The authors have declared that no competing interests exist.
About 62,000 dead or dying common murres (
Marine heatwaves (hereafter “heatwaves”), defined as prolonged periods where ocean temperatures are much warmer than usual [
During late 2013, a warm temperature anomaly developed in near-surface (upper ~100m) waters well offshore in the Gulf of Alaska (GOA) which grew to encompass a large area of the northeast Pacific Ocean [
Some immediate biological effects of this unprecedented heatwave were equally extreme. For example, phytoplankton production in the central north Pacific waters was greatly reduced [
We report on another extreme biological impact of the 2014–2016 heatwave: The widespread die-off and chronic reproductive failure of a trophically dominant piscivorous marine bird, the common murre (
Here we document the magnitude of the 2015–2016 murre die-off in terms of its spatial extent, duration, absolute numbers of dead or dying birds recovered, and relative magnitude of deposition on beaches relative to long-term baselines. We used data collected by systematic and repeated surveys for beached birds conducted by citizen science participants in the northeast Pacific [
We used sea surface temperature (SST) data from the Hadley Center Sea Ice and Sea Surface Temperature (HadISST) data set [
Specimens that were salvaged for necropsies and testing for disease, biotoxins, etc., were collected under U.S. Fish and Wildlife Service (USFWS) permit, i.e., the Migratory Bird Master Permit/Import-Export (MB025076-0) and State of Alaska Department of Fish and Game (ADF&G) Scientific Permit issued to the USFWS Regional Director. U.S. Geological Survey personnel operated under the same permits, as issued to the Alaska Biological Science Center USFWS (MB789758-2) and the ADF&G.
Effort-standardized surveys for beachcast marine birds at monthly or more frequent intervals were conducted at predetermined survey sites by participants of three beached bird citizen science programs: Beach Coastal Ocean Mammal/Bird Education and Research Surveys (BeachCOMBERS) [
Sources: COASST- Coastal Observation and Seabird Survey Team, DOI- Department of the Interior agencies, B.Watch- Beach Watch, COMBERS- BeachCOMBERS.
Location | Standardized beach surveys | Number of murres reported | |||||||
---|---|---|---|---|---|---|---|---|---|
Source | Start | #sites | #surv | #km tot | Surveys | Public | Rehab | TOTAL | |
Chukchi | COASST | 2006 | 5 | 25 | 47 | 14 | 0 | 14 | |
Aleutians | COASST | 2006 | 7 | 34 | 39 | 10 | 200 | 0 | 210 |
Bering Sea | COASST | 2006 | 9 | 110 | 114 | 23 | 300 | 0 | 323 |
Gulf of Alaska | COASST | 2006 | 52 | 423 | 359 | 4289 | 0 | 4289 | |
Gulf of Alaska | DOI | 2015 | 114 | 164 | 381 | 20240 | 21435 | 552 | 42227 |
SE Alaska | COASST | 2006 | 11 | 91 | 124 | 11 | 383 | 394 | |
Salish Sea | COASST | 1999 | 210 | 1771 | 2075 | 10 | 10 | ||
N Washington | COASST | 1999 | 37 | 310 | 492 | 884 | 24 | 908 | |
S Washington | COASST | 1999 | 41 | 381 | 564 | 914 | 108 | 1022 | |
N Oregon | COASST | 2001 | 55 | 476 | 645 | 1249 | 332 | 1581 | |
S Oregon | COASST | 2001 | 34 | 232 | 345 | 353 | 3 | 356 | |
N California | COASST | 2006 | 47 | 403 | 428 | 357 | 52 | 409 | |
C California | B.Watch | 1994 | 40 | 975 | 1516 | 2927 | 680 | 3607 | |
SC California | COMBERS | 1994 | 46 | 403 | 1125 | 5071 | 1614 | 6685 | |
TOTALS | 708 | 5798 | 8253 | 36352 | 22318 | 3365 | 62035 |
All carcasses found were marked or removed to prevent recounting. For analysis of carcass deposition per effort (km walked), we divided program surveys into two categories: event (May 1, 2015 to April 30, 2016) and baseline (all surveys prior to May 2015). Because each program was created at different points in time (
Additional surveys were undertaken in Alaska, including 164 standardized [
Specimens were collected during the study period (May 1, 2015 to April 30, 2016) and examined in three different efforts:
We contacted 72 bird rescue and rehabilitation (“rehab”) centers from southern California to Alaska (
Common murre carcasses were collected from multiple coastline locations along the GOA (n = 89) and southern Bering Sea (n = 14) and shipped to the NWHC for diagnostic examination by American Veterinary Medical Association certified pathologists [
