This file was created by scanning the printed publication.
Errors identified by the software have been corrected;
however, some errors may remain.
The Effects of Prolonged Flooding on the Riparian
Plant Community in Grand Canyon1
Lawrence E.
Stevens and Gwendolyn L.
Waring 2
Abstract.
Flood-induced removal, drowning,
and estimated total mortality levels of perennial
riparian plants were high, with significant differences between species following flooding of the
Colorado River corridor downstream from Glen Canyon
Dam in 1983-1984.
Proximity to the river (duration
of inundation), substrate type, and plant height
(age) were correlated with mortality.
Differential
mortality and colonization by seedlings may result in
riparian plant community change in this system.
depletion and other changes in inundated
soils (Ponnamperuma 1984), duration and
stage of flooding and turbidity (reduced
1 i g h t i n ten sit y).
INTRODUCTION
Flooding is the most ubiquitous
form of disturbance in riparian
ecosystems.
In unregulated streams it
limits riparian plant community
development (Campbell and Green 1968),
while reduced flooding in dam-controlled
streams permits plant life to colonize
streambanks (Turner and Karpiscak 1980),
creating ecologically and recreationally
valuable riparian habitat (Johnson and
Jones 1977).
Flooding events subsequent
to discharge regulation alter riparian
plant community structure through damage
and mortality of streamside plants.
Recent flooding events in the
Colorado River corridor downstream from
Glen Canyon Dam provided an opportunity
to study the effects of flooding on
riparian vegetation in a dam-controlled
system.
On 29 June, 1983 discharge from
Glen Canyon Dam reached 2,621m 3 /sec in
the Colorado River corridor in Grand
Canyon, and flows remained at twice the
normal level through 1984.
The flood
peak was the largest to pass throu$h
Grand Canyon in the post-dam (post-1963)
era of regulated discharge, and the
event exerted significant impacts on the
riparian plant community there.
Numerous factors influence floodrelated plant mortality.
Mor.tality
varies with plant age and between
species, and inundation resistance
increases with plant age (Hosner 1958;
Horton et ale 1960; Warren and Turner
1975; Kozlowski 1984).
Prolonged
flooding negatively affects leaf, shoot,
cambial and root growth and morphology,
and successful seedling establishment
varies ~idely between plant species
following flooding (Kozlowski 1984).
Abiotic factors that influence mortality
include water temperature, oxygen
To determine the impact of this
flooding event on the riparian plant
community in the Colorado River corridor
in Grand Canyon, we posed the following
questions:
1) Do all riparian plant
species respond to flooding in a similar
fashion in this system?
2) Do
floodstage, reach type, substrate type,
distance from Glen Canyon Dam, stem
density, and stem height influence
flood-induced plant mortality?
3) How
do different plant growth and
reproductive strategies (e.g. sexual
versus clonal strategies) affect
survival and recovery?
4) Can floodinduced colonization compensate for
differential mortality of adult
plants?
5) Lastly, did riparian plant
community composition change as a result
of the flooding event?
We present
preliminary results of our findings in
this paper.
1Paper presented at the symposium
on riparian ecosystems and their
management.
[University of Arizona,
Tucson, April 16-18, 1985].
2Lawrence E. Stevens and Gwendolyn
L. Waring are graduate students in the
Department of Biology at Northern
Arizona University, Flagstaff, Arizona.
81
Mortality due to drowning of all
perennial riparian species was measured
on 47 transects in 1984.
Transect sites
were selected in the four reach types
throughout the river corridor.
Each
transect was 30m in length and extended
to the top of Floodzone C.
The number
and heights of live and dead plants
(including seedlings) of each species
were measured in each floodzone of each
transect, and transect width was
measur~d.
The unflooded zone above the
2,400m /sec line was not an appropriate
control against which to compare flooded
plants because growing conditions and
sources of mortality were differen~
there.
Data were analyzed using X
statistics, multiple linear regression
and analyses of variance.
METHODS
To answer these questions we
collected data from several sources.
River-based surveys of the riparian
corridor in 1983 and 1984 provided three
data sets on removal of plants by
scouring flood waters:
1) the presence
and condition of plants under
observation since 1980
was determined
in 1984 follow~ng the final subsidence
of flows >700m /sec; 2) three reaches
were censused by counting all shrubs and
trees in 1982 and 1984, between miles
60-61, miles 166.5-179.5 (for Prosopis
only), and miles 196.5-198.0; and 3)
nine 10m x 30-40m study sites, each
situated with its long axis parallel to
the §iver and less than 5m from the
700m /sec stage were censused.
