Factors influencing nitrogen concentrations in soil water in a North American tallgrass prairie

Sol1 Broi. &ochem. Vol. 20. So. 5, pp. Z-729. Printed :n Great Bntam Ail rights rexned

0038-071785 53.00- 0.00 Copbnght r’ 1985Pcrgamon Ress plc

193.3

FACTORS INFLUENCING NITROGEN CONCENTRATIONS IN SOIL WATER IN A NORTH AMERICAN TALLGRASS PRAIRIE T. R. SEASTEDT and D. C. HAYES* Division

of Biology,

Kansas State LJniversity. (rlccepted

IO Dvcemher

hlanhattan.

KS 66506.

U.S.A.

1987)

Summary-The influence of roots, soil fauna and microbes on nitrogen concentrations in soil water in annually-burned tallgrass prairie were evaluated using porous cup lysimeters and a series of manipulations involving an insecticide. clipping of the vegetation. and C and N additions to the soil surface. An organophosphate insecticide (isofenphos), which significantly reduced densities of certain nematode and arthropod groups, resulted in small but statistically significant declines in soil water NO_; concentrations during Z of 3 yr of study. Organic N. the dominant form of soil water N, was unaffected by treatment. Clipping the foliage weekly during the spring resulted in significant reductions in annual foliage production, but failed to change concentrations of soil water NO; or organic N. Nitrogen additions (log N m-r as NH,NO,) greatly increased soil water NO; concentrations; However, when 250 gC sucrose m-r were concurrently added to plots. there was a significant reduction in soil water NO; concentrations. Soil water NH; concentrations were unaffected by treatment. These studies emphasize the importance of microbes as the dominant factor affecting soil water N concentrations in carbon-rich soils. Plants. soil fauna and nitrogen fertilizers do affect N dynamics, but. under annually-burned conditions in the tallgrass prairie, these etTects are obscured by microbial processes.

1NTRODL’CTION The tallgrass prairie is a fire-maintained ecosystem. Producttvity oi the prairie grasses declines in the absence of fire, and woody plant species often invade unburned areas (Knapp and Seastedt, 1986). Thus, fire is a common management tool in most remaining prairie remnants. Both root and shoot production of the dominant grasses are stimulated by spring burning (e.g. Hadley and Kieckhefer, 1963; Kucera et al.. 1967). The grasses of annually-burned prairie are low in N content (Hayes, 1985; Seastedt, 1988) and the high annual productivity and turnover of foilage and roots contribute to N immobilization by microbes feeding on these low N substrates. The tallgrass prairie is the biome that gave rise to some of the richest agricultural lands in North America and is a system that contains very large soil organic N pools (Risser and Parton, 1982). Only recently, however, has research focused on N fluxes rather than amounts of N within this system (Ojima et al.. 1988; Seastedt. 1988). Disturbances to vegetation are known to result in enhanced concentrations of inorganic N in soil and stream water. presumably because of a reduction in plant uptake of NH; and NO; ions (Vitousek and Melillo, 1979; Vitousek ef al.. 1982). Inorganic N concentrations in soils can also be enhanced by fauna, both by contributing readily-mineralizable N sources and excreting N in an inorganic form (e.g. Swank et al.. 1981; Ingham er oi.. 1985; Anderson er al., 1985). In general, any activity that stimulates ‘Present address: Center for Coastal and Environmental Studies, Doolittle Hall, Rutgers L’niversity. Sew Brunswick. NJ 08903. U.S.A.

mineralization while concurrently failing to stimulate plant uptake should increase inorganic N in soil water. The magnitude of change caused by disturbances to SOIIS or plants remains unknown for many ecosystems, and in particular the N dynamics of natural, unfertilized and unplowed grasslands have been poorly studied. Soil water and stream water inorganic N concentrations are low in the tallgrass prairie (McArthur er 01.. 1985: Knapp and Seastedt, 1986), with stream inorganic N levels averaging well under 100 pg N I-’ (C. M. Tate, unpublished Ph.D. thesis, Kansas State University, 1985) compared with rainfall inputs averaging over 1000~glW’ of inorganic N (Seastedt, 1985; Gilliam, 1987). Given the large reservoir of N in these soils, this system should possess the potential for large pulses of N export in response to disturbance such as occurs when the soil is used in greenhouse experiments (e.g. James and Seastedt. 1986). This study used porous cup lysimeters in a series of experiments to evaluate the relative importance of plant, soil invertebrate fauna and microbe activities on the N concentrations of soil water of tallgrass prairie. These data also provide a reference for comparing the N concentrations of soil water in the tallgrass prairie with other natural and managed systems.

