Global change is fundamentally altering flows of natural and anthropogenic subsidies across space and time. After a pointed call for research on subsidies in the 1990s, an industry of empirical work has documented the ubiquitous role subsidies play in ecosystem structure, stability and function. Here, we argue that physical constraints (e.g., water temperature) and species traits can govern a species’ accessibility to resource subsidies, which has been largely overlooked in the subsidy literature. We examined the input of a high quality, point-source anthropogenic subsidy (aquaculture feed) into a recipient freshwater lake food web. By using a combined bio-tracer approach, we detect a gradient in accessibility of the anthropogenic subsidy within the surrounding food web driven by the thermal preferences of three constituent species, effectively rewiring the recipient lake food web. Since aquaculture is predicted to increase significantly in coming decades to support growing human populations, and global change is altering temperature regimes, then this form of food web alteration may be expected to occur frequently. We argue that subsidy accessibility is a key characteristic of recipient food web interactions that must be considered when trying to understand the impacts of subsidies on ecosystem stability and function under continued global change.

Biomass Distribution Data:

The proportion of total biomass of each species sampled during the Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry (MNDMNRF) 2017 Broad-scale Monitoring program across five sites in Georgian Bay were extracted from the Upper Great Lakes Management Unit Lake Huron Report PS-LHA-2017-01 (MNDMNRF, 2018). Data was extracted from Figure 4, Figure 7, Figure 10, Figure 12, and Figure 14 using WebPlot Digitizer. Data was normalized to 100 and summed for each thermal guild. 

Fatty Acid Data:

Muscle tissue samples were collected from fish and whole invertebrates for fatty acid (FA) analysis from both Parry Sound and control sites in the summer of 2016 and 2017. Fish samples were collected using Ontario multi-mesh gill nets following a modified broad-scale monitoring (BsM) protocol of the fish community (Sandstrom et al., 2013) and supplemented with targeted angling. Both overnight and daytime sets were employed in the sampling protocol. All gill nets were set for a duration of ~12 h before retrieving fish. Weight (g), length (mm), and two muscle tissue plugs from behind the dorsal fin were collected from fish. Snails and mussels (Unionidae sp., and Dreissena polymorpha) were collected from each sampling site to provide baseline stable isotope values. Additionally, feed and muscle tissue from farmed rainbow trout was also collected each year from Aqua-Cage Fisheries in Parry Sound. All samples were stored at -20˚C after collection until further analysis. Muscle tissues from a subset of the fish collected were sent for fatty acid analysis, along with feed and baseline samples. Frozen samples were delivered to Ryerson University – The Arts Lab (Toronto, ON, Canada) in 2016, and both the Laboratory of Aquatic Sciences (Chicoutimi, QC, Canada) and Lipid Analytical Services (Guelph, ON, Canada) in 2017. All labs conducted fatty acid analysis using a combination of Bligh & Dyer and Morrison & Smith methods (Bligh & Dyer, 1959; Morrison & Smith, 1964). Individual FA weights (ug/g) were converted to a % FA composition and fatty acids with >1% presence were retained for analysis. See Appendix S2: Section S2 of manuscript for details on statistical methods.

Stable Isotope Data:

A combination of data collected by the McCann Lab and the MNDMNRF were used. Note these methods pertain to samples collected by the McCann Lab in 2016 and 2017. Details on sample collection of additional Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry (MNDMNRF) samples used in stable isotope analysis can be found in the Annual Report of Fisheries Assessment Projects Conducted on Lake Huron, 2017 (MNDMNRF, 2018).Muscle tissue samples were collected from fish and whole invertebrates for stable isotope (SI) analysis from both Parry Sound and control sites in the summer 2017. Fish samples were collected using Ontario multi-mesh gill nets following a modified broad-scale monitoring (BsM) protocol of the fish community (Sandstrom et al., 2013) and supplemented with targeted angling. Both overnight and daytime sets were employed in the sampling protocol. All gill nets were set for a duration of ~12 h before retrieving fish. Weight (g), length (mm), and two muscle tissue plugs from behind the dorsal fin were collected from fish. Snails and mussels (Unionidae sp., and Dreissena polymorpha) were collected from each sampling site to provide baseline stable isotope values. Additionally, feed and muscle tissue from farmed rainbow trout was also collected each year from Aqua-Cage Fisheries in Parry Sound. All samples were stored at -20˚C after collection until further analysis. One muscle plug from all fish, feed, and baseline samples collected were prepared for stable isotope analysis. Samples were thawed in the lab and dried at 60 °C in a drying oven for 48 h. Once dried, they were individually ground using a mortar and pestle and scooped into a labelled centrifuge tube. The samples were sent to the University of Windsor GLIER (Windsor, ON, Canada) laboratories for carbon and nitrogen isotopic analysis. See Appendix S2: Section S4 of manuscript for details on statistical methods.

Hydroacoustic Data:

In July 2017, both night-time and day-time hydroacoustic surveys were conducted in Lake Huron along four transects at increasing distances from the Aqua-Cage Fisheries cage-culture (Parry Sound, Ontario, Canada; Fig. 6). The central (C), transect started closest to the net-pen and was considered the experimental transect with the strongest predicted influence of net-pen, the west (W), east (E), and far eastern (EE) transects were considered reference transects. The C, W and E transects had a maximum depth of 111m, the EE transect had a maximum depth of 91m.  The day-time surveys were conducted from 2:30pm to 7pm, while the night-time surveys were conducted from 11:45pm to 4 am. The thermocline was estimated based on temperature profiles, sitting at approximately 10m below the surface.  All transects started well below the thermocline (all start depths >40m). Hydroacoustic procedures were based on Parker-Stetter et al.’s (2009) “Standard operating procedures for fisheries acoustic surveys in the great lakes”. See Appendix S2: Section S3 of manuscript for full explanation of hydro acoustic survey and data post-processing.