Sunflower Sea Star (Pycnopodia helianthoides): COSEWIC assessment and status report 2025

Official title: COSEWIC assessment and status report on the Sunflower Sea Star (Pycnopodia helianthoides) in Canada

Endangered

2025

COSEWIC assessment summary

Assessment summary – May 2025

Common name: Sunflower Sea Star

Scientific name: Pycnopodia helianthoides

Status: Endangered

Reason for designation: Historically found in the inshore waters along much of the Pacific Coast of North America, this is the world's largest sea star and the only member of its genus. The species was stricken by Sea Star Wasting Disease following abnormally high ocean temperatures during 2014 to 2015. Extensive diving survey data suggest a loss of over 75% of the Canadian population, with no evidence of recovery since 2015. The species is largely gone from the U.S. portion of its range to the south and Alaskan subpopulations to the north have also declined precipitously, limiting the possibility of rescue. Given these recent declines, continuing disease outbreaks and little potential for rescue, this species is Endangered and at risk of extirpation in Canada.

Occurrence: Pacific Ocean, British Columbia

Status history: Designated Endangered in May 2025.

COSEWIC executive summary

Sunflower Sea Star

Pycnopodia helianthoides

Wildlife species description and significance

Sunflower Sea Star (Pycnopodia helianthoides) is the largest known sea star. Adults have up to 24 arms and reach up to 130 cm in diameter. The species comes in various colours, including red, purple, orange, and brown, with a body surface covered in short spines. Sunflower Sea Star preys on a wide range of ecologically, culturally, recreationally, and commercially important species, such as clams, mussels, abalone, crabs, sea urchins, and sea cucumbers. Between 2013 and 2015, Sunflower Sea Star experienced a mass mortality across its entire range due to sea star wasting disease (SSWD), likely linked to a marine heatwave. Sunflower Sea Star prey on sea urchins, which in turn feed on kelp; in places where the sea star has disappeared due to SSWD, urchin populations have increased dramatically, resulting in kelp decline.

Distribution

Sunflower Sea Star is endemic to the west coast of North America. It previously ranged from Baja California, Mexico, to Alaska, U.S.A., but is nearly extirpated from Mexico, California, Oregon, and the outer coast of Washington following SSWD. In Canada, Sunflower Sea Star is still found along the entire coast of British Columbia (B.C.), including Haida Gwaii and Vancouver Island, albeit at lower than historical densities.

Habitat

Sunflower Sea Star is a habitat generalist, inhabiting mud, sand, and gravel substrates in kelp forests, in eelgrass meadows, and on rocky reefs. It is found from the low intertidal to a depth of approximately 500 m, although it is most abundant at depths of less than 75 m.

Biology

Sunflower Sea Star breeds in winter and spring. It releases its gametes in the water column, where fertilization occurs. The larvae remain in the water column for approximately 8 weeks and then settle on the bottom to metamorphose into five-armed juveniles. The number of arms increases with age. The age at maturity of Sunflower Sea Star is not known. One estimate of lifespan is 48 to 68 years, with a corresponding generation time of 27 to 37 years. However, a lifespan of 11 to 14 years may be more common, with a corresponding generation time of 8 to 10 years. Sunflower Sea Star is one of the most active sea stars: individuals follow tide movements, hunt prey, and escape predators or unsuitable conditions.

Population sizes and trends

The population size of Sunflower Sea Star in Canada cannot be estimated accurately. Based on aggregated time series spanning 2000 to 2021 and including more than 10,000 underwater surveys, a conservative estimate of the decline in the abundance of Sunflower Sea Star in Canada over this period is 82.1% (95% CI: 76.0, 87.2%), with most of the decline occurring over two years (that is, 2014 and 2015). The estimated decline in the probability of sighting over the same period is 68.5% (95% CI: 53.9, 79.5%), with most of the decline occurring in 2014 and 2015. Prior to 2013, Sunflower Sea Star was sighted in three of every four surveys; after SSWD, they were recorded in one of every five surveys. Although the available data do not span what is likely to be three generations (which may be 30 years or more), the stability of the population in the first 13 years of the time series indicates that the decline associated with SSWD has caused the population size to fall well below historical levels.

Threats and limiting factors

The most serious and plausible threat to Sunflower Sea Star is SSWD, which first presents as skin lesions before rapidly progressing to necrosis, loss of limbs, and death. The causative agent of what is a transmissible disease remains unidentified. Ongoing controlled challenge experiments have shown that SSWD can be controlled and transmitted from a wasting Sunflower Sea Star via co-housing, water, and inoculation with tissue homogenate and coelomic fluid from a wasting Sunflower Sea Star. Sunflower Sea Star specimens kept in isolation and exposed to heat-treated components of tissue homogenate and coelomic fluid from a wasting Sunflower Sea Star do not develop signs of SSWD. Through co-housing, SSWD has been transmitted from Sunflower Sea Star individuals to Ochre Sea Star (Pisaster ochraceus) individuals and vice versa. These results suggest a single causative agent is necessary for signs of disease. Small-scale outbreaks of SSWD in Sunflower Sea Star are ongoing.

The onset of SSWD coincided with a climate change–induced warm-water event (“the blob”) in the northeastern Pacific. Locally, disease outbreaks occurred more quickly and caused more severe declines in warmer regions. Sunflower Sea Star declines in shallow nearshore waters from California to Alaska were more likely when sea surface temperatures had recently been warmer than average. These observations are consistent with the SSWD outbreak’s being linked to climate change.

As broadcast spawners, the fertilization success of Sunflower Sea Star may be limited by inter-individual distance. Low population densities may result in distances between males and females that are too large to permit a high rate of gamete encounter.

Protection, status and ranks

Approximately 24.5% of B.C.’s coastline receives some form of protection, and most of B.C.’s 200 or so marine protected areas (MPAs) overlap with the range of Sunflower Sea Star. The levels of protection are variable, and few exclude all extractive activities. However, even the strictest MPAs do not protect Sunflower Sea Star from SSWD or climate change. The species is legally protected under the Canada National Parks Act in the Pacific Rim and Gulf Islands national park reserves and under the Canada National Marine Conservation Areas Act in Gwaii Haanas. Sunflower Sea Star was assessed as Critically Endangered on the IUCN Red List of Threatened Species in 2021. The species has been proposed for listing under the U.S. Endangered Species Act.

Technical summary

^{Footnote 1}

Pycnopodia helianthoides

Sunflower Sea Star

Solaster géant

No Indigenous names found for the species

Range of occurrence in Canada (province/territory/ocean): Pacific Ocean and British Columbia

Demographic information

Generation time (usually average age of parents in the population; indicate if another method of estimating generation time indicated in the IUCN guidelines (2011) is being used)

Published IUCN Red List estimate is 27 to 37 years; more recent observations suggest 10 to 20 years, used here.

Estimate is weak; see Biology – Life cycle and reproduction.

Is there an [observed, inferred, or projected] continuing decline in number of mature individuals?

Unknown

Surveys show no decline in number of mature individuals since 2015. Generation time estimate weak.

[Observed, estimated, or projected] percent of continuing decline in total number of mature individuals within 3 years [or 1 generation; whichever is longer up to a maximum of 100 years]

Unknown

Surveys show no decline in number of mature individuals since 2015. Generation time estimate is weak.

Estimated percent of continuing decline in total number of mature individuals within 5 years [or 2 generations; whichever is longer up to a maximum of 100 years]

Unknown

Surveys show no decline in number of mature individuals since 2015. Generation time estimate is weak.

[Observed, estimated, inferred, or suspected] percent [reduction or increase] in total number of mature individuals over the last [10 years, or 3 generations; whichever is longer].

Data not available for 3 generations. Estimates of percent decline 2000 to 2021:

• -82.1 in abundance (95% CI: -76.0, -87.2)
• -68.5 in prob(sighting) (95% CI: -53.9, -79.5)

Stable population before and after decline consistent with observed percent reduction capturing 3-generation decline. See Population Sizes and Trends.

[Projected, inferred, or suspected]

percent [reduction or increase] in total number of mature individuals over the next [10 years, or 3 generations, up to a maximum of 100 years].

Unknown

Surveys show no decline in number of mature individuals since 2015. Generation time estimate is weak.

[Observed, estimated, inferred, or suspected] percent [reduction or increase] in total number of mature individuals over any period [10 years, or 3 generations, whichever is longer up to a maximum of 100 years], including both the past and the future.

Unknown

Surveys show no decline in number of mature individuals since 2015. Generation time estimate is weak.

Are the causes of the decline clearly reversible?

No

SSWD identified as cause of recent decline, but agent not identified. Climate change–mediated warming implicated.

Are the causes of the decline clearly understood?

Partially

SSWD identified as cause of recent decline, but agent not identified. Climate change–mediated warming implicated.

Have the causes of the decline clearly ceased?

No

Local SSWD outbreaks continue, and climate change–mediated warming implicated.

Are there extreme fluctuations in number of mature individuals?

Unknown

None observed or expected based on biology; however, observation time is limited to few generations.

Extent and occupancy information

Estimated extent of occurrence (EOO)

219,452 km²

Calculated based on minimum convex polygon around all known occurrences since 1972; see main text.

Index of area of occupancy (IAO) (Always report 2x2 grid value).

7,652 km²

Calculated based on all known occurrences since 1972; see main text.

Is the population “severely fragmented” that is, is >50% of its total area of occupancy in habitat patches that are (a) smaller than would be required to support a viable population, and (b) separated from other habitat patches by a distance larger than the species can be expected to disperse?

1. No
2. No

Data and biology all consistent with wide-ranging dispersal.

Number of “locations”^*(use plausible range to reflect uncertainty if appropriate)

1

Based on SSWD.

Is there an [observed, inferred, or projected] decline in extent of occurrence?

Unknown

Is there an [observed, inferred, or projected] decline in index of area of occupancy?

Yes

Recent decline observed: loss from 31% of areas surveyed before and after SSWD; gain in only 2%.

Is there an [observed, inferred, or projected] decline in number of subpopulations?

N/A

Only one subpopulation in the population.

Is there an [observed, inferred, or projected] decline in number of “locations”?

N/A

Only one location based on main threat.

Is there an [observed, inferred, or projected] decline in [area, extent and/or quality] of habitat?

Yes

Observed decline in quality of habitat due to main threat (SSWD) and climate change.

Are there extreme fluctuations in number of subpopulations?

N/A

Only one subpopulation in the population.

Are there extreme fluctuations in number of "locations"?

N/A

Only one location based on major threat.

Are there extreme fluctuations in extent of occurrence?

Unknown

Are there extreme fluctuations in area of occupancy?

Unknown

Number of mature individuals (in each subpopulation)

Subpopulations (give plausible ranges)

N/A (only one subpopulation)

Total

Unknown; see main text

Quantitative analysis

Is the probability of extinction in the wild at least [20% within 20 years or 5 generations whichever is longer up to a maximum of 100 years, or 10% within 100 years]?

N/A (not done)

Threats

Was a threats calculator completed for this species?

Yes (see Appendix 1)

Overall threat: High–Medium

Key threats were identified as:

8.6 Diseases of Unknown Cause: Very high – Medium

9.2 Industrial and Military Effluents: Low

11.1 Habitat Shifting and Alteration: Low

11.3 Temperature Extremes: Low

5.4 Fishing and Harvesting Aquatic Resources: Unknown

8.3 Introduced Genetic Material: Unknown

11.4 Storms and Flooding: Unknown

What limiting factors are relevant?

Low population density could limit fertilization success.

Rescue effect (immigration from outside Canada)

Status of outside population(s) most likely to provide immigrants to Canada.

Declining

Subpopulations south thought largely extirpated; subpopulations north reportedly declined >90%.

Is immigration known or possible?

Yes

Possible via larval dispersal.

Would immigrants be adapted to survive in Canada?

Yes

In the absence of main threat (SSWD), very likely.

Is there sufficient habitat for immigrants in Canada?

Yes

In the absence of main threat (SSWD), very likely.

Are conditions deteriorating in Canada?

Unknown

Sea temperatures increasing, but net effect not known.

Are conditions for the source (that is, outside) population deteriorating?⁺

Unknown

Sea temperatures increasing, but net effect not known.

Is the Canadian population considered to be a sink?⁺

No

No data to suggest Canadian portion of range different in kind.

Is rescue from outside Canada likely, such that it could lead to a change in status?

No

Main threat (SSWD) has hit entire range of species.

Wildlife species with sensitive occurrence data

Could release of certain occurrence data result in increased harm to the Wildlife Species or its habitat?

No

Current status

COSEWIC: N/A

Year assessed: N/A

COSEWIC status history: n/a

Criteria: N/A

Reasons for designation: N/A

Status and reasons for designation:

Status: Endangered

Alpha-numeric code: A2be

Numeric code for change of status: N/A

Reasons for designation: Historically found in the inshore waters along much of the Pacific Coast of North America, this is the world's largest sea star and the only member of its genus. The species was stricken by Sea Star Wasting Disease following abnormally high ocean temperatures during 2014 to 2015. Extensive diving survey data suggest a loss of over 75% of the Canadian population, with no evidence of recovery since 2015. The species is largely gone from the U.S. portion of its range to the south and Alaskan subpopulations to the north have also declined precipitously, limiting the possibility of rescue. Given these recent declines, continuing disease outbreaks and little potential for rescue, this species is Endangered and at risk of extirpation in Canada.

Applicability of criteria

Criterion A (Decline in total number of mature individuals):

Meets Endangered, A2be. Estimated >50% decrease in total population size over last three generations where cause may not have ceased, is not fully understood, and may not be reversible (A2), as measured by indices appropriate to the taxon (b), due to a pathogen (e). Note, change in IAO is considered tentative, so subcriterion (c) not included here.

Criterion B (Small distribution range and decline or fluctuation):

Not applicable. EOO and IAO exceed thresholds.

Criterion C (Small and declining number of mature individuals):

Not applicable. Number of mature of individuals very likely much greater than 10,000. Insufficient data to determine population size, and no evidence of current decline.

Criterion D (Very small or restricted population):

Not applicable. Does not meet D1, but one location based on main threat (SSWD). A dramatic crash has been documented due to this threat, offering precedent for extreme events over short periods of time. May meet D2 Threatened.

Criterion E (Quantitative analysis):

Not applicable. Analysis not done.

COSEWIC history

The Committee on the Status of Endangered Wildlife in Canada (COSEWIC) was created in 1977 as a result of a recommendation at the Federal-Provincial Wildlife Conference held in 1976. It arose from the need for a single, official, scientifically sound, national listing of wildlife species at risk. In 1978, COSEWIC designated its first species and produced its first list of Canadian species at risk. Species designated at meetings of the full committee are added to the list. On June 5, 2003, the Species at Risk Act (SARA) was proclaimed. SARA establishes COSEWIC as an advisory body ensuring that species will continue to be assessed under a rigorous and independent scientific process.

