Transactions of the American Fisheries Society 129:262-281, 2000 Copyright by the

American Fisheries Society 2000

Assessing Salmonid Spawning Gravel Quality

G. MATHIAS KONDOLF*
Department of Landscape Architecture and Environmental Planning and Department of Geography.
University of California, Berkeley, California 94720, USA

Abstract.Much of the recent literature on salmonid spawning gravels has been devoted to the search for a single
statistic drawn or computed from the streambed particle size distribution to serve as an index of gravel quality.
However, a natural gravel mixture cannot be fully described by any single statistic, because gravel requirements of
salmonids differ with life stage, and thus the appropriate descriptor will vary with the functions of gravel at each life
stage. To assess whether gravels are small enough to be moved by a given salmonid to construct a redd, the size of the
framework gravels (the larger gravels that make up the structure of the deposit) is of interest, and the d50 or d84 of the
study gravel (the sizes at which 50% or 84% of the sediments are finer) should be compared with the spawning gravel
sizes observed for the species elsewhere. To assess whether the interstitial fine sediment content is so high as to
interfere with incubation or emergence, the percentage of fine sediment of the potential spawning gravel should be
adjusted for probable cleansing effects during redd construction, and then compared with rough standards drawn from
laboratory and field studies of incubation and emergence success. An assessment should also consider that the fine
sediment content of gravel can increase during incubation by infiltration, that the gravels may become armored over
time, or that downwelling and upwelling currents may be inadequate. These considerations are incorporated in a nine-
step, life-stage-specific assessment approach proposed here.

The size of available streambed gravels can limit the single statistics drawn or computed from the size distribution
success of spawning by salmonids (Groot and Margolis curves that describe gravel mixtures (e.g., Lotspeich and Ev-
1991). The bed material may be too coarse for spawning fish erest 1981; Shirazi and Seim 1981, 1982; Beschta 1982).
to move, a problem particularly common where dams These single-variable statistics are easier to report than full
eliminate supplies of smaller, mobile gravels (e.g., Parfitt and size distributions and provide convenient independent
Buer 1980). Excessive levels of interstitial fine sediment may variables against which to compare incubation and
clog spawning gravels, effects that have been do cumented emergence success in field and laboratory studies. However,
downstream of land uses that increase sediment yields, such there is no reason to expect that any single statistic can fully
as timber harvest and road construction (Cederholm and Salo represent the attributes of the gravel size distribution relevant
1979; Everest et al. 1987; Meehan 1991). to the distinct functions of redd construc tion, embryo
Because of these problems, there is frequently a need incubation, and fry emergence. Gravel size plays a different
to assess the quality of spawnin g gravels to determine role in each life stage, and thus the relevant size attributes
whether gravel size limits spawning success. Any such differ.
assessment involves comparison of gravel size on site with In this paper I consider the gravel requirements of
information on gravel size suitability from laboratory studies each life stage and the need for comparability among studies,
or field observations elsewhere. Although many of the and review size descriptors proposed as indices of gravel
fundamental questions are essentially sedimento-logical and quality from a geomorphological and sedimentological
geomorphological, these disciplines have not been involved viewpoint. I recommend that potential spawning gravel
in many spawning gravel assessments; instead, such quality assessors consider the distinct requirements for
assessments are typically conducted by fish biologists. different life stages of salmonids. I also propose a step-by-
In an effort to provide useful measures for eval- step procedure for assessing spawning gravel quality and
uating gravels, much of the literature on spawning gravels has demonstrate the approach with a case study from the
concerned single-variable indices of gravel quality, that is, Colorado River and tributaries in Grand Canyon National
Park.

Attributes of Gravel Size Distributions

Email: [email protected]
Received November 4, 1997; accepted May 15. 1999 Natural streambed gravels consist of a mixture of
sizes. If silt and clay are present in the mixture,
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

FIGURE 1- Cumulative size distribution curves for spawning gravels available to rainbow trout Oncorhynchus mykiss, drawn
from the case study in the main -stem Colorado River (solid lines, three samples) and Nankoweap Creek (broken lines, five
samples), a tributary downstream of Glen Canyon Dam. Size descriptors (d 50 , d 84 , etc.) are obtained from a curve by reading the
grain size corresponding to the indicated percentile. In the example shown, a potential spawning gravel from Nankoweap Creek
has a d 84 of 20 mm and a d 50 of 6.8 mm (i.e., 84% of grains are smaller than 20 mm and 50% of grains are smaller than 6.8 mm).

particle diameter may range over five orders of magnitude. present in a given gravel, it is unwieldy to use them to
Many sediments (and sedimentary rocks) are characterized compare gravels, and it is impossible to present more than a
by larger particles that make up the structure of the deposit few similar distributions on the same graph because the lines
(the framework grains) with finer sediments filling the pore overlap and obscure characteristics of indiv idual size
spaces between the framework grains (the matrix). Some distributions. Size distributions can also be presented as
sediments contain so much matrix that most framework modified box-and-whisker plots (Tu-key 1977; Kondolf and
grains are not touching and thus are not carrying the weight Wolman 1993), which permit summarization of multiple
of the deposit; these are termed "matrix -supported" deposits distributions on the same graph without overlap (Figure 2).
(Williams et al. 1982). The threshold size between matrix
sediment and framework gravel should be a function of the To facilitate comparison among size distributions,
pore sizes in the framework. In a bimodal distribution, the we commonly develop statistics from the curves. For
distinction between framework and matrix may be example, the median particle diameter, d50, is commonly used
straightforward. Otherwise, defining the upper size limit of in hydrology, geomor-phology, and engineering as a measure
matrix sediment may be arbitrary. of central tendency of the distribution because it is easily
The range of sizes present in natural gravels is read from distributions and unambiguously interpreted
typically presented in cumulative size distribution curves (Inman 1952; Vanoni 1975). Also commonly reported are the
(Figure 1). Although these cumulative size distribution d 16 and d 84 , the sizes at which 16% and 84% of the sample,
curves provide complete information on the range of sizes respectively, are finer. The range of sizes in natural gravels is
 KONDOLF

FIGURE 2.Box-and-whisker plots for rainbow trout spawning gravels from the case study in the Colorado River and
tributaries downstream of Glen Canyon Dam, and averages for other rainbow trout spawning gravels. (Summary numerical
values are given in Table 2). For each sample, the rectangle (box) encompasses the middle 50% of the sample, from the d25, to
d75 values, termed the "hinges." The median diameter, d 50 , is represented by a horizontal line through the box. Above and below
the box are lines (whiskers) extending to the d 90 and d10 values, a modification of the standard box-and-whisker plot of Tukey
(1977). Numbers below the plots refer to samples in Kondolf et al. (1989). Box-and-whisker plots are easier to read than
cumulative size distribution curves when several similar distributions are plotted on the same graph.

so great that data are usually log-transformed (or plotted on is skewed off a normal or lognormal distribution. It is
log-transformed graph paper). Gravel size distributions tend commonly calculated as the geometric skewness coefficient
to resemble log-transformed normal, gamma, or Weibull sk = [[log10 (d g /d 50 )] /[log10 (sg )] (Inman 1952). Gravel size
distributions rather than untransformed normal distributions distributions tend to be positively skewed, whereas log-
(Kondolf and Adhikari, in press). In lieu of an arithmetic transformed distributions (as reflected in the values of sk)
mean, sedimentologists have used the geometric mean, dg = tend to be negatively skewed, which is reflected in the
(d 84 * d16 ) 0.5 (Inman 1952), which is another measure of tendency of geometric mean diameters to be less than
central tendency, but one more influenced by extremes of the median diameters (Kondolf and Wolman 1993).
distribution than the median. Gravel Requirements of Salmonids
Other commonly reported attributes of size dis- The spawning gravel requirements of salmonids
tributions are sorting and skewness. Sorting, or dispersion, differ during redd construction, incubation, and emergence
refers to the degree to which fluvial processes have collected (Figure 3). The spawning female must be able to move
similarly sized particles together. In downstream reaches of gravels to excavate a depression in the bed to create the redd.
larger river systems, some deposits may be entirely of gravel, Fish need not move all rocks present (some larger particles
others entirely of sand. These deposits would be considered can remain unmoved as a lag deposit), but most of the
"well sorted" with low dispersion. Sorting is commonly particles present must be movable or the redd cannot be
reported as the geometric sorting coefficient, sg = (d84/d16)0.5, excavated. Thus, most framework grains should be movable,
which increases with dispersion (and thus decreases with a requirement that effectively sets an upper size limit to
sort ing). Skewness refers to how much the distribution suitable spawning gravels. Larger fish are capable of moving
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

FIGURE 3 - Flow chart showing gravel requirements of salmonids during redd construction, incubation, and emergence. The
intragravel flow equation is denned in Figure 5.

