# Endangered and Threatened Wildlife and Plants; Notice of Remanded Determination of Status for the Sacramento splittail (Pogonichthys macrolepidotus)

> Briefs, arguments, decisions, and more.

URL: https://www.frixlaw.com/law-library/documents/fr%3A03-23919

## Record

- **Collection:** Federal Register
- **Document type:** Rule
- **Published:** September 22, 2003
- **Citation:** 68 FR 55140

## Text

DEPARTMENT OF THE INTERIOR
Fish and Wildlife Service
50 CFR Part 17
[RIN 1018-AH73]

Endangered and Threatened Wildlife and Plants; Notice of Remanded Determination of Status for the Sacramento splittail (
Pogonichthys macrolepidotus
)

AGENCY:

Fish and Wildlife Service, Interior.

ACTION:

Final rule; revised determination.

SUMMARY:

On January 6, 1994, we, the U.S. Fish and Wildlife Service (Service) proposed to list the Sacramento splittail (
Pogonichthys macrolepidotus
), a fish species native to central California, as a threatened species under the Endangered Species Act of 1973, as amended (Act). We published a final rule to list the species as threatened on February 8, 1999. Our final decision to list the Sacramento splittail was subsequently challenged in the cases
San Luis & Delta-Mendota Water Authority
v.
Anne Badgley, et al.
and
State Water Contractors, et al.
v.
Michael Spear, et al.
On June 23, 2000, the Federal Eastern District Court of California found our final rule to be unlawful and on September 22, 2000, remanded the determination back to us for a re-evaluation of our final decision. However, because the District Court did not vacate our previous final decision, the decision remained in place until we issued a new determination. After a thorough review and consideration of all the best scientific and commercial information available, we are removing the Sacramento splittail from the list of threatened species. In accordance with the Administrative Procedure Act, the Service has determined that this rule relieves an existing restriction, and good cause exists to make the effective date of this rule immediate.

EFFECTIVE DATE:

In compliance with the Federal Eastern District Court of California order, this rule is effective September 22, 2003.

ADDRESSES:

Comments and materials received, as well as supporting documentation used in the preparation of this final decision, are available for public inspection, by appointment, during normal business hours at the Sacramento Fish and Wildlife Office, U.S. Fish and Wildlife Service, 2800 Cottage Way, Suite W-2605, Sacramento, CA 95825.

FOR FURTHER INFORMATION CONTACT:

Wayne White (
see

ADDRESSES
), (telephone: 916/414-6600; facsimile: 916/414-6713). Information is available in alternate formats upon request.

SUPPLEMENTARY INFORMATION:

Background

The Sacramento splittail (hereafter referred to as splittail) is a fish species native to central California and represents the only extant species in its genus in North America. We have previously discussed the taxonomic history of the splittail along with the physical description of the taxon in our final listing rule (64 FR 5963). Please refer to that document for a detailed discussion of these subjects. It is our intent, in this document, to reiterate and discuss only those topics directly relevant to this decision.

To assist the reader in understanding terminology used in this determination, we have provided below several terms with their corresponding definitions as they are used in this document. As used in this determination, the term “Delta” refers to all tidal waters contained within the legal definition of the San Francisco Bay-Sacramento-San Joaquin River Delta, as delineated by section 12220 of the State of California's Water Code. Generally, the Delta is contained within a triangular area that extends south from the City of Sacramento to the confluence of the Stanislaus and San Joaquin Rivers at the southeast corner and Chipps Island in Suisun Bay at the southwest corner. The term “Estuary,” as used in this determination, refers to tidal waters contained in the Sacramento and San Joaquin Rivers, the Delta, and San Pablo and San Francisco bays. “Export facilities,” as used in this determination, refers to the Bureau of Reclamation (USBR) Central Valley Project (CVP) and the California Department of Water Resources (CDWR) State Water Project (SWP) water export facilities in the South Delta.

Splittail are native to California's Central Valley. Historically, splittail were found as far north as Redding on the Sacramento River (Rutter 1908). Splittail were also found in the tributaries of the Sacramento River as far as the current Oroville Dam site on the Feather River and Folsom Dam site on the American River (Rutter 1908). Along the San Joaquin River, historic distribution is unclear. Girard (1854) reported two
Pogonichthys
species in the San Joaquin River. These reports do not make a distinction between which of the two species was found at particular locations on the San Joaquin River. In the southern Central Valley, Tulare Lake was likely to have supported many native fish species, including splittail (Moyle 1976) but has since been drained and reclaimed. Splittail were present within Buena Vista and Kern Lakes (Moyle 2002), both of which are reclaimed.

Some researchers (Sommer
et al.
(1997)) indicate that splittail still occur, at least during optimal conditions, through as much as 78 percent of their former range in terms of river reaches. However, others (Moyle and Yoshiyama 1992) believe the species appears to be restricted to a small portion of its former range, with dams and diversions preventing access to upstream habitat in large rivers and streams beyond the valley floor (Moyle and Yoshiyama 1992). The State of California indicates that splittail still occur in a large portion of its range (80% in the Sacramento, and 70% in the San Joaquin). There appears to be consensus that at least 20% and possibly more of the species range has been reduced. Baxter (2001b) found that the range of the splittail extends away from the Delta, though detections on the periphery of its range appear to be part of a single, mobile, Sacramento and San Joaquin River/Bay-Delta population that includes fish from the Napa and Petaluma River systems. Their distribution in the Estuary suggests that brackish water may characterize optimal rearing habitat for fish greater than 75 millimeters (mm) (3.0 inches (in)) standard length (SL) (Moyle
et al.
2001). Suisun Marsh includes the largest areal extent of shallow water habitat available to the splittail and likely has the greatest concentrations of the species.

Splittail are relatively long-lived and larger fish may be 8 to 10 years old (Moyle 2002). Splittail reach about 110 mm (4.3 in) SL in their first year, 170 mm (6.6 in) SL in their second year, and 215 mm (8.4 in) SL in their third year (Moyle 2002). Male and female splittail may mature by the end of their second year (Daniels and Moyle 1983), but some males mature in their first year and some females do not mature until their third year (Caywood 1974).

The largest females can produce over 250,000 eggs per year (Daniels and Moyle 1983). Other and more current estimates of splittail fecundity have shown high variability and occasionally, lower numbers. Caywood (1974) found a mean of 165 eggs per mm (6.5 in) of SL of fish sampled and reported a maximum of 100,800 eggs in one female. Daniels and Moyle (1983) observed approximately 17,500 to 266,000 eggs per female splittail. Feyrer and Baxter (1998) found a mean of 261 eggs per mm (10.2 in) of SL and estimated maximum fecundity at 150,000 eggs. Bailey
et al.
(1999) examined fish held for a considerable time in captivity and found that

fecundity ranged from 24,753 to 72,314 eggs per female, which agrees with Caywood's (1974) observations.

Although primarily a freshwater species, splittail can tolerate salinities as high as 10 to 18 parts per thousand (ppt) (Moyle 1976; Moyle and Yoshiyama 1992). Salinity tolerance in splittail increases in proportion to length; adults can tolerate salinities as high as 29 ppt for short periods (Young and Cech 1996). Splittail populations fluctuate annually, depending on spawning success, which is well correlated with freshwater outflow and the availability of shallow water habitat with submerged vegetation (Daniels and Moyle 1983; Sommer
et al.
1997). Fish typically reach sexual maturity by the end of their second year. The onset of spawning is associated with rising water levels, increasing water temperatures, and increasing day length. Peak spawning occurs from February through May, although records of spawning exist for late January to early July (Wang 1986). In some years, most spawning may take place within a limited period of time. For instance, in 1995, a year of high spawning activity, most splittail spawned over a short period in April, even though larval splittail were captured from February through early July (Moyle
et al.
2001). Within each spawning season older fish reproduce first, followed by younger individuals (Caywood 1974).

Splittail spawning occurs over flooded vegetation in tidal freshwater and brackish water habitats of estuarine marshes and sloughs and slow-moving, shallow reaches of large rivers. Observations of splittail spawning have indicated spawning at depths of less than 1.5 meters (m) (4.9 feet (ft)) in the Cosumnes River floodplain (Moyle
et al.
2001), and at depths of less than 2 m (6.6 ft) in Sutter Bypass (Moyle
et al.
2001). Sommer and Harrell (1999) postulated that individual splittail may not spawn in the year following a successful effort.

Splittail larvae remain in shallow, weedy areas close to spawning sites for 10 to 14 days and move into deeper water as they mature and swimming ability increases (Wang 1986; Sommer
et al.
1997). Bailey (1994) has documented that splittail eggs hatch in 3 to 5 days at 18.5 degrees centigrade (°C), (65.3 degrees Fahrenheit (°F)). Bailey (1994) also found that at 5 to 7 days after hatching, the yolk sac is absorbed and the diet begins to include small rotifers. Moyle
et al.
(2001) states that splittail of 20 to 25 mm (0.8 to 1.0 in) total length (TL) “* * * are essentially small juveniles, capable of fairly active swimming” and that 4 to 5 weeks post-hatch are required to reach this size class.

It is speculated that Suisun Marsh is the likely late stage rearing area for juvenile splittail hatched and reared in the extensive spawning habitat found within the Yolo Bypass, as a hydrologic connection apparently exists between these waters (N. Monsen, unpubl. data referenced in Moyle
et al.
2001). Splittail use of Suisun Marsh varies with outflow (Baxter 1999a).

Splittail are benthic foragers. In Suisun Marsh, adults feed primarily on opossum shrimp (
Neomysis mercedis,
and presumably, non-native shrimp species of the genus
Acanthomysis
as well), benthic amphipods (
Corophium
spp.), and other small crustaceans, although detrital material makes up a large percentage of their stomach contents (Daniels and Moyle 1983). In the Delta, clams, crustaceans, insect larvae, and other invertebrates also are found in the adult diet. More recently, research has indicated a shift in adult splittail diet towards the non-native Asiatic clam (
Potamocorbula amurensis
) in Suisun Marsh.

Historically,
Eurytemora affinis,
the native euryhaline copepod, has been the most important food for larval fishes in the Estuary. Three non-native species of euryhaline copepods (
Sinocalanus doerrii, Pseudodiaptomus forbesi,
and
Pseudodiaptomus marinus
) became established in the Delta between 1978 and 1987 (Carlton
et al.
1990), while native
E. affinis
populations have declined since 1980. It is not known if the non-native species have displaced
E. affinis
or whether changes in the estuarine ecosystem now favor
S. doerrii
and the two
Pseudodiaptomus
species. Meng and Orsi (1991) reported that
S. doerrii
is more difficult for larval striped bass to catch than native copepods because it is fast swimming and has an effective escape response. It is not known if this difference in copepod swimming and escape behavior has affected the feeding success of young splittail. Zeug
et al.
(2002) and Hieb (2002) reported a high abundance of an introduced, predatory Palaemonid shrimp (
Exopalaemon modestus
) in the Yolo Bypass and Delta. It is not known what effect(s) this invasive species will have on the trophic (food) pyramid of the estuary, though Moyle (2002b) speculates it is likely to prey on mysid shrimp and thus, may compete with splittail for food. Juvenile feed mainly on plankton composed of small animals (zooplankton), and then small crustaceans and insect larvae as body size increases.

Predators of splittail include striped bass (
Morone saxatilis
), largemouth bass (
Micropterus salmoides
) and other centrarchids, and other native and non-native piscivores (Moyle 1976, Moyle 2002a). Introduced, non-native benthic foragers such as shokihaze goby (
Tridentiger barbatus
), chameleon goby (
T. trigonocephalus
), and yellowfin goby (
Acanthogobius flavimanus
), may feed on splittail eggs. Introduced planktivorous, threadfin shad (
Dorosoma petenense
) and inland silverside (
Menidia beryllina
), compete directly with larval and juvenile splittail for food. Other non-native cyprinids, such as golden shiner (
Notemigonus crysoleucas
), red shiner (
Notropis lutrensis
), and fathead minnow (
Pimephales promelas
) are also likely to compete with splittail. In recent years, splittail have been found most often in slow moving sections of rivers, sloughs, and in dead end sloughs (Moyle 1976, Daniels and Moyle 1983). Reports from the 1950's, however, mention Sacramento River spawning migrations and catches of splittail during fast tides in Suisun Bay (Caywood 1974). Current accounts place splittail as far upstream as the Red Bluff Diversion Dam on the Sacramento River (Baxter 1999a). Splittail have been recorded in recent times from within Salt Slough and at the Merced River confluence on the San Joaquin River, and within the Napa and Petaluma Rivers (Baxter 1999a, 1999b; USACE 2002a, 2002b).

