全 文 ：第1 9卷第1期
1 999年 1月
生 态 学 报
ACTA ECOLOGICA SINICA
Vo1．1 9，No．1
Jan．．199g
隆圭塞 H．Alexes Londo
河溪 结构分析
Rebecca A．Megown 张全发
W illiam J．Boelema，Ann M ．Hoefferle，Jacob J．LaCroix，
Andy J．Londo，Krista A．M arkovic，M ichele L．Olson，
Karen E．Owens
( chool of ores~ry and WoodProductsMich zganTeehno[oglcalU*~iverslty．Hoaghton MI，49931)
摘要 河毽结构及其关联的空间特征，是研究集水区中能量和物质穆动规律所不可缺少的内容 。利用美国地
质调查局的球 丈敬据库和地理信息系统(ARC／INFO)，对美国^陆5个主要 山地中典型集水区河溪的数量
密度 分布结构以及河溪岸边带的组戚 ．进行丁分析和 比较 5十山地舒别是 ：西北太平洋沿岸，西部卡斯卡特
山．中部落矶山，求 阿巴拉契 亚山和南部欧扎克山 发现这5个集水区中的河溪网络结掏非常相似 ．一级河
灌大约占诃溪总长度的6O 左右 东部 山地地质历史比较长-其集水区中河溪 营长度 密度 ．高差 厦河溪片
段敏量t都比西部相对年轻的集水区要高。此外．尽管萁中河溪边岸域 总面积随边岸域宽度线性增长．但它们
在面积 中的相 对比例仍然很小(
关键词 坷溪结构．山区集求区 妻目
STREAM STRUCTURE ACR0SS FIVE M 0UNTAIN0US
W ATERSHEDS IN THE C0NTINENTAL UNITED STATES
CHEN Ji Quan H．Alexes I．ondo Rebecca A．Megown ZHANG Quan—Fa，
William J．Boelema． Ann M ．Hoeferle Jacob J．LaCroix Andy J Londo
Krista A．M arkovic，M ichdle L．OIson Karen E．()wens
(School ofForestry andWood Prod~ts]l~ichtganTechnolo ca!Unitersity．Houghton，Ⅲ ．49931)
Abstract Stream geometry and related measurements are the baseline information needed
for ecological analysis of materials and energy movements within a watershed landscape．
Using the United States Geological Surveys(USGS)hydrography database and geograph—
ic 1nformation system (ARC／INFO )．we examined and compared the distribution of
streams．stream sections，total numbers，density，and riparian zones。in five m ountainous
watersheds in the continental United States：the Pacific Coastal Range，Cascades，Rockies，
Appalachian M ountains，and 0zarks．Stream networks were found to be similar among the
watersheds，with first order streams comprising up to 60 of the watersheds，The two wa
tersheds in the eastern US had smaller stream length，basin relief，and higher stream densi一
收稿 日期 ：】998 04 26
的
区
水
集
区 A 山
个
五
陆
大
国
美
维普资讯 http://www.cqvip.com
1期 陈吉泉等 ：美国大特五个山区集水区的河溪结构分析 3l
ty and section density than western watersheds．The amount of riparian area linearly in
creased with buffer width and did not account for larger amounts of the land area．Only
seven to twelve percent of a total watershed area was included in riparian areas when a
60m buffer width was applied．Riparian zone distribution was related to watershed geom or—
phology．Emprirical models were developed to predict the proportions of stream ．stream
density，and changes in riparian areas with stream order within each watershed．Applica
tions of these results in ecosystem analysis at watershed scales and riparian zone manage
ment are dise1】ssed．
Key words stream structure，watersheds
1 I IR0DUCTION
The importance of understanding ecosystem behaviors at large spatial and temporal scales for natural
resource management has recently received special attention in ecology，forestry，wildlife management_and
conservation Concurrently．watershed analysis．as pioneered by Bormann and IAkens[ j at the Hubbard
Brook Experimental Forest in New Hampshire，has been widely accepted as both a logical approach in e
cosystem science and as a basic unit in resource management ．Scientific investigations of many ecological
processes(e．g．，evap。transporati0n，mass movement，disturbance，vegetation dynamics of watersheds)are
especially recommended and sought in the literature[ ～ ．In practice—the Forest Ecosystem Management
Team[ 。uses the watershed approach with a central focus on stream networks and the associated riparian
habitats region wide as a part of integrated ecosystem management[ 。”．
The structure and function of a stream and riparian ecosystem are key issues in a watershed study
Many authors have suggested streams and riparian areas to be the “blood system”and “hot—spots”of a
landscape 9l ．These terms originated because riparian areas form a highly connected network which pre
dominarely affects the overall function of a landscape．Examples include distribution and movement of
species，nutrients．sediment，cumulative effects，and disturbance events ’“]．The development of“river
continuum concept”by Vannore et d，，and others[ ’’ 一 ．which defines the existence of gradients of ecosys
tem composhion，structure，and function as one moves from the head waters to large rivers．provides a n2e—
chanical framework for exploring stream functions．
