Restoring cottonwood &
willow riparian forests
A field-calibrated seedling recruitment
model for the lower San Joaquin Basin
In California’s Central Valley, widespread flow
regulation and land development have greatly
reduced the extent and sustainability of native
cottonwood and willow riparian forests, which
provide critical habitat for many species of wildlife
and fish. The results of a three-year study of
seedling recruitment processes were used to develop
an ecological modeling approach for supporting
restoration planning.
R
iparian zones are critical areas in the landscape that
connect and sustain river and terrestrial ecosystems.
Riparian trees stabilize streambanks, filter nutrients
and pollutants, cool nearby air and waters, contribute
nutritious leaf litter and large woody debris to the
aquatic ecosystem, and provide migration corridors and
vertical habitat for birds and wildlife.
O
ver the last 150 years, 90% of riparian habitat in
the Central Valley has been lost to land conversion,
channelization, and flow manipulation. In the lower
San Joaquin Basin, alteration of natural flow regimes for
flood control, irrigation, and hydropower has reduced
seedling establishment for the dominant cottonwood
and willow species. Because these are the first trees to
colonize young floodplain surfaces (‘pioneer species’)
and the fastest growing, they are among the most
important in the ecosystem. With decreased seedling
establishment, near-river forest patches have become
older, more fragmented, and more likely to be replaced
by other, less beneficial habitat types.
T
o combat the decline of riparian areas,
rehabilitation of channels and floodplains in
Our study area was
the lower San Joaquin
Basin in the Central
Valley of California,
with field sites on the
Tuolumne and San
Joaquin rivers.
rom 2002 to 2005, we studied the key physical and
ecological processes needed to restore riparian
cottonwood and willow ecosystems in the San Joaquin
Basin. Using a simple conceptual model as a guide, we
conducted innovative field studies and experiments to
quantify factors controlling seedling recruitment for the
three species dominant throughout the Basin: Fremont
cottonwood (Populus fremontii ssp. fremontii), Goodding’s
black willow (Salix gooddingii), and narrow-leaved willow
(Salix exigua). We used the data from these studies to
develop a predictive recruitment model that can make
restoration strategies more effective and less costly,
such as modifying the timing and magnitude of flow
releases to ensure the greatest possible benefit for tree
populations and their dependent riparian ecosystem.
SAN
FRANCISCO
n
Ri
r
r1
r
Ap
1
Ma
y1
FRIANT DAM
ve
4
Ma
r
Rive
ui
aq
8
1
ced
Mer
Jo
Summer baseflow
(supports
seedling survival
and root growth)
b
Fe
River
ne
Tuolum
n
20
Spring snowmelt
(promotes
germination and
early root growth)
0
2
10
12
1
NEW DON PEDRO DAM
MODESTO
0
16
n
Ja
r
ive
sR
lau
nis
Sta
Sa
unimpaired discharge (1000 cfs)
F
F n
nc
Winter Peak flows
(remove vegetation and
deposit fine sediment)
20
degraded river reaches has become an increasingly
frequent restoration activity in the lower San Joaquin
Basin. Numerous large projects have been implemented
on the Tuolumne and Merced rivers, with more planned
on these and other Basin rivers. However, these
efforts are limited by a lack of knowledge of ecological
requirements for native plant species and practical ways
to predict the effects of restoration actions.
n
Ju
1
l
Ju
1
g
Au
1
p
Se
1
30
40
50 Kilo
meters
Kilometers
FRESNO
Riparian tree recruitment
in this region is controlled
by river flow timing and
magnitude, soil conditions,
and climate. Characteristics
of the annual river flow
regime critical for successful
recruitment are noted in the
figure to the left.
Starting with a conceptual model:
the ‘recruitment box’
Seed release
period
ELEVATION ABOVE BASEFLOW
‘Recruitment box’
Maximum
discharge
Survivable
rate of water
table decline
river
stage
Banks dry during seed release
Water stress
Suitable
site
hydrology
Successful
recruitment
Summer baseflow
APR
MAY
JUN
JUL
AUG
SEP
T
he key ecological drivers of seedling recruitment in near-channel riparian zones include three components: site hydrology, seed release timing, and seedling tolerance to desiccation (Mahoney and Rood 1998). Site hydrology determines
the availability and condition of potential seedbeds and water table dynamics throughout the growing season. Because
these species have short-lived seeds, dispersal timing controls when and where on the riverbanks seeds germinate. Recruitment typically occurs as spring floodwaters recede, and avoiding or tolerating water stress is critical for long-term survival.