Splittail are frequently found in areas subject to flooding because they require flooded vegetation for spawning and rearing. Historically, the major flood basins (
e.g.
, Colusa, Sutter, American, and Yolo basins; Tulare, Buena Vista, and Kern lakes) distributed throughout the Sacramento and San Joaquin valleys provided spawning and rearing habitat. These flood basins have all been reclaimed or modified for flood control purposes (
i.e.
as bypasses), and much of the floodplain area adjacent to the rivers is now inaccessible behind levees. The Yolo Bypass may approximate some of the Yolo Basin's former role, and the Butte Creek, Butte Sink, Sutter Bypass system remains somewhat intact. Meng and Moyle (1995) reported that the core distribution of splittail extends from Suisun Bay and Marsh through the western Delta.

The Yolo and Sutter bypasses and the Cosumnes River floodplain serve as important splittail spawning and early rearing habitat (Sommer
et al.
1997), as they approximate the large, open, shallow water areas which have been extensively reduced. The Yolo and Sutter bypasses provide good habitat for fish, particularly splittail, when flooded for several weeks in March and April. To provide the best spawning

conditions for splittail, water must remain on the bypasses until fish have completed spawning, and larvae are able to swim out on their own, during the draining process. The Cosumnes River also possesses natural and restored floodplain features. This river is unique in that it is not dammed and the hydrograph is relatively natural. The contributions made by this habitat are somewhat limited by the fact that the Cosumnes River watershed is lower in elevation than most adjacent rivers. It is therefore somewhat less dominated by the extended spring peak flow characteristic of a higher altitude watershed with greater snowmelt potential.

In summary, the current distribution of splittail habitat is certainly reduced at least 20% and may be much more reduced in extent from that which may have historically been present. Clearly, perhaps the largest portion of the splittail's habitat is contained in the natural and newly restored floodplains of the Cosumnes River, managed floodplains such as the Yolo and Sutter bypasses, disjunct segments adjacent to the Sacramento and San Joaquin rivers and in lower reaches of their respective tributaries.

In years where the Yolo and Sutter bypasses are not sufficiently inundated, splittail spawning is confined primarily to the natural and newly restored floodplains of the Cosumnes River and the margins of rivers and other floodplain features that are inundated at lower river stages. These areas likely represent only a fraction of the area which was historically subject to inundation; levees preclude access to reclaimed floodplains and basins. There are indications, based on presence of larvae and juveniles, that spawning in the Sacramento River occurs relatively far upstream at Colusa (Baxter 1999a; 1999b). Splittail appear to utilize the San Joaquin River in wet years when appreciable runoff exceeds the capacity for storage and diversion of runoff. The Tuolumne, Cosumnes, Feather, American, Napa, and Petaluma rivers, and numerous other smaller waters support splittail spawning activity. Early indications are that the Napa River may contain a robust subpopulation of splittail (USACE 2002a, 2002b).

Abundance

Seven sampling programs capture splittail frequently enough to allow the calculation of useful abundance indices. These programs are: (1) CDFG's Fall Midwater Trawl (Fall MWT); (2) CDFG's San Francisco Bay Midwater Trawl (Bay Study MW); (3) CDFG's San Francisco Bay Otter Trawl (Bay Study OT); (4) University of California (UC) Davis's Suisun Marsh Otter Trawl (Suisun Marsh OT); (5) Service's Chipps Island Trawl survey (Chipps Is. Trawl); (6) fish salvage operations (which repatriate fish taken from water intake screens) at the CVP Tracy Fish Collection Facility (CVP); and (7) fish salvage at the SWP Skinner Delta Fish Protective Facility in the south Delta (SWP).

Four other sampling programs provide additional splittail information but the data are insufficient to support useful indices. These are: (1) Service's Delta Beach Seine Survey; (2) CDFG's Summer Townet Survey; (3) U.S. Army Corps of Engineers' (USACE) Napa River Survey; and (4) CDFG's Creel Census.

Surveys Employed in Abundance Analyses

The data available even today on splittail abundance are not optimal. There are a number of survey programs which generate data, each of which have more or less limiting factors. This has made analysis of the status of the species based on this survey data problematic. Descriptions of all fisheries sampling programs that routinely detect splittail follow, and are differentiated into two categories: those that were used in the calculation of abundance indices and those that were not.

Fall Midwater Trawl Survey

The Fall MWT was initiated by CDFG in 1967 to sample striped bass, a non-native sport fish. In addition to striped bass, CDFG has maintained records of other fish species captured in the samples in most years. This monitoring program currently samples 100 sites from San Pablo Bay in the west to Rio Vista on the lower Sacramento River and to Stockton on the San Joaquin River. Data are collected from September through December using a midwater trawl with a 3.7 square m (39.8 square ft) wide mouth. Unlike the summer townet survey, the Fall MWT survey catches all splittail size classes, although larger fish are more likely to evade capture. Catches of splittail are generally low in number because splittail generally reside and feed on the channel bottom. Furthermore, splittail apparently use shallow (less than 6 m (19.7 ft)) and near-shore waters to a higher degree than open channels. The Fall MWT does not sample edge waters, and the proportion of samples in shallow water stations varies by region: 20 of 35 stations in San Pablo Bay; 1 of 18 in Carquinez Strait; 8 of 25 in Suisun Bay/Marsh; and 1 of 38 in the Delta. A monthly abundance index for splittail captured by the Fall MWT is calculated by grouping the samples by area (17 areas) and then calculating an area weighted average catch from each area; the index is the sum of these area weighted mean catches. The annual Fall MWT Index is the sum of the four monthly indices. Splittail lengths were not recorded until 1975, so for data collected prior to 1975, Young Of Year (YOY) (age 1) fish could not be differentiated from other age classes. Fall MWT data from 1967 through 2002 was used in our abundance analysis.

San Francisco Bay Studies

The San Francisco Bay Studies sample waters west of the Delta seaward to south San Francisco Bay using both a midwater trawl (Bay Study MWT) and an otter trawl (Bay Study OT) (Baxter 1999a). These programs capture relatively few splittail, but are still considered important because they involve two types of sampling equipment and frequent sampling (Baxter 1999a). Much of the sampling takes place in San Francisco Bay in deep water channels that are not characteristic splittail habitat. Monthly indices are calculated as the sum of regional volume-weighted average catch per 10,000 cubic meters (m
3
) (353,147 cubic feet (cf)) for the Bay Study MWT and the sum of regional area-weighted average catch per 10,000 m
3
(353,147 cf) for the Bay Study OT (Sommer
et al.
1997). During the 1997 index period, the Bay Study MWT collected only one YOY, and the Bay Study OT collected none at index stations. The tremendous variability in this survey's catch is likely due to the rare or limited occurrence of individuals splittails at the periphery of its range, which would result in limited detectability during sampling. Splittail can be expected to be captured in San Francisco and San Pablo bays only during time of infrequent, high outflow, when captures appear to increase for all net-based gear types. San Francisco Bay Studies data from 1980 through 2002 was used in our abundance analysis.

Suisun Marsh Otter Trawl

The Suisun Marsh OT surveys began in 1979 and are conducted by the University of California (UC) Davis as part of a long-term study of the ecology of the entire fish community of the marsh. Data from the 1979 survey have been excluded from our abundance analysis as greater sampling effort was employed in 1979 than in all subsequent years (Dr. Peter Moyle, pers. comm.). The survey is funded by California Department of Water Resources (CDWR) in part to determine if management actions in Suisun Marsh are affecting fish communities. The program samples 21 sites monthly in

nine sloughs with an otter trawl that drags along the bottom and samples much of the water column in the shallow sloughs. In small sloughs, the trawl samples much of the cross sectional area; in large sloughs, the sampling fraction is smaller. A monthly abundance index is calculated as mean catch per trawl. The annual abundance index is calculated as the mean of the monthly index values (Sommer
et al.
1997). While the splittail catches are dominated by YOY, the sampling also consistently catches larger fish. In this regard, the Suisun Marsh OT sampling of splittail is perhaps the most thorough of the various sampling programs. Splittail collection in the Marsh is enhanced by reduced gear avoidance in narrow, relatively shallow sloughs sampled as part of the monthly survey. In such conditions, the net samples a larger proportion of the channel cross sectional area than in any other survey. Larger sizes of splittail, however, apparently become progressively less vulnerable to the trawls, a limitation shared by all trawl-based surveys. Spawning occurs only sporadically in the marsh, and in most years YOY recruit from upstream in the Sacramento River, including the bypasses (Sommer
et al.
1997). Recent modeling studies indicate that the Yolo Bypass, a major spawning and nursery area, may be hydrologically connected to Suisun Marsh (N. Monsen, Stanford University, unpubl. data) so juvenile trends in the marsh are likely to be heavily influenced by upstream production in the Yolo Bypass during those years when inundated for a sufficient period of time. Suisun Marsh Otter Trawl data from 1980 through 2001 was used in our abundance analysis.

Chipps Island Survey

The U.S. Fish and Wildlife Service conducts a sampling program for juvenile salmon in the deep water channel near Chipps Island at the western terminus of the Delta. A midwater trawl is pulled at the surface in ten 20 minute hauls per day during May and June (Sommer
et al.
1997). Data are compiled to produce an index based on the catch per hour of trawling for the months of May and June combined (Sommer
et al.
1997). The program was initiated in 1975, but data before 1979 must be viewed with some caution as many splittail were not measured (Baxter 1999a); as only data related to the number of splittail caught were recorded. Length data from 1987 through 1993 was recorded such that determinations of age from the data cannot be done, and is therefore inadequate to calculate age-specific abundance indices. The Age 0 index reached minor peaks in 1982 and 1986, declined to low levels during the 1987-1992 drought (based on total splittail catch), then increased sharply to a record level in 1995; minor peaks occurred in 1998 and 2000, and remaining data tracked water year variability. For Age 1 splittail, the Chipps Island index for the period 1976 to 2001 shows high variability.

The Chipps Island trawl seems to sample splittail best in high outflow years when all age groups are more vulnerable to trawls due to increased turbidity, as is likely true for all gear types and surveys. It is, however, difficult to discern actual abundance from year biases, and turbidity can be high at Chipps Island regardless of outflow. Regardless, because the trawl captures fish only in the top couple of meters (or yards) of water in open channels, relatively low numbers of the benthic-foraging splittail are caught. The indices are probably less precise at low population levels due to the infrequent captures of splittail, a characteristic shared by all surveys. The Chipps Island Survey data from 1976 through 2002 was used in our abundance analysis.

Central Valley and State Water Project (CVP and SWP) Fish Salvage

The CVP and SWP operate fish screening facilities to divert fish away from the pump intakes into holding facilities where they are counted, measured, and released. Data collection takes place at two hour intervals when the pumps are operating. Consequently, the fish salvage operations provide the highest number of splittail caught per survey, but the number of data points (annual indices) is comparable to the other surveys. All splittail age groups are collected, the surveys do not suffer from gear avoidance by fish, and sampling locations do not vary over time. Reliable CVP data and SWP data both start in 1979. The salvage abundance index is calculated based on the total number of fish salvaged divided by the volume of water pumped (Sommer
et al.
1997). However, the pumps are not operated as sampling programs
per se
so the amount of “sampling” is related to the amount of water exported, which in turn is related to the amount of water available, water demand, and, in recent years, changes in pump operations to protect migratory salmon, splittail, and delta smelt (
Hypomesus transpacificus
) and to maintain appropriate salinities in Suisun Bay and Marsh. Also, the Salvage index does not address catch per volume per unit time. Lacking a time factor, the Salvage index may not adequately describe the differential variability in catch that may occur as approach velocities at Clifton Court Forebay (SWP) or Old River (CVP) change.

