I
-'-
-
--
---
=- -
'-
-
=--
.
,>
: - , , ; t"-:t!':' fID:':r'\t"-"
''': :AW':: " "'i;:
; 'i.
· :.>( .
iMltf ;;\ '/' W
--_
------ifJ;. · v-
;==.--;
.
' .
'
.
----:-"
(.,
, '
,v-If."
.
,,', {'
l
'
I i.:U'(,
('/
.",
. .
If'l#tlal1d and l<Jparial't f}Xosyst-erns
ffu.I 'JItmtrican rwtst-'''
1
j
I
Wetland and Riparian Ecosystems
of the Alnerlcan West
Eighth Annual Meeting of the Society of Wetland Scientists
May 26-29, 1987 Seattle, Washington Technical Coordinators: Kathryn M. Mutz
Lyndon C. Lee
Proceedings Sponsored by:
The North American Riparian Council
Rocky Mountain Chapter of the Society of Wetland Scientists
Utah Division of all, Gas & Mining
Utah Division of Wildlife Resources
U.S. Environmental Protection Agency
U.S. Forest Service
Planning Information Corporation
Conference Sponsored by:
U.S. Environmental Protection Agency
North American Riparian Council
Coastal Zone '67
Shapiro and Associates
Environmental law Institute
U.S. Army Corps of Engineers Association of State Wetland Managers U.S. Fish and Wildlife Service Washington Department of Ecology U.S. Geological Survey
National Wildlife Federation
U.S. Department of Commerce, NOAA
' , . "
A TECHNIQUE FOR ASSESSING THE INFLUENCE
OF SHALLOW WATER TABLE LEVELS ON
RED ALDER {Alnus rubra Bong.} FArHLY PERFORl'lANCE
3
l
2
4
D. D. Hook , M. M. Murray , D. S. DeBel1 , and B. C. Wilson
Abstract.--Depth of rusting on iron rods placed in the soil near
red alder trees was used to assess the relationship between tree
height at age 5 years and the soil water table level during the dor­
mant season.
The iron rods proved to be a sensitive indicator of
water table level. By use of regression and covariance analyses
differences in family response were found over the range of 0 - to ­
30 cm deep water tables. It was concluded that the technique is
sensitive, simple to use, and inexpensive. With further calibration
on various soils, it should prove to be a useful tool in helping to
characterize a wide array of wetland sites.
r NTRODUCTI ON
Dalineation of wetlands are of interest for
',regulatory and management purposes. Consequent­
ly, considerable effort has been expended in deve­
loping techniques to characterize wetness or
aeration zones in wetlands (Theriot, 1987; Good
et a1., 1986; Faulkner et a1., 1986; Armstrong'
et a1., 1985; Clark and Benforado, 1981).
Furthermore, the Corps of Engineers have deve­
loped a comprehensive manual for the purpose of
delineation Section 404 wetlands (Environmental
Laboratory 1987).
All of the above methodolo­
gies are useful for these purposes but they
require:
a) considerable expertise; b) collec­
tion of data on water table levels, vegetation
and soils and/or; c) intensive laboratory analy­
sis of plant or soil material.
In short, a lot
of expertise, time, and/or expensive equi nent
are usually required to characterize soil aera­
tion or wetness zones within a wetland area
using current techniques.
McKee (1978) introduced the concept of using the depth of rusting on mild steel rods placed in soils with shallow water tables. "The 1
Donal D. Hook, Professor, Department of Forestry, Clemson University, Clemson, South Carolina 29634. 2Marshal1 D. Murray, Forester USDA Forest Service, Pacific Northwest Research Station, Olymp a, Washington 38502. Dean S. DeBell, Principal Silviculturist, USDA Forest Service, Pacific Northwest Research Stati n, Olympia, Washington 98502. Boyd C. Wilson, Geneticist, Washington State Department of r atural Resources, Olympia Washington 98502. s
theory behind the technique is that in a poorly
drained soil, an iron rod I'li11 rust rapidly in
the aerated zone but not in the saturated non­
aerated cr reduced zone. Thus, changes in the
appearance of the metal indicate the depth of
the aerated zone."