The area in yellow illustrates the ‘recruitment box’, where these three factors determine a temporal and spatial zone of
opportunity for successful seedling recruitment. Along riverbanks, we see this process in action as a band of dense seedlings
whose upper margin is limited by the magnitude of flooding
during the seed release period as well as rapid rates of water
table decline. The lower margin is limited by scour and
deposition from subsequent high flows.
In a nutshell...
Seedling recruitment is critical for sustaining
near-channel riparian forests along lower San
Joaquin Basin rivers, but extensive changes to
the natural flow regime reduce the successful
establishment of young trees.
We adapted an existing conceptual model of
seedling recruitment to quantify the 3 key
factors limiting cottonwood and willow establishment in the ecosystem:
• the influence of site hydrology versus biological processes;
• seed release patterns;
• seedling water stress thresholds.
Using empirical data from these studies, we
developed a model that simulates recruitment
patterns, and tested predictions for the lower
Tuolumne River in 2002–04 against independently-observed field data.
3
What we measured: site hydrology, seed release timing, & seedling water stress
Site hydrology
sets the stage
16
12
ecause cottonwoods and willows need
moist seedbeds to germinate and grow,
site hydrology­—the amount and pattern of
soil moisture available during the growing season—is a crucial limiting factor for successful
recruitment. We tested the influence of site
hydrology and other factors on recruitment at
three sandbar sites along the lower Tuolumne
River. Seedling occurrence and survival were
correlated positively with soil moisture, negatively with bank elevation, and not at all with
the density of competing plants.
500 cfs
2003
120
80
12
8
40
4
0
0
20
16
2004
120
80
250 cfs
100 cfs
50
40
60
decline of 1 centimeter per day
for 50 days. That
critical threshold
(50% survival after 50 days) is met
at 1.5 cm/day for
narrow-leaved willow, and 3.5 cm/day for Goodding’s
black willow. Almost no seedlings
survived water table decline rates 6
cm/day or greater.
A
0
r1
Ma
r1
Ap
y
Ma
1
n1
Ju
l1
Ju
g1
Au
p1
Se
t1
Oc
0
ottonwood and willow species have
short-lived seeds and no long-term soil
seed bank. For seedling recruitment to be
successful, trees must release seeds when
sandbars are saturated and clear of competing vegetation. Historically, this happened
in late spring as snowmelt floodwaters began to subside. Peak seed release needed to coincide with elevated river levels for seedlings to establish
high enough on river banks to escape scouring flows later in the year.
Temperature influences
seed release timing
W
e tracked the seed density and dispersal timing at six floodplain sites along the lower Tuolumne and San Joaquin rivers
from 2002–04. Cottonwood trees consistently released seeds before
the willows, and dispersal was earliest in 2004, the year when spring
air temperatures were warmest. We adapted a simple but robust
method—a ‘degree-day’ model commonly used in crop science—to
predict the annual timing of peak seed release (measured as a seed
release index; see caption below). We can use this approach and
the field data to predict the best dates to release river flows for maximizing seedling recruitment throughout a river corridor.
For three years, we observed seasonal seed release
patterns (right) and calculated a seed release index, measured as the number of open catkins per
tree during each survey (figure to left).
80
Declining water
tables impair seedling
growth and survival
he rate of water table decline
following peak runoff is a
critical limiting factor for seedlings. Stage declines that are too
rapid will desiccate roots. We grew
seedlings at five simulated river
drawdown rates (range 0–9 cm/
day) and measured survival, growth
and physiological response to water
stress. Results of the drawdown
experiment indicate that 50% of
Fremont cottonwood seedlings
will survive a constant water table
40
C
t all treatment levels, Goodding’s black willow seedlings
had the highest survival and
growth, which they achieved
by minimizing water vapor loss
through their leaves. Control
group plants (grown at a stable
water table) were the largest for
100
all species, and growth declined
with increased rates of water table
decline. Survival was greatest for
individuals with faster root growth
rates and more root biomass (relative to shoot mass), although there
was no evidence that seedlings
actively adjusted their growth in
response to water stress.
B
oth willow species tolerated water stress better than Fremont
cottonwood. This pattern is consistent with differences in the species’
environments during seedling
establishment. River flows are generally stable and high in the early
spring, during cottonwood peak
seed release and germination, and
decline more steeply during the
late spring, when willows establish.
