Nitrogen fixation in subarctic streams influenced by beaver (Castor canadensis)

Margaret M. Francis’, Robert J. Naimant & Jerry M. Melilld I Woods Hole Oceanographic Institution, Woods Hole, MA 02543, U.S.A. 2 The Ecosystems Center, Marine Biological Laboratory, Woods Hole, MA 02543, U.S.A.

Keywords: acetylene reduction, beaver, Castor canadensis, nitrogen fixation, streams, subarctic

Abstract

Nitrogen fixation was measured in four subarctic streams substantially modified by beaver (Castor canadensis) in Quebec. Acetylene-ethylene (C2H2 - C2H4) reduction techniques were used during the 1982 ice-free period (May-October) to estimate nitrogen fixation by microorganisms colonizing wood and sediment. Mean seasonal fixation rates were low and patchy, ranging from zero to 2.3 X 10m3 pmol C,H, . cmm2 . hm* for wood, and from zero to 7.0 X 1O-3 pmol CZH, . g AFDMmt . h-i for sediment; 77% of all wood and 63% of all sediment measurements showed no C2Hz reduction. Nonparametric statistical tests were unable to show a significant difference (p > 0.05) in C2H2 reduction rates between or within sites for wood species or by sediment depth.

Nitrogen contributed by microorganisms colonizing wood in riffles of beaver influenced watersheds was small (e.g., 0.207 g N . mm2 . y-l) but greater than that for wood in beaver ponds(e.g., 0.008 gN . m-2 . y-t) or for streams without beaver (e.g., 0.003 g N . m-2 . y-l). Although mass specific nitrogen fixation rates did not change significantly as beaver transform riffles into ponds, the nitrogen fixed by organisms colonizing sediment in pond areas (e.g., 5.1 g N . m-2 . yi) was greater than that in riffles (e.g., 0.42 g N . m-z. yi). The annual nitrogen contribution is proportional to the amount of sediment available for microbial colonization. We estimate that total nitrogen accumulation in sediment, per unit area, is enhanced 9 to 44 fold by beaver damming a section of stream.

Introduction

Beaver (Castor canadensis) significantly modify the structure and dynamics of streams through their wood cutting and dam building activities (Naiman & Melillo, 1984). An individual beaver annually cuts a metric ton of wood for growth and maintenance(Howard, 1982), usuallyselectingspe- ties such as aspen, alder or birch which are high in nitrogen (Melillo et al., 1983). Alterations to the nutrient cycle may result as wood and sediment trapped behind dams become more available for microbial colonization. In nutrient poor boreal forests nitrogen fixation may be a major pathway for sequestering nitrogen, with streams modified by

beaver presenting suitable conditions for fixation to occur.

Recently, biologically fixed atmospheric nitro- gen in terrestrial and aquatic ecosystems has been measured using the acetylene-ethylene reduction technique (MacGregor & Keeney, 1973; Vander- hoef et al., 1974; Flett et al., 1980; Waughmann & Bellamy, 1980). Several studies of stream systems have examined nitrogen fixation associated with allochthonous litter inputs (Buckley & Triska, 1978; Ashton, 1979; El Samra & Olah, 1979; Dierberg & Brezonik, 1981; Tam et al., 1981). Al- lochthonous litter acts as an available colonization substrate and as a carbon source for nitrogen fixing organisms. Nitrogen fixation rates associated with

Hydrobiologia 121, 193-202 (1985). 0 Dr W. Junk Publishers, Dordrecht. Printed in the Netherlands.

beaver mediated litter inputs and accumulated sed- iments in streams have not been investigated.

The objectives of this study were to measure nitrogen fixation rates in subarctic, boreal forest streams and to determine if beaver establish situa- tions where substantial nitrogen is sequestered by the ecosystem.

Study areas

The study sites are located -25 km east of Sept- Iles, Quebec, on the North Shore of the Gulf of the St. Lawrence (Fig. I). Streams here are character- ized by low concentrations of dissolved nitrogen (0.1 to 0.3 mg N:L, as nitrate) and phosphorus (-0.003 mg P:L, as orthophosphate), and have dis- solved organic carbon concentrations ranging from 5-1 5 mg C:L (Naiman, 1982a, l983a). All sites are typical of the brown water, low gradient, streams

MATAMEK RIVER SECOND

FALLS

BEAVER CREEK

RIVER

FIRST CHOICE A WOOD - 80rn

CREEK

LOD CRAN CARRE CREEK

Fig. 1. Study sites on the north shore of the Gulf of St. Lawrence, -25 km east of Sept-Iles, Quebec.