larger rocks, so this upper size limit varies with fish size
(Figure 4) (Kondolf and Wolman 1993).
Human impacts may also affect spawning habitat.
Trapping of gravel in reservoirs and release of clear water
downstream may cause the winnowing of smaller, mobile
grains from beds below dams, leaving only progressively
coarser particles. This process, termed armoring, may result
in gravels becoming too coarse for use by spawning salmon,
as documented on the Sacramento, Shasta, and Klamath
rivers in California (Parfitt and Buer 1980; Buer et al. 1981).
For successful incubation, gravel must be suf-
ficiently free of fine sediment that the flow of water through
the gravel is adequate to bring dissolved oxygen (DO) to
eggs and carry off metabolic wastes (see discussions in
Chevalier et al. 1984 and Groot and Margolis 1991). St udies
relating intragravel water properties to emergence success
FIGURE 4.Median diameter (d50 ) of spawning gravel indicate that minimum levels of DO necessary for survival
plotted against body length of a spawning salmonid. Solid vary (with temperature, in part), but generally fall between 2
squares denote samples from redds: open triangles are and 8 mg/L (Alderdice et al. 1958; Coble 1961; Shumway et
"unspawned gravels," which are potential spawning gravels al. 1964; Silver et al. 1965; Davis 1975;
sampled from the undisturbed bed near redds. (Modified
from Kondolf and Wolman 1993.)
 KONDOLF

Chevalier et al. 1984). Other studies have shown that blamed for reducing permeability are finer than 1 mm
interstitial fine sediment can reduce gravel permeability and (McNeil and Ahnell 1964; Cederholm and Salo 1979; Tagart
lead to less intragravel flow, which can result in lower levels 1984). Thus, emergence requirements set another limit to
of DO and suffocation of embryos (McNeil and Ahnell 1964; interstitial fine sediment, but of a coarser caliber than those of
Cooper 1965; Koski 1966; Chevalier et al. 1984). Thus, for concern for incubation.
successful incubation, the lower limits of acceptable Laboratory and field researchers have attempted to
spawning gravel size are denned not by framework size, but relate fine sediment content to incubation and emergence
by the amount of interstitial matrix present (and its effect on success, producing a wide range of results (Table 1). In a
permeability). comprehensive and influential review. Chapman (1988)
Chinook salmon Oncorhynchus tshawytscha (and suggested that this variability resulted from poor
some other salmonids) have been observed to preferentially understanding of the structure of the egg pocket (the small
spawn where stream water down-wells into the gravel bed area within the redd containing the eggs) and argued for in-
(e.g., Vronskiy 1972); chum salmon 0. keta (and some other tensive studies of egg pockets. Such studies would no doubt
species) often spawn where water upwells from the gravel prove helpful in better understanding processes within the
bed into the water column (e.g., Tautz and Groot 1975). As redd; however, study results might have only limited direct
emphasized by Healey (1991), the absence of downwelling or application to the common problem of evaluating the
upwelling currents may be an important reason why suitability of potential spawning gravels because, by
spawning fish do not use many seemingly excellent spawning definition, no egg pockets yet exist to be sampled. In a
gravels (e.g., Burner 1951). thoughtful comment. Young et al. (1990) noted that
Dye studies in the field and laboratory have con- variations in female fecundity and egg viability can affect the
firmed that irregularities in the bed profile tend to promote results of relations between egg survival and gravel size.
exchanges of water between the stream and the interstices of
the gravel bed (Cooper 1965; Vaux 1968). These patterns can Comparability of Assessment Methods and
be explained by a fundamental equation of groundwater flow, Attributes
Darcy's Law, which states that the rate of groundwater flow
(or Darcy velocity, V) is the product of the permeability (or Studies relating gravel size to successful spawning
hydraulic conductivity, K) and the hydraulic gradient dh/dl or survival to emergence are useful only to the extent that the
(Figure 5) (Freeze and Cherry 1979). The lower elevation of data can later be applied to gravels elsewhere. Similarly, to
the water surface in the riffle creates a hydraulic gradient that assess gravel quality at a new site, we must be able to apply
induces downwelling at the tail of the pool. The redd mound relations between gravel size and spawning derived
(or tailspill) produces a similar effect at a smaller scale, elsewhere. This transfer of information cannot be effected
inducing inflow of stream water into the mound. (Darcy's law without comparability in methods of sampling and reporting
also illustrates the importance of the matrix sediment, of data. When full size distributions are reported, subsequent
because it affects the hydraulic conductivity, K). workers can compute a statistic of choice for comparison with
After hatching, alevins live in the intragravel results elsewhere. When only one (or a few) summary statis-
environment for a period, then migrate through t he gravel to tics are reported, comparisons are impossible un less the same
the surface. Successful emergence requires connected pore statistics have been reported in all studies.
space through which the alevins can pass. Field and Comparability also requires recognition of the
laboratory studies have demonstrated that, in some gravels, distinction between redd gravels and potential spawning
although eggs may incubate successfully and alevins hatch gravels being sampled to determine their suitability. As
and live in the intragravel environment, alevins cannot females construct redds, they win now fine sediment from the
migrate upward to the surface because fine sediment blocks gravel. The gravel within the redd typically has less fine
intragravel pore spaces (e.g., Phillips et al. 1975; Hawke sediment than it did before redd construction (Figures 6, 7).
1978). The sediment sizes held responsible for blocking The reduction in fine sediment during spawning depends
emergence are typically between 1 and 10 mm (Bjornn 1969; largely on the amount of fine sediment initially present, and
Phillips et al. 1975; Harshbarger and Porter 1982), and those the reduction can in some cases transform unsuitable
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

FIGURE 5.Diagram of groundwater flow through the tail of a pool. The lower elevation of the water surface in the riffle
creates a hydraulic gradient that induces downwelling at the tail of the pool; V is Darcy velocity and K is hydraulic
conductivity. Vertical scale is greatly exaggerated. (From Keller and Kondolf 1990.)

gravels into suitable gravels (Kondolf et al. 1993). distribution is commonly sim ilar to that of the transported
Montgomery et al. (1996) have suggested that mass spawning bed load (Parker and Klingeman 1982). The framework
may change sediment characteristics and bed form such that grains of the surface are generally not larger than those of the
the bed is less subject to scour. underlying sediment, but the surface layer is typically
Laboratory and field emergence studies attempt to deficient in the finer fractions of the distribution. In part, this
represent conditions in redds, so before relations from these can be explained by selective transport of finer grains
studies are applied to potential spawn ing gravels, the fine exposed on the surface at flows too low to mobilize the entire
sediment content of the potential spawning gravels should be bed. However, some coarse surface layers are active features
adjusted for the probable cleaning effect of spawning. in that they persist (or reform) despite frequent mobilization
Moreover, as noted by Chapman (1988). the redd structure of of the bed. By contrast, some coarse surface layers are
coarse lag gravels encountered in many redds may not be inactive, being mobilized only by infrequent flows,
reflected in the homogenized sediment mixtures typically developing downstream of dams or in other situations of
used in laboratory studies. decreased sediment supply. Parker and Klingeman (1982)
Sampling Spawning Gravels termed the active surface layers pavements and the inactive
Various techniques have been used to sample ones armor layers, whereras Gomez (1984) argued that the
spawning gravels, and they range widely in the effort and terms should be used in the opposite sense.
cost required to use them. Most sampling methods involve The potential paucity of interstitial fine sediment
obtaining a gravel sample, which is then passed through a in the surface layer implies that framework size can be
series of sieves to determine the proportions of the sample in estimated by sampling the surface layer, but matrix
various size classes. The more expensive and seemingly assessment requires subsurface sampling.
sophisticated techniques are not necessarily better. The
selection of sampling technique should be driven by the Exclusion of Large Rocks
purpose of the study, adequacy of sample size, and Many workers have excluded large rocks from
comparability of results. their gravel samples, because individual large rocks can
Sampling methods for gravels (and specifically constitute a large percentage of the total sample weight and
for spawning gravels) have been described in detail by thus might "bias" the distribution. There may be arguments
various authors, including Kellerhals and Bray (1971), Lisle for excluding rocks above some threshold size when only
and Eads (1991), and Young et al. (1991). Here, I briefly fine sediment content is compared (Church et al. 1987), but
review some popular sampling methods after considering the complete size distribution (including large rocks) should
issues of surface versus subsurface layers, exclusion of large be reported to permit assessment of framework size as well.
rocks, and sample size. Exclusion of large rocks reduces the coarser fraction of the
Surface versus Subsurface Layers sample and thus increases the remaining finer fractions as a
The surface layer of gravel on river beds (here percentage of the total sample. At the very least, the decision
defined as the depth of one grain diameter, d84 ) is typically to exclude large rocks from the sample should be reported;
coarser than the underlying, subsurface layers, whose size this has not always been done, casting doubt on the
comparability of some studies.
 KONDOLF