COSEWIC mandate

The Committee on the Status of Endangered Wildlife in Canada (COSEWIC) assesses the national status of wild species, subspecies, varieties, or other designatable units that are considered to be at risk in Canada. Designations are made on native species for the following taxonomic groups: mammals, birds, reptiles, amphibians, fishes, arthropods, molluscs, vascular plants, mosses, and lichens.

COSEWIC membership

COSEWIC comprises members from each provincial and territorial government wildlife agency, four federal entities (Canadian Wildlife Service, Parks Canada Agency, Department of Fisheries and Oceans, and the Federal Biodiversity Information Partnership, chaired by the Canadian Museum of Nature), three non-government science members and the co-chairs of the species specialist subcommittees and the Aboriginal Traditional Knowledge subcommittee. The Committee meets to consider status reports on candidate species.

Definitions (2025)

Wildlife species

A species, subspecies, variety, or geographically or genetically distinct population of animal, plant or other organism, other than a bacterium or virus, that is wild by nature and is either native to Canada or has extended its range into Canada without human intervention and has been present in Canada for at least 50 years

Extinct (X)

A wildlife species that no longer exists

Extirpated (XT)

A wildlife species no longer existing in the wild in Canada, but occurring elsewhere

Endangered (E)

A wildlife species facing imminent extirpation or extinction

Threatened (T)

A wildlife species likely to become endangered if limiting factors are not reversed

Special concern (SC)*

A wildlife species that may become a threatened or an endangered species because of a combination of biological characteristics and identified threats

Not at risk (NAR)**

A wildlife species that has been evaluated and found to be not at risk of extinction given the current circumstances

Data deficient (DD)***

A category that applies when the available information is insufficient (a) to resolve a species’ eligibility for assessment or (b) to permit an assessment of the species’ risk of extinction

*

Formerly described as “Vulnerable” from 1990 to 1999, or “Rare” prior to 1990.

**

Formerly described as “Not In Any Category”, or “No Designation Required”

***

Formerly described as “Indeterminate” from 1994 to 1999 or “ISIBD” (insufficient scientific information on which to base a designation) prior to 1994. Definition of the (DD) category revised in 2006.

Wildlife species description and significance

Name and classification

Scientific name: Pycnopodia helianthoides (Brandt 1835)

English name: Sunflower Sea Star

French name: Solaster géant

No Indigenous names could be found for this species.

Classification: Phylum Echinodermata, Class Asteroidea, Order Forcipulatida, Family Asteriidae

Sunflower Sea Star is the only species in the genus Pycnopodia, moved from the genus Asterias when the former was erected in 1862. It is a member of the subfamily Pycnopodiinae, which includes only one other species, Lysastrosoma anthosticta (Mah 2021). Common names for Sunflower Sea Star include Rag Mop, Slime Star, Twenty-Rayed Starfish, Sunflower Star, and Sunflower Starfish (Gravem et al. 2021).

Morphological description

The dioecious Sunflower Sea Star (cover photograph) is the largest known sea star species (Mauzey et al. 1968; Herrlinger 1983; Shivji et al. 1983). After an average of 8 weeks spent in the water column as free-swimming, self-feeding, pelagic larvae, Sunflower Sea Star settle onto hard substrata and develop into juveniles with pentaradially symmetrical arms (Greer 1962; Hodin et al. 2021). Additional arms are added (by emerging, bifurcation, or budding) as body size increases, with up to 24 arms being documented when fully grown (Feder 1980). The maximum arm tip–to–arm tip diameter recorded to date is 130 cm (Mauzey et al. 1968). Adult colouration on the aboral surface is variable, ranging from red to orange, purple, or brown, with the central disc often being either lighter or darker than the arms, which in turn may be striped or mottled. The aboral surface is covered in short spines that vary in colour and prominence. The oral side is typically pale yellow or orange and covered with thousands of tube feet (Hodin et al. 2021). There is no known dimorphism between sexes.

Designatable units

There are no recognized subspecies or varieties of Sunflower Sea Star and no discrete populations have been identified. Preliminary genetic analyses of Sunflower Sea Star from California, Washington, British Columbia (B.C.), and Alaska show no large-scale geographic structure, with low F_st values (<0.05) among sampling sites, suggesting well-mixed populations (M. Dawson pers. comm. May 2023). Similarly, population genetic analyses of sea star species in the same family with the same broadcasting mode of reproduction and similar pelagic larval durations (for example, Ochre Sea Star [Pisaster ochraceus] and Mottled Sea Star [Evasterias troschelli]) show evidence for high gene flow and low genetic structure across populations spanning California to Alaska (Harley et al. 2006; Marko et al. 2010). Sunflower Sea Star is therefore assessed as a single designatable unit.

Special significance

Sunflower Sea Star is the largest known sea star and the only member of its genus and one of only two species in its subfamily, suggesting that it is an evolutionarily distinctive species. It is a generalist mesopredator that feeds on a wide range of commercially, ecologically, and culturally important species.

In habitats without top predators such as Sea Otter (Enhydra lutris), Sunflower Sea Star can become a keystone species because of its role in controlling herbivorous sea urchins (Burt et al. 2018). In some habitats where Sunflower Sea Star has disappeared due to Sea Star Wasting Disease (SSWD), urchin populations have increased dramatically, resulting in kelp declines and urchin barrens (Schultz et al. 2016; Burt et al. 2018; Harvell et al. 2019). This indicates that Sunflower Sea Star population loss may have a large cascading effect on ecosystems.

Aboriginal (Indigenous) knowledge

Aboriginal Traditional Knowledge (ATK) is relationship-based. It involves information on ecological relationships between humans and their environment, including characteristics of species, habitats, and locations. Laws and protocols for human relationships with the environment are passed on through teachings and stories, and Indigenous languages, and can be based on long-term observations. Place names provide information about harvesting areas, ecological processes, spiritual significance or the products of harvest. ATK can identify life history characteristics of a species or distinct differences between similar species.

Several First Nations (specifically the Haida, Heiltsuk, Kitasoo/Xai’xais, Nuxalk, and Wuikinuxv Nations) have been involved in efforts to monitor Sunflower Sea Star and have collected and shared data used in this report.

Cultural significance to Indigenous peoples

There is no species-specific ATK in the report. However, Sunflower Sea Star is important to Indigenous Peoples, who recognize the interrelationships of all species within the ecosystem.

Distribution

Global range

Sunflower Sea Star is endemic to the Pacific coast of North America. Its documented range extends from Bettles Bay (60.95°N,148.30°W) near Anchorage, Alaska, to the north, the Fox Islands near Nikolski (Aleutian Islands; 52.64°N, 169.13°W), Alaska, to the west, and Isla Natividad (27.84°N, 115.14°W), Baja California, Mexico, to the south (Figure 1A) (Gravem et al. 2021; Hamilton et al. 2021a). Sunflower Sea Star was previously present year-round within this geographic range.

The depth range of Sunflower Sea Star is from the intertidal to the continental shelf break (approximately 500 m), although most (see Habitat requirements) are found at depths of less than 75 m (Lambert 2000; Gravem et al. 2021; Hemery et al. 2021).

Since the outbreak of SSWD in 2013, most of the remaining Sunflower Sea Star individuals in shallow water are found in Alaska and B.C. (Gravem et al. 2021). No Sunflower Sea Star individuals have been recorded from Baja California to central California since 2015, and only a few isolated sightings have been reported from central California and Oregon since 2021 (Hamilton et al. 2021a; M. Miner pers. comm. no date). There have been more frequent sightings north of Oregon over the same period (iNaturalist no date). There are fewer data about Sunflower Sea Star beyond diving depths, but Harvell et al. (2019) reported a 96% decline in the Sunflower Sea Star biomass in offshore surveys that coincided with the onset of SSWD. From 2015 to 2017 (the last year considered), the Sunflower Sea Star biomass in trawls from 55 to 1,280 m in depth was virtually zero from southern California to northern Washington state.

Figure 1. (A) Global distribution of Sunflower Sea Star (reproduced from Gravem et al. 2021). The Canadian distribution is bounded by the square and shown at a finer scale in (B).

Canadian range

In Canada, Sunflower Sea Star has been found in marine environments along the entire coast of B.C., including Haida Gwaii and Vancouver Island (Figure 1B).

Population structure

There is no published information on the population structure of Sunflower Sea Star, although both its biology and unpublished genetic data are consistent with a well-mixed population across the extensive range (M. Dawson pers. comm. May 2023: see Dispersal and migration).

Extent of occurrence and area of occupancy

Extent of occurrence (EOO) and index of area of occupancy (IAO) were calculated by combining three types of sighting data: iNaturalist observations, dive surveys, and bycatch records from fisheries research surveys (see Sampling effort and methods and figures 2 and 3).

The EOO was calculated by constructing a minimum convex polygon around all recorded occurrences for Sunflower Sea Star in Canada from 1972 to 2021. This EOO for Sunflower Sea Star in Canada was estimated to be 219,452 km² (Figure 4), with no appreciable change from pre- to post-SSWD (Figure 4).

The IAO was generated by summing the total number of 2 × 2 km grid cells in Canada in which Sunflower Sea Star has been recorded from 1972 to 2021. The IAO was estimated at 7,652 km² (1,913 2 × 2 km grids). Post-SSWD, the IAO was not calculated (see Figure 4), because of differences in sampling effort across the range in the two periods.

 

Figure 2. Occurrence records (orange = presence; brown = absence) of Sunflower Sea Star in Canada, based on all observations available from 1972 to 2021 (see Table 1). Three hundred 2 × 2 km cells used to calculate the IAO had at least one survey both pre- and post-SSWD done within them (included as presences in this figure): 92 cells (31%) experienced a complete loss of Sunflower Sea Star; 6 cells (2%) posted a gain (that is, changing from absent pre-SSWD to present post-SSWD); and 202 cells (67%) showed no change in occupation.

Figure 3. Occurrence records (orange = presence; brown = absence) of Sunflower Sea Star in Canada. (A) Pre-SSWD (1972 to 2013); and (B) post-SSWD (2015 to 2021).

Figure 4. Extent of occurrence (EOO) of Sunflower Sea Star in Canada, based on 4,296 occurrence records from 1972 to 2021 from bycatch, iNaturalist, and dive survey data. The pale grey area indicates the limits of Canada’s Pacific EEZ.

Search effort

Information on the distribution of Sunflower Sea Star that generated occupancy estimates came from three primary sources: dive surveys, bycatch records from fisheries research surveys, and iNaturalist observations.

Dive survey data came from a variety of sources, including SCUBA diving surveys conducted by independent researchers, community scientists, and non-governmental organizations (see Sampling Efforts and Methods). These surveys were generally limited to depths of less than 30 m. All datasets (Table 1) targeted benthic invertebrate communities rather than Sunflower Sea Star specifically, but they all required the recording of Sunflower Sea Star when encountered. Most time series were short (2 to 5 years; Appendix 2) though a few extended to more than 10 years and straddled the mass mortality event, most notably the Reef Environmental Education Foundation (REEF) community science database, which included data from 2000 to 2021 from multiple sites. These surveys are considered to be highly reliable because species misidentification is unlikely (at least for adult Sunflower Sea Star) and because divers examine crevices and overhangs for the presence of cryptic species. These surveys provide a reliable record of presence/absence because an abundance of zero likely reflects a very low density such that Sunflower Sea Star could not be detected. These surveys can also generate indices of abundance because survey time or area is known.

* Indicates data obtained from publicly available IUCN Red List assessment dataset (Gravem et al. 2021; Hamilton et al. 2021b).

^a Multiple divers completing a survey at the same time and area/site are counted only once. The highest observed count was retained and a fixed effect of number of surveyors (with a score of 0 for observations derived from a single diver and a score of 1 for observations derived from multiple divers) was included in the model.

^b The mean count and highest presence scores were taken for each site in each year in which sampling occurred. Observations with clear mistakes in coordinates (for example, observations on land) or missing coordinates were removed.

^c Latitudes and longitudes were available only for the traps with Sunflower Sea Stars (that is, the presence-only data). Absence data (n = 10,797 for crabs and n = 19,388 for prawns) contained only information on the year of the survey, and they were used to estimate annual bycatch trends but were not included in any formal analysis.

^d The data were presence-absence and were therefore used only for the probability-of-occurrence analysis, not the abundance analysis.

^e The dataset originally contained 290 surveys but was trimmed to include only surveys with a measure of survey effort (that is, area surveyed) for the probability-of-occurrence model and surveys with a measure of both survey effort and depth for the abundance model.

^f The original dataset includes data from 1972 but was trimmed to start in 2001 for the times series analyses, when abundance was recorded consistently across surveys (that is, data from years in which more than 10% of the surveys recorded only presence/absence were removed to avoid biasing the results).

^g This dataset was removed from the abundance analysis because it was the only small dataset that used time instead of area as a metric of survey effort, and it therefore could not be included in the same model as the other small datasets. It was retained for the probability-of-occurrence analysis since the small datasets were combined with the large REEF and OW surveys, which also used time as a metric of effort.

^h This was the only dataset with consistent count surveys prior to 2000, but there were too few surveys in each year to allow for reliable model estimates in those early years. All the data were used for the spatial analyses, but the data were trimmed to the 140 post-1999 surveys for the sighting frequency analysis. Additionally, because the small datasets were modelled separately from the large datasets in the abundance analysis, this dataset had to be trimmed to the 77 post-2008 surveys, as there were too few pre-2009 observations to reliably estimate annual trends. Sighting frequency was consistently high in all years prior to 2000 and similar to the sighting frequencies in 2000 to 2020.

Fisheries and Oceans Canada (DFO) provided data primarily from non-targeted observations from 1977 to 2021 in areas inaccessible by diving (for example, Hecate Strait) and beyond SCUBA-diving depths (that is, beyond 30 m). These prawn and crab trap and bycatch records from groundfish longline data provide presence-absence data for these areas; however, reliability is considered low due to lack of rigorous control over counting and recording.

iNaturalist observation records were retrieved from the iNaturalist website (iNaturalist 2022) and are composed of “research-grade” observations (at least two thirds majority agreement on identification), with the bulk of observations in this dataset being made after the wasting event, around human-populated areas, and in shallow waters. These observations are considered reliable due to the ease of identification and the presence of photographic evidence; however, they are anecdotal, do not reveal absence, and cannot generate indices of abundance.

Potential unmeasured biases in estimates of distribution include the following:

1. the greater detectability of larger Sunflower Sea Star individuals: depleted, post-SSWD populations of Sunflower Sea Star were dominated by smaller recruits, which could have been more easily overlooked or misidentified by divers
2. greater awareness and interest among researchers and community scientists in Sunflower Sea Star post-SSWD years, which could have resulted in increased reporting of presences; and
3. more limited sampling in deep water than at diveable depths, although the abundance of Sunflower Sea Star in deep water was considerably lower than in shallow water before the mass mortality (see Habitat requirements below)

Habitat

Habitat requirements

Sunflower Sea Star is a habitat generalist and can be found in a variety of substrates and habitat types, including mud, sand, gravel, kelp forests, and rocky reefs (Mauzey et al. 1968; Morris et al. 1980; Britton-Simmons et al. 2012; Hemery et al. 2016; Gravem et al. 2021). This is supported by the REEF surveys, which reported sightings of Sunflower Sea Star across almost all habitat categories, including hard substrates such as rock/shale reefs, artificial reefs, cobblestone/boulder fields, rocky walls, and biogenic or soft substrates such as kelp, eelgrass, surfgrass, mud, sand, and silt (Figure 5).

