Environmental heterogeneity mediates scale-dependent declines in kelp diversity on intertidal rocky shores
Environmental heterogeneity mediates scale- dependent declines in kelp diversity on intertidal rocky shores
Samuel StarkoID 0 1
Lauren A. Bailey 1
Elandra Creviston 1
Katelyn A. James 1
Alison Warren 1
Megan K. BrophyID 1
Andreea Danasel 1
Megan P. Fass 0 1
James A. Townsend 1
Christopher J. NeufeldID 1
0 Department of Botany, University of British Columbia , Vancouver, British Columbia , Canada , 2 Bamfield Marine Sciences Centre , Bamfield, British Columbia , Canada , 3 Department of Biology, University of Victoria , Victoria, British Columbia , Canada , 4 Department of Biology, University of British Columbia Okanagan , Kelowna, British Columbia , Canada , 5 Department of Biological Sciences, University of Calgary , Calgary, Alberta , Canada , 6 Department of Biology, University of British Columbia , Vancouver, British Columbia , Canada
1 Editor: Judi Hewitt, University of Waikato , NEW ZEALAND
Data Availability Statement: All relevant data are
within the manuscript and its Supporting
Funding: The authors received no specific grants
for this work, but received in-kind support from
Bamfield Marine Sciences Centre.
Competing interests: The authors have declared
that no competing interests exist.
Biodiversity loss is driven by interacting factors operating at different spatial scales. Yet,
there remains uncertainty as to how fine-scale environmental conditions mediate biological
responses to broad-scale stressors. We surveyed intertidal rocky shore kelp beds situated
across a local gradient of wave action and evaluated changes in kelp diversity and
abundance after more than two decades of broad scale stressors, most notably the 2013?2016
heat wave. Across all sites, species were less abundant on average in 2017 and 2018 than
during 1993?1995 but changes in kelp diversity were dependent on wave exposure, with
wave exposed habitats remaining stable and wave sheltered habitats experiencing near
complete losses of kelp diversity. In this way, wave exposed sites have acted as refugia,
maintaining regional kelp diversity despite widespread local declines. Fucoids, seagrasses
and two stress-tolerant kelp species (Saccharina sessilis, Egregia menziesii) did not decline
as observed in other kelps, and the invasive species Sargassum muticum increased
significantly at wave sheltered sites. Long-term monitoring data from a centrally-located moderate
site suggest that kelp communities were negatively impacted by the recent heatwave which
may have driven observed losses throughout the region. Wave-sheltered shores, which saw
the largest declines, are a very common habitat type in the Northeast Pacific and may be
especially sensitive to losses in kelp diversity and abundance, with potential consequences
for coastal productivity. Our findings highlight the importance of fine-scale environmental
heterogeneity in mediating biological responses and demonstrate how incorporating
differences between habitat patches can be essential to capturing scale-dependent biodiversity
loss across the landscape.
Ongoing biodiversity loss is expected to reduce ecosystem functioning and services [
uncertainty remains about the spatial scale at which to investigate the environmental drivers of
such loss [
]. Global stressors can interact with local factors to exacerbate or ameliorate
community responses to ongoing global change [
]. Yet, fine-scale microclimatic
differences between sites are often ignored by both climate envelop models?which predict
systematic shifts in the latitudinal ranges of species [
]?, and in meta-analyses of local diversity
change?which group plots only by habitat-type (e.g. forest, marsh, grassland) or by region [
]. These common approaches, although insightful, may miss functionally important trends
in community diversity change or local abundance loss if the stresses associated with a habitat
patch depend more on local conditions than on regional patterns [
], or if even the most
consistent declines occur at only a subset of sites within each habitat type. Understanding how to
detect and predict functionally-relevant biodiversity changes will therefore depend on
determining the relative importance of both broad-scale and fine-scale stressors in driving
community shifts through time. While much work has focussed on how broad-scale stressors are
driving the biological responses of communities [
], fewer studies have examined the
role that local, fine-scale conditions play in magnifying or ameliorating them .
The rocky intertidal zone is predicted to be particularly sensitive to ongoing changes in
] because intertidal organisms live near their physiological limits [
] and are sensitive
to air temperatures, which tend to be more variable and extreme than seawater temperatures
]. Local environmental gradients also have profound effects on intertidal systems.
Environmental heterogeneity in the form of wave action plays a significant role in structuring
intertidal communities [
]. Water movement from waves can eliminate nutrient-depleted or
oxygen-rich boundary layers that are associated with low-flow environments, thereby
increasing productivity [
]. Furthermore, wave splash can ameliorate the stressors associated with
aerial exposure, such as desiccation and thermal stress [
]. Thermal profiles have suggested
that wave exposed intertidal sites experience reduced thermal stress and emersion times
compared to sheltered sites [
], suggesting that perhaps they are more resilient to rising air
temperatures. Given the importance of wave action to the physiology and ecology of organisms
that live along rocky shorelines, exposure to waves is likely to mediate the biological responses
of intertidal organisms to ongoing changes in environmental stressors. However, the scarcity
of appropriate baseline community data has made this hypothesis difficult to test in the field.
Here we investigate the influence of a local wave exposure gradient on temporal changes in
intertidal kelp bed habitats in Barkley Sound, British Columbia, Canada following 22 years of
broad-scale stressors and extreme temperature events (Fig 1) [
]. Rocky shore kelp beds
are composed of a wide variety of marine flora (e.g. seaweeds and seagrasses) and fauna (e.g.
mussels, barnacles, echinoderms) many of which compete for space on the shore [
these systems, kelps (hereafter referring only to Laminariales) act as foundation species in both
intertidal and subtidal regions , driving ecosystem productivity through rapid growth and
formation of habitat for many ecologically important species [
]. Kelps are sensitive to high
], however, and as such are expected to respond negatively to climate
change and climate change-amplified heat wave events [
]. This sensitivity to high
temperatures can be made worse by the tendency for heat waves to be associated with nitrogen
poor waters [
] that can magnify the impacts of increased temperatures [
]. Increases in
marine heat wave prevalence and intensity have begun to cause negative impacts on kelp
forests near their geographical range limits [
]. However, interactions between global,
regional and local processes have led to complex responses of kelp communities, with large
variability in the magnitude and direction of change [
]. Studies of local-scale
2 / 25
Fig 1. Temperature anomalies between 1990 and 2018. Data are presented for (A) air temperature and (B) sea surface
temperature (SST) in relation to the timing of surveys. Data were taken from lighthouses near the opening of Barkley Sound and
are calculated with respect to 33-year historical averages (dating to 1985). Also shown (as black lines) are one-year moving
averages of temperature anomaly. Note that between 2014 and 2018, air temperature anomalies were consistently positive and
reached an unprecedented extreme of more than 3?C in 2015. SST anomalies were also consistently positive between late 2014
3 / 25
temporal change in the abundance of kelps and other large brown algae are increasingly
] and have collectively demonstrated that local conditions can interact with
global stressors to drive variation in ecosystem responses [
]. However, these studies
have focused on a small number of species, generally in the subtidal zone, and have not
examined temporal changes in the diversity or composition of entire kelp assemblages. Moreover,
there is still broad uncertainty as to how natural variation in site-level environmental
conditions will influence the responses of kelp-dominated ecosystems to increases in the prevalence
of broad-scale stressors.
To assess temporal changes in the diversity and abundance of kelps on intertidal rocky
shores, we resurveyed sites (n = 49) in 2017 and 2018 that had previously been surveyed by
Druehl & Elliot between 1993 and 1995 [
]. Sites occurred broadly throughout the region,
and were situated across a range of wave exposures, slopes, aspects, and types of rocky
substrates. We also analysed other long-term monitoring data for one centrally located site
(Wizard Islet) to better assess the timing of any widespread changes in kelp bed composition,
diversity or abundance. We found substantial changes in the diversity of Barkley Sound kelp
communities and widespread declines in the abundance of many kelp species. We discuss
potential drivers and consequences of changes in kelp communities as they relate to gradual
], the recent marine heatwave [
] and possible changes in trophic
Barkley Sound, on the southwest coast of Vancouver Island, Canada, is a nearly 30 km wide
inlet containing hundreds of islands. As such, it provides a wide range of local microhabitats.
Both wave-sheltered and wave-exposed sites are located throughout the area both near and far
from the opening to the sound. Sites were accessed by boat. Historical survey data spans 1993?
1995 with most (n = 46) sites sampled twice in 1993 (n = 19) or 1994 (n = 27) and 1995.
However, three sites were only sampled in either 1993 (n = 2) or 1994 (n = 1) and not in 1995 (S1
Table). Sites were located using GPS coordinates, photographs, and descriptions recorded in
the original surveys. In particular, most sites were located using a photo that was often
annotated with the exact location of the transect. A thorough description of most sites was also
given in the original report and allowed for location of some sites that did not have
photographs. Sites were only resurveyed if they could be definitively located in at least one of these
two ways using distinct geographic landmarks.
