes, identifies aquatic invasive species (AIS) as one of ten priority issues (annexes) that must be addressed to ensure the chemical, physical, and biological integrity of the Great Lakes. The Agreement calls out the need for a comprehensive strategy for detecting and tracking new and potentially invasive species. Yet, with a surface water area of 95, 000 square miles (246, 049 square km) and shoreline length of 10, 210 miles (16, 431 km), the Great Lakes represent a daunting challenge for prioritizing where AIS surveillance activities should occur. Our goal was to develop a spatially-explicit and quantitative approach for identifying the highest risk sites for AIS introduction into the US waters of the Great Lakes based on the cumulative risk of new introductions (including range expansions) from a range of pathways and associated taxa. We estimate “invasion risk” scores for nearly 6,000 sites (9 km x 9 km) across the Great Lakes basin using proxy measures for propagule pressure weighted by the proportion of taxa associated with each proxy variable. Proxy variables include human population, number of ship visits, marina size, number of ponds, and number of natural or artificial aquatic connections. In total, we identify more than 1,800 sites with invasion risk scores >0. A small subset of these 1,800+ sites accounts for a majority of predicted propagule pressure and are therefore logical targets for future surveillance and AIS prevention efforts. Many of the highest risk sites are located in western Lake Erie, southern Lake Michigan, and the St. Clair-Detroit River System.

Introduction

The management of biological invasions is one of twenty targets included in the United Nation’s Convention on Biological Diversity Strategic Plan, the foremost guidance for national strategies to conserve and sustainably use global biodiversity. Aichi Target 9, concerning invasive species, identifies prioritization of species and pathway risks as a key element of invasive species management (UNEP 2011). Site prioritization, though not explicitly mentioned in Aichi Target 9, has been recognized as a critical third focus area for comprehensive invasive species prioritization (McGeogh et al. 2016). Whereas, species prioritization efforts are numerous and have been applied at various scales from global to regional (Roy et al. 2014, Nentwig et al. 2016), and pathway prioritization examples are increasing as pathway research clarifies terminology and links pathways with “real-world” data (Hulme et al. 2008, Essl et al. 2015), prioritization efforts that consider the interaction of species, pathways, and sites are less common (i.e. “integrated prioritization,” sensu McGeogh et al. 2016). Here we consider the combined risk of multiple species, from multiple pathways, across multiple sites to inform aquatic invasive species (AIS) surveillance and early detection efforts in the Laurentian Great Lakes, one of the most heavily invaded aquatic systems in the world (Mills et al. 1993, Ricciardi 2006). We describe a spatially explicit and quantitative approach for identifying the highest risk sites for AIS introduction based on the cumulative risk of new introductions (including range expansions) from a range of pathways and associated non-native species across sites spanning the US waters of the Great Lakes basin.

Policy changes appear to have slowed the rate of invasion in the Great Lakes (Bailey et al. 2011). Increased regulation and monitoring of ballast water transport by transoceanic vessels likely accounts for some of the observed decline in non-native species introductions, as the shipping pathway has accounted for the majority (~ 70%) of new species introductions to the Great Lakes in the last sixty years (Holeck et al. 2004). However, four new non-native plankton species have been detected in the basin since 2015. Vectors of introduction are not know for these species but some of them were likely introduced via contaminated ballast water from foreign ports of call (e.g., Thermocyclops crassus, Connolly et al. 2017; Brachionus leydigii, Connolly et al. 2018; Diaphanosoma fluviatile, Whitmore et al. 2019), whereas introduction of Mesocyclops pehpeiensis into Lake Erie was probably related to the ornamental aquatic plant trade or aquaculture (Connolly et al. 2019). Thus, management of the ballast pathway, while robust, does not provide complete protection against biological invasion, and imperfect management of non-shipping vectors leaves the Great Lakes vulnerable to new introductions. Several potentially invasive species are predicted to arrive in the Great Lakes over the next few decades (Pagnucco et al. 2015). A recent analysis of historical Great Lakes AIS detection data found that through time, detections are increasingly associated with population centers and less associated with maritime traffic, highlighting the growing importance of introduction pathways other than shipping (O’Malia et al. 2018). For invaders that are already established in North America, authorized and unauthorized release of AIS and spread via canals and natural aquatic connections are two key vectors for the Great Lakes (Rothlisberger & Lodge 2013).

Recognizing the continued and imminent threat of AIS to the Great Lakes, the US Environmental Protection Agency (EPA) called for the development of a comprehensive program for AIS early detection and the establishment of a coordinated, multi-species early detection network (USEPA 2014). In 2014, the Great Lakes states of Illinois, Indiana, Michigan, Minnesota, New York, Ohio, Pennsylvania and Wisconsin formed an Early Detection Rapid Response (EDRR) Team to collaborate on the development of tools and guiding documents to support state AIS management actions. Under the leadership of the Michigan Department of Environmental Quality, the EDRR Team secured a Great Lakes Restoration Initiative (GLRI) grant from US Fish and Wildlife Service (USFWS) and invited partners representing state and federal agencies, academic institutions, and non-governmental organizations to develop a watch list of species of concern and a surveillance site selection and prioritization method as a first step towards developing a comprehensive program for AIS early detection in the Great Lakes. This paper is the product of those efforts.