Sunflower Sea Star occurs from the low intertidal zone down to a depth of at least 455 m (Gravem et al. 2021). A single study has examined the depth distribution of Sunflower Sea Star in Canada; in Haida Gwaii, there was no association between Sunflower Sea Star abundance and depth, but diving surveys were conducted to only 10 m below chart datum (Lee et al. 2016). Similarly, two American studies (northern Channel Islands, California, diving surveys at 5 to 16 m, Bonaviri et al. 2017; and San Juan Islands, northern Washington state, Remotely Operated Vehicle [ROV] surveys to 131 m, Britton-Simmons et al. 2012) also reported no association between Sunflower Sea Star density and depth. In contrast, deep-water (55 to 1,280 m) bottom trawls by the National Marine Fisheries Service (NOAA) along the coasts of California, Oregon, and Washington indicated that, prior to SSWD, the sea star’s biomass was more than 3 orders of magnitude lower over that wide range of “deep” water than in shallower, diver-based surveys (Harvell et al. 2019).

A species distribution model (MaxEnt) of Sunflower Sea Star presence on the continental shelf from southern Washington state to northern California, U.S.A., derived from groundfish bottom trawl ROV data, suggested a higher probability of sighting at depths of less than 75 m than at deeper depths (up to 550 m) (Hemery et al. 2021). A second MaxEnt model by Hamilton et al. (2021a) that combined data from Britton-Simmons et al. (2012), Bonaviri et al. (2017) and Hemery et al. (2021) predicted an exponential decline in the predicted probability of Sunflower Sea Star sighting with increasing depth, with a probability nearing zero at a depth of 300 m.

As larvae, Sunflower Sea Star individuals live in the water column and feed on phytoplankton (Greer 1962; Hodin et al. 2021). Once metamorphosis has occurred, the sea stars settle on the bottom, mainly in sheltered areas (Shivji et al. 1983). Sewell and Watson (1993) found that Sunflower Sea Star recruits surveyed in Nootka Sound, on the west coast of Vancouver Island, occurred mainly on invasive Wireweed (Sargassum muticum). Shivji et al. (1983) also found that juvenile Sunflower Sea Star individuals in Barkley Sound, also on the west coast of Vancouver Island, occurred mostly on kelp, but kelp is not a strict requirement for their life cycle because some juveniles (approximately 20%) were also present on muddy substrates. There is no evidence to suggest that Sunflower Sea Star has specific habitat requirements as adults (Britton-Simmons et al. 2012; Hemery et al. 2021).

Figure 5. (A) Mean probability of sighting and (B) mean abundance scores of Sunflower Sea Star across different habitat types, derived from REEF survey records from 2000 to 2021 in Canadian waters. The sizes of the points are proportional to the number of surveys carried out in each habitat type. “Probability of occurrence” is the proportion of surveys in which Sunflower Sea Star individuals were recorded; “mean abundance score” is the average abundance index (0 to 4) recorded across surveys within a habitat type. Note that habitat type refers to the habitat in which most of the dive time was spent during a survey.

Figure 6. Annual index of population abundance for Sunflower Sea Star in Canada over the past two decades (2000 to 2021), based on the REEF Invertebrate and Algae Monitoring Program (n = 4,711 surveys) and Ocean Wise (OW) Pacific Marine Life Surveys Database (n = 2,281 surveys), DFO-Abalone data (n = 994 observations), DFO-Multispecies data (n = 521 observations), and the remaining small datasets from the IUCN assessment (n = 617 surveys; total n = 9,124 surveys). Blue points and lines represent the annual abundance estimate from each dataset (that is, the 𝛼_𝑡 terms in the ordered logistic regression, defined in Appendix 2), scaled to a standardized mean survey time of 49.12 minutes, while the dashed green line and ribbon show the estimated true population state (that is, the index of abundance without observation error) and 95% credible interval. The number of observations included each year across all datasets is shown above the x-axis. No linear rate of decline is shown, as the decline is primarily restricted to two years. Note that a scaling term has been applied to the year-to-year estimates of all datasets except REEF for visualization (see Appendix 2 for details and Appendix 4 for unscaled dataset-specific time series with 95% credible intervals).

Figure 7. Annual probability of sighting for Sunflower Sea Star in Canada over the past two decades (2000 to 2021), scaled to a standardized mean survey area of 51.9 m², based on the REEF Invertebrate and Algae Monitoring Program (n = 4,711 surveys), OW Pacific Marine Life Surveys Database (n = 2,281 surveys), DFO-Abalone data (n = 994 data points), DFO-Multispecies data (n = 521 data points), DFO-Benthic Habitat Mapping data (n = 1,150 data points) and all other available datasets (n = 825 surveys; total observations = 10,482 surveys). Blue points and line represent the annual probability-of-occurrence estimate for all data (that is, the 𝛼_𝑡 term defined in Appendix 2), while the dashed green line and ribbon show the estimated underlying population state and 95% credible interval. The number of observations included each year across all datasets is shown above the x-axis. No linear rate of decline is shown, as the decline is primarily restricted to two consecutive years.

Water temperature is an important if complex environmental determinant of the spatial distribution of Sunflower Sea Star. From 2006 to 2011, prior to the mass mortality, Sunflower Sea Star was abundant at sites across the northern Channel Islands, California, where the mean annual in situ water temperature (that is, at the depth of the sea star surveys) was below 14 °C, and was either absent or occurred at lower density at warmer sites (Bonaviri et al. 2017). Water temperature was also the most important parameter explaining the pre-SSWD distribution of Sunflower Sea Star across its range in the United States (including Alaska), with presence associated most strongly with temperatures between 9.5 and 10.5 °C (Hemery et al. 2016). Hemery et al. (2016) also found an association between Sunflower Sea Star presence and salinities of 33.0 to 33.4 PSU (practical salinity units) prior to SSWD. After the mortality event (associated with the elevated 2013 to 2014 Pacific Ocean temperatures known as the “blob”), predictors from a separate MaxEnt model that compared pre- and post-SSWD differed (Hamilton et al. 2021a). After the event, the importance of temperature as a determinant of Sunflower Sea Star occurrence increased fourfold, with the modelled probability of sighting declining steadily between approximately 2 °C and 25 °C rather than showing a unimodal peak at 16 °C. In contrast, the importance of both depth and salinity as predictors declined compared to pre-event modelling: although the probability of sighting still fell with depth and became zero near 300 m, its linear association with salinity shifted from positive prior to SSWD to negative after SSWD across salinities ranging from 25 to 33.8 PSU. Note that the entire distribution range of Sunflower Sea Star experienced large positive anomalies (+2.0 to +3.5 °C) in sea surface and subsurface temperatures between 2014 and 2016 (Freeland and Ross 2019), coinciding with the onset of SSWD (Harvell et al. 2019).

Habitat trends

Historical changes in habitat availability for Sunflower Sea Star in Canada are largely unknown. However, because Sunflower Sea Star is a habitat generalist (Britton-Simmons et al. 2012; Hemery et al. 2021) that can readily and rapidly move between habitat types (Hodin et al. 2021), it seems likely that the availability of physical habitat has not changed substantially over the last three generations (see Life cycle and reproduction below). Nevertheless, the strong association between Sunflower Sea Star presence and abundance post-SSWD with low temperatures (Bonaviri et al. 2017; Hamilton et al. 2021a; Hemery et al. 2021) and between SSWD and anomalously high temperatures (Harvell et al. 2019; Aalto et al. 2020; Hamilton et al. 2021a) point to the critical role of temperature in determining current habitat suitability. Given the chronic warming of B.C. waters over the past decade (bottom temperature increases averaging +0.6 °C per decade [95% quantile range: -0.2 to 1.8 °C]) with higher rates of warming where sea temperatures are already warmer (for example, +1.3 °C in areas shallower than 50 m) (English et al. 2022), as well as the increased frequency of marine heatwaves (Smale et al. 2019), the habitat quality for Sunflower Sea Star in B.C. has likely declined in the past decade and will continue to do so in the future. That said (and SSWD notwithstanding) this is not known with certainty, and it is not possible to produce quantitative estimates. Importantly, there is no evidence that Sunflower Sea Star is moving deeper to cooler depths, likely because there are other habitat constraints such as food availability.

Biology

Life cycle and reproduction

Sunflower Sea Star is a dioecious (that is, separate sexes) broadcast spawner (Morris et al. 1980). Fertilization occurs in the water column. While in some species of echinoderms, fertilization rates depend on local mating abundance (for example, spawning synchrony, the number of males present, and close proximity) leading to selection for aggregation, there is at present no clear evidence to suggest that Sunflower Sea Star aggregates to spawn. Whether Sunflower Sea Star is subject to an Allee effect (lower-than-expected population growth rates at low density) due to lower overall abundance is unknown.

Feder and Christensen (1966) hypothesized that Sunflower Sea Star spawns once a year, in May and June, based on surveys around Vancouver Island. This is likely to be an underestimate of the length of the breeding season: Hodin et al. (2021) found mature and viable oocytes in Sunflower Sea Star females around San Juan Island, Washington state, from November to May, while males had testes containing mature sperm throughout the year. Reproductive females were found in July in southeastern Alaska (Hodin et al. 2021). There may therefore be a temperature-related gradient of reproductive seasonality from south to north along the Canadian coast.

After fertilization in the water column, Sunflower Sea Star proceeds through embryogenesis, bipinnaria, and brachiolaria larval stages before settling on the substrate after a 7 to 22 week pelagic period (Greer 1962; Strathmann 1978; Hodin et al. 2021). There is evidence for shorter pelagic larval durations in Alaska than in Washington state—an average of 8 weeks versus 11 weeks, respectively (Hodin et al. 2021), which could reduce the potential for dispersal in more northern regions.

Sea stars have indeterminate growth, and their growth rate and maximum size depend on environmental factors such as temperature and food supply, making size an imprecise estimate of age. In the case of Sunflower Sea Star, estimating growth is also complicated by the inability to tag individuals for long-term monitoring and the difficulty in rearing this species in captivity. Based on anecdotal evidence and information for other sea star and urchin species, Gravem et al. (2021) suggested that Sunflower Sea Star growth is sigmoidal, reaching 3 to 8 cm in the first year, after which the growth rate slows towards an asymptote. Mid-sized (30 to 60 cm in diameter) Sunflower Sea Star individuals grow 2 to 3 cm per year (Gravem et al. 2021). By following the growth of known cohorts of Sunflower Sea Star in Washington state, Gravem confirmed the sigmoidal growth pattern but noted that growth appears to be faster than initially estimated, with sea stars reaching 30 cm in approximately 4 years (S. Gravem pers. comm. 2023).

The age at maturity of Sunflower Sea Star is not yet known. Gravem et al. (2021) estimated it to be at least 5 years in their IUCN Red List entry, given the overlap in habitat and diet and the similarity of reproduction strategy between it and Ochre Sea Star (Menge 1975). More recently, Gravem examined gonad sizes of a small number of Sunflower Sea Star specimens collected in March in Washington state. Gonad size increases rapidly with body size in sea stars that are larger than 34 cm in diameter, indicating sexual maturity at a size reached in about 4 years (S. Gravem pers. comm. 2023).

Age-specific survival and reproduction rates for Sunflower Sea Star are not known. Nevertheless, Gravem et al. (2021) used anecdotal growth estimates and two growth models (von Bertalanffy and Richards) to estimate lifespan and generation time. Their estimates of lifespan, defined as the time to reach 100 cm in diameter, ranged from 48 to 68 years depending on the model used, while lifespan defined as the time to reach 50 cm in diameter (a more usual maximum size in natural populations) ranged from 11 to 14.5 years.

Based on these lifespan estimates across two models, Gravem et al. (2021) estimated generation time for Sunflower Sea Star to be 8.5 to 10.25 years using “middlemost estimates” where lifespan reflects growth to 50 cm in diameter and 27 to 37 years where lifespan reflects growth to 100 cm. Therefore, Gravem et al. (2021) suggested that the duration of three generations for Sunflower Sea Star is 81 to 111 years (61.5 to 195 years using minimum and maximum lifespans estimated by growth models for growth to 100 cm). Because the estimates of growth underpinning these calculations are mainly from captive Sunflower Sea Star specimens, it is not possible to estimate the uncertainty around these figures. Moreover, based on the faster growth rates documented by Gravem (S. Gravem unpublished data, pers. comm. 2023), published estimates of lifespan and generation time are likely to be inflated, with revised estimates of generation time closer to 10 to 20 years, such that three generations would be 30 to 60 years (S. Gravem pers. comm. June 2024).

Physiology and adaptability

The main physiological requirements of Sunflower Sea Star are described under Habitat requirements. Briefly, depth, temperature, and salinity were the three main determinants of Sunflower Sea Star presence prior to the mortality event. Along the U.S. west coast, the species was normally associated with shallow depths (<75 m), temperatures of 9.5 to 10.5 °C, and salinities of 33 to 33.4 PSU (Hemery et al. 2021). Since the mortality event, correlates of presence have changed, with a marked increase in the importance of temperature (Hamilton et al. 2021a). The highest probabilities of Sunflower Sea Star presence are now associated with low temperatures, low salinities and, as previously, shallow depths (Hamilton et al. 2021a). Sunflower Sea Star is one of the most active sea stars and has been recorded crawling at speeds of 1 to 2 m/min, allowing them to follow tide movements and escape predators or unsuitable conditions (Hemery et al. 2016; Hodin et al. 2021).

There is limited literature on the level of environmental change tolerated by Sunflower Sea Star. Hemery et al. (2016) hypothesized that Sunflower Sea Star would migrate to deeper refuges if temperatures were beyond its optimal temperature range, but there is no documented evidence of such temperature-related movements. Harvell et al. (2019) found that Sunflower Sea Star individuals in shallow, nearshore waters were more susceptible to SSWD when water temperatures were warmer, which is consistent with the idea that warmer-than-normal temperatures impose higher metabolic demands and metabolic stress on sea stars generally (Fly et al. 2012).

Hodin et al. (2021) have recently initiated a captive-rearing program of Sunflower Sea Star at Friday Harbor Laboratories, Washington state. They have now successfully reared Sunflower Sea Star specimens from egg to one-year post-settlement juvenile. The goal is to eventually complete the full egg-to-egg life cycle in captivity and explore the potential for the introduction of captive-bred animals into the wild. The program is also raising Sunflower Sea Star larvae in warmer water temperatures to examine the potential for acclimation/adaptation (Ma and Taguchi 2021).

Dispersal and migration

Sunflower Sea Star disperse during their pelagic larval phase. The relatively long pelagic larval duration offers the potential for long-distance dispersal via water currents (Shanks 2009). However, the extent to which a long pelagic larval duration translates into long-distance dispersal depends on several factors, including larval behaviour (Leis 2021). There is currently no information on the behaviour of Sunflower Sea Star larvae that could inform this.