STUDY

SITE

ASD

4iETHODS

Research was conducted at Konza Prairie Research Natural Area, an intensively studied native tallgrass prairie site located about 10 km south of Kansas State University in the Flint Hills of northeastern Kansas (39-N lat. 96-W long). Vegetation is

‘16

T. R.

%XSTEDT

dominated by big bluestem f.-tndropogon gerardii Vitman), little bluestem (,d. scoprius Michx.) and Indiangrass [Sorgasrrwt2 ~ZUMS IL.) Nash). Rainfall averages 83 cm yr-i. of which about 7% occurs in the growing season of April-September. Streamflow is estimated to export about 20% of rainfall: the remainder is lost via evapotranspiration (J. K. Koelliker, Kansas State University, personal communication). Rainfall is sufficient to leach solils derived from limestone, with surface (top 5 cm) soils averaging a pH of about 6.3, with deeper horizons generally averaging pH values over 6.0. All sites used in these studies were burned in April of each year. Porous cup lysimeters (Soil Moisture Corp.. San Diego, Calif.) that had been acid washed and soaked in deionized water were used to collect soil water samples. Lysimeters were pumped to a vacuum of -60 kPa and sampled weekly whenever soil water could be extracted. A minimum of 4 weeks passed between the time of lysimeter installation and the time of sampling to avoid short-term soil disturbance effects. Porous cup lysimeters were capable of obtaining soil water samples during only a portion of the year. In winter, samples froze within the tubes: during summer the soil water tension was often too low to obtain a sample. Microbial activity within the lysimeters was not believed to influence N concentrations (Vitousek et al.. 1982). Experiments to manipulate soil invcrtcbratc densities were initiated in 1983 and continued for 3 yr. Lysimeters were installed at 2 ridgetop sites with silty clay soils (Typic Natrustolls). These collectors were installed at 20 and 80 cm soil depths on one site. and at 20 and 40 cm depths on the second site. The 20 and XOcm depths were selected to obtain samples from the active rooting zone and from a zone below 90% of the roots (e.g. Dahlman and Kuccra. 1965). The 40 cm depth at the second site was a necessity due to the presence of limestone below this depth. A randomized design was used at both sites, with I6 collectors (8 at 20 cm. 8 at 80 cm) installed at the first site and 48 collectors (24 at 20 cm, 24 at 40 cm) at the second site. An organophosphate (isofenphos. “Oftanol”, Mobay Chemical Co.) was applied twice during the growing season at a rate of 5 g m-’ active ingredient to half of the plots at each site. The chemical initially reduced nematode densities about 20%. but was less effective through time (Seastedt er of.. 1987). Subsequent experiments using organophosphates indicated at least short-term reductions of 50% in microarthropod densities in the top 5 cm of soil, and in greenhouse leaching experiments the chemical did not affect the inorganic N content of leachates (Seastedt, unpublished). The effects of plant root dynamics on N concentrations in soil water were investigated using the 20 cm deep lysimeters at site 2 in 1985 by clipping half of the plots weekly for 6 weeks during late spring (May and June). A 33 x 33 cm square quadrat was placed around the lysimeters and vegetation was clipped to surface level. Clippings were saved from each plot and all plots were clipped at the end of the growing season to compare biomass production with untreated plots. Root and rhizome data were not collected, but previous studies (Seastedt er al.. 1986 and unpublished) indicated that this clipping in-