Unlike the CVP and SWP salvage, several surveys do not account for the volume per unit time sampled. Trawl data, presented as fish captured per volume of water sampled, do not describe the trawl speed, or the perceived trawl approach speed when pulled against a current. Seine indices are expressed as catch of fish per haul and do not include factors for catch per unit volume and/or per unit time. Seines are employed at sites with low water velocities, but variation in velocity within and between sampling locations likely exists. Trawls and seines may be more effective when employed through higher velocity waters; splittail may be more vulnerable to capture when already navigating swifter currents. Trawls, seines, and pumps therefore share a common difficulty in expressing catch per unit volume per unit time. Each of these techniques may also differentially detect splittail under turbid conditions. The pumps differ from trawling and seining, however, in that the pumps may differentially entrain (collect) weak swimming juvenile and fatigued post-spawn adult splittail as velocities towards the facilities vary. Regardless of boat or current speed, or turbidity, trawls and seines do not draw fish towards them, whereas the pumps may. The SWP catch also does not account for the predation that occurs in the Clifton Court Forebay, nor the latent mortality that may occur when salvaged fish are released.

Comparisons between CVP and SWP salvage and other sampling operations have to be made with caution. Nevertheless, the general patterns are similar to other studies, with diminished catches of both adults and juveniles during periods of drought and large catches of juveniles following wet winters. The CVP and SWP fish salvage data from 1979 through 2002 was used in our abundance analysis.

Surveys Not Employed in Abundance Analyses

US Fish and Wildlife Service Beach Seine Survey

The survey provides the broadest geographical coverage of all of the sampling programs but is focused on outmigrating juvenile salmonids. The beach seine primarily captures YOY splittail but any fish less than 25 mm (1 in) long are not identified. The limited data show low catches of splittail during

dry years and higher catches during wet years, reinforcing the concept of a strong outflow-production relationship. This general relationship may, however, be due to other factors. For example, turbidity may be higher in high outflow years, thus rendering fish more vulnerable to capture.

Summer Townet Survey

The CDFG summer townet survey began in 1959 to provide an index of striped bass abundance. It samples YOY fish twice monthly at 30 sites using oblique tows in mid-channel. Starting and ending dates vary from year to year. Sample sites are located throughout the Delta, Suisun Bay, and San Pablo Bay. Data for species other than striped bass were not regularly recorded until after 1962, but were also not recorded in 1966, 1967, and 1968 (Sommer
et al.
1997). The survey catches only low numbers of YOY splittail, presumably because it focuses on pelagic (open water) habitats while splittail are benthic in orientation. Not surprisingly, splittail catch varies widely and the index reflects only gross changes in YOY splittail abundance. The index peaked in 1982, was low during the 1987 to 1992 drought years, and abruptly rebounded in 1995 and 1998 (Baxter 1999a, 1999b).

Napa River Survey

This survey exists in association with a flood control and ecosystem restoration project in the Napa River. It is performed by consultants under contract to USACE, and involves a range of sampling techniques including beach seine, purse seine, otter trawl, fyke nets, and a 20 mm (0.8 in) size class surveys. The Napa River Survey began sampling in March 2001 and has detected splittail (USACE 2002a, 2002b) but the data are too recent and of too short a term (two years, including 2002 unpublished data) to be useful for an abundance index. The survey is scheduled to be completed in 2007 or 2008, after 7 years of data collection. Additionally, the Napa River is less well understood in terms of relationships between outflow, splittail habitat, and splittail production, than are the Central Valley rivers and the Delta. As such, the variables employed in our current analysis of abundance and trend (
see
Abundance section, below) cannot be applied to this distinct river system at this time.

California Department of Fish and Game Creel Census

CDFG collects creel census data in association with the Sacramento River System Angler Survey. This survey was initially conducted from August, 1989, to December, 1994, and was resumed in 1999 and 2000. Adult splittail catch data were only recorded during 1991 through 1994, and in 1999 and 2000. This survey collected angler count, fishing effort and fish catch information on the Sacramento River from Redding to Carquinez Bridge year round with the same effort, 4 week days and 4 weekend days per month per section, so changes in catch can reflect fish presence related to angler effort.

To reflect only the presence of migrating fish, Baxter (2001b) analyzed only catch data from Garcia Bend (RKM 80 (RM 50)) and upstream. Creel census data from 1991 through 1994 indicated a total annual catch of 114, 266, 498, and 110 splittail, respectively. The 1999 and 2000 censuses yielded an annual catch of 103 and 232 splittail, respectively. These catches represent 96 days of survey effort each year and are useful primarily to help establish the periods in which adult splittail migrate upstream. No abundance indices were calculated by any agency, organization, or individual from these data, as they fail to meet the criteria established by Meng and Moyle (1995) and are generally considered inadequate to the task of quantifying splittail abundance.

Survey Summary

All fish sampling methods may inherently suffer from a selection bias. This bias results from the particular method and must be considered when interpreting results. Because none of the surveys were designed specifically to monitor splittail populations, the survey equipment, survey locations, and sampling frequency must all be taken into consideration when interpreting the data. All the survey methodologies appear to sample young of the year (YOY) most effectively. As a result conclusions regarding YOY abundance appear to be the most accurate and reliable. Combined information from all survey efforts suggest that some successful reproduction occurs every year, but large numbers of young are produced only during years of relatively high outflow (wet years). This suggests that the majority of adult fish in the population result from spawning in wet years and lowest numbers are produced during drought years. The distribution and timing of YOY in the surveys also indicates that most spawning takes place in the bypasses, rivers or upper Delta, although some sporadic spawning also takes place in Suisun Marsh. It must be recognized, however, that YOY abundance may not be an entirely accurate indicator of adult abundance because there exists no observed stock-recruitment relationship (relationship between the number of adult fish and the number of offspring typically expected to join the adult population) in splittail (Sommer
et al.
1997; Moyle 2002). Consequently, YOY abundance may not describe the current of future population sizes or trends.

Abundance Trend Analyses

We initially evaluated and analyzed the aforementioned data series using a method published by Meng and Moyle (1995) in the Transactions of the American Fisheries Society. This method was used during the initial status review for the splittail and was again employed during the development of the proposed rule to list the splittail (59 FR 862). This same method was replicated during the development of the final listing rule published on February 8, 1999, (64 FR 5963) using abundance data provided and updated by CDFG, CDWR, and UC Davis. The Meng and Moyle (1995) methodology (
see
66 FR 2828 for complete description of methods) has been superceded by more current models employed by CDFG, and was not used to help us make this final determination. Further, this removal does not discuss the more recently available analytical methods such as permutation-based exact calculations of p-values for stratified (as opposed to unstratified) Mann-Whitney U-tests, as appeared in the August 17, 2001, notice (66 FR 43145) where we presented an updated statistical analysis of abundance data for the Sacramento splittail and requested comments on it. While these stratified Mann-Whitney U-tests represented an improvement on what essentially remained a Meng and Moyle (1995) statistical approach, and presented a major alternative to the categorical (
i.e.
, “before” and “after”) approaches of both Meng and Moyle (1995) and Sommer
et al.
(1997), substantive scientific and statistical issues raised during the August 17, 2001, (66 FR 43145) public comment period resulted in our using an alternative statistical analysis to help us make this final determination. The following details the history and findings of the current analysis.

In an August 17, 2001, notice (66 FR 43145) we presented an updated statistical analysis of abundance data for the Sacramento splittail and invited public comments on the analysis and data, in specific technical review of the information. We concurrently sought peer review on the statistical analysis from five subject-area experts affiliated with a total of five agencies and organizations. Requests for peer review

were sent to: (1) Dr. Peter B. Moyle of UC Davis, Davis, California; (2) Dr. Charles H. Hanson of Hanson Environmental, Inc., Walnut Creek, California; (3) Randall D. Baxter of CDFG, Central Valley/Bay-Delta Branch, Stockton, California; (4) Michael Chotkowski of the USBR, Mid-Pacific Region, Sacramento, California; and (5) Ted R. Sommer of CDWR, Environmental Services Office, Sacramento, California.

Following careful consideration of comments received from numerous respondents to the August 17, 2001, notice, including those provided through the peer review process, we concluded that the abundance indices and Multiple Linear Regression (MLR) model jointly developed and submitted by CDFG (2001) and USBR (2001), hereafter referred to as the CDFG/USBR MLR Model, provided the best scientific data (method) available, for statistically evaluating temporal trends of splittail abundance information. The CDFG/USGS MLR Model thus superceded the permutation-based exact calculations of p-values for stratified (as opposed to unstratified) Mann-Whitney U-tests.

On March 21, 2002, (67 FR 13095), we reopened the public comment period (67 FR 13095;67 FR 15337) to solicit comments on the CDFG/USBR MLR Model. We again sought peer review on the statistical analysis from the five individuals identified above. We have retained the CDFG/USBR MLR Model, albeit in a slightly modified form, after consideration of all public comments received, inclusive of this and preceding comment periods.

The CDFG/USBR MLR Model includes HYDROLOGY and TIME (year) as independent variables and ABUNDANCE INDICES as the dependent variable. We consider this statistical approach superior to the previous practice of using Mann-Whitney U tests (Meng and Moyle 1995; Sommer
et al.
1997) because it does not require arbitrarily dividing an inherently continuous data set into “before” and “after” categories (
see
previous discussion of this issue in the August 17, 2001, notice; 66 FR 43145). We consider the CDFG/USBR MLR Model superior to the polynomial regression model presented in the August 17, 2001, notice (66 FR 43145) because existing abundance index monitoring programs have not been conducted for a sufficient duration to provide for reasonably conclusive application of the polynomial model (as concluded in the August 17, 2001, notice (66 FR 43145)). We also support use of the CDFG/USBR MLR Model because of the facility with which it can be applied to all sets of splittail age class data from all seven applicable abundance monitoring data sets (Fall MWT, Bay Study OT, Bay Study MWT, Chipps Island, Suisun Marsh, CVP salvage, and SWP salvage). The seven surveys include a total of 20 discrete sets of age-specific abundance monitoring data. These 20 datasets consist of the 2 age classes (0 and 1 or more) for the Suisun OT, in addition to the 3 age classes (0, 1, and 2 or more) for each of the other 6 surveys.

The CDFG/USBR MLR Model explicitly controls for potential confounding effects of hydrological year type, the factor that is nearly unanimously viewed as the single strongest predictor of splittail year class strengths (
e.g.
, Moyle
et al.
2001), by utilizing the number of days total delta inflow (DAYFLOW, California Department of Water Resources' mathematical hydrology model) exceeds 1,557 cubic meters per second (cms) (55,000 cubic feet per second (cfs)) during the February through May spawning/rearing period as a predictor (independent variable). The 1,557 cms (55,000 cfs) variable was selected because it approximates the critical inflow value above which Delta floodplains, especially the key splittail spawning area in the Yolo Bypass, become inundated. The 1,557 cms (55,000 cfs) variable thus captures the existence of appreciable bypass and spawning habitat inundation. This is conceptually comparable, yet superior, to the stratified Mann-Whitney U tests presented in the August 17, 2001, notice (66 FR 43145), which also controlled for hydrological year type. There is, however, one potentially important assumption associated with the CDFG/USBR MLR Model that remains untested, and that concerns the assumption of a lack of interaction between the HYDROLOGY and TIME variables. In essence, the CDFG/USBR MLR Model assumes that the long term probabilities of high and low flow water years are random.

Discussion of CDFG/USBR MLR Model results

The results addressed in this discussion differ somewhat from those published previously (67 FR 13095) due to the inclusion of new data for 2001 and 2002 in some of the indices as it has become available (see discussion of each survey, above). We also removed from the analysis data taken for the Suisun OT in 1979, based on comments received from the USBR (2002) indicating that differing survey protocols were used in 1979 as compared to other years.

The question of how to analyze the less-than-optimal data we have on splittail was vexing. In large part we have accepted the statistical model provided to us by CDFG and USBR. However, while our approach was generally consistent with theirs, there are two major differences. First, we used all 20 data sets weighted equally; whereas the BOR and CDFG recommended that the data sets be weighted by their relative importance. Second, we accepted a 20 percent risk that we would wrongly conclude there is a downward trend in the population for each of the 20 data sets in order to reduce the risk that we would fail to detect a trend if, in fact, one exist. We used this approach in order to ensure our assumptions were conservative. The effect was to establish a “worse-case” scenario with respect to the status of the populations when we conducted our threats analysis. As a result, our interpretation of the model results differs from theirs.