195

found on the Canadian Precambrian Shield. TWO of the four streams (Beaver Creek, Matamek River) are in the Matamek River watershed; First Choice Creek and Cran Carre Creek are adjacent to the watershed. Various physiochemical and biological characteristics of these streams have been described previously(Naiman& Sedell, 198 1; Naiman, 1982a, 1983a, b, c; Conners & Naiman, 1984; Naiman et al., 1984; Melillo et al., 1983; Naiman & Melillo, 1984).

First Choice Creek is a first order stream with no history of beaver activity. Most organic inputs (3077534 g DM . mm2 . y-t as leaves, needles, wood and herbaceous vegetation) originate from the closely associated forest. The stream begins as a series of seeps, has an organic detrital substrate, and a mean annual discharge of -0.013 ms s-t. The temperature ranges between 0.4 and 11 .O o C, with about 1700 degree days per year (” C . y-l); the pH ranges annually from 6.2-7.2.

Beaver Creek is a second order stream with 12 abandoned or recolonized beaver dams sepa- rated by short riffles over 1.3 km. Most wood (-70% of the -39.7 kg m-2) in the stream channel was put there, directly or indirectly, by beaver. Sediment behind dams is flocculent detritus alternately layered with fine wood particles. Mean annual dis- charge is -0.033 m3 sst; the temperature ranges from 0.1 to 19.0 “C with about 2 100 OC . y-t; pH ranges from 6.0 to 7.0.

Cran Carrt Creek is a third order stream. Thir- teen dams, most of which are situated in open meadows, are separated by short, low-gradient riffles. Most wood in the stream was put there by beaver; loosely packed silt characterizes most ponds. Mean annual discharge is -0.95 m3 s-t; the temperature ranges from 0.1 to 22.0 o C with about 1939 o C . y-t; pH ranges between 5.8 and 6.5.

The Matamek River is a sixth order stream. There are no beaver dams due to seasonally high discharge but beaver do cut wood along the banks of slower flowing reaches. The mean annual dis- charge is -13.7 m3 sst. Wood cut by beaver makes up a minor proportion of the total wood mass. Sediment in slow flowing reaches is firmly packed clay and silt. Temperatures range from 0.1 to 19.8 o C; there are about 2 300 o C . y-t, and pH ranges from 4.9 to 6.1.

Methods and materials

Five tree species (black spruce, Picea mariana; balsam fir, Abies balsamea; speckled alder, Alnus rugosa; trembling aspen, Populus tremuloides; and paper birch, Betulapapyrifera) were debarked, cut into cross sections, attached to monofilament line, and anchored at five sites in June 198 1. The sites were: First Choice Creek (FCC), Beaver Creek lower pond (BCl), Beaver Creek riffle (BCR), Cran Carre Creek lower pond (Ccl), and Matamek River (MTK). The percentage of nitrogen ranged by spe- cies from 0.04 to 0.32% with the lignin: nitrogen ratio ranging from 42 to 647 (Melillo et al., 1983).

Sediment samples were taken from each of the five wood incubation sites, from two additional pond sites on Cran Carrt Creek (CC2, CC3) and an additional pond on Beaver Creek (BC2). All sam- ples were collected biweekly during the 1982 ice- free (May-October) period. Similar samples taken during 1981 for technique and procedure develop- ment are not reported here.

Incubation. The acetylene-ethylene reduction tech- nique was used for estimating nitrogen fixation potentials (reviewed by Hardy et al., 1968, 1973). Controls were run for inherent ethylene production from wood and sediment. Gas samples were stored for analysis in 3 ml disposable plastic syringes rather than fixing wood and sediment with tri- chloroacetic acid, which may cause ethylene ab- sorption by rubber stoppers.

Wood. Samples were placed in plastic bags with unaltered stream water and returned immediately to the laboratory. At least three replicates per wood species per site were used. Wood pieces, taken from the outer perimeter of the disks, were measured for exposed surface area (-12 cm2) and placed in 60 ml serum bottles with 5% C2H2 and -5 ml filtered (0.2 pm) water from the sample site. Initial gas samples were collected in disposable plastic sy- ringes for later analysis. Samples were incubated at the ambient stream temperature for a precise time period (either 2, 5, or 7 h depending upon expected activity) before final gas samples were extracted for determination of C2H, concentration. About 620 samples were examined during 1982.