and D. C. H~vss tensity would reduce belowground biomass to levels at or below 60% of the controls, The effects of nitrogen (IO g N m-’ as NH,NO,) and carbon (250 gC m-: of table sugar, sucrose) additions were studied in 1986 by using these 2 treatments in a 2 x 2 factorial design on a total of I6 plots. A lowland site with a silty clay loam (Pachic Argiustoll) was used. A single lysimeter was installed at a 2Ocm depth on each of the I6 plots. Fertilizer and sugar were applied in granular form and then watered into the plots. Water was misted onto the plots to avoid surface flow, and all plots received the same amount of water. We estimate that an additional 4cm of H,O was added to the soil during the first week of the experiment. inorganic N content of soil water was monitored for I week before treatment, and for 4 weeks after treatment. Plant foliage biomass was also harvested at the end of the growing season. All water samples were stored at 5’C and, if analysed for NH;, were processed within l-3 d after collection using the phenolhypochlorite procedure (Solorzano, 1969). With the exception of the fertilizer study, however, we found that soil water NH: values were low. often under IO pg N I-’ and usually below I /lg N I-‘. Therefore. NH: values wcrc not routinely obtained. Nitrate (and whatever traces of nitrite were present) was measured with the cadmium reductionsulfanilamide procedure using a Technicon Autoanalyzer. Organic N, when measured, was estimated with the total pcrsulfatc nitrogen procedure (D’Elia er crl., 1977, as modified by R. T. Edwards, University of Georgia. personal communication). The persulfate procedure measures total N. and organic N was assumed to equal total N minus NO, or NO; plus NH: concentrations. Samples that wcrc not analyscd for all forms of K within several days of being sampled were preserved with phenylmercuric acetate at the rate of I mgl .’ (Vitousek ef al.. 1952). Lysimeters obtain point estimates of N concentrations in soil water which tend to be quite variable in space and time. We attempted to homogenize some of the temporal variancr in the insecticide experiment by using monthly average values from each collector rather than weekly values. Statistical comparisons between treatments were made after applying logsrithmic transformations to homogenize variances. In the fertilizer experiment the pretreatment and posttreatment data wcrc sach ranked, and treatment differences were determined on the basis of an analysis of variance of the ranks rather than the actual N values. RESULTS

Modest reductions in arthropod and nematode densities in the prairie soil resulted in statistically significant declines in soil water NO; concentrations at 2Ocm depths on both study sites. (Two-way ANOVA by site for treatment and date, P < 0.05 for both variables with no interactions: Fig. I). A seasonal pattern, with higher NO; values seen during the nongrowing season, is also evident in these results. A reduction in detritivore densities at insecticide-treated sites corresponded with NO; values about half of those seen in the controls. A repeat

233 (151 n

OCT

NO’4

1

APFI

,

MAY

1983

(121

JUN

JUL

1884

Fig. I. Monthly volume-weighted nitrate concentrations from 20 cm deep lysimeters on 2 insecticide treated (+. n ) sites (0, a). Site i is the relatively and 2 untreated unproductive study site (0, l ) (data collected in autumn 1983 and summer 1984, n = I2 per treatment). Note Logarithmic scale. Bars are I SE. Site 2, is the refatively productive site (0, I) (data collected in 1953, n = 4 per treatment).

Fig. 2. Summary of lysimeter volumes, nitrate concentrations and organic nitrogen concentrations from insecticide treated (shaded bars) and untreated (open bars) plots. Values are means of II Iysimeters per treatment. with SE shown in parentheses.

Clipping. the plots 6 times in May and June fdiled erthcr NOI or organic N concentrations (Table I). Clipped plots averaged 4 ,IL~N 1-l as nitrate and 755 /Lg N I-’ organic N while unclipped plots averaged 7 pg N I _’ as nitrate and 288 j&g N I-’ organic N (P > 0.10 for both). Foliage production (including the clippings) for the growing season averaged 244 g m-’ on clipped plots vs 399 gm ml on unclipped plots (P < O.OOl). Clipping likely resulted in large decreases in root production and increased root mortality (c.p. Scastedt er uf., 1987). However, these postulated changes did not affect the soil water N concentrations. Adding nirrogcn and cubon to the soil in a factorial design demonstrated the strength of the soil microflora in regulating inorganic N dynamics of soil water. The addition of sucrose significantly reduced mean soil water NO; levels (two-way ANOVA of ranks P <0.05), while a similar analysis did not to affect

to produce of the experiment in 1985 failed diffcrcnces between treatments. We found no ditTcrcnces in plant biomass at the end of the growing season on these sites (Seastedt et al., 1987). Hence, results lack any apparent biological these signiticancc. No treatment effects at eithrr the 40 or 80cm depths wcrc observed. Unlike NO; concentrations. no seasonal patterns in soil organic N concentrations were found, and overall treatment means for organic N were not significantly different. (Two-way ANOVA; Fig. 2. Note: insufficient data were collected from site I for this analysis.) Significant reductions in organic N concentrations occurred with soil depth (Fig. 2). but any mechanism responsible for this reduction is unknou-n. Table

