C05__Attachment_11_AFWO-F-02-03.pdf

U.S. Fish & Wildlife Service U.S. Fish and Wildlife Service

Klamath River Fish Die-off September 2002

Causative Factors of Mortality

Klamath River Fish Die-off September 2002

Report Number AFWO-F-02-03 ii

Executive Summary

This report provides information describing the biological, hydrological, meteorological, and water quality conditions associated with the die-off of an estimated 34,056 fish in the Klamath River, California in September 2002. The proximate cause of death was heavy infections of two fish pathogens, Ich and columnaris. However, given that these ubiquitous pathogens are normally found in the Klamath River, additional factors must have played a role for them to have become lethal. It is our conclusion based on multiple lines of evidence that the fish die-off in the lower Klamath River in 2002 was a result of a combination of factors that began with an early peak in the return of a large run of fall Chinook salmon. Low river discharges apparently did not provide suitable attraction flows for migrating adult salmon, resulting in large numbers of fish congregating in the warm waters of the lower River. The high density of fish, low discharges, warm water temperatures, and possible extended residence time of salmon created optimal conditions for parasite proliferation and precipitated an epizootic of Ich and columnaris. Based on a review of available literature and historical records, this was the largest known pre-spawning adult salmonid die-off recorded for the Klamath River and possibly the Pacific coast.

iii

TABLE OF CONTENTS

INTRODUCTION

METHODS

Literature Review Fish Mortality Assessment Supplemental Field Observations Fish Pathology Tidal Data Meteorological Information River Discharge Water Quality

Continuous Water Quality Monitoring Grab Samples

Fish Tissue Samples for Contaminants Hazardous Materials Spills and Reports Permitted Point Source Discharges Hazardous Waste Sites Comparison of Salmon Run Size with Environmental Factors

Salmon Run Size River Discharge Cluster Analysis

Fall-Run Chinook Timing Fish Passage

RESULTS AND DISCUSSION

Mortality Estimates Supplemental Field Observations

Phytoplankton Odors and Discolored Water Green Sturgeon Tracking Data

Fish Pathology Ichthyophthirius multifiliis or Ich Flavobacterium columnare or Columnaris Relationship of Pathogens and Spawning Activity

Tidal Stages Precipitation River Discharges

Mainstem Discharges, August through October 2002 Tributaries, August through October 2002 iv

Combined Hoopa and Orleans Gage Discharges, August through October 2002

River Discharge Summary, August through September 2002 Historical versus 2002 discharges River Discharges Summary

Air Temperature Water temperature

Water Temperature - Weitchpec (Klamath RM 43.5) - 2002-2001 Water Temperature - Martin’s Ferry (Klamath RM 40.4) - 2002-2001 Water Temperature - Terwer (Klamath RM 6.7) - 2002-2001 Water Temperature - Inter-annual Station Comparisons All Sites Water Temperature - USGS Report Water Temperature - Grab Samples Water Temperature Summary

Dissolved Oxygen Dissolved Oxygen - Weitchpec - (Klamath RM 43.5) - 2002-2001 Dissolved Oxygen - Martin’s Ferry - (Klamath RM 40.4) - 2002-2001 Dissolved Oxygen - Terwer - (Klamath RM 6.7) - 2002-2001 Dissolved Oxygen - Grab Samples Dissolved Oxygen Summary pH pH - Weitchpec (Klamath RM 43.5) 2002 - 2001 pH - Martin’s Ferry (Klamath RM 40.4) 2002 - 2001 pH - Terwer - (Klamath RM 6.7) 2002 - 2001 pH - Grab Samples pH Summary

Specific Conductance Specific Conductance - Weitchpec (Klamath RM 43.5), Martin’s Ferry (Klamath

RM 40.4), Terwer (Klamath RM 6.7) Specific Conductance - Grab Samples

Other Water Quality Measurements - Grab Samples Chlorophyll-a Ammonia Nitrogen Biochemical Oxygen Demand (BOD) Turbidity Total Suspended Solids Nitrates Total Phosphorus and Orthophosphorus Periphyton Chlorinated Hydrocarbons and Pesticides in Water

Fish Tissue Sample v

Chemical Spill Reports Permitted Discharges Hazardous Waste Sites Relationship among Run Size, River Discharge and Precipitation

Run Size and Discharge Run Size and Precipitation Summary

Cluster Analysis - Run Size 1978 to 2002 versus Discharge and Precipitation Fish Density: Effects of Run Timing, Harvest, and Fish Behavior

Summary Fish Passage

Summary Historic Fish Die-offs

Salmonids Smallscale Sucker Mortality - 1997

CONCLUSIONS

LITERATURE CITED

vi

LIST OF FIGURES

Figure 1. Klamath River basin showing fish die-off sampling areas

Figure 2. Map of lower Klamath River depicting reaches surveyed during fish die-off.

investigations conducted during September and October 2002

Figure 3. Life cycle of Ichthyophthirius multifiliis. Trophonts not to scale. Trophonts normally appear as white grain sized spots on fish. From (Durborow et al. 1998)(Drawing by Wyvette Williams and Drew Mitchell)

Figure 4. Tidal stages measured from 9/1/02 to 10/31/02 at the NOAA tide gage in Crescent City, California

Figure 5. Daily precipitation measured at the National Weather Service meteorological stations located in Fort Jones near the Scott River, Yreka near the Shasta River and Klamath near the lower Klamath River. No precipitation was recorded at the Hoopa (Trinity River), Sawyers Bar (Salmon River) or Orleans gages (Klamath River) during this period of time.