Fresh (see above) common murre carcasses (n = 117) collected off Alaskan beaches (n = 88) between 1 November 2015 and 11 April 2016 or provided by Alaska rehab centers (n = 29), were necropsied by agency biologists in Anchorage, Alaska (
Reproductive success (
Under a wide range of prey densities, average common murre
To illustrate spatial magnitude of the event, carcass count data of murres on both systematic and opportunistic beaches sampled, and records of live and moribund murres rescued from point locations were compiled for the event period (May 1, 2015 to Apr 30, 2016). Raw data were binned into 75x75 km cells for mapping; a scale big enough to aggregate higher resolution (km) beach surveys or point samples and prevent excessive overlap of adjacent abundance circles, but small enough to track coast and island geography (individual beach survey sample sites are plotted in
Deposition of carcasses on beaches is a measure that incorporates both detection and persistence and is proxied by carcass encounter rate. To examine the seasonal variation of relative carcass abundance within and among large geographic regions, raw beach survey data from citizen science programs (above) were standardized to carcasses encountered per linear km of coastline by survey [
In order to determine whether carcass encounter rates in 2015/16 were significantly higher than previous years we calculated bootstrap 95% confidence intervals of mean encounter rate at the region and month-year scale. Each bootstrap estimate was calculated by drawing n samples (with replacement) of survey-specific encounter rate from the pool of available surveys for that month-year and region (Gulf of Alaska, Outer coast of Washington, Oregon, N California, C California and SC California–see
The average for each calendar month (i.e. the baseline) was then calculated by a second round of bootstrap resampling using the distributions generated in the previous step (
The observed warming in the GOA from winter 2014 through winter 2016 was unprecedented in the period since instrumental record-keeping began (1870–2017;
Average monthly time series (Jan 1870 to Dec 2018) of sea surface temperature anomalies (SSTa) in the (a) Gulf of Alaska (GOA); and in (c) northern (nCCS); (e) central (cCCS); and (g) southern (sCCS) waters of the California Current System (CCS). Solid black lines are 24-month running averages. Also presented are 5-year running averages of annual SST differences, an index of SST change rate, for the (b) GOA, and (d) nCCS, (f) cCCS and (h) sCCS. Average long-term SST values for each region are: GOA 8.1° C, nCCS 12.5° C, cCCS 13.9° C, sCCS 16.2° C.
During the event year (May 2015 to April 2016) ~62,000 murre carcasses were reported from a vast range of coastline spanning more than 6,000 km (
Numbers of dead or moribund common murres observed on beaches that were surveyed systematically (gold circles; ~monthly) and with opportunistic beach surveys and rehab captures (red circles). Areas in which zero dead murres were encountered during surveys are indicated by white circles. All remaining coastlines (without any circles) were not surveyed. Note the California Current System is divided roughly into 3 sections: north (nCCS), central (cCCs) and south (sCCS).
Encounter rates from southcentral California to the GOA were significantly elevated relative to baseline (
Monthly averaged encounter rates (carcasses per km, gray bars) for the (A) Gulf of Alaska, (B) Washington, (C) Oregon, (D) northern California, (E) central California, and, (F) southern California coastlines. Black lines are baseline encounter rates, yellow shadings are 95% confidence intervals, and red lines show relative magnitude of encounter rates in 2015/2016 compared to the 2006–2015 baseline. Colored squares indicate whether month-averaged encounter rate was significantly higher than baseline and whether they were significantly higher than any prior year of data collection for that calendar month. Asterisks indicate that the mean encounter rate in the corresponding month was the highest value ever recorded. Note that the GOA baseline is so low that it cannot be seen on the chart. Relative magnitude was calculated as the 2015/2016 encounter rates divided by the baseline mean value.