Recognition of the importance of stream networks ln landscapes has also lead to scientific interest in
the examination of riparian zones and their influences on both aquatic and terrestrial ecosystems．Riparian
zone functions include providing shade，fine and coarse organic materials，nutrients，stream sediments，and
diverse habitats for p[ants and wildlife—and corridors／or seed and animal dispersal。。 i s,~2 3．Management ef-
forts to help riparian zone functions include leaving riparian buffers OD both sides of a stream during timber
harvest and avoiding detrimental buman activities．These management efforts have become a generally
practied management guideline in North America with the area involved sometimes referred to as riparian
management zones ．For the Pacific Northwest．FEMAT proposes that buffer width be dependent upon
stream order，geological settings．and surrounding topography．The buffer width may vary according to geo
morphic features which has resutred in the proposed buffer width of 50ft to 300ft(】．e ，≈ lO~ 50m )．Man
agement objectives and guidelines concerning riparian zones are optional and differ between timber indus
tries and environmental groups ．
Despite the theoretical and practical debates on management criteria／or buffers，there is a lack of basic
●
-
‘
维普资讯 http://www.cqvip.com
32 生 态 学 报 19卷
information about the structure and spatial distributions within a watershed or landscape．For example，
how many streams and how much riparian areas given specific widths are in watersheds?Are stream seg—
merits proportionaly divided by stream orders?For mountainous areas．what Elre the roles of basin relief(1．
e．，elevation)in characterizing stream and riparian networks?
Using public—domain hydrological and elevation databases maintained by the United Staes Geological
Service(USGS)，this study was initiated to examine the stream networks of mountainous watersheds in five
major mountain ranges in the continental United States：the Pacific Coastal Range，Cascades，Rockies，Ap
palachian Mountains．and Ozarks．Specifically our objectives were to：(1)compare the differences and simi一
1arities of stream networks】n mountainous watersheds (2)evaluate the influences of streams and rivers by
order in the above landscapes(i．e．．elevation)by developing width varying riparian gones；and(3)provide a
conceptual framework for understanding streaFfl8，riparian zones，and their potential applications in ecologi
col research or natural resource management．Our working hypothesis was that streams and stream net—
works of the five mountainous 1andscapes have similar structures and influences( ．e．，riparian zones)across
watersheds in the above range s．
2 M ETH0DS
2．1 Study Areas
Five watersheds were included in this study(Fig．1)：the Queets River watershed in the western pot—
tion oftheOlympic Peninsula(W A)．theMcKenzie Riverwatershedinthe central Cascades(OR)，the Stil—
water River watershed in the northern Rockies(M O)，the Current River watershed in the western Ozarks
(Mo and AK)，and the Litter River watershed in the southern portion of the Appalachian M ountains(TN
and NC)．Selection of these watersheds was based on extensive research determined in the past and 0n—go—
ing research projects at these sites．
The Queets River watershed in located on the west end of the Olympic Peninsula(124。10 N and 47。40
W )．The climate is mild，between OC and 25U and rarely freezes．The yearly average rainfall is close to
500cm．Elevation ranges from sea level to 2434m on Mount Olympus over a horizontal distance of 55kin．
The high rainfall and effects of the Pacific Ocean jnintly produce the lare temperate rain forest that is dom-
inated by sitka spruce(Picea sitchensis)，western hemlock(Tsuga heterophylta)．and western redcedar(
plicata)．Donglas—fir(Pseudotsuga menzisti)t subapine fir(Ahies laciocarpa)，and Alaska yellow cedar
(Chamaecyparis nootkatensis)beeome more dominant with increasing elevation．The unique active glaciers，
maritime climate，rainforest，and sol1 provide diverse habitats for flora and fauna．
The M eKer~ie River watershed is loeated in the densely wooded，central portion of the western slope
within Oregon s Cascade Range(44。1 5JN and 122。10 W )．Elevation ranges from 420～ 1630m and the ter