Sediment. Cores were taken in the ponds behind

196

beaver dams and in riffles with a 6.5 cm diameter Plexiglas tube. Redox potential and the pH of sur- face and 10 cm depth sediment were measured in the laboratory using a Coleman 80 Metrion pH meter. Four replicate samples of each aerobic and anaerobic sediment (occupying -10% of the reac- tion volume) were incubated at the ambient stream temperature in 120 ml serum bottles with -5 ml filtered (0.2 pm) stream water. Water and serum bottles used for incubation of anaerobic sediment was flushed with argon for 5 min to remove oxygen both before and after adding sediment, and prior to CzHz injection. Following final gas collection, sedi- ment samples were dried at 50 o C, weighed for dry mass (DM), and weighed again after 12 h at 500 ’ C to calculate ash-free-dry-mass (AFDM). About 460 samples were examined during 1982.

Gas analysis. Ethylene concentrations were meas- ured within48 h on a Carle AGC211 Flame Ioniza- tion Detector Gas Chromatograph using a propo- pak Q column at 75 o C. Pre- and post-incubation ethylene concentrations ([C,H,]), reported on a Hewlett Packard 3390A integrator, were corrected for C2H, absorption by plastic syringes over time using the following equation:

(X,/X3 F-(X,/X;) I = Ethylene production

where: X, = mean [CzH4] in 5% C2H, at t = 0 in syringe, X, = mean [C,H,] in 5% C2H, at t = time final

samples remained in syringe prior to analysis, Xi = mean [C2H4] in 5% C2H, at t = time initial

samples remained in syringe prior to analysis. F = post-incubation [CzH& and I = pre-incubation [CZH4].

Ethylene production was calculated perg AFDM and g DM of sediment, or per surface area (cm2) of wood, per hour. There was no relationship between sediment sample volume and the specific rate of ethylene production on an AFDM (r2 = 0.01) or a DM (r2 = 0.01) basis, or between wood surface area and ethylene production. Mean acetylene reduction rates were used for data interpretation since para- metric Student-Newman-Keuls analysis of variance tests were not valid (based on Bartletts test of homogeneity of variances; Sokal & Rohlf, 1969). Non-parametric Kruskal-Wallis analysis of var- iance procedures were used to test for significant differences (p < 0.05) between sites, between wood

species, and between aerobic and anaerobic sedi- ment (Sokal& Rohlf, 1969). Nitrogen fixation po- tentials were estimated from C2H2 reduction rates using a I:3 conversion factor (Bergensen, 1970, 1980), although the universality of this ratio for specific microbes and environmental situations can be disputed (Hardy et al., 1971; Graham et al., 1980). Higher ratios (1:6-l 5) have been reported for waterlogged conditions (Rice & Paul, 1971); thus, estimates for nitrogen fixation reported in this paper may be the potential maximum.

Data extrapolations

Estimates of the total nitrogen contribution by sediment organisms were derived from mean sea- sonal reduction rates, an equation predicting aver- age stream width as a function of stream order (Naiman, 1983a), and data on sediment volume at each sample site (Naiman, 1982b). Sediment vol- umes accumulated behind dams were calculated using survey techniques and trigonometric inter- polations of valley contours. Sediment mass was calculated using the empirically determined factor of 1 cm3 = 0.25 g DM, and assuming that sediment in riffles extents to only 15 cm depth.

An estimate was also made of the total nitrogen contribution by organisms associated with wood by combining mean seasonal fixation rates with data on wood surface area in each stream (Naiman, 1982b). The volume of small wood (1 cm-10 cm diameter) was estimated by measuring dimensions of all wood in 1 m transects every 10 m of stream length. Wood >lO cm diameter was measured for either the entire stream or over 500 m of channel. Each piece of wood was identified to species and categorized by the method of input to the stream (e.g., erosion, wind, directly by beaver, indirectly by beaver). Wood surface area was calculated from diameter and length measurements treating each piece as a cylinder. Wood buried in sediments or as a structural part of dams was not accounted for in this survey; therefore, estimates of nitrogen contri- butions are underestimated in beaver ponds where high water levels and accumulated sediment cover significant wood accumulations.

Nitrogen fixation studies in field situations have historically yielded patchy data which are incom- patable with standard methods of statistical analy- sis. In many studies investigators have discarded

197

zero values, or reported only maximum values to avoid problems with statistical analysis; few have included long term repetitive sample schedules. These shortcomings mislead readers into believing that highest rates are the norm. Thus, we believe it is necessary to report observed trends, even if statisti- cally insignificant, in order to gain insights to the subtle multidimensional processes of stream eco- systems.