Figure 6. Precipitation measured at Orleans, Klamath and Hoopa during August, September and October 2002 compared with historical average values. Data obtained from the Western Regional Climate Center (2003)

Figure 7. Comparison of discharge regimes between major tributaries and the Klamath River mainstem below Iron Gate Dam from July 1 through November 1, 2002. Data obtained from USGS data web site

Figure 8. Combined mean daily discharges measured at U.S. Geological Survey gages located in the Klamath River Basin near Orleans (Klamath RM 59.1) and Hoopa (Klamath RM 43.5, Trinity RM 12.4) from July 1 to October 31, 2002. Information based partly on provisional discharge data. Discharges from day before and the Terwer gage (Klamath RM 6.7) site are also plotted for comparison. Data obtained from USGS web site

Figure 9. Air temperatures (maximum, mean, median, and minimum) observed at the Yurok Weitchpec weather station. Continuous line denotes mean value. Data generated from raw half hour readings

Figure 10. Maximum, mean, and minimum air temperatures observed at the Orleans National Weather Service weather station during August through September 2002. Summary data provided by National Weather Service vii

Figure 11. Maximum, mean and minimum air temperatures observed at the Klamath National Weather Service weather station during August and September, 2002. Summary data provided by National Weather Service

Figure 12. Average daily air temperature measured at the Orleans National Weather station from 1940 to 2002 during August through October. August, September and October temperatures averaged 22.6, 20.1 and 14.8 °C during the period of record

Figure 13. Average daily air temperature measured at the Klamath, California National Weather station from 1940 to 2002 during August through October. August, September and October temperatures averaged 15.3, 14.8 and 12.9 °C during the period of record. . . 66

Figure 14. Daily average, maximum, and minimum water temperatures and average daily discharge recorded at the Weitchpec (RM 43.5), Martin’s Ferry (RM 40.4) and Terwer (RM 6.7) on the Klamath River during May to October in 2001 and 2002 using Hydrolab® datasondes. Discharges derived from USGS gage sites at Orleans (RM 59.2) , Hoopa/Orleans (Klamath RM 43.5 + Trinity RM 12.4) , and Terwer gages respectively.

Figure 15. Maximum daily water temperatures measured at the Weitchpec (2001WE and 2002WE, RM 43.5), Martin’s Ferry (2001MF and 2002MF, RM 40.4), Terwer (2001 and 2002, T, RM 6.7), below Blue Creek (1995, BC, RM 16.4), below Coon Creek (1995, CC, RM 35.9), below confluence with Omagar Creek (1997-1999, RM 10.5) using Hydrolab® datasondes or Hobo® tidbits (1997-99). Recommended EPA criteria for migrating salmonids (21°C) and reduction of probability of severe disease outbreak (18 °C) are depicted (EPA, 2002)

Figure 16. Box and means plot of August and September maximum daily water temperatures measured at the Weitchpec (2001WE and 2002WE, RM 43.5), Martin’s Ferry (2001MF and 2002MF, RM 40.4), Terwer (2001 and 2002, T, RM 6.7), below Blue Creek (1995, BC, RM 16.4), below Coon Creek (1995, CC, RM 35.9), and Omagar Creek (1997-1999, RM 10.5) using Hydrolab® datasondes or Hobo® tidbits (1997-1999). Horizontal bar within box is the median; box is 25th and 75th percentiles; vertical line is remaining non-extreme values; * denotes extreme values; C denotes mean. Proposed Region 10 EPA criteria depicted at 21 °C (spawning migration) and 18 °C (reduction in disease) viii

Figure 17. Daily average, maximum, and minimum dissolved oxygen and average daily discharge recorded at the Weitchpec (RM 43.5), Martin’s Ferry (RM 40.4) and Terwer (Klamath RM 6.7) on the Klamath River during May to October in 2001 and 2002 using Hydrolab® datasondes. Discharges derived from USGS gage sites at Orleans (RM 59.2), Hoopa/Orleans (Klamath RM 43.5 + Trinity RM 12.4), and Terwer gages respectively.

Figure 18. Daily average, maximum, and minimum pH and average daily discharge recorded at the Weitchpec (RM 43.5), Martin’s Ferry (RM 40.4) and Terwer (Klamath RM 6.7) during May to October in 2001 and 2002 using Hydrolab® datasondes. Discharges derived from USGS gage sites at Orleans (RM 59.2) , Hoopa/Orleans (Klamath RM 43.5 + Trinity RM 12.4), and Terwer gages respectively

Figure 19. Total and adult in-river Chinook run during 1978 through 2002. Run size data compiled from CDFG megatable and KRTAT records (CDFG 2003a). The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure 20. Relationship between total in-river Chinook run size and daily average August discharges measured at the Terwer gage during 1978 through 2002. Run size data compiled from CDFG megatable and KRTAT records (CDFG 2003a). Discharges during 1996 and 1997 were estimated from up river discharges using the following equation:

Terwer gage discharge = - 171 + 1.33 (Hoopa gage discharge) + 1.36 (Orleans gage discharge), r2 = 92.9. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure 21. Relationship between total in-river Chinook run size and daily average September discharges at the Terwer gage during 1978 through 2002. Run size data compiled from CDFG megatable and KRTAT records (CDFG 2003a). Discharges during 1996 and 1997 were estimated from up river discharges using the following equation: Terwer gage discharge = - 123 + 1.12 (Hoopa gage discharge) + (1.39 Orleans gage discharge), r2 = 88.

The 2002 run size may be biased low due to conservative fish die-off number estimates.

