6 Modules

The following sections present the data and approaches used to prepare the data modules to perform both a species-scale cumulative effects assessment and a network-scale cumulative effects assessment, i.e. 1) the distribution of species (\(S_i\)), 2) the normalized distribution and intensity of environmental drivers (\(D_j\)), and 3) the species-specific sensitivity of each species to each driver (\(\mu_{i,j}\)); it further adds 2 modules to complete the assessment: 4) the metaweb of species interactions, i.e. the network of binary biotic interactions structuring local food webs, and 5) the trophic sensitivity of species (\(T_i\)), i.e. their sensitivity to trophically-mediated indirect effects (see below for more details).

6.1 Species

The assessment considers 172 marine, 8 marine mammal, and 25 seabird taxa, for a total of 205 taxa considered. We also assumed that phytoplankton and zooplankton species were present throughout the St. Lawrence System since these taxa are missing from our dataset and are required to properly consider trophic dynamics. The following sections describe the data and approaches used to evaluate their distribution on the Scotian Shelf.

6.1.1 Marine species

6.1.1.1 Biotic data

We used taxa occurrence data from Fisheries and Oceans Canada (DFO) ecosystem spring (Fisheries and ceans Canada, 2020c), summer (Fisheries and ceans Canada, 2020d), and 4vsw (Fisheries and ceans Canada, 2020a) surveys. These surveys collect data to monitor the distribution and abundance of fish and invertebrates through the Scotian Shelf, Bay of Fundy and Georges Bank. We formatted the list of taxa considered for the analyses by combining and removing taxa based on expert knowledge and bibliographic research (e.g. species of the same genus and hard to distinguish in situ were combined). Only taxa with at least 50 observations were considered for the assessment. This process yielded 65764 occurrences and 556191 absences for 172 on the Scotian Shelf (Table 6.1). All species scientific names were resolved using the taxize R package (Chamberlain et al., 2019; Chamberlain and Szöcs, 2013).

Table 6.1: List of marine species obtained through Fisheries and Oceans Canada (DF) ecosystem spring (Fisheries and ceans Canada, 2020c), summer (Fisheries and ceans Canada, 2020d), and 4vsw (Fisheries and ceans Canada, 2020a) surveys on the Scotian Shelf, Bay of Fundy and Goerges Bank and model fit for species distribution models.

Aphia ID	Scientific name	Presences (n)	Absences (n)	\(R^2\)
125837	Aldrovandia	28	3588	0.75
158667	Alosa aestivalis	501	3110	0.23
158669	Alosa pseudoharengus	79	3532	-0.05
158670	Alosa sapidissima	676	2935	0.41
105865	Amblyraja radiata	622	2989	0.20
125909	Ammodytes	636	2975	0.26
126757	Anarhichas denticulatus	7	3604	-0.08
126758	Anarhichas lupus	261	3350	0.10
157181	Aphrodita hastata	361	3250	0.05
105727	Apristurus	34	3584	0.71
138802	Arctica islandica	662	2949	0.14
126715	Argentina silus	142	3469	0.07
107550	Argis dentata	272	3339	0.32
126147	Artediellus	184	3431	0.02
1839	Ascidiacea	788	2935	0.31
145541	Ascophyllum nodosum	212	3399	0.26
159459	Aspidophoroides monopterygius	26	3585	0.04
137683	Astarte	79	3532	-0.13
123219	Asterias	1634	2102	0.18
158490	Astropecten americanus	65	3546	0.41
158351	Atlantopandalus propinqvus	134	3477	-0.13
106057	Balanidae	26	3585	-0.13
138265	Bathypolypus	89	3522	-0.06
125669	Bathypterois	10	3601	0.28
100698	Bolocera	79	3532	-0.15
124404	Brisaster fragilis	76	3535	0.50
126447	Brosme brosme	79	3532	0.07
146142	Bryozoa	64	3547	-0.19
137701	Buccinum	48	3563	-0.22
158056	Cancer borealis	407	3204	-0.03
158057	Cancer irroratus	1176	2435	0.10
105906	Centroscyllium fabricii	78	3533	0.72
124020	Ceramaster granularis	310	3301	0.21
158407	Chaceon quinquedens	252	3359	-0.20
127338	Chauliodus sloani	89	3522	-0.14
125956	Chiasmodon	21	3591	0.40
107315	Chionoecetes opilio	723	2888	0.57
140692	Chlamys islandica	67	3544	0.03
126336	Chlorophthalmus agassizi	54	3557	-0.01
158791	Citharichthys arctifrons	31	3580	-0.13
126417	Clupea harengus	1959	1652	0.27
137704	Colus	684	2938	0.37
158960	Coryphaenoides rupestris	493	3118	0.44
127235	Cottunculus microps	16	3595	-0.25
158355	Crangon septemspinosa	148	3463	0.13
124154	Crossaster papposus	781	2830	0.41
159675	Cryptacanthodes maculatus	25	3586	-0.13
123915	Ctenodiscus crispatus	361	3250	0.38
124612	Cucumaria frondosa	645	2966	0.46
135301	Cyanea capillata	61	3550	0.38
127214	Cyclopterus lumpus	48	3563	0.05
158356	Dichelopandalus leptocerus	351	3260	0.06
124128	Diplopteraster multipes	45	3566	-0.19
158548	Dipturus laevis	68	3543	-0.15
158062	Echinarachnius parma	826	2785	0.32
126450	Enchelyopus cimbrius	266	3345	0.24
101027	Epizoanthus erdmanni	69	3542	-0.05
159817	Eumesogrammus praecisus	109	3502	0.02
127217	Eumicrotremus spinosus	41	3570	0.00
515738	Eusergestes arcticus	19	3592	-0.17
135114	Flabellum	56	3555	-0.15
111367	Flustra foliacea	116	3495	0.26
126436	Gadus morhua	1655	1956	0.15
125743	Gaidropsarus	11	3600	0.27
156103	Gersemia rubiformis	131	3480	0.42
127136	Glyptocephalus cynoglossus	53	3558	0.19
125601	Gonostomatidae	45	3572	0.60
123586	Gorgonocephalus	287	3324	0.50
127251	Helicolenus dactylopterus	8	3603	-0.15
159518	Hemitripterus americanus	455	3156	0.05
123276	Henricia	1337	2278	0.12
124043	Hippasteria phrygiana	768	2843	0.25
158833	Hippoglossina oblonga	19	3592	-0.25
127137	Hippoglossoides platessoides	1407	2204	0.36
127138	Hippoglossus hippoglossus	1689	1922	0.19
156134	Homarus americanus	102	3509	-0.02
107322	Hyas araneus	421	3190	0.00
107323	Hyas coarctatus	535	3076	0.07
126150	Icelus	28	3583	-0.09
153087	Illex illecebrosus	179	3432	0.47
107520	Lebbeus groenlandicus	43	3568	-0.08
107521	Lebbeus polaris	217	3394	-0.10
158762	Lepophidium profundorum	25	3586	-0.05
127191	Leptagonus decagonus	326	3285	0.60
123222	Leptasterias	495	3137	0.20
158551	Leucoraja erinacea	356	3255	0.02
158553	Leucoraja ocellata	847	2764	0.35
158879	Limanda ferruginea	35	3576	-0.11
159524	Liparis atlanticus	964	2647	0.26
127218	Liparis fabricii	78	3533	0.10
159526	Liparis gibbus	36	3575	-0.21
107205	Lithodes maja	331	3280	0.16
157025	Loligo pealeii	1523	2088	0.56
159184	Lophius americanus	2267	1344	0.37
154675	Lumpenus lampretaeformis	1704	1907	0.01
305353	Lumpenus maculatus	238	3373	0.55
402541	Lycenchelys verrilli	24	3587	-0.16
127107	Lycodes lavalaei	45	3566	-0.11
127117	Lycodes terraenova	NA	3611	NA
127118	Lycodes vahlii	5	3606	-0.10
158554	Malacoraja senta	815	2796	0.22
126735	Mallotus villosus	108	3503	0.09
127312	Maurolicus muelleri	25	3586	0.69
126437	Melanogrammus aeglefinus	2629	982	0.52
127120	Melanostigma atlanticum	119	3492	0.22
158748	Merluccius albidus	57	3554	0.45
158962	Merluccius bilinearis	36	3575	0.57
140467	Modiolus modiolus	64	3547	-0.14
125498	Myctophidae	386	3367	0.61
126152	Myoxocephalus	1582	2135	0.25
138228	Mytilus	234	3377	-0.05
101170	Myxine glutinosa	369	3242	0.10
145	Naticidae	416	3196	0.35
126306	Nemichthys scolopaceus	64	3547	0.18
137710	Neptunea	138	3473	-0.07
183289	Nezumia bairdii	58	3553	-0.12
147037	Notolepis rissoi	270	3354	-0.10
125125	Ophiopholis aculeata	266	3345	-0.14
123574	Ophiura	175	3436	-0.06
293649	Osmerus mordax mordax	19	3592	-0.19
106854	Pagurus	159	3458	-0.07
107649	Pandalus borealis	455	3156	0.38
107651	Pandalus montagui	1662	1949	0.16
125418	Paragorgia arborea	45	3566	0.01
158826	Paralichthys dentatus	294	3317	0.12
158868	Parasudis truculenta	23	3588	0.02
107676	Pasiphaea multidentata	296	3315	0.43
128515	Pennatula aculeata	46	3565	0.53
159828	Peprilus triacanthus	78	3533	-0.05
101174	Petromyzon marinus	64	3547	-0.01
126441	Pollachius virens	1001	2610	0.21
158840	Polyipnus clarus	204	3407	0.28
939	Polynoidae	92	3519	-0.22
107563	Pontophilus norvegicus	186	3425	0.10
125166	Porania pulvillus	312	3299	0.26
123321	Poraniomorpha	166	3446	-0.05
558	Porifera	2000	1893	0.23
159571	Prionotus carolinus	34	3577	-0.14
124085	Pseudarchaster parelii	232	3383	0.07
158885	Pseudopleuronectes americanus	1174	2437	0.44
123908	Psilaster andromeda	299	3312	0.06
124147	Pteraster militaris	239	3372	-0.03
124151	Pteraster pulvillus	75	3536	-0.07
127144	Reinhardtius hippoglossoides	775	2836	0.11
178806	Sclerasterias tanneri	128	3483	0.30
107568	Sclerocrangon boreas	64	3547	0.18
127023	Scomber scombrus	628	2983	-0.03
236461	Scomberesox saurus saurus	211	3400	0.57
127278	Scopelogadus beanii	400	3211	0.58
158907	Scophthalmus aquosus	811	2800	0.51
126175	Sebastes	1526	2085	0.49
107136	Sergia robusta	29	3585	0.34
126319	Serrivomer beanii	37	3574	0.83
124160	Solaster endeca	409	3202	0.14
107531	Spirontocaris liljeborgii	122	3489	-0.01
107533	Spirontocaris spinus	87	3524	-0.10
105923	Squalus acanthias	701	2910	0.15
158737	Stomias boa ferox	371	3240	0.08
123390	Strongylocentrotus	981	2634	0.51
159358	Symphurus diomedeanus	427	3184	-0.01
126328	Synaphobranchus kaupii	53	3558	0.64
159451	Syngnathus fuscus	53	3558	-0.03
159785	Tautogolabrus adspersus	34	3577	-0.16
126154	Triglops	339	3277	0.16
159821	Ulvaria subbifurcata	24	3587	-0.14
123815	Urasterias lincki	53	3558	-0.09
301114	Urophycis chesteri	25	3586	-0.24
126503	Urophycis chuss	1273	2338	0.32
126504	Urophycis tenuis	1091	2520	0.40
172121	Vazella pourtalesi	423	3188	0.08
307180	Zenopsis ocellata	141	3470	0.31
159267	Zoarces americanus	624	2987	0.06

