Reviews in Fish Biology and Fisheries

, Volume 22, Issue 1, pp 95–135

Hypotheses concerning the decline and poor recovery of Pacific herring in Prince William Sound, Alaska

  • Walter H. Pearson
  • Richard B. Deriso
  • Ralph A. Elston
  • Sharon E. Hook
  • Keith R. Parker
  • Jack W. Anderson
Open Access
Reviews

DOI: 10.1007/s11160-011-9225-7

Cite this article as:
Pearson, W.H., Deriso, R.B., Elston, R.A. et al. Rev Fish Biol Fisheries (2012) 22: 95. doi:10.1007/s11160-011-9225-7

Abstract

This paper updates previous reviews of the 1993 stock decline of Pacific herring (Clupea pallasi) in Prince William Sound, Alaska, and focuses on hypotheses about subsequent poor recovery. Recent age structured assessment modeling with covariate analysis indicates that the population dynamics of the sound’s herring are influenced by oceanic factors, nutrition, and, most substantially, hatchery releases of juvenile pink salmon. For the 1993 decline, poor nutrition remains the most probable cause with disease a secondary response. Concerning poor recovery, we examined 16 potential factors and found three to be causal: oceanic factors, poor nutrition, and hatchery releases of juvenile pink salmon. Absences of strong year classes at both Sitka and Prince William Sound after 1993 indicate the action of large-scale ocean processes. Beyond regional-scale environmental factors, two factors specific to the sound influence the population dynamics of herring and are likely impeding recovery. First, pink salmon fry releases have increased to about 600 million annually and may disrupt feeding in young herring, which require adequate nutrition for growth and overwintering survival. Juvenile pink salmon and age-1 herring co-occur in nearshore areas of bays in late spring and summer, and available data on dietary overlap indicates potential competition between the age-1 juvenile herring and juvenile pink salmon. Field studies demonstrate that juvenile herring reduce food intake substantially in the presence of juvenile pink salmon. Second, overwintering humpback whales may consume potentially large amounts of adult herring, but further studies must confirm to what extent whale predation reduces herring biomass.

Keywords

Pacific herring Clupea pallasi Fisheries collapse Fisheries recovery Prince William Sound Alaska Exxon Valdez oil spill 

Introduction

Because of the commercial and ecological importance of the Pacific herring (Clupea pallasi) population in Prince William Sound (PWS) Alaska, there is concern for the present and future states of PWS herring. In a comprehensive review, Hay et al. (2001) found that during the last century many herring stocks around the world have collapsed. Intense fishing has often been associated with such collapses, but habitat loss and environmental change have also led to collapses of herring and other species (Hilborn 1997). Recovery of most collapsed herring stocks usually took a decade or more (Hay et al. 2001). For example, after 15 years of no fishing, the Georges Bank herring returned to stock levels above those before the 1977 collapse. Some stocks, however, have not returned to their former abundance even after several decades. For example, the Hokkaido-Sakhalin stock has remained extremely depressed since its decline in the late 1950s. Most collapsed stocks appear to recover, but the present status of some stocks is “worrisome.” One stock considered worrisome is that in PWS (Hay et al. 2001).

PWS herring biomass increased steadily through the 1980s (Fig. 1). Funk (1993, 1994) used ASA (age structured assessment) modeling to predict that the 1993 PWS herring spawning biomass would be a record high of over 121,000 metric tons. However, the predicted biomass failed to materialize (Brown et al. 1994, 1996; Carls et al. 2002). Brown et al. (1994) report that the observed PWS spawning biomass was about 27,000 metric tons in spring 1993 or about 20% of expectation.
Fig. 1

Herring biomass indices for Prince William Sound by calendar year. Data from: ADFG (unpublished data, See Appendix), Ashe et al. (2005), Donaldson et al. (1993), Moffitt (2006)

After the decline, the fishery was closed from 1994 through 1996. The PWS herring stock increased to the point where modest harvests of 3,629 to 4,536 metric tons were permitted in 1997 and 1998. Then, in 1999, only half the predicted PWS biomass was found (Marty et al. 2003). In 1999, the PWS herring fishery was closed and remains closed because of insufficient biomass.

Before the early 1990s, the PWS herring population and fishery were sustained by recruitment of strong year classes about every 4 years (Funk and Sandone 1990; Funk 1994; Williams and Quinn 2000a, b). Recovery of PWS herring would be greatly aided by the recruitment of a highly successful year class. However, no strong year class has emerged in PWS since the 1988 year class recruited in 1991 and 1992 (Fig. 2).
Fig. 2

Number of age-3 herring recruits for Prince William Sound and Sitka, Alaska, estimated by ADFG ASA models by calendar year (Dressel 2006, Moffitt 2006)

The two aims of this paper are (1) to briefly update the review of Pearson et al. (1999) concerning the decline of PWS Pacific herring (Clupea pallasi) stock in 1992–1993 (Table 1) and (2) to assess hypotheses concerning the subsequent poor recovery of the stock. For the first aim, the decline, we focus on information that has become available since the late 1990s, particularly work subsequent to Pearson et al. (1999), which includes studies by Marty et al. (2003, 2004, 2007, 2010) on disease, a comprehensive disease review by Elston and Meyers (2009), and ASA modeling focused on PWS herring (Quinn et al. 2001; Deriso et al. 2007, 2008; Hulson et al. 2008). Deriso et al. (2008) used ASA modeling with covariate analysis to quantitatively assess 19 decline and recovery hypotheses. ASA models have been well described in the literature (Quinn and Deriso 1999; Deriso et al. 2007, 2008) and will not be discussed in detail here. An important aspect of these models is that their statistical nature enables them to be used appropriately to provide estimates of uncertainty and to test hypotheses. Deriso et al. (2008) used 19 different time series of data generally for the period from the early 1970s to 2005 as single and multi-factor covariates in their ASA modeling. Modeling results revealed the need to assess predation and competition in greater detail and led to substantial effort and analysis in those areas in this paper.
Table 1

Hypotheses assessed by Pearson et al. (1999)

ID

Hypothesis

(i)

Three hypotheses postulate that the 1989 oil spill caused the 1993 decline

H1

Acute toxic effects of oil exposure in 1989 caused the decline

H2

Toxic effects of oil lingering in PWS through 1992 or 1993 produced the decline

H2a

Lingering oil affected the herring directly

H2b

Lingering oil affected the herring indirectly by reducing their food supply

H3

Delayed effects of the 1989 oil spill induced a disease that caused the decline

(ii)

Two hypotheses postulate that a harvesting effect caused the decline

H4

Overfishing led to the decline of the PWS herring stock

H5

Underfishing (1989 Fisheries Closure) allowed the herring biomass to increase to the point that density-dependent mechanisms led to the decline

(iii)

Four hypotheses postulate that a series or combination of natural factors caused the decline

H6

Poor nutrition alone or in conjunction with a disease caused by an opportunistic pathogen produced substantially lower herring survival that in turn caused the decline

H6a

Recently increasing herring biomass increased intraspecific competition for prey (a density dependent response)

H6b

Recently increasing biomass of other fish increased interspecific competition for prey

H6c

Large-scale oceanic processes reduced food supply that in turn lowered the nutritional status of PWS herring

H7

An infectious disease caused the decline

H7a

Disease alone caused the decline

H7a

Low temperatures triggered a disease outbreak that in turn reduced survival

H8

Increased predation reduced the population level

H9

Natural chaotic population dynamics produced the decline, perhaps by a stochastic process in which small changes in the physical or biological environment produced a large, abrupt change in the PWS herring population

Identifying the hypotheses

We have identified hypotheses from literature reviews and have distinguished the hypotheses about the decline from those about poor recovery (Table 2). Herring have supported important fisheries worldwide for centuries (Hay et al. 2001; Stephenson 2001), and high variability in abundance has been known for decades to be a common and inherent feature of herring stocks (Hjort 1914; Smith 1988; Rothschild 1986; Southward et al. 1988; Hay et al. 2001). Prominent among the various causes proposed for herring stock fluctuations have been the influence of ocean climate on recruitment processes and the effects of intense fisheries (Hay et al. 2001; Stephenson 2001). Unlike other declines that involve recruitment failure (Hilborn 1997), the PWS 1992–1993 decline was a one-time event with an acute increase in adult mortality, not a recruitment failure (Pearson et al. 1999; Quinn et al. 2001; Carls et al. 2002; Deriso et al. 2008; Hulson et al. 2008). Hypotheses concerning the decline must provide mechanisms that increase mortality in all adult age classes between the spring of 1992 and the spring of 1993. Mechanisms governing poor recovery cannot be assumed to be the same as those for the decline. Hypotheses for poor recovery must provide mechanisms for (1) no recruitment of strong year classes after the 1988 year class recruited and (2) continued low herring biomass.
Table 2

List of hypotheses concerning the decline of PWS herring and subsequent poor recovery

 

Decline hypotheses

Narrative

A

 1

Population dynamics

Natural chaotic population dynamics produced the decline, perhaps by a stochastic process in which small changes in the physical or biological environment produced a large, abrupt change in the PWS herring population

 2

Harvest—overfishing

Overfishing of PWS herring led to the decline

 3

Harvest—fisheries closure

The 1989 fishing closure led to high biomasses that exceeded the carrying capacity of PWS for herring

 4

Spawning habitat Loss

Loss of spawning habitat led to the decline

 5

Disease and I. hoferi

Mortality from disease from parasite I. hoferi caused the decline

 6

Disease and VHSV

Mortality from disease from pathogen VHSV caused the decline

 7

Disease and EOK pound fishery

Confinement in net pens for EOK fishery induced VHS and disease mortality led to the decline

 8

Oil Exposure and VHSV

Oil exposure induced VHS and disease mortality led to the decline

 9

Oil Exposure Alone

Oil exposure alone directly caused mortality and the decline

 10

Competition or predation by pink salmon

Increased releases of hatchery pink salmon fry increased predation and/or competition with adult herring and led to the decline

 11

Predation by sea lions

Mortality from predation by sea lions led to the decline

 12

Poor nutrition

Poor nutritional state of adult herring caused the decline

 13

Ocean factors

Changes in ocean factors caused the decline

 

Recovery hypotheses

 

B

 1

Population dynamics

Natural chaotic population dynamics are keeping PWS herring at a low level, perhaps by a stochastic process in which small changes in the physical or biological environment moved the PWS herring population to a new set point

 2

Harvest—overfishing

Overfishing of PWS herring impairs recruitment and recovery

 3

Harvesting—fisheries closure

Fishing closures impair recruitment and recovery

 4

Spawning habitat loss

Loss of spawning habitat impairs recruitment and recovery

 5

Behavioral conservatism

Behavioral conservatism impairs adjustment to environmental changes and reduces recruitment and recovery

 6

Disease and VHSV

Disease mortality from VHSV impairs recruitment and recovery

 7

Disease and I. hoferi

Disease mortality from I. hoferi impairs recruitment and recovery

 8

Disease and EOK Pound Fishery

Disease mortality from VHSV induced by net pen confinement in EOK fishery impairs recruitment and recovery

 9

Oil Exposure and VHSV

Exposure to residual oil in PWS induces VHS, which impairs recruitment and recovery

 11

Whales as predator in predator pit

Predation by humpback whales has PWS herring in a predator pit

 12

Predation by adult coho salmon

Mortality from predation by adult coho salmon impairs recruitment and recovery

 13

Predation by adult pink salmon

Mortality from predation by adult pink salmon impairs recruitment and recovery

 14

Predation by juvenile pink salmon

Mortality from predation by pink salmon fry released from hatcheries impairs recruitment and recovery

 15

Poor nutrition

Poor nutritional conditions in PWS larvae and juveniles impair recruitment and recovery

 16

Ocean factors

Changes in ocean factors have changed the recruitment processes and their outcomes

 17

Competition with juvenile pink salmon

Competition with pink salmon fry released from hatcheries produced food limitation that impairs growth and survival of juvenile herring and impairs recruitment and recovery

Hypotheses for the decline of PWS herring

The hypotheses concerning the decline and acceptance and rejection of certain hypotheses have evolved substantially since 1993. The ASA models of Quinn et al. (2001), Hulson et al. (2008), and Deriso et al. (2008) all require an abrupt increase in mortality in 1992–1993 to achieve reasonable model fits. Any hypothesis for the decline must explain this abrupt increase in adult mortality between 1992 and 1993. We briefly describe the evolution of the hypotheses below and assess the two hypotheses, oil exposure from the 1989 Exxon Valdez oil spill (EVOS) and disease, about which some debate continues.