1. Nitrate

and

unmowed

organic

tallgrass

nitrogen

prairie.

concentrations

Values

are means,

m

oi

lysrmetrrs

SE and

sample

mowed

and

size’

Treatment .

Unclipped ..__I-I

pgNI-’

Date 7 May

n

.I

(1)

7

5

II)

4

Organic-N

313

(691

8

272

(4-I)

IO

27 115

(7) (46,

7 3

17 290

(6) 140)

10 R

(7)

7

SO;

I4

May

NO, Organic-N

?I

May

NO Organic-N

8

i

29 M3y

50,

IO

SO!

17 June

25

June

water

19

SE

n

$0 &Ill 2

42

data

(23) So

3

data

15 406

(51 (5%)

9 7

7 29.l

(2) (10)

9 9

IO ‘58

(6) (W)

9 9

! 240

(1) (L?)

IQ IO

SO> Org.mx-N

6 337

(3) (@)

8 8

2 303

(1) (41)

Y

3

$2;?

7

1

(I)

Y

6

283

(37)

SO>

variable

’

Organic-N

Orgaflic-N ‘The

2

r_IgNI

data (31

No

SO! Orgamc-N

IO June

(0) No

Organs-N 3 June

CiiPpLi

SE

sample

during

268 size is caused by malfunctions

the sampling

interval.

of the Iysimeters

or Inadequate

9

S soli

T. R. SEASXDT and D. C. HAIS Table :

indicate a N fertilizer cffcct (P = 0.1 I). However. a large interaction between carbon and nitrogen additions occurred (P < 0.001). and a subsequent analysis of ranks was performed on individual treatment means (Table 2). These results demonstrate that plots with inorganic N additions showed massive increases in NO; concentrations. but plots that received both nitrogen and carbon did not. Foliage mass harvested at the end of the groaing season indtcated that values obtained on the nitrogen addition plots were significantly higher than either the carbon plus nitrogen or control plots (Table 2). Foliage mass on the carbon. carbon plus nitrogen and the control plots were not significantly different from each other, while mass on the carbon plots was also not significantly different from that on the nitrogen plots. These results indicate that the carbon plus nitrogen additions did not, together. stimulate plant growth. suggesting that the addition of fixed carbon may have stimulated dcnitrification or produced some other. unknown effect. The microbial responses to nitrogen and carbon additions are complex and not always conducive to simple interpretations (Flanagan, 1986; Flanagan and Van Cleve. 1983). Interestingly, NH; values failed to show any treatment effects. suggesting that the ammonium ion is quickly removed by some combination of plant or microbial uptake, fixed on cation exchange sites, or rapidly nitrified (Vitousek and Melillo. 1979). DtSCCSSlOS

Basefow (non-storm) stream NO,- lebels at our tallgrass site range from 3 to 29 ilg N I_! during the growing season to values of 15-111 jig N I-’ during winter (Tate, unpublished). Ammonium ions in baseflow are usually below detection limits ( < 2 {tg N I -’ ). and organic S concentrations in streams exhibited less seasonality than NO;. with values ranging from 81 FgN I-’ in the growing season to 134klg N I-’ during winter. These values are similar or lower than N values observed in the 40cm deep lysimeters and suggest that the prairie acts as an efficient filter for dissolved N as it percolates through the soil. Our data represent only concentrations; absolute estimates of production or removal of dissolved N down the soil column cannot be made. However. the overall pattern is that inorganic X disappears quickly in the upper soil horizons while organic N exhibits a less dramatic but significant decline in concentrations