Results

Wood. Acetylene reduction rates were consistently low and patchy with 77% of all wood incubations showing no activity. Figure 2 (paper birch at BCR) is an example of the patchiness and the distribu- tional range of reduction rates observed at all sites (e.g., 15 zero values for 22 incubations; maximum = 2.3 X 1O-2 I.tmol C2H4 a cm-2 . h-‘). Patchiness ac- counts for there being no homogeneity of variances and no significant difference between reduction rates for wood species and sample sites (x2 < 9.5, p > 0.05).

Mean rates ranged from 4.7 X 10m5 pmol C2H4 . cm-2 . hm’ for microorganisms associated with black spruce to 4.9 X 1O-4 pmol GH,. cm-z. h-1 for microorganisms associated with paper birch

10.'

10-q t . I

Fig. 2. CzH2 reduction rates associated with paper birch wood at the Beaver Creek riffle area. Graph depicts typical patchiness of fixation rates during the 1982 ice-free season.

(Table 1). Microorganisms on hard woods (birch, aspen, alder), which have low 1ignin:nitrogen ratios, showed higher mean activity than microorganisms on soft woods (spruce, balsam; Table 1). Mean reduction rates for each wood species increased downstream for every species at Beaver Creek. For example, C2H2 reduction by microorganisms asso- ciated with aspen at BCR (2.42 X 10m4 pmol C2H4 . cmm2 . hm’) was higher than for the same species at BCl (2.3 X 10-S pmol C2H4 . cm-2 . hm’; Table 1).

Sediment. Acetylene reduction rates were also low and patchy; 63% of the sediment samples showed no nitrogen fixation. Mean fixation rates ranged from 0.5 X 10m3 to 5.5 X 10m3 pm01 C2H4. g AFDM-’ e h-’ for microorganisms in anaerobic sediments and from 0.6 X 10m3 to 7.0 X 10m3 pmol C2H4 * g AFDM-’ * h-’ for microorganisms in aerobic sediments (Table 2). Mean rates for aerobic and anaerobic sediments tended to increase pro- ceeding downstream in beaver manipulated water- sheds (Table 2). Microorganisms from three dams on Cran Carri: Creek showed aerobic nitrogen fixa- tion rates consecutively increasing downstream from 1.0 X 10m3 (CC3) - 3.3 X 10m3 (CC2) - 5.0 X 10m3 (Ccl) pmol C2H4 . g AFDM-’ * hm’, and for microorganisms in anaerobic sediments increas- ing from 0.8 X 10m3 (CC3) - 1.1 X 1O-3 (CC2) - 2.8 X 10m3 (Ccl) C,H, . g AFDM-’ . h-l. The same situation occurred in Beaver Creek, where mean rates for aerobic sediments increase from0.6 X lo-3 (BC2) - 5.3 X 1O-3 (BCl) - 7.0 x 10-J (BCR) /*moles C2H4 . g AFDM-’ . hm’, while for microor- ganisms in anaerobic sediment the fixation rates increase from0.5 X 10m3 (BC2) - 2.8 X 1O-3 (BCl) - 5.2 X 10m3 (BCR) pmol C2H4 * g AFDM-’ * h-‘.

There is no statistically significant difference in mean nitrogen fixation rates between aerobic and anaerobic sediments for watersheds with or without beaver. However, at First Choice Creek, a non- beaver area, microorganisms in anaerobic sedi- ments show a higher mean rate (5.5 X 10-j pmol C2H4 . g AFDM-’ . h-l) than in aerobic sediments (2.4 X 10m3 pmol C2H4 . g AFDM-’ . hm’, Table 2). In contrast, sites at Beaver Creek and Cran Car& Creek, where beaver are present, have higher C2H2 reduction rates for aerobic sediments than for an- aerobic sediments (Table 2).

Table

I.

Mean

se

ason

al &H

z re

ducti

on

rates

of

orga

nisms

co

lonizi

ng

wood

sp

ecies

in

1982

; R

= me

an

rate

(IOm4

pmo

l C2

H4

cm-2

h 1)

; SE

M

= sta

ndar

d err

or;

n =

samp

le siz

e; 0

= pe

rcent

of to

tal

incub

ation

s sh

owing

no

C2

H2

redu

ction

.

Site

Bl

ack

Spru

ce

Balsa

m Fir

Sp

eckle

d Al

der

Trem

bling

As

pen

Pape

r Bi

rch

All

spec

ies

n R

SEM

0

n R

SEM

0

n 8

SEM

0

n R

SEM

0

n ?