These
study sites were located between miles
43 and 170 (downstream from Lees Ferry)
and were sampled for plant density and
species composition in 1982 and 1984.
To evaluate the change in community
structure resulting from this flooding
event, we calculated Stander's community
similarity index (Sullivan, 1975):
Information gathered from all
sources included plant species, height,
and condition, with distinct clumps
considered as single individuals.
The
proximity of plants to the river (a
measure of the period of inundation) was
determined by dividing the total
inundated area into three floodzones,
all of which lay in Carothers et ale
(1979) Zones 3 and 4 3 •
Floodzone A
570m 3 /sec to 1,130m3/~ec; Floodzone B
1,130m 3 /sec to 1,700m J /sec; and
Floodzone C = 1,700m 3 /sec to
2,400m 3 /sec.
Reach type categories
included eddy, straight, riffle, or
rapid settings.
Substrate types
included silt, sand, mixed sand and
cobble, cobble, and bedrock. Distance
downstream from Glen Canyon Dam was
noted.
Because virtually all of the
post-dam riparian vegetation occurred
below the 1,700m 3 /sec stage, data on
plants larger than the seedling sizes
from the floodzones A and B were pooled
for analysis of removal.
X2 analyses
with the Yates correction for continuity
(Brower and Zar 1977) were used to
determine if removal was significant for
each species.
3Carothers et al. (1979) described
several parallel zones of post-dam
riparian vegetation along the Colorado
River in Grand Canyon.
Below the desert
talus slope vegetation (Zone 1), a predam bel t of Prosopis .glandulosa, Acacia
greggii and some Tamarix chinensis (Zone
2) grew down to the 3,100m 3 /sec water
line (our estimate).
Zone 3 (between
the 3,100m 3 /sec and 1,700m 3 /sec stages)
was only ~parsely vegetated.
Their Zone
4 (1,700m /sec to the water line)
consisted of Tamarix, Salix, four
species of Baccharis, Tessaria and other
obligate riparian species (sensu Johnson
and Lowe, this volume).
SIMI
t
Pij
I
where Pij and Pin are the proportions of
species 1 in samples j and n,
respectively.
This statistic varies
from 0 for entirely dissimilar
communities to 1.0 for identical
communities.
We also calculated J'
(Pielou 1977), a measure of the evenness
of species' distributions which varies
from 0 for highly uneven communities to
1.0 for communities with equal
abundances of all species.
RESULTS
Discharge Data
The 1983-1984 flooding event was a
prolonged, elevated discharge of cold,
clear, well-oxygenated water.
Minimum
average current speed in straight
reaches exceeded 3.0m/sec during the
1983 flood peak.
Discharge data from
Glen Canyon Dam (measured at the U.S.
Geological Survey gauging station at
Lees Ferry, Arizona) for this flood are
presented in table 1.
Table 1:
Discharge data from Glen Canyon Dam during the 1983-1984 flooding event,
~:i::~~ed at the U. S. Geological gauging station at Lees Ferry,
MEAN
PERIOD
1 October, 1982 to
30 September, 1983
(1982 Water Year)
1 October, 1983 to
30 September, 1984
(1983 Water Year)
D~!~~~~
NUMBER OF ~!~~s~~) HIGH DISCHARGE
)700
)1,130
)1,700
683.7
147
56
35
750.0
99
64
)2,260
MAXIMUM
ANNUAL
FLOW
(m3 !sec)
2,603.6
1,282.0
4Mean discharge from 1963 to 1974 = 359.8m3 !sec; mean annual maximum flow =
789.5m 3 !sec. (Howard and Dolan, 1981).
82
Plant Mortality Prior
Levels of removal by scouring were
significant within most species at
p<0.005 and df=l; however, numbers of
Phragmites communis genets, Salix
gooddingii, Prosopis, and Acacia were
not statistically different before and
after 1983.
Susceptibility to removal
varied greatly between species. Three of
the four species with deep tap roots
suffered lower rates of removal than did
shallow-rooted species.
Removal data
for Tamarix indicate that removal
occurred at a significantly greater rate
in Floodzone A (p(0.005, df=2).
Because
Tamarix is extremely well anchored, the
trend of higher levels of removal at
lower floodstages is probably valid for
the other plant species as well.