with soil dcprh. These soil water N concentrations are low and comparable with soil water values observed in natural ecosystems lacking large numbers of N fixers or large sources of nitric acid inputs (cf. Sollins rt al.. 1980: Vitousek t’f crl.. 1982). These low concentrations in the tallgrass prairie arc particularly imdeep prairie soil pressivc given that moderately averages over IO00 g m ’ of particulate N (Risser and Parton. IYSZ). The importance of prairie plant and fauna (earthworm) activities on soil N dynamics was demonstratcd in a microcosm study using Konza Prairie plants, fauna and soil (James and Seastcdt. 1986). The study indicated that plant roots could reduce N leaching losses about j-fold, while earthworm activitics tcndcd to enhance losses. In our field lysimeter study of clipped and untreated plants, no plant effects wers detected. while a modest decrease in soil NO, levels was correlated with a reduction in nematode and arthropod densities. Treatment differences, when present, lacked the dramatic changes observed in the laborutory microcosms and suggested that regulator) mechanisms controlling N dynamics under field conditions were markedly different from those in the laboratory. The fertilizer experiment (Table 2) demonstrated that large pulses of inorganic N can be transmitted through intact prairie soil. but that this pulse can be absorbed by a microflora stimulated by carbon (sucross) additinns. Based on ranked data. sucrose additions signiticantly reduced soil Lvater NO; levels irrespective of N additions. Our field results on plant and t‘;luna t3Tcct.s therefore should be interpreted in the context of the microbial snbironment of annually burned prairie soils. These particular prairie sites accumulate large masses of high carbon. low nitrogen root detritus (Ojima er tr/.. 1988). and microbes both immobilize the N on this detritus and recruit N from other sources (Srastedt. 1988). Pulses of inorganic b from localized sites of root mineralization or mineralization resulting from fauna feeding activities are occasionally observed in individual lysimeter samples. However. microbes often habc the opportunity to immobilize much of the N in soil water before the sample is pulled into a Iysimeter. The soil fauna in particular may therefore have a much more important, loca!izsd effect on N mincrulization than here (e.g. indicated b> mean va!ucs reported Anderson rr (IL.. IYYS). Organic N inputs to thr soil from cunopk leachates

Soil water nitrogen in

of prairie are relatively high, averaging between 90+17OO~g N I-’ (Seastedt, 1985). and organic X remains the dominant form of N in soil water and in streams of the tallgrass prairie (Fig. 2). Our lysimeter data indicate that organic N concentrations diminish by about one-third when passing through the XMOcm rooting zone, and rate’s (unpublished) results suggest another one-third disappears before the water emerges in streams. The volume of this water is also much reduced compared to volumes of rainfall inputs. The fate(s) of the dissolved N and its associated carbon substrates introduced in rainfall and canopy leachates remain a major unknown in the N cycle of the tallgrass prairie. Denitrification is possible but unlikely to account for a significant fraction of the annual input (Wuodmansee. 1978). Instead, we suspect this material is either abiotically or biotically removed from solution, thereby contributing, at least temporarily, to the particulate reservoir of N found in these prairie soils. Acknow/edgementr-We

appreciate the help of Rosemary Ramundo in field and laboratory. David J. Gibson provided useful comments on a previous draft of the manuscript. This research was supported by grants from the National Science Foundation to Kansas State University. REFERENCES

Anderson J. M., Huish S. A.. Ineson P.. Leonard ht. A. and Splat P. R. (1985) Interactions of invertebrates, microorganisms and tree roots in nitrogen and mineral element fluxes in deciduous woodland soils. In Ecological Interucricms m Soil (A. H. Fitter, D. Atkinson. D. J. Read and M. B. Usher, Ed,). pp. 377-3Y?. Blackwell Scientific Publications. Oxford. Dahlman R. C. and Kucera C. L. (1965) Root productivity and turnover in native prairie. Ecology 36, 8+89. D’Elia C. F.. Steudler P. A. and Corwin N. (1977) Determination 0r total nitrogen in aqueous samples using persulfate digestion. Limnolog~ und Oceunugruph~ 22, 76&i&I.