We drew down
water levels
for 50 days
and tracked
seedling survival, growth,
and physiological response
(left). For
each species,
empirical data
(top right)
were used to
model cohort
survival (bottom right).
cohort survival (%)
100
river discharge (m³/sec)
4
16
4
20
T
0
0
20
8
150
80
40
4
12
200
120
80
60 drawdown
rate
0 cm/d
1 cm/d
3 cm/d
6 cm/d
9 cm/d
modeled
40
20
0
0
cohort survival (%)
bank elevation
above summer baseflow (cm)
250
unimpaired discharge (1000 cfs)
8
B
We developed
stage-discharge relationships at 3
sites in order
to convert
annual flow
records to
relative bank
elevation
(photo above
& figure to
right).
Fremont cottonwood
Goodding's black willow
Narrow-leaved willow
unimpaired discharge
2002
seed release index (open catkins/tree)
20
10
20
30
40
50
days since start of drawdown
100
80
60
40
20
0
0
dr 2
ra aw 4
te d
(cmow 6
8 0
/d n
)
10
50
40
30 e
20 ys sinc down
da draw
t of
r
a
t
s
5
Quantifying the conceptual model
We used results from the field studies
and experiments to make the conceptual recruitment box model (Mahoney
and Rood 1998) a quantitative tool
for predicting seedling recruitment
patterns. The table to the right summarizes the key conceptual model
processes, the studies we conducted to
measure them, and how we translated
the empirical data as input functions
to the quantitative recruitment model.
The resulting model predicts, for each
species, seedling density and rooting
elevation along riverbanks at the end
of summer.
Conceptual
Recruitment Box
Model Component
Site hydrology
Seed dispersal
timing
Seedling water
stress thresholds
Empirical Data
Collected
Recruitment
Model Input
Function
-Pressure transducer
stage data
-Tuolumne River daily
flow records
-Stage-discharge
relationships
-Daily flow records
-Daily stage hydrographs
-Seed release data (open
catkins/tree)
-Continuous air temperature
-Seed release index
-Degree-day model
-Seedling survival,
growth and physiology
for 5 rates of water
table decline (0, 1, 3,
6, and 9 cm/day)
-Seedling cohort
survival model for a
continuous range of
water table decline
rates
Testing the model
W
e tested the model by predicting species-specific recruitment patterns for the lower Tuolumne River in 2002-04 and
comparing predictions to seedling distributions observed independently in annual boat surveys conducted along a 20-km reach.
T
he model predictions capture the basic recruitment patterns
among species and between years. In both model predictions
and field data, seedling densities and rooting elevation were highest in 2004 and lowest in 2003. Both predicted and observed data
showed Goodding’s black willow densities to be highest and narrow-leaved willow the lowest in all years.
This research was funded by a CALFED Ecosystem Restoration Program grant to Stillwater Sciences and by academic grants secured by Dr. John Stella as part of a doctoral dissertation in the Department of
Environmental Science, Policy and Management (ESPM) at the University of California at Berkeley. Drs. John Battles and Joe McBride of ESPM provided scientific oversight for the research as dissertation
advisors to Dr. Stella. Academic funding sources include the CALFED Science Program, the National Science Foundation, and the Colman Fellowship in Watershed Management at UC Berkeley. In-kind support and facility use was provided by the Center for Forestry and Center for Stable Isotope Biogeochemistry at UC Berkeley. Property owners who graciously granted access include: R. Ott, T. Venn, G. Austin, S.
Howard, Lakewood Memorial Park, and the San Luis National Wildlife Refuge. The Turlock Irrigation District and McBain & Trush provided unimpaired flow data.
References Cited
Mahoney, J. M., and S. B. Rood. 1998. Streamflow requirements for cottonwood seedling recruitment--an integrative model. Wetlands
18:634–645.
Rood, S. B., G. M. Samuelson, J. H. Braatne, C. R. Gourley, F. M. R. Hughes, and J. M. Mahoney. 2005. Managing river flows to restore floodplain forests. Frontiers in Ecology and the Environment 3:193-201.
Stella, J.C., J.J. Battles, B.K. Orr, J.R. McBride. (in press). Synchrony of seed dispersal, hydrology and local climate in a semi-arid river reach in
California. Ecosystems.
Stella J.C. 2005. A field-calibrated model of pioneer riparian tree recruitment for the San Joaquin Basin, CA. Ph.D. Disseration. University of
California, Berkeley.
Stillwater Sciences. 2006. Restoring recruitment processes for riparian cottonwoods and willows: A field-calibrated predictive model for the
lower San Joaquin Basin. Prepared for CALFED, Sacramento, California by Stillwater Sciences and Dr. John Stella, in conjunction with Drs.
John Battles, and Joe McBride, Department of Environmental Science, Policy, & Management, University of California, Berkeley.