At the time of the decline, the major hypotheses concerning its cause(s) centered on two factors: oil exposure and disease. After identifying viral hemorrhagic septicemia virus (VHSV) in Pacific herring, Meyers et al. (1994) concluded that the role of VHSV in the 1992–1993 decline could not be established with the 1993 data. Brown et al. (1996) conjectured that the decline was due to an epizootic, and Marty et al. (1998) implicated VHSV disease as the cause of the 1992–1993 decline.

Later, Pearson et al. (1999) reviewed 13 hypotheses concerning the 1993 decline (Table 1). These 13 hypotheses fell into groups related to the EVOS, harvest, and natural processes, especially disease. Pearson et al. rejected hypotheses concerning oil-spill effects based on the high population levels following the spill, the lack of change from the expected age-class structure, and the low level of oil exposure following the spill. They also found no evidence to support hypotheses about harvesting effects and judged hypotheses concerning water temperature, predation, and competition to have insufficient information to accept or reject them. Condition, as weight at specific length, is a measure of nutritional state of fish (Winters and Wheeler 1994; Cardinale et al. 2003). The 1993 mortality was preceded by sharply declining condition (weight at specific length) in PWS herring from about 1986 (Fig. 3) and by lower zooplankton abundance (Fig. 4). Pearson et al. concluded that poor nutrition, either alone or in combination with disease or other natural factors, probably caused the decline.
Fig. 3

Herring condition as weight at specific length in spring for Prince William Sound and Sitka by calendar year. Weight at specific length was calculated following Winters and Wheeler (1994). Data from: ADFG (unpublished data, See Appendix), Biggs et al. (1992), Brady et al. (1987, 1988, 1990, 1991a, b), Donaldson et al. (1992), Dressel (2006), Sandone (1988)

Fig. 4

Settled volume of spring zooplankton in Prince William Sound. Data from: Cooney et al. (2001)

Since the earlier review (Pearson et al. 1999), new information is available about the decline relative to oil exposure, harvest, disease, poor nutrition, and oceanic factors. Hypotheses relating the decline to oil-spill effects, harvest effects, spawning-habitat loss, or the spawn-on-kelp fishery continue to be unsupported by the available evidence (Table 3). The declining herring condition and low zooplankton abundance (Figs. 3 and 4) described in Pearson et al. (1999) still support the conclusion that poor nutrition was a major factor in the 1993 decline. The results of the ASA modeling of Deriso et al. (2008) offer evidence for both oceanic factors (winter sea-surface temperature) and poor nutrition in the 1993 decline (Table 4). Since the 1999 review, other investigators also report that the 1993 decline was most likely derived from poor nutrition with disease as a secondary factor (Carls et al. 2002; Marty et al. 2003; Hulson et al. 2008; Rice and Carls 2007; Elston and Meyers 2009). Poor nutrition remains the best explanation of the 1992–1993 decline, although some debate remains about potential influence from the oil spill and about the role of disease agents.
Table 3

Summary of evidence for hypotheses concerning the decline of PWS herring and subsequent poor recovery

 

Decline

Hypothesis

Correlation or association?

Used in ASA model (Deriso et al. 2008)?

Significant single factor in randomization tests (Deriso et al. 2008)?

Factor in multifactor model (Deriso et al. 2008)?

Mechanism biologically plausible?

Supporting evidence

Contrary evidence

Role in decline?

1

Population Dynamics

NA

N

NA

NA

Not testable, subsumed under ocean regime

NA

NA

Subsumed under ocean factors

2

Harvest—overfishing

Weak, high harvests for three years preceding the decline

N

NA

NA

Y

Some overfishing occurred in early 1990s

Amount of overfishing not sufficient to account for decline

Not likely

3

Harvest—fisheries closure

Weak, Some underfishing in the 1980s and no harvest in 1989

N

NA

NA

Y

Some underharvesting occurred in 1980s

Timing of underharvesting and decline do not align. Amount insufficient

Not likely

4

Spawning habitat loss

Weak, Spawning habitat in good condition and not limiting

N, Indirectly see H9

NA

NA

Y

Some spawning stopped in some PWS areas

Expansion and contraction of spawning locations as expected with changes in stock levels

Not likely

5

Disease and I. hoferi

Weak, disease was observed, prevalence low and not correlated with biomass

Y, age 3 and older plus age 5 and older

Y, when additional mortality factor in model

Y, in two factor model when additional mortality factor in model

Y

I. h. can lead to mortality in older fish

Loss of older age classes was not evident in decline so that observed age structure does not support role in decline. A factor with mortality but not without mortality factor suggests no role in decline. Pattern in prevalences contrary to epizootic

Not likely

6

Disease and VHSV

Weak, Disease was observed, prevalence low and not correlated with biomass. No estimate of VHSV prevalence for 1993 and before

Y, age 3 and older

N, not for age 3 and older fish

N, not for age 3 and older fish

Y

VHS disease can occur and lead to mortality when fish are stressed

VHSV affects primarily young fish. Fish that survive viral exposure at early age acquire immunity. Mortality in 1992/1993 was not primarily on young fish but even among age classes 3 to 9+. Observed age structure not consistent with a role in decline. Not a significant factor in ASA model of Deriso et al. (2008)

Not a cause of the decline. Probably secondary response after nutritional or other stress

7

Disease and EOK pound fishery

Moderate, EOK pound fishery increased to peak in 1992 and with high harvest in 1993

Y, Ages 3 and older, Ages 3 and 4

N

N

Y

Pounding demonstrated to increase prevalence of VHSV and disease

VHSV mortality mainly on younger fish. Mortality in 1992/1993 was even among age classes 3 to 9+. Observed age structure not consistent with a role in decline

Not likely

8

Oil exposure and VHSV

Weak, VHSV observed in 1993. No significant correlation between PWS herring biomass and PAHs in nearshore waters from shoreline mussels

Y, Age 3 and older

N

N

Y

One experimental study suggests oil exposure may induce VHS disease

Four subsequent experimental studies fail to show VHSV induced by oil. Laboratory experiments failed to demonstrate a link between oil exposure and VHSV. Even if link exists, oil exposure was not high enough to induce effects

Not likely

9

Oil exposure alone

Weak, no significant correlation between PWS herring biomass and PAHs in nearshore waters from shoreline mussels

Y, age 3 and older

N

N

Maybe

Thorne and Thomas (2008) argue that decline started in 1989 and was due in part to adult mortality from exposure to water surface during nightly gulping

Water column concentrations not sufficient in 1991 and afterward to cause adult mortality. Exposure through gulping too intermittent in time and space to induce mortality. Mechanism not demonstrated. Avoidance of oil slicks and sheens would reduce exposure. Timing of exposure and decline not aligned

Not likely

9

Competition or predation by pink salmon

Strong, Pink salmon fry releases and associated adult returns increased dramatically before the decline

N

NA

NA

N

Dramatic increase in adult returns of hatchery pink salmon preceding the decline

No mechanism to induce adult herring mortality. Pink salmon interactions a significant variable in Deriso et al. (2008) ASA model for recruitment

Not likely for decline; Likely for poor recovery

10

Predation by sea lions

Weak, Sea lion abundance and distribution appears to positively correlate with herring biomass

Y, Ages 3 and older

N, Effect in wrong direction

N

Y

Sea lions prey on subadult and adult herring in PWS

Sea lion distribution appears to be driven by herring biomass rather than vice versa. Sea lion abundance too small to be significant source of mortality by predation. Sea lion decline in PWS parallels Alaska-wide declines starting in 1980s

Not likely

11

Poor nutrition

Strong, weight at length progressively decreased before the decline and recovered after the decline

Y

Y, spring condition

Y, spring condition

Y

Decreases in weight and condition precede the decline. Condition in PWS lower than at Sitka

Source of mortality still not directly demonstrated

Likely

12

Ocean factors

Moderate, regime shift occurred, Biomass increase in herring after 1976 associated with regime shift. Decreasing zooplankton abundance after late 1980s

Y

Y, GOA winter water temperature without Additional Mortality Factor

Y, GOA winter water temperature without additional mortality factor

Y

Parallel changes in biomass and year-class strength at PWS and sitka indicate response to gulf-wide environmental changes

Source of mortality still not directly demonstrated

Likely

 

Recovery

Hypothesis

Correlation or association?

Used in ASA Model (Deriso et al. 2008)?

Significant single factor in randomization tests (Deriso et al. 2008)?

Factor in multifactor model (Deriso et al. 2008)?

Biologically plausible mechanism?

Supporting evidence

Contradictory evidence

Role in poor recovery?

1

Population dynamics

NA

N

NA

NA

NA

NA

NA

Lack of strong year classes at PWS and Sitka indicate domain shift at regional scale; subsumed under ocean regime

2

Harvest—overfishing

No, little or no fishing after 1993

N

NA

NA

N

Some overfishing occurred in early 1990s

Sitka as well as PWS showed lack of strong year classes after early 1990s but Sitka has harvest

Not likely

3

Harvesting—fisheries closure

No, little or no fishing after 1993

N

NA

NA

N

Some underharvesting occurred in 1980s

Because of non-aligned timing, can not be a factor in lack of strong year classes

Not likely

4

Spawning habitat loss

No

N

NA

NA

Y

Some spawning stopped in some oiled areas post decline

Spawning habitat in good condition and not limiting; Spawning changes not related to oiled/unoiled conditions; changes in spawning location is expected with changes in stock levels

Not likely

5

Behavioral conservatism

No information

N

NA

NA

Maybe

Little information

Little information. If any conservatism occurred, too short-lived to explain time course of lack of strong year classes over 10 years

Not likely

6

Disease and VHSV

Y and N

Y, ages 3 and 4

N

Y

Y

VHS disease affects younger fish. VHS can occur and lead to mortality when fish are stressed

Not a cause of the lack of strong year classes. Not supported by disease analysis of Elston and Meyers (2009). Some influence but not a major source of mortality in ASA modeling of Hulson et al. (2008). VHS effect on juvenile herring mortality enters ASA model of Deriso et al. (2008) after pink salmon interactions, poor nutrition, and GOA winter temperature in multifactor ASA model and has influence in wrong direction

Not likely

7

Disease and I. hoferi

N

Y, ages 3 and 4

N

N

N

I. hoferi can lead to mortality in older fish. Some disruption of age class structure in late 1990s. influential in ASA models of Hulson et al. (2008) and Marty et al. (2010)

Loss of older age classes not consistent post decline. Age structure as expected in 2000s. Loss of older fish not a direct effect on recruitment

Likely an influence on older fish but not a cause of poor recruitment. Maybe secondary response to poor nutrition and influence on adult biomass

8

Disease and EOK pound fishery

No EOK fishery has occurred since the decline

Y, ages 3 and 4

N

N

N

Pounding can increase prevalence of VHSV

With PWS fisheries closed, the EOK pound fishery can not adversely affect recovery

Not Likely

9

Oil exposure and VHSV

No, PAHs at background after mid-1990s

Y, Ages 3 and 4

N

N

N

See decline

See decline. VHSV not linked to oil exposure. There is not sufficient oil exposure

Not likely

10

Whales as predator in predator pit

Humpback whales co-occur with herring on herring overwintering grounds. Whales target schooling forage fish

N

NA

NA

Y

Overwintering humpback whales observed in herring overwintering area in PWS. Whale abundance increasing. Overwintering diet now appears to be primarily herring

Recent preliminary data is supportive but estimates of consumption vary broadly. More study on overwintering whale abundance and diet is needed to refine consumption estimates

Likely as contributing factor reducing adult herring biomass. Influence on recruitment unlikely

11

Predation by adult coho salmon

Weak, adult coho returns have increased in recent years

Y, ages 1 and 2

N

N

Y

Increases in coho salmon in recent years may have increased extent of predation on young herring

None available

Not likely

12

Predation by adult pink salmon

Strong, Pink salmon returning adults increased dramatically before the decline and have remained high after decline

Y, Ages 0, 1, and 1

Y, Pink Salmon Returning Adults

NA

N

ASA model finds pink salmon adults a significant single factor

No mechanism for adult pink salmon to affect juvenile herring. Adult pink salmon do not feed when returning to PWS. Effect is through pink salmon fry

Not Likely

13

Predation by juvenile pink salmon

Strong, Pink salmon fry releases increased dramatically before the decline and have remained high after decline