rain is extremely rugged with steep slopes and deeply incised streambeds．The area has a quasi—Mediter-
ranean climate with mi ld moist winters and warm dry summers．The average annual temperature is 8．5℃．
Only six percent of the~lean yearly precipitation(230cm)falls from June to August．Most soils in the area
are classified asInceptisols，but someAlfisols are present．These soils are highly porous．with 60 ～ 70
porosity in surface soil and 50 ～ 6O porosity in subsoils．High porosity a[so provides storage for 3O～
40cm of water in the upper 120cm soil，which serves as a water source for the forest during summer
drought⋯ ．Vegetation in this area is stratified into two major zones，resulting from the altitude tempera—
ture gradient．The Tsuga heteropyla(Raf)．Sarg．zone is generally below 1050m and has abundant western
hemlock and Douglas fir[ ．The Pacific silver fir(Abies amabihs Doug1．ex Forbes)zone is found above
l 050m ．
The Stillwater River watershed(4 8030 N and ll4 42 W )is located in the northern Rockies and serves
维普资讯 http://www.cqvip.com
1期 陈吉泉等 ：美国太陆五个 山区集水区的河溪结构分析
as a drainage for the Beartooth M ountains．Starting elevations of the headwaters(1ocated 40km south of
Cooke City，M T)are in excess of 3300m．Precipitation is mostly in the form of snow at or near the headwa—
ters and exceeds 247cm annually．The Stillwater River flows northeast for approximately 105km hefore
emptying into the Yelowstone River east of Columbus，Montana．The elevation of this juncture is estimat—
ed to be 990m with annual precipitation here being only 35cm．Geologic formations of the watershed are
comprised of some of the oldest known rocks．Vegetation along this watershed varies greatly and is mostly
determined by elevation．Lodgepole pine(Pinus contorta)is the most frequent and dominant species in the
watershed with subalpine fir(^ ．taciocarpa)and whitebark pine(P．atbicaulis)in the subalpine areas．M uch
of the 1odgepole ar~rl is in a regenerative state due to the 1988 Yellowstone fires．Over 55 680hm of lodge—
pole forests in the Stilwater watershed were burned and are recently showing signs of regeneration．
The Current River watershed is located in the eastern portion of the Ozark M ountains in southeastern
Missouri and northeastern Arkansas(36。1 5 N and 90。33’w )．The region has a humid continental climate
with hot．humid summers and cool winters．Average annual precipitation is 112cm，the majority being rain
during spring and summer ．Dolomitic limestone imbedded with large quantities of chert dominated of the
watershed．The soil is day or clay loams containing chert on the surface[ ．Southern hardwood forests re,-
main relatively unfragmented ，and importain species include oaks(0uercus spp．)，hickories(Carya spp．)，
and shortleaf pine(Pings echinata)．
The Little River watershed is one of 45 m jot watersheds in the Great Smoky Mountains(83。35 W
and 35。40 N)．Aquatic resources include gradient from extremely soft—water．first order streams to mole
buffered third and fourth order rivers．The Smokies are located along the border of Tennessee and North
Carolina and lie along the backbone of the southern Appalachian M ountains．It is underlain by a diversity of
igneous，metamorphic and sedimentary bed rock types．The climate is characterized by high precipitation(>
150cm per year)，co l summers，and co ld winters．Soils are shallow Inceptisols with a thich organic hori-
zon．The Great Smoky M ountains are also rich in plant and wildlife diversity．Approximately 1200 native
vascular plant species and about 130 trees．450 known bryophyte and over 3400 lichen species【 ]exist
there．A variety of forest cover types are distributed within the watershed．including the largest undis—
turbed afea of the remnant red spruce—fraser fir(P ∞ rgbeJ,z$，and Ahies frsaeri)forest 1n the world 。 ’”-．
2．2 Data Acquisiton and Analysis
Our major data sources included 1：100 000 scale digtal line graph(DLG)of hydrology and 1：250 000
scale digital elevation m~dels(DEM )of the GeoDatahased maintained hy the USGS through electronic list-
serve．These data were retrieved using anonymous File Transfes Pr0toc01(FTP)，deco ded，and imported in-
to an ARC／INFO Geographic Informations System (GlS)on a UNIX platform．Originaly，these data were