SEM

0

n ??

SE

M

0

First

Choic

e Cr

eek

24

0.70

0.21

63

24

0.40

0.19

83

24

7.77

3.67

63

24

0.30

0.09

83

27

0.49

0.19

81

123

1.93

0.20

75

Beav

er

Cree

k, Lo

wer

Dam

25

0.28

0.08

76

25

0.31

0.12

68

28

0.23

0.06

79

23

0.23

0.06

78

26

0.16

0.04

69

127

0.24

0.01

74

Beav

er

Cree

k, Ri

ffle

27

0.68

0.23

78

26

1.14

0.56

81

27

I.16

0.29

56

24

2.42

1.05

71

26

23.0

7 11

.78

73

130

5.69

0.85

72

Cran

Ca

rri:

Cree

k, Lo

wer

Dam

23

0.41

0.11

70

23

0.27

0.13

61

23

0.31

0.15

78

23

0.80

0.27

61

23

0.45

0.15

65

115

0.45

0.02

67

Matam

ek

Rive

r 26

0.2

8 0.0

8 77

26

0.7

4 0.3

5 85

26

0.2

3 0.9

0 77

22

0.7

3 0.3

2 77

26

0.0

9 0.0

3 85

12

6 0.4

1 0.0

3 80

All

sites

12

5 0.4

7 0.0

1 73

12

4 0.5

7 0.0

3 76

12

8 1.9

4 0.2

9 71

11

6 0.8

9 0.0

8 74

12

8 4.8

5 0.9

0 75

62

1 I.7

5 0.0

9

199

Table 2. Mean seasonal CzH2 reduction rates for sediment organisms in 1982: j? = mean rate (IO-3 pmol CzH,t . g AFDM-t . h-t); SEM = standard error; n = sample size; 0 = percent of total incubations showing no CzH2 reduction.

Site

First Choice Creek

Beaver Creek, Upper Dam

Beaver Creek, Lower Dam

Beaver Creek, Riffle

Cran Carre Creek, Upper Dam

Cran Carre Creek, Middle Dam

Cran CarrC Creek, Lower Dam

Matamek River

All sites

Aerobic sediment

n X SEM

30 2.4 0.8

28 0.6 0.1

30 5.3 1.9

29 7.0 1.8

23 1.0 0.3

29 3.3 0.7

31 5.0 1.0

25 2.3 0.6

235 3.4 0.1

0

77

61

60

59

74

59

61

64

64

Anaerobic sediment Both sediment types

n X SEM 0 n x SEM 0

30 5.5 1.5 70 60 4.0 0.3 74

28 0.5 I .4 57 56 0.6 0.1 59

30 2.8 1.1 53 60 4.1 0.2 57

29 5.2 1 .o 55 58 5.1 0.2 57

30 0.8 0.2 63 53 1.0 0.0 69

20 1.1 0.5 90 49 2.2 0.2 75

30 2.8 0.6 77 61 3.9 0.2 69

26 2.8 0.4 38 51 2.6 0.1 51

223 2.7 0.1 63 458 3.0 1.8 64

Discussion

Beaver physically influence the structure and dy- Estimates of annual nitrogen contributions by namics of stream ecosystems by creating new habi- microorganisms colonizing wood in Beaver Creek tats which, as a function of area or volume, de- and First Choice Creek were calculated from mean termines the magnitude of change in ecosystem- acetylene reduction rates (Table 1) and the surface level processes. Absolute increases in biologically area of wood in those streams (Table 3). More bio- fixed nitrogen largely result from the creation of logically fixed nitrogen is produced by microorgan- newly available colonization substrates, more so isms on wood in Beaver Creek riffle (20.7 X IO-2 g

than to mass specific increases in nitrogen fixation rates.

Table 3. Estimated nitrogen contributions by organisms colonizing wood in First Choice Creek (no beaver), Beaver Creek riffle and Beaver Creek Dam. (Units: wood surface area = cm* . m- *; CzHz reduction rate = lO-4 pmol CzH4 . cm-* . h-t; N input = IO-2 g N . m-2 . y-t).