Clonal
Phragmites, Salix exigua and Tessaria
suffered high levels of areal loss of
ramets, but because flooding rarely
removed all of a clone's ramets, genet
(total clone) mortality levels were low,
ranging from 6.8% for Salix exigua to
31.4% for Phragmites.
Clonal
macrophytes, such as Scirpus and Typha,
that occupied the river's edge prior to
1983, suffered removal rates of 88.9% to
100%.
to 1983
In general, all plant species
encountered in floodzones A and B were
growing vigorously in 1982, with
mortality levels less than 2%.
Low
density stands revealed pre-flood stem
mortality levels; for example, mortality
levels for Tamarix, Prosopis, and
Baccharis species were 1.9% (n= 494),
2.2% (n=45), and 0.0% (n=448),
respectively.
Relatively high
proportions of dead stems were
encountered only in dense stands of
Tamarix (38.6% to 44.0%, n=3 stands),
Salix exigua (0.94% to 27.4%, mean=7.4%
for 6 stands), and Tessaria (50.8%,
n=I).
We avoided using data from dense
stands in our analyses.
Flood-induced Plant Mortality
The percent mortality due to
removal, drowning, and total estimated
mortality of each common riparian plant
species are presented in Table 2.
Data
were pooled for eddy and straight
reaches in floodzones A and B in which
most of the riparian corridor vegetation
occurs.
Estimates of total mortality
are based on combined removal and
drowning mortalities.
Where removal
data were not available (i.e. for less
common species), removal was considered
to be 0; therefore, the total mortality
estimates are conservative.
Table 2:
Percent removal, percent of remaining plants drowned, estimated total
percent mortality, ~nd seedling de~sity/m2 of common perennial plant
species in the 700m / sec to 1, 700m / sec riparian floodzone in the Colorado
River in Grand Canyon.
SPECIES
Deep Tap Roots
Tamarix chinensis
Prosopis !llandulosa
Acacia~
Salix !l0oddin!lii
PERCENT
REMOVAL (n)
30.4
0.9
16.7
0.0
(4344)
(l08)*
(35)*
( 13)*
Clonal, Shallow Roots
Salix exi!lua (ramet)
88.8 (12890)
(44)
Salix exi!lua (clone)
6.8
(313)
Tessaria ~ (ramet)
74.7
Tessaria sericea (clone)
(13)
23.1
Aster spinosus
Phra!lmites communis (clone) 3[.4
(20)
(11)
Typha sp. (clone)
88.9
(3)
Scirpus sp. (clone)
100.0
Shallow Roots
Baccharis salicifolia +
emoryi
Baccharis sarothroides
Baccharis ser!liloides
Brickellia lon!lifolia
Aplopappus acradenius
Gutierrezia spp.
Other mesic-adapted species
Other >eeric-adapted species
MEAN TOTAL
85.7
52.1
(S67)
(1010)
51.8 ([9358)
PERCENT
DROWNED (n)
ESTIMATED
TOTAL PERCENT
MORTALITY
20.7 (1981)
49.1 ( 118)
37.6 (198)*
( 13)*
0.0
44.8
49.6
48.0
0.0*
6.7 (874)*
(41)*
0.0
1l.8 (5285)
( 11)*
18.2
11. 7 (922)
89.6
6.8*
77.6
33.3
11.7
31.4*
88.9
100.0
77.4
5S.2
63.6
75.7
72.8
30.2
15.8
42.9
(S65)
(721)
(33)
(399)
(184)
(630)
(76)*
(308)
34.1 ([2348)
SEEDLING
DENSITY/m 2
0.491
0.001
0.003
0.000
0.002 s
Prior to 1983, large riverside
beaches in eddy settings were usually
occupied by Salix exigua, Tessaria,
Tamarix and Baccharis, other perennials,
herbs, and grasses.
All plants on 12 of
15 such beaches were scoured away, and
one of the three remaining beaches was
left with only one Salix stem.
The two
remaining beaches lay on the inside of
river meanders and were somewhat
protected from substrate erosion.
Excavations on four of five previously
vegetated beaches revealed no root
structure to at least 1.5m depth, and
changes in sediment texture and bedding
indicate that beach surface sediments
were scoured and totally replaced in
many instances.
In several cases, the
morphology of beaches redeposited by
subsiding floodwaters was remarkably
similar to that prior to the flood.