High dispersal of the larval stage of Sunflower Sea Star could allow propagule connectivity among post-SSWD refuges (Gravem et al. 2021). In Canada, the population is not likely to be considered severely fragmented as per COSEWIC’s definition because of its habitat generalist nature, pelagic stage, and genetic patterns (Gravem et al. 2021; M. Dawson pers. comm. May 2023).

Interspecific interactions

During the larval stage, Sunflower Sea Star is planktivorous (Strathmann 1978; Morris et al. 1980; see Life cycle and reproduction). As a juvenile, Sunflower Sea Star becomes a generalist carnivore at a very small size (approximately 1 cm in diameter), successfully opening small Manila Clam (Venerupis philippinarum) and Pacific Oyster (Magallana gigas; Hodin et al. 2021). As an adult, Sunflower Sea Star has a broad diet that includes scavenging and predation on polychaetes, bivalves, gastropods, crabs, sea urchins, sea cucumbers, and other sea stars (Feder and Christensen 1966; Moitoza and Phillips 1979; Morris et al. 1980; Duggins 1983; Herrlinger 1983; Shivji et al. 1983; Brewer and Konar 2005; Lee et al. 2016; reviewed in Gravem et al. 2021). Sunflower Sea Star can swallow small prey whole (for example, sea urchins and bivalves) and prey on larger organisms (for example, clams) by everting its stomach (Feder and Christensen 1966; Mauzey et al. 1968). The diet of the Sunflower Sea Star includes commercially important species: Butter Clam (Saxidomus gigantea), California and Common Blue Mussel (Mytilus californianus and M. trossulus), Red Rock Crab (Cancer productus), Purple, Green, and Red Sea Urchin (Strongylocentrotus purpuratus, S. droebachiensis, and Mesocentrotus franciscanus), California Sea Cucumber (Apostichopus californicus) (Moitoza and Phillips 1979; Shivji et al. 1983), and Northern Abalone (Haliotis kamtschatkana) (Lee et al. 2016), a species listed as Endangered under SARA (Species at Risk Act 2011). As a predator of small and medium-sized urchins, Sunflower Sea Star plays an important role in the persistence of kelp, complementary to that of large-urchin–eating Sea Otter (Burt et al. 2018; Galloway et al. 2023). At least in Alaska, Sunflower Sea Star contributes as much to clam mortality as Sea Otter, and its digging to excavate buried clams likely influences soft substrate dynamics (Traiger et al. 2016).

There are only a few marine species that have been observed preying on adult Sunflower Sea Star individuals in Canada, including Morning Sun Star (Solaster dawsoni) and Red King Crab (Paralithodes camtschaticus) (Gravem et al. 2021). However, during its gamete and larval stages, Sunflower Sea Star is free-floating in the water column, leaving it vulnerable to planktonic predators and benthic filter-feeders.

Population sizes and trends

Sampling effort and methods

Data sources

Sunflower Sea Star was recently assessed by the International Union for the Conservation of Nature (IUCN) Red List as Critically Endangered across its global range, including Canada (Gravem et al. 2021). The IUCN assessors compiled a dataset from 31 sources across the west coast of North America and have since published an updated version of their analyses and made the dataset publicly available (Hamilton et al. 2021a,b). Ten of these sources contained data from the Canadian range and were retained for this COSEWIC status assessment, including data collected by First Nations (the Central Coast Indigenous Resource Alliance [CCIRA] and Gwaii Haanas datasets), academics (the Simon Fraser—Lee, Simon Fraser—Salomon, Simon Fraser—Schultz, and VIU datasets), governmental and non-governmental scientists (the Gwaii Haanas and Hakai datasets), and through community science initiatives (the REEF, Ocean Wise [OW] Pacific Marine Life Surveys, and MARINe datasets; Table 1). The IUCN compilation was supplemented with additional academic surveys (Simon Fraser—Côté in Table 1), dive surveys conducted by DFO, and datasets on Sunflower Sea Star bycatch from DFO research trap and trawl data on crabs, prawns, and groundfish.

All dive surveys followed either a belt transect technique, in which divers record all Sunflower Sea Star individuals observed in a predetermined area, which varied between sources, or a transect-free roving technique, standardized by the length of time spent recording, which also varied between sources. Most surveys were intended to census multiple species rather than Sunflower Sea Star exclusively, but all surveys used in this report included standardized instructions to record any Sunflower Sea Star individuals observed (that is, zeros included in these datasets represent the absence of Sunflower Sea Star on surveys, rather than a lack of data). The REEF and OW datasets reported counts in categories (Appendix 3); all other datasets reported actual counts. In total, 9,124 SCUBA surveys were used to assess the abundance trend of Sunflower Sea Star in Canada, and 10,482 surveys were used to assess the probability-of-sighting trend (“prob[sighting]” in the technical summary, also sometimes called “probability of occurrence” or “detection probability”). The probability of sighting is a complementary indicator based on presence/absence that reveals both numerical and spatial aspects of abundance (for example, Schultz et al. 2016; Harvey et al. 2018).

The crab, prawn, and groundfish tow and trap data from DFO do not specifically target sea stars, but sea star bycatch is instead opportunistically recorded in the process of sampling for the commercially valuable target species. Therefore, the absence of any recorded individuals during a tow or in a trap may mean either that there was a true absence or that bycatch was not recorded. As such, these data are not included in the main analyses. The raw trends are nevertheless presented separately because they provide a comparatively long time series of sightings extending well before and after the onset of SSWD as well as a glimpse of the status of Sunflower Sea Star at deeper depths (>30 m).

Analyses

See Appendix 3 for a detailed explanation of the methodology used in these analyses and its advantages over methods used previously. Appendix 5 offers additional tests of assumptions.

To determine the trajectory of the Sunflower Sea Star population in Canada over the last two decades (that is, the timespan of available data), two time series analyses were constructed: one modelling the change in abundance over time (using count per standardized survey as an index of abundance and modelling these counts as either ordinal logistic [for binned data] or a negative binomial [for raw counts]), and the other modelling the change in probability of sighting on a survey over time (that is, whether Sunflower Sea Star individuals were observed at all on a given survey, modelled as a standard logistic regression). These analyses are an expansion of those presented in Gravem et al. (2021) and Hamilton et al. (2021a), which aggregated the data into pre-, mid-, or post-crash periods and conducted frequentist analyses on these groups rather than individual years. By disaggregating the data from periods into years, estimates of long-term trends and interannual variability can be obtained in addition to the estimate of the magnitude of the crash. If the population was relatively stable prior to the onset of SSWD, the magnitude of the crash can then be determined by comparing the mean index of abundance/sighting probability across all pre-crash years with the mean index of abundance/sighting probability across the post-crash years.

Both sets of models were run in a Bayesian framework using the program Stan (Stan Development Team 2022, 2023) implemented in R (R Core Team 2021). A Bayesian approach was chosen because it allowed for greater flexibility in developing the model and allowed for the calculation of quantities of interest (for example, the percent decline from pre- to post-crash years) and their uncertainty directly from the posterior distributions (Krushke 2021).

In each dataset, if multiple surveys were conducted at the same site on the same day, only the greatest recorded abundance was used in the analyses to avoid pseudo-replication. This approach was used because (1) there were relatively few cases where this occurred (resulting in difficulties in accounting for multiple observations directly in the model structure); (2) both logistic and ordinal logistic regressions require counts as integer responses (for example, counts or scores, making it inappropriate to use the average value when it is a non-integer, unless model weights are applied in the logistic regression); and (3) it is more likely for a diver to miss an animal on a survey than it is to overcount (making the higher estimate of abundance a better estimate of the true population). A term indicating whether multiple observations were collapsed into the single observation used in the survey was included as a fixed effect in the model to account for potential bias towards higher counts on dives surveyed by multiple divers. Most datasets did not include information on habitat and substrate type; those that did often had considerable within-site variability that was difficult to capture in the models. However, an analysis of a California Department of Fish and Wildlife and Marine Applied Research and Exploration dataset by Gravem et al. (2021) showed that Sunflower Sea Star inhabits a wide variety of substrates and that variation in abundance was much greater across depths (17 to 93 m) and latitudes (the whole California coast) than among habitats. Consequently, habitat was not included as a covariate in the models.

For the abundance analysis, a multivariate autoregressive state-space (MARSS) model was run. This modelling framework allows for an independent time series analysis of multiple datasets, while estimating an integrated underlying “true” population trend across all datasets after accounting for observation error in each time series. While there is likely spatial variation in the population dynamics, Sunflower Sea Star is assessed as a single designatable unit composed of a single population and, therefore, the population trend was not modelled by region.

Four datasets were large enough (that is, contained more than 50 surveys per survey year) to be treated as independent datasets in the MARSS models: the REEF Invertebrate and Algae Monitoring Program, Ocean Wise (OW) Pacific Marine Life Surveys Database, DFO Abalone Dive Surveys, and DFO Multispecies Dive Surveys (Table 1). The remaining datasets were too small to allow for reasonable estimates of annual abundances if modelled separately; instead, these smaller datasets were grouped together into a fifth “dataset,” while adding a random effect of data source to this time series to allow for differences between individual datasets. For all datasets, some measure of survey effort was required to standardize the analyses: for the REEF and OW datasets, this was the time spent on the roving survey; for the DFO datasets, this was the area surveyed. The majority of small datasets also used the area surveyed as a metric of survey effort; consequently, any surveys within these datasets that did not record the area were removed from this section of the analysis, as there were too few of these surveys to model independently (see Table 1). Unscaled, dataset-specific trends in abundance are shown with error estimates (that is, 95% credible intervals (“CI” in the Technical Summary and below) in Appendix 4.

For the probability-of-sighting analysis, all five datasets mentioned above were used, in addition to the DFO benthic habitat mapping surveys, which recorded only presence/absence (Table 1). A univariate version (that is, with only a single observation model) of the state-space model was run due to difficulties scaling between datasets in the logit space. Therefore, all surveys were grouped into a single dataset, and source-specific biases were accounted for with a random effect in the model. A logistic regression was used on the combined data to model the probability of detecting individuals on a survey, and a state-space component was used to estimate both observation error and process noise. Many of the smaller datasets from the IUCN assessment produced substantially higher probabilities of sighting than non-targeted surveys like REEF and OW. This was accounted for in the model with a random effect of source. Because Sunflower Sea Star is assessed as a single designatable unit, the analysis is not separated by region; however, site-specific differences in abundance are accounted for with a random effect in the model.

Population size

Total population size is challenging to estimate for marine species, particularly given the roving dive technique often used to survey Sunflower Sea Star. While most surveyors recorded total dive times, fewer recorded the total area surveyed. Using a subset of the available survey data that did contain area estimates, Gravem et al. (2021) estimated the total population size of Sunflower Sea Star in Canada excluding the Salish Sea (that is, the Strait of Georgia, Strait of Juan de Fuca, Puget Sound, and connecting channels and adjoining waterways) to have been 209,310,484 individuals prior to the crash and 21,054,457 individuals afterwards (a 90% decline). They estimated the population size of Sunflower Sea Star in the Salish Sea (encompassing both Canadian and U.S. territory) to have been 402,098,568 individuals prior to the crash and 32,764,937 afterwards (a 92% decline). These estimates were calculated by taking the mean density observed in each region across all pre-SSWD and all post-SSWD years and multiplying them by the total seafloor area less than 25 m in depth as estimated from bathymetry maps. No estimates of uncertainty around the declines were presented.

However, the standard error around each of the density estimates used to calculate these population sizes were extremely large, making it difficult to accurately estimate the magnitude of the decline. While the analysis presented in this report makes a reasonable estimation of the uncertainty around the percent decline possible, scaling up to population-size estimates remains challenging, as there is even greater uncertainty about the total available habitat for Sunflower Sea Star. Instead, a standardized index of abundance (that is, the number of individuals observed per dive, standardized by dive length and accounting for site- and year-level variation) and probability of sighting (that is, if any individuals were observed on a dive) were used as indices of abundance and abundance change appropriate to this taxon.

Fluctuations and trends

In both analyses, the general population trend was not different from zero prior to and following the crash period in Canada ($u_{\text{pre-wasting}}$  = 0.03, 95% CI: -0.10, 0.14 and $u_{\text{post-wasting}}$  = -0.10, 95% CI: -0.28, 0.10 for the abundance model, and ($u_{\text{pre-wasting}}$ = 0.04, 95% CI: -0.10, 0.17 and $u_{\text{post-wasting}}$ = -0.07, 95% CI: -0.27, 0.15, for the probability-of-sighting model), where $u$ is the estimate of average annual change. The mean pre- and post-crash values and the percent decline between them were calculated once for each iteration of the model, and then the median percent decline (and 2.5% and 97.5% quantiles) was drawn from the resulting distribution to determine the estimated effects and the 95% CI.

Abundance

In Canada, the estimated decline in the index of abundance of Sunflower Sea Star between 2000 and 2021 was 82.1% (95% CI: 76.0%, 87.2%), with most of the modelled decline occurring in a single year (2014; Figure 6). This is a conservative estimate of the true decline, because counts were estimated by multiplying the categorical probabilities from the ordinal logistic regression by the lowest abundance value in each abundance category (Appendix 3). The less conservative approach of using the highest abundance value in each category increases the magnitude of the abundance (that is, y-axis) but does not greatly affect the estimate of decline (83.7%; 95% CI: 77.8%, 88.5%). Prior to 2014, the population appeared relatively stable (Figure 6) and annual fluctuations in each dataset appeared to be primarily the result of observation error rather than true changes in the population state (SD around process noise = 0.15, SDs around dataset-specific observation errors = 0.49, 0.63, 0.30, 0.58, 0.81; see Dennis et al. 2006 for a description of how observation error and population trends are distinguished in multivariate state-space models). Thus, although these data do not span three generations (that is, 30 to 60 years), the stability of the population in the first 13 years of the dataset improves confidence that the crash initiated by SSWD has caused the population size to fall well below historical levels (from approximately 4 down to 0.77 Sunflower Sea Star per survey, on average, since 2015 [Figure 6]). While there are some small individuals, and therefore some recruitment, there is no sign of recovery in the dataset, and there is anecdotal evidence of continued outbreaks of disease (A. Gehman pers. obs. Sept. 2024).

Probability of sighting and change in IAO

In Canada, the estimated decline in the probability of sighting of Sunflower Sea Star on dives between 2000 and 2021 was 68.5% (95% CI: 53.9, 79.5%), with the majority of the decline occurring in the two years following the onset of SSWD (that is, 2014 and 2015; Figure 7). Prior to 2013, Sunflower Sea Star was sighted, on average, in three of every four surveys; after SSWD, it was recorded in one of every five surveys.