Species First Choice Creek

Wood C2H2 surface reduction area rate

N input

Beaver Creek

Riffle

Wood C2H2 surface reduction area rate

N input

Dam

Wood C2H2 N input surface reduction area rate

Alder 1105 1.70 6.96 831 1.16 0.79 1154 0.23 0.22 Balsam 1453 0.19 0.23 129 1.14 0.12 324 0.31 0.08 Spruce 396 0.70 0.23 0 0.68 0 256 0.28 0.06 Aspen 138 0.30 0.03 484 2.42 0.96 132 0.23 0.14 Birch 1636 0.49 0.66 999 23.07 18.84 1879 0.16 0.25

Total 4728 8.11 2443 20.71 4345 0.75

200

N . mm* . y-t) than by microorganisms on wood in average streams width (Naiman, 1983a), estimates First Choice Creek (8.1 X IO-2 g N . m-2 . y-t); there were made of the nitrogen contribution at each is twice the wood surface in the Beaver Creek riffle ponded area should that area be a typical riffle (i.e., (Table 3). Surprisingly, microorganisms on wood with sediment volume calculated from an average in the beaver pond contribute only 0.8 X lo--* g stream width, a 15 cm sediment depth and a stream N . m-2 . y-1. These estimates are somewhat lower length corresponding to the pond length). Total than reports of 70.0 X IO-* g N + m * . y-l in a wood nitrogen contributions were enhanced 9 (at BCI) to choked Oregon stream (Buckley & Triska, 1978; 44 fold (at CC2) due to beaver damming riffles Triska et al., 1983) but higher than the 0.2 X lo-* g (Table4). The calculated increase is proportional to N . m-2 . y-1 reported by Meyer et al. (1981) for a the volume of sediment trapped by the dams rather deciduous forest stream in New Hampshire. than to changes in nitrogen fixation rates.

The volume of sediment in a ponded reach of stream is vastly increased by beaver activity (Naiman & Melillo, 1984). Annual nitrogen contri- butions by sediment microorganisms, calculated from mean reduction rates (Table 2) and survey data of sediment volumes (Table 4) are much larger than those by wood microorganisms (5.1 g N . mm* for sediment, <O.l g N . m-* for wood at BCl). Nitrogen contributed to the ecosystem from 3 of 13 ponded areas on Cran Car& Creek is 52.4 kg N . y-t, assuming mean fixation rates occur in all accumulated sediments. Nitrogen contributions from sediment microorganisms behind two dams on Beaver Creek (5.8 kg N . y-t) are less, but still significantly greater than N inputs from the entire mass of wood in the stream system (Table4). Nitro- gen contributed to the system from ponded areas (4 to 5 I kg N . hectare-t . y-t based on BCl) is similar to the amount of nitrogen added (85 kg N. hec- tare-l . y-t) to an abandoned mill pond by alder in northeastern Connecticut (Voigt & Stevcek, 1969).

Litter inputs from the forest also represent a substantial source of nitrogen to these streams (Naiman & Melillo, 1984; Conners & Naiman, 1984). The riffle in Beaver Creek receives about 4.lg N. m-* . y t as direct litter fall and 1.9 g N . mm* . y-l from lateral movement along the forest floor, while the lower beaver pond receives 1.7 g N . mm* . y-t in direct fall and 0.1 g N . m-* . y-t from lateral movements. It is impor- tant to note, however, that even though nitrogen from litter inputs is substantially greater than from nitrogen fixation, it is not in a readily available form; rather, it is slowly released as decomposition proceeds. Nitrogen fixation on wood and surface sediments, although at low levels, may be more accessible to higher trophic levels because it is available for immediate assimilation. Conversely, nitrogen fixing organisms, by using allochthonous inputs as carbon sources, may enhance litter de- composition rate and thus its availability to the stream community.

To what degree do beaver influence the amount of nitrogen sequestered annually by a reach of stream by building dams and trapping sediment? Applying an equation relating stream order to

It also appears that nitrogen fixation may be enhanced downstream of beaver influenced areas. In fact, both the nitrogen fixation and wood de- composition studies (Melillo et al., unpubhshed

Table 4. Estimated nitrogen contributions by organisms colonizing sediment accumulated behind dams and sediment in riffles of comparable length.

Upper Dam Middle Dam Lower Dam

Total

Cran Carrt Creek Beaver Creek

Beaver No Beaver Ratio Beaver No Beaver Ratio

Sediment N input Sediment N input Sediment N input Sediment N mput volume (kg N Y’) volume (kg N Y’) volume (kg N y’) volume (kg N . Y’) Cm-‘) Cm31 Cm-‘) (m3)

5919 23.4 525 1.9 12.3 580 2.1 25 0.1 21.0 3811 21.9 81 0.5 43.8 6535 7.1 242 0.3 23.7 202 3.7 21.5 0.4 9.3

52.4 2.7 5.8 0.5

201

data) show increasing rates proceeding downstream within a beaver manipulated watershed. Either a lack of sufficient nitrogen, or an abundance of available phosphorus with ample carbon, could ini- tiate this trend. There are indications that both mechanisms may be responsible, depending on the local situation (Melillo et al., 1984; Naiman & Melillo, 1984).