stored in eight files per pixel of oned egree regulary speced intervals．ARC coverages imported from these
files were merged into one file for each basin．Using USGS 1：100 000，scale topo graphy maps as a hard—
copy reference，dangling arcs+lakes·reservoirs，man—made canals and other anomalies were removed．
Stream order was then assigned to each stream section based on Strahler L ．Standard buffer widths of 10．
20，3O，45，6O，90 and 120m were applied to sections across the watersheds to generate riparian coverages．
Varisble widths were used to examine the amount of riverine forests in the watershed／landscape under dif
ferent management scenarios that are currently applied on public Lands．
The DEM 1attice was used t0 delineate watershed boundaries using FLoW DIRECTION and“W A—
TERSHED commands of the GRID Module in ARC／INFO This bo undary coverage was used to calculate
watershed size and to clip a’DEM of the watershed．The DEM lattice then was used to generate an eleva—
tion coverage，which was later unioned with stream and riparian coverages to explore the spatial distrihu—
维普资讯 http://www.cqvip.com
34 生 态 学 报 19卷
t[on of streams and riparian zones in each watershed．All coverages and lattices were projected to the Uni—
versal Transverse M ercator(UTM )coordinate system
Watershed size，stream length by order，area of riparian zot~es，and their spatial distribution of streams
along the elevation were downloaded into ASC Ⅱ ／ties from the INFO Module for further analysis．Abso—
lute and relative length of streams by stream order tand area of riparian zones by seven widths and eleva—
tion were tabulated and summarized using Statistical Analysis System (SAS)．Linear regression analysis
was completed to develop empirical models to predict the changes in stream length，stream density，and area
of riparian zones with stream order in each watershed．
3 RESULTS
The five mountainous watersheds are clearly different in their size(423~ 6888km )，shape，basin relief
(4308 to 3142m )一and stream density(Fig．1．Table 1)．The Current River and the Little River watersheds
are two extreames in their size and topography．The Curreni River watershed is the largest(6888．04km )
with the smatlest retief ratio，while the Little River watershed is the smallest(423．47km )with Larger relief
ratio．The Current Rriver watershed has the lowest elevation difference(408m)compared to a drop of over
3000m in the M cKenzie River watershed(Table 1)．The two watersheds(the Little and Current rivers)in
the eastern states seemed tO have significantly higher stream density，shorter streams thigher section densi—
ty，and smaller basin relief than the three western watersheds．A simpte logarithmic linear regression
proved to be an effective mode[to predict the changes in streHm length and density for each stream order
(Table 2)．
Table 1 W atershed locationtareatelevation．1engthtand density calculations for five mountainous rtp4wian
areos of the continental Unlted States
Table 2 Parameters from logarithmic and linear regressional analysis asing buffer width independent
variable to estimate pro portion，stream dens ity，and propo rtion riparian area of the specified watershe d
①Model used：ln(y)=bo+btln(x)一￡
@Model used： =bt +￡
维普资讯 http://www.cqvip.com
1期 陈吉泉等 ；美国大陆五个 山区集水区的河溪结构分析 35
Queec
Fig 1 Geogaphie distribution of streams in flve mountainous watersheds of the continental United States
Two of the smaller watersheds(Queets and Little)extend to fourth order streams only．while the Still—
water and Current extend tO sixth—order streams(Table 3)．As expected，the total length of strealLs was
linearly related tO basin size(R =95．8)and the proportion of streams in all five watersheds decreased expo—
nentially with stream order(60 ，18％ ，and 1 0H for first．second，and thrid order streams，respectively)
(Fig．2a)．There is nO clear difference in the distribution of streams hy order among the five watersheds
Cha nges in stream density with stream order also show exponentially decreasing trends at these watersheds
but the Current and Little River watersheds have higher density for the first order streams(Fig．2b)．
Stream density for higher order streams among the five watersheads WSS not different．
The amount of area contained within the 7 buffer zones increased linearly with ha sin size (R =95．4