Our analysis relies heavily on the predictive power of propagule pressure and history of invasion as indicators of invasion success (sensu Williamson and Fitter 1996, Ricciardi et al. 2011, Kumschick et al. 2015, O’Malia et al. 2018). “Invasion risk” scores are modeled for nearly 6,000 sites (9 km x 9 km) across the Great Lakes basin using proxy measures for propagule pressure weighted by the proportion of established and potential future invasive taxa present in the pathway(s) associated with each proxy variable. Proxy variables applied at each site include human population, number of ship visits, marina size, number of ponds, and number of natural or artificial aquatic connections. Sites are ranked on the basis of highest cumulative risk of invasion from all potential pathways of AIS introduction and secondary spread within the Great Lakes Basin.

Methods

A systematic spatial (geo-referenced) prioritization method was developed for attributing a weighted index of invasion pressure to each of the 5900+ 9 km x 9km grid squares across the Great Lakes Basin (Figure 1).

Figure 1.

Conceptual diagram showing the systematic spatial prioritization method for attributing a weighted “index of invasion pressure” for each 9 × 9 km grid cell.

Spatial framework

We used the Great Lakes Aquatic Habitat Framework (GLAHF) 9,000-meter grid (Wang et al. 2015) as our underlying spatial framework. The original raster grid was converted to a polygon layer and cells (9 km per side) were attributed with country, state/province, and lake basin based on the location of cell centroids. This grid was subsequently attributed with surrogate variables thought to influence the risk of AIS invasion from various potential pathways. We selected surrogates that account for all the major vectors and activities that have been associated with the introduction of non-native species into the Great Lakes (Table 1; Mills 1993, Ricciardi 2006). Grid cells were attributed according to features occurring locally in the grid cell. Coastal cells that included a river mouth were also attributed with the features in upstream contributing areas (watersheds). The grid was restricted to waters of the Great Lakes, connecting channels, and inland streams up to the first major barrier. The first major barrier was identified using a draft version of the Fish Works hydrography and barriers data layers (Moody et al. 2017).

Table 1.

Data sets, data type, and data source for all pathway surrogates used for determining the “index of invasion pressure”.

Data Set	Data Type	Source	Associated AIS vector categories
9,000-meter grid	Raster converted to polygon	Great Lakes Aquatic Habitat Framework (GLAHF) glahf_9000m_grid GLAHF_spatial_framework_v1d1.gdb http://glahf.org/framework/	n/a
Great Lakes Basin Population (2010/2011)	Polygon	Great Lakes Aquatic Habitat Framework (GLAHF) US/Canadian integrated census data. Credits: U.S. Census Bureau 2010 Census Demographic Profile 1; Statistica Canada 2011 Census Profile. Apportioned to GLAHF US/Canadian land cover using Dasymetric Mapping Toolbox from EPA EnviroAtlas.	Organisms in trade pathways including, aquarium release and accidental release (e.g. ornamental escape)
Great Lakes Basin GLAHF Land Use (2010/2011)	Raster	Great Lakes Aquatic Habitat Framework (GLAHF) glahf_glb_land_cover_11_12_00_nlcd_solris_plo glahf_land_cover_11_12_00_nlcd_solris_plo.gdb http://glahf.org/data/	n/a
Shipping vessel trips to port (2004–2013)	Point	Data provided by: Elon O’Malia and Dr. Joel Hoffman (EPA). Data gathered by E. O’Malia, University of Minnesota Duluth/EPA — (2014) from National Ballast Water Clearing House (NBIC) (http://invasion.si.edu./nbic).	Ballast and hull fouling
In-lake discharge events (2004–2009)	Point	Data provided by: John Bossenbroek (University of Toledo).	Ballast
Marina size (# of boat slips)	Point	Data provided by Caitlin Dickinson at the Great Lakes Environmental and Mapping Project. Allan JD. et al. 2013. Joint analysis of stressors and ecosystem services to enhance restoration effectiveness. Proceedings of the National Academy of Sciences 110(1):372–377. Digital data.	Recreational boating and associated activities including, attachment to hulls, entanglement of fishing gear or anchor chains, and transport of standing water (e.g. live wells, bilge, bait buckets)
Boat launch size (# of parking spaces)	Point	Data provided by Caitlin Dickinson at the Great Lakes Environmental and Mapping Project. Allan JD. et al. 2013. Joint analysis of stressors and ecosystem services to enhance restoration effectiveness. Proceedings of the National Academy of Sciences 110(1):372–377. Digital data.	Recreational boating and associated activities including, attachment to hulls and trailers, entanglement of fishing gear or anchor chains, and transport of standing water (e.g. live wells, bilge, bait buckets)
Ponds	Polygon converted to point	National Wetlands Inventory (NWI) “excavated” freshwater ponds. U.S. Fish and Wildlife Service. 2015. Classification of Wetlands and Deepwater Habitats of the United States. U.S. Department of the Interior, Fish and Wildlife Service, Washington, DC. FWS/OBS-79/31. Publication date: 2015–05-01. Digital data.	Deliberate release associated with cultivation or stocking
Major canals	Point	Chicago Area Waterway System and Erie Canal. Internally developed.	Natural dispersal
Interbasin Headwater Connections	Point	Great Lakes and Mississippi River Interbasin Study. Internally developed using information provided in GLMRIS reports.	Natural dispersal
GLAHF watersheds	Polygon	Great Lakes Hydrography Dataset Version 1 (GLHDv1.0). 2014. Watersheds and pour point features. Digital data. Forsyth et al. 2015. A consistent binational watershed delineation and synthetic network dataset for the Great Lakes Basin. In review.	n/a