In addition, beaver ponds may act as sources or sinks of essential nutrients (P and N) and carbon, the relative levels of which influence nitrogen fixa- tion rates and decomposition dynamics. Beaver utilize woods which decompose rapidly (Jenkins, 1975; Melillo et al., 1983) and thus continuously supply downstream waters with carbon (Naiman, 1983~). Phosphorus may be a contributing factor to the enhancement of downstream fixation and de- composition rates since there appears to be a con- tinuous leaching of phosphorus as beaver ponds cyclically dry and flood over many years (Baxter, 1977).

Finally, it is clear that beaver not only influence the structure of stream ecosystems but also have a pronounced influence upon the magnitude of stream processes. The amount of nitrogen annually accrued to small watersheds is substantially in- creased, primarily from the massive volume of sedi- ment trapped behind dams. The implications for aquatic productivity, nutrient cycling, and ultimate terrestrial succession are significant. Beaver activity results in increased absolute productivity and biotic diversity in headwater streams, and has a substan- tial effect on downstream communities as ponds and meadows pass through their ontogenetic de- velopment (Naiman & Melillo, 1984). With beaver, the productivity and biotic diversity of aquatic sys- tems and their adjacent riparian zones are in- creased, and the effects are evident in nearly all ecosystem components.

Acknowledgements

We thank R. Morin and J. E. Hobbie for helpful suggestions on the manuscript, B. S., Farr for as- sistance in surveying stream sites and E. M. Conners for allochthonous input data. This re- search was supported by NSF Grant DEB 8 I-05677 and the Matamek Research Program. Contribu- tion No. 5543 of the Woods Hole Oceanographic

Institution and No. 92 of the Institution’s Matamek Research Station.

References

Ashton, P. J., 1979. Nitrogen fixation in a nitrogen-limited im- poundment. J. Wat. Pollut. Cont. Fed. 51: 570-579.

Baxter, R. M., 1977. Environmental effects of dams and im- poundments. Ann. Rev. Ecol. Syst. 8: 2555283.

Bergersen, F. J., 1970. The quantitative relationship between nitrogen fixation and the acetylene reduction assay. Aust. J. biol. Sci. 23: 1015~1025.

Bergersen, F. J. (ed.), 1980. Methods for Evaluating Biological Nitrogen Fixation. John Wiley L Sons, N.Y., 702 pp.

Buckley, B. M. & F. J. Triska, 1978. Presence and ecological role of nitrogen fixing bacteria associated with wood decay in streams. Verh. int. Ver. Limnol. 20: 132441332.

Conners, M. E. & R. J. Naiman, 1984. Allochthonous inputs to the Matamek River watershed, Quebec: the influence of stream size. Can. J. Fish. aquat. Sci. 41: 1473-1484.

Dierberg, F. E. & P. L. Brezonik, 1981. Nitrogen fixation (acetylene reduction) associated with decaying leaves of pond Cyprus (Taxodium distichum var. nutans) in a natural and a sewage-enriched Cyprus dome. Appl. envir. Microbial. 41: 1413-1418.

El Samra, M. I. & J. Olah, 1979. Significance of nitrogen fixa- tion in fish ponds. Aquaculture 18: 367-372.

Flett, R. J., D. W. Schindler, R. D. Hamilton&N. E. R. Camp- bell, 1980. Nitrogen fixation in Canadian Precambrian Shield lakes. Can. J. Fish. aquat. Sci. 37: 494-505.

Graham, B. M., R. D. Hamilton & N. E. R. Campbell, 1980. Comparison of the nitrogen-15 uptake and acetylene reduc- tion methods for estimating the rates of nitrogen fixation by freshwater bluegreen algae. Can. J. Fish. aquat. Sci. 37: 488-493.

Hardy, R. W. F., R. D. Holsten, E. K. Jackson & R. C. Burns, 1968. The acetylene-ethylene assay for nitrogen fixation: laboratory and field evaluation. PI. Physiol. 43: 123-128.