Surveys were conducted following the methods of the original surveyors and were mostly
restricted to species in the order Laminariales (i.e. kelps). Three non-kelp species that are
expected to be more resilient to heat stress than kelps [
] were also included in the surveys
but not in analyses of kelp diversity: Phyllospadix spp. (Alismatales), Fucus distichus (Fucales)
and Sargassum muticum (Fucales). Surveys (performed between June 20 and Sept 9, 2017 and
between July 10 and Aug 17, 2018) were conducted on 20?50 m stretches of coastline and
included the entire intertidal region, from Lower Low Water Large Tide (LLWLT) to the
upper limit of marine organisms. All sites were surveyed in 2017 and all but two (n = 47) were
surveyed in 2018. A subset of these sites (N = 17) were surveyed in both June and September of
2017 and no differences in kelp community composition were detected during this time.
Survey sites were uniform lengths of shoreline and included the area between the high tide line
and LLWLT (approximately 3 m vertical distance). Presence and absence of all kelp species
4 / 25
were determined for the entire survey area by carefully identifying all kelp species present by
morphology. Kelps are large, seasonally persistent and are easy to distinguish based on
conspicuous morphological features [
]. Thus, both our surveys and those done by the original
surveyors were likely to result in unbiased, reproducible data. In order to quantify abundance,
the intertidal was blocked into four zones: high intertidal (approx. > 2.5 m), mid intertidal
(approx. 1.2?2.5 m), low intertidal (approx. 0.2?1.2 m) and shallow subtidal (0?0.2 m).
Abundance of each species, in each zone, was then quantified based on visual estimation of
percentage cover categories: absent (0%), rare ( 5%), common (6?20%) and (21?100%). A species?
assigned abundance was then taken from the zone of its greatest abundance.
Wave exposure quantification
Quantifying wave exposure is a known challenge to intertidal biologists, as local topography
can influence water velocities in ways that many geographical indices fail to capture [
characterization of sites, we used the site-specific wave-exposure categories provided by the
original Barkley Sound surveyors [
] that were modelled after the categories of Topinka et al.
]. However, we grouped sites that they had ranked as ?Sheltered? and ?Moderately
Sheltered? (into ?Sheltered? grouping) since few sites were assigned to the former category. Sites
were categorized by the original surveyors based on direct, qualitative observations of water
] and the presence of indicator species. In order to test the validity of these
categories, we measured two known proxies of wave exposure at subsets of sites. First, we used a
cartographical method previously developed [
] and tested [
] in Barkley Sound. In brief, this
method is a continuous index derived from the angle of unimpeded exposure to the
predominant direction of offshore swell (southwest). It is therefore only effective for SW facing sites
]. We used all of our SW (180-270o) facing sites (N = 26) to ground-truth these wave
exposure categories (Kruskal-Wallis test: X2 = 16.451, p < 0.001; S1 Fig) and showed that sites
categorized as ?Sheltered? had significantly lower wave exposure index measures than ?Moderate?
and ?Exposed? (Dunn?s test: p < 0.001 for both). There was a near significant trend suggesting
a difference between ?Moderate? and ?Exposed? sites (Dunn?s test: p = 0.0728). Secondly, we
measured the upper limit of barnacles at a majority of sites (n = 47) in each wave exposure
category. This was accomplished by measuring the distance between the top of the barnacle band
and the water using either a stadia rod and sight level, or?if a surface was vertical?a transect
tape. Tidal predictions from the closest tide station (either Bamfield Inlet or Effingham Island)
were used to calculate the height of the water relative to LLWLT for each site at the time of the
survey. There was a significant effect of wave exposure category on the upper limit of
barnacles, with all means differing significantly (Kruskal-Wallis test: X2 = 21.195, df = 2, p < 0.001,
S2 Fig). Together, these additional measures of wave exposure suggest that our categories were
In order to determine how any wave exposure-specific responses might scale up across the
landscape, we examined the distribution of rocky habitats of different wave exposures across
the North American Pacific coast using a comprehensive georeferenced linear shoreline
dataset called ShoreZone [
]. ShoreZone data span from Oregon to Alaska and are based on
expert classification of shoreline units using low-elevation aerial imagery obtained from
fixedwing aircraft or helicopter and relevant geographic features. During segment classification,
each shoreline unit is assigned a substrate class from high-resolution imagery, and assigned a
wave-exposure class using a combination of fetch calculations and geographic and biotic
features. For the current study, predominantly rocky shoreline was identified by selecting all
5 / 25
shoreline units from this dataset which contained at least 25% rocky substrate (ShoreZone
coastal classes 1?20). Regional totals of the extent of shoreline containing only bedrock
(ShoreZone coastal classes 1?5) were 44?78% shorter than regional totals from mostly rocky
shoreline but produced the same patterns of relative habitat types between regions. Because the
average shoreline unit in the ShoreZone dataset is between 300m and 500m long (much longer
than our 20?50 m surveys), the ShoreZone wave exposure classification is not able to resolve
small scale differences in exposure that fall within a single shoreline unit. Importantly,
ShoreZone produced similar categorizations as Druehl & Elliot when we grouped ?Very Protected?,
?Moderately Protected? and ?Protected? Shorezone categories (hereafter ?Sheltered?): 96% of
sites were within one wave exposure category of one another (and classifications at 67% of
sites agreed completely). Sites that differed between methods included a tidepool that was set
back from the shore and protected from incoming waves, three exposed headland sites in areas
that were otherwise largely sheltered from waves, and two sites that were located on the
wavesheltered side of islands that were near the mouth of the sound where overall wave exposure is
greater. Barring these few exceptions that arose largely due to differences in the scale at which
wave exposure was assessed, the overall concordance of the two independent approaches
suggests that scaling up to the broader region using the ShoreZone dataset is appropriate.
Long term data from Wizard Islet
In order to better evaluate the timing of any changes in kelp communities, we analyzed
long-term monitoring data from a centrally-located site (Wizard Islet; 48.857983N,
125.160793W). This long-term dataset was collected by researchers at Bamfield Marine
Sciences Centre and includes seven time-points between 1997 and 2017 (publically
available at: https://doi.org/10.5683/SP2/C8G480). These data were collected by using
randomly placed vertical transects (n 8 per year) and estimating percent cover of species in
25 cm x 25 cm quadrats at 0.5 metre tidal height increments. We summed the cover of
kelps present at this site (Egregia menziesii, Alaria marginata, Saccharina sessilis, and
Laminaria setchellii) and compared total kelp cover through time at tidal heights of 0.5,
1.0 and 1.5m above LLWLT. Wizard Islet was not included in the original surveys of
Druehl and Elliot and so was not assigned by them to a wave exposure category. While
ShoreZone classifies all of Wizard Islet as ?sheltered?, we suspect that the site is actually
moderately exposed given the authors? visual observations of the site, and based on
previously reported dynamometer readings [
]. While species that are indicative of wave
exposed sites (e.g. Lessoniopsis littoralis, Pelvetiopsis limitata, Postelsia palmaeformis) are
not present here, L. setchellii and S .sessilis are present, which are not generally found at
sheltered sites but can be common at moderately exposed sites.
Analysis of air and water temperature data
We analyzed air temperature and sea surface temperature (SST) from nearby lighthouses to
assess whether temperature increases from gradual heating on the west coast of British
] or persistent anomalies such as the 2013?2016 heatwave [
] have influenced
thermal conditions in our study region. Air temperature data were taken from Cape Beale
Lighthouse at the southern opening of Barkley Sound, and SST data were taken from
Amphitrite Lighthouse at the northern opening of Barkley Sound. Air temperature data are not
available from Amphitrite Lighthouse and SST data from Cape Beale have several gaps over the
period of interest.
6 / 25
All analyses were performed in R 3.4.1 [
]. To assess whether wave exposure and survey year
significantly influenced the richness and abundance of kelps at each site, we used
KruskalWallis rank sum tests on site-level proportional responses (ratio of historic to modern values)
with wave exposure (fixed factor, three levels) as an explanatory variable. Means were then
compared using Dunn?s tests correcting for multiple comparisons. Comparisons were made to
averages of historical survey data when sites were sampled in more than one year during the
1993?1995 (n = 46) or 2017?2018 (n = 47) surveys. To rule out spatial effects, we tested for
spatial correlation of proportional richness responses using Moran?s I in the R package ?ape? [
Average abundance was calculated for each site by averaging the species-specific abundance
ranks of all species present. Thus, a lower value indicates a higher proportion of rare species.