6.1.1.2 Abiotic data

We used environmental data characterizing the bottom-water and surface-water conditions on the Scotian Shelf using data from 1) Bio-ORACLE, which provides geophysical, biotic and environmental data layers for surface and benthic marine environments (Assis et al., 2018; Bosch and Fernandez, 2022; Tyberghein et al., 2012), 2) the General Bathymetric Chart of the Oceans (GEBCO) offering a global terrain model for ocean and land, providing elevation data, in meters, on a 15 arc-second interval grid (Group, 2021), and 3) the Bedford Institute of Oceanography North Atlantic Model (BNAM) averaged over the 1990 to 2015 period for monthly temperatures and salinities at the bottom and at the surface (Wang et al., 2018a, 2018d, 2018c). See table 6.2 for more details.

Table 6.2: List of abiotic variable considered to model the distribution of taxa on the Scotian Shelf.

Name	Source
Bottom chlorophyll	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Bottom iron	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Bottom light	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Bottom nitrate	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Bottom phosphate	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Bottom phytoplankton	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Bottom primary productivity	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Bottom silicate	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface calcite	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface chlorophyll	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface dissolved oxygen	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface iron	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface nitrate	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface ph	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface phosphate	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface phytoplankton	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface primary productivity	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Surface silicate	Bosch and Fernandez (2022); Tyberghein et al. (2012); Assis et al. (2018)
Bathymetry	Group (2021)
Bottom temperature	Wang et al. (2018a); Wang et al. (2018d)
Surface temperature	Wang et al. (2018a); Wang et al. (2018d)
Bottom salinity	Wang et al. (2018a); Wang et al. (2018c)
Surface salinity	Wang et al. (2018a); Wang et al. (2018c)

6.1.1.3 Taxa distribution

We evaluated the distribution of taxa on the Scotian Shelf using the Random Forest ensemble learner (Breiman, 2001). We used the default parameters of the randomForest R package to classify the presence or absence of taxa: 500 trees and the number of variables in the random subset at each node set to the square root of the number of variables (Liaw and Wiener, 2002). All 23 abiotic variables (Table 6.2) were used to model the distribution of species. The performance of each model was evaluated using the \(R^2\) of each species’ regression tree [Table 6.1; Allouche et al. (2006)]. We predicted the spatial distribution of all taxa within the grid of our area of interest and performed a smoothing of predictions using a bisquare kernel smoothing approach (Dos Santos et al., 2018) with a 5 km radius to avoid granular distributions; this is done to minimize the effect on the assessment of species co-occurrences necessary for the subsequent network-scale cumulative effects assessment. A threshold of 0.5 was used to transform continuous predictions to binary distributions, identifying areas where taxa are most likely to be found.