～ 96．5 )and buffer width in all five watersheds(Table 3)．For example，in our smallest watershed ( ．e．，
the Little River)．usinz Bmedium bufferwidth of 60m or about 200ft．there are 5235hm。of riparian area．
Riparianarea increased tO 79350hm in the Current River watershed，which is the largest(Table 4)．Two
distinct increasing patterns in hnffered area with burfer width were apparent as noted in stream length in
Fig．2b．Changes ix proportion of riparian area of the watershed are greater in the two eastern watersheds
than those in the west(Fig．3)．Our confidence in predicting the amount of riparian area hy stream order
was greater than 99．9 (Table 2)．
Spatial distribution of stream and riparian areo~in the watersheds indicate a very unique pattern with—
in the Current River watershed(Fig．4 ．The nmjority of riparian areas are distributed in the mid to lower
维普资讯 http://www.cqvip.com
生 态 学 报 19卷
areas of the 1andscapes and decreases exponen
tially with increasing in elevation．In the Cur
rent River watershed，however．the distribution
of riparian area in the landscape is relatively
even within the range nf 200m to 400m in ele—
ration with smaller amounts extending into the
high and low extrerles．The above relationship
appeared to be independent of buffer width．
Similar distribution of land areas along the ele—
vatiorml gradient in each watershed existed
(Fig．5)+indicating that mountainous streams
are independent of spatial locations and eleva—
tion．
4 DISCUSSION
Streams and related concepts and theories
ate becoming the key components in ecology as
we extend our interest in ecosystem processes
to broader scalesE Many issues，such as water
quality control and maintenance of biological di—
versity．cannot be properly addressed without
baseline informa tion on the ometrical struc—
ture nf stl~eam networks in a watershed[ ．A
combination nf the DLG information from US-
GS GeoDatabase and applications of geographic
information systems are promising for quantify—
ing stream structure and associated riparian
zones at the watershed．1andscape，and regional
scales[ ．Completion of the GeoDatabase coun
try—wide in the near future will allow us to sys—
tematica lly examine the stream networks of
each geological region．
A limitation of using the GeoDatabase to
delineate stream networks is that HlOSt first and
second order stream s are not [ncluded in the
database because nf its coarse resolution(i．e．，1
：100000)．0ther more accurate approaches．
might involve delineation nf stream networks
Fig 3 Proportion of watershed area included for each
buHer widths for five mountainous watershed from the
continenta【Uhi red States
using high resolution digital elevation models
based on geomorphic and hydrologic process—
es[ “]
． mappingtopographic cues on fire scale maps(i e．，1：24000)，or resorting Io field mapping to
ground truth the date lB]Despite the controversial debates 0n various models and approaches㈨ ，the Geo—
Database is the only available database to provide relative good quality and consisten prec[son when making
com parisons across regions．
维普资讯 http://www.cqvip.com
1期 陈吉泉等 美国大聩五个山区集水区的河溪结构分析
Table 3 Stream length by order and total stream length for five mountainous watershed a，e8s in the confi．
nenlal Unlted Stales
Table 4 Watershed area(km。)calculated for five mountainous watersheds mstng 10，20，30，45，6O，90 a叫
120m burfer dths
；
： j
Fig．4 Total a a distribution by elevation of each buffer zone for the live mountainous watershed investigated
Our results may therefore be significantly biased due to missing smaller streams in the GeoDatabase．
For the Coastal range and Cascade M ountains in the Pacific Northwest，the FEM AT teamE l reported that
stream density ranged from 1．82 tO 5．56km per square km based on 53 small watershed s in W ashingtonand
Oregon．Results from the GeoData base for the Queets and MeKenzie River watersheds are only 0．652 and
0．623kin／kin ，respectively(Table 1)，suggesting that the GeoData base probably missed at least 60 of
维普资讯 http://www.cqvip.com
生 态 学 报 l9卷
small streams．As a result，the amount of smaller
streams and associated riparian areas reported in the
study is underestimated．However，it appears that the
pattern of changes in stream length by order remains