Rarefaction and regional species pool extrapolation were performed using the ?vegan? package
] in R and were performed for each year, combining 1993 and 1994 surveys into a single
time-point. In order to determine whether individual species have changed in abundance or
whether their distribution has changed across the wave exposure gradient, we used
proportional odds models of abundance (ordered factor, 4 levels) versus year (fixed factor, 4 levels),
wave exposure (fixed factor, 3 levels) and their interaction in the package ?VGAM? [
] for the
sites that were surveyed in all four years (n = 43). These models were fit to the 9 most common
kelp species (Alaria marginata, Costaria costata, Ecklonia arborea, Egregia menziesii,
Laminaria setchellii, Lessoniopsis littoralis, Macrocystis pyrifera, Nereocystis luetkeana and
Saccharina sessilis) as well as Phyllospadix spp., Fucus distichus and Sargassum muticum. Trends in
temperature anomaly from nearby lighthouses (see previous section) were assessed using a
simple moving average with a 12-month window in the R package ?TTR? [
]. Changes in
kelp cover on Wizard Islet were assessed in two ways but in both cases statistics were
performed separately for data from each tidal height (0.5m, 1.0m and 1.5m above LLWLT). First,
to determine whether there were gradual changes in kelp cover through time, a linear
regression was fit between annual averages of total kelp cover and year (as a continuous variable). If
no linear relationship was found, then data from the most recent survey year (2017) were
compared to years prior to the 2013?2016 heatwave to determine whether recent temperature
] have influenced kelp cover. This was tested using an ANOVA and planned
contrast between 2017 and the grand mean of all years sampled prior to the 2013?2016
Permission for sampling in the Pacific Rim National Park was granted by Parks Canada.
Permission for sampling on Huu-uy-aht First Nations (HFN) territory was given by the HFN. No
permission was required for sampling sites outside of these areas.
Declines in kelp diversity and abundance were greatest at wave sheltered
Across the study system, we found widespread declines in kelp species richness and abundance
mediated by local variation in wave action (Fig 2). Kelp species richness has not changed
significantly at wave-exposed sites (Paired t-test: t = 0.78779, df = 9, p = 0.4511), while kelp
communities at wave sheltered sites have been reduced to between zero and three species,
regardless of their historical diversity (Fig 3A), leading to a significant effect of wave exposure
on proportional change in richness (Kruskal-Wallis test: X2 = 19.561, df = 2, p < 0.001, Fig 3A,
7 / 25
Fig 2. Study region. Study sites (n = 49) are coded for magnitude of change in kelp species richness (colour) and relative exposure to waves (shape). Due to the
close proximity of some sites, some symbols have been moved slightly to avoid obscuring overlapping symbols. There was no effect of the spatial distribution of
sites on proportional kelp richness change (Moran?s I: I = 0.0303, p = 0.49676). Stars in the inset indicate the location of Cape Beale (South) and Amphitrite
(North) lighthouses. The arrow labelled ?W? indicates the location of Wizard Islet long-term monitoring site.
S3 Fig). At all wave exposures, average abundance also declined with significantly larger
declines having occurred at sheltered sites than at moderate or sheltered sites (Fig 3B, S3 Fig;
Kruskal-Wallis test: X2 = 7.6663, p = 0.0216).
Habitat-dependent declines in kelp diversity were further demonstrated by changes in the
shape of species accumulation curves in 2017 and 2018 relative to 1993?1995 (Fig 4). While
rarefaction curves of the entire region are similar between years across all sites, sheltered sites
had rarefaction curves with lower asymptotes and shallower slopes in 2017?2018 than in
1993?1995 (Fig 3), indicating that sheltered communities consisted of fewer total species
during the resurveys. While at exposed sites, the total number of species detected in 2017 was
slightly larger than in 1995, the total species pool at sheltered sites declined from 11 species to
8 / 25
Fig 3. Diversity and abundance of kelp beds during 1993?1995 and 2017?2018. (A) Richness and (B) average abundance of species at each site (absent = 0,
rare = 1, common = 2, abundant = 3) for historical (average of 1993?1995) and modern (2017) observations. There is a significant effect of wave exposure on
site-wise proportional changes in richness (ANOVA: F47,1 = 17.27, P = 0.000136) and abundance (ANOVA: F43,1 = 4.396, P = 0.0420).
3 (Fig 4; Macrocystis pyrifera, Egregia menziesii, Alaria marginata) with one of those species
(A. marginata) only present at one site. In 2018, a total of five species were detected across all
sheltered sites with only one observation of each species that was not detected at sheltered sites
in 2017 (Costaria costata and Saccharina latissima).
Despite these widespread declines in the number of species found at sheltered and
moderate sites between the 1993?1995 and 2017?2018 surveys, species richness across the
region has not changed (Fig 4A inset, S4 Fig). Only two species were detected in 1993?
1995 that were not detected during our resurveys: Laminaria ephemera and Agarum
fimbriatum. These species were inconsistently found at a small number of sites during 1993?
1995 and have been observed elsewhere in Barkley Sound recently. Laminaria ephemera
was collected from Edward King Island in 2015 [
], near some of our sites, and was
found in the wrack at nearby Pachena Bay (N 48.790481, W -125.120173) in 2016 (Starko,
pers obs). Subtidal Neoagarum fimbriatum was observed reliably in Bamfield Inlet
between 2012 and 2018 [
] (Starko, pers obs) and was found intertidally by three of the
authors at a small island in the Broken Group Islands (N48.923916, W125.255136) in
August, 2018 (Starko, Brophy & Townsend pers obs). Thus, we find no evidence that the
diversity of kelps throughout all of Barkley Sound has changed, despite widespread local
losses of kelp species. Our results therefore show a disconnect in how kelp diversity has
changed across scales and habitats: richness of the regional kelp assemblage has remained
constant, while local richness and average abundance has declined markedly, with
diversity loss concentrated on wave-sheltered and moderate shores.
9 / 25
Fig 4. Rarefaction curves (10,000 iterations) for modern (2017) and historic (1995 for 46 sites, 1993 for 2 sites and 1994 for 1 site) surveys of sites in
Barkley Sound. Analyses are broken down by (A) all sites, and then for (B) wave exposed, (C) moderately exposed sites and (D) sheltered sites plotted
separately. The inset in (A) shows the results of a bootstrap extrapolation of total regional species richness of the Sound.
Most species differed in distribution between years
Species distributions differed between years for eight of the nine kelp species that were
statistically tested using proportional odds models (Table 1; Fig 5). There was a significant effect of
year on the abundance of all kelp species analyzed except for the high intertidal kelp,
Saccharina sessilis (Table 1), and there was an effect of wave exposure on all species analyzed except
Ecklonia arborea (Table 1). Although the abundance of Egregia menziesii differed between
years, this was driven largely by a high abundance in 1993?1994 that was not found in 1995.
When comparing data from 2017?2018 only to 1995, no significant effect of year on
abundance was detected (Proportional odds model: Coefficient = -0.0444, P = 0.0828) for E.
menziesii. There were particularly strong effects of exposure observed in Lessoniopsis littoralis and
Saccharina sessilis which are restricted to more wave exposed sites (Fig 5). There was a
significant interaction between wave exposure and year for Alaria marginata (P = 0.0155, Table 1,
10 / 25
Fig 5), with a possible interaction for Costaria costata. (P = 0.0806, Table 1, Fig 5). These
species did not decline in abundance at exposed sites but did substantially at sheltered and
moderate sites (Fig 5). There was no significant effect of year or interaction between year and wave
exposure for abundances of Fucus distichus or Phyllospadix spp. There was a near significant
increase in the abundance of Sargassum muticum between 1993?1995 and 2017?2018 across
11 / 25
Fig 5. Species-specific abundance versus year for the nine most abundant kelp species at our survey sites (n = 8). Data
are shown only for the 43 sites that were sampled for abundance in all four years. Data are ordinal (0 = absent, 1 = rare,
2 = common, 3 = abundant) and associated statistics are provided in Table 1.
12 / 25
Fig 6. Species-specific abundance versus year for three non-kelp species (fucoids and seagrass) investigated in this study (n = 3). Data are shown only for
the 44 sites that were sampled for abundance in all four years. Data are ordinal (0 = absent, 1 = rare, 2 = common, 3 = abundant) and associated statistics are
provided in Table 1.
all sites (Proportional odds model: Coefficient = 0.055, P = 0.0656, Table 1, Fig 6) and a
significant increase across sheltered sites only (Proportional odds model: Coefficient = 1.0441,
P = 0.0387). There was also a significant effect of wave exposure on abundance of Fucus
distichus and Sargassum muticum across all years. This effect was particularly strong in Sargassum
muticum that was restricted largely to sheltered areas but was found at one exposed site in
both 1995 and 2017?8 (but not 1994) and was restricted to a tidepool in 2017?2018.
Temperature data show clear evidence of the 2013?2016 heatwave
Public data from nearby lighthouses show that both sea surface (SST) and air temperatures in
Barkley Sound have reached abnormal highs between 1995 and 2017 with especially high
temperatures occurring during the 2013?2016 heatwave (Fig 1). Anomalies have lasted longer and
have been more extreme for air temperature than for SST (Fig 1). This heatwave is regarded as
the largest on record in the north Pacific [
]. It resulted from reduced surface cooling and
equatorward Eckman transport during a period of unusually high pressure in 2013?2014
(termed ?the Blob?) [
], followed by one of the most intense El Ni?os on record in 2015?