He found a strong correla­
tion {r
0.87} between depth of rust on an iron
rod and depth to the water table level among
several soil series under controlled waterlogged
conditions. Carnell and Anderson (1986) stated
that corrosion of iron in soil depends on soil
resistivity, redox potential, moisture content,
salt content, hydrogen uptake, soluble iron con­
tent, pH value, organic material content, oxygen
transfer and soil compactiDn. However, they
concluded that for any given soil the only fac­
tors that appear to vary over time and place
I'lere pll, redox potential, moisture content, and
oxygen transfer. Oxygen transfer was assumed to
have the greatest influence on corrosion rate.
Also, they found a close relationship beh/een
the depth of rust on a rod and·the depth of
rooting of Sitka spruce under controlled
I'la terlogged condit ions.
Hook et a1. (1987)
using the "rusty rod" technique were able to
separate the performance of several red alder
falid 1 ies on a wet site by the depth of rusting
on an iron rod placed by each tree.
=
Therefore, the use of depth of rusting on
iron rods placed in the soil appears to offer a
simple-inexpensive alternative to the more
complex methods mentioned above for characteriz­
ing soil I'le tness.
Results of the research I'lith
red alder on shallow water tables are reported
herein find offered as an example of how the
technique can be used to quantify the rela­
tionship between soil water table level and
performance of vegetation sensitive to soil
waterlogging.
Our results are also discussed in
r.!!l;)tion to otlH'r nata on nepth of rusting on iron
r(lds il\ IIf"!t ';oi 1 5 tn illustrate the strenuths and
''/f llknessses of this technique.
I1ETHOIJS
Experimental Site and lIateria1
A small alluvial flat of mature-mixen
flouu1as-fir and ren alder near IIcCLeary,
l/ater Tah1e Leve 1
Estimates
flild steel iron rods (about 89 cm long and
3.0 (TTl1 in diameter) vlere cleaned of 0; 1, IilX,
and dirt before installation.
Rods were pushed
into the soil within 15 em of each tree stem to
a depth of 75 cm or until a barrier was encoun­
In the wettest portions of the site,
tered.
rods were placed on two sides of each tree.
They were installed on tlovember 17-18, 1983 and
one rod was removed from the side of each tree
The depth of penetration
on February 6, 1984.
into the soil was measured as A and length of
unrusted portion of the rod below the soil sur­
face as B.
Depth to water table was estimated
hy subtracting B from A.
The second rod in
each set 11as removed on flay 24, 1984 and
measured in the same manner.
0 shington was 10UUed in the mid 1970's.
An
experilllental plantation of 33 open-pollinated red
alder families was planted on the site in tlarch
1979.
Prior to 10Uging there was no evidence of
Iaterlo\)\)ing problems on the site but since 10\)gging the lower lying areas (a Siffon,
gra­
ve111y silt loam soil, wet-variant) have been I'laterlo�HJed most of the dormant season each year and Some of the growinu season during the wetter .years. The surrounding high areas (a Dahob, very
yravelly loam soil series) have not been water­
loguen.
The -prevalence of \'Iater10\HJing in the
10l-/er areas since lo gging mi1Y have resulted from
redllced tra nspi ration losses brolluht about by the
rel'loval of the mature stand of mixed Douglas-fir
and red alder.
lIe restricted our analyses to trees on
microsites having water tables within 30 em of
the soil surface hecause:
(1) at some loca­
tions rods could not be pushed into the soil
more than 30 cm; (2) in
He restricted our
analyses to trees on microsites having \'later
tnh1es within 30 cm of the soil surface
hecause:
(1) at some locations rods could not
be pushed into the soil more than 30 cm; (2) in
other areas rods were pushed in deeper and the
entire length of the rod in the soil rusted,
hence, depth to the water table level could not
be determined; and (3) our primary interest was
to evaluate response to shallow water table
levels.
In addition, families with less than
13 observations of water table levels within 30
cm of the soil surface were excluded from our
analysis.
After all rejections, data from 24
families (consisting of 13 to 25 trees in each
family) was used in our investigation.
These
24 families represented six collection sites.
1\11 families includen in the plantation orignated within a 40 km radius of McCleary, I/ashi,ngton (see lIook et al. 1987 for description and location of parent areas).