TABLE 1.Fine sediment percentages corresponding to 50% emergence of salmonids in various studies.
Maximum percentage of grains finer than:

Reference or statistic Speciesa 0.83 mm 2.0 mm 3.35 mm 6.35 mm 9.5 mm

Hausle and Coble (1976) Brook trout 10

Weaver and White (1985) Bull trout 16,40
Bjornn (1969) Chinook salmon 15, 26
Tappel and Bjornn (1983) Chinook salmon 40
McCuddin (1977) Chinook salmon 30, 35
Koski (1975, 1981) Chum salmon 27
Cederholm and Salo (1979) Coho salmon 7.5, 17
Koski (1966) Coho salmon 21 30
Phillips et al. (1975) Coho salmon 36
Tagart (1984) Coho salmon 11
Irving and Bjomn (1984) Cutthroat trout 20
Irving and Bjornn (1984) Kokanee 33
Irving and Bjornn (1984) Rainbow trout 30
NCASI (1984) Rainbow trout 12 40
Bjornn (1969) Steelhead 25
Tappel and Bjornn (1983) Steelhead 39
McCuddin (1977) Steelhead 27
Phillips et al. (1975) Steelhead 25
Mean 13.7 10.0 29.5 30.3 28.0
SD 4.7 0.0 4.2 7.4 12.0

a
Scientific names: brook trout Salvelinus fontinalis; bull trout S. confluentus; chinook salmon Oncorhynchus tshaw-ytscha: chum salmon 0. keta;
coho salmon 0. kisutch; cutthroat trout 0. clarki; kokanee 0. nerka; rainbow trout (nonadromous) and steelhead (anadromous) 0. mykiss.

Sample Size to obtain truly representative samples, and proposed a simple

Adequate sample size increases with particle size. rule that the largest particle should not constitute more than
Church et al. (1987) noted that "the largest class of grains 1% of the total sample weight. The pebble count method
present in the sample should define the sample size since (described below) was proposed by Wolman (1954) as an al-
they will be the fewest in number, hence least well ternative to large bulk samples for estimating sur face grain
represented." They reviewed sample size requirements, noted size distributions; however, Wolman's principal interest was
that for typical river gravels, more than 200 kg are required in estimating grain roughness, not in determining interstitial

FIGURE 6.Percentage of sediment finer than 1 mm in FIGURE 7.Percentage of sediment finer than 4 mm from
redds and potential (comparable, unspawned) gravels. The pairs of redd and potential spawning gravel sampled by
data point for Evans Creek is excluded from the regression. Chambers et al. (1954) (squares) and by Kondolf et al. (1993)
(See Kondolf et al. 1993 for sources of data.) (triangles).
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY
fine sediment content.
Many trout redds, especially those constructed in usually shortened to permit the operator's arms to reach the
pocket gravels in steep channels, do not contain enough bottom of the sampler), have been used (e.g., Chambers et al.
gravel to satisfy sample size requirements, which poses a 1954, 1955; Orcutt et al. 1968). Other variants of cylindrical
fundamental problem in attempting to apply minimum size core samplers have included a 60-cm length of 46-cm-
requirements to sampling spawning gravels. In such cases, diameter well casing with a serrated lower edge and handles
obtaining as large a size as possible from the site is probably attached to the top (Horton and Rogers 1969), a 53-cm length
the best approach. of 35-cm-diameter pipe with a serrated lower edge (W. F.
Van Woert and E. J. Smith, California Department of Fish
Sampling Methods and Game, unpublished), a 25-cm-diameter Hess-type core
Pebble counts and visual estimates provide a sampler (Shirazi et al. 1981), a 25-cm-diameter "bottomless
measure of the surficial grain size only, and cannot measure bucket" (Kondolf et al. 1989), and a 75-cm length of 15-cm-
fine sediment content of the subsurface gravel. Visual diameter galvanized stove pipe (Peterson 1978).
estimates ("ocular assessments") are widely employed by fish The most popular bulk core sampler among fish
biologists and are typically used as input to the PHABSIM biologists has been the FRI or McNeil sampler, constructed
fish habitat model (Bovee 1982). However, there is no ev- from a 50-cm drum with a 15- to 30-cm-diameter pipe
idence that these subjective estimates of percentages of welded on the bottom. The smaller pipe is worked into the
various size classes in the bed are reproducible among bed, the gravel is removed by hand, and the muddy water
different investigators. Moreover, the results are usually within the sampler is retained to sample suspended fine sed-
reported in the form of "dominant" and "subdominant" size- iments (McNeil and Ahnell 1964). Geomorphol-ogists have
classes or as percentages of classes such as "80% cobble, used bottomless 50-cm oil drums in various forms to obtain
10% sand, and 10% silt." Even if these estimates are sufficiently large samples, such as the 140-240-kg samples
accurate, they are not reported in a form that can be readily collected by Wilcock et al. (1996). The "cookie-cutter"
compared with sediment sizes reported in the engineering sampler is a 50-cm drum sampler with an underwater sample
and geomorphic literature, in which statistics are drawn from box that has a screen to collect fine material washed
standard size distributions. downstream (Klingeman and Emmett 1982), and the "barrel"
The pebble count method (Wolman 1954; Kon- sampler is a 46-cm drum sampler fitted with a 152-cm-long
dolf 1997) involves measurement of the diameter of 100 hood of filter mesh to collect fine sediment (Milhous et al.
stones randomly selected from specific geomorphic features 1995). When gravel is removed from drum samplers, it is
on the bed surface. Pebble counts provide reproducible possible to remove the surface layer first and analyze it
surface grain size distributions and can be readily adapted for separately. Curtin (1978) fitted a hood on a shovel to retain
use in fish habitat studies as an alternative to visual estimates fine sediment. In the event the gravel is exposed on a bar
(Kondolf and Li 1992). A recent modification, the zigzag (usually not t he case for spawning gravels), it can be easily
count (Bevenger and King 1995), should be avoided because sampled by shovel or backhoe.
it mixes sample points from many different channel features Freeze core sampling involves driving steel
(i.e., this method would typically mix data from spawning probes into the bed, discharging a cooling agent (such as
riffles, intervening pools, and banks), does not yield adequate liquid CO2 or nitrogen) into the probes to freeze the
sample sizes for individual populations of gravel, and does interstitial water adjacent to the probe, and withdrawing the
not yield reproducible size distributions. Thus, the zigzag probes (with gravel samples frozen to them) from the bed
count (and similar modifications) are not true pebble counts with a tripod-mounted winch (Everest et al. 1980). The
and are not good methods for assessing spawnin g gravel method was developed to obtain gravel samples that pre-
quality (Kondolf 1997). served vertical stratification of the sediments, although
Bulk core sampling involves driving a cylindrical laboratory experiments have shown that driving the probes
core sampler into the bed and removing (by hand) the into the bed can disrupt the existing stratification (Beschta
material within it down to a predetermined depth. Drums 50 and Jackson 1979). Freeze core samples tend to have a
cm in diameter, with the top and bottom removed (and "ragged edge" with larger particles protruding from the
 KONDOLF