6
For more information, and
full text of report, go to
www.stillwatersci.com.
Generating recruitment predictions
250 cfs
100 cfs
stage-discharge
relationships
1
p
Se
1
0
Ma
r1
Ap
r1
Ma
Recruitment
Model
Ju
n 1 J ul 1
river
stage
MAY
JUN
JUL
p1
AUG
SEP
May
60
2003
0
0
10
10
20
20
30
30
40
40
50
2002
Predicted
Observed
June
50
60
-final seedling density
-seedbed elevation
-species composition
elevation above
baseflow (cm)
Se
Survivable
rate of water
table decline
Model Predictions
-seedling surveys
-observed distributions
g1
50
40
30 ce
20 ys sin down
da f draw
10
to
star
‘Recruitment box’
APR
Validate Predictions
Au
0
2
radraw 4
te d
(cmow 6
8 0
/d n
)
Seed release
period
ELEVATION ABOVE BASEFLOW
Suitable
site
hydrology
y1
60
l
Ju
0
20
40
60
80
100
2004
50
1
40
y
Ma
30
1
25
20
r
Ma
50
100
80
60
40
20
0
10
0
75
0
25
Fremont cottonwood
Goodding's black willow
Narrow-leaved willow
100
cohort survival (%)
500 cfs
seed release index
elevation above
baseflow (cm)
75
50
Seedling Water
Stress Thresholds
Seed Release
Timing
Site
Hydrology
0
20
40
60
80
100
0
20
40
60
80
100
seedling density (stems/m²)
Now what?
The predictive, field-tested recruitment model we developed is directly relevant
to river management and conservation. By combining the recruitment model
with geographic information systems and other analytical tools, we can address a
number of important needs, such as designing efficient and ecologically-beneficial
flow releases, predicting spatial recruitment patterns at restoration sites, and
simulating the impacts on tree populations of climate-driven changes in river
hydrology and seasonal temperature.
7
Applying it
Managing water for
ecosystem benefit
W
elevation above
summer base flow
-2 to 0 ft
0 to 2 ft
2 to 4 ft
4 to 6 ft
6 to 8 ft
8 to 10 ft
10 to 20 ft
idespread flow regulation along Central Valley
rivers contributes to the decline of pioneer
riparian vegetation populations, but also provides a
critical opportunity for their recovery. Our research
findings on tree reproductive timing and life history
traits can be used to make flow releases more effective in
promoting seedling recruitment at the lowest water cost.
Because seed production is abundant every year, these
‘recruitment flows’ may be needed only in wet years,
when natural water surpluses can meet both human water
demand and ecosystem needs.
cottonwood seed
release
20
R
iver corridor planning and large-scale restoration
projects are increasingly common in the lower
San Joaquin Basin, where channels and riparian areas
are degraded by land development, flood control
structures, and aggregate mining. The recruitment
model can be a useful design tool for revegetating
floodplains by predicting where seedlings are likely to
establish under a particular flow regime.
F
or these projects, we would use GIS-based digital
topography and stage-discharge data from field sites
or hydraulic models to model areas of the floodplain
where seedlings would germinate and survive. In the
example above, colored floodplain zones are defined
by elevation above the summer low-flow stage. Using
this coverage with the recruitment model will allow us
to evaluate which discharge patterns would maximize
the spatial extent of seedling recruitment. This
approach has been applied successfully to predict
riparian vegetation response to natural and managed
flow regimes along other Western rivers.
willow
seed release
16
discharge (1000 cfs)
River corridor
restoration planning
unimpaired
regulated
recruitment flow
12
8
4
0
n
Ja
1
b
Fe
1
Ma
r1
r
Ap
1
Ma
y1
n
Ju
1
T
1
g
Au
1
p
Se
1
he figure above illustrates a proposed recruitment
flow (yellow line), with unimpaired discharge (blue
shaded area) and regulated flow (orange line) for a
high-flow year (1986) in a representative river reach in
the San Joaquin Basin. The recruitment flow uses the
same volume of water as actually released but with the
seasonal timing and flow recession modified to address
riparian seedling needs. The proposed release peaks at
the beginning of Fremont cottonwood’s peak seed release
period and recedes gradually until the river reaches its
summer low-flow stage. The recession rate is less rapid
initially to sustain young cottonwood seedlings, which
have slower root growth than willows. Recruitment flows
will need to be coordinated with water management and
other ecosystem needs (e.g., flow releases for native fish).
Prepared by Stillwater Sciences © 2006. Visit www.stillwatersci.com for more information.
Printed on recycled paper using soy-based inks.
8
l
Ju