The experiment was a rdandomized incon lete blOCk design with five blocks.
umber of families varied from 24 to 33 per block. -Howpver, all families used in this report I'/ere planted in all five hlocks and tire sample trees occurred in three or more blocks.
Each family was planted as a row plot consisting of six trees in each hlock.
Spacing between trees was 2 x 2 m and the lanting stock was 1-1. Survival was assessed after the first In the wetter portion of the plantation,
water table levels were found to be similar
during the 12-week, November-February and the
2fi-l'leek, tlovember-flay measurement periods.
Therefore, only the measurements from the
s horter period (November-February) /ere used in
the analyses.
gr0l1ing season.
tlorta1ity vias 5% and occurred randomly throughout the plantation.
Dead seedlings I'lere replaced with "surplus" seedlings that had been planted in the outside row of each block. A fter four years, variation in grOl-/th /as
evident between the low wet areas that dissected
the plantation and the higher areas.
HOI-/ever,
some families seemed to be affected more than
others. Since the rows of the plantation ran
neRrly perpendicular to the waterlogged areas,
most families had individuals within waterlogged
and non-waterlogged areas in each hlock.
Because
of the random nature of the planting and the
distribution of the waterloyued and non­
waterlogged areas in each block.
BecRuse of the
random nature of the plilntin!l and the distribu­
tion of the 11aterloyyed and non-llaterlogged
areas, Ie Here able to evaltlRt" illld compare the
growth of 24 families over a range of 0 to 30 c
'later table depths.
lIeight at the end of the
fifth grOl-linu season Ias IIsed as the dependent
variable in this experiment.
ANALYSIS
Relationship behleen tree gro /th Rnd depth
to the I-later table /ere examined using ,1 four­
step process:
(1) plotting of growth vnriahles
vs. depth to water tah1e to examine general
patterns of the relationships; (2) developinu a
correlation matrix consisting of growth
variables, (height, diameter, diameter squared
Pililtiplied hy height), depth to I-later table,
and the natllral logarithms of growth and 11ater
tahle variahles; (3) selecting the most
appropriate functions for describing the
general relation-ship between growth and water
table for each fanily and developing regres­
249
sion equation therefqre, (the selection of func­
tions or variahles was based on "the plotting and
correlations); and (4) testing the slope of line r
r('gressions of all fanilies by the assumpti(1n of
hOl'1ogeneity as a null hypotheses in an analysis of
covariance where water table levels was the
covariate.
lable 2.
selected fa.,llles.
Family
Humb r
Relght
Water labl! Level
30
0
em
em
flelght
Decrease
(m)
Bei ght
Decrease
Ho SignifIcant Response
10
62
RESULTS
Projected heights from the re re"lons of lable 1 at 30
em and 0 em ",tor table levels and projected decre"es In
height as the "ater table level changed from 30 to 0 em for
7.0
7.9
6.7
8.5
11
21
0.9
1.8
lin.ar Response ("eak)
9
42
The correlation between height and water
tlhle level was significant (r
0.62) when all
fanilies were included in the analysis.
"Iiollever, there vias considerable variation in
correlation responses among families. Slope
;lOpulations Here heterogeneous by the covariance
analysis, indicating the variation was real.
In
Hddition, three patterns of response were evi­
dent al'1ong the 24 families. First, three fami­
lies showed no significant relationship between
height and water table level. Their height did
not change significantly as the water table
changed from 30 cm deep to 0 cm deep (Table 1
Second, fifteen families showed
and Figure 1).
a linear decrease in height as the water table
changed frOM 30 cm to 0 cm. The response among
0.03; r
the e families varied from weak (slope
0.06) to strong (slope
0.18; r
0.83; Table
I and Figure 1 ). Third, six families showed a
r;'Jrvilinear response between height and water
t hle level (Table 1 ).
5.6
5.5
8.9
8.5
3.3
3.0
37
35
=
lin.ar Respone (strong)
8
67
9.7
9.8
4
11
7.9
7.6
4.3
4.7
5.4
5.1
56
52
Curvtl t nellr Response
'Percent
decrease ·
4.4
4.3
44
43
3.5
3.3
(ht. 3 0 e m - ht. 0 em x 100)/(ht. 30 em)
=
=
=
=
lable 1. Regression relationships betwe.n height .nd \later table level and
.,ean heights and water table levels for selected red alder families.