Hardy, R. W. F., R. C. Burns, R. R. Hebert, R. D. Holstein & E. K. Jackson, 1971. Biological nitrogen fixation: a key to world protein. Pl. Soil, Spec. Volume: 561-590.

Hardy, R. W. F., R. C. Burns & R. D. Holsten, 1973. Applica- tions of the acetylene-ethylene assay for measurement of nitrogen fixation. Soil. Biol. Biochem. 5: 47-8 I.

Howard, R. J., 1982. Beaver habitat classification in Massachu- setts. M.S. Thesis, Univ. Massachusetts, Amherst, 67 pp.

Jenkins, S. H., 1975. Food selection by beavers: A multidimen- sional contingency table analysis. Oecologia 21: 157-173.

MacGregor, A. N. & D. R. Keeney, 1973. Acetylene reduction assay of anaerobic nitrogen fixation by sediments of selected Wisconsin lakes. J. envir. Qual. 2: 438-440.

Melillo, J. M., R. J. Naiman, J. D. Aber & K. N. Eshleman, 1983. The influence of substrate quality and stream size on wood decomposition dynamics. Oecologia 58: 28 l-285.

Melillo, J. M., R. J. Naiman, J. D. Aber & A. F. Linkins, 1984. Decay of plant litter in northern streams. Bull. mar. Sci. (in press).

202

Meyer, J. L., G. E. Likens&J. Sloane, 1981. Phosphorus, nitro- gen, and organic carbon flux in a headwater stream. Arch. Hydrobiol. 91: 28-44.

Naiman, R. J., 1982a. Characteristics of sediment and organic carbon export from pristine boreal forest watersheds. Can. J. Fish. aquat. Sci. 39: 1699-1718.

Naiman, R. J., 1982b. The Matamek Research Program: An- nual Report for 1981. Woods Hole Oceanogr. Inst. tech. Rep. 8 I-29,234 pp.

Naiman, R. J., 1983a. The annual pattern and spatial distribu- tion of aquatic oxygen metabolism in boreal forest wa- tersheds. Ecol. Monogr. 53: 73-94.

Naiman, R. J., 1983b. A geomorphic approach for examining the role of periphyton in large watersheds, p. 191-198. In R. G. Wetzel (ed.), Periphyton of Freshwater Ecosystems. Dr. W. Junk b.v. Publishers, The Hague.

Naiman, R. J., 1983~. The influence of stream size on the food quality of seston. Can. J. Zool. 61: 1195-2010.

Naiman, R. J., D. M. McDowell & B. S. Farr, 1984. The influ- ence of beaver (Casfor canadensis) on the production dy- namics of aquatic insects. Verh. int. Ver. Limnol. 22 (in press).

Naiman, R. J. & J. M. Melillo, 1984. Nitrogen budget of a subarctic stream altered by beaver (Castor canadensis). Oecologia 62: 150-155.

Naiman, R. J. & J. R. Sedell, 1981. Stream ecosystem research in a watershed perspective. Verh. int. Ver. Limnol. 21:

804-8 I 1. Rice, W. A. & E. A. Paul, 1971. The acetylene reduction assay

for measuring nitrogen fixation in waterlogged soil. Can. J. Microbial. 17: 1049-1056.

Sokal, R. & R. J. Rohlf, 1969. Biometry: the Principles and Practice of Statistics in Biological Research. W. H. Free- man & Co., San Francisco, 776 pp.

Tam, T. Y., C. I. Mayfield & W. E. Inniss, 1981. Nitrogen fixa- tion and methane metabolism in stream sediment - water system amended with leaf material. Can. J. Microbial. 27: 51 I-516.

Triska, F. J., J. R. Sedell, K. Cromack, S. V. Gregory & F. M. McCorison, 1983. Nitrogen budget for a small coniferous forest stream. Ecol. Monogr. 54: 119-140.

Vanderhoef, L. N., C. Huang, R. Musil & J. Williams, 1974. Nitrogen fixation (acetylene reduction) by phytoplankton in Green Bay, Lake Michigan, in relation to nutrient concentra- tions. Limnol. Oceanogr. 19: 119-125.

Voigt, Cl. K. & G. L. Stevcek, 1969. Nitrogen accumulation in alder ecosystems. Soil. Sci. Am. 33: 946-949.

Waughman, G. J. & D. J. Bellamy, 1980. Nitrogen fixation and the nitrogen balance in peatland ecosystems. Ecology 61: 1188-1198.

Received I2 December 1983; in revised form II May 1984; accepted 11 May 1984.