2016 . It then dissipated by September 2016 [
]. This led to positive temperature
anomalies that lasted several years (Fig 1). Although gradual heating has been documented on the
west coast of Vancouver Island over longer timescales [
], there was no clear gradual
temperature increase over the 22 year period of interest. However, both air temperatures and water
temperatures were higher between the 5-year period of 2013?2017 than between 1991?1995
(S5 Fig) indicating that climatic conditions differed between the two survey periods.
13 / 25
Fig 7. Long term trends on Wizard Islet. Total percent kelp cover through time for three tidal heights at a moderately exposed site on
Wizard Islet. Data points represent average kelp cover and error bars represent 95% confidence intervals. The dotted line indicates the
grand mean of kelp cover prior to the 2013?2016 heat wave. P-values indicate the significance of planned comparisons between 2017 data
(mean shown in red) and all data collected prior to the 2013?2016 heat wave. There was no significant correlation between kelp cover and
year, as a continuous variable (see text).
14 / 25
Declines on Wizard Islet occurred recently
In 2017, kelp cover on Wizard Islet was lower than any other year examined (Fig 7). On
average, total kelp cover was 59%, 56% and 17% at 0.5m, 1.0m and 1.5m tidal elevations
respectively, prior to the 2013?2016 heatwave. In 2017, these elevations had percent cover estimates
of 20%, 17% and 4%, respectively, indicating a more than two and a half-fold reduction in kelp
cover at this site. Mean cover in 2017 was significantly lower than the grand mean of previous
years at 0.5m and 1.0 m elevation, indicating that there was a significant drop in kelp cover at
this site between 2009 and 2017 (0.5 m: ANOVA with planned comparison: F = 7.111, df = 1,
p = 0.0093; 1 m: ANOVA with planned comparison: F = 14.247, df = 1, p = 0.0003). Although
not significant at 1.5m elevation, there was a trend towards reduced cover in 2017 relative to
the grand mean of previous years (ANOVA with planned comparison: 3.461, df = 1,
p = 0.0678). There was no significant effect of year (as a continuous variable) on total kelp
cover (0.5 m: Linear model: F = 4.688, df = 4, p = 0.0963; 1.0 m: Linear model: F = 0.6793,
df = 5, p = 0.447; 1.5 m: Linear model: F = 1.229, df = 5, p = 0.318), likely indicating no gradual,
linear decline in kelp cover through time. Thus, declines on Wizard Islet occurred between
2009 and 2017, consistent with the timing of positive temperature anomalies (Fig 1).
Timing and causes of declines
Between 1993?1995 and 2017?2018 kelp beds in Barkley Sound have changed substantially
with losses in kelp diversity at wave-sheltered and moderate sites. Most kelp species were
found at fewer sites in 2017 and 2018 than during 1993?1995 and kelp communities in 2017?
2018 tended to consist of fewer species that were less abundant on average. The spatial extent
and magnitude of species loss, as well as the multiannual life cycle of many investigated kelp
species suggest that these declines are a result of widespread responses to broad-scale stressors
that are occurring or have occurred throughout Barkley Sound. While temperatures in Barkley
Sound have gradually increased over the past century [
], this gradual change is not
detectable over the 22 year period between 1995 and 2017 (Fig 1). Instead, temperature data bear a
clear signal of the 2013?2016 marine heatwave, with anomalously warm temperatures detected
consistently between 2013 and 2018. Thermally tolerant fucoids (Sargassum muticum) and
seagrasses (Phyllospadix spp.) showed greater persistence than most kelp species, and have not
declined significantly between the 1990s and 2017. Climate-mediated shifts from
kelp-dominated to Sargassum-dominated communities have been documented elsewhere [
Therefore, these data are consistent with the hypothesis that changes in kelp communities
have resulted from increases in climate stress. Data from Wizard Islet also support this
hypothesis, demonstrating a substantial drop in kelp cover between 2009 and 2017, consistent with
the timing of temperature anomalies. Splashing of cool water at exposed sites could alleviate
air temperature stress during low tide, leading to the patterns that we show here  or local
mixing at sites with increased water motion could mediate these stresses by preventing pockets
of warm water from forming at small scales.
Out of the nine common kelp species that we investigated using proportional odds models,
only two species have not declined since 1993?1995: Saccharina sessilis and Egregia menziesii.
Both of these species are found higher in the intertidal zone than most other kelps, suggesting
resistance to desiccation and thermal stress at low tide [
]. Egregia menziesii has been
described as the kelp with the highest upper limit , although Postelsia palmaeformis,
Saccharina sessilis and a wave exposed ecotype of Alaria marginata (i.e. A. nana) may be found as
high or higher at wave exposed sites [
]. Both S. sessilis and E. menziesii possess complex
15 / 25
three-dimensional morphologies that could promote the retention of water and may therefore
improve survival during particularly stressful low tides. Although S. sessilis specializes in the
high intertidal zone, it is less resistant to warm water temperatures than some species that
experienced significant declines at moderate and sheltered shores (e.g. Alaria marginata and
Costaria costata) [
]. This suggests that air temperature is likely a stronger driver of the
observed patterns of kelp loss than SST. During the recent 2013?2014 ?Blob? and the 2015?
2016 El Ni?o, nitrogen levels were also abnormally low [
]. Nutrient availability may limit
productivity  and influence thermal tolerance of kelp species [
]. So, multiple stressors
could have interacted to result in the declines that we observed [
]. Given the multiple
stressors associated with the heatwave, it is possible that different species have declined in
abundance as a result of distinctive broad-scale drivers.
An alternative hypothesis, separate from the direct effects of recent temperature anomalies,
is that kelp declines were caused by changes in the trophic dynamics of intertidal kelp beds.
Sea stars have declined in abundance along the coast of British Columbia as a result of sea star
wasting disease [
], an epidemic that was possibly amplified by the 2013?2016 heatwave
]. This loss of sea stars has led to increases in sea urchin biomass and declines in kelp
abundance in some areas [
]. While it is well established that herbivory by urchins can
cause declines in kelp abundance, urchins are generally absent in the intertidal zone in our
system, with the exception of tidepools, and therefore are not likely to be responsible for observed
kelp losses. Herbivory by intertidal grazers however, especially Katharina tunicata, has been
shown to influence kelp bed diversity and species composition in some areas [
]. It is
unknown whether intertidal grazers are more abundant following sea star wasting disease
outbreaks of 2013?2014, and it is possible that changes in trophic dynamics could have
contributed to kelp losses. However, K. tunicata is predominantly found at wave exposed sites, rather
than at sheltered sites [
], and therefore cannot have driven the ubiquity of kelp declines
at moderate and sheltered sites. Declines at sheltered and moderately exposed sites occurred
regardless of substratum (boulder versus bedrock) or slope (steep versus shallow), factors
known to influence the distribution of invertebrates [
]. Therefore, the observed declines
are likely too widespread to have resulted from increases in abundance of a single grazer
species. Moreover, increases in grazers would have been expected to influence fucoids, such as
Sargassum, along with kelps [
], a result which did not occur. Although we cannot rule out a
role of herbivory in driving some declines, it is unlikely to be the most important driver.
Local stressors caused by human activity such as run-off or pollution are also unlikely to be
drivers of the declines that we document. Barkley Sound has very low population densities,
limiting human disturbance [
] and many of our sites occurred within Pacific Rim National
Park, a region that is largely uninhabited and protected from human disturbance.
In sum, while changes in kelp bed composition, diversity and abundance may have resulted
from multiple interacting factors, evidence is consistent with the hypothesis that temperature
anomalies during the 2013?2016 heatwave drove widespread declines in kelp bed diversity and
species abundance. Regardless of the timescale over which these declines occurred or the exact
combination of factors that have driven them, our results suggest that wave-sheltered habitats
are more sensitive to regional stressors than wave exposed habitats.
Implications of kelp loss
Given the important ecological role of kelp [
], the substantial declines that we
document are likely to have cascading effects on the diversity of other organisms and on ecosystem
functioning and productivity of intertidal communities [
]. While the affected kelp
communities may yet recover following the 2013?2016 heatwave, our results offer a novel prediction
16 / 25
for how communities will be affected by increasing climatic stressors. In particular, these
results suggest that kelp communities at wave-sheltered sites may be particularly sensitive to
the increasing prevalence of broad-scale stressors, such as more frequent and intense marine
heat waves [
]. It could be hypothesized that declines on wave-sheltered shores may not
affect regional productivity or habitat availability as much as would declines on wave-exposed
shorelines, which are more diverse and more productive . Yet, positive interactions
generated by kelp canopies may be especially important on wave-sheltered shores because these
shores are more physiologically stressful [
]. Furthermore, the lower diversity and
productivity of sheltered shorelines is far outweighed by their sheer abundance in the Northeast Pacific
Approximately 57,000 km of wave-sheltered rocky shoreline exists from Oregon to central
Alaska, virtually all of which (99.8%) occurs north (and east) of Washington?s outer coast (Fig
8A). Therefore, even small changes in kelp diversity on more sensitive wave-sheltered shores
could have large effects on intertidal productivity if magnified across the landscape. While
some of this shoreline may not be suitable kelp habitat due to limitations from salinity and
other factors, it is clear from our analyses that wave sheltered shorelines are common and
extensive in northern Washington, British Columbia and Alaska. Given that these types of
habitats are uncommon further south, it is likely that some northern shorelines will experience
losses in kelp abundance and diversity before southern ones.