6.1.2 Marine mammals

The distribution of marine mammals in the area of interest comes from the second edition of a guidebook published for mariners who frequent the Northwest Atlantic (Le WWF-Canada et le R’eseau d’observation de mammif‘eres marins, 2021a). The purpose of this guide is to inform mariners on the problem of collisions between vessels and whales, how to minimize such incidents, to present the various species of whales and leatherback turtles in the area and where they might be present, and on areas where heightened vigilance is required. Nine species of marine mammals were selected for this guide based on their conservation status and their known risk of collision with vessels (Le WWF-Canada et le R’eseau d’observation de mammif‘eres marins, 2021a, 2021b).

The distribution of each marine mammal was assessed through the integration of 14 datasets from scientific and opportunistic surveys between 2010-2015 (Table 6.3; Le WWF-Canada et le R’eseau d’observation de mammif‘eres marins (2021c)). The area of interest covered by the guide covers the geographic boundaries of \(40^o\) to \(55^o\) N and -72\(^o\) to -48\(^o\) W. The area of interest was divided into a regular 0.05\(^o\) x 0.05\(^o\) grid and the number of sightings recorded in the 14 available databases was calculated for each species considered. A Gaussian smoothing – i.e. an interpolation method based on a normal distribution – of 0.2\(^o\) was then applied to the grid. Values in each grid cell were then normalized by the total number of sightings for a species and log-transformed to minimize the effect of extreme values.

The caveats that accompany these maps in the guide are reported verbatim in this report:

• Sighting effort was not quantified and varies considerably in time and space. The data represent the relative occurrence of reported sightings rather than the actual density or abundance of the species.

• The quality of some of the sighting data is unknown. Sightings are reported from individuals with varying degrees of expertise in identifying marine mammals.

Table 6.3: List of scientific and opportunistic datasets used to generate maps, including informations on the period covered by the data, the number of species and the number of observations of one or more individuals available for the analysis, and the owner of the dataset. Table from: Le WWF-Canada et le R’eseau d’observation de mammif‘eres marins (2021c).

Name of dataset	Temporal series	Owner	Species (n)	Observations (n)	Scientific (S) or Opportunistic (O)
Whitehead, Université Dalhousie	1988-2019	Université Dalhousie	27	2464	S/O
WWAM	2015-2019	Parcs Canada	7	1353	S
WWAM_C	2015-2019	Parcs Canada	7	1774	S
Predator pelagic prey	2015-2018	Parcs Canada	7	816	S
NAISS	2016	Pêches et Océans Canada	39	2508	S
NARW	2017	Pêches et Océans Canada	31	2910	S
NARW	2018	Pêches et Océans Canada	39	7312	S
WSDB	1963-2019	Base de données des observations de mammifères marins et d’animaux pélagiques, Pêches et Océans Canada, Dartmouth, N.-É. [2020/02/13]	67	24538	O
TC	2018	Transports Canada	19	3517	S
MICS	2014-2018	Station de recherche des îles Mingan	11	4808	S
NOAA	2018-2020	National Oceanic and Atmospheric Administration	5	2699	S
NARWC	2015-2019	North Atlantic Right Whale Consortium	16	18190	S
AOM_MMON	2014-2019	Réseau d’observation de mammifères marins	8	4705	S
Shipping_MMON	2015-2019	Réseau d’observation de mammifères marins	35	5891	O

Of the 9 species available in the guide, 8 are present on the Scotian Shelf (Table 6.4). For our study, we resampled the distributional maps available in the guide to integrate them into our own grid. The values were then normalized between 0 and 1 by dividing each value by the maximum value observed within the grid for each species. The resulting values can be interpreted as the relative density of marine mammal sightings, with values of 0 representing a low relative density and a value of 1 a high relative density. All cells with values greater than 0 were used to identify the binary distribution of marine mammals.

Table 6.4: List of marine mammal distributions obtained through the the second edition of a guidebook published for mariners who frequent the Northwest Atlantic (Le WWF-Canada et le R’eseau d’observation de mammif‘eres marins, 2021a, 2021b, 2021c).

Aphia ID	Scientific name	Common name
137090	Balaenoptera musculus	Blue whale
137091	Balaenoptera physalus	Fin whale
137092	Megaptera novaeangliae	Humpback whale
137087	Balaenoptera acutorostrata	Minke whale
159023	Eubalaena glacialis	North atlantic right whale
343899	Hyperoodon ampullatus	Northern bottlenose whale
137088	Balaenoptera borealis	Sei whale
137119	Physeter macrocephalus	Sperm whale

6.1.3 Seabirds

The distribution of seabirds was obtained through vessel-based and aerial-based surveys and available through the Eastern Canada Seabirds at Sea (ECSAS) database (Gjerdrum et al., 2012; Service et al., 2022). All sightings of individual bird species or bird functional groups from the vessel and aerial surveys between 2006 and 2022 were combined to evaluate the distribution of birds in the area of interest. A total of 25 bird species / groups were obtained (Table 6.5).

Table 6.5: List of seabirds obtained through the Eastern Canada Seabirds at Sea (ECSAS) database (Gjerdrum et al., 2012; Service et al., 2022).

Alpha	Aphia ID	Scientific name	Observations (n)
RAZO	137128	Alca torda	151
ALCI	136987	Alcidae	396
DOVE	137129	Alle alle	2387
COSH	1348497	Calonectris borealis	761
BLGU	137130	Cepphus grylle	238
LTDU	137071	Clangula hyemalis	150
ATPU	137131	Fratercula arctica	409
NOFU	137195	Fulmarus glacialis	1683
UNLO	136996	Gaviidae	205
UNSP	148794	Hydrobatidae	3776
UNLA	137043	Larus	273
HERG	137138	Larus argentatus	2519
GBBG	137146	Larus marinus	1302
UNSC	137002	Melanitta	177
NOGA	148776	Morus bassanus	2111
UNCO	137054	Phalacrocorax	240
UNPH	137049	Phalaropus	988
GRSH	137201	Puffinus gravis	3767
SOSH	137202	Puffinus griseus	320
MASH	137203	Puffinus puffinus	74
BLKI	137156	Rissa tridactyla	632
COEI	137074	Somateria mollissima	487
UNJA	137051	Stercorarius	192
UNTE	148764	Sternidae	92
UNMU	137041	Uria	1694

To evaluate the distribution of birds, a similar approach to that used for marine mammals was used. The area of interest was divided into a regular 0.05\(^o\) x 0.05\(^o\) degrees grid. For each species or species group, the number of sightings recorded in the ECSAS database was evaluated in each grid cell. A Gaussian smoothing - i.e. an interpolation method based on a normal distribution - of 20 \(km\) was then applied to the grid. A log transformation was then applied to minimize the effect of extreme values on the resulting maps. Cell values were then normalized between 0 and 1 by dividing by the maximum value observed for each bird across the area of interest. All cells with values greater than 0 were used to identify the binary distribution of seabirds.