frozen mass, implying that all fractions of the distribution are blocking fry emergence through intragravel pores (Everest et
not sampled proportionately. Most importantly, however, al. 1987).
freeze core samples are typically less than 10 kg, too small to The use of "percent fines" in the fish biology
accurately represent gravels that include particles 64 mm and literature originated with a study relating incubation success
greater (Church et al. 1987). to gravel size by McNeil and Ahnell (1964), who found
Bulk core sampling is simple (although labor- incubation success was inversely related to the percentage
intensive), can yield large samples, and does not suffer from finer than 0.83 mm. The value of 0.83 mm was an arbitrary
the "ragged edge" of freeze core sampling. In a comparison of cutoff, simply an artifact of the set of Tyier sieves used in the
shovel, bulk core (McNeil), and freeze-core sampling, Young study. It is not a physically significant threshold, nor does it
et al. (1991) found that the bulk core samples most frequently correspond to a break in size-classes on the standard
approximately the true substrate composition. Wentworth scale (Vanoni 1975). However, many subsequent
Standpipes can be used to directly measure per- authors apparently accepted 0.83 (or 0.85) mm as a physically
meability, intragravel flow velocity, and DO in situ. The most significant size threshold. It is preferable to use sieves sized
widely used type is the standpipe of Terhune (1958) or later in whole mm, and to round 0.83 or 0.85 mm to 1 mm.
variants, such as the substitution of stainless steel for Tappel and Bjornn (1983) proposed that the
aluminum by Bar-nard and McBain (1994). The pipe, quality of spawning gravel be assessed based on the
perforated to allow seepage of water from the adjacent gravel percentages finer than 0.83 mm and finer than 9.5 mm out of
into the pipe, is inserted into the gravel bed to the depth at the portion of the size distribution finer than 25.4 mm. This
which the permeability measurement is desired. Water is approach is an improvement over simple percent-fines
pumped from the standpipe, maintaining a constant hydraulic measures in that it offers a more complete description of the
gradient into the pipe, and the rate at which water flows into percentage of fine sediment, albeit one in which the influence
the pipe is measured and used as a basis to compute of framework particles greater than 25 mm is explicitly
permeability. The dissolved oxygen content of the inflowing ignored.
water can also be measured. The percentage of fine sediment below a given
size is influenced not only by the amount of fine sediment,
Size Descriptors Proposed as Indices of but by the other sizes present as well, because it is simply a
Gravel Quality percentage of the total. Thus, exclusion of large grains
artificially increases the percentage of remaining, finer
Once obtained, gravel samples are usually sieved and fractions, and we could expect a given amount of fine
weighed to yield a size distribution. Because these size sediment to have different effects on permeability of gravels
distributions are unwieldy, statistics drawn (or computed) depending on framework size.
from the distributions have been used as indices of gravel
quality and as the independent variables against which the de- Geometric Mean Diameter and Fredle Index
pendent variables of incubation or emergence suc cess are In response to shortcomings in the "percent fines"
plotted in laboratory studies. Because these indices are so measure, Shirazi and Seim (1981) proposed the geometric
widely employed with so little discussion of their relation to mean diameter, d g , as an index of gravel quality and a
the complete size distribution, a review of the evolution of "unifying substrate statistic," because it reflected the
these indices and their attributes may be useful. complete size distribution and because emergence success in
laboratory experiments was found to be related to dg.
However, the experimental gravels differed principally in the
Percent Fines
fine sediment content added, so lower values of dg, reflected
In general, the literature suggests that interstitial sediments
greater fine sediment contents. Lotspeich and Everest (1981)
finer than about 1 mm reduce the permeability of the gravel
proposed the fredle index (fi), which combined a measure of
and can prevent intragravel flow from providing sufficient central tendency (d g ) with a measure of dispersion. The fredle
oxygen to embryos and removing metabolic wastes.
index is calculated as Fi = dg /S T, where ST is the Trask
Sediments in the 1-10-mm size range have been implicated in
sorting coefficient, given by ST = [(d75 )/(d25 )] 0.5 (Inman
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

1952), where d75 and d 25 are the sizes at which 75 and 25% of Gravel Quality Criteria Drawn from
the distribution, respectively, are finer. Emergence Studies
The geometric mean can be similar for very dif-
ferent gravel mixtures because a large d 84 can offset a small Compilation of Laboratory and Field Study
d 16 if it contains some very coarse framework gravels (and Results
thus a large d 84 ). A gravel mixture with a large content of fine Results of laboratory and field experiments of
sediment (and thus a small d 16 ) could have a d g similar to a incubation and emergence success have generally been
gravel mixture with little fine sediment. Large chinook presented as plots relating percent fine sediment content to
salmon have spawned in gravels with median diameters of percent successful incubation or emergence. Table 1 presents
over 50 mm, but they also spawn in smaller gravels (Kondolf fine sediment percentages corresponding to 50% emergence
and Wolman 1993). Thus, suitable spawning gravels (i.e., drawn from such plots from 2 field studies (Koski 1966;
movable by fish and free of fine sediment) could have very Tagart 1984) and 11 laboratory trough studies. The choice of
different values of d g solely because of different framework 50% emergence is arbitrary, but can be justified because
sizes, implying nothing about fine sediment content. redds with at least 50% emergence success would probably
As with the geometric mean, similar values of be considered as productive by most biologists. Moreover,
fredle index could derive from a wide range of different size the range of emergence reported in these studies always en-
distributions, so it is similarly unsuitable as a unifying compassed 50%, but not necessarily lower or high er
substrate statistic. The fredle index has other disadvantages emergence values (i.e., some studies had no emergence
as follows. values less than 20%, and some had none more than 80%).
1) There is no physical reason to expect a measure of However, it is worth noting that in some streams with
successful natural reproduction, emergence measured in
central tendency divided by a measure of dispersion
natural redds is considerably less than 50% (see NCASI 1984
(sorting) to yield a meaningful index of gravel
quality. The fredle index is effectively an inverse for a review).
coefficient of variation with dimensions of length. One of the most striking features of Table 1 is the
variation among studies in the definition of "fine sediment,"
which ranges from 0.83 mm to 9.5 mm. In some cases, "fine
2) The measure of central tendency, d g , is calculated sediment" was defined based on the sieve size that best
from (d 84 and d 16 and is thus influenced by the correlated to emergence; in other cases it was defined at the
extremes of the distribution. A more robust and outset of the study, and experimental gravel mixtures were
preferable measure of central tendency would be the prepared with varying percentages of sediment smaller than
median size, d 50 . this size. Gravel mixtures varied among studies, and some
size distribution curves were atypical for natural spawning
3) The measure of dispersion, S T is calculated from d75 gravels (Figure 8).
and d 25 and thus reflects the spread of only the
middle 50% of the distribution, and is thus Gravel Quality Criteria
insensitive to ecologically significant differences in Gravel quality criteria were inconsistent among
fine sediment contents less than 25%, these studies (and even among replicates of the same studies),
so to define precise thresholds for fine sediment content is
4) The fredle index has no physical reality, unlike the probably not justified. However, it is possible to generalize
median diameter or percentage finer than 1 mm. from these studies. The percentage finer than 1 mm (or 0.83
Because it is more complicated and harder to mm) was about 14% for 50% emergence, close to the
comprehend, it may appear more so phisticated than standard of 12% indicated by McNeil and Ah-nell (1964) and
(and thus preferable to) sim pler, more from extensive field observations by J. Cederholm
straightforward, descriptors of gravel size. (Washington Department of Nat ural Resources, personal
communication 1986).
Gravel quality is by nature complex, due to various meanings Results for the effect of coarser fine sediment on
of quality (Figure 3) and the natural complexity of sediments, emergence are less consistent. Values associated with 50%
so it is unreasonable to expect any single-variable descriptor emergence averaged about 30% for sediment finer than both
to be a good index.
 KONDOLF

FIGURE 8.Cumulative size distribution plots of some gravels used in emergence studies: (a) Phillips et al. (1975); (b) Tappel
and Bjornn (1983) and Irving and Bjornn (1984), mixtures 0:0 -50:20: (c) Cederholm and Salo (1979), 1975 experiments,
troughs 6-12; (d) Weaver and White (1985). (From Kondolf 1988.)

3.35 mm and 6.35 mm. (We might expect that more of the Changes in Gravel Size over Time
coarser fine sediment could be present before negatively af- Gravel size can change seasonally and from year
fecting gravel quality, but in the artificial gravel mixtures to year, affecting the applicability of observed gravel sizes to
used in most of these studies, 3-mm and 6-mm sediment may actual conditions during incubation or emergence. The
have similar effects in blo cking pore space.) amount of interstitial fine sediment can increase during the
incubation period by infiltration into the redd (Carling and
Influence of Fish Size McCahon 1987; Sear 1993) or by scour and fill (Lisle 1989).
While the framework size movable by a fish will Thus, the timing of sediment transport in the channel in
depend on the size of the fish, the effect of fine sediment on relation to incubation of salmonid embryos is very important
gravel permeability should be a function of the physics of in determining spawning success. Timing may be especially
groundwater flow, which would be independent offish size. important with fine sediment inputs from human activities,
However, larger eggs (of larger fish) may require more because these may occur during low flows in the channel.
irrigation, potentially making them more sensitive to reduced Most naturally produced fine sediment enters the channel
permeability. For emergence, larger alevins may have more during high flows, when there is adequate stream power to
difficulty than smaller alevins in passing through intragravel transport and disperse it. However, anthropogenic sources
pore spaces decreased by interstitial fine sediments, but they (such as irrigation return flow) may occur during base flow,
may also be stronger (Phillips et al. 1975; Tappel and Bjornn when the fine sediment is more likely to settle out and in -
1983). filtrate.
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

FIGURE 9 - Flow chart illustrating nine discrete steps in evaluating salmonid spawning gravel quality.