Three hundred of the 2 × 2 km cells used to calculate the IAO had at least one survey done within them both pre- and post-SSWD. Of these, 92 cells (31%) experienced the complete loss of Sunflower Sea Star; 2% (6 cells) posted a gain (that is, from absent pre-SSWD to present post-SSWD); the remaining 202 cells (67%) showed no change in occupation (whether absence or presence). We consider this as an index of change in IAO. An overview of recorded presences and absences are presented in Figure 2 and Figure 3.

Bycatch in DFO crab, prawn and groundfish surveys

Because of the opportunistic nature and low reliability (that is, it is not certain whether a lack of Sunflower Sea Star data on a given tow or in a given trap means an absence or that a presence was simply not recorded by an observer) of the bycatch data from DFO crab, prawn, and groundfish datasets, mean data trends are presented without attempting to estimate uncertainty. Crab traps (Figure 8A) were set at depths ranging from 10.4 m above to 260 m below chart datum; prawn traps (Figure 8B) ranged in depths from 0 to 130 m below chart datum; and groundfish tows (Figure 8 C) ranged from 20 to 268.8 m below chart datum. All three datasets show a drop in the number of Sunflower Sea Star recorded as bycatch per tow or trap between 2014 and 2015 (Figure 8), which is consistent with the decline in Sunflower Sea Star biomass noted in deep offshore trawl data (n = 8,968 tows) compiled by the National Marine Fisheries Service off Washington, Oregon, and California between 2004 and 2016 (Harvell et al. 2019).

Figure 8. Annual estimates of the mean number of individuals (A and C) or total mass of Sunflower Sea Star caught as bycatch per unit of effort (that is, number of sea stars per trap per hour for crabs and prawns and per hook per hour for groundfish) over the past two decades, based on DFO research surveys for (A) crabs, (B) prawns, and (C) groundfish. Estimates may not reflect true population trends as bycatch was likely not always recorded, and the presence of part of an individual was recorded as a full individual. The number of observations (that is, traps for crabs and prawns, or longlines for groundfish) included each year is shown at the bottom of each panel.

Rescue effect

The southern Sunflower Sea Star populations from Mexico to Washington state declined more than 92% after SSWD (Gravem et al. 2021; Hamilton et al. 2021a). Both shallow- and deep-water populations were severely affected, and there has been no evidence of recovery since 2017; therefore, a rescue of the Canadian population by larval dispersal from the southern portion of the range is unlikely (Harvell et al. 2019; Gravem et al. 2021; Hamilton et al. 2021a). The change in probabilities of sighting of Sunflower Sea Star on Canada’s Pacific Coast (-60% based on the trend analysis) and across regions from the Salish Sea northward (<-69%, Hamilton et al. 2021a) were less negative than the southern range estimates. The population in southeast Alaska declined in probability of sighting by 20.8% (Hamilton et al. 2021a); this suggests that Sunflower Sea Star is still relatively broadly distributed in parts of Alaska. However, abundance is reported to have declined by 96%, which limits the potential for rescue by larval dispersal. Alaskan figures are based on relatively few surveys and so uncertainty is high for some areas (Hamilton et al. 2021a). The correlation between density and fertilization rates and the density threshold that could trigger an Allee effect via low fertilization are unknown for Sunflower Sea Star (Lundquist and Botsford 2004), making predictions about rescue from Alaska difficult.

Threats and limiting factors

Threats

The nature, scope, and severity of the threats to Sunflower Sea Star are described in the threats calculator in Appendix 1, following the IUCN-CMP (International Union for the Conservation of Nature–Conservation Measures Partnership) unified threats classification system. Impacts for each of 11 main categories of threats and their subcategories were assessed, based on the scope (proportion of population exposed to the threat over the next 10-year period), severity (predicted population decline within the scope during the next 10 years or 3 generations, whichever is longer, up to approximately 100 years), and timing (from ceased to continuing) of each threat. The overall threat impact was calculated by taking into account the separate impacts of all threat categories and was adjusted by the species experts participating in the threats evaluation.

The overall threat impact for Sunflower Sea Star is considered to be High–Medium, corresponding to an anticipated further moderate decline (3 to 70%) over the next three generations. This assessment should be interpreted with caution, as it is in part based on subjective information such as expert opinion.

Category 8: Invasive and other problematic species, genes and diseases

8.6 Diseases of unknown cause (threat impact: very high–medium)

Episodic, localized die-offs of various species of sea stars owing to disease have been reported from the northeast Pacific since the late 1970s (Dungan et al. 1982; Bates et al. 2009; Hewson et al. 2019). However, the SSWD-induced mortality that began in 2013 on the west coast of North America is viewed as the largest known marine wildlife epizootic to date because of its geographic extent and the number of sea star species (approximately 20) affected (Hewson et al. 2014; Harvell et al. 2019). Sunflower Sea Star experienced the largest documented declines of any of the affected species (Montecino-Latorre et al. 2016; Schultz et al. 2016; Harvell et al. 2019; Gravem et al. 2021).

Sea star wasting disease first presents as behavioural changes, including lethargy and arm curling, followed by epidermal lesions and necrosis, body deflation, loss of limbs and gonads, and finally death and rapid decay (Hewson et al. 2014; Schultz et al. 2016; Harvell et al. 2019). The disease progresses from the onset of clinical signs to death in only a few days (Miner et al. 2018).

The causative agent of SSWD remains unknown, with several competing hypotheses and active research. Early experiments suggested that SSWD was caused by a densovirus (Hewson et al. 2014, 2018), but further analysis of the original work shows no association between the putative densovirus and SSWD (Hewson et al. 2024). Several laboratory and field studies have found dysbiosis associated with signs of disease (Lloyd and Pespeni 2018; Aquino et al. 2021; McCracken et al. 2023); however, whether dysbiosis is causative or responsive to disease outbreak remains unknown. One hypothesis suggests that a trigger for SSWD could be low oxygen conditions on the surface boundary layer of sea stars (Aquino et al. 2021). However, ongoing controlled challenge experiments have shown that SSWD can be controlled and transmitted from a wasting Sunflower Star via co-housing, water, and inoculation with tissue homogenate and coelomic fluid from a wasting Sunflower Sea Star (Gehman unpubl. data). Importantly, these studies show that Sunflower Sea Star specimens kept in isolation and exposed to heat-killed components of tissue homogenate and coelomic fluid from a wasting sunflower star do not develop signs of SSWD (Gehman unpubl. data). Through co-housing, SSWD has been transmitted from Sunflower Sea Star individuals to Ochre Star individuals and vice versa (Crandall and Gehman unpubl. data). These results suggest a single causative agent is necessary for signs of disease, and ongoing genomic work is exploring the identity of that agent (Prentice, Gehman et al. unpubl. data). However, this does not rule out the potential for interactions between a causative agent and later stages of dysbiosis. Intriguingly, aquarists report having developed successful methods for treating stars with signs of disease by using a sterilizing agent (“Seachem Reef Dip”) and housing them within a probiotic and pH-controlled environment (Rudek and Collura 2023). Future work combining controlled exposure experiments with these husbandry techniques could elucidate the role of dysbiosis in the development of signs of SSWD.

Infection can alter hosts’ thermal performance across a range of species (Gehman et al. 2018; Cohen et al. 2020). In the field, Sunflower Sea Star has shown a change in distribution following the outbreak of SSWD. Prior to the SSWD outbreak peak Sunflower Sea Star abundance was associated with water temperatures of 16 °C, but after the SSWD outbreak, the peak was around 5 °C (Hamilton et al. 2021a). Combined with the association of the original outbreak with anomalously high temperatures (Harvell et al. 2019), this suggests that SSWD could be altering infected Sunflower Sea Star individuals’ thermal tolerance.

There is anecdotal evidence that Sunflower Sea Star mortality is ongoing at local scales. After several years of relative stability (from 2019), remnant groups of Sunflower Sea Star surveyed in remote, cold fjords in Canada have recently declined severely (A. Gehman, pers. obs. 2024, contra Gehman et al., 2025, a study based on data up to 2023).

8.3 Introduced genetic material (threat impact: unknown)

Research into captive breeding for release is ongoing in the United States, but whether this is a potential threat (versus being harmless or advantageous) to the Canadian portion of the population is unknown.

Category 9: Pollution

9.2 Industrial and military effluents (threat impact: low)

Oil spills at sea are a potential local threat, but the effect of the small fraction of oil that settles on the bottom is unknown. Catastrophic spills could occur any time or even beyond the 3-generation time frame, and such spills within the range have happened in the past, though effects on sea stars are unknown. Impacts of small chronic spills and acute catastrophic spills differ and depend on the time of year, weather and sea state condition, where the spills occur, and what kind of spill (for example, diesel versus other fuel types, including diluted bitumen). Uncertainty is high for severity and scope. The effects of chronic leaks and spills are also unknown.

Category 11: Climate change and severe weather

11.1 Habitat shifting and alteration (threat impact: low)

Climate change effects are potentially limiting recruitment in southern latitudes, and generally warming seas may increase the incidence of SSWD, though this is conjecture. Ocean acidification negatively affects some sea star prey (for example, shellfish). Sunflower Sea Star is mobile and can accommodate sea level rise. While the scope and timing of habitat changes due to climate change are both high and ongoing, there is uncertainty in projections regarding impacts (severity) on sea star adults and larvae over 3 generations. Over this time frame, expert opinion is that larvae are more likely to be negatively affected.

11.3 Temperature extremes (threat impact: low)

There is some evidence of a link between warming temperature and both the progression of SSWD and declines in sea star abundance. Laboratory experiments have shown that higher temperatures are associated with faster disease progression and higher mortality in Ochre Sea Star (Eisenlord et al. 2016; Kohl et al. 2016). However, for Sunflower Sea Star, the evidence is correlational and mixed. Mixed information on the influence of temperature could be driven by pandemic transmission dynamics, where high numbers of susceptible individuals can swamp out the effect of abiotic drivers (Carlson et al. 2020). That said, Harvell et al. (2019) and Hamilton et al. (2021a) both suggest an effect of temperature. Before the outbreak, water depth was the most important predictor of Sunflower Sea Star occurrence across its range but, after the outbreak, temperature became as important as depth, suggesting that temperature drove the observed latitudinal patterns of SSWD timing and severity (Hamilton et al. 2021a). Sunflower Sea Star declines in shallow nearshore waters from California to Alaska were associated with the maximum sea surface temperature (SST) anomaly occurring within 2 months of each survey and low counts were more likely when SSTs had recently been above average (Harvell et al. 2019). The SST anomaly metric accounted for 38% of the variance in Sunflower Sea Star counts, suggesting a role of temperature but also of other unidentified variables. However, not all outbreaks of SSWD were associated with warmer-than-average temperatures (Menge et al. 2016; Hewson et al. 2018).

Whether temperature is primarily influencing the putative pathogen, the host, or both remains unresolved. Models of early spread of SSWD that included both temperature-mediated pathogen spread and a temperature trigger were the most consistent with field observed patterns (Aalto et al. 2019), potentially indicating a role of temperature for both host and pathogen. Overall, the evidence suggests that warming exacerbates SSWD (Harvell et al. 2019; Gravem et al. 2021; Hamilton et al. 2021a, Gehman et al., 2025).

11.4 Storms and flooding (threat impact: unknown)

Storms and flooding are both projected to increase over the next 3 generations, but direct impacts are likely low, and indirect impacts (for example, potentially causing oil spills) are unknown.

To the extent that increased mean water temperatures and temperature extremes affect the exposure to and progression of SSWD, climate change may be an important threat to Sunflower Sea Star. However, the Foden et al. (2013) framework (exposure, severity, and adaptive capacity) offers moderate guidance for assessing this threat for this species: exposure and severity may both be high, but adaptive capacity (that is the ability to colonize new habitat via larval dispersal) is unknown.

Category 5: Biological resource use

5.4 Fishing and harvesting aquatic resources (threat impact: unknown)

There is no targeted fishing of Sunflower Sea Star in Canada or elsewhere within its range (Gravem et al. 2021). However, it is regularly recorded as bycatch in DFO prawn and crab research data (Figure 5) and is commonly encountered in pot/trap and trawl/seine test fisheries in Washington state (Gravem et al. 2021). The extent of bycatch and survival rate of any sea stars released are unknown.

Limiting factors

As Sunflower Sea Star is a dioecious broadcast spawner, its reproductive success may be limited by inter-individual distance and, hence, by the currently low population density.

Number of locations

The most serious and plausible threat to Sunflower Sea Star is SSWD. There may be some Sunflower Sea Star refuges in Canada and Alaska (Hamilton et al. 2021a), which prompted Gravem et al. (2021) to suggest a maximum of 10 locations across the northern part of the global range of Sunflower Sea Star. However, given that the 2013‒2015 SSWD outbreak rapidly affected this species globally and that there is minimal pre-SSWD information on Sunflower Sea Star abundance in possible refuges, the number of locations in Canada is estimated to be 1, which was also suggested for the global range (Gravem et al. 2021).

Protection, status and ranks

Legal protection and status

Sunflower Sea Star is not legally protected under any legislation in Canada, except in the Pacific Rim and Gulf Islands national park reserves under the Canada National Parks Act and in Gwaii Haanas National Park Reserve under the Canada National Marine Conservation Areas Act.

Non-legal status and ranks

Sunflower Sea Star was assessed as Critically Endangered on the IUCN Red List of Threatened Species 2021, due to the drastic decline in global populations and range after the outbreak of SSWD (Gravem et al. 2021). The U.S. National Oceanic and Atmospheric Administration has submitted a proposal to list Sunflower Sea Star as Threatened under the U.S. Endangered Species Act. NatureServe (2024) has assigned N2 (Imperilled) status for Sunflower Sea Star in Canada; global and United States ranks have not yet been assigned.

Habitat protection and ownership

Approximately 24.5% of Canada’s Pacific coastal and marine waters receive some form of protection (Statistics Canada 2021), and most of the 200 or so marine protected areas (MPAs) in Canada’s Pacific overlap with the range of Sunflower Sea Star. The vast majority of MPAs are managed either by British Columbia Parks in designated ecological reserves, provincial parks, protected areas, conservancies, or recreation areas, or by the federal government in designated national park reserves, national marine conservation area reserves, national wildlife areas, marine protected areas, or migratory bird sanctuaries. The extent of protection from anthropogenic effects varies among MPAs, and few exclude all extractive activities. However, even the strictest MPAs do not protect Sunflower Sea Star from SSWD, which is the main cause of population decline.

Acknowledgements

Financial support was provided by Fisheries and Oceans Canada. We thank Jackie Hildering and Fiona Francis for spearheading the idea of assessing Sunflower Sea Star, and Cathryn Clarke Murray, Lucie Hannah, and Fiona Francis of DFO for facilitating this report. The Asteroidea (ad hoc) SSC consisted of Christina Czembor, Jennifer Diment, Purnima Govindarajulu, Jennifer Heron, Chris Johnson, Bruce Leaman, Dwayne Lepitski, Nicholas Mandrak and Arne Mooers. The authorities listed below provided valuable data and/or advice for this report.