6.2 Environmental drivers

We characterized the distribution and intensity of 17 environmental drivers on the Scotian Shelf during two distinct periods, i.e. 2010 to 2015 and 2016 to 2021. Temporal data was unavailable for certain drivers; in such instances, we used the same data for both periods (e.g. shipping). Drivers are divided in 4 groups: climate (n = 4), land-based / coastal (n = 6), fisheries (n = 5) and marine traffic (n = 2; Table 6.6). Drivers were log-transformed to avoid underestimating intermediate intensity values. All drivers were also scaled between 0 and 1 to obtain relative intensities and allow driver comparisons. Scaling was performed using the 99th quantile of intensity distribution as the upper bound to control for extreme values. For more details, see Beauchesne et al. (2020) and Halpern et al. (2019). The following subsections describe the approach used to evaluate the distribution and the intensity of each driver considered for the cumulative effects assessment.

Table 6.6: List of environmental drivers considered for the cumulative effects assessment of global changes on the ecological communities of the Scotian Shelf.

Groups	Drivers	Spatial resolution	Temporal resolution	Years	Units	Source
Climate	Negative sea bottom temperature anomalies	0.2 degree	Annual	2010-2019	negative anomalies	Fisheries and ceans Canada (2022)
Climate	Negative sea surface temperature anomalies	~2 \(km^2\)	Monthly	2010-2021	negative anomalies	Fisheries and ceans Canada (2021b)
Climate	Positive sea bottom temperature anomalies	0.2 degree	Annual	2010-2019	positive anomalies	Fisheries and ceans Canada (2022)
Climate	Positive sea surface temperature anomalies	~2 \(km^2\)	Monthly	2010-2021	positive anomalies	Fisheries and ceans Canada (2021b)
Coastal	Coastal development	15 arcsecond	Annual	2012-2021	\(nanoWatts\) \(cm^{−2}\) \(sr^{−1}\)	Elvidge et al. (2021)
Coastal	Direct human impact	< 1 to > 40000 \(km^2\)	Annual	2016,2021	population count	Canada (2016a); Canada (2016b); Canada (2017); Canada (2022a); Canada (2022b); Canada (2022c)
Coastal	Inorganic pollution	Modeled $100 \(m^2\)	-	Betwen 2010 and 2019	% cover agriculture land	Guijarro-Sabaniel and Kelly (2022)
Coastal	Nutrient input	Modeled $100 \(m^2\)	-	Betwen 2010 and 2019	\(kg\) \(N\) \(yr^{-1}\)	Kelly et al. (2021); Guijarro-Sabaniel and Kelly (2022)
Coastal	Organic pollution	Modeled $100 \(m^2\)	-	Betwen 2010 and 2019	% cover impervious surface	Guijarro-Sabaniel and Kelly (2022)
Coastal	Population density	Modeled $100 \(m^2\)	-	2015,2019	person \(ha^{-1}\)	Guijarro-Sabaniel and Kelly (2022)
Fisheries	Demersal, destructive	Lat/Lon	Event based	2010-2020	\(kg\)	Fisheries and ceans Canada (2021b)
Fisheries	Demersal, non-destructive, high-bycatch	Lat/Lon	Event based	2010-2020	\(kg\)	Fisheries and ceans Canada (2021b)
Fisheries	Demersal, non-destructive, low-bycatch	Lat/Lon	Event based	2010-2020	\(kg\)	Fisheries and ceans Canada (2021b)
Fisheries	Pelagic, high-bycatch	Lat/Lon	Event based	2010-2020	\(kg\)	Fisheries and ceans Canada (2021b)
Fisheries	Pelagic, low-bycatch	Lat/Lon	Event based	2010-2020	\(kg\)	Fisheries and ceans Canada (2021b)
Marine traffic	Invasive species	0.01 degree	Annual	-	\(n\) species	Lyons et al. (2020a)
Marine traffic	Shipping	0.1 degree	Monthly	2017-2020	\(n\) ship lanes	Global Fishing Watch (2022)

6.2.1 Climate

6.2.1.1 Sea surface temperature anomalies

The data used to characterize sea surface temperature anomalies come from the Atlantic Zone Monitoring Program [AZMP; Galbraith et al. (2021)] from the Department of Fisheries and Oceans (DFO). A full description of the data and methods can be found in Galbraith et al. (2021), and we used a similar approach to that described by Beauchesne et al. (2020) to evaluate positive and negative temperature anomalies. Here, we provide a brief summary of the approach.

The surface layer is characterized using sea surface-water temperature (SST) monthly composites from Advanced Very High Resolution Radiometer (AVHRR) satellite images obtained from the National Oceanic and Atmospheric Administration (NOAA) and European Organization for the Exploitation of Meteorological Satellites (EUMETSAT). Images used are from DFO’s Maurice Lamontagne Institute at a 1 km resolution from 1985-2013 and from DFO’s Bedford Institute of Oceanography (BIO) Operational Remote Sensing group at a 1.5 km resolution since 2014. Monthly anomalies were constructed as the difference between monthly averages and the 1985-2010 climatological mean for each month.

We used temperature anomalies, i.e. deviations from long-term normal conditions between 1985 and 2010, to evaluate an annual index of stress associated with extreme temperatures between 2010 and 2021. Temperature anomalies were calculated using the difference between grid cell values with 1985-2010 climatological averages. Anomaly time series were normalized by their standard deviation (SD) to allow comparisons between areas with different temperature ranges. Grid cells whose monthly value exceeded ±0.5 standard deviation (SD) from the long-term average were considered as anomalous (Galbraith et al., 2021). Outliers in the data were defined as those that fell beyond the interquartile range * 3, identified as extreme outliers by Tukey (1977). Outlier values were capped to correspond to the 5th and 95th percentile values. Anomalies were divided into positive and negative anomalies and the absolute value of anomalies were used as the intensity of the surface-water temperature anomalies in a given cell. Only the months of May to November were included to avoid biases associated with the presence of ice cover. The sum of anomalies between 2010 and 2015, and between 2016 and 2021, were used to obtain a single value per cell per period considered for the assessment. The anomalies were then resampled on the study grid used for this assessment.

6.2.1.2 Sea bottom temperature anomalies

The data used to characterize bottom-water temperature anomalies come from the Department of Fisheries and Oceans’ (DFO) temperature data from the summer groundfish survey on the Scotian Shelf. Bottom-water temperatures are interpolated from conductivity-temperature-depth (CTD) sampling performed annually at a 0.2 degrees resolution (Fisheries and ceans Canada, 2022). Anomalies are evaluated as a cell value for a given year minus that cell’s value from a historical climatology. Anomalies were divided into positive and negative anomalies and the absolute value of anomalies were used as the intensity of the surface-water temperature anomalies in a given cell. Grid cells that exceeded 0.5 were considered as anomalous for this assessment. The mean anomaly intensity between 2010 and 2015, and between 2016 and 2019, for each grid cell was used to generate the final index of bottom-water temperature anomalies for each period considered.