Surrogate data and processing steps

The data representing surrogates for potential pathways of AIS invasion were acquired from multiple sources (Table 1). The data were attributed to grid squares as follows: Most point datasets originated as tabular data and were converted to geospatial layer points using latitude and longitude coordinates contained in the data using ArcGIS version 10.3 (ESRI 2015). Census population and land cover for the Great Lakes Basin were acquired from the Great Lakes Aquatic Habitat Framework dataset (GLAHF; https://www.glahf.org/). We used the Dasymetric Mapping Toolbox tools (from EPA EnviroAtlas; https://www.epa.gov/enviroatlas/dasymetric-toolbox) to apportion the census unit population data to appropriate land covers to get a more refined geospatial representation of population across the basin. This tool apportions the census block unit population to those areas within that block that have ‘developed’ land uses (30m cell size resolution). This way, waterbodies and undeveloped land get little to no population assigned and the developed areas get most of the population. For our work, which quantifies population in every watershed, this provides a more accurate assessment of the population in each watershed. These refined population data were then attributed to GLAHF watersheds and our grid cells. The Chicago metropolitan area is situated mostly outside of the basin, but because of the artificial connections created by the Chicago Area Waterway System, much of the population is effectively connected to the basin. We therefore included the population within two 8-digit hydrologic units (07120003 & 07120004), which are hydrologically connected to Lake Michigan, to more accurately account for the population risk in the Chicago area.

Data located within the boundaries of a Great Lake or along the coastline were assigned to the grid cell in which they occur and attributed with a count of the feature in that grid cell (e.g., population size) or a total amount of an attribute of the feature (e.g., total number of marina boat slips). Data located inland were first attributed to watershed polygons developed as part of the GLAHF, and then transferred to the appropriate grid cell using the outlet pour point of those watersheds that intersected the grid. To create risk scores that did not over emphasize a single pathway, we combined the rescaled data for marina size and boat launch size (all surrogates for the dispersed and bait release pathway) into a single variable. The shipping surrogate data layer is a combination of the number of ship visits to a given port and open water discharge events, with the latter being treated as equivalent to a ship visit. The canal surrogate data layer consisted of point locations representing smaller headwater or large canal inter-basin connections. The connection points were weighted based on perceived risk derived from a pathway viability assessment (USACE 2013); low-risk connections were assigned a weight of 1, medium-risk connections were assigned a weight of 10, and high-risk sites (major canals) were assigned a weight of 100. As with the other risk variables, upstream connections were summed to the drainage outlet.

After all data were attached to the grid, each variable was divided by the maximum for that variable and multiplied by 100 to rescale from zero (none) to 100 (maximum). This approach assumes that the maximum values for each pathway are equal in risk. Risk weighting factors were then applied to the rescaled surrogate data based on past or predicted future pathway-related patterns of invasion (see “AIS risk weighting factors” below). Surrogate values were summed to generate weighted risk scores for each site (i.e. grid). The two weighted risk scores (i.e. based on future and historic patterns of introduction) were averaged for each site and the average score was used as the final “index of invasion pressure.” All sites were then ranked from highest to lowest risk according to the index score for each grid square.

AIS risk weighting factors

Retrospective and prospective analyses of non-native species introductions to the Great Lakes demonstrate that some pathways pose a higher risk than others as vectors of species introduction (Pagnucco et al. 2015, Ricciardi et al. 2006). Thus, we derived two sets of weighting factors for surrogate data layers, one based on predicted future patterns of non-native species introduction and a second based on historic pathways of introduction.

1. ”Future invaders” risk weighting factor: In this approach our aim was to weight re-scaled surrogate values for each site based on predicted future pathways of invasive species introduction to the Great Lakes. We compiled a list of potential future invaders from various sources including, regulated species lists for Great Lakes’ jurisdictions, the US Fish and Wildlife Service’s Ecological Risk Screening Summaries, Fisheries and Oceans Canada’s screening level assessments for fish, mollusks, and plants, the Great Lakes Aquatic Nonindigenous Species Information System (GLANSIS) Watchlist and GLANSIS Nonindigenous and Range Expander species, etc. (see Supplement 1). A priori exclusions included viruses, bacteria, marine and tropical species, species established in all five lake basins, and species with no known history of invasion or impacts (among other criteria; see Supplement 1).

   We then used a semi-quantitative questionnaire-based risk assessment methodology developed by Davidson et al. (2017) to evaluate the potential for negative ecological or socio-cultural impacts and to assign a pathway (or pathways) of introduction for every species on the candidate future invaders list. The risk assessment method from Davidson et al. scores risk for each of three “assessment components” (introduction, establishment, and impact) based on the results of a literature review and expert opinion. In total, 236 species were assessed, of which 147 were identified as high- or medium-risk species (i.e. “future invaders”; based on impact scores). The future invaders risk weighting factor for each surrogate (i.e. pathway) was then derived based on the relative proportion of all future invaders assigned to each pathway (Table 2). Because the probability of introduction and adverse impacts varies across the major taxonomic groups of concern, separate weightings were derived for all taxonomic groups combined and for each major taxonomic group independently (fish, plants, and invertebrates).