Y, ages 0, 1, and 2

Y, Pink Salmon Fry

Y, Pink Salmon Fry

Y

ASA modeling finds juvenile pink salmon significant in single factor and multifactor models. Predation by juvenile pink salmon on age-0 herring in July and August likely

Not empirically demonstrated

Possibly. Predation by juvenile pink salmon and juvenile herring likely, but extent of predation is unclear

14

Poor nutrition

Moderate, poor nutrition in PWS in 1990. Recovered in Sitka but not PWS

Y, ages 3 and older

Y, spring condition without additional Mortality Factor

Y, spring condition without additional mortality factor

Y

Continued low zooplankton abundance in PWS. Condition in PWS lower than at Sitka. Nutrition for overwintering age 0 and age 1 herring is key to year class strength

Not empirically demonstrated

Likely in association with ocean factors and pink salmon fry releases

15

Ocean Factors

Strong, natural log of numbers and biomass of age 4 herring significant quadratic relationship with GOA winter temperature

Y, Age 3 and older

Y, GOA Winter Water Temperature

Y, GOA Winter Water Temperature

Y

Changes at PWS and Sitka indicate response to Gulf-wide environmental changes. No strong year classes at either PWS or Sitka after 1993

Not empirically demonstrated

Likely in association with ocean factors and pink salmon fry releases; Likely for lack of strong year classes

16

Competition

Strong, pink salmon fry releases increased dramatically before the decline and have remained high after decline

Y, ages 0, 1, and 2

Y, Pink Salmon

Y, Pink Salmon

Y

ASA modeling finds juvenile pink salmon significant in single factor and multifactor models. Juvenile pink salmon and age 1 herring overlap in space and time and share prey. Field observations in PWS show significant disruption of herring feeding when juvenile pink salmon co-occur

Not empirically demonstrated

Likely in association with ocean factors and poor nutrition; More likely and more substantial than ocean factors

Table 4

Summary of outcomes from ASA modeling of Deriso et al. (2008) sorting by rank in randomization test witout an additional mortality factor. NS indicates a non-significant factor

ID from Deriso et al. (2008)

Factor

Factor affects ages

Rank based on Chi-Square

Rank based on randomization test

Rank in multifactor model

w/mortality factor

w/o mortality factor

w/partial recruitment

w/o partial recruitment

w/partial recruitment

w/o partial recruitment

w/mortality factor

w/o mortality factor

w/mortality factor

w/o mortality factor

15

PWS pink salmon hatchery fry releases

0 and 1

1

3

1

3

1

1.5

1 and 2

1, 2, and 3

16

PWS pink salmon hatchery adult returns

0, 1, and 2

2

4

2

4

2

1.5

NA

NA

2

GOA winter water temperature

3 and older

NS

NS

3

NS

NS

3

NO

2 and 3

1

Poor nutrition

3 and older

NS

NS

4

NS

NS

4

NO

3

8

VHSV-FSR severity index

3 and 4

NS

NS

6

NS

NS

5

NO

4

9

Ichthyophonus hoferi infection rate

All

3

6

NS

5

3

NS

2

NA

3

Spring–summer PDO

3 and older

NS

1

NS

1

NS

NS

NA

NA

4

Fall–winter PDO

3 and older

NS

2

5

2

NS

NS

NA

NA

5

Spring–summer NOI

3 and older

NS

NS

NS

NS

NS

NS

NA

NA

6

Fall–winter NOI

3 and older

NS

5

NS

6

NS

NS

NA

NA

7

VHSV-FSR severity index

3 and older

NS

NS

NS

NS

NS

NS

NA

NA

10

Ichthyophonus hoferi infection rate

5 and older

NS

NS

NS

NS

NS

NS

NA

NA

11

Oil exposure index, PAH in nearshore mussels

3 and older

NS

NS

NS

NS

NS

NS

NA

NA

12

Oil exposure index, PAH in nearshore mussels

Age 0

NS

NS

NS

NS

NS

NS

NA

NA

13

Spawn-on-kelp closed pound fishery

3 and older

NS

NS

NS

NS

NS

NS

NA

NA

14

Spawn-on-kelp closed pound fishery

3 and 4

NS

NS

NS

NS

NS

NS

NA

NA

17

PWS total pink salmon returns (wild and hatchery)

0, 1, and 2

NS

NS

NS

NS

NS

NS

NA

NA

18

PWS coho salmon adult returns plus recreational catch

1 and 2

NS

NS

NS

NS

NS

NS

NA

NA

19

Sea lion abundance (not including pups)

3 and older

NS

NS

NS

NS

NS

NS

NA

NA

Oil exposure

Recent evidence continues to support the earlier conclusions of Pearson et al. (1999) that oil exposure during or after the EVOS did not play a role in the 1992–1993 decline. Boehm et al. (2007) synthesized the available information on the water-column concentrations of total polycyclic aromatic hydrocarbons (TPAH) with emphasis on nearshore waters in the spill path and concluded that “water column concentrations of TPAH resulting for the spill returned to background levels by 1990, ranging from 0.001 to 0.10 ppb TPAH.” An oil-exposure index derived from the TPAH burdens in mussels confirms that oil exposure was not a significant factor affecting adult mortality in the single-factor or multi-factor ASA models of Deriso et al. (2008) (Table 4). Carls et al. (2002), Marty et al. (2003), Rice and Carls (2007) and Hulson et al. (2008) find factors other than the EVOS (e.g., poor nutrition) to be the likely direct cause of the 1992–1993 decline.

Thomas and Thorne (2003) and Thorne and Thomas (2008) suggest that the decline actually began in 1989 and was associated with the EVOS. As a mechanism for exposure to oil and to subsequent mortality, they postulate direct contact with oil when herring break the water surface during “gulping” of air to fill their swim bladders. At dusk, herring schools ascend from deeper waters to surface waters to feed (Blaxter 1985). While at the surface, herring may gulp air to fill their swim bladders and regain buoyancy. The extent of gulping is unknown.

Gulping is a plausible way for herring to contact oil on the water surface, but such contact would be brief and intermittent. Behavioral avoidance of slicks and sheens is also plausible and would prevent or minimize exposure. Herring avoid suspended dredged materials (Johnston and Wildish 1981), but the extent to which herring avoid oil is unknown.

Assuming herring were unable to avoid slicks and sheens, the window for significant direct contact with EVOS oil in slicks and sheens would have been limited primarily to spring 1989 and would have ended when EVOS sheens decreased to background by the fall of 1990. The oil slicks from the Exxon Valdez began to exit PWS about one week after the beginning of the spill (Wolfe et al. 1994). By April 1989, floating oil within PWS had fallen to primarily surface sheens (Wolfe et al. 1994).

At their greatest extent, sheens associated with EVOS covered only a very small area of PWS on a few days. The occurrence, volume, and source of PWS sheens were surveyed from summer 1989 until fall 1990 (Taft et al. 1995). The majority of the sheens (75%) were small (<1 l) with only 19 of the total 827 sheens being greater than 1,000 l in oil volume. A rainbow-colored sheen has a thickness of 0.0003 mm (Taft et al. 1995) so that a rainbow sheen of 1,000 l would cover an area of about 3,000 m2. Of these 19 sheens, 13 large-volume sheens associated with the EVOS occurred on October 24, 1989, after a storm had released oil from beach sediments. These 13 sheens had volumes from 1,000 to 54,000 l and covered areas from 0.003 km2 about 0.34 km2. If all 13 were of the maximum observed extent, these sheens would have covered about 2.2 km2 or about 0.049% of PWS total area of 9,059 km2. After March 1990, no sheens from EVOS were greater than 65 l in oil volume. By September of 1990, PWS sheens from EVOS had decreased to smaller volumes than those associated with normal vessel traffic.

If direct and lethal contact between herring and oil had been widespread in 1989, then large, observable herring kills would be expected, but none was observed. In contrast, mass mortalities of herring have been observed during epizootics (Tibbo and Graham 1963; Patterson 1996; Meyers et al. 1999) and an industrial spill of elemental phosphorus (Winters et al. 1986). However, even with the massive boat traffic throughout PWS during the 1989 and 1990 cleanup—and with scientists specifically looking for fish kills—no herring (or other fish) kills were observed.

If herring had direct and lethal contact with spilled oil, adult herring survival would have been lowered in 1989. In fact, adult herring survival was highest in 1989 and lowest in 1992 (Hulson et al. 2008)―opposite to the expectation if direct contact with slicks was inducing substantial adult herring mortality.

Viral hemorrhagic septicemia virus

Viral hemorrhagic septicemia virus (VHSV) infects Pacific herring and other marine fish species along the west coast of North America (Meyers et al. 1994; Hedrick et al. 2003). Recent studies suggest that VHS is primarily a disease of age-0 herring and that the disease is either rapidly resolved or fatal within 2 weeks following infection (Kocan et al. 2001; Hershberger et al. 2007; Elston and Meyers 2009). Fish surviving VHS acquire immunity. VHSV is an opportunistic pathogen in adult fish that produces disease when fish are stressed by environmental or other factors (Meyers et al. 1994; Elston and Meyers 2009).

Based on an analysis of VHSV-positive herring samples in 1993 from PWS and other areas, Meyers et al. (1994) concluded that the role of VHSV in the 1992–1993 decline could not be established with the 1993 data. A critical and comprehensive review of the role of disease in PWS herring population dynamics from 1989 to 2005 led Elston and Meyers (2009) to conclude that VHS was not a primary cause of the 1992–1993 decline for several reasons: (1) VHSV is principally a pathogen of juvenile herring (Kocan et al. 1999a); (2) the mortality in 1992–1993 was distributed across all age classes including those age classes resistant to the disease (Elston et al. 1997); (3) no mass mortality was observed; (4) the poor condition of the herring itself was sufficient to cause the decline; and (5) most herring simply did not return to spawning grounds in 1993 and could not be examined for cause of death. The designation of the herring population crash of 1992–1993 as an “epidemic,” implying a primary infectious cause (e.g., Marty et al. 2003; Marty 2007) is without basis. Elston and Meyers (2009) evaluated all evidence regarding VHS in PWS herring from 1989 to 2005 and found no evidence to support any assignment of significant effects of VHSV on PWS herring biomass or recruitment.

Additionally, hypothesis tests using the ASA modeling by Deriso et al. (2008) did not find VHSV to be a significant factor when tested in the model as a single factor. The VHSV indices used for younger fish were selected by the hypothesis test procedure as the fourth factor in a four-factor model, however the regression coefficient had the incorrect (i.e., positive) sign.

Viral hemorrhagic septicemia virus and oil exposure

Pearson et al. (1999) found no evidence that either acute or chronic oil exposure from the EVOS contributed to the 1993 decline. Carls et al. (2002) stated that the consequences of high biomass, disease, and poor nutrition are the more likely explanations of the 1993 decline but indicated that indirect links to the EVOS could not be entirely ruled out. Carls et al. (2002) and Marty et al. (2007) find a link from oil exposure to the decline unlikely, in part because of the time delay between oil exposure and the decline.

An indirect link from oil exposure through disease is suggested by one experiment of Carls et al. (1998) but is not supported by subsequent experiments. In reviewing experiments that attempted to induce or exacerbate VHSV in herring by weathered or fresh crude oil exposure (Carls et al. 1998; Kennedy et al. 1999; Kocan et al. 1999b; Sanders 2005), Elston and Meyers (2009) found that none of the experiments with juvenile and adult herring demonstrated a causal link between any oil effect and VHSV at oil-exposure concentrations as high as 178 to 328 ppb TPAH (range of highest treatments used) for between 14 and 28 days. Rice and Carls (2007) now also consider the EVOS not to be a causal or contributing factor in the decline.

Pathogen Ichthyophonus hoferi

The pathogen Ichthyophonus hoferi has long been associated with epizootics leading to mass mortalities and low population levels in Atlantic herring (Tibbo and Graham 1963; Patterson 1996; Kramer-Schadt et al. 2010). An alternative or additional causative agent, I. hoferi, was identified by Meyers et al. (1994) and Marty et al. (1998) but was believed an incidental finding at the time and subsequently (Quinn et al. 2001; Marty et al. 2003).

Recent work by Kramer-Schadt et al. (2010), studying Norwegian spring spawning herring (NSSH), and a reiteration of the ASA model (Marty et al. 2010) suggest that I. hoferi can be an important factor in herring population declines. However, Kramer-Schadt et al. (2010) demonstrate that representative sampling of herring populations to estimate prevalence, severity, and impact of I. hoferi requires extensive effort to account for broad geographic migration patterns, differential distribution of the disease by age class, and clustering of infected fish in schools more vulnerable to sampling.