We are grateful to Sara Hamilton (Oregon State University) and the IUCN report team (Sarah Gravem, Walter Heady, Vienna Saccomanno, Kristen Alvstad, Alyssa Gehman, and Taylor Frierson) as well as the additional authors of the subsequent Proc B paper (Steve Lonhart, Rodrigo Beas-Luna, Fiona Francis, Lynn Lee, Laura Rogers-Bennett, and Anne Salomon) for completing the massive initial undertaking of compiling Sunflower Sea Star data across its global distribution range and for placing these data in an open-access archive.

Thank you to all the data contributors who made this evaluation of Sunflower Sea Star in Canada possible and who gave us permission to use their data: Alejandro Frid on behalf of the Heiltsuk, Kitasoo/Xai’xais, Nuxalk and Wuikinuxv nations under the umbrella of the Central Coast Indigenous Resource Alliance; Fiona Francis for compiling the Fisheries and Oceans datasets, Dominique Bureau and Erin Herder for providing access to the Abalone Survey data, Janet Lochead for the Multispecies Dive data, and Joanne Lessard and Ashley Park for the Benthic Habitat Mapping data; Lynn Lee, Dan McNeill, and Dan Okamoto for the Gwaii Haanas National Park Reserve, National Marine Conservation Area, and Haida Heritage Site (“Gwaii Haanas”), the Council of the Haida Nation, the Parks Canada Agency, Fisheries and Oceans Canada, the Haida Fisheries Program divers and contract divers including Vanessa Bellis, Richard Smith, Ben Penna, Jaasaljuus Yakgujanaas, Shaun Edgars, Ondine Pontier, Doug Swanston, Erika Paradis, Tristan Blaine, Ryan Miller, Leandre Vigneault and Candice St. Germain, and many Gwaii Haanas field support staff including data support from Charlotte Houston, Niisii Guujaaw, Marilyn Deschenes and Chavonne Guthrie; Alyssa Gehman for the Hakai dataset with support from the Hakai Institute with study design and data curation by Margot Hessing-Lewis, Alyssa Gehman, Ondine Pontier, and Tyrel Froese, and field research conducted by Jenn Burt, Ondine Pontier, Angeleen Olson, Zach Monteith, Derek Van Maanen, Tanya Prinzing, Kyle Hall, Andrew McCurdy, Tristan Blaine, Krystal Bachen, Neha Acharya-Patel, and other Hakai staff; Melissa Miner for the MARINe dataset with support from all contributing divers; Donna Gibbs for the Ocean Wise Pacific Marine Life Surveys database with support from Ocean Wise at the Vancouver Aquarium, Charlie Gibbs, and Andy Lamb; Christy Pattengill-Semmens for the REEF dataset with support from all contributing divers; Lynn Lee for the Simon Fraser—Lee dataset from Gwaii Haanas, the Council of the Haida Nation, the Heiltsuk Integrated Resource Management Department, Fisheries and Oceans Canada, the Natural Sciences and Engineering Research Council of Canada, the Hakai Institute/Tula Foundation, Marine Toad Enterprises Inc., Simon Fraser University, with support from Anne Salomon, Rowan Trebilco, Alejandro Frid, Hannah Stewart, Matt Drake, Stu Humchitt, Eric White, Joel White, Jane Watson, Leah Saville, Erin Rechsteiner, Joanne Lessard, Seaton Taylor, Mike Atkins, Leandre Vigneault, and Taimen Lee Vigneault, with much appreciation for and acknowledgement of the Haida, Heiltsuk and Nuu-chah-nulth nations, on whose traditional territories these data were collected; Anne Salomon, Hannah Stewart, and Lynn Lee for the Simon Fraser—Salomon dataset with support from Simon Fraser University, the Natural Sciences and Engineering Research Council of Canada, the Canadian Foundation for Innovation, the Hakai Institute, Gwaii Haanas, Fisheries and Oceans Canada, the Haida, Heiltsuk, and Wuikinuxv nations, with field leadership from Jenn Burt, Kyle Demes, Margot Hessing-Lewis, Britt Keeling, Hannah Stewart, and Rowan Trebilco; Jessica Schultz and Ryan Cloutier for the Simon Fraser—Schultz dataset with support from Simon Fraser University, the Ocean Wise Conservation Association, the Howe Sound Research and Conservation Program at the Vancouver Aquarium, the Natural Sciences and Engineering Research Council of Canada, and a Discovery Grant to Isabelle Côté; Jane Watson for the VIU dataset with support from the Vancouver Island University, Leah Saville, Gina Lemiux, and Erin Rechsteiner, with support from Friends of the Ecological Reserves, Fisheries and Oceans Canada, British Columbia Parks, Ecological Reserves Unit, who collected data in the traditional territories of the Huu-ay-aht, Kyuquot, and Checleseht First Nations.

Authorities contacted

• Arnett, K. Executive Director. Strawberry Isle Marine Research Society. Tofino, British Columbia
• Dawson, M.N. Professor. University of California, Merced. Merced, California, U.S.A
• Francis, F. Aquatic Biologist. Groundfish Section, Stock Assessment and Research Division, Fisheries and Oceans Canada. Vancouver, British Columbia
• Gibbs, D. Pacific Marine Life Surveys Inc. Vancouver, British Columbia
• Gravem, S. Research Associate. Department of Integrative Biology, Oregon State University. Corvallis, Oregon, U.S.A
• Greenberg, D. Research Associate. Reef Environmental Education Foundation. Key Largo, Florida, U.S.A
• Hamilton, S. NSF-GRFP Ph.D. Candidate. Oregon State University. Corvallis, Oregon, U.S.A
• Pattengill-Semmens, C. Co-Executive Director, Science and Engagement. Reef Environmental Education Foundation. San Diego, California, U.S.A
• Yakimishyn, J. Resource Management Officer II. Parks Canada. Ucluelet, British Columbia

Information sources

Aalto, E.A., K.D. Lafferty, S.H. Sokolow, R.E. Grewelle, T. Ben-Horin, C.A. Boch, P.T. Raimondi, S.J. Bograd, E.L. Hazen, M.G. Jacox, F. Micheli, and G.A. De Leo. 2020. Models with environmental drivers offer a plausible mechanism for the rapid spread of infectious disease outbreaks in marine organisms. Scientific Reports 10:5975.

Aquino, C.A., R.M. Besemer, C.M. DeRito, J. Kocian, I.R Porter, P.T. Raimondi, J.E. Rede, L.M. Schiebelhut, J.P. Sparks, J.P. Wares, and I. Hewson. 2021. Evidence that microorganisms at the animal-water interface drive sea star wasting disease. Frontiers in Microbiology 11:610009.

Bates, A.E., B.J. Hilton, and C.D.G. Harley. 2009. Effects of temperature, season and locality on wasting disease in the keystone predatory sea star Pisaster ochraceus. Diseases of Aquatic Organisms 86:245 to 251.

Bonaviri, C., M. Graham, P. Gianguzza, and N.T. Shears. 2017. Warmer temperatures reduce the influence of an important keystone predator. Journal of Animal Ecology 86:490 to 500.

Brewer, R., and B. Konar. 2005. Chemosensory responses and foraging behavior of the seastar Pycnopodia helianthoides. Marine Biology 147:789 to 795.

British Columbia. 2023. Protected lands and waters in BC, Environmental Reporting BC. Website: https://www.env.gov.bc.ca/soe/indicators/land/protected-lands-and-waters.html [accessed April 2023].

Britton-Simmons, K.H., A.L. Rhoades, R.E. Pacunski, A.W.E. Galloway, A.T. Lowe, E.A. Sosik, M.N. Dethier, and D.O. Duggins. 2012. Habitat and bathymetry influence the landscape-scale distribution and abundance of drift macrophytes and associated invertebrates. Limnology and Oceanography 57:176 to 184.

Burt, J.M., M.T. Tinker, D.K. Okamoto, K.W. Demes, K. Holmes, and A.K. Salomon. 2018. Sudden collapse of a mesopredator reveals its complementary role in mediating rocky reef regime shifts. Proceedings of the Royal Society B 285:20180553.

Carlson, C.J., A.C.R. Gomez, S. Bansal, and S.J. Ryan. 2020. Misconceptions about weather and seasonality must not misguide COVID-19 response. Nature Communications 11:4312.

Cohen, J.M., E.L. Sauer, O. Santiago, S. Spencer, and J.R. Rohr. 2020. Divergent impacts of warming weather on wildlife disease risk across climates. Science 370: eabb1702.

Dennis, B., J.M. Ponciano, S.R. Lele, M.L. Taper, and D.F. Staples. 2006. Estimating density dependence, process noise, and observation error. Ecological Monographs 76:323 to 341.

Duggins, D.O. 1983. Starfish predation and the creation of mosaic patterns in a kelp-dominated community. Ecology 64:1610 to 1619.

Dungan, M.L., T.E. Miller, and D.A. Thomson. 1982. Catastrophic decline of a top carnivore in the Gulf of California rocky intertidal zone. Science 216:989 to 991.

Eisenlord, M.E., M.L. Groner, R.M. Yoshioka, J. Elliott, J. Maynard, S. Fradkin, M. Turner, K. Pyne, N. Rivlin, R. Van Hooidonk, and C.D. Harvell. 2016. Ochre star mortality during the 2014 wasting disease epizootic: role of population size structure and temperature. Philosophical Transactions of the Royal Society B 371:20150212.

English, P.A., E.J. Ward, C.N. Rooper, R.E. Forrest, L.A. Rogers, K.L. Hunter, A.M. Edwards, B.M. Connors, and S.C. Anderson. 2022. Contrasting climate velocity impacts in warm and cool locations show that effects of marine warming are worse in already warmer temperate waters. Fish and Fisheries 23:239 to 255.

Feder, H.M. 1980. Asteroidea: the sea stars. Pp. 117 to 135, in R.H. Morris, D.P. Abbott, and E.C. Haderlie (. Intertidal Invertebrates of California. Stanford University Press, Stanford, California.

Feder, H.M. and A.M. Christensen. 1966. Aspects of asteroid biology. Pp. 87 to 127, in R.A. Boolootian (ed.). Physiology of Echinodermata. Interscience Publishers, New York, New York.

Fly, E.K., C.J. Monaco, S. Pincebourde, and A. Tullis. 2012. The influence of intertidal location and temperature on the metabolic cost of emersion in Pisaster ochraceus. Journal of Experimental Marine Biology and Ecology 422 to 423:20 to 28.

Foden, W.B., S.H.M. Butchart, S.N. Stuart, J.-C. Vié, H.R. Akçakaya, A. Angulo, L.M. DeVantier, A. Gutsche, E. Turak, L. Cao, S.D. Donner, V. Katariya, R. Bernard, R.A. Holland, A.F. Hughes, S.E. O’Hanlon, S.T. Garnett, Ç.H. Şekercioğlu, and G.M. Mace. 2013. Identifying the world’s most climate change vulnerable species: a systematic trait-based assessment of all birds, amphibians and corals. PLoS ONE 8(6): e65427.

Freeland, H., and T. Ross. 2019. ‘The Blob’ - or, how unusual were ocean temperatures in the Northeast Pacific during 2014-2018? Deep-Sea Research Part 1 150:103061.

Galloway A.W.E., S.A. Gravem, J.N. Kobelt, W.N. Heady, D.K. Okamoto, D.M. Sivitilli, V.R. Saccomanno, J. Hodin, and R. Whippo. 2023. Sunflower sea star predation on urchins can facilitate kelp forest recovery. Proceedings of the Royal Society B 290:20221897.

Gehman A.L.M., O. Pontier, T. Froese, D. VanMaanen, T. Blaine, G. Sadlier-Brown, A.M. Olson, Z.L. Monteith, K. Bachen, C. Prentice, M. Hessing-Lewis, and J.M. Jackson. 2025. Fjord oceanographic dynamics provide refuge for critically endangered Pycnopodia helianthoides. Proceedings of the Royal Society B 292:20242770.

Gravem, S.A., W.N. Heady, V.R. Saccomanno, K.F. Alvstad, A.L.M. Gehman, T.N. Frierson, and S.L. Hamilton. 2021. Pycnopodia helianthoides. IUCN Red List of Threatened Species 2021.

Greer, D.L. 1962. Studies on the embryology of Pycnopodia helianthoides (Brandt) Stimpson. Pacific Science 16:280 to 285.

Greenberg, D.A., C. Pattengill-Semmens, B.X. Semmens. In prep. Validating citizen science natural history observations as a tool for assessing the status and trends of marine species.

Hamilton, S.L., V.R. Saccomanno, W.N. Heady, A.L. Gehman, S.I. Lonhart, R. Beas- Luna, F.T. Francis, L. Lee, L. Rogers-Bennett, A.K. Salomon, and S.A. Gravem. 2021a. Disease-driven mass mortality event leads to widespread extirpation and variable recovery potential of a marine predator across the eastern Pacific. Proceedings of the Royal Society B 288:20211195.

Hamilton, S.L., V.R. Saccomanno, W.N. Heady, A.L. Gehman, S.I. Lonhart, R. Beas- Luna, F.T. Francis, L. Lee, L. Rogers-Bennett, A.K. Salomon, and S.A. Gravem. 2021b. Disease-driven mass mortality event leads to widespread extirpation and variable recovery potential of a marine predator across the eastern Pacific [Dataset]. Dryad.

Harley, C.D.G., M.S. Pankey, J.P. Wares, R.K. Grosberg, and M.J. Wonham. 2006. Color polymorphism and genetic structure in the Sea Star Pisaster ochraceus. The Biological Bulletin 211:248 to 262.

Harvell, C.D., D. Montecino-Latorre, J.M. Caldwell, J.M. Burt, K. Bosley, A. Keller, S.F. Heron, A.K. Salomon, L. Lee, O. Pontier, C. Pattengill-Semmens, and J.K. Gaydos. 2019. Disease epidemic and a marine heat wave are associated with the continental-scale collapse of a pivotal predator (Pycnopodia helianthoides). Science Advances 5: eaau7042.

Harvey, G.K.A., T.A. Nelson, P.C. Paquet, C.J. Ferster, and C.H. Fox. 2018. Comparing citizen science reports and systematic surveys of marine mammal distributions and densities. Biological Conservation 226:92 to 100.

Hemery, L.G., S.R. Marion, C.G. Romsos, A.L. Kurapov, and S.K. Henkel. 2016. Ecological niche and species distribution modelling of sea stars along the Pacific Northwest continental shelf. Diversity and Distributions 22:1314 to 1327.

Herrlinger, T.J. 1983. The diet and predator-prey relationships of the sea star Pycnopodia helianthoides (Brandt) from a central California kelp forest. M.A. thesis, San Jose State University, San Jose, California. 26 pp.

Hewson, I., J.B. Button, B.M. Gudenkauf, B. Miner, A.L. Newton, J.K. Gaydos, J. Wynne, C.L. Groves, G. Hendler, M. Murray, S. Fradkin, M. Breitbart, E. Fahsbender, K.D. Lafferty, A.M. Kilpatrick, C.M. Miner, P. Raimondi, L. Lahner, C.S. Friedman, S. Daniels, M. Haulena, J. Marliave, C.A. Burge, M.E. Eisenlord, and C.D. Harvell. 2014. Densovirus associated with sea-star wasting disease and mass mortality. Proceedings of the National Academy of Sciences of the United States of America 111:17278 to 17283.