6.2.2 Coastal

6.2.2.1 Coastal development

We used lights at night as a proxy of the presence of coastal infrastructure development, as stable lights mostly capture human settlements and industrial sites. We used data from the Nighttime Lights Time Series. Nighttime light products are compiled by the Earth Observation Group at the National Oceanic and Atmospheric Administration’s (NOAA) National Centers for Environmental Information (NCEI). They use globally available nighttime data obtained from the Visible Infrared Imaging Radiometer Suite (VIIRS) Day/Night Band (DNB) of the Defense Meteorological Satellite Program (DMSP) to characterize global average radiance (\(nanoWatts\) \(cm^{-2}\) \(sr^{-1}\)) composite images at a 15 arc-second (~200 m) resolution. We used the annual Version 2 Nighttime VIIRS DNB composites between 2012 and 2021 (Elvidge et al., 2021) to characterize coastal development in coastal areas of the Scotian Shelf. As the effects of coastal development are likely acute in its direct vicinity, we extracted average radiance values using a 2 km buffer around grid cells within 2 km of the coast. In each grid cell, we used the mean annual radiance as a proxy of the effects of coastal development in coastal areas. The first period (2010-2015) only contains 4 years of data, since radiance values are unavailable prior to 2012.

6.2.2.2 Direct human impact

We used the sum of coastal populations as a proxy of direct human impact in coastal areas of the Scotian Shelf. We used Statistics Canada dissemination area and associated population counts from the 2016 (Canada, 2016a, 2016b, 2017) and 2021 (Canada, 2022a, 2022b, 2022c) censuses to evaluated the population size in coastal areas. Dissemination areas are the smallest standard geographic area in which census data are disseminated. They combine to cover all of Canada and are highly variable in shapes and sizes. The census provides population count within the boundary of each dissemination area, which we used to evaluate total coastal population.

As the effects of direct human impacts are likely acute mostly in coastal areas we calculated total population in grid cells within 2 km of the coast. Total population was measured in a 10 km buffer around each coastal cell. The total population (\(DHI\)) in each buffer was the sum of intersecting dissemination areas divided by the intersection area between buffers and dissemination areas:

\[DHI_j = \sum_{k=1}^{n_j} P_k * \frac{A_{j,k}}{A_{tot, k}}\]

where \(j\) is a buffered grid cell, \(k\) is a dissemination area intersecting \(j\), \(P\) is the population in \(k\), \(A\) is the area of the \(k\) overlapping with \(j\) and \(A_{tot}\) is the total area of \(k\). We used this approach to reduce the effects of very large dissemination areas overlapping with buffers on a very small percentage of their total area. We used the mean values from the 2016 and 2021 census for both periods of the assessment, as we did not have the data for the period between 2010 and 2015.

6.2.2.3 Land-based drivers

The data used to characterize the intensity of inorganic pollution, organic pollution, population density and nutrient input in coastal areas come from an atlas of land use for coastal watershed in the Maritimes Region, hereafter the Land Use Atlas (Guijarro-Sabaniel and Kelly, 2022). We provide a brief summary of data and methods in this document. For more details, refer to Guijarro-Sabaniel and Kelly (2022).

Coastal watershed boundaries in the Scotian Shelf Biogeographic Marine Management Sub-region were delineated using a digital elevation model (DEM) for each province was combined with national hydrographic data and calculating the flow direction and accumulation for each DEM. Stream orders for the stream network were then evaluated to locate the highest flow accumulation value for each watershed and given a value between 1 (lowest) and 4 (highest). The locations of these values were then used to identify pour points (e.g. river mouths) along the coast for each watershed. Spread of land stressors into coastal and marine environments was then modeled using a diffusive plume model using a density decay buffer based on stream order of the main pour points.

The following sections describe how this model was used to characterize the intensity of inorganic pollution, nutrient input, organic pollution and population density in coastal areas of the Scotian Shelf.

6.2.2.3.1 Inorganic pollution

The intensity of inorganic pollution was modeled using the percentage of impervious area in each watershed.. Impervious surfaces refer to artificial structures such as roads and parking lots and industrial areas such as ports that are covered by water-resistant materials. Impervious surfaces are used as a proxy of inorganic pollution under the assumption that most of this pollution comes from urban runoff. Impervious surfaces were identified in watersheds using a combination of multiple datasets from the provinces New Brunswick and Nova Scotia. Refer to the Land cover section and Table 1 of the Material and Methods in Guijarro-Sabaniel and Kelly (2022) for more details.

6.2.2.3.2 Organic pollution

We used the percent cover of agriculture land use in each watershed as a proxy of the intensity of organic pollution in each watershed. The percent area covered by agricultural land in each watershed was identified in watershed using a combination of multiple datasets from the provinces New Brunswick and Nova Scotia. Refer to the Land cover section and Table 1 of the Material and Methods in Guijarro-Sabaniel and Kelly (2022) for more details.

6.2.2.3.3 Population density

This intensity of stress caused by watershed population in coastal areas was modeled using the population density in each watershed in \(person * ha^{-1}\). Total population was estimated as the number of civic address datasets from New Brunswick and Nova Scotia multiplied by the average number of residents per household (i.e. 2.3) obtained from Statistics Canada. Population density was then evaluated as the total population in a watershed divided by the watershed area in hectares.

6.2.2.3.4 Nutrient input

We used data from Kelly et al. (2021) to obtain the intensity of nutrient input in coastal areas of the Scotian Shelf. Kelly et al. (2021) used data from the Land Use Atlas to estimate nitrogen loading in coastal areas with the Nitrogen Loading Model (NLM) framework (McIver et al., 2015; Nagel et al., 2018; Valiela et al., 1997; Valiela et al., 2000) to estimate total (dissolved) nitrogen from point and nonpoint source inputs in coastal areas from atmospheric deposition, fertilizer use, and wastewater disposal. Data and methods used to apply the NLM framework are described in detail in Kelly et al. (2021).

6.2.3 Fisheries

The intensity and distribution of commercial fisheries on the Scotian Shelf was assessed using the Department of Fisheries and Oceans logbook program (Fisheries and ceans Canada, 2021b); these logbooks provide a thorough, if not exhaustive, overview of fisheries in eastern Canada. We used data between 2010 and 2021 to characterize the distribution and the intensity of commercial fisheries on the Scotian Shelf. This period was characterized by 2986667 fishing activities targeting 80 species and a total of 111 species were captured when considering bycatch.

Fishing activities are performed using a variety of gear types (e.g. trap, trawl, dredge, driftnet, hand line, longline, scuba diving, purse seine, seine, beach seine and jig fishing), which can have different effects on species and habitats. Fisheries activities were therefore divided based on the type of environmental effects attributed to specific gear types according to the categories suggested by Halpern et al. (2008) and used by Beauchesne et al. (2020) for the Estuary and Gulf of St. Lawrence (Tables 6.7 and 6.8): demersal, destructive, high bycatch (DD) demersal, non-destructive, high bycatch (DNH) demersal, non-destructive, low bycatch (DNL) pelagic, high bycatch (PHB) pelagic, low bycatch (PLB)

Gear types can be further classified into fixed or mobile gear (Table 6.8). We used the type of mobility to evaluate an expected area of effect for each fishing event, using radii of 200 and 2000 meters for fixed and mobile gear types, respectively (Beauchesne et al., 2020). This approach, while decreasing precision, considers the potential uncertainty associated with fishing activity coordinates, gear mobility, and the absence of start and end coordinates for mobile gear.