Sampling of PWS herring for the presence of I. hoferi prior to the decline was limited but two points derived from all available data are contrary to the I. hoferi as the causative agent in the decline. First, mass mortalities with large numbers of moribund and dead fish in shallow areas have been associated with I. hoferi in Atlantic herring (Sindermann 1958), but the geographically limited mortality observed in PWS herring in 1993 (Meyers et al. 1994) showed a very low prevalence of I. hoferi. Removal of moribund fish by predation, as observed by Kramer-Schadt et al. (2010) for Atlantic herring, remains a possibility, but it also must remain an unknown for which there is no evidence in PWS in 1993. Second, the prevalence of I. hoferi infection in PWS (based on very limited sampling, and determined by histology, Table 5) before and during the decline was low, i.e., 5.7% in spring 1992 and 5.1% in spring 1993 but increased to 29% in 1994 after the decline (Table 5). This pattern of low levels before the decline and high levels after the decline is the reverse of the expectation of high prevalence before and lower prevalence after the decline if I. hoferi was causal in the decline. The expectation would be that the prevalence decreases after the epizootic because infected fish succumb to disease. The prevalence data is contrary to the notion that the I. hoferi infection was causative in the PWS decline (Table 5).
Table 5

Prevalence of Ichthyophonus hoferi in PWS Pacific herring samples and mean age of fish sampled

Year

I. hoferi prevalencea

Method usedb

Mean age of herring (years)c

1989

13.0

Histology

ND

1990

ND

ND

ND

1991

3.4

Histology

7.3

1992

5.7

Histology

4.7

1993

5.1

Histology

ND

1994

23.1

Histology

6.2

1995

21.7

Histology

6.2

1996

20.4

Histology

5.8

1997

16.2

Histology

5.4

1998

17.6

Histology

4.7

1999

24.3

Histology

6.1

2000

21.7

Histology

6.2

2001

38.0

Histology

5.7

2002

15.0

Histology

4.0

2003

20.8

Gross/Calc

4.2

2004

28.4

Gross/Calc

5.1

2005

50.6

Gross/Calc

5.7

2006

40.0

Gross/Calc

6.2

aValues for I. hoferi prevalence for 1989–2006 from Marty et al. (2010). ND no data available

bPrevalence of I. hoferi for 1994–2002 is the prevalence of cases with sum-I. hoferi score >0 by histology. Prevalence of I. hoferi for 2003 through 2006 was calculated using values for I. hoferi prevalence determined by gross examination of the heart and the first order linear regression of gross I. hoferi prevalence (x) versus prevalence determined by sum-I. hoferi score >0 (y) for samples from 1997 through 2002 (y = 1.81x + 2.05; r2 = 0.89) from Marty et al. (2007) and as cited in Marty et al. (2010)

cMean age values from Marty et al. (2010). ND no data available

Hypotheses concerning poor recovery

Hypotheses for poor recovery must provide mechanisms for (1) no recruitment of strong year classes in PWS after recruitment of the 1988 year class and (2) low herring biomass. Recovery hypotheses examined (Table 3) include oil exposure, harvesting, disease, oceanic factors, predation, and competition.

Oil exposure

Among the post-decline exposure pathways that could limit development of strong year classes, and hence recruitment, the exposure of most concern is that of juvenile (age-0+ and 1+) herring to water-column TPAH concentrations. Nearshore areas of PWS bays and fjords provide nursery areas for juvenile herring (Stokesbury et al. 1999, 2000; Norcross et al. 2001). In such locations, mussels provide composite water sampling of intertidal and shallow subtidal areas.

The use of bivalves for monitoring the marine environment dates back to the 1970s (Farrington and Quinn 1973; Dunn and Stich 1976; Dunn and Young 1976). Mussel tissues have been used extensively in determining levels of total TPAH in the waters of PWS (Short and Harris 2006; Boehm et al. 2004, 2007). For PWS, Neff and Burns (1996) applied an equilibrium-partitioning technique to relate tissue burdens of TPAH in mussels to contaminant levels in the water column. Direct water-column sampling was discontinued several years after the EVOS because TPAH concentrations were at background or below method detection limits. Boehm et al. (2004, 2007) produced an extensive database of the concentrations of TPAH in mussel samples collected from 1989 to 2005 from oiled and unoiled shorelines including sites of past human activity (e.g., mines, canneries) in PWS.

Figure 5 shows the arithmetic means of TPAH water concentrations derived from analyses of mussel tissues from oiled and unoiled shorelines using the conversion of Neff and Burns (1996). The profile of exposure concentrations in Fig. 5 shows the peak of 3.3 ppb in 1989 (the year of the spill), followed by exponential decay to less than 0.1 ppb in 1990 and to low background levels (<0.02 ppb) since 1995.
Fig. 5

Arithmetic mean TPAH concentrations (ppb) in Prince William Sound seawater. Mussel samples were collected from coastal waters oiled in 1989 and unoiled areas. Tissue values from Boehm et al. (2007) were converted to seawater concentrations using the method of Neff and Burns (1996)

The TPAH levels were too low to cause toxic effects. Even in 1989, the TPAH levels (3.3 ppb) were below the 28 ppb that Carls et al. (1998) found associated with VHSV in one laboratory experiment. For all years except 1989, water concentrations were also below the value of 0.4 ppb, the value which Carls et al. (1999) postulated as the threshold level for the development of abnormal larvae in herring eggs exposed to oil. While evidence of localized effects on herring eggs were observed by Pearson et al. (1995) in 1989, there was no evidence of effects on eggs in 1990 (Brown et al. 1996; Pearson et al. 1995). In 1995, Johnson et al. (1997) incubated eggs from mature herring collected from oiled and unoiled areas of PWS and found no regional differences in fertilization, hatching times, hatching success, or larval viability or abnormalities. The water TPAH data and experimental studies provide no evidence that exposure to oil in the water column has affected herring or prevented their recovery in PWS.

Subsurface oil residues are not a source of exposure for herring. Juvenile herring forage in the water column and do not dig in sediments. Other factors also limit the potential of subsurface oil residues as an exposure source for herring. First, there was little overlap between herring spawning grounds and oiling at the time of the EVOS (Fig. 6) (Pearson et al. 1995, 1999) and much less between the residual oiling and spawning areas (Rice and Carls 2007). Second, the bioavailability of subsurface oil residues is low. In 2002, Neff et al. (2006) analyzed beach sediment, beach pits, and intertidal biota for 17 sites heavily oiled in 1989 and found low TPAH concentrations in biota (e.g., means of 22.4 ± 59 (SD) ppb in fish from oiled sites, 6.4 ± 4.9 ppb in unoiled sites, and 108 ± 165 ppb in unoiled sites with human activity). In 2002, TPAH concentrations in the tissue of the cockscomb gunnel, which inhabits the intertidal and shallow subtidal zones of PWS, and in the tissues of clams, mussels, and worms, did not differ significantly between previously oiled sites and unoiled reference sites and are predominantly from combustion sources—not from crude oil (Neff et al. 2006).
Fig. 6

Herring spawning areas and oiled shorelines in PWS during 1989. Base map from Pearson et al. (1999) reproduced with permission. The boundaries of the Herring Spawning Areas designated by ADFG have changed slightly over the year, and the 1989 boundaries given by Biggs et al. (1992) appear as an overlay on the base map

The ASA modeling results do not support an effect of oil exposure on recruitment. Deriso et al. (2008) used TPAH concentrations in PWS nearshore areas as a covariate in their ASA modeling. The TPAH levels fell during the spill and have been at background levels for over two decades. The oil-exposure index derived from the TPAH burdens in mussels was not a significant factor affecting recruitment in the single-factor or multi-factor ASA models (Deriso et al. 2008).

Harvest

The evidence that harvest effects are not involved in the poor recovery of PWS herring derives from the lack of strong year classes after 1993 in both PWS and Sitka (Fig. 2). Fishing has continued at Sitka but not PWS. This parallelism between PWS and Sitka indicates that the lack of strong year classes in PWS since 1993 is due to some regional-scale environmental change rather than a harvest effect. Also, Deriso et al. (2008) found that the commercial harvest of PWS herring was commensurate with harvests being actively managed to be below 20% of the spawning biomass.

Spawning habitat loss

Loss of spawning habitat is one of the three main general causes of fishery declines listed by Hilborn (1997) and is an impediment to stock success (Stephenson 1997). Spawning habitat, however, is not limited and has not been lost in PWS. Spawning sites in PWS are not subject to shoreline urbanization or other widespread human disturbances that degrade or fragment herring spawning habitat in other regions.

Shifts in PWS herring spawning sites since the 1970s (Fig. 7) are typical of those in other herring stocks. Spawning sites change over the years, to be abandoned and then reoccupied (McQuinn 1997; Hay et al. 2001). Both Atlantic and Pacific herring are repeat spawners at particular sites but with varying degrees of site fidelity (McQuinn 1997; Hay et al. 2001). This repeat spawning is not natal homing because herring do not necessarily return to the site where the fish were hatched. Herring spawning habitat in PWS appears to be in excess of what herring require, and, in any event, little if any spawning habitat was impacted or lost as a consequence of the 1989 spill. The shifts in spawning locations (Fig. 7) are not related to oiled or unoiled condition following EVOS. For these reasons, we reject the hypothesis that the poor recovery of herring can be attributed to habitat loss from the 1989 oil spill.
Fig. 7

Kilometer-days of herring spawn in Prince William Sound by subarea by calendar year. The 1989 boundaries for the ADFG Herring Areas appear in Fig. 6. Data from: Dillingham et al. (2004)

Viral hemorrhagic septicemia virus

VHSV primarily affects juvenile herring so that the potential role of VHSV in reducing recruitment and limiting recovery needs to be examined. Quinn et al. (2001) incorporated disease indices in their ASA model of PWS herring and found that negative correlations between VHSV prevalence and recruitment or the logarithm of recruitment at lags of 0 year and 1 year were not significant. Lagging the VHSV indices by 1 year improved the negative correlation with recruitment, but the correlation was still not significant. Quinn et al. (2001) suggest that disease may still be a factor in recruitment, the effect of which was masked by other factors in their study. Based on a later version of the ASA model, Marty et al. (2003) conclude that both VHSV and Ichthyophonus hoferi may have affected the PWS herring population since the 1993 decline. Elston and Meyers (2009) reviewed all VHSV data from PWS herring from 1989 through 2005, including correlation analysis with recruitment lags of 0, 1, 2, and 3 years and concluded that there was no evidence that VHSV affected recruitment processes after the decline. These authors also provided corrected prevalence data showing that the prevalence of VHSV in PWS herring never exceeded 7.0% from 1994 to 2002 in contrast to Marty et al. (2003). The ASA modeling of Deriso et al. (2008) does not find VHSV adversely influencing ages-3 and -4 herring recruitment. The ASA modeling of Hulson et al. (2008) finds only small influence of VHSV prevalence on herring mortality in PWS and supports Elston and Meyers (2009). As discussed in more detail in Elston and Meyers (2009), the evidence is not sufficient to assign VHSV as a major factor limiting recovery.

Viral hemorrhagic septicemia virus and oil exposure

As with the decline, the experimental evidence does not support any causal relationship between oil exposure and VHSV in the poor recovery of PWS herring. With the oil levels in PWS at background levels throughout the period of poor recovery and lack of evidence for a significant effect of VHSV on herring with or without oil present (Elston and Meyers 2009), a linkage between oil exposure, VHSV, and poor recovery is not warranted.

Pathogen Ichthyophonus hoferi

The marked and progressive increase in observed infection rate of herring by of I. hoferi after the decline (Table 5) suggests some involvement of disease in the recovery process. Ichthyophonus hoferi is a pathogen producing chronic disease that progressively affects older age classes of herring more severely (Marty et al. 1998; Hershberger et al. 2002). I. hoferi is associated with adult Atlantic herring mortality, in some cases on a massive scale (Tibbo and Graham 1963; Patterson 1996). Marty et al. (2003) concluded from ASA modeling that the PWS prevalence of I. hoferi did not affect adult herring survival from 1994 to 2000. The available PWS disease data suggest that the I. hoferi infection rate is more influential in mortality of adult PWS herring than VHSV, especially since 2000 (Marty et al. 2010).