Our model results indicate that fifteen of twenty data sets have a downward trend, more downward trending data sets than we would expect based on chance. Typically, statisticians decide whether such trends are “statistically significant” or not. Interpreting the model results using the classic statistical standard (p # 0.05) for determining significance, we find that five of the fifteen downward trends are statistically significant. CDFG and USBR believe that this result is insufficient to make a determination that the splittail is declining in abundance. By adopting the more relaxed standard (p # 0.20), we increase the likelihood that a significant result will be identified, a conservative approach. Taking this approach (p # 0.20), we find nine significant downward trending data sets and two significant upward trending data sets. We believe that the existing data sets constitute the best available scientific information and that our more conservative approach indicates a number of significant declining splittail population trends exist. Coupled with the CDFG and USBR results, we have bracketed the range of possibility regarding the population status of the species as a whole. We believe this range is the best context for us to use when we conduct our threats analysis.

We fully concur with the statements of various respondents that abundance monitoring data for splittail have methodological weaknesses of one sort or another; none of the surveys were designed specifically to rigorously measure splittail population numbers (see Moyle
et al.
2001; Meng and Moyle 1995; and Sommer
et al.
1997 for

descriptions of surveys). However, existing data sets do constitute the best available scientific information for the species.

While our conservative approach to analyzing that information is more likely to produce results indicating that significant declining splittail population trends exist, we believe that using this “worst case” scenario in analyzing the impacts reported in the section entitled Summary of Factors Affecting the Species is most likely to result in a listing finding that is robust.

Because we have chosen to adopt the CDFG/USBR MLR Model jointly submitted by CDFG and USBR (as our primary basis for abundance analyses), and are no longer using our analysis in our August 17, 2001 notice (66 FR 43145), specific comments on our analysis in our August 17, 2001 notice (66 FR 43145) will not be addressed in the section entitled Summary of Comments and Recommendations.

Previous Federal Action

On February 8, 1999, we published a final rule listing the splittail as threatened under the Act (64 FR 5963). Please refer to the final rule for a discussion of Federal actions prior to the publication of the final rule. At the time of our final determination of threatened status for the splittail, the splittail population had declined in both numbers and range and was primarily threatened by changes in water flow and water quality resulting from the export of water from the Sacramento and San Joaquin Rivers, periodic prolonged drought, loss of shallow water habitat, introduced aquatic species, and agricultural and industrial pollutants.

Subsequent to the publication of the final rule, plaintiffs in the cases San Luis &
Delta-Mendota Water Authority
v.
Anne Badgley,
°
et al.
and State Water Contractors,
et al.
v.
Michael Spear
,
et al.
commenced action in the Federal Eastern District Court of California, challenging the listing of the splittail as threatened, alleging various violations of the Act and of the Administrative Procedure Act (5 U.S.C 551
et seq.
), specifically that we: (1) Failed to use the best scientific and commercial data available; (2) ignored all pre-1980 and post-1992 data available and that we used only selected data from the 1980 to 1992 period; (3) did not publish a summary of the available data, which data we considered, and the relationship between the data and our decision on the final rule; and (4) promulgated the final rule in a manner that was arbitrary, capricious, and not in accordance with law, in that the splittail did not meet the definition of a threatened species as set forth in the Act.

On June 23, 2000, the Court rendered summary judgment in the two cases in favor of the plaintiffs, finding that our promulgation of the final rule listing the splittail as threatened was unlawful. On September 22, 2000, the court remanded the determination of whether or not the splittail is a threatened or endangered species to us. The court ordered us to re-evaluate our final determination and publish a new finding within 6 months of the date of the remand order, and kept the rule in effect during that period. The court used its equitable powers to retain the protections of the Act for the species during the remand of the rule to the Service.

On January 12, 2001, we reopened the comment period for 30 days to seek information regarding the splittail's status, abundance and distribution, as well as information regarding issues identified by the District Court in its June 23, 2000, judgment (66 FR 2828). At that time, we were subject to a court-ordered deadline of March 22, 2001. On March 16, 2001, we received an extension from the District Court until June 22, 2001, so that we could reopen the comment period. Subsequent to that extension, we reopened the comment period for the second time since the remand, from May 8, 2001 to June 7, 2001 (66 FR 23181). On June 28, 2001, we received an additional extension from the court so that the comment period could be reopened and we could have additional time to obtain reviews of the revised statistical analyses which we employed in response to prior comments. The comment period was then opened on August 17, 2001 (66 FR 43145); while the court ordered decision date was established as January 31, 2002. We later received an additional extension from the court until October 15, 2002, so that we could seek comments on the MLR Model submitted by CDFG and USBR during the August 17, 2001, comment period. On March 21, 2002, we reopened the comment period for the fourth time since the remand (67 FR 13095) and on April 1, 2002, we corrected the duration of the comment period to reflect 60 days (67 FR 15337). On October 31, 2002, we received an additional extension from the court so that the comment period could be reopened for a fifth time since the remand (67 FR 66344) to solicit comments on the revised statistical analysis we had done, as described in our March 21, 2002 document (67 FR 13095). Finally, on February 28, 2003, the court approved a joint stipulation requiring us to submit our final determination to the
Federal Register
for publication on or before September 15, 2003. This final determination is in compliance with that joint stipulation agreement.

Summary of Comments and Recommendations

During the five comment periods following the remand, we contacted all appropriate State and Federal agencies, Tribes, county governments, elected officials, and other interested parties and invited them to comment. We have requested that all interested parties submit factual reports or information that might contribute to the development of a final determination. In addition, we have invited public comment through the publication of notices in various newspapers. We published notice of the January 12, 2001, reopening of the comment period in the
Sacramento Bee, Fresno Bee
and
Contra Costa Times
newspapers. For the May 8, 2001, notice, we invited public comment through publication of notices in the
Antioch Ledger-Dispatch, the Marysville Appeal-Democrat,
the
Fresno Bee,
and the
Sacramento Bee.
For the August 17, 2001, reopening notice we invited public comment through publication of notices in the
Marysville Appeal-Democrat,
the
Fresno Bee,
and the
Sacramento Bee.
An electronic mail address for submission of comments was provided in the May 8, 2001, and August 17, 2001, notices and was posted on the Sacramento Fish and Wildlife Office's official web site. For the March 21, 2002 reopening notice, we invited public comment through publication of notices on March 27, 2002, in the
Marysville Appeal-Democrat,
the
Sacramento Bee,
and the
Fresno Bee.
An electronic mail address was not provided for the March 21, 2002, reopening due to uncertainties regarding our internet access. An electronic mail address was, however, provided with our April 1, 2002, correction, and with our October 31, 2002, reopening. We also sent out notices of each reopening of the comment period to all parties on a mailing list for Sacramento splittail information.

During the five comment periods opened since the remand, we received a total of 33 written comment letters representing 1 Federal agency, 2 State agencies, 2 local governments, and 13 private individuals or organizations. We reviewed all comments received for substantive issues and new information regarding the status of the Sacramento splittail. Of the comments we received, only 3 supported listing. Information contained in these comments was reviewed to determine if it raised any

new substantive issues that had not been raised in comments previously submitted, and subsequently addressed in this final determination.

The following is a summary of comments we received during the 197 days associated with the five comment periods opened since the remand of the final listing rule. For additional information on comments received during three previous comment periods before the current litigation, please see the previous final listing rule (64 FR 5963). Substantive comments and information raised or provided during the public comments periods have either been incorporated directly into this notice or addressed below.

Peer Review

As previously discussed in the above abundance section, we requested 5 biologists to provide scientific review of the proposed listing of the splittail as threatened. Technical data provided by the peer reviewers have been incorporated into or addressed in this document, while other issues raised by the peer reviewers are addressed below.

Peer Reviewer Comment 1:
A peer reviewer cited the “White Paper” (Moyle
et al.
2001) for splittail as raising the possibility that abundance may not be a reliable measure of population status for the splittail.

Our Response:
We acknowledge that abundance may not be the most reliable measure of population status, but assert that it is the best scientific measure available. The utility of abundance as a measure of splittail population status is reflected in its continued use by the scientific community including researchers (Meng and Moyle 1995, Sommer
et al.
1997) and agencies (CDFG, CDWR, USBR).

Peer Reviewer Comment 2:
A peer reviewer cited the “White Paper” (Moyle
et al.
2001) for splittail as reporting a tentative population model result that stated, “* * * a long series of dry years is unlikely to drive the splittail to extinction, even if the population is greatly reduced.” Another peer reviewer asserted that if the splittail were truly going extinct, all surveys would show a decline.

Our Response:
A species warrants listing as threatened under the Act if is in danger of becoming endangered in the foreseeable future throughout all or a significant portion of its range (16 U.S.C. 1532(20)). It is possible for the splittail to be undergoing threats or declines in a significant portion of its range without declines showing in all surveys. Alternatively, threats to the splittail may support listing even in the absence of our ability to document current population declines. However, even considering our conservative analysis of the apparent splittail population declines and the threats analysis, we believe the conservation elements of the California State and Federal cooperative program (CALFED) and the Central Valley Project Improvement Act (CVPIA) programs adequately mitigate for these threats (please refer to Summary of Factors Affecting the Species section for a detailed discussion of CALFED and the CVPIA).

Peer Reviewer Comment 3:
A peer reviewer submitted comments that included an analysis using a modified version of Meng and Moyle's (1995) pre-decline and post-decline method. The peer reviewer also divided the data by year class and used data available from all years and requested we consider these analyses.

Our Response:
As discussed earlier in this notice, we acknowledge that there are other methods by which to analyze the available data, but that we have now employed an analysis using the CDFG/USBR MLR Model data series to describe population trends of the splittail. We refer the peer reviewer to our Abundance section for a discussion of our most recent statistical analysis of the species population trends.

Peer Reviewer Comment 4:
A peer reviewer criticized us for evaluating the results of the CDFG/USBR MLR Model for all 20 data series of splittail abundance index data, instead of limiting the evaluation to the nine data series that the respondents view as most representative of overall splittail populations. Another peer reviewer stated that Bay Study OT and Fall MWT data were more indicative of splittail abundance trends, rather than the trends made evident by data collected at the SWP Salvage facilities, Chipps Island, and in Suisun Marsh, which the respondent felt were narrow in geographic scope.

Our Response:
We note that these and other respondents have previously criticized us, while employing different analysis, for not treating all 20 data series equally and for not including all available data series in statistical evaluations of abundance trends. We refer the commentor to the section entitled Abundance for a discussion of our treatment of the data series.

Peer Reviewer Comment 5:
A peer reviewer reiterated his assessment that the statistical evidence for a declining trend in splittail abundance is weak, and cited an analysis that asserted that evidence for a time trend in 7 of 20 data series is not a compelling factor in determining that declines exist. The peer reviewer specifically cited Manly (2002) which states “The Service claims that lack of power to detect a trend gives a reason for using a 20 [percent] level of significance in assessing whether or not there is evidence of a trend with individual series. This then allows [the Service] to claim evidence for a trend for 7 of the 20 series. Although this sounds impressive, it is less so when it is realized that by chance alone 4 of the 20 series (
i.e.,
20 [percent] of them) are expected to give a significant result if this level of significance is used.” The peer reviewer also asserted that the weak nature of the MLR Model regression coefficients will be demonstrated with the calculation of splittail abundance indices for 2000, 2001, and 2002 and their inclusion into the models.

Our Response:
Using the most recent data, our analysis now indicates that 9 of 20 indices show significant negative trends at the 20 percent level of significance, while 2 of 20 show significant positive trends at that significance level. As we noted earlier in the analysis, we achieved these results by a conscious choice of a variable that accepted a higher risk of incorrectly identifying downward trends in population in order to take a conservative position in our threats analysis.