The sensitivity of wave-sheltered sites in our system is contrasted by the apparent resilience
of wave exposed kelp beds, a novel finding that has important implications for conservation
and management. Wave exposed sites are highly productive and often represent hotspots of
diversity in our system [
]. Our results demonstrate that these sites may also be especially
resilient against broad-scale stressors. Wave exposed sites may act as refugia during times of
stress, potentially buffering kelp ecosystems against regional extinctions and playing a key role
in maintaining regional species diversity. The role of climatic refuges in maintaining species
diversity through geological time has been widely discussed in the paleoecological literature
]. However, few studies illustrate this phenomenon under ongoing global change.
Recent efforts to understand biodiversity change in ecologically important biogenic habitats
have identified areas where ecosystems are performing substantially better (?bright spots?) or
worse (?dark spots?) than average [
]. Our results demonstrate that a fine-scale
environmental gradient?one that can vary over tens of metres [
]?has mediated the formation of bright
spots and dark spots in kelp-dominated ecosystems. Importantly, given the distribution of
wave exposure in the Northeast Pacific (Fig 8), such dark spots are likely to be much more
common than bright spots across the landscape. As a result, although wave exposed sites
might maintain regional diversity, abundant dark spots could have profound effects on
ecosystem functioning and coastal productivity [
In addition to reductions in diversity, we document widespread declines in the abundance
of intertidal kelps in Barkley Sound. While the magnitude of decline was dependent on wave
exposure (Fig 3B) and varied between species (Figs 6 and 7), sites from all wave exposure
categories declined significantly in average kelp abundance and 7 of the 9 species most common
species declined in abundance. Losses of kelp cover are common worldwide [
] and a recent
global meta-analysis found that more than one third of published subtidal kelp bed surveys
showed declines over the past 50 years?significantly more than had increased [
many negatively affected kelp forest ecosystems are found near the warm-edge of kelps?
latitudinal range [
], our data suggest that similar declines have occurred in the intertidal
zones of British Columbia, reasonably far from the warmer latitudinal limit of northeast Pacific
kelp ecosystems [
]. This supports previous work suggesting that central-and not just
edgepopulations of brown algae may be susceptible to broad-scale stressors brought on by heat
17 / 25
Fig 8. Northeast Pacific intertidal habitat classified by wave exposure. Wave-sheltered habitat makes up the majority of Northeast Pacific shorelines (A) and
is abundant in British Columbia (C) and Alaska (B) but rare along the outer coast of Washington and Oregon (D). Bar plots show the proportions of rocky
shoreline of different wave-exposures and are accompanied by the length of predominantly rocky shoreline in each region (A) or in each inset (B-D). Mosaic
plot inset in A shows the relative proportions of rocky shoreline of different wave-exposures scaled by the length of coastline for each region. Regions from
South to North: (1) Oregon coast, (2) Washington outer coast, (3) Salish Sea, Puget Sound, and Strait of Georgia, (4) western Vancouver Island, (5) northern
British Columbia, (6) southern Alaska, (7) central Alaska.
]. This may be especially true for intertidal communities that show limited
correlations between latitude and thermal stress .
Although declines may be attributable to stressors occurring over short timescales [
1), rather than a response to gradual warming, the recovery from ecosystem-wide declines
may not occur rapidly in either case. Four of our sites lost all kelp species and thirteen others
were reduced to a single, sometimes rare (< 5% cover) species. Thus, many of our sites have
experienced complete or near-complete collapses of kelp-dominated communities. For the 17
18 / 25
sites that had the fewest kelps in 2017, similar results were found in 2018: four sites with no
kelps and 13 sites with only one kelp species (data in S1 and S2 Tables). Thus, even if declines
did occur recently, they have persisted for two years, indicating that recovery has not occurred
immediately following the heatwave. Kelp bed collapses have been documented previously in
various regions worldwide and many have yet to recover following initial kelp bed collapse
Scale-dependence of diversity loss and the importance of local gradients
A broader implication of our results, one that extends beyond rocky shores, is that important
biodiversity loss could easily remain hidden from studies not specifically designed with
environmental heterogeneity in mind. We found that the total diversity of kelps in Barkley Sound
has not changed throughout the region, yet a majority of sites experienced large losses in local
diversity. This clearly demonstrates how declines in diversity can be concentrated in only
some habitats that may be stressful and lower diversity to begin with. Studies that focus on
regional patterns or only investigate certain types of sites could miss losses mediated by local
gradients. Thus, differences between local conditions in distinct habitat patches may directly
contribute to the disconnect between diversity measurements taken at different spatial scales
]. In our study, species accumulation curves demonstrate how we could have missed the
widespread biotic homogenization that has occurred only at wave-sheltered sites were we to
assess all sites together (Fig 5). Capturing these losses in between-site diversity can be essential
to monitoring and conservation efforts because ecosystem functionality can depend on having
many species combinations across the landscape [
]. Yet, while our results support growing
evidence that local environmental heterogeneity explains important variation in diversity loss
], few studies that examine responses to ongoing global change incorporate these
gradients into their analyses. Our results point to the need for a framework that better incorporates
the interacting effects of stressors at different scales. Such an approach would hold much
promise for identifying and predicting diversity loss and changes in abundance not only at
species range edges but also along local gradients throughout the range of each species.
Heterogeneity in environmental variables, like wave exposure, is ubiquitous in the natural
world, but its importance in determining the responses of communities to broad scale stressors
is often underappreciated [
]. As global change continues to drive shifts in ecosystem
structure, heterogeneity of habitats will lead to variation in microclimates [
] and could strongly
affect the biological responses of organisms. Rather than assessing the average responses across
all communities in a region or across the globe [
], we should work to identify the habitats
that are most vulnerable to declines and determine whether they are abundant enough to
influence ecosystem functioning across the landscape. Consistent declines across all habitat patches
or at the most diverse, high quality habitats may not be reasonable predictions for how
communities will respond to global change [
]. Instead, some sites may act as refugia, while
diversity is lost from marginal habitats; if these sensitive habitats are common, as they are in our
system, then the consequences to ecosystem functioning could be profound.
S1 Table. Summary of survey data for all sites sampled in this study.
S2 Table. Raw data for communities surveyed in this study. Excel spreadsheet with
abundance and presence/absence data for all five survey years.
19 / 25
S1 Fig. Boxplot of cartographical Wave Exposure Index measure versus the wave exposure
categories used in this study, for SW facing sites (n = 26). Letters represent significant
differences between means as determined by a Kruskal-Wallis rank sum test followed by a Dunn?s
S2 Fig. Boxplot of the upper limit of barnacles at sites (n = 47) of each wave exposure
category. Letters represent significant differences between means as determined by a
Kruskal-Wallis rank sum test followed by a Dunn?s posthoc test.
S3 Fig. Proportional, site-level changes in (A) richness and (B) average abundance, broken
down by wave exposure. Both panels display ratios of modern and average historic
observations and red lines indicate zero change.
S4 Fig. Results of species pool bootstrap extrapolation between years. Estimates shown for
(A) all sites, (B) exposed, (C) moderate and (D) sheltered species pools, broken down by year,
as calculated using the specpool function in the R package ?vegan?. Points represent
bootstrapped estimates of species richness and error bars represent 95% confidence intervals.
S5 Fig. Maximum air and water temperatures between the periods of 1991?1995 and
2013?2017. Panel (A) shows the maximum daily air temperature averaged by month and by
time-period; data is from Cape Beale Lighthouse. Panel (B) shows the maximum monthly sea
surface temperature averaged by time-period; data is from Amphitrite Lighthouse. Both
lighthouses are located on the outer edge of Barkley Sound, British Columbia.
This work would not have been possible without the initial surveys which were conducted by
Druehl and Elliot and funded by Parks Canada. We thank the Huu-ay-aht First Nations, S.
Rogers, B. Anholt, S. Gray, B. Rogers, T. Eastham, N. Wiewel and other Bamfield Marine
Sciences Centre (BMSC) staff for making this work logistically feasible. Students in the 2017
Coastal Community Ecology class and the 2018 Marine Ecology class at BMSC and L.
Campbell contributed to field logistics and planning. We thank the many instructors, researchers,
and students who contributed to the collection of long-term monitoring data on Wizard Islet.