This hypothesis may ultimately be proven correct, but the findings from Kramer-Schadt et al. (2010) indicate the I. hoferi prevalence data for PWS suffer in two aspects that preclude drawing strong conclusions from the prevalence data. First, the field observational technique used to identify infected fish needs to be corrected for sensitivity to achieve an estimated match with histology diagnosis (a high significant correlation was shown between field observations and histology [Marty et al. 2010]). Measurements with the field technique underestimate infection rate (Table 5; Kocan et al. 2011; Kramer-Schadt et al. 2010) as determined by the more accurate tissue explant method. The time series produced by correction from field observation to histology appears reasonably consistent, but total infection rate likely remains underestimated (Kocan et al. 1999a, b, 2011). There is also a lack of correlation between prevalence by histology or visual examination and prevalence as determined by tissue explant culture, so that the extent to which prevalence was underestimated by the first two techniques likely varied substantially between samples. Kocan et al. (2011) visualized tissue sections from infected hosts three-dimensionally, and found non-random distribution of the parasite, establishing a mechanism for the variability in sensitivity encountered when using histology for the detection of I. hoferi in infected hosts. Second, sampling error for the PWS data is likely large because Holst (1996) and Kramer-Schadt et al. (2010) found infection rates were up to 3 times higher in small compared to large catches. Swimming impairment in infected fish appears to separate them from the larger schools and can also be expected to make infected fish more vulnerable to sampling. Finally, the consequences of infection severities and the time course of the disease under various infection scenarios remain poorly understood. The observed spatial and seasonal patchiness of infected fish suggests that sampling herring for prevalence without taking such variation into account could yield sampling bias and erroneous conclusions.

Multiple disease factors

Several authors have introduced the hypothesis that filamentous bacteria played a role in the decline and lack of recovery of PWS herring populations (Hulson et al. 2008; Marty 2007; Marty et al. 2007). In the only reported, confirmed infection of this type in Pacific herring, Carls et al. (1998) reported a low mortality rate (3%), scale loss, and raised white cutaneous foci associated with filamentous bacterial rods in captive herring from near Juneau, Alaska. However, culture of kidneys from PWS herring sampled from 1994 to 2002 failed to produce any filamentous bacterial isolates. Filamentous marine bacteria (such as Tenacibaculum maritimum, formerly Flexibacter maritimus) are a part of the normal external flora of marine fish and often, if not always, require an injury to the tissues as a portal of entry (Avendaño-Herrera et al. 2006). However, once established, T. maritimum can cause an aggressive infection and contribute to mortality. Herring are susceptible to stress, which can cause blood vessel dilation, hemorrhaging, and external injuries to skin and fins, opening the portals for secondary pathogens. There is no hard evidence to show that T. maritimum was responsible for the skin ulcers observed in PWS herring between 1993 and 2005 (Elston and Meyers 2009), but it could have played a role in exacerbating lesions in stressed fish. Given the low prevalence of observed ulcers in PWS herring and the complete absence of any data confirming infection by filamentous bacteria, the assertion of a primary role of filamentous bacteria in controlling PWS herring populations is not supported.

Predation

The synthesis of Spies (2007) provides substantial detail on trophic interactions in PWS and GOA (Gulf of Alaska) and a background for assessing effects of predation on herring. Forage fish in PWS include juvenile Pacific herring, Pacific sand lance, capelin, eulachon, juvenile salmon, juvenile pollock, and some small mesopelagic fishes. Predators utilizing this complex of forage fish include marine mammals, marine birds, and larger fish. The high fat content of sand lance, capelin, and juvenile herring make these fish attractive prey to predators; for example, Pacific herring, sand lance, and capelin comprise 58% of the diet by weight of 14 seabirds in PWS (Spies 2007).

The ASA modeling of Deriso et al. (2008) finds PWS hatchery releases of juvenile pink salmon to consistently be the single factor affecting recruitment (Table 4) but could not discern whether the mechanism was predation or competition; the paper states “The salmon factors always were highly significant and reduced the spawning biomass to 20% of its size as calculated by removal of the hatchery effect in the ASA [model].” In contrast, the ASA modeling did not find adult coho predation applied to age-1 and age-2 herring to be a significant factor influencing recruitment. Similarly, the modeling did not find adult sea lion predation to be a significant factor influencing adult herring mortality.

Whale predation on herring can be significant. Minke whale predation on Norwegian spring-spawning herring varies annually but can account for 45% of the mortality (Tjelmeland and Lindstrom 2005). For Atlantic herring on Georges Bank, both fin and humpback whale predation removes substantial portions of the herring biomass while seabird predation removed about 100 times less biomass (Overholtz and Link 2007). This section examines recovery hypotheses concerning predation on herring by pink salmon and humpback whales.

Pink salmon predation

Unlike all other herring fishery areas, PWS has pink salmon hatcheries that contribute substantial numbers of fry above and beyond wild fry entering the Sound. Juvenile pink salmon releases from PWS hatcheries began in the mid-1970s. The annual releases first reached 600 million in 1989 and have been about 600 million since the mid-1990s (Fig. 8). Here we examine the potential for predation by juvenile pink salmon on juvenile herring. We do not consider predation by returning adult pink salmon because previous studies (Okey and Pauly 1999) report that returning pink salmon “feed little while in [PWS], if at all.”
Fig. 8

Prince William Sound herring ASA biomass and pink salmon fry released from hatcheries by calendar year. Data from: Ashe et al. (2005), Donaldson et al. (1995), Farrington (2004), Moffitt (2006), White (2005)

In contrast to adult pink salmon, which feed primarily on zooplankton and micronekton (Ruggerone and Nielsen 2004), juvenile pink salmon in PWS consume substantial amounts of fish (Sturdevant 1999; Willette et al. 1999a). Stomach analyses of PWS juvenile pink salmon by Sturdevant (1999) reveal that fish are 28% of the diet by weight over all months but are as high as 80% in July. Juvenile pink salmon and age-0 herring overlap in the nearshore nursery areas from July through at least August.

Fish in general select prey that are 20–30% of their body size, but they can select prey that are up to 50% of their size when they commence piscivory (Willette et al. 1999b). Juvenile herring are smaller than pink salmon when the herring enter PWS. Age-0 herring enter the Sound in May at less than 0.2 g average body weight (Okey and Pauly 1999) and grow to 6.9 g the first summer (Sturdevant et al. 2001). Pink salmon, by contrast, enter the Sound in mid-May at 0.26 and 0.53-g average body weights for wild and hatchery fish, respectively, and grow to 99 g by October (Cross et al. 2005). In May, age-1+ herring weigh 12 g and are much larger than any of the other 0+ forage fish. Juvenile pink salmon grow more quickly than the age-1+ herring but do not outweigh them until August. Predation by juvenile pink salmon most likely affects age-0 rather than age-1 herring.

Humpback whale predation

The humpback whale (Megaptera novaeangliae) is a predator that targets herring aggregations (Hjort and Ruud 1929). Such targeted predation leads Brown (2004) to suggest that PWS herring are now in “a predator pit,” where recruitment cannot exceed losses from predation.

Humpback whales visiting PWS are part of the Central North Pacific Stock of humpback whales, which has increased in recent years, primarily from increases in Southeast Alaska (Angliss and Outlaw 2005). A recent estimate for the Central North Pacific Stock has 2,036 whales with 147 comprising the PWS population. Whales observed in PWS are transient, being observed elsewhere in the GOA during the same summer (Von Ziegesar et al. 1994; Angliss and Outlaw 2005) and in other parts of the range in other seasons (Calambokidis and Steiger 2001).

Whale abundance in PWS peaks in July and falls to low levels by mid-September. The number of humpbacks identified in PWS from 1988 to 1993 averaged 49 with a range from 26 to 66 (Von Ziegesar et al. 1994; Calambokidis and Steiger 2001). Von Ziegesar (pers. comm.) estimates that PWS has 50 to 70 whales during the summer and 10 to 20 whales during the winter. Von Ziegesar et al. (1994) estimated summer residence time for individual whales in PWS to be 49 days for summer resident whales and 27 days for other more transient whales. During summer feeding in PWS, humpback distribution in 1988 to 1990 centered on the southern end of Knight Island and Montague Strait (Von Ziegesar et al. 1994).

Preliminary results from ongoing studies of humpback whales in PWS indicate that the overwintering population is higher in recent years (2008, 2009) than that reported by Von Ziegesar et al. (1994) for the 1990’s. PWS-wide surveys estimated the numbers of whales overwintering in PWS to be 67 in September 2008 and with a peak of 112 in October 2008 (Rice 2009). Slightly over 50 whales were estimated to be overwintering in November and December 2008, and only 8 whales, in January through March 2009. The total PWS whale population also appears to have increased to 160 (Rice 2009).

The diet of humpback whales in their North Pacific feeding grounds consists mainly of zooplankton and small schooling fish (Angliss and Outlaw 2005). Reconstruction of current humpback diets near Kodiak Island estimated that the bulk of the diet was comprised of walleye pollock (37%), capelin (29%), and euphausiids (22%) with herring being only 0.03% of the diet (Witteveen et al. 2006). Stomach analyses of humpback whales conducted in the 1920s near Kodiak Island indicated that surf smelt and euphausiids were the major dietary items (Witteveen et al. 2006). Based on sonar surveys and scat analysis, the majority of the potential prey in PWS during the summer were euphausiids, but herring and sand lance also were present (Matkin and Hobbs 1999). Based on observations of the winter distribution of herring and humpbacks, Matkin and Hobbs (1999) and Brown (2004) have inferred that the overwintering whales are feeding on herring. However, studies have shown that high euphausiid abundance leads to aggregations of humpback and other whales (Hjort and Ruud 1929 cited in Sverdrup et al. 1942; Everson 2000). Also plausible is the inference that the observed winter co-occurrence results from pursuit by both herring and humpbacks of euphausiids, probably a major, if not the dominant, zooplankton in winter. Preliminary results of ongoing visual and acoustic observations along with fatty acid analyses indicate that overwintering whales in PWS frequently feed on herring (Rice 2009).

Tables 6 and 7 present estimates of the consumption of herring by humpbacks in PWS during the summer feeding season and the overwintering period. Median value for the standing stock of humpbacks of 60 whales in the summer is taken from the range given by Von Ziegesar et al. (2006 pers. comm.). The monthly values for overwintering whales are taken from Rice (2009). A range of values for the daily consumption rates was used to estimate consumption. The information on the diet composition for summering whales leads us to use a worst-case high estimate of consumption by assuming that 50% of the diet is fish and that 95% of the fish are herring. Similarly, we estimate predation by overwintering whales, assuming that the winter diet is 80% herring.
Table 6

Estimated consumption of herring by Humpback whales in PWS during summer

Element/source

Feeding rate (kg per day)

Consumption (kg per season) assuming season (days) = 122

Proportion of diet as fish

Number of herring consumed per whale assuming weight (g) of individual herring = 74.4

Number of herring consumed in PWS assuming number of whales = 60

Number of herring consumed as % of subadult population = 81,000,000 (%)

Matkin and Hobbs (1999)

890

108,580

0.48

700,516

42,030,968

52

Matkin and Hobbs (1999)

890

108,580

0.10

145,941

8,756,452

11

Tynan (2004)

951

116,022

0.48

748,529

44,911,742

55

Tynan (2004)

2,200

268,400

0.02

72,151

4,329,032

5

Witteveen et al. (2006)

370

45,140

0.48

291,226

17,473,548

22

Witteveen et al. (2006)

370

45,140

0.0003

182

10,921

0.013

Mean

945

115,310

0.26

326,424

19,585,444

24

Minimum

  

0.03

182

10,921

0.013

Maximum

  

48.00

748,529

44,911,742

55

Table 7

Estimated consumption of herring by Humpback whales in PWS during fall and winter

Element/source

Feeding rate (mt per day)

Proportion of diet as fish

Weight (mt) of herring consumed by number of whales observed, 2008, 2009

Total weight (mt) of herring consumed for the over wintering period

Number of herring consumed per whale assuming weight (kg) of individual herring = 0.1584

Number of herring consumed as % of adult population = 200,000,000 (%)

SEP

67

OCT

112

NOV

53

DEC

51

JAN

8

FEB

8

MAR

8

Matkin and Hobbs (1999)

0.890

0.80

1,431

2,392

1,132

1,089

171

171

171

6,558

41,398,485

21

Tynan (2004)

0.951

0.80

1,529

2,556

1,210

1,164

183

183

183

7,007

44,235,909

22

Tynan (2004)

2.200

0.80

3,538

5,914

2,798

2,693

422

422

422

16,210

102,333,333

51

Witteveen et al. (2006)

0.370

0.80

595

995

471

453

71

71

71

2,726

17,210,606

9

Witteveen et al. (2006)

0.370

0.80

595

995

471

453

71

71

71

2,726

17,210,606

9

Mean

1.103

0.80

1,773

2,964

1,403

1,350

212

212

212

8,125

51,294,583

26

Minimum

0.370

595

995

471

453

71

71

71

2,726

17,210,606

9

Maximum

2.200

3,538

5,914

2,798

2,693

422

422

422

16,210

102,333,333

51

Number of whales observed from Rice (2009)

In the summer, the whales would feed primarily on age-2 herring because the younger age classes are in nearshore nursery areas inaccessible to whales and adult age classes are outside PWS or in eastern parts of PWS where whales do not aggregate in summer (Stokesbury et al. 1999; Brown et al. 2002). Assuming an estimated weight of 74 g for individual age-2 herring, estimated consumption by 60 summering whales would range from 0.011 to 44.9 million sub-adult herring with an average of about 19.6 million sub-adult herring (Table 6). Assuming an average weight of 158.4 g for individual adult herring, estimated consumption by overwintering whales ranged from 17 to 102 million herring with an average of about 51.3 million adult herring (Table 7).