Figure 22. Relationship between total in-river Chinook run size and combined August average discharges measured at the Hoopa/Orleans gages during 1978 through 2002. Run size data compiled from CDFG megatable and KRTAT records (CDFG 2003a). The 2002 run size may be biased low due to conservative fish die-off number estimates ix

Figure 23. Relationship between total in-river Chinook run size and combined September average discharges measured at the Hoopa/Orleans gages during 1978 through 2002. Run size data compiled from CDFG megatable and KRTAT records (CDFG 2003a). Orleans gage average discharges were adjusted by deleting data from September 28-31, 2002, which removes the influence of the pulse discharge. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure 24. Relationship among the eight highest total in-river Chinook run sizes and August and September discharges measured at Lewiston (Le), Hoopa (Ho), Iron Gate (IG), Orleans (Or), and Terwer (Te) gages. Combined Hoopa/Orleans (OrH) discharges were also calculated. Run size data was compiled from CDFG megatable (CDFG 2003a). Iron Gate, Orleans, and Hoopa/Orleans average discharges were adjusted by deleting data from the September 2002 pulse discharge. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure 25. Relationship among the eight highest total in-river Chinook run sizes and August and September percentage of Hoopa/Orleans (OrHo) and Terwer discharges composed of Iron Gate discharges. Run size data was compiled from CDFG megatable (CDFG 2003a). Iron Gate, Orleans, and Hoopa/Orleans average discharges were adjusted by deleting data from the September 2002 pulse discharge. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure 26. Relationship among the four years with the lowest flows and high total in-river Chinook run sizes and August and September discharges measured at Lewiston (Le), Hoopa (Ho), Iron Gate (IG), Orleans (Or), and Terwer (Te) gages. Combined Orleans and Hoopa (OrH) discharges were also calculated. Run size data was compiled from CDFG megatable (CDFG 2003a). Iron Gate, Orleans, and combined Orleans and Hoopa average discharges were adjusted by deleting data from the September 2002 pulse discharge. The 2002 run size may be biased low due to conservative fish die-off number estimates. . . 79

Figure 27. Comparison of precipitation measured at Klamath and Orleans during August and September in 2002 and years with larger total runs of Chinook. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure. 28. Dendrogram produced from cluster analysis showing classification of various years.

Ward’s clustering algorithm and squared Euclidean distance were used. Variables included August and September average discharges at Scott River, Klamath River at Iron Gate Dam, Shasta River, Salmon River, and the Trinity River at Hoopa. September Iron Gate discharges were adjusted for pulse discharge release from Iron Gate on 9-27-02. . 81 x

Figure 29. Comparison of CDFG lower Klamath River data recreational creel catch per unit effort for 2002 (group 2), 2001 (Group 1), 1988 (Group 1), 1992 (Group 2), 1991 (Group 2), 1994 (Group 2), 1995, and 1996. Data Source: CDFG (2003b)

Figure 30. Comparison of tribal net harvest catch per unit effort for 1994-2002 in the lower Klamath River and estuary for the August 1 - September 28 time period, except for the periods that the fishery was closed during 1994, 1995, and 2000

Figure 31. Karuk Tribal Fishery catch per unit effort data collected at Ishi Pishi falls (RM 66.5) during 2002. Daily average discharge data obtained from USGS Orleans station

Figure 32. Willow Creek weir counts of migrating salmon during 2002, 2001 and the four lowest years of record based on September discharges at Klamath (Terwer). Data source: CDFG (2003b) xi

LIST OF TABLES

Table 1. Summary of green sturgeon movement obtained from preliminary combined radiotelemetry and acoustic monitoring data

Table 2. Comparison of discharge regimes at various gage sites for the period of record during August and September. Orleans Hoopa is the combined discharge from the Hoopa and Orleans gage

Table 3. Results of water quality monitoring for water temperature, specific conductance, pH, dissolved oxygen, chlorophyll-a, and ammonia nitrogen during 2001 and 2002 in the lower Klamath River

Table 4. Results of water quality monitoring for turbidity (NTU), 5 day biochemical oxygen demand (BOD), nitrate nitrogen (NO3-N), total suspended solids (TSS), total phosphate phosphorus (total-P) and orthophosphate phosphorus (Ortho-P) during 2001 and 2002 in the lower Klamath

Table 5. Periphyton growth at Terwer during the period of the fish die-off. Slides were incubated for approximately two weeks prior to removal

Table 6. List of analytes examined in water samples collected during 9/26/02 and 10/8/02. . . 91

Table 7. Results of chemical analysis of coho gill tissue CDFG laboratory. Coho collected on 10- 3-02 at Blue Creek mouth. Fish was 77 cm FL and possessed right maxillary clip

Table 8. Estimated immigration timing of major stocks of salmonids into the lower Klamath River based on review of literature

Table 9. Fall Chinook spawning escapement (jacks and adults) for the Klamath and Trinity sub-basins, 1978-2002

Table 10. Total fall Chinook Tribal harvest and semi-monthly catch effort (Chinook/net-hour) for Agust and September in the Klamath River Estuary Area, 1984-1992, 1994-2002 xii

APPENDICES

APPENDIX A. CLUSTER ANALYSIS: RUN SIZE AND RELATIONSHIP TO DISCHARGE

AND PRECIPITATION

Methods Results Comparison of IGD Gage Discharges by Cluster Groups Comparison of Orleans Gage Discharges by Cluster Groups Comparison of Terwer Gage Discharges by Cluster Groups Comparison of Combined Orleans and Hoopa Gage Discharges by Cluster Groups . . . 99 Tributary - Shasta River at Yreka Tributary - Scott River at Fort Jones Tributary - Salmon River at Somes Bar Tributary - Trinity River at Lewiston Dam Comparison of Run Sizes by Cluster Groups Comparison of Run Size, Discharges, and Precipitation by Cluster Groups - Overall

Patterns Summary Figure A.1. Comparison of two major clusters of years using August and September discharges, including pulseflow, from Iron Gate and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates.