Peer Reviewer Comment 6:
A peer reviewer criticized our acceptance of the “sign” (
i.e.,
positive or negative) results of the CDFG/USBR MLR Model coefficients at face value because in most cases (16 of 20) the true signs (
i.e.,
positive or negative) were just as likely to be positive as negative.

Our Response:
We cannot apply the respondent's reasoning to the available data. The p-value for a coefficient is what statistical analysis has indicated it should be; simply because a given p-value does not rise to the level of 95 percent significance criterion, does not indicate that the p-value automatically reverts to 50 percent.

Peer Reviewer Comment 7:
It was noted by a peer reviewer that in half the CDFG/USBR MLR Model runs the dependent variable was significantly non-normal and that as a consequence probability statements will be “slightly” in error.

Our Response:
We believe that the peer reviewer's comment is correct. This type of error alone, however, would not necessarily invalidate our evaluations of the signs and magnitudes of the regression coefficients. The error would have to be of a nature that creates bias. The peer reviewer did not provide any statistical or other argument to explain why such error would necessarily result

in bias. The unknown statistical effects of non-normality in half the model runs constitutes just the sort of uncertainty that leads us to be cautious about giving undue weight to any conclusions regarding the abundance index data for splittail.

Peer Reviewer Comment 9:
A peer reviewer believes that the extended drought of 1984 to 1992 created only a perception of decline and that it was the “* * * accidental juxtaposition of a series of wet, strong splittail years with a series of dry, weak years that prompted [our] interest in the first place.”

Our Response:
We disagree with the peer reviewer's claims that the period of extended drought has been ignored, as well as with the contention that the splittail's drought-driven declines are the sole factor under consideration in our determination. We first note that the period of continuous drought is considered by most authoritative sources to have begun in 1987 (Moyle
et al.
2001; Baxter 1999a; Sommer
et al.
1997), not 1984 as reported by the respondent. We note, however, that 1985 and 1986 were dry years (Cannon 2001 in prep.).

The declines noted during the 1987 to 1992 drought were the likely result of a paucity of spawning habitat being available. The drought decreased the amount of floodplain (
i.e.
Yolo Bypass and mainstem river margins) available for spawning and thus, spawning output was lower. Low splittail population densities were aggravated by the CVP and SWP's diversion of a greater proportion of water from the Delta than in prior years; fish were entrained at the facilities and the entrapment zone (location where fish become vulnerable to the export facilities' effect on currents in the Delta), was located well upstream of Suisun Marsh in increasingly suboptimal habitat. These events are described in detail in our February 8, 1999, final listing rule (64 FR 5963).

The basis for the peer reviewer's claim that we are disproportionately concerned with splittail declines noted during the 1984 (or 1987) to 1992 drought is unclear. True, the “accidental juxtaposition” of wet and dry years resulted in abundance data that appeared to illustrate a precipitous drop in the splittail population. There are, however, up to 10 years of pre-drought as well as up to 8 to 10 years of post-drought data. The data collected during six years of continuous drought are but a subset of the nearly 20 years of extant splittail data. The splittail's relatively long life span and resilience following unfavorable conditions renders the declines exhibited during a discrete drought unlikely to influence the analytical findings from an ever-lengthening period of record. Most importantly, we now employ the CDFG/USBR MLR Model, which explicitly controls for potential confounding effects of hydrological year type. The respondent's concern would be more applicable to abandoned analytical techniques. The arbitrary pre- and post-decline cut point approach of Meng and Moyle (1985) was driven by trends noticed during the 1987 to 1992 drought, as was a formerly touted alternative analysis that involved the use of 1987 (the beginning of the drought) as a cut point (Sommer
et al.
1997) for determining percent declines.

We also disagree with the contention that the 1987 to 1992 drought serves as the only factor which triggered our investigations of the splittail's status. Our interest in the splittail was prompted initially by the statement in Daniels and Moyle (1983) that the splittail's and delta smelt's “* * * abundance could decline rapidly if environmental conditions become unfavorable for them, possibly making them candidates for listing as threatened species.” We subsequently included the Sacramento splittail as a category 2 candidate species for possible future listing as endangered or threatened in the January 6, 1989, Animal Notice of Review (54 FR 554). The candidate category system was abandoned on February 28, 1996 (61 FR 7457), and species meeting the definition of the former category 2 (such as splittail) were no longer considered candidates. Our administrative proceedings on splittail resumed on November 5, 1992, when we received a petition from the Natural Heritage Institute to add the Sacramento splittail to the List of Endangered and Threatened Wildlife and to designate critical habitat for this species in the Sacramento and San Joaquin Rivers and associated estuary.

Peer Reviewer Comment 10:
A peer reviewer, in response to our March 21, 2002 (67 FR 13095) notice, believed that we should not have adopted the CDFG/USBR MLR Model which was jointly submitted in CDFG's and USBR's respective peer review and comment letters. The CDFG/USBR MLR Model was advocated by its submitting agencies as an approach superior to our Meng and Moyle (1985) method utilized in our 1994 proposed listing (59 FR 862) and 1999 final listing (64 FR 5963) rules, the polynomial regression technique discussed in our August 17, 2001 (66 FR 43145) notice, or the Sommer
et al.
(1997) technique formerly forwarded by CDFG and CDWR.

Our Response:
We agree that USBR's submission was labeled
A Sample Alternative Model of Sacramento Splittail Abundance.
However, USBR (2001) included no language in their agency comment letter and peer review submission to suggest their intent was to have us retain the polynomial regression analysis (66 FR 43145), revert to the Meng and Moyle (1985) analysis, adopt the Sommer
et al.
(1997) analysis, or employ any other analytical technique until the CDFG/USBR MLR Model and results reached a greater state of refinement.

To the contrary, USBR's peer review and comment letter states, “Results presented in Table 1, include actual p-values for the Service's inspection.” (USBR 2001). To advocate we abandon the model is to advocate we abandon analysis of p-values. Furthermore, USBR scientifically derived and submitted multiple conclusions in their peer review and comment letter, such as, “In summary, the results [of the CDFG/USBR MLR Model] presented here clearly indicate that hydrologic variability strongly affects YOY splittail indices, and also affects some adult indices in succeeding years as cohorts propagate through the population.” (USBR 2001). These conclusions were not accompanied by any disclaimers that the conclusions should be disregarded because the model was not yet sufficiently developed or that the conclusions should not be applied to the review of the splittail's population trends.

The CDFG/USBR MLR Model was also submitted to us by CDFG. Consistent with the USBR peer review and comment letter, CDFG also derived and submitted multiple conclusions based on the specific runs of the CDFG/USBR MLR Model that the USBR is now criticizing us for accepting. CDFG advocated the use of the CDFG/USBR MLR Model (as submitted) in their peer review and comment letter (CDFG 2001) with the statement, “Our response is composed of two parts: a discussion of individual analyses presented in our August 17, 2001, notice (66 FR 43145), and a summary of the results of a multiple regression analysis [the CDFG/USBR MLR Model] that we believe is more useful in evaluating trends in survey indices.” Again, consistent with the USBR's peer review and comment letter, CDFG's peer review and comment letter did not qualify any of the conclusions they derived from the CDFG/USBR MLR Model with disclaimers about the inappropriateness of employing the model.

We independantly evaluated the structure and findings of the CDFG/

USBR MLR Model and determined that it represented the best available scientific and commercial information. We retain our conclusions regarding our analysis and meta-analysis of the model's results, regardless of its developers' current desire to secondarily qualify its application.

Peer Reviewer Comment 11:
A peer reviewer commented that it was unclear whether we had independently re-derived the CDFG/USBR MLR Model results submitted jointly by CDFG and USBR.

Our Response:
We did not independently re-derive those results. We accepted the results presented in CDFG (2001) and USBR (2001) at face value, as they were developed by subject area experts within CDFG and USBR during a peer review and public comment process.

Peer Reviewer Comment 12:
A peer reviewer believed that our that our statement, “* * * [the] traditional [alpha-value] criteria assume a much higher standard of statistical power than the splittail data are able to meet * * *” in our March 21, 2002, notice (67 FR 13095) is erroneous.

Our Response:
We agree with the peer reviewer that in a strictly literal sense, the choice of an alpha-value criterion can be made without any regard for statistical power. However, in practice, researchers are concerned with both type I error (determined by the choice of an alpha value) and type II error (directly related to the statistical power of a study). When conducting our analysis, we made a conscious choice to use the more conservative nontraditional approach of using an alpha value of 0.20.

Peer Reviewer Comment 13:
A peer reviewer asserted that the purpose of statistical hypothesis testing in the case of the MLR Model is to decide whether trends do or do not exist, not to evaluate gradients of reliability in evaluating trends.

Our Response:
The CDFG/USBR MLR Model is a probabilistic approach to examining time trends, it is not a categorical “either/or” approach (as the respondent appears to assert). We chose to evaluate the probabilities associated with competing hypotheses concerning the abundance status of splittail. It is for this reason that we stated that all trends, not just trends meeting an arbitrary traditional confidence criterion (95 percent confidence, or alpha-value of 0.05) were evaluated.

Peer Reviewer Comment 14:
A peer reviewer believes that a trawl's declining catch efficiency for adult splittail as compared to juvenile and YOY rendered trawl surveys less likely to reflect trends and stated that adult and juvenile indices should not be combined. The peer reviewer also suggested that Bay Study OT and Fall MWT were more representative measures of abundance.

Our Response:
While we concur that declining catch efficiency may be a characteristic of trawls, we do not agree that it should be used to exclude a trawl survey's data. Declining catch efficiency within a given trawl survey is expected to be uniform from year to year, thus rendering inter-annual analysis valid. Although an age bias will make data series for older age-class fish less sensitive for detecting change, it will not produce a long-term directional bias (
i.e.,
we have no reason to believe that the capture efficiency for older age class splittail is becoming progressively worse over time). Thus, any trends in the older age class data series with a substantive p-value can be viewed to be roughly as accurate and reliable as for the Age-0 class of splittail showing trends at comparable p-values.

We also concur that trawls' declining catch efficiency does preclude the combination of age class data. We report each index separately herein and do not combine adult and juvenile indices other than for meta-analytical purposes. We also acknowledge that, in certain situations, adult abundance for different age classes (of adults) is combined and reported because the data are collected in that manner,
i.e.,
salvage data are reported as Age-1 and as Age-2 and greater with no differentiation made for individuals greater-than Age-2 classes. Situations such as this represent a relict of the sampling methodology but remain the best available information. We continue to believe that as long as the degree of age-based capture bias is constant over a survey period, all age classes should show approximately the same trends, and that combining age classes for meta-level statistical analyses is not problematic.

We reiterate that the Suisun Marsh OT, which combines an efficient, bottom trawling technique with focused surveys in a small habitat at the core of the splittail's range, is the most likely to detect a trend and likely suffers from less sampling inefficiency than the Bay Study OT (low detection of splittail at periphery of range) and Fall MWT (unlikely to detect benthic fish and does not sample shallow water or near-shore areas).

Peer Reviewer Comment 15:
A peer reviewer asserted that the peer review process for scientific publications doesn't necessarily ensure that published papers are unbiased, scientifically sound, and without errors. The
Transactions of the American Fisheries Society
does not use the double-blind method for peer review. This issue was raised in regards to our past use of the Meng and Moyle (1995) methodology to determine splittail abundance.

Our Response:
We agree with this assertion. Each piece of scientific work, whether a peer reviewed published paper or an unpublished, unreviewed, draft report, must be objectively evaluated for the scientific merit of its content alone. Peer reviewed publication provides no guarantee of scientific merit. The test of time, following publication, provides the ultimate measure of scientific merit. Indeed, subsequent iterative examination of the splittail's status has resulted in our abandonment of Meng and Moyle (1995), Sommer
et al.,
(1997) and our permutation-based exact calculations of p-values for stratified (as opposed to unstratified) Mann-Whitney U-tests (66 FR 43145).

Peer Reviewer Comment 16:
A peer reviewer claimed we ignored the draft “White Paper” published by Moyle
et al.
(2001, in prep.)

Our Response:
We use the various findings and hypotheses found in the draft and revised White Paper (Moyle
et al.
2001 in prep.) extensively throughout this document.

Peer Reviewer Comment 17:
A peer reviewer stated that the range of the splittail is wider than was previously thought.