The framework sizes of gravel may also undergo through sieves and weighed to obtain size distributions
changes, more likely on a longer timescale of years to (Vanoni 1975). In either case, the size distribution should be
decades, as a result of changes in coarse sediment supply or plotted as a cumulative frequency curve; to compare multiple
local shear stress. For example, at spawning areas distributions, box-and-whisker plots can be plotted from
downstream of dams, the bed may coarsen due to decreased percentile values drawn from the cumulative distributions.
supply of sand and gravel from upstream, such that size
distributions may no longer be valid several years after their Determine Whether Gravel Is Movable by
measurement. Similarly, channel straightening, levee con- Spawning Fish (Step 3)
struction, or upstream urbanization could increase local shear Whether the framework gravels are too large for
stress and thus lead to a coarser bed material. the fish to move can be determined by comparing the d50 or
d84 with those reported for the species elsewhere and with the
A Procedure to Assess Spawning Gravel maximum movable size predicted by Figure 4, which
Quality suggests that spawning fish can move gravels with a median
Gravel requirements should be considered sep- diameter up to about 10% of their body length. In some
arately for redd construction, incubation, and emergence, and channels, gravels may be compacted or cemented, rendering
gravel size distribution curves should be examined for otherwise suitable sizes unsuitable. No widely accepted or
information relevant to the specific requirements of these life easily applied method has been developed to quantify this
stages. As indicated in Figure 9 and discussed below, this phenomenon, so it should be evaluated qualitatively.
life-stage-specific approach can be broken down into nine
discrete steps. Determine Whether Fine Sediment Content Is
Excessive for Incubation (Steps 4-5).
Sample the Gravel and Develop a Size The question is whether the amount of sediment
Distribution (Steps 1-2) finer than 1 mm is so great that gravel permeability, and thus
The sampling method depends upon the purpose intragravel flow, is negatively affected. The percentage finer
of the assessment. If the concerns are limited to whether the than 1 mm should be drawn from the grain size distribution
fish can move the gravels, pebble counts may be adequate, curves and adjusted downward (using Figure 6) to reflect the
although such values (obtained from the surface layer) may probable cleaning effect of redd construction be fore fine
be larger than those from bulk samples, because the latter sediment content is evaluated.
would be influenced by interstitial fine sediment in the The resulting values can be compared with values
subsurface. More commonly, however, the fine sediment reported from redds elsewhere and with standards drawn
content is also of concern, in which case subsurface samples from laboratory and field studies of incubation and
must be obtained. Because of the drawbacks of freeze core emergence in Table 1 (showing values for 50% survival).
sampling discussed earlier, bulk core samples (of adequate They also can be evaluated against conclusions drawn from
size) are preferable. Pebble counts directly yield size dis- field observations by McNeil and Ahnell (1964) and
tributions, but bulk subsurface samples must be passed Cederholm and Salo (1979) that less than 12-14% of gravels
 KONDOLF

FIGURE 10.Location map of Colorado River and tributaries, Grand Canyon National Park. Arizona. Spawning gravel
sample locations (on Nankoweap, Clear. Crystal, Shinumo, and Tapeats creeks, and on the main stem) are shown in open
circles. Other important tributaries also shown for reference. (Adapted from Kondolf et al. 1989.)

should be finer than 1 mm (or 0.83 mm) for suc cessful Consider Changes in Gravel Size after Spawning (Step 8)
incubation. Potential changes in sediment yield and local
sediment transport capacity should be evaluated at the
Determine Whether Fine Sediment Content Is watershed scale to identify potential sources of fine sediment
Excessive for Emergence (Steps 6-7) during the incubation period and to evaluate the potential for
To assess whether the fine sediment will block the bed scour or coarsening. Field studies to monitor changes in
upward migration of fry, the percentage finer than 3, 6, or 10 fine sediment percentages over the course of the incubation
mm can be compared with values reported from redds season (Adams and Beschta 1980; Lisle and Eads 1991) may
elsewhere and with standards drawn from laboratory and be appropriate. Because the future applicability of gravel size
field studies of incubation and emergence. However, data collected may be compromised by long-term changes in
although the fine sediment (< 1 mm) threshold for incubation bed material size, monitoring of bed material sizes in future
effects can be estimated at 12-14%, the upper limits of the years may also be appropriate.
(larger) fine sediments affecting emergence (percentages less
t h a n 3-10 mm) are more difficult to select, showing Evaluate Intragravel Flow Conditions (Step 9)
considerable variability (Table1). Intragravel flow depends both on the gravel per-
As with the percentage of sediment less than 1 meability and the hydraulic gradient. The former is affected
mm, the percentages less than 3, 6, or 10 mm should be by fine sediment content and thus is partly addressed in steps
adjusted downward to reflect the probable cleaning effect of 4-5. The hydraulic gradient is more complex to evaluate
redd construction, but the effects of redd building on these because it depends on flow level, channel bed geometry, and
sizes are more variable than they are upon the percentage possibly on large-scale groundwater circulation patterns.
finer than 1 mm (Figure 7) (Kondolf et al. 1993). Standpipe measurements, dye studies, or examination of the
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

channel bed geometry could all be used to shed light on

intragravel flow conditions.

A Case Study: Assessing Spawning Gravels

for Rainbow Trout in the Colorado River and
Tributaries below Glen Canyon Dam
Case Study Site Description
Since closure of Glen Canyon Dam on the Col-
orado River in 1963, a popular sport fishery for rainbow trout
has developed in the Colorado River downstream in Grand
Canyon National Park. The fishery is especially productive in
the tailwater reach between the dam and the Paria River con-
fluence (Figure 10), where consistently cold-water releases
have produced a nearly complete change in species
composition from native warmwater fishes to introduced
rainbow trout. The trout spawn both in the main stem and in
tributaries. Spawning habitat is limited to some large main-
stem gravel bars and to gravel deposits in tributary reaches
downstream of migration barriers (Figure 11), many of which
are "pocket gravels" within bo ulder-dominated channels
(Figure 12). The quality of these spawning gravels had not
been assessed prior to the study described here.

Case Study Methods

As part of a larger research effort to examine the
fish resources (native and exotic) of the Colorado River

FIGURE 11.Lower reaches of Nankoweap Creek, looking

upstream from right bank, about 50 m upstream from the
confluence with the Colorado River. (Photograph by author,
December 1985.)

FIGURE 12.Pocket gravel on lower Nankoweap Creek, view from right bank. (Photograph by author, December 1985.)
 KONDOLF

TABLE 2 - Grain size statistics for rainbow trout spawning gravels reported in the literature and in the Colorado River
and tributaries (from Kondolf et al. 1989).
Partical size (mm) Geometric Geometric Percentage
sorting skew-ness of grains
coefficient coefficient
finer than
Fish Median geometric
size mean
Locationa Reference (cm) N (d50 ) (dg ) (sg ) (sk) 0.85 mm 3.40mm b

Lardeau River, Hartman and

British Columbia Galbraith (1970) 75 6 23.5 14.7 3.6 -0.37 6.3 15.2
North Fork Boise River, Idaho Platts et al. (1979) 30 45 20 12.4 6.5 -0.25 7 24
Missouri River, Montana Spoon (1985) 44 27 12.5 8.3 4.6 -0.27 11.1 ND
Beaver Creek, Montana Spoon (1985) 44 19 15 9.3 4.9 -0.3 9.8 ND
Colorado River, Arizona Kondolf et al.
Tributary, redds (1989) 40 9 33.4 24.5 3.04 -0.27 1.5 7.7
Tributary, unspawned 40 8 21.9 14.9 4.21 -0.25 6.5 17.8
Main stem, redds 45 2 10.5 5.6 5.4 -0.4 6.8 15.2
Main stem. unspawned 45 1 16 5.2 10.5 -0.47 15 25

a
See Figure 2 for box-and-whisker plots by site.
b
ND means "no data."

system in the Grand Canyon and the effects of fluctuating generally not at issue. The d 50 s were similar to those reported
flows upon them (Maddux et al. 1987), Kondolf et al. (1989) in the literature for rainbow trout elsewhere (Figure 2; Table
sampled trout redds and potential spawning gravels in the 2) and fall within the range of d50 s expected for these
Colorado River and tributaries below Glen Canyon Dam in spawning females, which average 40-45 cm long (Figure 4).
1985 (Figure 10). Main -stem gravels were sampled by shovel In assessing the fine sediment content, the po-
from three gravel bars exposed by low river levels, removed tential spawning gravels had less than 7% finer than 1 mm,
to the laboratory, dried, and sieved. Tributary samples were and the redds even less, values well below the standard from
obtained within a few hundred meters of the Colorado River laboratory studies and other values reported for rainbow trout
confluence with a 25-cm polyvinyl chloride bucket with the even before adjustments were made for the probable cleaning
bottom removed. The latter sampling method was dictated by effect of redd digging (Table 2). In the main stem, redd
logistics of the remote tributary sites, the small size of the gravels had less than 7% finer than 1 mm, but one sample of
trout redds, and the often limited extent of gravel deposits in potential spawning gravel had 15% finer than 1 mm. This
which they occurred. Samples were obtained from both redd would exceed the standard of 12%, but with the expected
and potential spawning gravels, dried by sun or campfire, and effects of spawning taken into account (Figure 6), the per-
sieved and weighed on site, except for subsamples of the fine centage of fine sediment in the redd gravels would be less
fraction retained for laboratory sieving. than 10%. Thus, the quality of these gravels was quite good.