Adjusted September 2002 discharges were 760 cfs after removing the pulse flow versus 813 cfs (unadjusted)

Figure A.2. Comparison of two major clusters of years using August and September discharges from Orleans and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates. Adjusted September 2002 discharges were 1,287 cfs after removing the pulse flow versus 1,305 (unadjusted)

Figure A.3. Comparison of two major clusters of years using August and September monthly average discharges, including pulse flow, measured at Terwer (Klamath) and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates. The adjusted average discharge for September 2002, was 1,987 cfs versus 1,993 cfs (unadjusted)

Figure A.4. Comparison of two major clusters of years using August and September combined discharges, including pulse flow, measured at Hoopa/Orleans and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates. Adjusted September 2002 dishcharges were 1,918 cfs after removing the pulse flow versus 1,936 cfs (unadjusted) xiii

Figure A.5. Comparison of two major clusters of years using August and September discharges measured at Shasta River at Yreka gage and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates.107

Figure A.6. Comparison of two major clusters of years using August and September discharges measured at Scott River Fort Jones gage and total Chinook run size.

The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure A.7. Comparison of two major clusters of years using August and September discharges measured at the Salmon River at Somes Bar gage and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure A.8. Comparison of two major clusters of years using August and September discharges measured at the Trinity River at Lewiston Dam gage and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure A.9. Comparison of two major clusters of years using August and September discharges measured at the Hoopa gage on the Trinity River and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure A.10. Comparison of two major clusters of years using August and September monthly total precipitation measured at the Orleans weather station, and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates

Figure A.11. Comparison of two major clusters of years using August and September monthly total precipitation measured at the Klamath weather station, and total Chinook run size. The 2002 run size may be biased low due to conservative fish die-off number estimates

Table A.1. Comparison of yearly cluster groups 1 and 2 using major mainstem gage stations

Table A.2. Comparison of yearly cluster groups 1 and 2 using tributary gage stations and mainstem weather stations

INTRODUCTION

The California Department of Fish and Game (CDFG) received a verbal report of dead and dying salmon in the lower Klamath River on September 19, 2002. The caller indicated that on the previous day he had seen large numbers of dead and dying salmon in the lower river. The CDFG then notified the U.S. Fish and Wildlife Service (USFWS) about the incident. The subsequent investigation involved collaboration with representatives of the Yurok Tribal Fisheries Program, Hoopa Valley Tribe, Karuk Tribe of California, and CDFG.

Various factors were investigated as potential causes for the fish die-off. These factors included poor water quality, pathogens, reduced river discharges, large run size and associated crowding, impaired migration, run timing, and environmental contaminants. This report evaluates and identifies those factors most likely associated with and/or contributing to the observed fish die-off.

Information examined included recent and historical water quality, hydrology, meteorology, fisheries data, agency reports, scientific articles, and pathology reports. This report provides an analysis of the immediate and contributing factors associated with this fish die-off, and conclusions regarding the most likely causative factors.

METHODS

Literature Review

In order to evaluate the relative influence of each potential factor, a search of existing data and other pertinent background literature on anadromous salmonid die-offs and associated variables related to pathogen infection including discharge, run size, run timing, and water quality was conducted. Reports of massive die-offs or cases of pre-spawning mortality of salmon are rare.

Fish Mortality Assessment

Methods for estimating numbers of fish that died during the die-off are described in more detail in a report by Guillen (2003). Biological data collected during the die-off included presence or absence of dead fish, species identification, numbers of dead fish, coded wire tags (CWT) for determining hatchery composition, length measurements, scales for aging, disease incidence, and decay condition. These data were collected during three primary survey dates (September 20, 24, and 27, 2002) at four mainstem survey reaches below Coon Creek at Klamath river mile (RM) 36 (Figures 1 and 2). These observations were supplemented by various other surveys conducted throughout the event.

Supplemental Field Observations

During the response to the fish die-off, biologists made observations regarding the appearance of the Klamath River such as color, turbidity, and odors of the water, and examined a few samples of water for phytoplankton. In addition, green sturgeon in the area of the die-off were tracked using mobile radio receivers and stationary acoustic receivers. These fish had been tagged previously and were tracked from May through December 2002 by the Yurok Tribal Fisheries Program and U.S. Fish and Wildlife Service (USFWS). Preliminary data were examined to determine potential impacts of the die-off on this species.

Fish Pathology

Fish tissue samples were collected and external necropsy examinations were conducted on specimens collected on September 26, 2002, near Blake’s Riffle (Klamath RM 7.3) and at the confluence of Blue Creek (Klamath RM 16.4) by USFWS and CDFG fish pathologists (Foott 2002; Veek 2002). Necropsy samples were collected from dead and moribund fish. In addition, numerous field pathological examinations were conducted by USFWS, Yurok Tribal Fisheries, and CDFG biologists while counting dead fish.

Tidal Data

Tidal data were obtained from the National Oceanic Atmospheric Administration (NOAA) real-time tide gages located at Crescent City, approximately 25 miles north of the mouth of the Klamath River. Our purpose was to determine tidal conditions during and prior to the die-off and possible mechanisms that could affect river discharge in the lower river or salmon immigration.