Table 6.7: Description of commercial fishing catagories used to characterize the intensity and distribution of fisheries on the Scotian Shelf.

Accronym	Category	Description
DD	Demersal, destructive, high-bycatch	Commercial fishing activities using demersal fishing gear that may damage habitats or substrate, e.g. trawling and dragging.
DNL	Demersal, non-destructive, low-bycatch	Commercial fishing activities using demersal fishing gear with little or no bycatch and not causing habitat modification, e.g., deep-sea fishing.
DNH	Demersal, non-destructive, high-bycatch	Commercial fishing activities using demersal fishing gear with high bycatch and not causing habitat modification, e.g., trap and seine.
PLB	Pelagic, low-bycatch	Commercial fishing activities using pelagic fishing gear with little or no bycatch and not causing habitat modification, e.g., line fishing, purse seine.
PHB	Pelagic, high-bycatch	Commercial fishing activities using pelagic fishing gear with high bycatch and not causing habitat modification, e.g., gillnet and longline.

Table 6.8: Classification of gear types based on their environmental effects and mobility. DD: demersal, destructive, high bycatch; DNH: demersal, non-destructive, high bycatch; DNL: demersal, non-destructive, low bycatch; PHB: pelagic, high bycatch; PLB: pelagic, low bycatch. Adapted from Beauchesne et al. (2020).

Type of gear	Category	Mobility
Trap	DNH	Fixed
Bottom trawl	DD	Mobile
Drag	DD	Mobile
Gillnet	PHB	Fixed
Line fishing	PLB	Fixed
Longline	PHB	Fixed
Diving	DNL	Fixed
Purse seine	PLB	Fixed
Danish or Scottish seine	DNH	Fixed
Shoreline seine	DNH	Fixed
Trap	DNH	Fixed
Jigger	PLB	Fixed

To characterize the intensity of fishing activities, we used a biomass yield density index by summing the total annual biomass captured in each grid cell covering the Scotian Shelf, yielding an assessment in kg per cell surface, i.e. \(kg * km^{-2}\) in this case. For each period considered in the assessment (2010-2015 and 2016-2021) and each fishing category, we used the mean annual biomass captured per grid cell to capture the intensity and distribution of fisheries on the Scotian Shelf.

6.2.4 Marine traffic

6.2.4.1 Shipping

The data used to characterize shipping come from the Global Fishing Watch, which provides shipping data from Automatic Identification System (AIS) data globally. We obtained data between 2017 and 2020 that characterize monthly commercial shipping (e.g. cargo, reefer, tanker, bunker) as rasters with a 0.1 degree resolution, either as the number of vessels or the total hours of vessel presence in each grid cell (Global Fishing Watch, 2022). The data was also available either as direct observations from the AIS data, or interpolated to a regular interval of five minutes between points. We used the sum of the monthly interpolated number of all vessel types in each grid cell to obtain an assessment of annual shipping intensity on the Scotian Shelf. We then resampled the 0.1 degree resolution grid to our own study grid. We used the mean annual shipping intensity value between 2017 and 2020 for both periods considered in the assessment.

6.3 Species-specific sensitivity

We evaluated the relative species-specific sensitivity of all 205 taxa considered in the assessment to each stressor using a trait-matching approach, as in Beauchesne (2020). For each driver considered, Beauchesne (2020) identified traits that were known or suspected to influence the sensitivity of a species to the direct effects of that particular driver. For instance, the feeding strategy of an organism can affect its sensitivity to certain stressors. Traits were categorized to reflect their relative contribution to the sensitivity of a taxa to the effects of a stressor and allow for a comparison between taxa. For example, suspension feeders are generally more affected by nutrients and metals than deposit feeders (Ellis et al., 2017).

Beauchesne (2020) documented the body composition, the maximal body size, the type of marine environment in which species are found, the feeding mode, the mobility and the phylum of 391 taxa from the St. Lawrence. Traits data were extracted manually and automatically from several sources. Primary sources were the World Register of Marine Species [WoRMS; WoRMS Editorial Board (2017)], FishBase (Froese and Pauly, 2019), SeaLifeBase (Palomares and Pauly, 2019), the Encyclopedia of Life [EoL; Encyclopedia of Life (2020)] and the Global Biotic Interaction (GloBI) database (Poelen et al., 2014; Poelen et al., 2019). We used the taxize (Chamberlain et al., 2019; Chamberlain and Szöcs, 2013), worrms (Chamberlain, 2020) and rfishbase (Boettiger et al., 2012) R packages to extract traits data. Manual searches, mainly on the WoRMS and EoL web portals, were performed and documented when programmatic extractions were unavailable. The traits dataset from Beauchesne (2020) was updated for this assessment by using the same approach and resources to consider species of the Scotian Shelf, and to include species of marine mammals and seabirds that were not considered in Beauchesne (2020) (Table 6.9). Species that were targeted or caught as bycatch on the Scotian Shelf were also identified using the fisheries logbook data from DFO (Fisheries and ceans Canada, 2021b). The code associated with traits extraction is available at https://github.com/Ecosystem-Assessments/Species_Traits.

For each trait and taxa combination, traits were given a weight between 0 and 1 to reflect the relative contribution of that trait to a species’ sensitivity to a stressor. A weight of 0 means that a taxa possessing that trait is insensitive to a stressor, whereas a weight of 1 means that that trait renders a taxa highly sensitive to the effect of a stressor. In instances when a taxa had more than one trait (e.g. crawler and swimmer), the trait with the maximal sensitivity weight was retained. The sensitivity assessment was informed by expert knowledge and bibliographic research. Trait-matching rules and weights for each trait-stressor combination are available in Table 6.10. The relative sensitivity of each taxon was evaluated as the product of the weight given to all traits considered to assess the sensitivity of a given stressor. This process yielded a relative sensitivity assessment ranging between 0 and 1. The code to evaluate the relative sensitivity of taxa on the Scotian Shelf is available at https://github.com/Ecosystem-Assessments/Species_direct_sensitivity.

Table 6.9: List of traits used to evaluate the relative sensitivity of taxa to the effects of environmental drivers. Adapted from Beauchesne (2020).