2. “Historic invaders” risk weighting factor: In this approach, a pathway of introduction was assigned to every Nonindigenous and Range Expander species currently established in the Great Lakes (GLANSIS 2017). Pathways were assigned per the pathway categories defined by GLANSIS (e.g. aquaculture, aquarium release, bait release, canals, etc.; see Table 3). Pathway weightings were then derived from the relative proportion of all established Nonindigenous and Range Expanders species in each pathway.

Table 2.

“Future invaders” risk weighting factors applied to each re-scaled spatial surrogate value and used to derive invasion risk index scores based on future predicted patterns of invasion (per Davidson et al. 2017). The risk weighting factors were calculated as the proportion of species in each taxonomic group that is predicted to arrive by the specified pathway(s). When more than one pathway is indicated, the risk weighting factor is the sum of all pathways combined.

Spatial surrogates	Davidson et al. 2017 pathways	All taxa	Fish	Inverts	Plants
U.S. Population (2013)	Unauthorized intentional release (INT) Escape from recreational culture (ESC)	0.67	0.57	0.14	0.98
Shipping vessel trips to port (2004–2013)	Shipping (SH)	0.31	0.33	0.74	0.09
Marina size (# of boat slips)	Hitchhiking/Fouling (HF)	0.59	0.33	0.40	0.84
Boat launch size (# of parking spaces)
Ponds	Unauthorized Intentional release (INT) Escape from recreational culture (ESC) Escape from commercial culture (COMM)	0.74	0.67	0.14	1.08
Canals	Dispersal (D)	0.44	0.43	0.34	0.49

Table 3.

“Historic invaders” risk weighting factors applied to each re-scaled spatial surrogate value and used to derive invasion risk index scores based on historic invasion patterns (per the GLANSIS “Nonindigenous + Range Expanders” list). The risk weighting factors were calculated as the proportion of species in each taxonomic group for which introduction has been assigned to a given pathway (s). When more than one pathway is indicated, the weighting is the sum of all pathways combined.

Spatial surrogates	GLANSIS “Nonindigenous + Range expanders” pathways	All taxa	Fish	Inverts	Plants
U.S. Population (2013)	Aquarium release Pet release Stocked Planted	0.37	0.68	0.13	0.47
Shipping vessel trips to port (2004–2013)	Shipping	0.43	0.16	0.67	0.24
Marina size (# of boat slips)	Dispersed Bait release	0.42	0.51	0.27	0.66
Boat launch size (# of parking spaces)
Ponds	Aquaculture Planted Stocked	0.29	0.61	0.04	0.39
Canals	Canals	0.17	0.40	0.13	0.12

Statistical analyses

The non-parametric Mann-Kendall test was used to assess the strength of the relationship between binary combinations of pathway surrogates. For each pairwise comparison data were sorted by one surrogate and the Mann-Kendall test was used to look for a trend in the other surrogate. The analysis was conducted using the Kendall package (McLeod 2011) in RStudio (RStudio Team 2015 version 1.2.1335). A linear regression model was used to compare weighted models (i.e. future versus historic invaders; Microsoft Excel for Office 365, Version 16.0.11328.20362).

For fish, the proportion of total propagule pressure that could be accounted for by sampling increasingly greater numbers of sites was calculated for each pathway surrogate and on average across all pathway surrogates. As an example, for the population surrogate: First, all sites were ranked based on the index of invasion pressure score for fish. Then the sum total population for the “top 5” highest ranked sites was calculated. The “top 5” total population was divided by the total population for all 5,900+ sites across the basin. The resulting quotient is the “proportion of propagule pressure” (for pathways associated with the population surrogate) that could be accounted for if the “top 5” highest risk sites were surveyed. The “proportion of propagule pressure” quotient was derived for 5, 10, 20, 30, 40 sites and so on up to 640 sites. The number of sites and corresponding proportion of propagule pressure were plotted against each other for each surrogate and a logistic model was fitted to the data.

Results

Surrogates

Surrogate values vary across the basin (see Supplement 2; Figures S1-S5), but most grid squares have low values for each surrogate, relative to basin-wide maximum values (Table 4). All surrogate combinations are significantly correlated but Mann-Kendall tau values are low, indicating that while there is some monotonic relationship between each pairwise set of surrogates, any one surrogate is not necessarily a good predictor of another (Table 4). Furthermore, pathway ranks vary considerably within any given site (Table 5, for the “top 25 highest risk” sites).

Table 4.

Summary statistics based on raw data values for select surrogate variables in each 9km x 9km grid square (based on raw data and including contributions from catchment). Column headers are: Minimum, 25^th Percentile, Median, 75^th Percentile, Maximum.

	Min	P25	Median	P75	Max
Population	0	265	1414	5395	6670987
Ponds	0	0	9	52	18517
Marina + Boat Launch	0	0	0	0	5710
Connections	0	0	0	0	100
Shipping	0	0	0	0	8298

Table 5.

Mann-Kendall Tau (correlation coefficient) and two-sided p-value for each pairwise combination of surrogates.