Flanagan P. W. (1986) Substrate quality influences onmicrobial activity and mineral availability in taiga forest floors. In Foresl Ecosystems in rhe Alurkurt Trliga: A Synrhesis of Srructure and Function (K. Van Cleve, F. S. Chapin 111. P. W. Flanagan. L. A. Viercck and C. T. Dyrness. Eds). pp. 138-1.51. Springer. New York. Flanagan P. W. and Van Cleve K. (1983) Nutrient cycling in relation to decomposition and organic-matter quality in taiga ecosystems. Cunudiun Journul of F
tallgrassprairie

729

Insham R. E.. Trofymow J. A., lngham E. R. and Coleman D. C. (,1985) Interactions of bacteria. fungi. and their nematode grazers: effects on nutrient cycling and plant growth. Ecological Monographs 55. Il9-140.

James S. W. and Seastedl T. R. (1986) Nitrogen mineralization by native and introduced earthworms: et%cts on big bluestem growth. Ecology 67. 1091-1097. Knapp A, K. and Seastedt T. R. (1986) Detritus accumulation limits productivity of tallgrass prairie. Bioscience 36, 662668. Kucera C. L.. Dahlman R. C. and Koeiling M. R. (1967) Total net productivity and turnover on an energy basis for tallgrass prairie. Ecology 48. 536531. hIcArthur J. V.. Gurtz Xl. E.. Tale C. M. and Gilliam F. S. (19S5) The interaction of biological and hydrologic phenomena that mediate the quality of water draining native tallgrass prairie on the Konza Prairie Research Satural Ayea. Proceedings of National Symposium on Non-Point Source Pollution. Kansas Citv. MO.. 22 Mav lY85. North American Lakes Managemdnt Society. . Ojima D. S., Parton W. J.. Schimel D. S. and Owensby C. E. (198s) Simulating long-term impact of burning on C, N and P cycling in a tallgrass prairie. Seventh International Symposium on Environmental Chemistry. September 1985. Rome. Italy. Nostrand Reinhold. New York. In press. Risser P. G. and Parton W. J. (1981)Ecosystem analysis of the tallgrass prairie: nitrogen cycle. Ecolog! 63, 13-l-1351. Scastedt T. R. (1985) Canopy interception of nitrogen in bulk precipitation by annually burned and unburned tallgrass prairie. Oeculnyiu 66, 88-92. Seastedt T. R. (1988) Mass. nitrogen and phosphorus dynamics in foliage and root detritus in tallgrass prairie. Ewiog\. 69, 59-65. Srastcdt i. R., Hayes D. C. and Petersen N. J. (1986) EITects of vegetation, burning and mowing on soil arthropods of tallgrass prairie. In Proceedings of the Ninrh Annuul Pruirie Con/&xce (G. K. Clambey and R. H. Pemblc, Eds). Tri-Collcgc Press. Fargo. Seastedt T. R., Todd T. C. and James S. W. (1987) Experimental manipulations of arthropod. nematode and earthworm communities in a North American tallerass prairie. Pedoohiologia 30. 9--17. Sollins P., Grier C. C., M&orison F. M.. Cromack Jr K.. Fonel R. and Fredrikson R. L. (1980) the internal element cycles of an old-growth Douglas-fir ecosystem in western Oregon. Ecological Monographs SO, X1-285. Solorzano L. (1969) Determination of ammonia in natural waters by the phenol-hypochlorite method. Limnok)g,v and Oceanography 14, 799 80 I. Sv.ank W. T., Waide J. B.. Crossley D. A. Jr and Todd R. L. (1981) Insect defoliation enhances nitrate export from forest ccosystcms. Orculo,qiu 51, 297-?YY. Vitousek P. &l. and Melillo J. IM. (IY7Y) Nitrate losses from disturbed forests: patterns and mechanisms. Forest .Science 25, 37638

I.

Vitousek P. M.. Gosz J. R., Grier C. C.. Melillo J. M. and Reiners W. A. (1982) A comparative analysis of potential nitrification and nitrate mobility in forest ecosystems. Ecoloaicul .i~onoaraphs 52. 155-l 77. Woodr&see R. G-(1$78) Ahditions and losses of nitrogen in grassland ecosystems. BioScience 28. 48453.