0.083 s
0.017
0.000
0.000
0.000
96.8
76.6
63.6
7S.7
72.8
30.2
lS.8*
42.9
0.008
0.004
0.000
O.OlS
0.005
0.006
59.2
0.643
Mortality due to Drowning
Percent refUoval, drowning and total mortality values are significant within species
*-
at p<0.005 (df=!) unless otherwise indicated.
p values not statistically significant (p > 0.05, df=l) for pre- versus post-
s -
flood counts.
new shoots, not seedlings.
83
Rates of mortality due to drowning
varied significantly between species
(p(O.OOI, df=13,737) and within most
species.
All species except Acacia,
Salix exigua ramets, and pooled
miscellaneous species showed a
significant decrease in density due to
drowning (p<0.005, df=l for each
species).
Salix exigua (6.7%
mortality), Tamarix (20.7%), and several
other riparian species were relatively
tolerant of inundation, while Prosopis
(49.1%), Baccharis spp. (55.2% to
77.4%), Aplopappus acredenius (72.8%)
and Brickellia (75.7%) were intolerant
of inundation stress.
Three of the four
species with deep tap roots suffered
relatively low levels of drowning
mortality.
Nearly all xeric-adapted
species that had colonized post-dam
beaches from the surrounding desert were
intolerant of flooding.
Desert
Compositae, such as ~odia
pentachaeta, Gutierrezia ~thrae,
~.microcarpa, Aplopappus spinosus,
Encelia farinosa, and Peucephyllum
schottii suffered moderate to high
levels of mortality, as did Ephedra
spp., Larrea and various cacti species.
Mortality due
substrates.
Substrate type and reach
type (a measure of relative current
velocity) are intercorrelated in this
system: for example, sand or cobble
substrates occur in eddy or riffle
reaches, respectively.
Two-way analysis
of variance using factors of substrate
type and reach type showed that drowning
mortality decreased in sand substrates
as current velocity increased, but
mortality increased with velocity in
cobble substrates.
Two-way analysis of variance of the
mortality due to drowning of all species
was also run for floodstage and
substrate types.
This analysis showed
the highest levels of mortality (68.4%)
occurred in cobble substrates in
Floodstage A.
This trend is further
corroborated with data from cobble
islands near miles 53 and 73, which had
mean removal rates of 52.3% for Tamarix
and 100% for Baccharis spp., and 93.7%
mortality of remaining stems.
to Burial
Mortality due to burial by newly
deposited beach sediments could not be
distinguished from drowning with these
data; however, plant species were
observed to respond differentially to
this source of mortality.
Many Tamarix
plants that had been all but completely
buried produced new shoots and appeared
to be surviving in 1984.
A Salix exigua
clone at Mile 122.1R that had been
buried in 1983 and then re-exposed in
1984, produced vigorous new growth.
No
Baccharis sarothroides plants that had
been buried in 1983 were alive ' in 1984.
Analysis of variance showed that
mortality due to drowning was negatively
correlated with Tamarix plant height
(R 2 =.236, p<.OOl, df=9,220).
The
percent variation in Tamarix mortality
explained by reach type, floodstage,
substrate, stem density, and distance
from Glen Canyon Dam was greatest in
plants 3m or more in height (R2=41.2%,
p<0.004, df=8,40) and R2 values
decreased with plant height.
Factors Influencing Mortality due to
Drowning
Transect data were used to assess
the influence of plant density, plant
height, distance from Glen Canyon Dam,
reach type, floodstage (period of
inundation), and substrate type on
levels of mortality due to drowning.
Analyses of variance showed that the
latter two factors were significantly
correlated with mortality due to
drowning.
Drowning was strongly
correlated with floodstage for all
species and locations (p<O.OOl,
df=2,748), with 49.4% of all plants
drowned in Floodzone A, 26.2% drowned in
Floodzone B, and 17.7% drowned in
Floodzone C.
Range tests showed that
mortality was significntly different in
each of the three floodzones.
~ata for
Tamarix by itself also showed that
mortality attributed to drowning was
strongly correlated with floodstage
(p<0.005, df=2,168).
Colonization
Colonization is believed to be
directly related to flooding events in
this system (Hayden unpublished 1976).
Following flooding in 1980, mean
seedling densities of mixed species
reached 2,921/10 2 (n=6) on previously
uncolonized beaches.
In September, 1983
dense Tamarix seedling beds were
observed beneath the canopies of both
the Tamarix study sites that had been
inundated by floodwaters.
This was the
first colonization at these sites in 5
years of observation.