""pon lII:
-- non.
-..
Regressl0n Rehbonshlps
Family
Humber
Intercept
6,96
6.68
10
62
Slope
Coefl •
Fa"lly
Height
lIeans
lilL'
24
18
7.3
7.5
12.4
13.7
21
18
6.5
6.6
7.7
10.7
18
15
6.0
6.8
9.4
12.8
6.1
6.9
10.3
15.5
Humber
Cootf icient
P for
Observ.
Slope
of dete"".
Non-sensitive Families
0.06
0.07
0.03
0.06
0.25
0.28
tb
2'0
Water tabla lev,,! (em)
Hoderately Sensitive Families
5.62
5.54
9
2
0.33
0.32
0.11
0.10
0.01
0.01
nn.or
- curvilinear
3
Figllre l.--Projected regression relationships of
three types of responses of red alder fami
lies to shallow water table levels.
Highly Sensitive Families
4.26
4.71
8
67
0.18
0.11
0.83
0.82
0.00
0.00
DISCUSSIONS ANn CONCLUSIONS
ram1l1 es with Curvlll near Response
11
4
4.26
4.44
'\lTL
•
O. 98
1.01
0.62
0.54
0.00
0.00
18
18
The "rusty rod" technique proved to be a
sensitive indicator of water table level on a
single site in vlestern lIashington. Uater table
level estimates obtained by this method \ ere
related to the height of red alder at age 5
years bllt more importantly, such estimates were
sensitive enough to detect differences in fftmily
response to water table levels.
This level of
sensitivity should be qllite helpflll for most \ et
site characterization purposes. Hov/ever, Illore
experience is m!eded with the techni Cl"e in other
regions and for other seasons of the year hefore
its c pahilitip.s and liMit tions are fully
reftlized. For inst nce, this techniqlle hAS been
used by the seniur author in coastal South
Carolinil to dHterr'linE' the rclfttionship of thrl!e­
lIater lable level
Decrease in height varied from 2 1 % or less in
the no significant response group to 43 to 56
percent in the strong linear and curvilinear
r sponse groups over the range of 0 to 30 cm
(Table 2).
In this population height decreased
only slightly as the water table changed from 30
CM to 1 0 cm (about 1.0 m), but it decreased
sharply from 1 0 cm to 0 cm (about 2.9 1'1; Figure
1)
•
250
Y"M-old lnhlo1 h ;line -(Pinlls taed L.) height
'3rl)\lth to shall 0\/ \'latEr 1a1ilrleve-rs dllri ng the
. dor lant ann grov';l1 seasons. Generally, the rela­
tionship hetween height and water table level was
h st during the dor ant season and better on
wptter sites than on drier sites. Several factors
c(lu1n account for these differences. lIater tilh1e
levels tend to be more stable and tend to be
closer to the soil surface during the normant
season·. llhen I'/ater table levels are stable (i.e.,
I/hen they do not fluctuate I'ddely), the color de-.
marcations on rods are clearer and easier to
interpret. Also, higher water table levels place
a greater stress on the tree, thereby pushing the
plant-water table level relationship into a higher
stress zone than lower water table levels. 11cKee
(1978) pointed out that certain soil aeration
c haracteristics could cause differences in rusting
among soil series.
The correlation between height and water table level was not as strong for loblolly pine as it was for red alder. This may indicate that red alder is, in general, more sensitive to s hallow water levels than loblolly pine. To classify a site as a Section 404 wetland,
It rlust meet hydric vegetation, soil, hydrology
criteria. Soils and vegetation can usually be
quantified by hydric soil lists and color charts
and hydrophytic vegetation lists for specific
regions. Hydrology is usually more subjective.
Clues for hydrology generally depend on visual
observations of inundation or some evidence
thereof. The use of rust on iron rods placed in
the soil for short durations could greatly
improve the hydrology phase of the delineation
process.
The "rusty rod" technique is a reliable
indicator of average water table levels for spe­
cific periods of time and for specific sites.