We thank C. Harley, P. Thompson and M. Whalen for their many insightful comments on an
earlier version of this paper, and thank J. Sunday for helpful feedback on data interpretation.
SS and CJN designed the study, collected and analyzed the data, and wrote the paper with
input from all authors. LB, EC, KJ, AW, MB, AD, MF and JT assisted with data collection, data
curation and sampling logistics. The authors declare no competing interests.
Conceptualization: Samuel Starko, Christopher J. Neufeld.
Data curation: Samuel Starko, Lauren A. Bailey, Elandra Creviston, Katelyn A. James, Alison
Warren, Megan K. Brophy, Andreea Danasel, Megan P. Fass, James A. Townsend,
Christopher J. Neufeld.
Formal analysis: Samuel Starko, Christopher J. Neufeld.
20 / 25
Investigation: Samuel Starko, Lauren A. Bailey, Elandra Creviston, Katelyn A. James, Alison
Warren, Megan K. Brophy, Andreea Danasel, Megan P. Fass, James A. Townsend,
Christopher J. Neufeld.
Methodology: Samuel Starko, Christopher J. Neufeld.
Project administration: Christopher J. Neufeld.
Supervision: Samuel Starko, Christopher J. Neufeld.
Validation: Samuel Starko.
Visualization: Samuel Starko, Christopher J. Neufeld.
Writing ? original draft: Samuel Starko, Christopher J. Neufeld.
Writing ? review & editing: Samuel Starko, Lauren A. Bailey, Elandra Creviston, Katelyn A.
James, Alison Warren, Megan K. Brophy, Andreea Danasel, Megan P. Fass, James A.
Townsend, Christopher J. Neufeld.
21 / 25
22 / 25
23 / 25
24 / 25
1. Isbell F , Gonzalez A , Loreau M , Cowles J , D??az S , Hector A , et al. Linking the influence and dependence of people on biodiversity across scales . Nature . 2017 ; 546 : 65 - 72 . https://doi.org/10.1038/nature22899 PMID: 28569811
2. Thomas CD . Local diversity stays about the same, regional diversity increases, and global diversity declines . Proc Natl Acad Sci . 2013 ; 110 : 19187 - 19188 . https://doi.org/10.1073/pnas.1319304110 PMID: 24248347
3. Gonzalez A , Cardinale BJ , Allington GRH , Byrnes J , Arthur Endsley K , Brown DG , et al. Estimating local biodiversity change: a critique of papers claiming no net loss of local diversity . Ecology . 2016 ; 97 : 1949 - 1960 . https://doi.org/10. 1890 /15- 1759 .1 PMID: 27859190
4. Cardinale BJ , Gonzalez A , Allington GRH , Loreau M. Is local biodiversity declining or not? A summary of the debate over analysis of species richness time trends . Biol Conserv . 2018 ; 219 : 175 - 183 . https:// doi.org/10.1016/j.biocon. 2017 . 12 .021
5. Helmuth B , Harley CDG , Halpin PM , O'Donnell M , Hofmann GE , Blanchette CA. Climate change and latitudinal patterns of intertidal thermal stress . Science . 2002 ; 298 : 1015 - 1017 . https://doi.org/10.1126/ science.1076814 PMID: 12411702
6. Helmuth B , Russell BD , Connell SD , Dong Y , Harley CD , Lima FP , et al. Beyond long-term averages: making biological sense of a rapidly changing world . Clim Change Responses . 2014 ; 1 . https://doi.org/ 10.1186/s40665-014-0006-0
7. Russell BD , Connell SD . Origins and consequences of global and local stressors: incorporating climatic and non-climatic phenomena that buffer or accelerate ecological change . Mar Biol . 2012 ; 159 : 2633 - 2639 . https://doi.org/10.1007/s00227-011-1863-8
8. Parmesan C , Yohe G. A globally coherent fingerprint of climate change impacts across natural systems . Nature . 2003 ; 421 : 37 - 42 . https://doi.org/10.1038/nature01286 PMID: 12511946
9. Dornelas M , Gotelli NJ , McGill B , Shimadzu H , Moyes F , Sievers C , et al. Assemblage Time Series Reveal Biodiversity Change but Not Systematic Loss. Science . 2014 ; 344 : 296 . https://doi.org/10.1126/ science.1248484 PMID: 24744374
10. Vellend M , Baeten L , Myers-Smith IH , Elmendorf SC , Beause?jour R , Brown CD , et al. Global meta-analysis reveals no net change in local-scale plant biodiversity over time . Proc Natl Acad Sci . 2013 ; 110 : 19456 - 19459 . https://doi.org/10.1073/pnas.1312779110 PMID: 24167259
11. Krumhansl KA , Okamoto DK , Rassweiler A , Novak M , Bolton JJ , Cavanaugh KC , et al. Global patterns of kelp forest change over the past half-century . Proc Natl Acad Sci . 2016 ; 113 : 13785 - 13790 . https:// doi.org/10.1073/pnas.1606102113 PMID: 27849580
12. Tewksbury JJ , Huey RB , Deutsch CA. Putting the heat on tropical animals . Science . 2008 ; 320 : 1296 - 1297 . https://doi.org/10.1126/science.1159328 PMID: 18535231
13. Wernberg T , Bennett S , Babcock RC , Bettignies T de, Cure K , Depczynski M , et al. Climate-driven regime shift of a temperate marine ecosystem . Science . 2016 ; 353 : 169 - 172 . https://doi.org/10.1126/ science.aad8745 PMID: 27387951
14. Harley CDG . Climate change, keystone predation, and biodiversity loss . Science . 2011 ; 334 : 1124 - 1127 . https://doi.org/10.1126/science.1210199 PMID: 22116885
15. Helmuth B , Choi F , Matzelle A , Torossian JL , Morello SL , Mislan K a. S, et al. Long-term, high frequency in situ measurements of intertidal mussel bed temperatures using biomimetic sensors . Sci Data . 2016 ; 3 : 160087 . https://doi.org/10.1038/sdata. 2016 .87 PMID: 27727238
16. Denny MW . Ocean waves, nearshore ecology, and natural selection . Aquat Ecol . 2006 ; 40 : 439 - 461 .
17. Leigh EG , Paine RT , Quinn JF , Suchanek TH . Wave energy and intertidal productivity . Proc Natl Acad Sci . 1987 ; 84 : 1314 - 1318 . PMID: 16593813
18. Hurd CL . Shaken and stirred: the fundamental role of water motion in resource acquisition and seaweed productivity . Perspect Phycol . 2017 ; 73 - 81 . https://doi.org/10.1127/pip/2017/0072
19. Fitzhenry T , Halpin PM , Helmuth B. Testing the effects of wave exposure, site, and behavior on intertidal mussel body temperatures: applications and limits of temperature logger design . Mar Biol . 2004 ; 145 : 339 - 349 . https://doi.org/10.1007/s00227-004-1318-6
20. Bond NA , Cronin MF , Freeland H , Mantua N. Causes and impacts of the 2014 warm anomaly in the NE Pacific . Geophys Res Lett . 2015 ; 42 : 3414 - 3420 .
21. Levine AFZ , McPhaden MJ . How the July 2014 easterly wind burst gave the 2015-2016 El Ni?o a head start . Geophys Res Lett . 2016 ; 43 : 6503 - 6510 . https://doi.org/10.1002/2016GL069204
22. Teagle H , Hawkins SJ , Moore PJ , Smale DA . The role of kelp species as biogenic habitat formers in coastal marine ecosystems . J Exp Mar Biol Ecol . 2017 ; 492 : 81 - 98 . https://doi.org/10.1016/j.jembe. 2017 . 01 .017
23. Steneck RS , Graham MH , Bourque BJ , Corbett D , Erlandson JM , Estes JA , et al. Kelp forest ecosystems: biodiversity, stability, resilience and future . Environ Conserv . 2002 ; 29 : 436 - 459 .
24. Graham MH . Effects of local deforestation on the diversity and structure of southern California giant kelp forest food webs . Ecosystems . 2004 ; 7 : 341 - 357 .