Comparison	Tau	p-value
Population v Ponds	0.615	0.00
Population v Marina + Boat launch	0.257	0.00
Population v Connections	0.127	0.00
Population v Shipping	−0.042	0.03
Ponds v Marina + Boat launch	0.199	0.00
Ponds v Connections	0.097	0.00
Ponds v Shipping	−0.082	0.00
Marina + Boat launch v Connections	0.256	0.00
Marina + Boat launch v Shipping	0.235	0.00
Connections v Shipping	0.119	0.00

Weightings

Based on historic introduction patterns and future predicted patterns of introduction, every vector has been or is likely to be associated with the introduction of at least some non-native species (Table 2 & 3, Figure 2; all pathway weightings > 0). However, the primary pathways of introduction vary across taxa and time. Invertebrates, both historically (GLANSIS NAS + RE) and based on predicted invasion patterns (future invaders list), are strongly associated with the shipping pathway, by a nearly 2:1 margin relative to all other pathways. Approximately 60% of fish have been or are likely to be introduced through the pathways associated with ponds (e.g. stocking) and human population, especially the trade in live organisms (including ornamental escape). For plants, more than 90% of potential future invaders are associated with intentional introduction, escape, and commercial cultivation pathways, and more than 80% of future invader plant species could be introduced via hitchhiking or hull fouling associated with recreational watercraft. This contrasts with historic patterns of invasion that were overall more skewed towards recreational boating with relatively smaller contributions from the vectors associated with ponds and connections (Figure 2).

Figure 2.

Proportion of “Future invaders” (dark bars) and “Historic invaders” (light bars) occurring in each pathway, within each taxonomic group and for all taxa combined. The “Future invaders” are indicative of predicted future pathways of AIS introductions to the Great Lakes, whereas “Historic invaders” indicate past pathways of introduction (i.e. based on GLANSIS Nonindigenous and Range Expander species).

Rankings

The spatial framework was comprised of 5,953 grid squares in the U.S. Great Lakes basin, of which there were 3,487 coastal grid squares (shoreline and tributaries up to the first barrier). Of the coastal grid squares, approximately 1,800 have attributes resulting in index scores greater than zero.

Weighted model scores based on historic and potential future invaders were strongly correlated but the “index of invasion pressure” for each taxonomic group or for all taxa ultimately uses the average of the two weighted risk scores because of the apparent variation in historic and predicted future pathways of invasion (Figure 3 and see Weightings above). Some sites contribute disproportionately to total risk, based on proportion of total propagule pressure within any given pathway or when averaged across all pathways (Table 7, for fish). Averaging across all pathways for fish introduction, the top 30 (out of more than 1,800) highest risk sites account for >50 % of estimated fish propagule pressure. Sites with large values for any given surrogate substantially increase the proportion of propagule pressure that is accounted for within some pathways. For example, including Duluth, MN (25^th highest risk site for fish introduction, but the largest shipping port in the Great Lakes) in a “portfolio of surveillance sites,” increases the proportion of propagule pressure that is accounted for from the shipping pathway by nearly 20%.

Figure 3.

Regression plots of weighted risk scores for Great Lakes coastal sites based on “future invaders” risk weighting factor and “historic invaders” risk weighting factor for each group, a) fish, b) invertebrates, c) plants, and d) all taxa (r² value shown). Risk scores are strongly correlated but regression slopes indicate that only for fish is the original scale preserved between weightings (i.e. the slope = 1).

Table 7.

Proportion of propagule pressure accounted for within each pathway (and on average across all pathways) for a given number of survey sites, based on invasion pressure scores (ranked 1 to n, for fish only).

No. of sites	Population	Ponds	RecBoat	Connections	Shipping	Average
5	0.23	0.13	0.04	0.49	0.09	0.20
10	0.31	0.24	0.10	0.74	0.17	0.31
30	0.52	0.44	0.39	0.98	0.41	0.55
80	0.68	0.59	0.71	0.98	0.78	0.75
90	0.69	0.61	0.75	0.99	0.82	0.77
320	0.87	0.80	0.97	1.00	1.00	0.93
640	0.96	0.93	0.99	1.00	1.00	0.98

The highest risk sites vary by taxa, but in general, the same subset of twenty to twenty-five sites consistently rank among the highest risk across all taxa (Table 8). When risk scores are averaged across all taxa, the ten highest risk sites rank no lower than 17 (out of 1,800+) within any one taxonomic group, and for fish these sites account for 31% of total propagule pressure (Table 7). The top 30 sites represent more than 50% of propagule pressure for fish. The number of sites that account for 90% or more of propagule pressure exceeds 300.

Table 8.

Index of Invasion Pressure rank order (1= highest risk) inclusive of top 25 sites in each taxonomic grouping.