Meteorological Information

Air temperature and precipitation data were obtained from National Weather Service (NWS) stations located at Klamath (Klamath River), Hoopa (Trinity River), Orleans (Klamath River), Yreka (Shasta River), Sawyers Bar (Salmon River), and Fort Jones (Scott River) (Figure 1). In addition, recent data collected at the Yurok weather station at Weitchpec were made available to the USFWS by the Yurok Tribe.

River Discharge

River discharge refers to the volume of water passing through a channel during a given time, usually measured in cubic feet per second (cfs). Average daily and monthly river discharge measurements were obtained from the U.S. Geological Survey (USGS) gage sites located below Iron Gate Dam (Klamath RM 189.8), Shasta River (Shasta RM 0.5), Scott River near Ft. Jones (Scott RM 21), Salmon River (Salmon RM 1.01), Orleans (Klamath RM 59.1), Trinity River at

Lewiston Dam (Trinity RM 111), Trinity River at Hoopa (Trinity RM 12.4), and Terwer (also called Turwar or Klamath) (Klamath RM 6.7) (Figure 1). Historical and recent provisional data were obtained from the USGS web site.

Water Quality

Continuous Water Quality Monitoring

Continuous water quality data including water temperature, dissolved oxygen, pH, and specific conductance were measured using Hydrolab® datasondes operated by the Yurok Tribe and the USFWS monitoring network in the lower Klamath River. These data were collected at several fixed stations in the lower River including Weitchpec (Klamath RM 43.5), Martins Ferry (Klamath RM 40.4), and Terwer (Klamath RM 6.7) (Figure 2). These were monitored daily at 30-minute intervals continuously from June through October 2002. Data from August through October 2002, are presented in this report. Prior to deployment and upon retrieval the instruments were calibrated. In some cases Hydrolab® datasonde temperature data were augmented with data collected from co-located Hobo® tidbit thermistors.

Data collected during 2002 were compared with the 2001 USFWS/Yurok Tribal water quality monitoring data from the Martin’s Ferry and Terwer sites. We also compared the 2002 data with the data from 1995 collected by the USFWS at mainstem sites located below Blue Creek (Klamath RM 16.4), and below Coon Creek (Klamath RM 35.9) using Hydrolab® datasondes.

Finally, continuous water temperature data collected during 1997 through 1999 were provided by the Yurok Tribal Fisheries Program for Omagar Creek (Klamath RM 10.5). These studies used Hobo® tidbit thermistors.

Grab Samples

Grab samples are a single water sample drawn over a short time period. Water grab samples were collected and analyzed from July 30 to October 8 during 2002 as part of the USFWS/Yurok routine sampling program and in support of the die-off investigation on a bi-weekly to weekly basis. The primary sites sampled were Weitchpec, Martin’s Ferry, and Terwer, which are also continous water quality monitoring sites as noted above. Water quality variables measured included water temperature, dissolved oxygen, pH, specific conductance, phytoplankton chlorophyll-a, ammonia nitrogen, nitrate nitrogen, turbidity, total suspended solids, total phosphorus, orthophosphate phosphorus, periphyton biomass, and chlorophyll-a. Grab sample data from the 2001 monitoring season at Weitchpec, Terwer, and Martin’s Ferry were also included for comparison. Data from 2001 were used for comparison, because they were collected at the same sites and with the same protocol as the 2002 data.

Additional sites within the die-off zone were monitored less frequently. Variables monitored at those sites included water temperature, dissolved oxygen, pH, specific conductance, chlorophyll-a levels, and ammonia nitrogen. Water samples were collected and analyzed for a suite of herbicides and pesticides. These samples were collected on September 26, 2002, in the mouths of two tributaries of the Klamath, including Tectah Creek (Klamath RM 22.1) and Blue Creek (Klamath RM 16.4), in the Klamath River about 0.25 mile downstream of Blue Creek (Klamath RM 16.1), and in the Klamath River at the 101 Bridge (Klamath RM 2.8). On October 8, 2002, additional water samples were collected in the estuary (Klamath RM 1.8), and mainstem Klamath at Terwer (Klamath RM 6.7) and analyzed for the same variables. Analysis of these samples was conducted by the North Coast Laboratories Ltd. in Arcata, California, under contract to the North Coast Regional Water Quality Control Board (NCRWQCB) or the Service.

Grab water samples were collected using standard operating procedures to ensure good data quality (American Public Health Association (APHA) 1998, USGS 1999). In addition, each laboratory sample was subjected to internal laboratory quality control and quality assurance procedures including matrix spikes, blanks, and duplicate analyses. Water sampling, in general, consisted of collecting a sample using a USGS composite collection churn. After water was collected it was stirred carefully to ensure complete mixing. Water samples were then transferred through the drain spigot to individual sample containers. Sample containers were preserved by placing them on ice and for selected parameters preserved with sulfuric acid. Measurements of water temperature, specific conductance, pH, turbidity, and dissolved oxygen were made using either pre-calibrated YSI® and Hydrolab® meters, or, in the case of dissolved oxygen, a Winkler titration procedure. In addition, turbidity was measured using a LaMotte nephelometer.

These water quality samples were analyzed by the North Coast Laboratories Ltd. located in Arcata, California. Water quality data collected by continuous monitors and grab samples were compared to NCRWQCB State standards and/or proposed U.S. Environmental Protection Agency (EPA) Region 10 criteria for salmonids where such criteria exist.