25. Lu?ning K , Freshwater W. Temperature tolerance of Northeast Pacific marine algae . J Phycol . 1988 ; 24 : 310 - 315 . https://doi.org/10.1111/j.1529- 8817 . 1988 .tb04471.x
26. tom Dieck I. Temperature tolerance and survival in darkness of kelp gametophytes (Laminariales, Phaeophyta): ecological and biogeographical implications . Mar Ecol Prog Ser . 1993 ; 100 : 253 - 264 . 27 . Wernberg T , Smale DA , Tuya F , Thomsen MS , Langlois TJ , de Bettignies T , et al. An extreme climatic event alters marine ecosystem structure in a global biodiversity hotspot . Nat Clim Change . 2012 ; 3 : 78 - 82 . https://doi.org/10.1038/nclimate1627
28. Filbee-Dexter K , Wernberg T. Rise of turfs: a new battlefront for globally declining kelp forests . BioScience . 2018 ; 68 : 64 - 76 . https://doi.org/10.1093/biosci/bix147
29. Wernberg T , Krumhansl K , Filbee-Dexter K , Pedersen MF . Chapter 3-Status and Trends for the World's Kelp Forests . In: Sheppard C, editor. World Seas: an Environmental Evaluation ( Second Edition) . Academic Press; 2019 . pp. 57 - 78 . https://doi.org/10.1016/B978-0 -12-805052-1 . 00003 - 6
30. Barber RT , Chavez FP . Biological consequences of El Ni?o . Science . 1983 ; 222 : 1203 - 1210 . https:// doi.org/10.1126/science.222.4629.1203 PMID: 17806711
31. Dean TA , Jacobsen FR . Nutrient-limited growth of juvenile kelp, Macrocystis pyrifera, during the 1982- 1984 ?El Ni?o? in southern California . Mar Biol . 1986 ; 90 : 597 - 601 . https://doi.org/10.1007/ BF00409280
32. Zimmerman RC , Kremer JN . Episodic nutrient supply to a kelp forest ecosystem in Southern California . J Mar Res . 1984 ; 42 : 591 - 604 .
33. Gerard VA . The role of nitrogen nutrition in high-temperature tolerance of the kelp, Laminaria saccharina (Chromophyta) . J Phycol . 1997 ; 33 : 800 - 810 . https://doi.org/10.1111/j.0022- 3646 . 1997 . 00800 .x 34. 35. Wernberg T , Russell BD , Moore PJ , Ling SD , Smale DA , Campbell A , et al. Impacts of climate change in a global hotspot for temperate marine biodiversity and ocean warming . J Exp Mar Biol Ecol . 2011 ; 400 : 7 - 16 . https://doi.org/10.1016/j.jembe. 2011 . 02 .021
Witman JD , Lamb RW . Persistent differences between coastal and offshore kelp forest communities in a warming Gulf of Maine . PLoS ONE . 2018 ; 13 . https://doi.org/10.1371/journal.pone. 0189388 PMID: 29298307
36. Pfister CA , Berry HD , Mumford T. The dynamics of kelp forests in the Northeast Pacific Ocean and the relationship with environmental drivers . J Ecol . 2017 ; 106 : 1520 - 1533 . https://doi.org/10.1111/ 1365 - 2745 . 12908
37. Bennett S , Wernberg T , Joy BA , Bettignies T de, Campbell AH. Central and rear-edge populations can be equally vulnerable to warming . Nat Commun . 2015 ; 6 : 10280 . https://doi.org/10.1038/ ncomms10280 PMID: 26691184
38. Reed D , Washburn L , Rassweiler A , Miller R , Bell T , Harrer S. Extreme warming challenges sentinel status of kelp forests as indicators of climate change . Nat Commun . 2016 ; 7 : 13757 . https://doi.org/10. 1038/ncomms13757 PMID: 27958273
39. Druehl LD , Elliot CTJ . Parks Canada Barkley Sound Kelp Distribution . Parks Canada Contract # K3489 - 3- 0006 . Parks Canada Contract # K3489 -3-0006; 1996 .
40. Amos CL , Martino S , Sutherland TF , Al Rashidi T. Sea surface temperature trends in the coastal zone of British Columbia, Canada . J Coast Res . 2015 ; 300 : 434 - 446 . https://doi.org/10.2112/JCOASTRESD-14 -00114 . 1
41. Jacox MG , Hazen EL , Zaba KD , Rudnick DL , Edwards CA , Moore AM , et al. Impacts of the 2015-2016 El Ni?o on the California Current System: Early assessment and comparison to past events . Geophys Res Lett . 2016 ; 43 : 7072 - 7080 . https://doi.org/10.1002/2016GL069716
42. Schultz JA , Cloutier RN , Co?te? IM. Evidence for a trophic cascade on rocky reefs following sea star mass mortality in British Columbia . PeerJ . 2016 ; 4: e1980 . https://doi.org/10.7717/peerj.1980 PMID: 27168988
43. Burt JM , Tinker MT , Okamoto DK , Demes KW , Holmes K , Salomon AK . Sudden collapse of a mesopredator reveals its complementary role in mediating rocky reef regime shifts . Proc R Soc B . 2018 ; 285 : 20180553 . https://doi.org/10.1098/rspb. 2018 .0553 PMID: 30051864
44. Helmuth B , Denny MW . Predicting wave exposure in the rocky intertidal zone: do bigger waves always lead to larger forces? Limnol Oceanogr . 2003 ; 48 : 1338 - 1345 .
45. Topinka J , Tucker L , Korjeff W. The Distribution of fucoid macroalgal biomass along central coastal maine . Bot Mar . 2009 ; 24 : 311 - 320 . https://doi.org/10.1515/botm. 1981 . 24 .6. 311
46. Burns KC , Neufeld CJ . Plant extinction dynamics in an insular metacommunity . Oikos . 2009 ; 118 : 191 - 198 . https://doi.org/10.1111/j.1600- 0706 . 2008 . 16816 .x
47. Neufeld CJ , Starko S , Burns K. Disturbance and diversity in a continental archipelago: A mechanistic framework linking area, height, and exposure . Ecosphere . 2017 ; 8: e01957 . https://doi.org/10.1002/ ecs2.1957
48. Howes D. BC biophysical shore-zone mapping system-A systematic approach to characterize coastal habitats in the Pacific Northwest . Proceedings of the 2001 Puget Sound Research Conference , Seattle, Washington: 11p. 2001 .
49. Arsenault DJ , Marchinko KB , Palmer AR . Precise tuning of barnacle leg length to coastal wave action . Proc R Soc Lond B Biol Sci . 2001 ; 268 : 2149 - 2154 . https://doi.org/10.1098/rspb. 2001 .1776 PMID: 11600079
50. R Core Team. R: A language and environment for statistical computing [Internet] . Vienna, Austria: R Foundation for Statistical Computing; 2017 . Available: https://www.R-project.org
51. Paradis E , Claude J , Strimmer K. APE: Analyses of phylogenetics and evolution in R language . Bioinformatics . 2004 ; 20 : 289 - 290 . https://doi.org/10.1093/bioinformatics/btg412 PMID: 14734327
52. Oksanen J , Blanchet FG , Friendly M , Kindt R , Legendre P , McGlinn D , et al. vegan: Community Ecology Package [Internet] . 2018 . Available: https://CRAN.R-project.org/package=vegan
53. Yee TW . Vector Generalized Linear and Additive Models: With an Implementation in R [Internet]. New York: Springer-Verlag; 2015 . Available: //www.springer.com/gp/book/9781493928170
54. Ulrich J. TTR : Technical Trading Rules [Internet]. 2018 . Available: https://CRAN.R-project.org/ package=TTR
55. Starko S , Martone P. An empirical test of ?universal? biomass scaling relationships in kelps: Evidence of convergence with seed plants . 2016 ; https://doi.org/10.1111/nph.14120 PMID: 27479188
56. Lorenzo ED , Mantua N. Multi-year persistence of the 2014/15 North Pacific marine heatwave . Nat Clim Change . 2016 ; 6 : 1042 - 1047 . https://doi.org/10.1038/nclimate3082
57. Gentemann CL , Fewings MR , Garc??a-Reyes M. Satellite sea surface temperatures along the West Coast of the United States during the 2014-2016 northeast Pacific marine heat wave . Geophys Res Lett . 2017 ; 44 : 312 - 319 . https://doi.org/10.1002/2016GL071039
58. Staehr PA , Pedersen MF , Thomsen MS , Wernberg T , Krause-Jensen D . Invasion of Sargassum muticum in Limfjorden (Denmark) and its possible impact on the indigenous macroalgal community . Mar Ecol Prog Ser . 2000 ; 207 : 79 - 88 . https://doi.org/10.3354/meps207079
59. Viejo RM . The effects of colonization by Sargassum muticum on tidepool macroalgal assemblages . J Mar Biol Assoc U K . 1997 ; 77 : 325 - 340 . https://doi.org/10.1017/S0025315400071708
60. Riosmena-Rodr??guez R , Boo GH , Lo?pez- Vivas JM , Herna? ndez-Velasco A , Sa?enz-Arroyo A , Boo SM . The invasive seaweed Sargassum filicinum (Fucales, Phaeophyceae) is on the move along the Mexican Pacific coastline . Bot Mar . 2012 ; 55 . https://doi.org/10.1515/bot-2012 -0120
61. Cosson J . On the progressive disappearance of Laminaria digitata on the coasts of Calvados (France) . Cryptogam Algol . 1999 ; 1 : 35 - 42 .