Seedling
densities ranged from 4.5/m 2 to 330/m 2 ,
with the higher germination taking place
on a silt bed that had been deposited by
tributary flooding.
No Tamarix
seedlings have ever been observed to
germinate beneath the canopy of the
Tamarix stand that was not inundated in
1983.
Drowning mortality varied
significantly between the five substrate
types (p<O.Ol, df=3,747), with lowest
mortality on bedrock substrates (23.2%),
moderate mortality in silt, sand, and
sand-cobble mixed substrates (30.6% to
31.2%), and highest mortality on cobble
substrates (53.8%).
Range tests showed
that cobble substrates were
significantly different from the other
Analysis of transect data revealed
that colonization effort was unequal
between species (table 2).
Tamarix
seedlings were more than 5 times more
abundant than any other species;
however, subsequent mortality of Tamarix
seedlings is expected to be extreme.
At
a density of 0.003/m 2 , Acacia seedlings
84
were three times as abundant as Prosopis
seedlings and have relatively high
survivorship.
All clonal plant species
showed a vigorous production of new
shoots.
Rapid recolonization of beaches
was observed in Salix exigua, Tessaria,
Phragmites, and Aster spinosus.
Gutierezzia spp. and Dyssodia seedlings
were the only talus slope species to recolonize the post-flood beaches in
abundance, and recruitment may
compensate for the loss of adult plants
in these two species.
Agave utahensis
seedling density was significantly
higher in Floodzone B than in other
zones at some sites in Marble Canyon,
and this ~pecies demonstrated a rapid
and extensive colonization response to
flooding.
Changes
Floodstage, substrate type, and
reach type were abiotic factors that
correlated with mortality due to
drowning, and the highest levels of
mortality occurred in Floodzone A.
The
value of these factors in explaining
drowning mortality was improved by
excluding smaller height classes in the
Tamarix data set.
Distance downstream
from Glen Canyon Dam and plant density
were unimportant in explaining observed
mortality.
Disturbance by flooding is a
mechanism of community change in this
system.
It is evident from the results
presented above that the 1983 flooding
event served as a "weeding" event that
decreased overall plant densities by
scouring, drowning, and perhaps
burial.
Flooding decreased a small
population of yellow Mimulus cardinalis
at Mile 31.8R, but did not result in a
large-scale loss of species from this
system.
Flooding did cause a range
expansion of one species:
Corispermum
nitidum, was rare in the riparian zone
prior to 1983 but became common on
beaches throughout the river corridor in
1983 and 1984.
in Community Similarity
When adjusted for removal, pre- to
post-flood plant community similarity
decreased (SIMI=0.862), and evenness of
species composition decreased slightly
(J'pre-flood = 0.792 and J'post-flood =
0.761).
By combining seedling data with
adult plant data and recalculating these
indices (assuming complete survivorship
of seedlings), the maximum possible
change in community structure resulting
from this flooding event was
estimated.
The community similarity and
J' values decreased dramatically (SIMI
0.567; J' = 0.471), with the community
more strongly dominated by Tamarix.
Other community similarity and diversity
statistics were calculated and agreed
with these results.
The immediate change in riparian
plant community similarity was moderate
as a result of this flooding event;
however, long-term changes may have been
initiated through promotion of
colonization.
For example, Acacia
seedlings were previously rare compared
to Prosopis, but now outnumber Prosopis
seedlings on beaches and have a high
survivorship.
While recruitment of
seedling colonists is not expected to be
complete in this system, 80.0% of all
plants encountered in the flood zone in
1984 were seedlings and it is apparent
that this flooding event resulted in a
"juvenescence" of the riparian plant
community.
DISCUSSION
The results presented above show
that virtually all riparian plant
species along the Colorado River in
Grand Canyon are highly susceptible to
flooding stress; however, betweenspecies mortality rates are strongly
differential.
Shallow-rooted Baccharis
spp. (Gary 1963), Brickellia longifolia,
and Aplopappus acradenius, suffered
higher levels of drowning than did
species with deep tap-roots, such as
Salix gooddingii, Tamarix chinensis
(Gary 1963), Acacia greggii, and
Prosopis glandulosa.
Despite high
levels of areal loss among several
common clonal species (i.e. Phragmites
communis, Salix exigua, and Tessaria
sericea), a few ramets of most clones
persisted, and overall clonal'mortality
rates were low.
Xeric-adapted plant
species, such as Ephed~a spp., various
cacti, Larrea tridentata, and Encelia
farinosa, that had colonized riparian
beaches from the surrounding desert were
generally intolerant of inundation.