Case Study Results Case Study Discussion

Cumulative size distributions for all samples are Although potential spawning gravels had good
reported in Appendix Table A.I, curves for main -stem and quality, their extent was limited. Tributary spawning gravels
Nankoweap Creek gravels are shown in Figure 1, and box- were limited because of the small size of the channels and the
and-whisker plots of all samples in Figure 2. These often patchy distribution of gravels. Some of these gravels
distributions illustrate the wider range of gravel sizes in the may be inaccessible at low river stage because of migration
tributaries, reflecting greater variability in hydraulic barriers. Main-stem gravels were limited at the time of
conditions. These distributions also illustrate the larger sampling (1985), and the extent of suitable gravel bars
percentage of fine sediments in the main -stem gravels, as re- probably continues to decrease and the grain size to coarsen
flected in the fine tails of the main -stem distributions. as smaller, mobile gravels are transported from the reach by
Because most samples were taken at or adjacent high flows without replacement from upstream. In repeated
to redds, the ability of the fish to move the gravels was visual observations over 2 years, Maddux et al. (1987) not-
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

ed large variations in the fine sediment content of tributary References

gravels, presumably reflecting changes wrought by flash Adams, J. N., and R. L. Beschta. 1980. Gravel bed com-
floods. Thus, repeated gravel sampling might be warranted position in Oregon coastal streams. Canadian Journal of
here. We did not evaluate intragravel flow conditions. Fisheries and Aquatic Sciences 37:1514-1521.
Alderdice, D. W., W. P Wickett, and J. R. Brett. 1958. Some
Conclusion effects of exposure to low dissolved oxygen levels on
The literature on spawning gravels contains much Pacific salmon eggs. Journal of the Fish eries Research
debate over the best single-variable descriptor for gravel Board of Canada 15:229-250.
quality (e.g., Lotspeich and Everest 1981; Shirazi and Seim Barnard, K., and S. McBain. 1994. Standpipe to determine
permeability, dissolved oxygen, and vertical particle
1981, 1982; Beschta 1982;
size distribution in salmonid spawning gravels. U.S.
Chapman 1988, 1990; Young et al. 1990), but there can be no Fish and Wildlife Service, Fish Habitat Relationships,
single statistic that measures all aspects of gravel quality. The Technical Bulletin Currents 15: 1-12, Washington,
gravel requirements of salmonids differ with life stage as the D.C.
role of gravel changes. Rather than seek a single index that Beschta, R. L. 1982. Comment on 1stream system evaluation
can capture all characteristics relevant to salmon spawning with emphasis on spawning habitat for salmonids' by
success, assessment of gravel quality is more profitably Mostafa A. Shirazi and Wayne K. Seim. Water
approached by recognizing that the appropriate measures Resources Research 18:1292-1295.
depend on the questions being asked. Beschta, R. L., and W. L. Jackson. 1979. The intrusion of
To determine if the fish can dig redds in the fine sediments into a stable gravel bed. Journal of the
gravel, the framework size is important and can be compared Fisheries Research Board of Canada 36:204 210.
with framework sizes of gravels utilized by the same sized Bevenger, G. S., and R. M. King. 1995. A pebble count
procedure for assessing watershed cumulative effects.
fish elsewhere. To determine if the gravel contains too much
U.S. Forest Service Research Paper RM-RP-319.
fine sediment, the percentage of fine sediment can be
Bjornn, T. C. 1969. Embryo survival and emergence studies.
compared with values for the species elsewhere and with Idaho Fish and Game Department, Federal Aid in Fish
threshold from laboratory studies, although the percentage of Restoration, Project F-49-R-7, Job 5, Job Completion
fine sediment measured in potential spawning gravels should Report. Boise.
be adjusted downward to account for the cleaning effect of Bovee, K. D. 1982. A guide to stream habitat analysis using
the spawning fish. the instream flow incremental methodology. U.S. Fish
When gravel sizes are reported, the full size dis- and Wildlife Service FWS/OBS-82/86. (Instream Flow
tribution should be included (or made readily available) so Information Paper 12.)
that later workers can independently calculate size Buer, K., R. Scott, D. Parfitt, G. Serr, J. Haney, and L.
descriptors of choice for purposes of comparison. Thompson. 1981. Salmon spawning enhancement
studies on northern California rivers. Pages 149 154
Acknowledgments in T. J. Hassler, editor. Proceedings: propaga tion,
I have benefited greatly from conversations on enhancement, and rehabilitation of anadromous
salmonid populations and habitat in the Pacific
these topics with M. G. Wolman, M. J. Sale, J. G. Williams,
Northwest symposium, October 1981. California
S. Li, K. Vyverberg, L. B. Leopold, and others. The
Cooperative Fishery Research Unit, Arcata.
manuscript was much improved by review comments of S. Burner, C. J. 1951. Characteristics of spawning nests of
Railsback, D. Chapman, and two anonymous reviewers. This Columbia River salmon. U.S. Fish and Wildlife Service
publication is based in part on work performed at Oak Ridge Fishery Bulletin 61:97-110.
National Laboratory in the Laboratory Graduate Participation Carting, P. A., and C. P. McCahon. 1987. Natural silta-tion
Program under contract DE-AC05-760R00033 between the of brown trout (Saimo trutta) spawning gravels during
U.S. Department of Energy and Oak Ridge Associated low-flow conditions. Pages 229-244 in J. F. Craig and
Universities. Manuscript preparation was supported in part J. B. Kemper, editors. Regulated streams: advances in
by the Beatrix Farrand Fund of the Department of Landscape ecology. Plenum Press, New York.
Architecture and Environmental Planning, and by the Center Cederholm, C. J.. and E. 0. Salo. 1979. The effects of
for Environmental Design Research, both at the University of logging road landslide siltation on the salmon and trout
California, Berkeley. spawning gravels of Stequaleho Creek and the
Clearwater River basin, Jefferson County, Washington,
1972-1978. University of Washington, Fisheries
 KONDOLF

Research Institute. Report FRI -UW-7915, Seattle.