62. Harley CDG , Helmuth BST . Local- and regional-scale effects of wave exposure, thermal stress, and absolute versus effective shore level on patterns of intertidal zonation . Limnol Oceanogr . 2003 ; 48 : 1498 - 1508 . https://doi.org/10.4319/lo. 2003 . 48 .4. 1498
63. Druehl L , Clarkston B. Pacific Seaweeds [Internet]. Harbour Publishing; 2016 . Available: http://www. harbourpublishing.com/title/PacificSeaweeds2
64. Widdowson TB . Some aspects of the intertidal ecology of marine organisms on Vancouver Island between Victoria and Port Renfrew . [Internet]. University of British Columbia. 1959 . https://doi.org/10. 14288/1.0106061
65. MacMillan C. The kelps of Juan de Fuca. Postelsia . 1902 ; 1 : 193 - 220 .
66. Henley WJ , Dunton KH . Effects of nitrogen supply and continuous darkness on growth and photosynthesis of the arctic kelp Laminaria solidungula . Limnol Oceanogr . 1997 ; 42 : 209 - 216 . https://doi.org/10. 4319/lo. 1997 . 42 .2. 0209
67. Thomas MK , Aranguren-Gassis M , Kremer CT , Gould MR , Anderson K , Klausmeier CA , et al. Temperature-nutrient interactions exacerbate sensitivity to warming in phytoplankton . Glob Change Biol . 2017 ; 23 : 3269 - 3280 . https://doi.org/10.1111/gcb.13641 PMID: 28132424
68. Harvell CD , Montecino-Latorre D , Caldwell JM , Burt JM , Bosley K , Keller A , et al. Disease epidemic and a marine heat wave are associated with the continental-scale collapse of a pivotal predator (Pycnopodia helianthoides) . Sci Adv . 2019 ; 5: eaau7042 . https://doi.org/10.1126/sciadv.aau7042 PMID: 30729157
69. Bates AE , Hilton BJ , Harley CDG . Effects of temperature, season and locality on wasting disease in the keystone predatory sea star Pisaster ochraceus . Dis Aquat Organ . 2009 ; 86 : 245 - 251 . https://doi.org/ 10.3354/dao02125 PMID: 20066959
70. Paine RT . Trophic control of production in a rocky intertidal community . Science . 2002 ; 296 : 736 - 739 . https://doi.org/10.1126/science.1069811 PMID: 11976455
71. Dethier M , Duggins D. Variation in strong interactions in the intertidal zone along a geographical gradient: a Washington-Alaska comparison . Mar Ecol Prog Ser . 1988 ; 50 : 97 - 105 . https://doi.org/10.3354/ meps050097
72. Puchalski SS . Taphonomic approaches to interpretation of the fossil record of chitons (Mollusca, Polyplacophora) [Internet] . Ph.D. , Indiana University. 2010 . Available: http://search.proquest.com/docview/ 851550478/abstract/27B5AD460B57472CPQ/1
73. Sigwart JD , Green PA , Crofts SB . Functional morphology in chitons (Mollusca, Polyplacophora): influences of environment and ocean acidification . Mar Biol . 2015 ; 162 : 2257 - 2264 . https://doi.org/10.1007/ s00227-015-2761-2
74. Linsenmeyer TA . The resistance of five species of polyplacophorans to removal from natural and artificial surfaces . Veliger . 1975 ; 18 : 83 - 86 .
75. D'Antonio C. Role of sand in the domination of hard substrata by the intertidal alga Rhodomela larix . Mar Ecol Prog Ser . 1986 ; 27 : 263 - 275 . https://doi.org/10.3354/meps027263
76. Light SF . Intertidal invertebrates of the Central California Coast . University of California Press; 1954 .
77. Underwood AJ , Chapman MG . Scales of spatial patterns of distribution of intertidal invertebrates . Oecologia . 1996 ; 107 : 212 - 224 . https://doi.org/10.1007/BF00327905 PMID: 28307307
78. Britton-Simmons KH , Pister B , Sa?nchez I , Okamoto D. Response of a native, herbivorous snail to the introduced seaweed Sargassum muticum . Hydrobiologia . 2011 ; 661 : 187 - 196 . https://doi.org/10.1007/ s10750-010-0523-1
79. Iacarella JC , Adamczyk E , Bowen D , Chalifour L , Eger A , Heath W , et al. Anthropogenic disturbance homogenizes seagrass fish communities . Glob Change Biol . 2018 ; 24 : 1904 - 1918 . https://doi.org/10. 1111/gcb.14090 PMID: 29431880
80. Duggins DO , Simenstad CA , Estes JA . Magnification of secondary production by kelp detritus in coastal marine ecosystems . Science (Washington). 1989 ; 245 : 170 - 173 .
81. Fro? licher TL , Fischer EM , Gruber N. Marine heatwaves under global warming . Nature . 2018 ; 560 : 360 . https://doi.org/10.1038/s41586-018 -0383-9 PMID: 30111788
82. Oliver ECJ , Donat MG , Burrows MT , Moore PJ , Smale DA , Alexander LV , et al. Longer and more frequent marine heatwaves over the past century . Nat Commun . 2018 ; 9 : 1324 . https://doi.org/10.1038/ s41467-018 -03732-9 PMID: 29636482
83. Alexander MA , Deser C , Timlin MS . The reemergence of SST anomalies in the North Pacific Ocean . J Clim . 1999 ; 12 : 2419 - 2433 . https://doi.org/10.1175/ 1520 - 0442 ( 1999 ) 012 < 2419 : TROSAI>2.0 .CO; 2
84. Smale DA , Wernberg T , Oliver ECJ , Thomsen M , Harvey BP , Straub SC , et al. Marine heatwaves threaten global biodiversity and the provision of ecosystem services . Nat Clim Change . 2019 ; 1 . https:// doi.org/10.1038/s41558-019-0412-1
85. Bruno JF , Stachowicz JJ , Bertness MD . Inclusion of facilitation into ecological theory . Trends Ecol Evol . 2003 ; 18 : 119 - 125 . https://doi.org/10.1016/S0169- 5347 ( 02 ) 00045 - 9
86. Willis KJ , Whittaker RJ . The refugial debate . Science . 2000 ; 287 : 1406 - 1407 . https://doi.org/10.1126/ science.287.5457.1406 PMID: 10722388
87. Haffer J. Speciation in Amazonian forest birds . Science . 1969 ; 165 : 131 - 137 . https://doi.org/10.1126/ science.165.3889.131 PMID: 17834730
88. Cinner JE , Huchery C , MacNeil MA , Graham NAJ , McClanahan TR , Maina J , et al. Bright spots among the world's coral reefs . Nature . 2016 ; 535 : 416 - 419 . https://doi.org/10.1038/nature18607 PMID: 27309809
89. Loreau M , Naeem S , Inchausti P , Bengtsson J , Grime JP , Hector A , et al. Biodiversity and ecosystem functioning: current knowledge and future challenges . Science . 2001 ; 294 : 804 - 808 . https://doi.org/10. 1126/science.1064088 PMID: 11679658
90. Assis J , Berecibar E , Claro B , Alberto F , Reed D , Raimondi P , et al. Major shifts at the range edge of marine forests: the combined effects of climate changes and limited dispersal . Sci Rep . 2017 ; 7 : 44348 . https://doi.org/10.1038/srep44348 PMID: 28276501
91. Bolton JJ . The biogeography of kelps (Laminariales, Phaeophyceae): a global analysis with new insights from recent advances in molecular phylogenetics . Helgol Mar Res . 2010 ; 64 : 263 - 279 . https:// doi.org/10.1007/s10152-010-0211-6
92. Thomsen MS , Mondardini L , Alestra T , Gerrity S , Tait L , South PM , et al. Local Extinction of Bull Kelp (Durvillaea spp .) Due to a Marine Heatwave . Front Mar Sci . 2019 ; 6 . https://doi.org/10.3389/fmars. 2019 .00084
93. Filbee-Dexter K , Feehan CJ , Scheibling RE . Large-scale degradation of a kelp ecosystem in an ocean warming hotspot . Mar Ecol Prog Ser . 2016 ; 543 : 141 - 152 . https://doi.org/10.3354/meps11554
94. Raybaud V , Beaugrand G , Goberville E , Delebecq G , Destombe C , Valero M , et al. Decline in kelp in west Europe and climate . PloS One . 2013 ; 8: e66044 . https://doi.org/10.1371/journal.pone. 0066044 PMID: 23840397
95. Fanin N , Gundale MJ , Farrell M , Ciobanu M , Baldock JA , Nilsson M-C , et al. Consistent effects of biodiversity loss on multifunctionality across contrasting ecosystems . Nat Ecol Evol . 2017 ; 2 : 269 - 278 . https://doi.org/10.1038/s41559-017 -0415-0 PMID: 29255299
96. Elahi R , O'Connor MI , Byrnes JEK , Dunic J , Eriksson BK , Hensel MJS , et al. Recent trends in localscale marine biodiversity reflect community structure and human impacts . Curr Biol CB . 2015 ; 25 : 1938 - 1943 . https://doi.org/10.1016/j.cub. 2015 . 05 .030 PMID: 26166784