the same，suggesting that stream networks constructed
using the Gee Database are valid and comparisons can
be ma de among our five selected watersheds．Clearly，a
high quality stream network database is the key for fu—
ture studies relating stream networks and their rela-
tionships to various ecosystem processes and manage—
m ent practices．
Consistency in propo rtion of streams by order
sugge sis mountainous streams have very similar struc—
tures(Fig．2a and Table 4)，regardless of watershed
size and differences in stream density(Fig．2b)among
the five mountainous watersheds． There results
matched precisely with findings in coastal watersheds
§
§
善
主
Fig．5 Distribution of land a1．ea relative to basin size a
long the elevation．The x axis ents a relative ale—
ration gradient which is computed using elevation data
as：(maximum—minimum )，(basin relief)
of western W ashington where the percentages of first and second order streams are about 6O and 18 ．
respectivelym]．These resuhs support otlr hypothesis that streams are organized in a similar structure
across mountainous watersheds． ]related these similarities to a higher level of processes(i．e．，self-organi
zaticn)．W e argue that basin morphometry，such as the size and drops of sub basins and slope，may have e-
qual contributions 。 ．Obviously，conclusions cannot be made without including more watersheds and analy—
ses in similar and different geological settings．Similarly，our conclusion that the western watersheds have a
lower stream density also needs further research．
Special attention needs to be paid to the spatial distribution of streams within the context of elevation
in the watersheds．We found that a majority of streams were distributed in the lower elevations of the
Queets．McKenzi，Stillwater+and Little watersheds，where there are steep slopes and larger elevation
drops．However，the streams in the Current River watershed were more or less evenly distributed in the
landscape ．1argely because the geological settings are more uniform(e．g．．smaler basin relief，Fig．5)com—
pared with the other four watersheds[ ．Nevertheless，high correlations be twen the spatizl disribution of
riparian zones and total land areas．along elevation gradients(Fig．4 and Fig．5)suppo~s a conclusion that
predictable models can be develope d based on the size of the sub—watersheds in a basin；that is the relation
of reparizn zone with elevation is correlated with the total watershed area．
Direct applications of our results on stream structure in nutural resource management are muhi-foId．
Parameters estimated from this study．such as Stl~am length，basin relief，and relief ratio，are commonly
used to study basic geomorphic processes ．For example，mean annual runoff was suggested to be a lin-
ear function of stt~eam density；sed iment discharges can be calculated by basin size and relief ratio[ ]{and
the number of species，amount of suitable wildlife habitat，or catchment of materials by riparian vegetation
from terrestrial lands into the stream may be estimated based on the amount of rif，arian zones in the land-
scapeI
．
In ecology．extens ive efforts have been made to study the distribution of coarse woody debris[ 。 ·
aquatic species．productivity／respiration rate ” ·and nutrients along a continuous river network by se-
lecting and sampling a number of streams in the landscape．Results from these empirical studies can be
维普资讯 http://www.cqvip.com
l期 陈吉泉等 ：美国太陆五个 山区集水区的河谭结构分析
readily extrapolated to watershed scales using the stream geometry findings of this study-For example·
overlaying stream and riparian networks on top of a vegetation coverage will provide us with first hand in—
formation on the coatributions(e．g．，organic matter and quality)of each vegetation type to the watershed．
Such informarion is critical in forest practice when alternative management scenarios are developed and ap—
plied at landscape levels．At regional scales，estimation of stream flows，sediment sccumulatinn·and dis—
charge regimes are also possible using the empirical results(Fig．2～ 4)and models(Table 4)developed
here．Additional information 0n watershed size and distributions in a region can also be obtained through
the results of this$tudy[ ．
As denoted by high species diversity．riparian zones serve as important pools and corridors for the