Our Response:
The greater range of the splittail was acknowledged in the January 12, 2001, notice (66 FR 2828). The above Background section of this final document contains a discussion of the range of the splittail.

Peer Reviewer Comment 18:
Several peer reviewers felt that we should not classify the Yolo and Sutter bypasses as a threat to the splittail, as we did in the January 12, 2001, reopening of comment period (66 FR 2828), based primarily upon the data found in Sommer
et al.
(1997) and Sommer (2001a). The bypasses have demonstrated the capability of producing large numbers of splittail when inundated. One peer reviewer also felt that the bypasses cannot be considered a threat simply because the conditions could be better. Another peer reviewer claimed that current operations in the bypasses do not harm splittail or their habitat. Another peer reviewer felt that the bypasses are not to be considered a threat because even though their splittail habitat conditions are not optimal, they are still sufficient to provide substantial benefits to the species. Finally, another peer reviewer

stated that the Yolo and Sutter bypasses are a “net benefit” to the splittail in that without their existence, the species might not have persisted to the present day.

Our Response:
We have determined, based on consideration of scientific data and information provided by respondents, that the Yolo and Sutter bypasses are not, in and of themselves, a threat to the splittail. Our reevaluation of this issue is discussed in Factor E of the section entitled Summary of Factors Affecting the Species.

Peer Reviewer Comment 19:
A peer reviewer felt that our determination that the Sutter and Yolo bypasses would require inundation for at least 30 continuous days between March and April in order for them to be considered a beneficial splittail spawning habitat was inaccurate and could affect water supply and flood management.

Our Response:
We have not proposed inundation of the bypasses for any specific interval, duration, or frequency. Rather, we have speculated that the bypasses would have their greatest benefits to splittail if they became inundated at a frequency and duration that as closely as possible mimics the natural, precipitation-driven hydrograph. The reference to 30 days is a statement regarding how the inundation patterns of the bypasses at times do not meet the life history requirements of the splittail. Inundation of bypasses in dry years would reduce the effects of drought on the splittail. We also speculate that if the bypasses were inundated at a frequency and duration that as closely as possible mimics the natural, precipitation-driven hydrograph, then the numbers of non-native fish would be reduced, as non-native fishes favor ponded and continuously inundated habitats.

Peer Reviewer Comment 20:
A peer reviewer believed that full implementation of the CALFED Program would preclude the need to list the splittail and indicated that over $10 million had been spent on actions that could improve conditions for splittail.

Our Response:
We refer the peer reviewer to the section entitled Summary of Factors Affecting the Species.

Peer Reviewer Comment 21:
A peer reviewer asserted that the age-based capture bias argues against combining data from different age groups.

Our Response:
We assume this comment refers to the pooling of data series from all age classes for meta-level statistical evaluation. We believe that as long as the degree of age-based capture bias is constant over a survey period, all age classes should show about the same trends, and that combining age classes for meta-level statistical analysis will not be a problem.

State Agencies

We received comments from the following California State agencies: Department of Fish and Game (CDFG) and Department of Water Resources (CDWR). Technical data provided by the CDFG and CDWR have been incorporated into or addressed in this document, while other issues raised by State agencies are addressed below:

State Agency Comment 1:
CDFG submitted comments that included an analysis using a modified version of Meng and Moyle's (1995) pre-decline and post-decline method. CDFG also divided the data by year class and used data available from all years and requested we consider these analyses.

Our Response:
As discussed earlier in this notice, we acknowledge that there are other methods by which to analyze the available data, but that we have now employed an analysis using the CDFG/USBR MLR Model data series to describe population trends of the splittail. We refer CDFG to our Abundance section for a discussion of our most recent statistical analysis of the species population trends.

State Agency Comment 2:
CDFG reiterated their assessment that the statistical evidence for a declining trend in splittail abundance is weak. CDFG cited an analysis that asserted that evidence for a time trend in 7 of 20 data series is not a compelling factor in determining that declines exist. CDFG specifically cited Manly (2002) which states “The Service claims that lack of power to detect a trend gives a reason for using a 20 [percent] level of significance in assessing whether or not there is evidence of a trend with individual series. This then allows [the Service] to claim evidence for a trend for 7 of the 20 series. Although this sounds impressive, it is less so when it is realized that by chance alone 4 of the 20 series (
i.e.,
20 [percent] of them) are expected to give a significant result if this level of significance is used.” CDFG also asserted that the weak nature of the MLR Model regression coefficients will be demonstrated with the calculation of splittail abundance indices for 2000, 2001, and 2002 and their inclusion into the models.

Our Response:
As we note in our earlier analysis we made a conscious decision to use the more conservative, nontraditional 0.20 alpha for analysis purposes.

State Agency Comment 3:
CDWR claimed we ignored the “White Paper” published by Moyle
et al.
(2001).

Our Response:
We agree with and use many of the various findings and hypotheses found in the draft and revised White Paper (Moyle
et al.
2001) extensively throughout this document. We believe that the White Paper is a useful resource and contributes to the knowledge on splittail biology. The paper has been referenced throughout this document.

State Agency Comment 4:
CDWR stated that the hypothetical analytical model presented at the January 29, 2001, CALFED Bay-Delta Program (CALFED Program) Splittail Science Conference and described in the White Paper (Moyle
et al.,
2001) indicates that the splittail, even during severe and lengthy drought, is unlikely to be driven to extinction.

Our Response:
We ultimately arrive at the same conclusion as Dr. Moyle, that the splittail is unlikely to be driven to extinction. However, at this point we are unwilling to accept that premise solely on the basis of the White Paper. To date, there remains no proven scientific method for determining the current splittail population size primarily because no extant survey was designed specifically to monitor splittail populations or to determine their absolute numbers. Further, the splittail exhibits relatively wide variation in annual abundance in response to prevailing hydrologic conditions; it is likely that the population size exhibits appreciable year to year variability which would confound size estimates.

Calculating the current population's risk of and/or time to extinction would require estimates of absolute population size, rate of decline, and minimum viable or sustainable population size, none of which currently exist in a scientifically defensible form. Moreover, it must also be noted that the statutory and regulatory standard for ascertaining threatened status is not to determine whether or why a species will become extinct in the near future, but if, pursuant to section 3(19) of the Act, it “* * * is likely to become an endangered species within the foreseeable future throughout all or a significant portion of its range”. An endangered species, pursuant to section 3(19) of the Act, is that “* * * which is in danger of extinction throughout all or a significant portion of its range * * *”. Our analysis, including a nontraditional conservative approach to estimating population trends examines the factors identified in the Act and in fact we find that the splittail does not warrant listing at this time.

State Agency Comment 5:
CDWR felt that we should not classify the Yolo and Sutter bypasses as a threat to the

splittail, as we did in the January 12, 2001, reopening of comment period (66 FR 2828), based primarily upon the data found in Sommer
et al.
(1997) and Sommer (2001a). The bypasses have demonstrated the capability of producing large numbers of splittail when inundated.

Our Response:
We have determined, based on consideration of scientific data and information provided by respondents, that the Yolo and Sutter bypasses are not, in and of themselves, a threat to the splittail. Our reevaluation of this issue is discussed in Factor E of the section entitled Summary of Factors Affecting the Species.

State Agency Comment 6:
CDWR felt that our determination that the Sutter and Yolo bypasses would require inundation for at least 30 continuous days between March and April in order to for them to be considered a beneficial splittail spawning habitat was inaccurate and could affect water supply and flood management.

Our Response:
We have not proposed inundation of the bypasses for any specific interval, duration, or frequency. Rather, we have speculated that the bypasses would have their greatest benefits to splittail if they became inundated at a frequency and duration that as closely as possible mimics the natural, precipitation-driven hydrograph. The reference to 30 days is a statement regarding how the inundation patterns of the bypasses at times do not meet the life history requirements of the splittail. Inundation of bypasses in dry years would reduce the effects of drought on the splittail. We also speculate that if the bypasses were inundated at a frequency and duration that as closely as possible mimics the natural, precipitation-driven hydrograph, then the numbers of non-native fish would be reduced, as non-native fishes favor ponded and continuously inundated habitats.

State Agency Comment 7:
CDWR commented that our classification of the Yolo Bypass as a threat in the January 12, 2001, notice (66 FR 2828) would undermine potential ecosystem restoration actions that would benefit the splittail.

Our Response:
In this notice, we determine that the Sutter and Yolo bypasses are not in and of themselves threats.

State Agency Comment 8:
CDWR objected to our statements regarding the entrainment risks present in the bypasses based upon Sommer
et al.
's (1997) findings that entrainment is not a significant threat within the bypasses. It is thought that the splittail's evolutionarily-derived ability to emigrate prior to stranding reduces the risk of stranding. CDWR also felt that the magnitude of the entrainment threats presented by the bypasses was overestimated when we cited in the January 12, 2001, notice (66 FR 2828), the death of a number of juvenile splittail in an approximately 0.8 hectare (ha) (2 acre (ac)) borrow pit as statistically-significant and that the classification of “natural sinks” as a threat was in error.

Our Response:
We have considered these data and now agree that entrainment in the Yolo Bypass is less than was originally thought. Information presented at the January 29, 2001, CALFED Splittail Science Conference indicates that a modest degree of topographic variability within an inundated area may be beneficial, as it may create a diversity of flow patterns and velocities which in turn may allow juvenile splittail to evade predation and forage more effectively during egress.

State Agency Comment 9:
CDWR believed that full implementation of the CALFED Program would preclude the need to list the splittail and indicated that over $10 million had been spent on actions that could improve conditions for splittail.

Our Response:
We refer CDWR to the section entitled Summary of Factors Affecting the Species.

Other Public Comments and Responses

We address other substantive comments and accompanying information in the following summary. Relatively minor editing changes and reference updates suggested by commenters have been incorporated into this document, as appropriate.

Comment 1:
The court directed that we provide a more thorough response to the California Resources Agency comments, specifically comments submitted by CDFG and CDWR in July 1998. The court also directed that we address the perceived biases from the Meng and Moyle (1995) method. We also received specific comments on issues related to prior statistical analyses of abundance.

Our Response:
We have adopted a multiple linear regression approach proposed by CDFG and U.S. Bureau of Reclamation (USBR). CDFG, in comments submitted in association with the August 17, 2001, comment period, stated: “Although CDFG reported Mann-Whitney U test results in previous comments (February 8, 2001), we now suggest greater reliance on a multiple regression approach to trend analysis, described in a following section of our comments. We no longer support use of the Mann-Whitney U procedure of time trend analysis.” CDWR, in comments also submitted under the August 17, 2001, comment period, stated: “A more defensible alternative would be to develop a multivariate model incorporating the effects of both flow and time.” CDWR also made reference to the USBR application of regression techniques, which also were provided in USBR's comments. We have considered the CDFG, CDWR, and USBR recommendations to employ a multivariate, regression based model and have incorporated an analysis using the CDFG/USBR MLR Model data series as described in the section entitled Abundance. We will therefore forego providing responses to specific comments on the perceived bias of the Meng and Moyle (1995) and alternate methodologies previously employed because our analytical tools have been upgraded to utilize the modified methodology employed by CDFG and USBR.

The CDFG/USBR MLR Model provided in CDFG and USBR comments addresses the shortcomings of other methods, thus allowing our analysis using the CDFG/USBR MLR Model data series to supercede abundance analyses based upon methods appearing in prior rules. In combination with meta-analyses to analyze the distribution of MLR Model results across the 20 indices, statistical inferences based on the CDFG/USBR MLR data series are informative.

Our analysis using the CDFG/USBR MLR Model data series incorporates the results of seven surveys (Fall MWT, Bay Study OT, Bay Study MWT, Chipps Island, Suisun Marsh OT, CVP salvage, and SWP salvage), and includes separate indices of YOY, age 1 (juvenile) and age 2+ (adult) age class abundance. The independent examination of abundance of all age classes throughout these surveys helps mediate discrepancies among survey results, discrepancies that are a likely indication that splittail populations are not very evenly distributed over space and time and/or that different sampling methodologies are not very comparable. The model also does not require uninterrupted data; all available data from each survey's period of record is included. Further, our analysis controls for the confounding effects of hydrology, and involves no inherent or intentional bias towards either wet or dry water year types. Strict adherence to uniformity among all data series is also inconsistent with the precautionary nature of section 4 of the Act.