Traits	Categories	Description
Body composition	Biogenic silica	Organism with skeleton formed by microscopic particles of silica
	Bone	Organism with solid skeleton made of bones
	Cartilaginous	Organism with solid skeleton made of cartilage
	Chitinous	Organism with solid skeleton made of chitin
	Non-calcifying	Organism devoid of calcifying skeleton
	Soft-bodied, aragonite	Organism with aragonite in soft tissues
	Soft-bodied, calcite	Organism with calcite in soft tissues
	Soft-bodied, calcium phosphate	Organism with calcium phosphate in soft tissues
	Soft-bodied, calcium sulfate	Organism with calcium sulfate in soft tissues
	Skeleton, aragonite	Organism with solid skeleton made of or containing aragonite
	Skeleton, calcite	Organism with solid skeleton made of or containing calcite
	Skeleton, calcium phosphate	Organism with solid skeleton made of or containing calcium phosphate
	Skeleton, high-magnesium calcite	Organism with solid skeleton made of or containing high-magnesium calcite
	Skeleton, phosphatic	Organism with solid skeleton made of or containing phosphates
	Skeleton, gorgonin	Organism with solid skeleton made of or containing gorgonin
Environment	Bathydemersal	Living and/or feeding on or near the bottom, below 200 m
	Bathypelagic	Occurring mainly in open water below 200 m, not feeding on benthic organisms
	Benthic	Living and feeding on the bottom
	Benthopelagic	Living and/or feeding on or near the bottom, as well as in midwater, between 0 and 200 m
	Demersal	Living and/or feeding on or near the bottom, between 0 and 200 m
	Pelagic	Occurring mainly on the surface or in the water column between 0 and 200 m, not feeding on benthic organisms
	Coastal	Occurring mainly along the coast, specific to seabirds
	Terrestrial	Occurring mainly on land, specific to seabirds
Feeding type	Deposit feeder	Organism that lives on or in the sediments and consumes organic material on the sea floor
	Selective filter feeder	Organism that actively and selectively filters waters to consume plankton or nutrients suspended in the water
	Grazer	Organism that feeds on plants
	Parasite	Organism that lives in or on another organism, benefiting at the other organism’s expense
	Plankton	Organism feeding exclusively on plankton
	Predator	Organism that actively hunts prey
	Scavenger	Organism that feeds on dead plant or animal material, or refuse
	Suspension feeder	Organism that captures and consumes particules suspended in the water, such as plankton, bacteria, detritus and particulate organic matter
	Xylophagous	Organism feeding on or boring into wood
Fisheries landings	Targeted species	Applicable to fisheries only. Taxa targeted by fishing activities.
	Bycatch	Applicable to fisheries only. Taxa bycatch by fishing activities.
	Others	Applicable to fisheries only. Taxa not targeted or caught by fishing activities
Mobility	Burrower	Organism that lives in a burrow dug in the sediments
	Crawler	Organism that crawls slowly on the bottom
	Mobile	Free-ranging organism
	Sessile	Immobile or fixed organism
	Swimmer	Organism with limited swimming ability
	Flying	Flying organisms
Phylum	Annelida	-
	Arthropoda	-
	Brachiopoda	-
	Bryozoa	-
	Chlorophyta	-
	Chordata	-
	Cnidaria	-
	Ctenophora	-
	Echinodermata	-
	Echiura	-
	Mollusca	-
	Ochrophyta	-
	Porifera	-
	Rhodophyta	-
	Sipuncula	-
	Tracheophyta	-
Size	0-100 cm	Body length between 0 and 100 cm
	100-200 cm	Body length between 100 and 200 cm
	200-300 cm	Body length between 200 and 300 cm
	300+ cm	Body length greater than 300 cm

Table 6.10: Relative species-specific sensitivity weights for each environmental driver. Adapted from Beauchesne (2020).

Drivers	Traits	Categories	Sensitivity
Ocean acidification	Environment	Bathydemersal	1.00
		Bathypelagic	0.00
		Benthic	1.00
		Benthopelagic	0.50
		Demersal	1.00
		Pelagic	0.00
		Coastal	0.00
		Terrestrial	0.00
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
	Body composition	Biogenic silica	0.00
		Bone	0.00
		Cartilaginous	0.00
		Chitinous	0.00
		Non-calcifying	0.00
		Soft-bodied, aragonite	0.00
		Soft-bodied, calcite	0.00
		Soft-bodied, calcium phosphate	0.00
		Soft-bodied, calcium sulfate	0.00
		Skeleton, aragonite	0.90
		Skeleton, calcite	0.80
		Skeleton, calcium phosphate	0.80
		Skeleton, high-magnesium calcite	1.00
		Skeleton, phosphatic	0.00
		Skeleton, gorgonin	0.80
	Phylum	Annelida	0.00
		Arthropoda	0.50
		Brachiopoda	0.50
		Bryozoa	1.00
		Chlorophyta	0.00
		Chordata	0.00
		Cnidaria	1.00
		Ctenophora	0.00
		Echinodermata	1.00
		Echiura	0.00
		Mollusca	1.00
		Ochrophyta	0.00
		Porifera	0.00
		Rhodophyta	0.00
		Sipuncula	0.00
		Tracheophyta	0.00
Bottom-water temperature anomalies	Environment	Bathydemersal	1.00
		Bathypelagic	0.00
		Benthic	1.00
		Benthopelagic	0.50
		Demersal	1.00
		Pelagic	0.00
		Coastal	0.00
		Terrestrial	0.00
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
	Feeding type	Deposit feeder	1.00
		Selective filter feeder	0.00
		Grazer	1.00
		Parasite	0.00
		Plankton	0.50
		Predator	0.50
		Scavenger	0.50
		Suspension feeder	1.00
		Xylophagous	0.50
Surface-water temperature anomalies	Environment	Bathydemersal	0.00
		Bathypelagic	0.00
		Benthic	0.00
		Benthopelagic	0.50
		Demersal	0.00
		Pelagic	1.00
		Coastal	0.50
		Terrestrial	0.25
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
	Feeding type	Deposit feeder	1.00
		Selective filter feeder	0.00
		Grazer	1.00
		Parasite	0.00
		Plankton	0.50
		Predator	0.50
		Scavenger	0.50
		Suspension feeder	1.00
		Xylophagous	0.50
Hypoxia	Environment	Bathydemersal	1.00
		Bathypelagic	0.00
		Benthic	1.00
		Benthopelagic	0.50
		Demersal	1.00
		Pelagic	0.00
		Coastal	0.00
		Terrestrial	0.00
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
	Feeding type	Deposit feeder	1.00
		Selective filter feeder	0.00
		Grazer	1.00
		Parasite	0.00
		Plankton	0.50
		Predator	0.50
		Scavenger	0.50
		Suspension feeder	1.00
		Xylophagous	0.50
Coastal development	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.50
		Sessile	1.00
		Swimmer	0.75
		Flying	0.25
	Size	0-100 cm	0.25
		100-200 cm	0.50
		200-300 cm	0.75
		300+ cm	1.00
Direct human impact	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.50
		Sessile	1.00
		Swimmer	0.75
		Flying	0.25
	Size	0-100 cm	0.25
		100-200 cm	0.50
		200-300 cm	0.75
		300+ cm	1.00
Inorganic pollution	Feeding type	Deposit feeder	0.75
		Selective filter feeder	0.00
		Grazer	0.00
		Parasite	0.00
		Plankton	0.00
		Predator	0.00
		Scavenger	0.50
		Suspension feeder	1.00
		Xylophagous	0.00
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
Nutrient input	Feeding type	Deposit feeder	0.75
		Selective filter feeder	0.00
		Grazer	0.00
		Parasite	0.00
		Plankton	0.00
		Predator	0.00
		Scavenger	0.50
		Suspension feeder	1.00
		Xylophagous	0.00
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
Organic pollution	Feeding type	Deposit feeder	0.75
		Selective filter feeder	0.00
		Grazer	0.00
		Parasite	0.00
		Plankton	0.00
		Predator	0.00
		Scavenger	0.50
		Suspension feeder	1.00
		Xylophagous	0.00
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
Demersal destructive fisheries	Fisheries landings	Targeted species	1.00
		Bycatch	0.75
		Others	0.25
For Others only	Environment	Bathydemersal	1.00
		Bathypelagic	0.00
		Benthic	1.00
		Benthopelagic	0.50
		Demersal	1.00
		Pelagic	0.00
		Coastal	0.00
		Terrestrial	0.00
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
Demersal non-destructive high-bycatch fisheries	Fisheries landings	Targeted species	1.00
		Bycatch	0.75
		Others	0.00
Demersal non-destructive low-bycatch fisheries	Fisheries landings	Targeted species	1.00
		Bycatch	0.75
		Others	0.00
Pelagic high-bycatch fisheries	Fisheries landings	Targeted species	1.00
		Bycatch	0.75
		Others	0.00
Pelagic low-bycatch fisheries	Fisheries landings	Targeted species	1.00
		Bycatch	0.75
		Others	0.00
Marine pollution	Feeding type	Deposit feeder	0.75
		Selective filter feeder	0.00
		Grazer	0.00
		Parasite	0.00
		Plankton	0.00
		Predator	0.00
		Scavenger	0.50
		Suspension feeder	1.00
		Xylophagous	0.00
	Mobility	Burrower	0.75
		Crawler	0.75
		Mobile	0.25
		Sessile	1.00
		Swimmer	0.75
		Flying	0.00
Shipping	Environment	Bathydemersal	0.00
		Bathypelagic	0.00
		Benthic	0.00
		Benthopelagic	0.50
		Demersal	0.00
		Pelagic	1.00
		Coastal	0.50
		Terrestrial	0.25
	Size	0-100 cm	0.00
		100-200 cm	0.50
		200-300 cm	0.75
		300+ cm	1.00