plaⅡt and wildlife spe cies within landscapes．Gregory et a1．㈣ reported that total number of plant species
Dtaks 1n rioarian zones and decreases as one moves away from the stream．Nilsson found 13 ot Sweden’s
vascular Dlants are distributed along the streams and rivers．Naiman a1． also repo rted that 68 of
plant species sampled along gradients from small streams to the uplands in western W ashington were loeat—
ed within 10m of the stream．This suggests a majority of species can be properly ma intained by managing a
small proportion of the landscape(i．e ．the riparian zones)．
A frequently debated issue in forest landscapes concerns manipulations of riparian vegetation[ ]．
These areas contain high timber production but play crucial roles in maintaining water quality control and
providing diverse habitats tor wildlife，and so{orth ’ ．A general argument is that preservation of riparian
forests can only be ma de at the COSt of timber production．W hile such an argument is valid—economic gains
m ust be balanced witk ecological losses．More efort has to be ma de 1o quantity the amount of riparian
∞ nes in a landscape．For example．many management plansL recommend very narrow riparian buffers(<
6Ore )for reservation，indicating a minor proportion of the land areas will he omitted when calculating tim—
be r production．A management plan setting aside 21 of forests for reserve allows us to leave 60m buffers
f0r aI】streams in the Current or the Little watersheds in the eastern United States．and more than 100m
buffers in the Queets，M cKenzie．and Stillwater watersheds(Fig 3)．
Acknowledgm ents
W e thank Eric Heitzman．Christopher Pappas，Glade Soward．Kim Brosofske，Conghe Song，M ing Xu·
and Margaret Gale for their constructive comments On earlier drafts 0f the manuscript and help with data
ana lysis．This study was partially supported by USDA Forest Service Grant Contract No PNW 93—0314，
M issouri Department of Conservation(M DC)，and the Mclntire—Stermis Funds at Miehigan Technological
U niversity．
LITERATURE CITED
Like~ G E，Bormann F H ，Johnson N M ．etⅡ^ Eff~ ts of forest cutting and herbicide treatment On nutrient budget in
the Hubbard Brook Watershed ecosystem Ecologica1．?vtonographs·1970_40：船 ～ 47
Nai~ n R J (ed1．Wa．ters~d managem~t：balancing sustaina~lity and environ~ ntal change．Springer—Varlag，NY ，
USA．1992．542
Running S W and Coeg hlan J C A general model 0 forest ecosystem processes for regional applications I-HydrologJ
calhe lance，canopy gas exchange and prio ry production processes．EcologicalModelhng，1988，42}125~ 154
Band L E，Peter~onD L Runnning SW ．et al Forest ecosystem processes atthewatershed scale；basisfor distributed
simulation EcologicalM odelling·1991，56：】17I～ 196
／ohason B L．Richardson W B and Naimo T J．Past，present．and iuture concepts in large river ecology．BioS~ence
1995，45(3)：134～ 141
Forest Ecosystem M anagement Assessment Team
维普资讯 http://www.cqvip.com
生 态 学 报 19卷
and ~ iat d “啪 ￡ ．Portland，OR，USA 1993
7 Grant G，McCain C and Cissel J Summary of the watershed landscape analys~ workshop：H．J．Andrews Experimen—
tat Forest．．USDA General Technical Report，PNW GTR 338．1994．67
8 BergD R．Riparian silvicultural system design and assessmentinthePacifk NorthwestCascadeM cun~ins ，USA
．
colagicalApphcati~s，1995，5(1)；87～ 96
9 Stanford J A and Covich A P Commuuity strt．ctu~ aud fu t1oⅡin temparste t砷p峨 l Stl：eBAt~．Journqf q，‘“
North A m Benthological S~-iety，1988，7(4。】：261～ 529
10 footmanR T T．L口 如 ：珀f ecology oflandscapes and resins．Cambridge University Press．New York．USA
1995．832
11 R~ser PG n le 8tatug ofthescience examining ecoto~es．BioSol~ce，1995，4S(5)13l8～ 325
18 VannoteR L，Minsha]lG W ，Cummins K W a／ The⋯ r continuum concept．Canadian Journal 0，Fisheries and
Aquanc Sciences，1980，” -l30～ 137
13 Minshal[G W ，Cummins K W ，PetersenR C，et a1．Developmentsin~tream ecosystem theory．Canadian Journal口，
胁 cries删 Aquanc Science．1985，42：1045～ l055
14 Naim~ R J，Mellto JM ，LockM ， Ⅱf．Lgag]t~dinalpe ttex~s ~cosystem p~ocesses and commungty H 札 弛
a subarctic river continuum．髓 ，l987，68(5)：1139～ 1188
15 Ni L son C，Gretsson G ，Johansson M ，et a1．Patter~ of species richness along riverbanks．Ecologyt1989，70：77～ 84
18 Cummins K W Structure and fus tian of s【r㈣ ecosystems．BioScience，1974，24：831～ 641