In the CCS, murre carcass encounter rate is typically about 1–3 carcasses per km during late summer and early fall (July-October;
Of the common murre carcasses collected in Alaska and necropsied at the National Wildlife Health Center, 79% were AHY (68% adult,
Body mass of common murres collected: (a) at seven breeding colonies in the Gulf of Alaska (GOA), May-September 1988–1999; (b) after a large die-off of murres on the Kenai Peninsula, GOA, February-March 1993; (c) at scattered locations in the GOA and Bering Sea, May 2015-April 2016; (d) along the central coast of California (37°-42° N), May 2015-April 2016; and, (e) along the southern coast of California (32°-37° N), March 2015-April 2016. The vertical dashed line represents the approximate critical mass below which mortality is expected in starved common murres (phase III starvation). The vertical dotted line indicates the cutoff mass below which birds were likely young-of-the-year fledglings.
From subsamples necropsied at the ASC, we determined that 67% of birds were female, 89% were AHY birds (64% adult), and 11% were HY. Nearly all (97%) birds examined for pectoral condition were scored as emaciated and of those scored for subcutaneous fat most had zero (83%) or very little (17%). Only 7% of birds (n = 100) had food remains (trace amounts, mostly bone fragments) in the gizzard. Mass of all birds (n = 97) averaged 715.2 ±79.9 SD. Overall, body mass of murre carcasses collected during the event year were comparable to that of murre carcasses measured (n = 116 AHY, mean mass = 666.0 ±92.4 SD) during a similar die-off in Prince William Sound in March 1993 [
Of 3,365 murres examined at rehabilitation centers, 8% were dead on arrival or euthanized immediately; 47% were described primarily as emaciated; 5% were injured in some way (e.g. broken wing); and 5% were oiled. The remainder (35%) had non-specific information (e.g., beached, sick, weak). The frequency of these conditions varied from south to north, with 91–99% of birds classified as “starving” in Oregon, Washington and Alaska, while “other” causes increased in frequency in southcentral California (13%) and northcentral California (25%). Birds received at rehab centers in California exhibited a bimodal pattern of mass distribution (
Just under one quarter of the Alaskan common murre population resides on colonies regularly monitored for attendance and reproductive success (
Annual standardized deviations in reproductive success of common murres (on right) at continuously monitored colonies (on left, large red circles) distributed over ~6000 km in the NE Pacific Ocean. Red bars (on right) indicate every known year of reproductive failure (i.e., success 0.0 to <0.10 chicks/pair) from 1985 to 2014. Horizontal dashed lines indicate where reproductive success would fall two standard deviations below the long-term average (for pre-event average during 1996–2014, value at right of each plot). Smaller and/or irregularly monitored colonies (on left, small red circles) included (a) Round I., (b) Aiktak I., (c) Oil Creek, (d) Duck I., (e) Nord I., (f) Gull I., (g) Barwell Is., (h) Devil’s Slide Rock.
During the 2015 breeding season, two annually monitored colonies (Chowiet, Amatuli) failed completely, as did one occasionally monitored site: Gull Island (pre-2015 mean = 0.54 ch/p). In the GOA, these three sites comprise ~26% of the common murre population. Only two regularly monitored colonies reported an above average
In 2016, reproductive failures expanded in the GOA and Bering Sea. Seven of eight annually monitored colonies in the GOA and Bering Sea, and five intermittently monitored colonies in the GOA (Barwell, Nord, Gull, and Duck islands; Oil Creek) failed completely (0 ch/p). Of regularly monitored colonies, only Chowiet in the Semidi Islands (GOA) produced fledglings (0.48 ch/p) but far fewer birds attempted to breed (39% of pre-2015 high count of 4283 murres on monitoring plots). Additional signs of reproductive difficulties during 2016 in the GOA included late egg-laying (if it occurred at all), irregular attendance and total abandonment at some colonies.