Fish Tissue Samples for Contaminants

A single moribund coho salmon carcass was collected by the USFWS on October 3, 2002, at the mouth of Blue Creek for tissue analysis. The carcass was immediately placed on ice, then frozen and submitted to the CDFG Fish and Wildlife Water Pollution Control Laboratory in Rancho Cordova, California, for pesticide and organic contaminant analysis.

Hazardous Materials Spills and Reports

Data from August and September 2002 State and national pollution reports generated by the State of California Office of Emergency Service and the U.S. Coast Guard National Response Center were reviewed for the presence of any reported spills on the Klamath River.

Permitted Point Source Discharges

The EPA web-based database on National Pollutant Discharge Elimination System permits was reviewed for the presence of any permitted discharges in the vicinity of the fish die-off including Humboldt, Del Norte, and southwestern Siskiyou counties during August and September 2002.

Hazardous Waste Sites

The EPA web-based database Comprehensive Environmental Response, Compensation, and Liability Information System (CERCLIS) was reviewed for the presence of any unpermitted hazardous waste site in the vicinity of the fish die-off including Del Norte, Humboldt and southwestern Siskiyou counties. CERCLIS contains information on hazardous waste sites, potential hazardous waste sites, and remedial activities across the nation, including sites that are on the National Priorities List or being considered for this list.

Comparison of Salmon Run Size with Environmental Factors

We evaluated the relationship of run size with various environmental variables, including mainstem and tributary discharges and local precipitation. This analysis was limited to data from 1978 to 2002. These are years for which data on total instream run size are available from the CDFG (CDFG 2003a).

Salmon Run Size

Historical and recent run size estimates were obtained from the 2003 CDFG electronic database called the megatable (CDFG 2003a). It contains age-specific estimates of run size from 1978 to 2002 (Klamath River Technical Advisory Team) (KRTAT 2003). At the time of this writing the 2002 data are considered preliminary (Sinnen 2003). The Klamath River megatable is used for Klamath Chinook stock management purposes and estimation of escapement (Pierce 1998). It also includes estimates of hatchery and tributary composition, and projections for future years based on estimates derived by the KRTAT (Sinnen 2002). Run size estimates are obtained from hatchery returns, tribal harvest data, carcass surveys, and recreational creel surveys. The estimated number of fish that died during the 2002 die-off was included in the overall run estimate for 2002 (Guillen 2003).

River Discharge

River discharges used in our analysis were obtained from USGS gage stations via their internet web site (USGS 2003). Generally, discharges before 2002 are validated data. Discharges during most of 2002 are considered “provisional” and subject to further revision. Sites used in our analysis included significant gaged discharges including Iron Gate Dam (IGD), Trinity River at Hoopa, Salmon River, Scott River near Fort Jones, and the Shasta River. August and September monthly average discharges from each of these five sites during 1978 through 2002 were used in our evaluation. These historical discharges were used to classify various years of record from 1978 through 2002 based on the overall pattern of significant discharges. Using this approach, the influences of mainstem discharges at IGD and significant tributary sources were included in the classification of various years.

Cluster Analysis

In order to evaluate how 2002 may have been different from other years, we wished to focus our analysis on years that possessed similar hydrology, but did not have a fish die-off. For years with available run size estimates (1978-2002), we applied the statistical technique of cluster analysis to classify years based on the pattern of significant gaged discharges during August and September in the Klamath Basin below Iron Gate Dam. Cluster analysis is a multivariate statistical analysis designed to classify objects based on various common traits. It is widely used to conduct complex pattern recognition and classification (Legendre and Legendre 1998). For example, cluster analysis has been used to develop regional classifications of streamflow drought series (Stahl and Demuth 1999).

In order to examine the similarity of various years based on patterns of basin hydrology, we analyzed August and September discharges from significant tributaries and mainstem flows using a familiar cluster analysis technique with the SPSS ™ version 10.07 statistical package (SPSS 2000). The traits used were August and September monthly average discharges from the Klamath River at Iron Gate Dam, the Trinity River at Hoopa, the Salmon River, the Scott River, and the Shasta River. Monthly average discharge from the Klamath River at Iron Gate Dam during September 2002 was adjusted by recalculating this value using daily average flows from September 1 to 27, 2002. This removes the influence of the pulse flow during September 2002.

Therefore, a total of 10 traits or discharges were used to classify each year. The cluster analysis methods are described in more detail in Appendix A. After conducting the cluster analysis, we compared environmental factors including including run size, precipitation, and individual tributary discharges for the group that included 2002 with the next most similar group based on hydrologic conditions.

Fall-Run Chinook Timing

Data from CDFG’s lower Klamath River recreational creel survey and Trinity River Willow Creek weir counts (CDFG 2003b), and from the Yurok Tribal Fisheries monitoring program (Hillimier 2003) were used to determine the timing of the 2002 fall Chinook run. The Karuk Tribe provided data on the dip net fishery harvest upstream at Ishi Pishi Falls (Klamath RM 67) (Chamberlain 2002). The Karuk data were used to evaluate the response of the fall-run Chinook to the increased river discharge (pulse flow) provided in late September from the upper reservoirs to help stimulate fish movement. Published reports on general overall trends in spawning periodicity for the lower Klamath River were also reviewed.

Fish Passage

We reviewed the literature on minimum passage requirements to determine if there was a potential for reduced fish movement due to physical obstructions or insufficient river depths.

This information was compared to largely anecdotal information provided by various biologists working on the lower river during the die-off. No detailed surveys were conducted to evaluate physical passage conditions at critical locations during the investigation.

RESULTS AND DISCUSSION

In this section, we will describe the mortality event and the environmental conditions that existed at the time of the event. Conditions and factors that might have contributed to the die-off will be assessed individually, followed by evaluation of effects resulting from interactions between these variables.