17 Gregory S V，Swanson F J，McKee W A．et ．An ecosystem perspe ctive of rlpe rMn删 s．BiaSd~ce，1991，41(8)．
54O～ 551
18 Chen J．Riparian vegetation andtheir rolesin stream ecosystems．ChineseJournal of@ pliedEcology，1996，7(4)：489
～ 448
19 Brosofske K D，Cha n J Naimaa R J． al Effects of hat． sting on mleroclimatk gradients from stⅧ to uplands ln
weste【n"~rashingtoa USA．．Ecotoglcat Appl6catioas，1997，7(41 t1188～ l20O
铷 Amour C．ch— n 。n硼 RMz’s and UM ㈣ Ⅷwith 姆 i to u d“ bah／tat 1988 field report．Washington D}
partment of W i Ld Life，Olympia，W A．1989
21 Bierlmaler F A and M eKee A．Climatic summaries and documentation for the primary meteoro]ogTcal station．H．J
And s Eeperimenta[Porest，1972— 1984．USDA—FS Ge~ ral Techmcal Report，PNW —GTR一242，1989
22 Franklinl F andD m s C1 Natural vegetation ofOrego~and Washington USDAF．S Technical R PNW一
118．1973
船 Rafferty M D．The Ozarb —land and life．Urdversity of Oklahoma PresstNorman，oK，1990
24 Sauer C 0．The geography the Ozark highland of Missouri．Unlversity of Chicago Pros，Chicago，1L．1920．237
85 No~vinSC，Ri ga f J S audTwigsSM ．An indexed f㈣ database theGreat$mokyM~untaitxs，NorthCarolina
and Tennessee Nao~al Park Semite—South Regi~．T~hnical Report NPS／SERGRSM ／NRTR～93／08t NPS D
413．1993
26 Whjte P S．Natural disturbance and gap phase dynamics in southern Appalachian sprucefir forests．Canadian Journag
F删 Research，1985，15：233～ 240
27 BushingR T and W uX．Size—specificmortality，growth，and s【 ucture of a great SmokyM ouaUtins red—spruce popula—
t n．Canadi4 Journal 0厂ForestRe s~rch，l989，20：206～2l0
28 Strahter A N Dynamics basis of geomorpho]ogy Geological Society of Ampca Bu~lennt1958，63：923~928
29 Bren L J．Aspect of the geometry of riparian buffer strips and its significance to forestry operations ．Forest Ecology
and Management，1995，’5： l～ 10
30 M ontgomery D R，and Foufoula—Georgiou E．Chan el network gou[ce representation using digital elevation models．
W n r R Re uarch，1993t29(12)：3985～ 3934
31 lehoku C and Chorow~cz J．A numerical approach to the analysis and classification of cha nnel aetx~ork patterns．W arer
Reso~ces Researeh，1994，∞ (2) l61～ l74
维普资讯 http://www.cqvip.com
1期 陈吉泉等 美国大陆五十山区集水 区的河溪结构分析 41
32 Zksng W Montgomery D R·Distal e]evation model grid size-landscape representation．and hydrelo~cal simuh~
tions．Water Resources Research．1994．30(4)：1019～ 1028
40
41
Nsiman R JFetksrston K L McKsy S· a1．RJverine Forests．In Naimsn，R J．and R．E．Bflby(eds．)tEcology 4’
M蚰 gn删 S~Pams 4Id Ri~zrs in the Panic Northtoest Coamal Ecore#on．Springer—Verhg．1998
Stolum H H．Rivermeandering as a self-organization process．Sc／tace，1996。271(22)：1710~ 1713
RitterDF·Kc~helRC and Miller J且 Proces geomorphotng-j．Wm．C．B~x,ra Publishers，Dubuque，I A．1999
ToyT J andHadleyRF．Geomorphology and reclamation ofdisturbedlands．AcademicPros．NewYork．USA．1987．
480
HadleyR F and Schumm SA．Sediment aour。髑 and drainagebasin characterisitiesin upperCheyenneRiver basin：U．
5．C．eolo#calSurveyWater—S lyPaper，1961，1531-B
Swanson F J FrndrickseD R L．，McCoison F M．Materia】transfer in western Oi"~on forested watershnd．b1t Ed-
mend s，R．L．(ed．)，analysis of coaiferc~s ecosystemsinthe v~ternUnited5抽 ．US／1BP Syathesis Series14．
Hutehinso Ross Publishing Company，StroudsburgtPA．USA．1982
Swanson F J Neilson R P，and Grant G E．Some emerging issues in watershed management|landscape patterns，
species conservationtand climate change ．307 323tIN．Nairnz~，R．J．(ed．)tW atershed managem~~ ibalanciMg 5145一
~inability 删 棚 r。鲫 m删抽 change．Springe r—Verlag．New York。USA．1992
Fetherston K L·Nalman R JtBilby R E．Large woody debris，physical process，and riparian forest development in
montane river networks of the Pacific Northwest．C．eocnorpho／ogy，1995，131133~ 144
Nairn~n R J and And erson E C EeoLogieaL imp]icatiorm of]atitudinal variation in watershed size，stream flow．and 丑_
ter temperctureinthe constaltempe rate rainfo t otNorthAn~ iea．In|SchoomakertP．，and B Von FLagen(ed s．)．
the rainforests homel口 exploration of and place．1shnd P ∞stWashington．n C．，USA 1996
NaimanR J’FetherstonK L·Mckay S andChen J．1~ erineforests．InNaimsnR J and Bi1 R E，eds．Ecology and
Mana ge ment of Streams and Rivers in the Pacifc Northwest Co~sta]Ec0r 。n Springer Ver lsg，1998
驺 ；宾 钾 黯
维普资讯 http://www.cqvip.com