The abundance of the PWS adult herring population was about 900 million per year from 1980 through 1993 (estimated from Fig. 1) and about 200 million per year from 1994 to 2005 (Fig. 2). The overwintering whales appear to be consuming about 26% of the recent adult population. The abundance of age-3 herring in PWS was about 81 million fish per year from 1994 to 2005. We have no estimates of the age-2 population. Assuming the age-2 population to be generally greater than the age-3 population, the summering whales would then be consuming less than 24% of the age-2 population.

Poor nutrition and oceanic factors

Modeling by Rose et al. (2007) illustrates how temperature and other oceanic factors can influence weight at age and recruitment in herring populations. Oceanic factors have been found to influence herring recruitment in Alaska (Collie 1991; Williams and Quinn 2000a, b; Zebdi and Collie 1995). The analysis of 14 herring stocks by Williams and Quinn (2000a, b) show that the GOA stocks, which include the PWS and Sitka stocks, demonstrated similar interannual patterns in recruitment and weight at age that set them apart from herring stocks in the Bering Sea or British Columbia. The similarities among the GOA stocks indicate that they respond in a similar way to regional-scale environmental factors occurring over the GOA (Williams and Quinn 2000b).

Evidence for operation of oceanic factors in PWS and GOA comes from a comparison of recruitment (the numbers of herring at age-3 and age-4) that reveals that emergence of strong years occurred at the same time at PWS and Sitka from 1980 to 1993 (Fig. 2, Table 8). More importantly, strong year classes emerged at the same time in both PWS and Sitka until 1993; after 1993, strong year classes have not emerged in either PWS or Sitka (Fig. 2). Autocorrelation analysis supports such synchrony. The 4-year lag provides the highest positive autocorrelation for the numbers of both age-3 and age-4 herring at PWS and Sitka from 1980 through 1993 (Table 9). After 1993, the high positive autocorrelations at 4-year lag are no longer evident at either PWS or Sitka. The parallel trends are strong evidence that population dynamics of both PWS and Sitka herring are being influenced at the same time by GOA-wide ocean processes.
Table 8

Coefficients of determination for the numbers of herring at age 3 and age 4 between PWS and Sitka Sound, 1980 to 2005

Time period

Age

R-sq

P-value

All years

3

0.779

<0.001

4

0.727

<0.001

1993 and before

3

0.833

<0.001

4

0.823

<0.001

After 1993

3

0.449

0.017

4

0.282

0.067

Data from ADFG, Dressel (2006), Moffitt (2006)

Table 9

Autocorrelations in time series of the numbers of herring at age 3 and age 4 in PWS and Sitka Sound, 1980–2005

 

All years

1993 and before

After 1993

Lag

Autocorr

Lag

Autocorr

Lag

Autocorr

Sitka age 3

1

−0.06

1

−0.09

1

0.16

2

−0.3

2

−0.31

2

−0.18

3

−0.14

3

−0.07

3

−0.37

4

0.61

4

0.61

4

−0.13

5

−0.07

5

−0.09

5

0.37

6

−0.16

6

−0.24

6

0.17

Sitka age 4

1

−0.04

1

−0.08

1

0.31

2

−0.28

2

−0.22

2

−0.2

3

−0.13

3

−0.07

3

−0.23

4

0.59

4

0.61

4

−0.24

5

−0.06

5

−0.08

5

0.17

6

−0.16

6

−0.16

6

0.14

PWS age 3

1

−0.08

1

−0.28

1

0.08

2

−0.11

2

−0.33

2

−0.13

3

0.06

3

−0.02

3

−0.38

4

0.57

4

0.53

4

−0.18

5

−0.11

5

−0.3

5

0.18

6

−0.07

6

−0.26

6

0.07

PWS Age 4

1

−0.04

1

−0.29

1

0.09

2

−0.12

2

−0.28

2

−0.2

3

0.04

3

−0.09

3

−0.42

4

0.57

4

0.48

4

−0.24

5

−0.03

5

−0.29

5

0.17

6

−0.08

6

−0.16

6

0.14

Data from ADFG, Dressel (2006), Moffitt (2006)

Italics indicates significant difference at P ≤ 0.05

Further, the nutritional state of the PWS herring after the decline remains much poorer than that at Sitka (Table 10). With the data available, we may not be able to separate the relative roles of poor nutrition and other oceanic factors, but the field observations of Foy and Norcross (2001) and the analysis by Norcross and Brown (2001) suggest how these factors play a role in determining year-class strength in PWS herring. Elsewhere, changes in the ocean environment have apparently changed the zooplankton prey field and reduced overwintering larval survival in Atlantic herring, in turn leading to six sequential years of poor recruitment (Payne et al. 2009).
Table 10

Comparisons of weight at age and length at age of Pacific herring between PWS and Sitka Sound from 1983 to 1993 and from 1994 to 2005

Variable

1993 and before

P value

1994 and after

P value

PWS

Sitka

PWS

Sitka

Weight at age (g)

119.5

116.7

0.6

123.1

133.2

0.007

Length at age (mm)

205.7

202.2

0.02

210.6

209.8

0.59

Data from ADFG, Bruce (2006), Dressel (2006), Moffitt (2006)

Poor nutrition associated with oceanic factors is a plausible hypothesis for poor recovery in PWS herring and has empirical evidence for its occurrence. However, some additional process beyond environmentally-derived poor nutrition must be occurring in PWS but not at Sitka. The ASA modeling outcomes suggest that the effects of the hatchery releases of pink salmon fry operate as well and are of great influence on recruitment and recovery.

Competition with juvenile pink salmon

Results from the ASA modeling with covariates by Deriso et al. (2008) appear in Table 4. These results led us to evaluate mechanisms of competition and predation by which juvenile pink salmon could influence population dynamics in PWS herring. We assessed the competition hypotheses on the basis of both the ASA modeling results and the available information from field studies.

Juvenile pink salmon releases into PWS now average about 600 million annually (Fig. 8). ASA modeling by Deriso et al. (2008) found releases of hatchery pink salmon fry to consistently be the most influential single factor on PWS herring recruitment and the common factor in all their multi-factor models (Table 4). The impact analysis of Deriso et al. (2008) reveals that the pink salmon fry releases have the most substantial impact, reducing the spawning herring biomass to an estimated 20% of the level without the releases. However, Deriso et al. (2008) could did not address whether the mechanism behind these results was predation, competition, or a combination of the two.

Here we examine potential competition mechanisms by which the pink salmon fry releases influence PWS herring recruitment. To assess competition hypotheses, we examine several questions: (1) Do the competing species overlap in space and time? (spatio-temporal overlap), (2) Do the diets of the competitors include shared prey? (dietary overlap), (3) Does consumptive demand exceed food availability (consumptive demand), (4) Is the feeding of one competitor substantially reduced in the presence of the other? (feeding disruption), and (5) Does lowered prey availability or feeding disruption adversely affect growth and survival (growth and survival). To affirm a competition hypothesis, we must find evidence for all of the following: spatio-temporal overlap, dietary overlap, consumptive demand or feeding disruption, and effects of food limitation on herring growth and survival.

Spatio-temporal overlap

There is evidence for spatial and temporal overlap between juvenile pink salmon and juvenile herring. The degree and timing of overlap between pink salmon and age-0 herring differs from that between pink salmon and age-1 herring. Some previous discussions of the overlap (e.g., Cooney et al. 2001) have only considered the limited spatial overlap between age-0 herring and pink salmon in May and June.

Wild pink salmon fry enter the Sound in late April or May (Boldt and Haldorson 2003), and release of hatchery pink salmon fry is timed to optimally correspond with the spring zooplankton bloom, usually mid-May (Cross et al. 2005). Once salmon fry have entered the Sound, they initially aggregate in shallow, nearshore bays (Boldt and Haldorson 2003). Young-of-the-year salmon stay within the 20-m isobath (Okey and Pauly 1999). Furthermore, most of the hatchery-released pink salmon inhabit the corridor running from Port Valdez to Esther Island and to Knight Island Passage, roughly one-half of the Sound (Okey and Pauly 1999; Willette et al. 2001). Juvenile pink salmon can inhabit PWS through October but begin to migrate out of PWS into the coastal waters of the GOA in July (Armstrong et al. 2005). The vast majority of juvenile pink salmon have emigrated from the Sound by September.

The geographic distributions of age-0 and age-1 herring overlap substantially with the distribution of hatchery juvenile pink salmon within PWS. In Fig. 9, the one-dimensional migration route for hatchery pink salmon fry taken from Willette et al. (2001) is superimposed on the distribution of age-0 and age-1 herring in July of 1995 and 1996 taken from Norcross et al. (2001). Age-0 herring are in the larval drift in May and June and, after metamorphosis, enter PWS bays and fjords in June and July (Norcross et al. 2001; Stokesberry et al. 2000). From distribution studies over several years, Stokesbury et al. (1999, 2000) and Norcross et al. (2001) conclude that the age-0 and age-1 herring use the nearshore areas of PWS bays and fjords as nursery areas throughout the year. In western PWS, juvenile pink salmon overlap with age-0 herring from July onwards and with age-1 herring from May onwards. July is probably the time of maximum overlap when there are substantial numbers of juvenile pink salmon still within the PWS nearshore and both age-0 and age-1 herring are in PWS bays.
Fig. 9

The locations of Prince William Sound hatcheries, distribution of herring nursery areas and migration route of pink salmon. Migration route is redrawn from Willette et al. (2001); sympatric and allopatric aggregations, from Sturdevant (1999); herring distributions, from Norcross et al. (2001)

Several lines of evidence indicate that the extent of spatial and temporal overlap is biologically meaningful. First, Wertheimer and Celewycz (1996) state that their beach seining captured “large numbers” of Pacific herring and other fish “coincident with juvenile [pink] salmon” during April to June of 1989 and 1990 in PWS. Second, Willette et al. (1997) indicate that in August and September of 1994, at a low stock level for PWS herring, co-occurrence of juvenile pink salmon and juvenile herring in the same net hauls was 27%. Willette et al. (1997) used an anchovy purse seine that was 20-m deep; therefore, these are not beach samples. The mean length of the herring indicates that they were age-1 and with perhaps some age-2. Third, of the 17 stations described by Sturdevant (1999) where either allopatric or sympatric herring and pink salmon occurred, about 25% met a rigorous definition of sympatry, i.e., both species being captured in the same net haul (Fig. 9). Sturdevant’s (1999) study was not intended to be a survey of all the allopatric and sympatric conditions for juvenile herring and other species. In her discussion, Sturdevant suggests that sympatry among the forage fishes may be greater than her collection data suggest. Fourth, Moulton (1997) finds substantial numbers of juvenile pink salmon and juvenile herring occurring in the same tow net hauls in northern Cook Inlet. The tow nets were conducted away from the beaches and sampled to a depth of 3.1 m. Analysis of Moulton’s raw data shows that the sympatry between juvenile pink salmon and age-1 herring in tow net sets in the surface waters of northern Cook Inlet during the months of June and July is 81 and 58% (Table 11). Age-0 herring, but not juvenile pink salmon, are caught in September. The available field evidence indicates that the rate of co-occurrence of juvenile pink salmon and juvenile herring ranges from 25% to about 80% depending on timing.
Table 11

Montlhy catch of tow nets for juvenile Pacific herring and pink salmon in Cook Inlet, Alaska,1993, by month from raw data for Moulton (1997) provided by Moulton (pers. comm. 2006)

Category

Number of sets

Percent of sets with w/pink salmon and/or herring

June

July

September

June (%)

July (%)

September (%)

Total number of sets with pinks and/or herring

103

55

15

100.0

100.0

100.0

Sets with herring and pinks (co-occurrence)

83

32

0

80.6

58.2

0.0

Sets with herring and no pinks

6

12

15

5.8

21.8

100.0

Sets with pinks and no herring

14

11

0

13.6

20.0

0.0

Dietary overlap

Both herring and pink salmon juveniles are highly omnivorous, with both species consuming small and large copepods, larvaceans, and euphausiids (Sturdevant 1999; Boldt and Haldorson 2003). Both species prefer large copepods in May (Foy and Norcross 1999; Willette 2001). In general, pink salmon avoid small copepods (Cross et al. 2005), although small copepods may be underrepresented in diet studies because of their small size (Boldt and Haldorson 2003). The degree of dietary overlap varies, depending on the time of year, the date of sampling, and the study, but overlap in food resources appears greatest at the beginning of the summer, when both species are consuming the seasonally abundant calanoid copepods, and both are occupying nearshore habitats. Later in the summer, herring primarily consume smaller prey than that preferred by pink salmon but eat larger prey items such as larvaceans as a fraction of their diet. Larger prey items are not numerically abundant in the diets of juvenile herring, but these prey items contribute disproportionately to the overall amount of calories ingested and hence increase the energy content of juvenile herring (Sturdevant 1999).