Lake Basin	Location Name	State	Fish	Inverts	Plants	Average rank
Michigan	Chicago/Chicago River Mouth	IL	1	3	1	1.7
Erie	Toledo/Maumee River Mouth	OH	2	2	2	2
Ontario	Oswego/Oswego River Mouth	NY	3	9	8	6.7
Michigan	Portage/Portage-Burns Waterway	IN	4	5	12	7
Erie	Cleveland/Cuyahoga River Mouth	OH	9	4	9	7.3
Huron	Saginaw Bay/Saginaw River Mouth	MI	6	15	4	8.3
Erie	West Harbor/Marblehead/Lake Erie	OH	14	10	3	9
Erie	Sandusky/Sandusky Bay	OH	16	6	6	9.3
Erie	Buffalo/Niagara River	NY	5	11	14	10
Michigan	Calumet River Mouth/Lake Michigan	IN	8	12	17	12.3
Erie	Grosse Pointe Shores/Lake St. Clair	MI	18	16	7	13.7
Michigan	Benton Harbor/Saint Joseph River	MI	7	31	5	14.3
Michigan	East Chicago/Indiana Harbor Canal	IN	10	17	20	15.7
Michigan	Evanston/North Shore Channel Mouth	IL	11	18	21	16.7
Erie	Lake St. Clair/Clinton River Mouth	MI	21	21	10	17.3
Michigan	Milwaukee/Kinnickinnic River Mouth	WI	20	14	19	17.7
Ontario	Rochester/Genesee River Mouth	NY	12	20	24	18.7
Michigan	Green Bay/Fox River Mouth	WI	15	25	16	18.7
Erie	Lakeside/Lake St. Clair	MI	17	35	11	21
Superior	Duluth/St. Louis River Mouth	MN	23	1	39	21
Erie	Detroit River/Rouge River Mouth	MI	19	32	15	22
Erie	Fairport Harbor/Grand River Mouth	OH	22	33	18	24.3
Erie	Erie/Presque Isle Bay	PA	26	28	22	25.3
Erie	Lorain/Black River Mouth	OH	24	30	25	26.3
Michigan	Grand Haven/Grand River Mouth	MI	13	54	13	26.7
Erie	Toussaint River Mouth	OH	25	37	23	28.3
Erie	Ashtabula/Ashtabula River Mouth	OH	30	23	40	31
Superior	Marquette/Dead River Mouth	MI	37	7	57	33.7
Michigan	Chicago-Calumet Port	IL	40	13	55	36
Erie	Detroit/Detroit River	MI	38	8	64	36.7
Huron	Alpena/Thunder Bay River Mouth	MI	47	19	68	44.7
Huron	Rogers City/Calcite	MI	60	22	91	57.7
Superior	Two Harbors	MN	66	24	106	65.3

For fish, high risk sites are especially concentrated at the St. Clair-Detroit River System (SCDRS) (from Port Huron, MI to Sandusky, OH), in western basin Lake Erie, and in southern Lake Michigan (Figure 4). These sites are all associated with varying combinations of moderate population density within their contributing catchments, large marinas and boat ramps, or in some cases moderate to high shipping activity. The highest risk sites for invertebrates are major ports, including Duluth-Superior, Toledo, Chicago, and Cleveland (Figure 5). High risk plant sites are concentrated in southern Lake Michigan, western and central Lake Erie, and the SCDRS in areas with relatively high densities of natural and artificial connections, boat launches and marinas, and large population centers (Figure 6).

Figure 4.

AIS risk scores by grid square for Fish.

Figure 5.

AIS risk scores by grid square for Invertebrates.

Figure 6.

AIS risk scores by grid square for Plants.

Discussion

The findings we report here could improve surveillance prioritization and implementation across the US waters of the Great Lakes Basin in a few key ways. First, our approach facilitates spatially explicit predictions of current and future patterns of risk. The predictions are based on existing pathway surrogate data but could be modified to accommodate new data as pathway dynamics change. Second, predicting invasion pressure in this way helps to objectively prioritize surveillance efforts so that finite surveillance resources can be directed to the subset of “highest risk” sites, which appear to account for a substantial amount of AIS risk. Finally, although risk is concentrated in a few especially high-risk sites, patterns of risk vary by taxa, allowing stakeholders to make decisions about which taxa to target at any given location. These benefits of the site prioritization system, along with some limitations and areas for future improvement are discussed below.

Overall, our model of invasion pressure predicts risk based on the combination of all surrogates that represent a full suite of potential pathways. Although surrogates were generally highly correlated, a single surrogate may not be a good predictor of any other and in that sense each surrogate added value to clarify the potential propagule pressure at each survey unit. For example, based on population alone, West Harbor/Marblehead (Lake Erie), would be considered a low-risk site, but its status as a popular boating and fishing area (or “destination watershed” sensu Davis and Darling 2017) indicates that realized AIS propagule pressure is probably quite high. Conversely, if risk were modeled based solely on number of ponds or marina and boat ramp size, the Chicago River Mouth (Lake Michigan) survey unit would be considered a low-risk site. But including population surrogate data to capture the potential for AIS introduction via the trade in live organisms probably results in a more accurate prediction of real risk for the Chicago area. In general, the most robust predictions of invasion risk are those that consider propagule pressure from all potential pathways.

Pathway weightings could also be explicitly based upon models that quantify the predictive power of each surrogate against the distribution of non-native species in the Great Lakes basin (sensu Leathwick et al. 2016). In early iterations we ran a boosted regression model to assess this approach but given the spatial and taxonomic bias of existing non-native species’ distribution data across the Great Lakes we ultimately rejected the boosted regression method. Furthermore, such an approach would only consider historic patterns of invasion.