We recognize a distinct danger in controlling for hydrological effects in our analyses, because systematic

changes in hydrological regimes, due to human manipulation or long term climate change, could just as feasibly be a causative factor as a confounding source of “noise.” If systematic changes in hydrological regimes were occurring, it would not be prudent to control for that factor. Our since-superceded polynomial regression analysis of abundance data (See Abundance section of the August 17, 2001, notice) controlled for influences of hydrological cycles without discarding hydrology as a potential directional factor determining long term trends of splittail abundance. We expect that the polynomial regression analysis presented in the August 17, 2001, notice may eventually inform the understanding of the effects of changed hydrology on the splittail, once the future, cumulative hydrologic analyses for potential water development projects have been developed by the responsible agencies.

Comment 2:
The court directed us to show the relationship between the data used in our decision-making analysis and the original final rule and how we reached the conclusion that the splittail was threatened.

Our Response:
We have provided a more detailed analysis in the section entitled Summary of Factors Affecting the Species. The threats to the species have also been summarized in an additional section entitled Conclusion Regarding Abundance, Distribution, and Factors Affecting the Species. We have also included in the Abundance section of this notice a discussion of our most recent statistical analysis of the species population trends.

Comment 3:
Several respondents cited the draft White Paper (Moyle
et al.
2001 in prep) for splittail as reporting a tentative population model result that stated, “ * * * a long series of dry years is unlikely to drive the splittail to extinction, even if the population is greatly reduced.”

Our Response:
A species warrants listing as threatened under the Act if is in danger of becoming endangered in the foreseeable future throughout all or a significant portion of its range (16 U.S.C. 1532(20)). It is possible for the splittail to be undergoing threats or declines in a significant portion of its range without declines showing in all surveys. Alternatively, threats to the splittail may support listing even in the absence of our ability to document current population declines. Finally, we believe the conservation elements of the California State and Federal cooperative program (CALFED) and the Central Valley Project Improvement Act (CVPIA) programs adequately mitigate for these threats (refer to Factor A for a detailed discussion of the CALFED program and the CVPIA programs).

Comment 4:
A respondent informed us that CDFG re-analyzed the striped bass egg and larval survey and found that splittail spawn in the mainstem of the Sacramento River, especially in dry years. This indicates that splittail occur in the Sacramento River upstream from the Delta.

Our Response:
CDFG and our survey results confirm that splittail use river margin habitat in the mainstem Sacramento River. Indeed, recent indications are that river margin habitat is where splittail spawning occurs through periods of drought.

Comment 5:
A respondent stated that young of the year (YOY) abundance was at near record levels in 2000, thus inferring the splittail is not in decline.

Our Response:
Data presented in the Spring 2001
Interagency Ecological Program Newsletter
(Baxter 2001a), provided as an attachment to public comment submitted on this rulemaking, do indicate that splittail spawning was highly successful in 2000. This spike of juvenile fish is to be expected given the relatively wet conditions of 2000, and the splittail's ability to exploit suitable habitat when available. Also, YOY are generally the most reliably sampled fish in any given survey, since their raw abundance is temporarily high and YOY splittail are likely less effective at evading sampling equipment. Population level conclusions drawn from such a spike must be made with caution because, though extremes in YOY abundance appear to be reflected in 2 to 3 year subsequent adult abundances, the splittail appears to exhibit no stock-recruitment relationship (Sommer
et al.
1997). Possible reasons for the lack of a stock-recruitment relationship may be variation in female growth, survivorship and fecundity from such causes as inter- and intra-annual hydrologic variation, environmental contaminants, years of non-spawning, predation, etc., which may be exerting independent or synergistic influences on recruitment of splittail into the population. Regardless of cause, large portions of YOY fail to survive to the adult, spawning population age class. Juvenile abundance may therefore be inadequate to fully describe the size of the standing or future adult populations and may also be inadequate to describe the ability of the population to persist. Population abundance cannot be accurately predicted based upon examination of juvenile abundance alone.

We currently support use of the CDFG/USBR MLR Model because of the facility with which it can be applied to all sets of splittail age class data from all seven applicable abundance monitoring data sets (a total of 20 discrete sets of age-specific abundance monitoring data). This approach therefore includes consideration of YOY splittail without granting undue analytical weight to any single survey or age class or inappropriately combining different survey equipment types. Regardless of the strengths and weaknesses of year 2000 YOY abundance, these data were considered in our analysis using the CDFG/USBR MLR Model data series (
see
our Abundance section for a discussion of our most recent statistical analysis of the species population trends).

Comment 6:
The Court and numerous commenters requested that we address and clarify the issue of splittail resiliency and that the species may be able to withstand drought and produce high numbers of young of year (YOY) during wet periods.

Our Response:
We concur that splittail are a resilient species and that they can reproduce effectively in wet years. Sacramento splittail populations fluctuate annually depending upon the availability of shallow water habitat with submerged vegetation (Daniels and Moyle 1983). Meng and Moyle (1995) and Sommer
et al.
(1997) have found that splittail year-class abundance is positively correlated with freshwater outflow occurring during the species' late winter and spring spawning season. The evolutionary strategy of the splittail therefore appears to be one of opportunism, whereby the population collectively invades and exploits spawning habitats if and when they become available. Historically, this resilience is likely to have maintained the population of splittail through extended droughts. This resilience also has allowed the splittail to persist in spite of the significant loss of habitat that has occurred since the species was first described by Ayres.

Comment 7:
A respondent wished to know why the Bay Study and CVP and SWP salvage data showed an increase in splittail abundance, and the commenter requested that we explain the variation in the study results.

Our Response:
This comment pertains to the Meng and Moyle (1995) methodology employed in our previous analyses of splittail population. We refer the respondent to our Abundance section for a discussion of our most recent statistical analysis of the species population trends.

We believe that trends noted in the Bay Study are likely due to the large numbers of YOY fish that were collected during certain wet years. High outflows may transport juveniles from the Estuary to locations where Bay Study samples are collected. It is unclear what happens to these fish once they are transported to these areas. Fish transported to San Pablo Bay may survive to join, if not sustain, the Napa River and Petaluma River and Marsh subpopulations. Once located in these areas, it is not known what contribution is made to the Central Valley population as a whole.

In regard to trends in CVP and SWP salvage data, we believe that these too are driven by seasonal variation in hydrology. Though it is true that hydrology and production are strongly correlated, and that salvage would be expected to rise as populations rise, there are concerns with the data's application (see discussion of surveys under Abundance section, above).

In the case of splittail salvage, entrainment is likely influenced by the rate, or volume per unit of time, of export. As stated before, salvage data are expressed as fish captures per acre foot and lack a time value. At higher rates of export, splittail are likely to be disproportionately entrained because of higher velocities in the channels adjoining or approaching the facilities and thus, abundance could be overestimated. All sampling gears may be more effective at capturing splittail during high outflows due to increased velocity and turbidity, but only the pumps have the ability to draw fish towards them at different rates. The rate at which fish may become pulled towards the pumps cannot be described using existing data. Differing rates of export also introduce variability, which cannot be discerned without a time factor. Salvage data, as mentioned previously, do not effectively sample a large extent of the splittail population, as fish reared in the Sacramento River and/or Yolo Bypass are likely to largely avoid the pumps. Salvage data do however collect the largest number of splittail of any survey.

Comment 9:
Several respondents cited an analysis that took issue with us for adopting a non-traditional alpha-value of 0.20 (instead of 0.05) for evaluating results of the CDFG/USBR MLR Model.

Our Response:
Available literature customarily demands a rigid adherence to the traditional alpha value of 0.05. In this particular analysis, we chose to take a far more conservative approach in terms of how we evaluated the splittail's abundance. Accordingly, we used the non-traditional alpha value of 0.20. We believe while unusual it is conservative, and results in a more robust determination of whether the species should be listed.

Comment 10:
Several respondents cited an analysis that criticized our treatment of separate surveys of splittail abundance indices as statistically independent.

Our Response:
We followed a long established practice in the peer-reviewed literature on splittail of treating these surveys as statistically independent (
e.g.
, Meng and Moyle 1995; Sommer
et al.
1997) including papers repeatedly cited by the respondents in previously submitted comments. We accept at face value Manly's (2002) conclusion that an analysis of corrections among residuals provides evidence for some degree of interdependence among the different sets of survey data (Manly 2002:4-6). We also accept at face value Manly's (2002) attempt to correct our meta-analysis of survey results to account for the interdependence in the data sets. We have consistently stated that the abundance index data for splittail suffer from several fundamental inadequacies that make them far from ideal for decision-making purposes (an opinion with which the respondents and their statistical consultant concur (Manly 2002:3,8)).

Comment 11:
Several respondents criticized us for evaluating the results of the CDFG/USBR MLR Model for all 20 data series of splittail abundance index data, instead of limiting the evaluation to the nine data series that the respondents view as most representative of overall splittail populations.

Our Response:
We note that these and other respondents have previously criticized us, while employing different analysis, for not treating all 20 data series equally and for not including all available data series in statistical evaluations of abundance trends.

We are aware of no other party who has rigorously evaluated abundance index data (
e.g.
, Sommer
et al.
1997; Meng and Moyle 1995; Moyle
et al.
2001 in prep.) that has deemed it appropriate to limit the evaluation to the nine data series favored by the respondents. Further, CDFG and USBR elected to include all 20 data series in the CDFG/USBR MLR Model applications submitted to us as part of earlier comments.

We disagree with the respondent's suggestion that only data from a select group of nine survey indices that sample a wide geographic area (we assume the respondent is referring to three age classes each of the Bay Study MWT, Bay Study OT, and Fall MWT) should be given greater weight for making population-scale determinations. Weighting such a select group of surveys necessarily could require inappropriately combining their indices. The nine surveys are a composite of appreciably different gear types, some of which suffer from the same detection limitations as were used by other respondents to advocate against accepting certain other surveys. Mid-water trawling is an inappropriate match to splittail habitat preferences and other aspects of splittail biology, so even geographically extensive mid-water surveying would not necessarily be any more representative of overall splittail populations than geographically more restricted surveys better matched to splittail biology.

We disagree with respondents' claims that Bay Study MWT, Bay Study OT, and Fall MWT data are more indicative of splittail abundance trends than are those found in data collected at the SWP and CVP salvage facilities, Chipps Island, and in Suisun Marsh because they each suffer from gear or location difficulties. We postulate that each of these surveys is, to varying degrees, unsuited to the task of assessing splittail abundance. The Bay Study OT employs the efficient otter trawling technique but only infrequently captures splittail; surveys are conducted on the periphery of the species' range. The Bay Study MWT employs an inefficient (at capturing splittail) mid-water trawl. The Fall MWT fails to sample near-shore areas and the benthos (bottom), where splittail are most likely to occur. The Fall MWT does not sample shallow waters; in Suisun Bay/Marsh 8 of 25 sites are shallow, 1 of 38 in the Delta are shallow. We acknowledge that the Chipps Island Survey is a midwater trawl of deep channels and that it too would suffer from a similar bias. The CVP and SWP salvage data may suffer from an unquantifiable differential entrainment based on export rates (
see
Abundance section, above).

We also do not believe it is necessarily correct to infer that the wider geographical coverage of the nine surveys in question, alone, is sufficient to guarantee that those surveys are more representative of overall splittail populations. The Bay Study MWT, Bay Study OT, and Fall MWT are geographically wider in distribution, but given that estuarine conditions are specifically managed to maintain optimum habitat conditions within Suisun Marsh, the wider survey areas of the Bay Study MWT, Bay Study OT, and Fall MWT are not likely to contribute to a more informed trend analysis. Surveys need not cover large areas if a fixed

point is likely to result in detection of an appreciable number of individuals of a migratory species; splittail are as likely to arrive at a static survey point in a key location as they are to be captured by a mobile survey of varied habitats.