Chambers. J. S.. G. H. Alien, and R. T. Pressey. 1955. surfaces: some comments. Earth Surface Processes and
Research relating to study of spawning grounds in Landforms 9:19-24.
natural areas. Annual report to U.S. Army Co rps of Groot. C., and L. Margolis. 1991. Pacific salmon life
Engineers, Contract DA-35026-Eng-20572. (Available histories. University of British Columbia Press.
from Washington State Department of Fish eries, Vancouver.
Olympia.) Harshbarger, T J.. and P. E. Porter. 1982. Embryo sur vival
Chambers, J. S. R. T. Pressey, J. R. Donaldson. and W. R. and fry emergence from two methods of planting brown
McKinley. 1954. Research relating to study of trout eggs. North American Journal of Fisheries
spawning grounds in natural areas. Annual report to US Management 2:84-89.
Army Corps of Engineers, Contract DA-35026-Eng- Hartman. G. F, and D. M. Galbraith. 1970. The reproductive
20572. (Available from State of Wash ington. environment of the Gerrard stock rainbow trout. British
Department of Fisheries. Olympia.) Columbia Department of Recreation and Conservation,
Chapman. D. W. 1988. Critical review of variables used to Fisheries Research Section, Fish-cries Management
define effects of fines in redds of large salmonids. Publication 15, Victoria.
Transactions of the American Fisheries Society 1 17: Hausle, D. A.. and D. W. Coble. 1976. Influence of sand in
1-21. redds on survival and emergence of brook trout
Chapman, D. W. 1990. Response to Young et al. 1990. (Salvelinus fontinalis). Transactions of the American
Transactions of the American Fisheries Society 119: Fisheries Society 105:5763.
160-162. Hawke, S. P. 1978. Stranded redds of quinnat salmon in the
Chevalier, B. C. C. Carson, and W. J. Miller. 1984. Report of Mathias River, South Island. New Zealand. New
engineering and biological literature pertaining to the Zealand Journal of Marine and Freshwater Research
aquatic environment: with special emphasis on 12:167-171.
dissolved oxygen and sediment effects on salmonid Healey, M. C. 1991. Life history of chinook salmon. Pages 3)
habitat. Colorado State University. Department of 1-394 in C. Groot and L. Margolis. editors. Pacific
Agriculture and Chemical Engineering. ARS Project salmon life histories. University of British Columbia
5602-20813-008A, Fort Collins. Press, Vancouver.
Church. M. A., D. G. McLean, and J. F. Wolcott. 1987. Horton. J. L.. and D. W. Rogers. 1969. The optimum
River bed gravels: sampling and analysis. Pages 43-79 streamflow requirements for king salmon spawning in
in C. R. Thorne, J. C. Bathurst, and R. D. Hey. editors. the Van Duzen River, Humboldt County, California.
Sediment transport in gravel bed rivers. Wi-ley, New California Department of Fish and Game, Water
York. Projects Branch. Administrative Report 69-2,
Coble, D. W. 1961. Influence of water exchange and Sacramento.
dissolved oxygen in redds on survival of steelhead trout Inman, D. L. 1952. Measures for describing the size
embryos. Transactions of the American Fish eries distribution of sediments. Journal of Sedimentary
Society 90:469-474. Petrology 22:125-145.
Cooper, A. C. 1965. The effect of transported stream Irving, J. S.. and T. C. Bjornn. 1984. Effects of substrate size
sediments on the survival of sockeye and pink salmon composition on survival of Kokanee salmon and
eggs and alevins. International Pacific Salmon cutthroat and rainbow trout embryos. U.S. Forest
Committee, Bulletin 18. New Westminster. British Service Intermountain Forest and Range Experiment
Columbia. Station, Cooperative Agreement 12-11-204-1 1.
Curtin. G. C. 1978. A tubular-scoop sampler for stream Supplement 87, Completion Report, Boise. Idaho.
sediments. Journal of Geochemical Exploration 10:
Keller. E. A.. and G. M. Kondolf. 1990. Groundwater and
193-194.
Davis. J. C. 1975. Minimal dissolved oxygen requirements of fluvial processes: selected observations. Geological
aquatic life with emphasis on Canadian species. Journal Society of America Special Paper 252:319-340.
of the Fisheries Research Board of Canada 32:2295- Kellerhals, R.. and D. I. Bray. 1971. Samplingprocedures for
2332. coarse fluvial sediments. Journal of the Hydraulic
Everest, F. L., R. L. Beschla. J. C. Scrivener. K. V. Koski. J. Division 97:1165-1179. (American So ciety of Civil
R. Sedell, and C. J. Cederholm. 1987. Fine sediment Engineers. New York.)
and salmonid productiona paradox. Pages 98-142 in Klingeman. P. C.. and W. W. Emmett. 1982. Gravel bed-load
E. 0. Salo and T W. Cundy. editors. Streamside transport processes. Pages 141-179 in Hey et al..
management: forestry and fishery interactions. editors. Gravel bed rivers. Wiley. Chichester, UK.
University of Washington. College of Forest Resources. Kondolf. G. M. 1988. Salmonid spawning gravels: a
Seattle. geomorphic perspective on their distribution, size
Everest, F. L.. C. E. McLemore, and J. F. Ward. 1980. An
modification by spawning fish. and application of
improved tri-tube cryogenic gravel sampler. U.S. Forest
criteria for gravel quality. Doctoral dissertation. Johns
Service Research Note PNW -350.
Freeze, R. A., and J. A. Cherry. 1979. Groundwater. Prentice Hopkins University, Baltimore, Maryland.
Hall. Inc.. Englewood Cliffs, New Jersey. Kondolf. G. M. 1997. Application of the pebble count:
Gomez, B. 1984. Typology of segregated (armor/paved) reflections on purpose, method, and variants. Journal
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

of the American Water Resources Association

(formerly Water Reso urces Bulletin) 33:79-87. 1995. Sampling river-bed material: the barrel sampler.
Kondolf. G. M., and A. Adhikari. In press. Weibull vs. Rivers 5:239-249.
lognormal distributions for fluvial gravels. Journal of Montgomery, D. R.. J. M. Buffington, N. P. Peterson, D.
Sedimentary Research. Schuett-Hames, and T. P. Quinn. 1996. Stream-bed
Kondolf. G. M., S. S. Cook. H. R. Maddux, and W. R. scour, egg burial depths, and the influence of salmonid
Persons. 1989. Spawning gravels of rainbow trout in the spawning on bed surface mobility and embryo survival.
Grand Canyon. Arizona. Journal of the Arizona-Nevada Canadian Journal of Fishery and Aquatic Sciences
Academy of Science 23:19-28. 53:1061-1070.
Kondolf. G. M.. and S. Li. 1992. The pebble count technique NCASI (National Council of the Paper Industry for Air and
for quantifying surface bed material in instream flow Stream Improvement). 1984. A laboratory study of the
studies. Rivers 3:80-87. effects of sediments of different size characteristics on
Kondolf. G. M.. M. J. Sale and M, G. Wolman. 1993. survival of rainbow trout (Salmo gairdneri) embryos to
Modification of gravel size by spawning salmonids. fry emergence. NCASI, Technical Bulletin 429.
Water Resources Research 29:2265-2274. Orcutt, D. R., B. R. Pulliam, and A. Arp. 1968. Char-
Kondolf, G. M.. and M. G. Wolman. 1993. The sizes of acteristics of steelhead trout redds in Idaho streams.
salmonid spawning gravels. Water Resources Research Transactions of the American Fisheries Society 97: 42-
29:2275-2285. 45.
Koski, K. V. 1966. The survival of coho salmon Parfitt, D., and K. Buer. 1980. Upper Sacramento River
(Onchorhynchus kisutch) from egg deposition to emer- spawning gravel study. California Department of Water
gence in three Oregon coastal streams. Master's thesis. Resources, Northern Division, Red Bluff.
Oregon State University. Corvallis. Parker, G., and P. C. Klingeman. 1982. On why gravel bed
Koski, K. V. 1975. The survival and fitness of the two stocks streams are paved. Water Resources Research 18:1409-
of chum salmon (Oncorhynchus keta) from egg 1423.
deposition to emergence in a controlled stream Peterson, R. H. 1978. Physical characteristics of Atlantic
environment at Big Beef Creek. Washington. Doctoral salmon spawning gravels in some New Brunswick
dissertation. University of Washington. Seat tle. streams. Fisheries and Oceans Canada, Biological
Koski. K. V. 1981. The survival and quality of two stocks of Station. Fisheries and Marine Service Technical Report
chum salmon (Oncorhynchus keta) from egg deposition 785. St. Andrews. New Brunswick.
to emergence. Rapports et Proces-Verbaux des Phillips. R. W.. R. L. Lantz. E. W. Claire. and J. R. Moring.
Reunions Conseil International pour I'Exploration dc la 1975. Some effects of gravel mixtures on emergence of
Mer 178:330-333. coho salmon and steelhead trout fry. Transactions of the
Lisle. T. E. 1989. Sediment transport and resulting deposition
American Fisheries Society 104: 461-466.
in spawning gravels, north coastal California. Water
Resources Research 25:1303-1319. Platts. W. S., M. A. Shirazi. and D. H. Lewis. 1979.
Lisle, T. E.. and R. E. Eads. 1991. Methods to measure Sediment particle sizes used by salmon for spawning
sedimentation of spawning gravels. U.S. Forest Service. with methods for evaluation. U.S. Environmental
Research Note PSW -411. Protection Agency. Report EPA 600/3 -79-043.
Corvallis. Oregon.
Lotspeich, F. B.. and F. H. Everest. 1981. A new method for
reporting and interpreting textural composition of Sear, D. A. 1993. Fine sediment infiltration into gravel
spawning gravel. U.S. Forest Service. Research Note spawning beds within a regulated river experiencing
PNW -369. floods: ecological implications for salmonids. Reg-
Maddux. H. R., D. M. Kulby. J. C. deVos, Jr., W. R. Persons. ulated Rivers: Research and Management 8:373-390.
R. Staedicke. and R. L. Wright. 1987. Effects of varied Shirazi. M. A., and W. K. Seim. 1981. Stream system
flow regimes on aquatic resources of Glen and Grand evaluation with emphasis on spawning habitat for
canyons. U.S. Department of Interior, Bureau of salmonids. Water Resources Research 17:592-594.
Reclamation. Contract 4-AG-40-01810, Final Report. Shirazi, M. A., and W. R. Seim. 1982. Reply. Water
Washington, D.C. Resources Research 18:1296-1298.
McCuddin. M. E. 1977. Survival of salmon and trout Shirazi, M. A., W. K. Seim. and D. H. Lewis. 1981.
embryos and fry in gravel-sand mixtures. Master's Characterization of spawning gravel and stream system
thesis. University of Idaho. Moscow. (Not seen; cited in evaluation. Pages 227-278 in Salmon-spawning gravel:
Chapman 1988) a renewal resource in the Pacific Northwest?
McNeil. W. J., and W. H. Ahnell. 1964. Success of pink Washington State University, State of Washington
Water Research Center. Report 39, Pullman.
salmon spawning relative to size of spawning bed
materials. U.S. Fish and Wildlife Service. Special Shumway, D. L.. C. E. Warren, and P. Duodoroff. 1964.
Scientific Report -Fisheries 469. Influence of oxygen concentration and water movement
Meehan. W. R. 1991. Influences of forest and rangeland on the growth of steelhead trout and coho salmon
embryos. Transactions of the American Fisheries
management on salmonid fishes and their habitats.
American Fisheries Society. Special Publication 19. Society 93:342-356.
Bethesda. Maryland. Silver, J., C. E. Warren, and P. Duodoroff. 1965. Dissolved
Milhous. R. T, S. A. Hogan, S. R. Abt. and C C. Waison. oxygen requirements of developing steel-
 KONDOLF