Moving forward, pathway weightings and underlying species risk assessments could be modified to reflect the inevitable changes in pathway activity and importance stemming from policy changes, shifts in global trade or economic drivers, adjustments to surrogate values in a survey unit, or new additions to the list of future invaders. Our pathway weightings, based on vectors associated with past and future AIS, indicate that past drivers of non-native species introduction may be changing. The general patterns we derived are consistent with other backward- and forward-looking analyses of vector activity (e.g. Ricciardi 2006, Pagnucco et al. 2015). As compared to historic invaders, a smaller proportion of future invaders are predicted to be introduced via the shipping pathway (pathway weightings 0.43 vs. 0.31 respectively, for all taxa). In our analysis, this pattern appears to be driven primarily by the large proportion of plants on the list of future invaders (44%), as plants are generally not expected in the shipping pathway (weighting 0.09). However, improved ballast water policy might also contribute to an overall reduced risk of ship mediated invasions (Bailey et al. 2011). The concentration of risk at large population centers highlights the increasing importance of the human population vector, which we view as the best surrogate for the trade in live organisms and which has been shown to be a good surrogate for propagule pressure and an accurate predictor for invasive species introductions (Copp 2010, O’Malia 2018). Pathways associated with the population surrogate seem to be increasingly important drivers for species introductions across the entire basin (e.g. trade in live organisms), but shipping or other pathways will continue to be important drivers for some sites.

Another advantage of the spatial framework that we employed is that it allows managers to sort surveillance priorities to specific geographies based on a standardized 9 km x 9 km survey unit. Management resources are finite. Hence, it is important that surveillance efforts concentrate on those sites with the highest risk of introduction (Lodge et al. 2006). Yet, current surveillance efforts for AIS in the Great Lakes basin are often implemented across very large priority areas on the order of hundreds of kilometers (e.g. Green Bay, the SCDRS, and Western Lake Erie, USFWS 2014). In reality, risk is probably not spread evenly across such large areas and each location likely contains multiple high-risk sites. The method we describe is wholly compatible with current surveillance efforts which generally employ random survey designs to select sample stations (Hoffman et al. 2016). But it helps direct managers (and thus survey design) to the highest risk grid cells within larger priority areas. This approach maximizes detection sensitivity rather than spreading limited surveillance effort across wider geographies. Invasion risk scores were near zero for more than two-thirds of the approximately 6,000 sites evaluated, and the majority of risk is concentrated at the same subset of twenty to thirty sites for all taxa (fish, invertebrates, and plants). The relative concentration of risk at a handful of sites around the basin means that monitoring a reasonable number of sites (ca. 50 – 100) could account for a substantial amount of existing risk. Nevertheless, because the boundaries of grid cells are arbitrary, survey design at a site level will always need to consider local geography and features that define the spatial extent of a survey area (e.g. embayment, port area, etc.) and this will mean that a survey site could encompass the coastal area of parts of two or three neighboring grid cells.

Although risk is concentrated in a few especially high-risk sites, patterns of risk vary by taxa for these sites based on vector activity. Such information can be used to determine which taxa to target at a particular location. In general, the risk of invertebrate introduction will be greatest at sites with high levels of shipping activity, fish introduction will be strongly influenced by population density, and plants appear most likely to occur in “destination watersheds” (epicenters of hitchhiking and fouling vectors). For example, Duluth-Superior (MN/WI) ranks 14^th out of 27 US statistical areas (by population) in the Great Lakes basin (US Census Bureau 2018), but it is the busiest port in the Great Lakes and one of the busiest in the United States in terms of total tonnage per annum (Bureau of Transportation Statistics 2017). Consequently, while Duluth-Superior ranks outside the top twenty highest risk sites for fish (25) and plants (39), it is considered the single highest risk site for invertebrate introduction. Given limited surveillance resources, it would make sense that surveillance efforts for invertebrates be directed especially to Duluth-Superior harbor, whereas surveillance for fish or plants may be most productive if implemented in the highest risk areas for those taxa (e.g. Chicago and Toledo).

Separate ranked lists of high-risk sites were developed for fish, invertebrates and plants, in part because, 1) each taxonomic group is best sampled with taxon-specific gears and survey methods and, 2) surveillance efforts are undertaken by independent sampling teams in most instances (e.g. state natural resource agencies). Taxa specific survey designs and gear specifications have been developed for Great Lakes ports and coastal areas for fish (Hoffman et al. 2011, Hoffman et al. 2016) and invertebrates (Trebitz & Hoffman 2009, Trebitz et al. 2010). Survey design and methods are under development for plant surveillance in Great Lakes’ coastal areas (EDRR Team and personal comm, P. Kocovsky). Approaches for plant surveillance in Great Lakes’ coastal wetlands have been described (Trebitz and Taylor 2007) and detailed survey methods for fish and invertebrates in Great Lakes coastal wetlands have also been developed (Burton et al. 2008).