Hewson, I., K.S.I. Bistolas, E.M. Quijano Cardé, J.B. Button, P.J. Foster, J.M. Flanzenbaum, J. Kocian, and C.K. Lewis. 2018. Investigating the complex association between viral ecology, environment, and northeast Pacific sea star wasting. Frontiers in Marine Science 5:77.

Hewson, I., B. Sullivan, E.W. Jackson, Q. Xu, H. Long, C. Lin, E.M. Quijano Cardé, J. Seymour, N. Siboni, M.R.L. Jones, and M.A. Sewell. 2019. Perspective: something old, something new? Review of wasting and other mortality in Asteroidea (Echinodermata). Frontiers of Marine Science 6:406.

Hewson, I., M.R. Johnson, and B. Reyes-Chavez. 2024. Lessons learned from the sea star wasting disease investigation. Annual Review of Marine Science 17:257 to 279. https://doi.org/10.1146/annurev-marine-040623-082617

Hodin, J., A. Pearson-Lund, F.P. Anteau, P. Kitaeff, and S. Cefalu. 2021. Progress toward complete life-cycle culturing of the endangered Sunflower Star, Pycnopodia helianthoides. The Biological Bulletin 241:253 to 258.

iNaturalist. 2022.

Kohl, W.T., T.I. McClure, and B.G. Miner. 2016. Decreased temperature facilitates short-term sea star wasting disease survival in the keystone intertidal sea star Pisaster ochraceus. PLoS ONE 11(4): e0153670.

Lambert, P. 2000. Sea stars of British Columbia, Southeast Alaska and Puget Sound. University of British Columbia Press, Vancouver, British Columbia. vi + 186 pp.

Lee, L.C., J.C. Watson, R. Trebilco, and A.K. Salomon. 2016. Indirect effects and prey behavior mediate interactions between an endangered prey and recovering predator. Ecosphere 7(12): e01604.

Leis, J.M. 2021. Perspectives on larval behaviour in biophysical modelling of larval dispersal in marine, demersal fishes. Oceans 2:1 to 25.

Lloyd, M.M., and M.H. Pespeni. 2018. Microbiome shifts with onset and progression of Sea Star Wasting Disease revealed through time course sampling. Scientific Reports 8:16476.

Lundquist, C.J., and L.W. Botsford. 2004. Model projections of the fishery implications of the Allee effect in broadcast spawners. Ecological Applications 14:929 to 941.

Ma, M., and K. Taguchi. 2021. Thousands of baby sea stars born at UW lab are sign of hope for endangered species. University of Washington News, University of Washington. Website: https://www.washington.edu/news/2021/04/27/thousands-of-baby-sea-stars-born-at-uw-lab-are-sign-of-hope-for-endangered-species/ [accessed October 2021].

Mah, C.L. 2021. World Asteroidea Database. Lysastrosoma anthosticta Fisher, 1922. Website: https://www.marinespecies.org/asteroidea/aphia.php?p=taxdetails&id=292866

Marko, P.B., J.M. Hoffman, S.A. Emme, T.M. McGovern, C.C. Keever, and L.N. Cox. 2010. The ‘Expansion–Contraction’ model of Pleistocene biogeography: rocky shores suffer a sea change? Molecular Ecology 19:146 to 169.

Mauzey, K.P., C. Birkeland, and P.K. Dayton. 1968. Feeding behavior of asteroids and escape responses of their prey in the Puget Sound region. Ecology 49:603 to 619.

McCracken, A.R., B.M. Christensen, D. Munteanu, B.K.M. Case, M. Lloyd, K.P. Herbert, and M.H. Pespeni. 2023. Microbial dysbiosis precedes signs of sea star wasting disease in wild populations of Pycnopodia helianthoides. Frontiers in Marine Science 10:1130912.

Menge, B.A. 1975. Brood or broadcast? The adaptive significance of different reproductive strategies in the two intertidal sea stars Leptasterias hexactis and Pisaster ochraceus. Marine Biology 31:87 to 100.

Menge, B.A., E.B. Cerny-Chipman, A. Johnson, J. Sullivan, S. Gravem, and F. Chan. 2016. Sea star wasting disease in the keystone predator Pisaster ochraceus in Oregon: insights into differential population impacts, recovery, predation rate, and temperature effects from long-term research. PLoS ONE 11(5): e0153994.

Miner, C.M., J.L. Burnaford, R.F. Ambrose, L. Antrim, H. Bohlmann, C.A. Blanchette, J.M. Engle, S.C. Fradkin, R. Gaddam, C.D.G. Harley, B.G. Miner, S.N. Murray, J.R. Smith, S.G. Whitaker, and P.T. Raimondi. 2018. Large-scale impacts of sea star wasting disease (SSWD) on intertidal sea stars and implications for recovery. PLoS ONE 13(3): e0192870.

Montecino-Latorre, D., M.E. Eisenlord, M. Turner, R. Yoshioka, C.D. Harvell, C.V. Pattengill-Semmens, J.D. Nichols, and J.K. Gaydos. 2016. Devastating transboundary impacts of sea star wasting disease on subtidal asteroids. PLoS ONE 11(10): e0163190.

Moitoza, D.J., and D.W. Phillips. 1979. Prey defense, predator preference, and nonrandom diet: the interactions between Pycnopodia helianthoides and two species of sea urchins. Marine Biology 53:299 to 304.

Morris, R.H., D.P. Abbott, and E.C. Haderlie. 1980. Intertidal Invertebrates of California. Stanford University Press, Stanford, California. ix + 690 pp.

NatureServe. 2024. NatureServe Explorer. Website: https://explorer.natureserve.org/ [accessed July 2024].

R Core Team 2021. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

Rudek, T., and E.M. Collura. 2023. Sea star illness treatment protocol. protocols.io.

Schultz, J.A., R.N. Cloutier, and I.M. Côté. 2016. Evidence for a trophic cascade on rocky reefs following sea star mass mortality in British Columbia. PeerJ 4: e1980.

Sewell, M.A., and J.C. Watson. 1993. A “source” for asteroid larvae?: recruitment of Pisaster ochraceous, Pycnopodia helianthoides and Dermasterias imbricata in Nootka Sound, British Columbia. Marine Biology 117:387 to 398.

Shanks, A.L. 2009. Pelagic larval duration and dispersal distance revisited. Biological Bulletin 216:373 to 385.

Shivji, M., D. Parker, B. Hartwick, M.J. Smith, and N.A. Sloan.1983. Feeding and distribution study of the Sunflower Sea Star Pycnopodia helianthoides (Brandt, 1835). Pacific Science 37:133 to 140.

Smale, D.A., T. Wernberg, E.C.J. Oliver, M. Thomsen, P.B. Harvey, S.C. Straub, M.T. Burrows, L.V. Alexander, J.A. Benthuysen, M.G. Donat, M. Feng, A.J. Hobday, N.J. Holbrook, S.E. Perkins-Kirkpatrick, H.A. Scannell, A. Sen Gupta, B.L. Payne, and P.J. Moore. 2019. Marine heatwaves threaten global biodiversity and the provision of ecosystem services. Nature Climate Change 9:306 to 312.

Species at Risk Act: order amending Schedule 1 to the Species at Risk Act. 2011. Canada Gazette, Part II, 145(14):1227 to 1253.

Stan Development Team. 2022. RStan: The R Interface to Stan. R Package Version 2.21.5.

Stan Development Team. 2023. Stan Modeling Language Users Guide and Reference Manual, Version 2.29.2.

Strathmann, R. 1978. Length of pelagic period in echinoderms with feeding larvae from the Northeast Pacific. Journal of Experimental Marine Biology and Ecology 34:23 to 27.

Traiger, S.B., B. Konar, A. Doroff, and L. McCaslin. 2016. Sea otters versus sea stars as major clam predators: evidence from foraging pits and shell litter. Marine Ecology Progress Series 560:73 to 86.

Wobber, D.R. 1975. Agonism in asteroids. Biological Bulletin 148:483 to 496.

Biographical summary of report writers

Isabelle M. Côté is Professor of Marine Ecology at Simon Fraser University. She has written more than 200 papers on subjects related to marine ecology and conservation and pioneered the use of meta-analyses to address marine conservation issues. Her group was among the first to document the ecological consequences of Sunflower Sea Star mortality in B.C. waters. She coordinated the team and edited the final version of the report.

Alyssa Gehman is a Research Scientist with the Hakai Institute and Adjunct Professor at the University of British Columbia. She is a marine disease expert interested in the interactions between infected hosts and their communities, as well as the impact of temperature on host-parasite interactions. She has ongoing work evaluating the causative agent of sea star wasting disease, as well as monitoring sea star populations in the Central Coast of B.C.. She contributed the most up-to-date account and unpublished data on the nature of the pathogen causing sea star wasting disease.

Beth Oishi is an M.Sc. student in Dr. Côté’s group. Her research focuses on the effects of the invasive European Green Crab on eelgrass meadows. She previously conducted a large systematic review of the effect of acclimation on the responses of marine organisms to warming temperatures. She was responsible for compiling the data and writing the first draft of the report.

Hannah V. Watkins is a Ph.D. candidate in Dr. Côté’s group. Her research focuses on the ecology and sustainable management of sea cucumber populations in B.C. She conducted the novel Bayesian analyses.

Steven Brownlee is a Ph.D. candidate in Dr. Côté’s group with an M.Sc. in Geomatics. His research focuses on predicting the colonization and spread of invasive mussels in the Okanagan basin. He was responsible for the production of maps and occurrence statistics.

Collections examined

No museum specimens were examined in the preparation of this status report.

Appendix 1. Threats calculator for Sunflower Sea Star (Pycnopodia helianthoides)

Threats assessment worksheet

Species or Ecosystem scientific name: Sunflower Sea Star (Pycnopodia helianthoides)

Date: 6/20/2024

Assessor(s): Dwayne Lepitzki, Arne Mooers, Katarina Duke, Moriah Cootes, Alyssa Gehman, Isabelle Cote, Fiona Francis, Chris Johnson, Hannah Watkins, Iona Kearns, Melanie Prentice, Brad Johnson, Shannan May-McNally, Emma Pascoe, Sarah Gravem, Jennifer Heron, Purnima Govindarajulu

Assigned overall threat impact: BC = High - Medium

Impact adjustment reasons: Overall threat impact high range reduced from “Very high” to “High” because declines have halted and experts believe it unlikely that probability of extirpation or extinction is imminent.

Overall threat comments: IUCN CE A2ace 2021 because of dramatic declines in population size and range from sea star wasting disease (SSWD; pathogen is unknown but likely native); gen time is approximately 10 to 20 years; time frame for severity and timing used here is 30 to 50 years into the future; no decline (or recovery) observed since 2015; estimated decline in observed abundance and probability of sighting over past 22 years is 82% and 69%, respectively (no data for 3 gens), and future decline cannot be reliably estimated; 1 location is proposed because of SSWD; probable close linkage between SSWD and sea temperatures with heat-stressed individuals more prone to disease. Sunflower Sea Star eats commercially important clams, mussels, urchins, cucumbers, and Northern Abalone, and is a key predator species. Where top predator Sea Otter is absent, the sea star is the principal urchin predator; in absence of otters, absence of sea stars leads to overabundant urchins, which overeat kelp.

Appendix 2. Temporal distribution, sample sizes, and modelled probabilities of sightings of Sunflower Sea Star occurrence across Canadian datasets included in this report

Figure S1. Probability of sighting for Sunflower Sea Stars on dive surveys, measured as average presence score (where absent = 0 and present = 1) across individual surveys, across years in each of the 14 data sources included in the quantitative analyses. The sizes of dots reflect sample sizes (number of surveys). Information about each data source is presented in Table 1 in the main text.

Appendix 3. Modelling framework for abundance and probability-of-sighting analyses

The modelling approach used in this assessment has two major components: (1) generalized linear mixed-effects models (GLMMs) to model dataset-specific population changes; and (2) a multivariate autoregressive state-space model (Ward et al. 2010; Holmes et al. 2012; Greenberg et al. 2024), which allows an integrated population estimate to be made from the trends in the dataset-specific GLMMs.

1. Dataset-specific models (GLMMs)

Each dataset-specific model is a GLMM estimating a time-varying intercept for each year of survey data (that is, ${\alpha }_t$) that tracks mean abundance through time. A combination of random (for example, site) and fixed (for example, log-transformed time or area surveyed, depth) effects were included in each model, depending on the available data for each dataset. While most of the datasets recorded counts of individual Pycnopodia as a response variable, two of the largest datasets (that is, the Reef Environmental Education Foundation [REEF] and the Ocean Wise Pacific Marine Life Database [OW]) used discrete abundance scores instead, with REEF data allocated into $K=5$ categories (that is, 0, 1, 2 to 10, 11 to 100, and >100 individuals observed on a single survey), and OW data allocated into $K=7$ categories (that is, 0, 1 to 10, 11 to 25, 26 to 50, 51 to 100, 101 to 1,000, and >1,000 individuals). Therefore, two separate approaches were used for the dataset-specific GLMMs: ordinal logistic regressions for the datasets with abundance scores and negative binomial regressions for the datasets with counts.

1a. Ordinal logistic regressions

For each of the datasets with abundance scores, an ordinal logistic regression was developed, in which an underlying linear predictor $\eta$ of dataset i, site s, and year t, was estimated from the individual year- (${\alpha }_t$) and site-specific (${\alpha }_s$) abundances, as well as survey-specific covariates ($X_i$) that might influence detection ability, such as dive time, maximum survey depth, diver experience, visibility, and current:

(1)

$\mathrm{Ordered\; Logistic}\left(y_{i,s,t}\right|{\eta }_{i,s,t},c_{1:K-1})$

(2)

${\eta }_{i,s,t}={\alpha }_t+{\alpha }_s+\beta X_i$

All continuous variables were standardized (that is, centred and scaled), such that the year-specific abundances (${\alpha }_t$) represented an index of abundance at the mean level of survey effort and under typical survey conditions. Site-specific differences in abundance (${\alpha }_s$) were modelled to reduce some of the bias that could be introduced into the estimate of the population trajectory by the potential over-representation of consistently high or low abundance sites in post-crash years. Thus, ${\alpha }_t$ functioned as an index of abundance that was comparable across years, even when the sampling effort varied. The year-specific linear abundances ${\alpha }_t$ could then be used to determine the probability of an individual survey falling into each score ($k$) in a given year by shifting the estimated cut points ($c_k$):

(3)

$\begin{array}{l}\mathrm{Pr}(k=1)\\ \mathrm{Pr}(k=2)\\ \dots \\ \mathrm{Pr}(k=K-1)\\ \mathrm{Pr}(k=K-1)\end{array}=\{\begin{array}{l}{\mathrm{logit}}^{-1}(c_1-{\alpha }_t)\\ {\mathrm{logit}}^{-1}(c_2-{\alpha }_t)-{\mathrm{logit}}^{-1}(c_1-{\alpha }_t)\\ \dots \\ {\mathrm{logit}}^{-1}(c_{K-1}-{\alpha }_t)-{\mathrm{logit}}^{-1}(c_{K-2}-{\alpha }_t)\\ 1-{\mathrm{logit}}^{-1}(c_{K-1}-{\alpha }_t)\end{array}$

These probabilities could then be multiplied by a chosen count in each category to generate a conservative estimate of the index of abundance for each year. Since most observations in these datasets fell into the lower score bins (that is, those with the least variability in possible true counts), the choice of which value to use (for example, the mean or the highest count in each category) did not change the estimated rate of decline across years by more than 2%. Similarly, simulating a potential count from within each bin for each iteration of the model yielded results nearly identical to that obtained by using the mean count. Throughout this report, the results are presented using the lowest count in each bin, since this yields the most conservative estimate of decline.