Furthermore, the technique is sensitve enough to
determine subtle relationships between water
table levels and vegetation performance. Because
of its sensitivity, simplicity, and inexpensive
nature, it should prove to be a helpful supple­
mental tool for characterizing and delineating
wetlands.
L ITERI\TURE CITED
Spinale and 11cKee (1985) found that the for­
mation of ruston iron rods accurately measured the
average depth of seasonal water tables for the
evaluation of sites for septic tank absorption
fields. They concluded that iron rods were less
expensive in terms of material and manpower than
pieZtlr.leters and recommended them as a supp1ement
to the lise of conventional piezometers for deter­
mining seasonal water table measurements on imper­
fectly drained soils. As previously mentioned,
Carnell and Anderso (1986) found depth of rust on
iron rods to be strongly correlated with depth of
root penetration of Sitka spruce into a soil with
a shallow water table. Hence, the technique has
been tested in several nifferent environments and
for several different purposes.
SOl e limitations of the technique are: 1) the rons must be left in the soi1 for a fel'/ weeks to obtain an index of the water table clearly defined) and 2) they only provide an integrated index of soil neration (i.e., no quantitative easure of soil oxygen status, redox potential, etc. are ohtnined). But since it is so simple and inexpensive to use, it o ffers the opportunity to greatly increase sanpling intensity at a reasonahle cost. For instance, in the red alder data reported herein over 200 sampling points were taken and in the lohlolly pine study over 700 sampling points Such large samples with more con­
were tilken
ventional techniques would be prohibitive in cost and time. It appears that with Additional studies that quantify the relationships a ong soil srries, water table fluctuations, color on exposed rods, nrl minimu length of exposure to index a soil, th technique can be usen to c haracterize relative differences anong wetlann sites within a locale or rrgion. Armstrong, W., E. J. Wright, S. Lythe, and T. J.
Gaynard. 1985. Plant zonation and the
effect of the spring neap tide cycle on
soil aeration in a Humber salt marsh.
Ecol. 73:323-229.
Carnell, R., and N. A. Anderson; 1986. A tech
nique for extensive field measurement of
soil anaerobism by rusting of steel rods.
Forestry 59:129-140.
Clark, J. R., and J. Benforado (eds.). 1981.
lIetlands of bottomland hardwood forests.
385 p. Elsevier Sci. Publ. Co., New York.
Environmental Laboratory. 1987. Corps of
Engineers I'/etlands delineation manual.
Wet1anns Research Progam. Technical Rrport
Y -E7 -1. Depa rbne'lt of the Army, US Army
Corps of Engineers. Washington, DC 20314
1000.
Faulkner, S. P., II. H. Patrick, Jr., H. B.
Parker, E. l1altby, and R. Gambrell. 1986.
Characterization of soil processes in
bottomland hardwood wetland-nonwetland
transition zones in the lower lIississippi
river valley. 252 p. Contract No. DI\C\I
39-81-6-0032. Report to the Environmental
l.aboratory, tl. S. Army Corps of Engineers.
Ilaten ays Experinent Station, Vicksburg, I1S.
• •
Goon,
IL J., S. P. Faulkner, and 11. 11. Piltrick. Jr. 1986. Evaluation of green ash root responses as a soil wetness indicator. Soil Sci. Soc. I\ . ,I. 50:1570-1575. Hook, n. D., 11. 11. lIl/rray, D. S. DeBell, and B.
C. lIilson. 1987. Variation in growth of
ren alder families in relation to shallow
water tahle levels. For. Sci. 33:224-229.
251
Mckee, W. Hot Jr.
1978.
Rust on iron rods
indicate depth of soil water tahles.
In
Soil moisture-site productivity Sym. Proc.
area State and Private For.
r rtle, SC
in sewage disposal systems.
J. Env. Health
48:26-27 .
Theriot, R. F.
1987.
Flood tolerance indices
for Palustrine forest species.
In Ecology
and Hanagement of Wetlands. Vol:-1Ecology.
Spinale, F. G., and W. H. McKee, Jr.
1985.
Use
of iron rods to determine the depth 0'
seasonal water tables for aborption fields
D. O. Hook et a1. (eds.).
Croom Helm LTD
Publishers. Beckenham, Kent. UK. (in
press).
252