One limitation of our approach is that it focuses primarily on susceptibility of sites to invasion as a function of propagule pressure. The index of invasion pressure implicitly considers the probability of establishment at a broader basin scale because surrogate weightings are based on a list of species that are already established (historic invaders) or that are predicted to have a climate match between source and receiving environments (potential future invaders). But measures of site suitability (i.e. habitat invasibility) could be more explicitly incorporated into future iterations. For example, our approach does not consider implications of in-water habitats and physio-chemical environmental conditions at the point of introduction. Data on abiotic conditions are increasingly being used to predict suitability of the Great Lakes and inland waters to novel AIS based on published environmental tolerances (USEPA 2008, Vander Zanden et al. 2008, Kramer et al. 2017). The Great Lakes Aquatic Habitat Framework now contains over 300 abiotic variables (Wang et al. 2015) and provides an excellent resource to assess environmental suitability for individual taxa. Furthermore, biological invasion theory and empirical data suggest anthropogenic disturbance is correlated with increased establishment success (Marchetti et al. 2004, Havel et al. 2005). The Great Lakes Environmental Assessment and Mapping project (GLEAM; Allen et al. 2013) provides a measure of cumulative and individual human disturbance and could be used to develop a more explicit measure of establishment potential. Measures of species richness, which can in theory be related to the likelihood of species invasion (e.g. Fridley et al. 2007), could also be used to refine predictions of establishment potential.

Future iterations of the plan should also incorporate measures of site irreplaceability and vulnerability (i.e. “site sensitivity,” McGeogh et al. 2016). Margules and Pressey (2001) note the dual importance of preserving a full variety of species and natural processes in conservation planning. Sites with exceptional ecological or economic value (e.g., vulnerable uninvaded areas or areas supporting important fisheries, including large wetland nursery areas), irreplaceable areas with rare or threatened species, areas set aside as parks or wilderness areas, and areas of high biodiversity could be more explicitly targeted for surveillance in future models (Vander Zanden and Olden 2008, Collier et al. 2017, Panlasigui et al. 2018).

Additional spatial analyses are also warranted. For example, site connectivity and potential to promote the spread of novel AIS to other areas of high value based on proximity and the existence of dispersal corridors or pathways could be defined through nearest neighbor analysis or network models (Sieracki et al. 2014, Beletsky et al. 2017, Kvistad et al. In press). Thus, proximity of sites to areas with substantial current, key ballast water uptake zones, or connecting channels and canals that connect the Great Lakes or the Great Lakes Basin to other major catchments (e.g., CAWS-Mississippi River Basin or the Erie Canal) could be a relevant consideration in future site rankings. Finally, recognizing that the Great Lakes are a shared resource and that comprehensive early detection will require management actions on both sides of the US/Canada border, we recommend that future iterations of the model aggregate surrogate data for survey units under Canadian jurisdiction so that risk can be considered on the whole basin scale.

The framework described here provides a useful starting point for surveillance planning and implementation that can be adaptively improved. The surrogates that we use, based on meaningful measures of propagule pressure and invasion history, can be updated as new information regarding potential AIS and associated pathways becomes available. Because the surrogates are geo-referenced, priorities can be sorted to geographies that are most meaningful and manageable for the jurisdictional managers and decision makers tasked with implementing surveillance activities in the basin.

Figure 7.

AIS risk scores by grid square for All Taxa combined.

Table 6.

Pathway ranks based on rescaled data for each surrogate (i.e. 0–100) for the “top 25 highest risk” sites based on the All Taxa ranking. (−) in the “Shipping” pathway indicates the surrogate rank fell outside the top 25 for the given site. For “Connections,” rank order reflects multiple sites with the same surrogate values and (−) indicates absence of connections at the site.

Lake Basin	Location Name	State	Population	RecBoats	Ponds	Shipping	Connections
Michigan	Chicago/Chicago River Mouth	IL	1	5	24	15	1
Erie	Toledo/Maumee River Mouth	OH	6	13	1	3	2
Ontario	Oswego/Oswego River Mouth	NY	8	23	8	11	1
Michigan	Portage/Portage-Burns Waterway	IN	13	17	13	5	1
Erie	Cleveland/Cuyahoga River Mouth	OH	9	11	9	2	3
Huron	Saginaw Bay/Saginaw River Mouth	MI	5	8	4	8	-
Erie	West Harbor/Marblehead/Lake Erie	OH	25	1	18	-	-
Erie	Sandusky/Sandusky Bay	OH	23	2	16	7	-
Erie	Buffalo/Niagara River	NY	12	19	12	9	1
Michigan	Calumet River Mouth/Lake Michigan	IN	22	15	17	16	1
Erie	Grosse Pointe Shores/Lake St. Clair	MI	24	3	22	-	-
Michigan	Benton Harbor/Saint Joseph River	MI	10	9	2	-	-
Michigan	East Chicago/Indiana Harbor Canal	IN	17	20	19	-	1
Michigan	Evanston/North Shore Channel Mouth	IL	18	21	25	-	1
Erie	Lake St. Clair/Clinton River Mouth	MI	19	4	15	-	-
Michigan	Milwaukee/Kinnickinnic River Mouth	WI	7	16	10	4	5
Ontario	Rochester/Genesee River Mouth	NY	16	25	21	-	1
Michigan	Green Bay/Fox River Mouth	WI	11	18	5	6	4
Erie	Lakeside/Lake St. Clair	MI	4	10	7	-	-
Superior	Duluth/St. Louis River Mouth	MN	20	22	20	1	-
Erie	Detroit River/Rouge River Mouth	MI	2	7	14	14	-
Erie	Fairport Harbor/Grand River Mouth	OH	15	12	6	12	5
Erie	Erie/Presque Isle Bay	PA	21	6	23	13	-
Erie	Lorain/Black River Mouth	OH	14	14	11	10	3
Michigan	Grand Haven/Grand River Mouth	MI	3	24	3	17	-