Reproductive difficulties continued in 2017, as four out of seven regularly monitored colonies failed completely (Cape Pierce, Round, Aiktak, St. Lazaria), and two others (St. Paul 0.02 ch/p, Amatuli 0.15 ch/p) experienced failure or unusually low success (below 2 SD of long-term mean (
In summary, 13 common murre colonies in the GOA and Bering Sea experienced a complete failure (0 ch/p) in reproduction at least once during the event years (2015–2017). Multi-year failures were documented in 8 colonies. Out of 31 colony-years of
There is evidence that common murres in the CCS also experienced depressed reproductive success, albeit to a lesser degree and somewhat lagged in comparison to Alaska. Reproductive success in the CCS population is usually stable, as informed from three decades of monitoring (1985–2014) at 5 colonies representing ~64% of the CCS population (
The 2015–2016 common murre die-off in the northeast Pacific is unprecedented globally in magnitude, spatial extent and duration. It occurred during a heatwave that was also severe (Category III) in magnitude, spatial extent and duration (711 days, [
To put numbers into perspective with other mass mortalities in Alaska, biologists counted 22,800 emaciated (average 704 g) murre carcasses along an ~700 km stretch of coast on the southeast Bering Sea following a severe storm during April 1970 [
Few birds were recovered on beaches in British Columbia or southern southeast Alaska (
A few exceptionally large die-offs have also occurred elsewhere in the world. During winter of 2013–2014, a total of 54,982 seabirds, mostly (54%) Atlantic puffins (
Counts of dead seabirds on beaches following mass mortality incidents represent a minimum measure of total mortality. They do not include carcasses that sink at sea, or those washed ashore that are removed by scavengers or buried in sand and debris. Furthermore, the frequency and thoroughness of beach surveys ultimately determines how many carcasses will be discovered and counted [
In this study, owing to logistic constraints and geographic expanse, we measured few or none of the factors needed to model total mortality in the areas most affected. However, we can draw upon a comparable study of carcass counts [
On the West Coast, we don’t have a comparable model to estimate total mortality. However, we know that beach survey coverage was more comprehensive on the West Coast, and so we used the conservative range of expansion factors from experimental studies (above, [
Taken together, the total impact of the heatwave on common murre populations throughout all areas was likely between 0.53 and 1.2 million birds, or approximately 10–20% of total populations (~6 million). The fact that most birds killed in the die-off were probably breeding adults compounds the seriousness of the mortality for the population [
The extreme reproductive failures of common murres that occurred during summer 2015 and in the two years after the main die-off were also focused in Alaska and occurred less frequently in the California Current System (CCS). Considering the low number of birds encountered on beaches in the Bering Sea, breeding failures at colonies there were surprisingly similar in magnitude to those in the Gulf of Alaska (GOA). This reduction in juvenile production will significantly delay recovery of populations in all affected colonies [
The frequency of total reproductive failures (n = 22), overall reduced breeding success and decline in numbers that occurred at multiple colonies in the northeast Pacific during 2015–2017 is a cause for astonishment and alarm. The common murre is probably the most widely studied seabird in the Northern Hemisphere and total reproductive failures at well-established colonies have been rare during some 70+ years of detailed observations (
Several acute biological responses to this unprecedented heatwave were observed throughout the northeast Pacific. Phytoplankton biomass in the northeast Pacific transition zone waters was lower in winter 2014 than in any year measured since 1997 [
A common thread to most of these events was that they involved either a loss in productivity or a mass mortality of higher trophic-level animals, both of which point to problems in food production or availability. All the vertebrate predators affected also share a common dietary dependence on a few key forage species (see below) and this points to a bottleneck in the forage base. These events all occurred within, and for some years after, the time-frame of the 2014–2016 heatwave, and over an enormous spatial range involving three large marine ecosystems (CCS, GOA and Bering Sea). This calls for an explanation that is plausible for all species and regions, and that involves water temperature as a driving force—either directly or indirectly. With respect to murres, we offer three non-exclusive hypotheses to explain the cause of these events: 1) temperature-mediated changes in the distribution and quality of the prey base available to murres; 2) harmful algal blooms associated with warm water anomalies; and, 3) temperature-enhanced competition from ectothermic predators.