Mortality Estimates

Details regarding mortality counts and fish species and age composition were reported by Guillen (2003). Initial reports of fish mortality were received on September 19, 2002. According to reports, dead and dying fish were observed on September 18, 2002, by a fisherman in the lower Klamath River. Unconfirmed reports suggested dead fish may have been observed as early as September 16, 2002.

An estimated 34,056 fish died during the incident (Guillen 2003). Of 33,527 anadromous salmonids estimated to have succumbed during this event, about 97 percent (estimated 32,533) were fall-run Chinook salmon, Oncorhynchus tshawytscha, 2 percent (estimated 629) were steelhead, O. mykiss, and 1 percent (estimated 344) were coho salmon, O. kisutch. One coastal cutthroat, O. clarki clarki was found dead during the investigation.

The KRTAT (2003) estimated that dead fall-run Chinook salmon represented about 19 percent of the total (estimated 169,297) in-river Klamath-Trinity River run. The KRTAT estimated that about 7,060 (22 percent) of the dead Chinook were of hatchery origin. A total estimate of 2,921 (9 percent) of the dead Chinook were of Iron Gate (Klamath River) Hatchery origin and an estimated 4,139 (13 percent) of the dead Chinook were of Trinity River Hatchery origin.

Approximately 91 percent of the coho salmon, and 39 percent of the steelhead observed had marks indicating a hatchery origin (Guillen 2003). All hatchery coho were from the Trinity River Hatchery.

Other dead fish observed during the investigation included sculpins, Cottus spp. (87 fish), speckled dace, Rhinichthys osculus (9 fish), Klamath smallscale sucker, Catostomus rimiculus (311 fish), one American shad, Alosa sapidissima, and one green sturgeon, Acipencer medirostris.

In addition to the above counts, we estimated that a total of 120 unidentifiable carcasses were present during the die-off.

Throughout the investigation, live adult and juvenile fish of affected and unaffected species were observed in the River. Some species (e.g., American shad, speckled dace, and green sturgeon) did not appear to experience extensive mortality. Over 99 percent of the dead fish observed were adults or larger species of fish. In addition, dead invertebrates, amphibians, or terrestrial vertebrates were not observed during the incident.

The lack of dead juvenile salmonids may be explained by previous studies that documented low numbers of juveniles in the River during the time of the year when the fish die-off occurred (Wallace 1997; Weskamp et al. 1998). By September, most juvenile salmonids have emigrated to the estuary.

Dead and/or dying fish were observed by survey crews from at least September 20, 2002, through October 3, 2002. Even as late as October 3, 2002, a few recently dead fish were observed. Small numbers of dead fish were also observed by Yurok Tribal biologists after October 1, 2002, for about another week. However, the majority of fish had died by September 27, 2002. It appears that mortality occurred primarily over a 10-day period, with peak mortality occurring between September 18 and September 24, 2002.

Greater than 85 percent of the recently dead fish examined exhibited one or more outward gross signs of disease including white spots, gill necrosis, bacterial growth, sores, bloody vents, and ulcerations.

It should be noted that the conservative fish die-off estimate may have resulted in an underestimate of the actual total 2002 in-river run size (Guillen 2003). We evaluated the potential effect on estimated run size of underestimating the number of fish that died during the 2002 fish die-off using several estimators that were available, and supported by the literature on fish kills and spawner carcass surveys (AFS 1992; Pisano 1994). Using an estimator of an additional 100% of the fish die-off, the total 2002 in-river run becomes 203,353 fish, which is approximately equal to the 2001 run size, and becomes the sixth largest run during the period between 1978 to 2002.

Supplemental Field Observations

Phytoplankton

Very few phytoplankton cells were observed in the 100-ml sample collected near the mouth of Terwer Creek in the Klamath River (Klamath RM 5.3) on September 20, 2002. The lack of high numbers of phytoplankton cells indicate that an algal bloom was not occurring during the fish die-off.

Odors and Discolored Water

Besides decaying fish, unusual odor or discolored water was not observed by any investigator during the entire response between September 20 and October 1, 2002. The lack of any obvious odors or discolored water suggests that visible or odor-producing contaminants such as gasoline, diesel, or other pollutants were absent during the die-off.

Green Sturgeon Tracking Data

Nine green sturgeon were tagged with acoustic and radio transmitters during May through June 2002. In addition, one white sturgeon was tagged. During May through November 2002 these fish were tracked using a series of stationary acoustic receivers and mobile radio receivers. At least 8 of the 10 fish were observed throughout the area affected by the fish die-off during August through September (Table 1). All of these fish survived and eventually moved downstream to the estuary. The lack of any significant green sturgeon mortality suggests that contaminants or any other factors that would cause extensive non-selective mortality across all species were absent in the lower River where green sturgeon were present.

Fish Pathology

Examination of specimens collected during the die-off indicated that all the moribund or recently dead fish were infected by Ich (Icththyophthirius multifiliis) and/or columnaris (Flavobacterium columnare) (Foott 2002a; Veek 2002).

Many pathogens are ubiquitous along the northwestern Pacific coast of the United States in salmon populations. However, they are normally present at low levels and do not usually affect the host to the point of causing disease (Arkoosh 1998). Only when other stressors are present are there increased incidences of disease outbreaks. These stressors can include elevated water temperature, low dissolved oxygen, crowding, high levels of ammonia, and presence of pollutants (Wedemeyer 1974). The susceptibility of anadromous salmonids to these pathogens is also influenced by hydrological regime, behavior, and physiological changes associated with spawning activity.