head and chinook salmon embryos at different water Vaux, W. G. 1968. The flow and interchange of
velocities. Transactions of the American Fisheries water in a streambed. U.S. Fish and Wildlife
Society 92:327-343. Service Fishery Bulletin 66:479-489.
Spoon, R. L. 1985. Reproductive biology of brown and Vronskiy, B. B. 1972. Reproductive biology of the
rainbow trout below Hauser Dam, Missouri River, Kamchatka River chinook salmon
with reference to proposed hydroelectric peaking. [Oncorhynchus tschawytscha (Walbaum)].
Master's thesis. Montana State University, Boze-man. Journal of Ichthyology 12:259-273.
Tagart, J. V. 1984. Coho salmon survival from egg de - Weaver, T. M., and R. G. White. 1985. Coal Creek
position to fry emergence. Pages 173-181 in J. M. fisheries monitoring study no. III. Report to
Walton and D. B. Houston, editors. Proceedings of U.S. Forest Service, Flathead N. F.
the Olympic Wild Fish Conference, Port Angeles, (Available from Cooperative Fisheries
Washington, March 1983. Peninsula College, Fish - Research Unit, Montana State University.
eries Technical Program, Port Angeles, Washington. Bozeman.)
Tappel, P. D., and T. C. Bjornn. 1983. A new method of Wilcock, P. R., G. M. Kondolf, W. V. G.
relating size of spawning gravel to salmonid embryo Matthews, and A. F. Barta. 1996.
survival. North American Journal of Fisheries Specification of sediment main tenance flows
Management 3:123-135. for a large gravel-bed river. Water Resources
Tautz, A. F., and C. Groot. 1975. Spawning behavior of Research 32:2911-2921.
chum salmon (Oncorhynchus keta) and rainbow trout Williams, H., F. J. Turner, and C. M. Gilbert. 1982.
(Salmo gairdneri). Journal of the Fisheries Research Petrography: an introduction to the study of
Board of Canada 22:633-642. rocks in thin sections, 2nd edition. Freeman,
Terhune, L. B. D. 1958. The Mark VI groundwater San Francisco.
standpipe for measuring seepage though salmon
Wolman, M. G. 1954. A method of sampling
spawning gravel. Journal of the Fisheries Research
coarse river-bed material. Transactions,
Board of Canada 15:1027-1063.
American Geophysical Union 35:951-956.
Tukey, J. W. 1977. Exploratory data analysis. Addison-
Young. M. K., W. A. Hubert, and T. A. Wesche.
Wesley, Reading, Massachusetts.
1990. Fines in redds of large salmonids.
Vanoni, V. A. 1975. Sedimentation engineering. American
Transactions of the American Fisheries
Society of Civil Engineers, New York.
Society 119:156-162.
Young, M. K., W. A. Hubert, and T. A. Wesche.
1991. Biases associated with four stream
substrate samplers. Canadian Journal
Fisheries and Aquatic Sciences 48:1882-
1886.
 ASSESSING SALMONID SPAWNING GRAVEL QUALITY

Appendix: Gravel Size Distributions

TABLE A.I.Cumulative size distributions for rainbow trout spawning gravel samples, Colorado River and tributaries. Size
categories differ between tributaries and main stem.

Sample Stream or Redd or Cumulative percentage of grains finer than (mm):c

Rivera potentialb
number 0.063 0.25 0.85 2 4 9.5 12.5 19

1 Nankoweap Redd 0.3 0.6 3.2 25.8 17.2 34.2 42.5 58.4
2 Nankoweap Redd 0.0 0.1 0.3 4.6 4.0 15.1 22.0 40.1
3 Nankoweap Potential 0.1 0.2 1.6 31.1 25.7 62.9 71.5 83.9
4 Nankoweap Potential 0.5 1.0 3.4 29.2 21.1 41.6 48.8 65.5
5 Nankoweap Potential 0.3 1.0 6.6 38.3 24.6 42.4 50.1 64.8
6 Nankoweap fan Redd 0.1 0.5 2.3 12.8 9.3 21.7 26.5 38.3
7 Clear Redd 0.7 1.6 4.7 32.5 23.9 39.1 43.4 51.0
8 Clear Potential 2.0 3.8 6.9 35.7 24.1 38.3 43.0 51.6
9 Clear Redd 0.1 0.1 1.2 9.3 6.8 16.3 19.9 29.3
10 Clear Potential 1.3 2.2 4.9 17.8 10.7 14.8 18.3 25.2
11 Bright Angel Potential 1.1 2.5 8.4 36.3 22.1 32.5 36.7 44.6
12 Crystal Redd ND ND ND ND 3.7 9.1 11.4 17.0
13 Crystal Redd 0.3 0.5 0.8 4.3 4,2 18.3 25.1 36.3
14 Crystal Potential 0.5 1.9 3.7 11.9 8.2 22.4 29.3 45.5
15 Shinumo Redd 0.1 0.2 1.0 11.0 8.4 18.3 22.8 31.2
16 Shinumo Redd 0.0 00.1 0.5 7.4 5.7 12.1 14.8 20.2
17 Shinumo Potential 0.9 2.4 6.8 32.7 20.6 31.2 35.6 43.2
18 Tapeat.s Redd 0.0 0.4 2.3 14.0 10.8 28.9 35.5 47.6
19 Tape at s Redd 0.1 0.5 2.3 14.5 11.4 30.6 37.9 49.9
20 Tapeat.s Potential 0.6 3.9 10.6 32.2 19.3 39.8 45.4 57.0
21 Tape at s Potential 0.4 3.7 10.1 30.5 18.6 34.6 39.8 53.0
Stream or Redd or Cumulative percentage at grains finer than (mm):c
Rivera potentialb
Sample 25.4 32 45 64 90 128 180

1 Nankoweap Redd 69.2 72.2 84.4 97.3 100.0 100.0 100.0

2 Nankoweap Redd 57.4 69.9 86.6 100.0 100.0 100.0 100.0
3 Nankoweap Potential 88.0 88.8 92.2 100.0 100.0 100.0 100.0
4 Nankoweap Potential 76.5 80.4 91.5 93.1 100.0 100.0 100.0
5 Nankoweap Potential 74.6 78.8 92.8 100.0 100.0 100.0 100.0
6 Nankoweap fan Redd 50.6 54.9 69.6 88.1 100.0 100.0 100.0
7 Clear Redd 58.8 61.8 72.8 91.2 100.0 100.0 100.0
8 Clear Potential 61.7 64.6 72.8 84.5 90.3 100.0 100.0
9 Clear Redd 36.9 44.5 54.6 71.3 75.0 100.0 100.0
10 Clear Potential 33.6 38.2 49.4 68.9 80.9 100.0 100.0
11 Bright Angel Potential 50.5 54.1 66.0 79.4 100.0 100.0 100.0
12 Crystal Redd 24.1 29.8 46.1 72.5 88.6 100.0 100.0
13 Crystal Redd 45.0 49.3 58.1 88.0 88.0 100.0 100.0
14 Crystal Potential 61.3 71.6 84.7 100.0 100.0 100.0 100.0
15 Shinumo Redd 35.3 40.7 57.2 82.5 100.0 100.0 100.0
16 Shinumo Redd 24.6 26.5 37.9 50.4 65.1 86.0 100.0
17 Shinumo Potential 48.9 51.9 60.8 74.6 100.0 100.0 100.0
18 Tapeats Redd 57.6 64.4 77.6 83.7 100.0 100.0 100.0
19 Tape at s Redd 56.0 59.4 65.9 83.6 100.0 100.0 100.0
20 Tapeats Potential 63.7 66.7 76.0 85.5 100.0 100.0 100.0
21 Tape at s Potential 62.5 66.5 77.7 93.6 100.0 100.0 100.0
Sample Stream or Redd or Cumulative percentage at grains finer than (mm):c
number river potential
0.045 0.25 0.5 1 1.7 3 6 11 22 51

22 Main stemd Redd 0.0 3.0 9.0 14.9 18.3 24.6 36.8 53.5 78.6 100.0
23 Main steme Redd 0.1 5.2 11.2 16.6 19.1 23.6 34.3 49.8 72.3 100.0
24 Main stemf Potential 0.2 4.0 16.0 22.7 24.5 27.6 32.8 41.8 57.6 79.2

a
Tributary creeks for samples 1-21.
b
"Redd" denotes actual use; "potential" denotes undisturbed gravel near redds.
c
ND means "no data."
d
Four-mile bar.
e
Eight-mile bar.
f
Twelve-mile bar.