Consumptive demand and prey availability

Cross et al. (2005) suggest that the consumption demands of pink salmon may exceed food availability in PWS. They used a bioenergetics model (Hansen et al. 1997) to predict the consumption demand of juvenile pink salmon, and compared the demand for certain prey with the availability of these taxa within the prey fields. Estimated consumptive demand of high-energy-content prey, such as large copepods, larvaceans, and hyperiid amphipods, exceeded supply in PWS. One advantage of the Cross study is that they were able to take zooplankton tows at the same sites where they collected fish and thereby were able to compare demand and regional availability. Sturdevant (1999) also found locally decreased zooplankton populations in areas with high densities of sympatric juvenile salmon and herring.

Disruption of feeding during sympatry

Sturdevant’s (1999) examination of dietary overlap of 14 species of forage fish collected from PWS includes field observations of feeding disruption that provide strong evidence of food competition between juvenile pink salmon and herring. When Pacific herring occur sympatrically with pink salmon, Sturdevant observed significant and substantial decreases for herring in four feeding indices: median stomach fullness, prey weight in stomach as a percentage of body weight, median prey number, and median prey biomass (Table 12). These observations indicate that when herring are food-limited in the presence of salmon, they do not shift diet composition, but instead decrease feeding. The observations of decreased feeding indices under sympatry led Sturdevant (1999) to conclude that if these fishes regularly co-occur, the carrying capacity of PWS may be exceeded. The spatio-temporal overlap observations discussed above indicate that co-occurrence is common.
Table 12

Feeding indices for allopatric and sympatric pink salmon and Pacific herring in PWS from Sturdevant (1999)

Species

Condition

Median stomach fullness (%)

Median prey as % body Wt

Median prey number

Median prey biomass (mg)

Herring

Allopatric

75

1.5

383.5

19.97

Sympatric

Trace

0.4

24

1.68

Significance (P=)

<0.0001

<0.0001

0.0001

0.0035

Pink Salmon

Allopatric

75

1.6

288.5

24.81

Sympatric

50

0.8

123

25.1

Significance (P=)

0.0138

<0.0001

0.255

0.0371

Note that 0.0371 may be typographic error. A value of 0.371 would match the conclusion in the text in Sturdevant (1999)

Effects of food limitation on growth and survival of juvenile herring

Field observations in PWS indicate that food limitation influences growth and survival of juvenile herring; the ability of herring juveniles to gain sufficient fat reserves to live through the winter determines their overall recruitment (Cooney et al. 2001; Norcross et al. 2001).

Paul and Paul (1998) experimentally starved young-of-the-year herring to determine the critical energy content of herring juveniles to survive winter. They found that fish that died during laboratory starvation had a whole-body energy content (WBEC) of 3.2 kJ g−1, whereas the WBEC of young-of-the-year herring in PWS was 3.4 kJ g−1, suggesting that juvenile herring use most of their reserves during the winter (Paul and Paul 1998). Fish with low WBEC are inferred to be at greater risk of predation, because they probably leave predator refuges in search of food (Paul and Paul 1998).

Field, laboratory, and modeling studies have led Norcross et al. (2001) to consider adequate nutrition (stored energy) to be essential for overwintering survival of age-0 herring, which in turn is considered important to year-class strength. Foy and Norcross (2001) reported that warmer water temperatures in nursery bays in 1997 were associated with lower zooplankton densities and taxonomic diversity, and that the frequency of empty stomachs among juvenile herring (a mixture of age-0 and age-1 in unknown proportion) was three times higher in October of the year with warmer water temperatures. In the past, strong year classes have typically emerged in PWS and other GOA herring populations every 4 years (Funk and Sandone 1990; Williams and Quinn 2000a; Hay et al. 2001). The 1996 year class would have been expected to be a strong one in the four-year cycle. However, the 1996 year class (age-1 at the time of the warmer water temperatures) had one of the lowest recruitments (less than one million age-3 herring) in the PWS recruitment time series (Fig. 2). Norcross and Brown (2001) and Norcross et al. (2001) suggest that no one early life stage is the “key” to year class strength in PWS herring, and the above observations suggest that, at least in the case at hand, nutritional state, low zooplankton densities, and warmer water temperatures at age-1 were associated with an extremely weak year class (Brown 2004; Brown and Norcross 2001).

Discussion

Based on Platt’s (1964) guidance for making inferences, we organize the summary information in Table 3 under these four questions: (1) Is there correlation or constant association between the hypothesized factor and herring mortality, biomass, or recruitment? (constant association test), (2) Is there a biologically plausible mechanism for the hypothesized factor to produce the observed effect? (mechanism test), (3) Is there evidence for the occurrence of such a mechanism? (test of supporting evidence for mechanism), and (4) Is there contrary evidence? (test for contrary evidence).

Decline

We find no evidence that oil exposure, harvest effects, spawning habitat loss, or the spawn-on-kelp fishery caused or contributed to the 1993 decline (Table 3). The evidence supports the contention that at least a portion of the PWS herring population was stressed by poor nutrition. Further evidence supports the argument that infectious disease effects during the decline were a secondary response to poor nutrition.

Most investigators now agree with the conclusion of Pearson et al. (1999) that the decline was not related to the oil spill. The link between the decline and the oil spill postulated by Thomas and Thorne (2003) and Thorne and Thomas (2008) has a plausible mechanism for exposure to oil during the initial year of the spill but lacks empirical evidence for its occurrence in PWS. The limited distribution of oil slicks over the large area of PWS and intermittent nature of gulping at the water surface indicates a limited exposure via gulping during the spill year and essentially negligible exposure after the initial year. Further, the results of ASA modeling (Deriso et al. 2008; Hulson et al. 2008) and other available evidence is contrary to the notion of a mortality event in adult herring during the spill year or year other than 1992 or 1993. Consequently, we reject the hypothesis that the decline started in 1989 and was associated with the 1989 EVOS.

The sum of the observations is not sufficient to assign VHSV as a cause of the 1992–1993 decline. Recent detailed analysis (Elston and Meyers 2009) demonstrates that the preponderance of evidence does not support any primary role of infectious disease in the decline. VHSV appears to have been a secondary response in an undetermined portion of the PWS herring population when the herring had already been stressed by poor nutrition.

The prevalence patterns of I. hoferi are contrary to the notion that this pathogen caused or contributed to the decline, although there is evidence that I. hoferi may play a role in PWS herring population dynamics some years after the decline. Because of the potential for sampling bias and changing measurement techniques, the data on the prevalence of I. hoferi requires cautious use. Despite these limitations, the prevalence patterns during the decline do not match the expected pattern if I. hoferi was a causative agent in the decline.

Poor nutrition preceding severe I. hoferi disease is a plausible mechanism for the recent emergence of the disease in PWS. Kramer-Schadt et al. (2010) hypothesize that the interactions of several factors drive the progression from I. hoferi infection to disease and point out that a dominant causal factor did not emerge from their study of herring epizootics. They suggest that high fish density facilitates disease transmission and that poor condition precedes disease expression. Laboratory experiments by Kocan et al. (1999a, b) showed that I. hoferi prevalence can increase with fish age without an associated increase in mortality. This finding suggested to Kramer-Schadt et al. (2010) that Pacific herring in good condition appear able to endure high infection intensities, although the fish may be at high risk if external conditions become unfavorable. Similar to VHSV, disease expression and morbidity caused by I. hoferi infection may result from, and perhaps exacerbate, the effects of poor nutrition and decreased condition, but there is insufficient knowledge to fully understand the apparent interactive effects of these factors. Certainly, more focused research may reveal the possible synergistic mechanisms by which these factors affect herring health and condition, but the available evidence continues to indicate that poor nutrition preceded disease expression for both VHSV and I. hoferi during the decline.

Our re-examination of available information and recent modeling outcomes supports earlier conclusions (Pearson et al. 1999) that poor nutrition is the probable cause of the 1993 decline. Nutritional status of PWS herring clearly began to decline in the mid-1980s and reached a low in 1993. Zooplankton abundance in PWS was low before, during, and following the decline. In recent years, other investigators have come to conclude that the 1993 decline most likely derived from poor nutrition with disease as a secondary factor (Carls et al. 2002; Marty et al. 2003; Hulson et al. 2008; Rice and Carls 2007; Elston and Meyers 2009).

Recovery

We find no evidence that oil spill effects or oil exposure, harvest effects, spawning habitat loss, or VHSV caused or contributed to poor recovery of PWS herring after 1993 (Table 3). Oceanic factors, poor nutrition, and interactions with hatchery releases of juvenile pink salmon are supported by the evidence. Available information about disease from I. hoferi and predation by overwintering whales suggest roles for these factors in limiting recovery but more information on these potential factors is needed to confirm them as causal.

We attribute the lack of strong year classes in both PWS and Sitka principally to oceanic factors operating at the scale of the GOA, and the nutritional state of the adult PWS herring remains poorer than that at Sitka and probably influences biomass levels of adult herring in PWS. The condition of adult herring in PWS would have been expected to improve under the low biomass levels after the decline. Some factor specifically related to PWS appears to be acting beyond the regional-scale oceanic factors.

ASA modeling clearly and strongly indicates 600-million-per-year juvenile pink salmon releases as a substantial factor affecting PWS herring recruitment. Evidence presented above supports the position that juvenile pink salmon releases affect juvenile herring through predation on age-0 herring and food competition with the age-1 herring.

Predation by juvenile pink salmon on age-0+ herring may be intense in mid-summer. Because age-0 herring are the most abundant forage fish in PWS waters (Okey and Pauly 1999), a substantial fraction of juvenile pink salmon predation likely occurs on herring. Also, the extent to which avoidance of such predation limits feeding by age-0 herring is not known. More specific identification of the fish found in the PWS juvenile pink salmon stomachs by Sturdevant (1999) would be useful. The hypothesis that juvenile pink salmon predation substantially influences age-0 herring survival has supporting evidence and no contrary evidence.

Our findings that juvenile pink salmon and age-1 herring probably compete for limiting food resources is in contrast with some other published reports on the interactions between these species. Cooney (1993) studied the influence of the release of hatchery-reared pink salmon fry on the carrying capacity of PWS and concluded that the release of hatchery-reared fry are unlikely to have much impact on zooplankton populations. Hence, Cooney (1993) did not expect that the fry releases would cause competition with other forage-fish species. However, his estimates were based on the total production (as grams of carbon) of broad trophic classes of organisms, such as zooplankton or planktivores. This simple approach assumes that all organisms within a given trophic class are potential prey for the next trophic level, regardless of size, energy content, or the ease of capture. This approach also does not include the potential for differences in prey quality or the selection of high-energy-content prey, which studies indicate have a strong bearing on growth and survival of PWS juvenile herring (Paul and Paul 1998; Cooney et al. 2001; Norcross et al. 2001).