The riparian zone of the Colorado
River was created with discharge
regulation by Glen Canyon Dam; however,
it is an ecologically and recreationally
valuable naturalized riparian habitat
(Carothers et al. 1979).
Appropriate
management of discharge in this system
should include consideration of the
potential effects of flood duration and
stage on substrate erosion, differential
mortality of adult plants, recruitment
phenology and seedling survivorship, and
the effect of flooding on riparian plant
community structure
and dynamics.
It
is relevant to note that floodstage was
the abiotic factor most closely
correlated with mortality by drowning,
and is closely correlated with removal
of Tamarix and other species. Such
considerations are of obvious importance
If the life of this riparian ecosystem
lsto be prolonged.
85
ACKNOWLEDGMENTS
Hosner, J.F.
1958.
The effects of
complete inundation upon the
seedlings of six bottomland tree
species.
Ecology 39: 371-373.
This research was supported by the
Bureau of Reclamation/ National Park
Service cooperative impact study of Glen
Canyon Dam.
We wish to acknowledge the
support provided by David Wegner, John
Thomas, and R. Roy Johnson.
Valuable
criticism on this manuscript was
provided by Graydon Bell, Peter W.
Price, and Christopher Sacchi of
Northern Arizona University.
Howard, A. and R. Dolan.
1981.
Geomorphology of the Colorado River
in the Grand Canyon.
J. Geol. 89:
269-298.
Brower, J.E. and J.H. Zar.
1977.
Field
and laboratory methods for general
ecology.
194 p. W.C. Brown Co.,
Dubuque, Iowa.
Johnson, R.R. and D.A. Jones.
1977.
Importance, preservation and
management of riparian habitat:
a
symposium.
[Tucson, Ariz., July 9,
1977]
USDA Forest Service Gen.
Tech. Rept. RM-43. Tucson, 218 p.
Rocky Mountain Forest and Range
Experiment Station, Fort Collins,
Colo.
Campbell, C.J. and W. Green.
1968.
Perpetual succession of stream
channel vegetation in a semi-arid
region.
J. Ariz. Acad. Sci. 5: 8698.
Kozlowski, T.T.
1984.
Responses of
woody plants to flooding, p. 129164.
In T.T. Kozlowski, ed.
Flooding and plant growth.
356 p.
Academic Press, Orlando, Fla.
Carothers, S.W., S.W. Aitchison, and
R.R. Johnson.
1979.
Natural
resources, white water recreation
and river management alternatives
on the Colorado River, Grand Canyon
National Park, Arizoa.
In R.M.
Linn, ed.
Proc. of the first
conference on scientific research
in the National Parks, I: 253-260.
Pielou, E.C.
1977.
Mathematical
Ecology.
385 p. John Wiley & Sons.
New York, NY.
LITERATURE CITED
Ponnamperuma, E.N.
1984.
Effects of
flooding on soils.
p. 10-45.
In
T.T. Kozlowski, ed.
Flooding and
plant growth.
356 p. Academic
Press, Orlando, Fla.
Gary, H.L.
1963.
Root distribution of
five-stamen tamarisk, seepwillow
and arrowweed.
Forest Sci. 9: 311314.
Sullivan, M.J.
1975.
Diatom
communities from a Delaware salt
marsh.
J. Phycology 11: 384-390.
Turner, R. M. and M.M. Karpiscak.
1980.
Recent vegetation changes
along the Colorado River between
Glen Canyon Dam and Lake Mead,
Arizoa.
125 p.
U.S. Geol. Survey
Professional Paper 1132.
U.S.
Government Printing Office,
Washington, D.C.
Hayden, B.
1976.
The dynamics of an
exotic on a man-altered system:
Tamarix in the Grand Canyon.
Unpuplished National Park Service
Report, 8 p.
Grand Canyon, Ariz.
Horton, J.J., F.C. Mounts, and J.M.
Kraft.
1960.
Seed germination and
seedling establishment of
phreatophyte species.
F.ort
Collins, Colorado Forest Service
Station Paper No. 48, 29 p. Rocky
Mountain Forest and Range
Experiment Station, Fort Collins,
Colo.
Warren, D.K. and R.M. Turner.
1975.
Saltcedar (Tamarix chinensis) seed
production, seedling establishment
and response to inundation.
J.
Ariz. Acad. Sci. 10(3): 135-144.
86