6.4 Metaweb

We predicted the metaweb of the Scotian Shelf, i.e. the network of biotic interactions, using a recommender approach (Beauchesne et al., 2016b). Here, we provide a brief overview of the approach, but refer to Beauchesne et al. (2016b) for more details. The approach is built on a series of logical steps that predict a candidate resource list for each taxon considered based on empirically observed interactions and on the similarity between consumers and resources. It predicts pairwise interactions given taxonomic and dietary similarity between taxa using the K-nearest neighbor algorithm [KNN; Murphy (2012)].

Predictions are informed by a catalogue of empirical interactions from all over the world (Beauchesne et al., 2016b). The interactions catalogue was built using food web data (Brose et al., 2005; Kortsch et al., 2015; University of Canberra, 2016), predator-prey interactions (Barnes et al., 2008) and pairwise interactions from the GloBI database (Poelen et al., 2014; Poelen et al., 2019). We limited the compendium to taxa found in marine and coastal ecosystems. The original interaction catalogue was built for a cumulative effects assessment of global changes on the food web of the St. Lawrence System (Beauchesne, 2020). The interaction catalogue was updated for the assessment of the Scotian Shelf using the GloBI database. We used the taxonomic families from the list of species considered for the assessment to identify all member species of those families for which interactions were available on the GloBI dataset using the rglobi package (Poelen et al., 2019). We then combined the original interaction catalogue with the interactions extracted from GloBI to form a new catalogue of biotic interactions. This process yielded a total of 183 625 pairwise interactions between 14 870 taxa to inform interaction predictions.

Taxa similarity was evaluated from taxonomic classification and sets of consumers or resources using the Tanimoto similarity measure (\(T\)), which compares two vectors \(x\) and \(y\) with \(n = \left\vert x \right\vert = \left\vert y \right\vert\) elements, and is defined as the size of the intersection of two sets divided by their union:

\[T_{x,y} = \frac{\left \vert x \cap y \right \vert}{\left \vert x \cup y \right \vert}\]

where \(\cap\) is the intersect and \(\cup\) the union of the vectors. Adding a weighting scheme, we can measure the similarity using two different sets of vectors \(\{x,y\}\) and \(\{u,v\}\):

\[T_{x,y,u,v,w_t} = w_t T_{x,y} + (1-w_t) T_{u,v}\]

A weight of 0.5 was given to taxonomy and consumers or resources to consider them simultaneously (Desjardins-Proulx et al., 2016). The taxonomy of all taxa considered was accessed and validated from WoRMS (WoRMS Editorial Board, 2017) using the taxize package (Chamberlain et al., 2019; Chamberlain and Szöcs, 2013). We included the main phytoplankton and zooplankton taxa found in the St. Lawrence System to predict the metaweb (Morissette et al., 2003; Savenkoff et al., 2004; Savenkoff, 2012); we then grouped predictions under phytoplankton or zooplankton.

This yielded a total of 772 taxa (\(S\)) for which we predicted a metaweb structured by 18 021 links (\(L\)), a link density (\(L_{moy} = L/S\)) of 23.34 and a connectance (\(C = L/S^2\)) of 0.03, which is within range of most reported food webs (Dunne et al., 2002).

6.5 Trophic sensitivity

The approach used to evaluate trophic sensitivity comes from Beauchesne (2020) and Beauchesne et al. (2021). Here, we provide a brief description of the approach; a full description is available in those works. Beauchesne et al. (2021) developed a theoretical framework to evaluate a species’ sensitivity to multiple stressors as a function of its trophic position, which we refer to as a species’ trophic sensitivity. Beauchesne et al. (2021) simulated stressors on the most empirically abundant 3-species motifs — i.e. tri-trophic food chain, omnivory, exploitative competition and apparent competition – using Lotka-Volterra models (Gellner and McCann, 2016). Negative effects of environmental pressures were simulated using combinations of equilibria equation parameters for resource growth, mortality, attack and conversion rates; these represent effects to ecological processes and their combination are the pathways through which environmental drivers affect species directly and indirectly through their ecological interactions. All possible pathways of effects – i.e. all combinations of ecological processes – were simulated for each 3-species motif considered. The difference between equilibrium population abundance before and after simulations was used as a proxy of a species’ trophic sensitivity; This represents the net effects of stressors on species and integrates all direct effects to a species, and all indirect effects propagating through its interactions with other species (Abrams et al., 1996; Beauchesne et al., 2021).

To evaluate a species’ trophic sensitivity empirically, Beauchesne et al. (2020) used simulation results from Beauchesne et al. (2021) as heuristics to infer a species’ anticipated trophic sensitivity to the effects of environmental drivers as a function of its position in the ecological network of the Scotian Shelf. Beauchesne et al. (2020) used the absolute values of simulated trophic sensitivities and considered that any effect to a species’ population dynamics, whether negative or positive, can propagate and disturb the dynamics of an ecological community. Pathways of effects were also simplified to consider only effects to population density for ease of evaluation empirically, as effects on ecological processes such as attack and conversion rates are inherently challenging to evaluate empirically, especially for many species simultaneously. We thus averaged all trophic sensitivities to all pathways of effects for each position in the 3-species motifs considred. This resulted in 124 possible trophic sensitivity values based on the position and types of ecological interactions in which a species is involved. All trophic sensitivities were scaled between 0 and 1 for the analyses. Refer to Beauchesne (2020) and Beauchesne et al. (2021) for more details, and to the description of the network-scale assessment model for a description of how trophic sensitivity values are used in the assessment of the cumulative effects of global changes on the food webs of the Scotian Shelf.