We understand the respondent's logic in formulating a hypothesis that the nine surveys in question might be most representative of the overall splittail population due to geography, but note that at this point such an opinion is only a working hypothesis with no actual data available to either support or refute it. Until such data become available, we believe it is most conservative to follow the practice of evaluating all the data series rather than combining or rejecting discrete sets. We continue to believe that, of the individual indices, the Suisun Marsh Otter Trawl should be the most appropriate sampling method because it samples core splittail habitat, utilizes an effective, bottom-trawling gear, and samples a greater relative proportion of the habitat at the sampling site.

Comment 12:
A respondent claimed we employed “Shifting approaches to the splittail listing” in regard to statistical testing of available data.

Our Response:
Since we have published one listing notice for the splittail, on February 8, 1999 (64 FR 5963), we assume that this respondent is actually referring to our evolving evaluations of data relevant to the issue of whether the splittail should be listed or not, as have appeared in the January 12, 2001 (66 FR 2828); May 8, 2001 (66 FR 23181); August 17, 2001 (66 FR 43145); and March 21, 2002 (67 FR 13095); and October 31, 2002 (67 FR 66344), notices reopening public comment periods.

It is common practice in science to continually formulate and revise hypotheses in response to new information. We have applied this scientific process during the review of the splittail's status, as have certain respondents (
i.e.
CDFG, CDWR, USBR). The evolving results of our various statistical analyses and the background information describing the bases for those analyses have each appeared in successive notices. Notices are solicitations for public comment and information, not final agency actions. As a result of new scientific information and comments received during the many comment periods, we have updated our analytical methodology based on the best scientifically and commercially available information. Note also that neither we, nor respondents, have advocated nor implemented a return to the superceded techniques used by Meng and Moyle (1995), Sommer
et al.
(1997), or the permutation-based exact calculations of p-values for stratified Mann-Whitney U-tests, published on August 17, 2001 (66 FR 43145).

Comment 13:
A respondent claimed that we were incorrect in departing significantly from the analysis of CDFG and USBR.

Our Response:
We did not depart at all from the statistical analysis provided by CDFG and USBR in the form of the CDFG/USBR MLR Model. We have fully accepted the model results submitted by CDFG and USBR. We have noted earlier in our analysis where we have departed from the CDFG and BOR analysis and our reasons for doing so.

Comment 14:
Several respondents stated that the extended drought of 1984 to 1992 created only a perception of decline and that it was the “* * * accidental juxtaposition of a series of wet, strong splittail years with a series of dry, weak years that prompted [our] interest in the first place.”

Our Response:
We disagree with the respondent's claims that the period of extended drought has been ignored, as well as with the contention that the splittail's drought-driven declines are the sole factor under consideration in our determination. We first note that the period of continuous drought is considered by most authoritative sources to have begun in 1987 (Moyle
et al.
2001; Baxter 1999a; Sommer
et al.
1997), not 1984 as reported by the respondent. We note, however, that 1985 and 1986 were dry years (Cannon 2001 in prep.).

The declines noted during the 1987 to 1992 drought were the likely result of a paucity of spawning habitat being available. The drought decreased the amount of floodplain (
i.e.
Yolo Bypass and mainstem river margins) available for spawning and thus, spawning output was lower. Low splittail population densities were aggravated by the CVP and SWP's diversion of a greater proportion of water from the Delta than in prior years; fish were entrained at the facilities and the entrapment zone (location where fish become vulnerable to the export facilities' effect on currents in the Delta), was located well upstream of Suisun Marsh in increasingly suboptimal habitat. These events are described in detail in our February 8, 1999, final listing rule (64 FR 5963).

The basis for the respondent's claim that we are disproportionately concerned with splittail declines noted during the 1984 (or 1987) to 1992 drought is unclear. True, the “accidental juxtaposition” of wet and dry years resulted in abundance data that appeared to illustrate a precipitous drop in the splittail population. There are, however, up to 10 years of pre-drought as well as up to 8 to 10 years of post-drought data. The data collected during six years of continuous drought are but a subset of the nearly 20 years of extant splittail data. The splittail's relatively long life span and resilience following unfavorable conditions renders the declines exhibited during a discrete drought unlikely to influence the analytical findings from an ever-lengthening period of record. Most importantly, we now employ the CDFG/USBR MLR Model, which explicitly controls for potential confounding effects of hydrological year type. The respondent's concern would be more applicable to abandoned analytical techniques. The arbitrary pre- and post-decline cut point approach of Meng and Moyle (1985) was driven by trends noticed during the 1987 to 1992 drought, as was a formerly touted alternative analysis that involved the use of 1987 (the beginning of the drought) as a cut point (Sommer
et al.
1997) for determining percent declines.

We also disagree with the contention that the 1984 to 1992 drought serves as the only factor which triggered our investigations of the splittail's status. Our interest in the splittail was prompted initially by the statement in Daniels and Moyle (1983) that the splittail's and delta smelt's “ * * * abundance could decline rapidly if environmental conditions become unfavorable for them, possibly making them candidates for listing as threatened species.” We subsequently included the Sacramento splittail as a category 2 candidate species for possible future listing as endangered or threatened in the January 6, 1989, Animal Notice of Review (54 FR 554). The candidate category system was abandoned on February 28, 1996 (61 FR 7457), and species meeting the definition of the former category 2 (such as splittail) were no longer considered candidates. Our administrative proceedings on splittail resumed on November 5, 1992, when we received a petition from the Natural Heritage Institute to add the Sacramento splittail to the List of Endangered and Threatened Wildlife and to designate critical habitat for this species in the Sacramento and San Joaquin Rivers and associated estuary.

Comment 15:
A respondent questioned how the data collected relate to a conclusion that the species is threatened. We had not provided analyses of population level outcomes that could be linked to threats analyses. Another respondent believed that our threats analysis is speculative,

imprecise, and meaningless absent any data or analysis concerning population level effects and that the threats analysis does not show why the species is threatened because of the factors, as required under section 4 of the Act

Our Response:
We refer the respondent to the sections entitled Summary of Factors Affecting the Species and Conclusion Regarding Abundance, Distribution, and Factors Affecting the Species. We believe that the splittail does not qualify for threatened status at this time based on our analysis of the threats.

Comment 16:
Several respondents asserted that the peer review process for scientific publications doesn't necessarily ensure that published papers are unbiased, scientifically sound, and without errors. The
Transactions of the American Fisheries Society
does not use the double-blind method for peer review. This issue was raised in regards to our past use of the Meng and Moyle (1995) methodology to determine splittail abundance.

Our Response:
We agree with this assertion. Each piece of scientific work, whether a peer reviewed published paper or an unpublished, unreviewed, draft report, must be objectively evaluated for the scientific merit of its content alone. Peer reviewed publication provides no guarantee of scientific merit. The test of time, following publication, provides the ultimate measure of scientific merit. Indeed, subsequent iterative examination of the splittail's status has resulted in our abandonment of Meng and Moyle (1995), Sommer
et al.
, (1997) and our permutation-based exact calculations of p-values for stratified (as opposed to unstratified) Mann-Whitney U-tests (66 FR 43145).

Comment 17:
Several respondents claimed we ignored the “White Paper” published by Moyle
et al.
(2001)

Our Response:
We agree with and use the various findings and hypotheses found in the draft and revised White Paper (Moyle
et al.
2001) extensively throughout this document. We believe that the draft White Paper is a useful resource and contributes to the knowledge on splittail biology, though it has not yet been finalized. The paper has been referenced throughout this document.

Comment 18:
A respondent requested that we acknowledge that the Interagency Ecological Program (IEP) provides oversight for fisheries data collection.

Our Response:
We concur that the IEP has oversight of the various fishery programs. However, various agencies collect the data for the surveys mentioned previously in this document. CDFG conducts the Fall Midwater Trawl, summer townet, and the Bay study; we conduct the beach seine and Chipps Island Survey; UC Davis conducts the Suisun Marsh OT, and USBR and CDFG collect the salvage and creel census data.

Comment 19:
A respondent felt the 2000 Service beach seine survey data supported the respondent's earlier comments that splittail were not declining. The respondent stated that new insights include: (1) YOY abundance was at a near record level in 2000; (2) distribution data show that in years of low spring outflow (
e.g.
, 1992, 1994, and 1997), the largest catches of young splittail occurred upstream in the Sacramento River, upstream of many sampling programs; and (3) splittail spawn and recruit even in dry years.

Our Response:
YOY abundance for a species with naturally high juvenile mortality does not necessarily equate with high recruitment. The respondent's statement that distributional data show that in years of low spring outflow (
e.g.
1992, 1994, and 1997), the largest catches of young splittail occurred upstream in the Sacramento River, is inaccurate for two of the three years referenced, and faulty conclusions are drawn from the data.

Water year 1992 exhibited similar abundances of splittail in upper Sacramento River and Far North Delta locations, and moderate abundance overall. Water year 1994 did exhibit relatively higher abundance in upstream locations, but abundance was low throughout all locations. Water year 1997 was wet, not dry as stated by the respondent. Also, regardless of being a wet year, water year 1997 exhibited low splittail abundance in all locations. Further, we expect that YOY spawned higher in the Sacramento River to suffer higher mortality, relative to fish spawned in the Delta, as they migrate downstream through progressively-worsening habitat conditions to rejoin the core population. Increased mortality among splittail spawned upstream may explain why YOY tend to be captured less frequently in downstream trawl-based surveys in certain dry years. The final statement, that splittail spawned upstream exhibit successful spawning and recruitment in dry years, is not supported by survey data. While spawning success can be inferred from YOY abundance, YOY fish do not necessarily recruit to the adult population. There is some evidence that high or low YOY abundance is correlated, with a two to three year time lag, with adult abundance. For this reason, YOY abundance cannot be excluded entirely. Our we believe our analysis using the CDFG/USBR MLR Model data series (see section entitled Abundance) incorporates all applicable YOY and adult abundance data, though excludes the beach seine due to its lack of a reliable catch per unit time indicator (seine hauls do not accurately account for time, or unit area per time sampled). Beach seine data are best employed with regard to the splittail for determining range and timing of occurrence.

Comment 20:
A respondent stated that while splittail are able to persist in a few key areas, such as Suisun Marsh and the lower Sacramento River, during periods of low flow, the relatively smaller populations would be vulnerable to a large scale disaster (
e.g.
, toxic spill), habitat loss, entrainment mortality, reduced outflows, non-native species predation, and contaminants.

Our Response:
We refer the respondent to the section entitled Summary of Factors Affecting the Species.

Comment 21:
The respondent stated that the court requested that we provide: (1) An estimate of the current population size of the splittail; (2) determine whether or why the current populations size is inadequate to prevent extinction in the near future; (3) determine the rate of population decline of splittail; and (4) identify the minimum viable population size. In addition, a respondent stated that the hypothetical analytical model presented at the January 29, 2001, CALFED Bay-Delta Program (CALFED Program) Splittail Science Conference and described in the White Paper (Moyle
et al.
, 2001) indicates that the splittail, even during severe and lengthy drought, is unlikely to be driven to extinction.

Our Response:
There remains no proven scientific method for determining the current splittail population size primarily because no extant survey was designed specifically to monitor splittail populations or to determine their absolute numbers. Further, the splittail exhibits relatively wide variation in annual abundance in response to prevailing hydrologic conditions; it is likely that the population size exhibits appreciable year to year variability which would confound size estimates.

Calculating the current population's risk of and/or time to extinction would require estimates of absolute population size, rate of decline, and minimum viable or sustainable population size, none of which currently exist in a scientifically defensible form. Moreover, it must also be noted that the statutory and regulatory standard for ascertaining

threatened status is not to determine whether or why a species will become extinct in the near future, but if, pursuant to section 3(19) of the Act, it “* * * is likely to become an endangered species within the foreseeable future throughout all or a significant portion of its range”. An endangered species, pursuant to section 3(19) of the Act, is that “* * * which is in danger of extinction throughout all or a significant portion of its range * * *”.

As stated above, analytical techniques do not exist to determine the rate of splittail population decline with current splittail data. Again, the absence of survey methodologies specifically designed to monitor splittail populations is a limiting factor in

[Text truncated at 120,000 characters. The full text is on the page linked above.]

---

Source: Frix Law Library, https://www.frixlaw.com/law-library/documents/fr%3A03-23919. Public record. Not legal advice.