Reduction in primary production, and ultimately zooplankton or forage fish biomass, has been implicated in past seabird die-offs and reproductive failures (e.g., [
On the other hand, as endotherms living in hostile, cold environments, murres maintain high metabolic rates (2.14 kJ/g/d [
This is the ultimate “Achilles heel” for murres, and one that sets it far apart from competing ectothermic groundfish (eating typically 0.1–1% of BM/d) and endothermic marine mammals [
Warming of subarctic shelf waters may lead to both vertical (deepening) [
In the Gulf of Alaska (GOA), a similar introgression of smaller zooplankton was observed, along with a breakdown in established SST-phytoplankton-zooplankton dynamics after 2013 [
In addition to shifts in latitudinal abundance of specific taxa, warm water conditions diminished body condition and somatic growth of ectothermic forage fish. Body condition of capelin and sand lance in the GOA and CCS was reduced during the heatwave, resulting in smaller, less energy-dense prey for murres [
We propose that an unusually warm layer of water in the NE Pacific, persisting for more than 2 years, had a powerful cumulative effect on ectothermic groundfish (stimulating food intake rate) and ectothermic forage fish (reducing their quality) leading to a strong top-down and bottom up (vise-like) impact on murre survival and reproductive success.
Increased ocean temperatures during and following the heatwave have been associated with harmful algal blooms (HABs) [
During the common murre mass mortality event, an extensive HAB of a toxigenic diatom (
Furthermore, the likelihood that HAB toxins were a primary and acute causal factor in the die-off appears small given that the center of the murre mass mortality event was the GOA, and the extended duration of the die-off (9 months of 100x baseline) both preceded and extended well past peak HAB bloom windows [
In addition to affecting spatial distribution of large predatory groundfish [
Recent modeling [
The micronekton most commonly eaten by these groundfish include several species also favored by murres and other avian piscivores, especially capelin, sand lance, juvenile pollock, herring and euphausiids (especially in winter) [
If the GOA marine ecosystem was operating at “relative equilibrium” [
While murres were visibly dying
During the most powerful marine heatwave on record [
Our hypothesis to explain the wide-spread depletion of forage fish is based on three facts: 1) metabolic rate, food intake rate, somatic growth, fat storage, fecundity and survival in marine ectotherms are strongly modulated by water temperature, 2) these physiological and life history traits are adapted to function optimally over relatively small ranges of water temperature that are species-specific, and, 3) physiological efficiency declines markedly when ambient temperatures wander too far above or below the optimal temperature range. This is why fish seek out water with temperatures that optimize these parameters, e.g., by migrating vertically in and out of warm surface layers or migrating geographically to stay within waters that suit their tolerances and/or facilitate various activities (e.g., migration, foraging, predator avoidance, “hibernation”, etc.). Such effects were noted widely in zooplankton, ichthyoplankton, forage fish and groundfish during the heatwave. However, murres and many other marine predators are highly mobile and deep-diving, and it is unlikely that simple distributional changes in forage fish can account for all the widely observed starvation, mortality and breeding failures in murres.
A more plausible hypothesis is that persistently warm water temperatures modulated ectotherm physiology everywhere at the same time, and forage fish were caught in an “ectothermic vise” (
An ectothermic vise on forage fish ought not to be expected under every scenario of ocean warming, and much may depend on how well community thermal optima are aligned with temperature regimes before, during and after a heatwave occurs [
(DOCX)
(DOCX)
(TIF)
(TIF)
(TIF)
(DOCX)
(XLSX)
(XLSX)
Beach surveys and specimen collections were conducted by trained citizen scientists and professional biologists from southern California to Alaska, and we appreciate their diligent and persistent monitoring efforts. Reproductive success data were collected by the FWS Alaska Maritime National Wildlife Refuge (NWR), Togiak NWR, Becharof NWR, San Francisco Bay NWR Complex, Farallon Islands NWR, USGS, Institute for Seabird Research on Conservation, Oregon State University, Humboldt State University and Point Blue Conservation Science (Contribution #2196), and we thank all the staff, interns and collaborators who assisted in these efforts for more than 30 years. We also thank the staff and volunteers at 66 bird rehabilitation centers (