Ichthyophthirius multifiliis or Ich

Ich is a fresh-water ciliated protozoan. The life cycle includes the following stages: an attached parasitic stage (trophont), a detached reproductive stage (tomont), and a free-swimming infective stage or “tomite” (Figure 3) (Dickerson and Dawe 1995). Up to 2,000 tomites can be produced in a 12-hour period from one tomont at the optimal temperatures (Meyer 1974). The optimal temperature for Ich development is 21.1-23.9 degrees centigrade (°C) (70 - 75 degrees Farenheit (oF)) (Meyer 1974). Within this optimal range, the higher the temperature the faster the parasite replicates (Gratzek 1993).

The parasitic phase of Ich can encyst in the skin or gill. Damage to the skin and gill results in osmoregulatory and respiratory distress (Ewing et al. 1985). As the infection progresses, the capacity of infected fish to absorb oxygen and excrete ammonia is severely reduced and their blood ammonia levels rise because of gill impairment. Mortality from Ich may be caused by the parasite when the gills are too damaged to function. Breaching the protective barrier of the skin by Ich may also allow opportunistic bacteria or fungi access to underlying tissues resulting in death from secondary infections (Post 1987).

Bodensteiner et al. (2000) described a study in which increasing the discharge in catfish raceways reduced mortality due to Ich. Velocities greater than 75 cm/min (0.041 ft/s), resulting in a turnover rate of greater than 1.9 volumes per hour, prevented any disease outbreaks. The increased discharges and velocities reduced the probability of the tomites finding a host, since the parasite was swept away downstream. Ich outbreaks in epidemic proportions are rare in flowing rivers and streams (Allison and Kelly 1963).

Outbreaks of Ich occur when conditions are favorable for rapid multiplication of the parasite.

This includes a suitable environment and susceptible fish. There may be a requirement for some minimum number of fishes before an epizootic occurs (McCallum 1985). Ich epizootics occur when fishes are stressed, densities are high, and the water temperature is relatively elevated (Dickerson and Dawe 1995). A wide variety of factors can induce stress in fishes including crowding, high temperature, low dissolved oxygen, changes in conductivity or salinity, chemical pollutants, and spawning activities. Frequently, spawning adults of only one species are affected (Pickering and Christie 1980; Wurtsbaugh and Tapia 1988). Outbreaks are most common during warmer months and also during periods when fish are spawning. Increased stress associated with spawning migrations of salmonids has been shown to increase the prevalence of disease outbreaks (Fagerlund et al. 1995).

Ich is a freshwater parasite. Thus, in the Klamath River die-off, Ich would have been transmitted to returning adults from resident freshwater species such as suckers or sculpin, or possibly from resident or previously-infected emigrating salmonids.

Flavobacterium columnare or Columnaris

Columnaris is the common name for the bacterial pathogen Flavobacterium columnare, formerly Flexibacter columnare. The earliest sign of columnaris disease in fish is a thickening of the mucus at various spots on the head, opercula and fins. Well-developed columnaris disease on the skin usually has tiny bloody spots or petechia within the lesions. Fringes of gill filaments are lost to advancing necrosis and sloughing of gill tissue. Eventually respiratory and osmoregulatory function is lost at the gill surface (Post 1987).

Resident non-salmonids, such as suckers, have been reported to be carriers of columnaris and a source of infection for migrating adult salmon in the Columbia River (Becker and Fujihara 1978).

Columnaris is usually pathogenic at temperatures higher than 15 °C (Noga 2000). Columnaris outbreaks are common in adult salmon populations held at hatcheries in warm waters (15 to 18 °C) (Foott 2002b). In laboratory studies, host immune and nutritional status have been demonstrated to affect mortality from columnaris infections. Further, mortality rates in groups of juvenile Chinook salmon challenged with columnaris were related to the density of fish (Fujihara et al. 1971).

Normal, healthy fish are usually resistant to columnaris (Shotts and Starliper 1999). However, it can develop as a secondary infection due to environmental stress or trauma. Stress can include crowded conditions, handling stress, low dissolved oxygen, elevated temperatures, and high organic loads (Thune 1993). Columnaris often appears in association with one or more other pathogens and is secondary to the primary disease organism (Plumb 1999), which can include ectoparasites such as Ich.

Extensive mortality of summer-run Chinook salmon, coho salmon, suckers, and sockeye salmon has occurred in the Columbia River as fish migrated into warm upstream waters (Becker and Fujihara 1978). Becker and Fujihara (1978) found that highest incidents of infection and mortality occurred during periods of high spring water temperatures, hot summers, and associated low river flows. Epizootics involving columnaris and high temperatures also have been documented in the Rogue River, Oregon (Oregon Department of Fish and Wildlife, (ODFW)1992). There, they found that the incidence of mortality and columnaris infection were more correlated with high water temperature than fish density. They also found that river flows were highly correlated with water temperature and therefore affected the probability of contracting this pathogen. Increased flows were used to reduce water temperatures, fish stress, and resulting infections due to columnaris. ODFW (1992) also concluded that extensive prespawning mortality of Pacific salmon has been observed in other streams and seems to occur primarily when streams are unusually warm during periods of low flow.

Relationship of Pathogens and Spawning Activity

In Pacific salmon, the spawning migration from the sea to the natal stream induces stress from changes in osmoregulation, increased activity, starvation, and sexual maturation. Cumulative stress debilitates the fish sufficiently that they become highly susceptible to a variety of diseases and die soon after spawning (Smith 1993).