1b. Negative binomial regressions

Each of the remaining datasets, which report true counts of Pycnopodia, were modelled using a GLMM with a negative binomial distribution and a log link function:

(4)

$\mathrm{Negative\; Binomial}(y_{i,s,t}|{\mu }_{i,s,t},\varphi )$

(5)

${\mu }_{i,s,t}=e^{({\alpha }_{i,t}+{\alpha }_{i,s}+{\beta }_i{X}_i)}$

or a zero-inflated negative binomial regression, if the posterior predictive check of the standard negative binomial model substantially underpredicted the number of zeros in the data (Kéry and Royle 2015):

(6)

$ZI-\mathrm{NegBin}(y_{i,s,t}|ϴ,{\text{µ}}_{i,s,t},\varphi )=\{\begin{array}{l}ϴ+(1-ϴ)\ast \mathrm{NegBin}(0|{\mu }_{i,s,t},\varphi )\mathrm{if}{\mathrm{ y}}_{i,s,t}=0\\ (1-\text{ϴ})\ast \mathrm{NegBin}(y_{i,s,t}|{\text{µ}}_{i,s,t},\varphi )\mathrm{if}{\mathrm{ y}}_{i,s,t})>0\end{array}$

(7)

${\mu }_{i,s,t}=e^{({\alpha }_{i,t}+{\alpha }_{i,s}+{\beta }_i{X}_i)}$

As with the ordinal logistic regressions, ${\alpha }_{i,t}$ functioned as an index of abundance that was comparable across years, even when the sampling effort varied.

1c. Logistic regressions

For the probability-of-sighting model, all datasets were modelled together using a Bernoulli probability mass function:

$\mathrm{Bernoulli}\left(y_{i,s,t}\right|{ϴ}_{i,s,t})={\mathrm{logit}}^{-1}({ϴ}_{i,s,t})$

$$\Theta {}_{i,s,t}=\alpha {}_t+\alpha {}_s+\alpha {}_d+\beta {}_{\mathrm{time}}*x{}_{\mathrm{time}}+a{}_{\mathrm{time}}$$ (when survey effort is measured in time)

$\Theta {}_{i,s,t}=\alpha {}_t+\alpha {}_s+\alpha {}_d+\beta {}_{\mathrm{area}}*x{}_{\mathrm{area}}$ (when survey effort is measured in area)

A univariate approach (combining all the datasets and modelling them as a single time series) was chosen for the probability-of-occurrence analysis because of the difficulty in applying scaling terms in the logit space. Instead, a random effect ${\alpha }_d$ accounted for variation between the datasets. While the DFO datasets and most of the small datasets used area as a metric of survey effort, the remaining datasets used time, making it necessary to partition the data into two observation models, one with a fixed effect of time and the other with a fixed effect of area. Both models contained the same ${\alpha }_t$, ${\alpha }_s$, and ${\alpha }_d$ parameters so that all data contributed to the estimates of year-, site-, and dataset-specific trends. Although both time and area were standardized, the mean survey effort across the surveys that used time as a metric of effort might not be equivalent to the mean survey effort of the surveys that used area. Consequently, we included a scaling term, ${\alpha }_{\mathrm{time}}$, to account for differences between the two types of surveys.

2. Multivariate state-space model

The annual abundance estimates from the GLMMs (${\alpha }_{i,t}$) reflected an underlying population state $X_t$ (that is, the estimated index of the true population) after accounting for measurement error. A single underlying population state was modelled, as Pycnopodia in Canada were considered to belong to a single designatable unit, with no evidence for large-scale genetic structure. Each dataset had its own independent annual estimates of measurement error ($v_{\mathrm{dataset},t}$, which accounted for sources of variation such as missed observations or targeting sites that were unrepresentative of the true population), each of which was assumed to be drawn from a normal distribution centred on 0 with variance $r_{\mathrm{dataset}}$. This underlying population state itself was derived from a first-order autoregressive process, in which the population size at a given year $t$ (that is, $X_t$) differed from the previous year’s population size (that is, $x{}_{t-1}$) by an average annual change $u_{\mathrm{period}}$ (representing the long-term, consistent trend across a given time period) and an annual process noise $W_t$ (representing year-specific demographic fluctuations). This annual process noise was assumed to be drawn from a normal distribution centred around 0 with variance $q$. The average annual change $U_{\mathrm{period}}$ represented a vector of three different values: one for the years prior to the start of the sea star population crash, in 2014, one during the years of the crash in Canada, from 2014 to 2015, and one after the crash ended, from 2015 to 2021, the last year of surveys considered here. This means that the extreme declines observed during the crash years did not artificially inflate the estimate of $q$ and therefore did not lead to an underestimation of observation error. It also allowed for assessments of how stable the population was prior to the crash as well as of any signs of recovery or further decline following the crash.

(8)

$x_t=x_{t-1}+u_{\mathrm{period}}+w_t$

$w_t\sim N(0,q)$

(9)

${\left[\begin{array}{c}{\alpha }_{\mathrm{REEF}}\\ {\alpha }_{\mathrm{OW}}\\ {\alpha }_{\mathrm{small}}\\ {\alpha }_{\mathrm{DFOa}}\\ {\alpha }_{\mathrm{DFOm}}\end{array}\right]}_t=x_t+\left[\begin{array}{c}0\\ a_1\\ a_2\\ a_3\\ a_4\end{array}\right]+{\left[\begin{array}{c}{v}_{\mathrm{REEF}}\\ v_{\mathrm{OW}}\\ v_{\mathrm{small}}\\ v_{\mathrm{DFOa}}\\ v_{\mathrm{DFOm}}\end{array}\right]}_t$

$\left[\begin{array}{c}{v}_{\mathrm{REEF},t}\sim N(0,r_{\mathrm{REEF}})\\ v_{\mathrm{OW},t}\sim N(0,r_{\mathrm{OW}})\\ v_{\mathrm{small},t}\sim N(0,r_{\mathrm{small}})\\ v_{\mathrm{DFOa},t}\sim N(0,r_{\mathrm{DFOa}})\\ v_{\mathrm{DFOm},t}\sim N(0,r_{\mathrm{DFOm}})\end{array}\right]$

A scaling term $a$ was applied to all but one of the datasets to account for consistent differences between datasets (for example, in cases where certain datasets targeted consistently high-density sites or had consistently higher sampling effort). The REEF dataset was chosen as the baseline as it was the dataset with the largest number of surveys, encompassed the largest geographic range of survey sites, and was not solely targeting Pycnopodia (to avoid overestimates of decline caused by site-selection bias; Fournier et al. 2019). Consequently, the underlying population states and all annual estimates of abundance presented in this assessment are scaled to the REEF dataset for ease of visualization.

A univariate version of the state-space model described above was used for the probability of occurrence analysis to estimate the combined observation error across datasets:

$x_t=x_{t-1}+u_{\mathrm{period}}+w_t$

$w_t\sim N(0,q)$ ,

where $u_{\mathrm{period}=1}$ when year < 2014,

$u_{\mathrm{period}=2}$ when year = 2014 or 2015,

and $u_{\mathrm{period}=3}$ when year> 2015

${\alpha }_t=x_t+v_t$

$v_t\sim N(0,r)$

Due to the rapid and well-defined period of the population decline, the mean pre- and post-crash indices of abundance and the percent reduction between them were calculated once for each iteration of the model. The median, 2.5%, and 97.5% quantiles of the reduction were then drawn from the resulting distribution to determine the total estimated population reduction and the 95% credible interval.

The entire model was run in Stan, version 2.32.2 (Stan Development Team, 2023), using the cmdstanr package (Gabry et al., 2023) in R, version 4.3.1 (R Core Team, 2023), with four chains and 3,000 post-warmup iterations for each chain. Model convergence was assessed using Gelman-Rubin statistics with a threshold of 1.01, visual inspection of trace plots, and effective sample sizes greater than 1,000 (Gelman et al., 2013). The models were examined for fit with posterior predictive checks and an examination of the residuals using the DHARMa package (Hartig 2022). Priors for each parameter were normally distributed and nondirectional for all parameters that could be negative or positive (for example, fixed effects).

For parameters that had to be positive, either a gamma distribution (for standard deviations) or inverse gamma distribution (for dispersion parameters) was used to constrain model estimates above zero. For the cut points in the ordinal logistic regression, an induced Dirichlet prior was used to ensure that they remained ordered. All priors were chosen to be weakly regularizing (that is, allowing the data to inform the estimates more than any prior expectations), and the sensitivity of the model results to the chosen priors was tested by comparing the final model to one with flatter priors. In all cases, prior choice did not substantially affect the estimated year-to-year changes or any of the parameter estimates, so the slightly regularizing priors—which constrain the sampler enough to avoid problems with divergent transitions and reduce the time needed to run the model—were chosen instead of flat priors. A comparison of model results using flat priors and using regularizing priors is detailed in Appendix 5.

See Hannah V. Watkins, Pycnopodia helianthoides COSEWIC Report for annotated code.

Advantages of modelling approach

The modelling approach used here better quantifies uncertainty around estimated declines than previous approaches for two reasons. First, it takes into account the non-normal distributions of data and, second, it avoids aggregating data as much as possible, which allows for the inclusion of more survey-specific parameters in each observation model, thereby reducing uncertainty around the mean annual estimates. Previous estimates of Sunflower Sea Star declines across the global range were either not accompanied by estimates of uncertainty or presented uncertainty as standard errors of means calculated from pre- versus post-wasting data that were not normally distributed and aggregated across many years. This resulted in standard errors that were more than half the size of the estimates themselves (and confidence intervals on pre-wasting abundances that would have crossed 0). The method used here also accounts for both variation in data sources and in sites (and a handful of other survey-specific factors), which makes the percent decline reported here more reliable than previous estimates.

Our modelling approach also treats the data as a time series, rather than present single pre- and post-wasting estimates (for example, Appendix 4). The advantage of the time series is that visualizing the relative stability of the population pre-wasting and estimating a single pre-wasting trend, $u_{\mathrm{period}=1}$, helps justify the claim that the population was doing consistently well in the long term and was not typically subject to large annual fluctuations.

Finally, because the REEF and OW abundance data, which are recorded on an ordinal scale, could be used in a quantitative manner in this framework, the sample size underpinning the abundance change estimate is much higher than previous attempts (but similar for the probability-of-occurrence analysis, except for the additional surveys of Simon Fraser—Côté and DFO). Previous estimates of abundance decline for Canada were based on 1,642 to 1,978 surveys (an unknown number of which were made in the U.S. portion of the Salish Sea). By contrast, the present abundance analysis includes 9,124 surveys conducted exclusively in Canada.

Appendix 4. Unscaled dataset-specific trends in abundance

Figure S2. Annual index of population abundance for Sunflower Sea Star in Canada over the past two decades (2000 to 2021), based on the REEF Invertebrate and Algae Monitoring Program (n = 4,711 surveys) and Ocean Wise (OW) Pacific Marine Life Surveys Database (n = 2,281 surveys), DFO-Abalone surveys (n = 994 surveys), DFO-Multispecies surveys (n = 521 surveys), and the remaining small datasets from the IUCN assessment (n = 617 surveys; total observations = 9,124 surveys). Points and lines represent the annual abundance estimate from each dataset (that is, the 𝛼_𝑡 terms in the ordinal logistic regression, defined in Appendix 2) and are scaled to the dataset-specific mean sampling effort (that is, 49.12 minutes for REEF, 45.59 minutes for OW, 21.44 m²for DFO-Multispecies, 15.19 m²for DFO-Abalone, and 189.24 m²for the combined small datasets). Shaded areas are 95% credible intervals (that is, the 2.5% and 97.5% quantiles of the estimates for each year across all post-warmup model iterations). The number of surveys conducted each year in each dataset is shown at the bottom of each panel. Note that the scale of the y-axis is different in each panel. The onset of SSWD in Canada (that is, 2014) is marked with a vertical bar in each panel. Note that some surveys conducted in 2014 occurred before the onset of SSWD.

Appendix 5. Sensitivity analysis of priors

*  Note that priors were applied to the standard deviations on the process noise and observation error rather than to their variances.

** See GitHub code for full function details.

Note: Nearly all parameter estimates in the model with flat priors are unreliable due to sampling issues. The model is too complex to estimate individual parameters reliably without setting realistic constraints on variables through regularizing priors, although it can roughly capture the overall trends in the time series.

Information sources - Appendix 2, 3, 4 and 5

Fournier, A.M.V., E.R. White, and S.B. Heard. 2019. Site-selection bias and apparent population declines in long-term studies. Conservation Biology 33:1370 to 1379.

Gabry, J., and T. Mahr. 2022. bayesplot: Plotting for Bayesian Models.

Gabry, J., R. Češnovar, and A. Johnson. 2023. cmdstanr: R Interface to “CmdStan”.

Gelman, A., J.B. Carlin, H.S. Stern, D.B. Dunson, A. Vehtari, and D.B. Rubin. 2013. Bayesian Data Analysis, Third Edition. CRC Press. 675 pp.

Greenberg, D.A., C.V. Pattengill-Semmens, and B.X. Semmens. 2024. Assessing the value of citizen scientist observations in tracking the abundance of marine fishes. Conservation Letters 17: e13009.

Hartig, F. 2022. DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. R package version 0.4.5.

Holmes, E.E., M.D. Scheuerell, and E.J. Ward. 2021. Applied time series analysis for fisheries and environmental sciences.

Kéry, M., and J.A. Royle. 2015. Applied Hierarchical Modeling in Ecology: Analysis of Distribution, Abundance and Species Richness in R and BUGS. Volume 1: Prelude and Static Models. Academic Press. 808 pp.

R Core Team. 2023. R: A Language and Environment for Statistical Computing.

Stan Development Team. 2023. Stan Modeling Language Users Guide and Reference Manual, Version 2.29.2.

Ward, E.J., R. Hilborn, R.G. Towell, and L. Gerber. 2007. A state–space mixture approach for estimating catastrophic events in time series data. Canadian Journal of Fisheries and Aquatic Sciences 64:899 to 910.