If PWS adult herring condition has been declining since the mid-1980s, how does the strong PWS 1988 year class reconcile with oceanic factors and juvenile pink salmon releases as impediments to recovery? Pink salmon interactions are a combination of predation by pink salmon juveniles on age-0 herring and competition of pink salmon juveniles with age-1 herring. The 1984 and 1988 year classes were 1 year old in 1985 and 1989, respectively, which were two years of the highest zooplankton abundance (Fig. 4). The pink salmon fry releases first reached 600 million in 1989. There are no years of high zooplankton abundance in the years following 1989 (Fig. 4).

There is evidence for food competition between juvenile herring and juvenile pink salmon in PWS. Age-0 herring and pink salmon fry overlap in the nearshore areas of PWS from July through the fall. Age-1 herring and juvenile pink salmon overlap in PWS nearshore areas from May through the fall. There is sufficient overlap in the preferred food items among the juvenile forage fishes in PWS to indicate food competition, especially for prey of high-energy content. The field studies of Sturdevant (1999) found statistically significant and substantial decreases in four feeding indices of juvenile herring when they co-occur with juvenile pink salmon. Poor nutrition in juvenile herring, which could result from competition with pink salmon, is related to growth, survival, and recruitment (Foy and Norcross 2001; Norcross et al. 2001; Brown and Norcross 2001; Norcross and Brown 2001). Because juvenile herring need sufficient energetic reserves to survive the winter, decreased nutrition from competition in the spring and summer, followed by a poor prey field in the fall, likely would seriously compromise overwintering ability and, in turn, PWS herring year-class strength.

Marty et al. (2010) posit that I. hoferi affects both recruitment and biomass in PWS herring, but a mechanism for the effect on recruitment was not offered and is not clear. The potential for a biologically plausible mechanism for I. hoferi to produce low biomass appears to be greater. Poor condition in Atlantic herring was associated with severe I. hoferi infections (Kramer-Schadt et al. 2010). Also, I. hoferi affects older fish and a loss of older herring could decrease the biomass. But as discussed previously, poor nutrition appears to precede disease expression.

Marty et al. (2010) used ASA modeling with two disease covariates to assess the relative roles of VHSV and I. hoferi in the decline and poor recovery and concluded that both VHSV and I. hoferi had a substantial influence on the population dynamics of PWS herring―reducing the stock biomass to levels about three times below those that otherwise might have been. Deriso et al. (2008) also used ASA modeling but with 19 covariates and found some support for both VHSV and I. hoferi using the chi-square test but not when using the more accurate randomization test. Interactions with hatchery-release juvenile pink salmon proved to be the most consistent and strongest influence on PWS herring population dynamics (Deriso et al. 2008). Marty et al. (2010) does not include the hatchery releases in their ASA model so that the assessment of the higher relative influence of juvenile pink salmon from hatchery releases versus disease remains that of Deriso et al. (2008).

The impact of I. hoferi infection on herring populations is well documented regarding Pacific herring (Hershberger et al. 2006), and the ASA modeling of Deriso et al. (2008) (Table 3) and Hulson et al. (2008) indicate that I. hoferi could be influencing the population dynamics of PWS herring. The observed prevalence patterns of I. hoferi in PWS herring through 2006 (Table 5) suggest that the pathogen could be limiting survival of herring, particularly in the older year classes (Hershberger et al. 2002). Given the poorly understood relationship between infection intensity and prevalence on population mortality risk, we conclude that these factors need to be more fully understood before accurate conclusions and predictions relating I. hoferi infection to population effect can be made.

The influence of I. hoferi on PWS herring recruitment is not supported, but some secondary role of I. hoferi in producing low biomass in PWS herring cannot be dismissed. The increase in prevalence of I. hoferi after 2000 offers some evidence of a role and the outcomes of the more recent ASA models posit this hypothesis. The loss of older fish to disease from I. hoferi offers a plausible mechanism for lower biomass. Experimental examination of the disease process to determine whether poor condition is the precursor to disease related morbidity and mortality is needed before we can more definitively assess whether there is a causal role for I. hoferi in the low post-decline biomass levels in PWS herring.

Whale predation appears to be reducing adult biomass rather than the sub-adult biomass. Summer feeding by humpback whales appears to be primarily on euphausiids. Whale predation during the summer is on the 2-year-old herring because age 0 and age 1 herring are in the nursery areas and the adults are in the GOA. Our estimates indicate that summering whales may be taking less than 24% of the PWS recent sub-adult herring population. Because there are no estimates of the age-2 biomass or numbers, we used the age-3 abundance in our estimate. Consequently, a more realistic estimate from the available information would be that summering whales remove substantially less than 24% of the sub-adult herring population.

Our estimates are sensitive to the daily consumption rate, the proportion of herring in the diet and the number of whales. Our estimates for the overwintering period are especially sensitive to the high number of whales (112) observed in October and by the preliminary diet assumption that 80% of the overwintering whale diet is herring. This predation falls on adult herring overwintering in PWS. Because of the wide range of potential influence that whale predation may have on adult herring mortality, we recommend assessments of overwintering whale abundance and diet over several years and some studies to confirm the summer diet.

From ongoing studies, overwintering whales now appear to be in greater abundance than previously supposed and appear to feed on herring in the fall and winter. Our estimates indicate that overwintering whales may be taking about 26% of the recent levels for the adult population. Preliminary estimates from ongoing bioenergetics modeling suggest that overwintering whales take 18% to 23% of the pre-winter PWS herring biomass (Rice 2008). Whereas these estimated levels of whale predation are approximately equivalent to a 20% commercial harvest and worthy of concern, the wide range of estimates, the variety of assumptions, and the preliminary nature of recent observations make it premature to declare that a classic “predator pit” from whale predation is the cause of poor recovery in PWS herring.

Furthermore, whale predation cannot be the sole factor impairing recovery in PWS herring. Condition and weight at age has not rebounded in PWS. Enhanced growth under lower population levels would have occurred if whale predation was removing large portions of the sub-adult and adult population and other factors remained the same. Perhaps, overwintering whales take not only herring but also consume large amounts of the euphausiid prey also needed for growth and survival of overwintering adult herring. Whereas whale predation appears to be a credible hypothesis for contributing to low biomass, how whale predation influences recruitment and lack of strong year classes remains less clear. We need to examine other factors in PWS that would affect recruitment.

The recovery hypotheses remaining are not mutually exclusive and could be acting on different aspects of herring population dynamics at the same time. A multifactor hypothesis would include the following elements. First, regional-scale oceanic factors set the limits for the prey field and, in turn, the growth and survival of PWS herring. Second, interactions with pink salmon juveniles released from the hatcheries reduce age 0 herring survival through predation and reduce age 1 growth and survival through food competition and feeding disruption. Third, disease as a secondary response following poor nutrition could reduce biomass levels. Fourth, overwintering whales could remove large portions of adult herring biomass through predation.

Implications for management and restoration

To be most effective, any management or restoration efforts should address the mechanism behind the decline and lack of recovery. If the mechanisms are still operational, then restoration efforts that enhance the population before any bottleneck will be wasted effort. For example, because spawning habitat is not limiting in PWS, enhancing spawning habitat is unlikely to improve herring recovery.

When oceanic factors are influencing recovery, then little can be done other than to manage the commercial herring harvest. It is important to remember that several important herring stocks took decades to recover even under severe fisheries management efforts. Recovery of PWS herring could well depend on a future change in the conditions of the marine environment and, especially, the prey field.

Similarly, how to influence recovery if disease is either a primary or secondary casual factor is elusive. Perhaps all the fisheries manager can do is to incorporate the disease variables, along with all other available and relevant variables, into an ASA model to predict the spawning biomass more accurately. If the disease variables are a proxy for a population showing the effects of an environmental stressor rather than a causal factor, their incorporation in the ASA models may still be of practical use in fisheries management. Marty et al. (2010) cites recent instances of forecasts that prevented unnecessary mobilization of the herring fishing fleet. The available information indicates that closer attention to sampling design and the use of the improved techniques to measure I. hoferi need to be incorporated into any field efforts to monitor infectious diseases in PWS herring.

Whale predation may prove to be removing large amounts of PWS herring, but protected status precludes manipulating whale populations or individuals. The only viable remedy to any effects of whale predation may still be to manage the commercial herring harvest with improved forecasts of the influence of whale predation. Tjelmeland and Lindstrom (2005) found that including minke whale predation in a stock assessment model of Barents Sea herring was both feasible and beneficial in improving stock assessments to support management decisions. Adding whale predation to PWS assessment models will require ongoing studies of whale abundance and predation to be completed over a sufficient number of years to provide reliable long-term information.

Improving nutrition among overwintering juvenile herring clearly would be beneficial to PWS herring recruitment, but how to do so on a scale that would be effective is not immediately obvious. One approach may be to switch the timing of hatchery pink salmon fry release from the peak of plankton production to smaller releases over a longer time period. Reduction in the total amount of releases may also be effective in preventing juvenile pink salmon from depleting any zooplankton stocks that also benefit juvenile herring. If ocean survival of juvenile and adult salmon is the main factor governing the magnitude of returns, then an evenly timed release or even a reduced total release may not have a great influence on the overall returns of adult pink salmon to PWS.

In their review of the lessons learned from the collapse and recovery of North Sea herring, Dickey-Collas et al. (2010) call for not only more research on the mechanisms for collapse and recovery but also for a move away from single-species management approaches to multispecies management and, in the long term, to ecosystem management. When commitments to the management of several species are in conflict (which appears to be the case with PWS), reconciling the conflicting commitments will be best done at the ecosystem level (Dickey-Collas et al. 2010).

Conclusions

We have assessed the evidence for and against the principal decline and poor recovery hypotheses and find no evidence that oil exposure from the Exxon Valdez oil spill, harvest effects, spawning habitat loss, the spawn-on-kelp fishery, or disease have led to either the decline or poor recovery of PWS herring.

Our re-examination of available information and recent modeling outcomes supports earlier conclusions (Pearson et al. 1999) that poor nutrition is the probable cause of the 1993 decline. Nutritional status of PWS herring clearly began to decline in the mid-1980s and reached a low in 1993 and was associated with low zooplankton abundance. Recent information, especially the disease analysis of Elston and Meyers (2009), decreases the role of disease in the decline. The evidence supports the contention that disease during the decline was a secondary response after a portion of the PWS herring population was stressed by poor nutrition.

Poor recovery probably results from several factors. Since 1993, no strong year classes have emerged in the GOA herring populations, including that at PWS. This lack of strong year classes appears to derive from regional-scale ocean environmental factors. Beyond the regional-scale factors, two other factors specific to PWS appear to be reducing herring biomass and recruitment. First, predation by an increasing number of overwintering humpback whales may prove to be removing a substantial proportion of the adult herring in PWS. Second, interactions with juvenile pink salmon released from PWS hatcheries may be influencing nutrition in juvenile herring and their subsequent growth, survival, and recruitment. Continued research is recommended on PWS humpback whale predation and on herring recruitment processes. Also recommended is a move away from single-species management to multi-species management, if not to the management of PWS as a whole ecosystem.

Acknowledgments

We thank Michael Anderson, Michael Cobb, Lee Miller, Ann Skillman, Kathryn Sobocinski, and John Southard of Battelle Marine Sciences Laboratory in Sequim, Washington, and Karen Humphrey of Aquatechnics, for their dedicated efforts in compiling and presenting the information in this review. We thank Mark N. Maunder of Quantitative Resource Assessment LLC for his modeling aid in this effort. We thank Lawrence L. Moulton of MJM Research for sharing his raw data on the co-occurrence of pink salmon and herring in his tow net samples. We thank Kenneth A. Rose of Louisiana State University and John R. Skalski of the University of Washington for valuable discussions and for reviewing an earlier version of this paper. We also thank Douglas E. Hay of Nearshore Consulting for reviewing earlier manuscripts. Exxon Mobil Corporation supported this effort.

Open Access

This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.

Copyright information

© The Author(s) 2011

Authors and Affiliations

  • Walter H. Pearson
    • 1
  • Richard B. Deriso
    • 2
  • Ralph A. Elston
    • 3
  • Sharon E. Hook
    • 4
  • Keith R. Parker
    • 5
  • Jack W. Anderson
    • 4
  1. 1.Peapod ResearchAllendaleUSA
  2. 2.Inter-American Tropical Tuna Commission (IATTC)Scripps Institution of OceanographyLa JollaUSA
  3. 3.AquaTechnics, Inc.CarlsborgUSA
  4. 4.Battelle Marine Sciences LaboratorySequimUSA
  5. 5.Data Analysis GroupCoverdaleUSA

Personalised recommendations