Primary Production and Abiotic Controls in Forests, Grasslands, and ...

Primary Production and Abiotic Controls in Forests, Grasslands, and Desert Ecosystems inthe United StatesAuthor(s): Warren L. Webb, William K. Lauenroth, Stan R. Szarek and Russell S. KinersonSource: Ecology, Vol. 64, No. 1 (Feb., 1983), pp. 134-151Published by: Ecological Society of AmericaStable URL: http://www.jstor.org/stable/1937336 .

Accessed: 03/09/2013 16:13

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.

http://www.jstor.org

This content downloaded from 128.193.8.24 on Tue, 3 Sep 2013 16:13:46 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=esa

http://www.jstor.org/stable/1937336?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


tEology, 64(i). 1983. pp. 134-15I (? 1983 by the Ecological Society of America

PRIMARY PRODUCTION AND ABIOTIC CONTROLS IN FORESTS, GRASSLANDS, AND DESERT ECOSYSTEMS

IN THE UNITED STATES'

WARREN L. WEBB2 Department of Forest Science, School of Forestry, Oregon State University,

Corvallis, Oregon 97331 USA

WILLIAM K. LAUENROTH3 Natural Resource Ecology Laboratory and Range Science Department,

Colorado State Universitv, Fort Collins, Colorado 80523 USA

STAN R. SZAREK Department ofBotany and Microbiology, A ri ona State Uni'ersitv,

Tempe, Arizonn 85281 USA

AND

RUSSELL S. KINERSON Department of Botany and Plant Pathology, University of Net' Hampshire,

Durham, Newi Hampshire 03814 USA

Abstract. This paper, a synthesis based on data generated by the International Biological Pro- gram, deals with the relationships among biotic and abiotic factors at the ecosystem level. Emphasis is placed on aboveground net primary production (ANPP), a major component of energy that drives ecosystem processes, and on potential evapotranspiration (PET), the abiotic variable most often used to explain variation in ANPP. The question addressed is: can ANPP be related to combinations of biotic and abiotic factors such that the relationships are independent of ecosystem type, whether it be forest, grassland, or desert'?

ANPP as a function of peak foliar standing crop (FSC) was best explained by models which showed a reduction in ANPP/FSC as FSC increased. Thus, deserts had a higher ANPP per unit of FSC than did other systems. As expected, photosynthetic efficiency (PE) was highest for forests, -ZOO times greater than for deserts. However, when PE was evaluated per unit of foliage, the differences in PE of ecosystems were much less. In fact, a hot-desert site had the highest PE/FSC. In terms of a theoretical maximum, the PE of forests was only 6-25% of the maximum value. Systems with nearly steady-state aboveground standing crop (ASC) showed an exponential decrease with decreased water availability (potential evapotranspiration minus precipitation). For these same systems, the ratio of ANPP to ASC increased with decreased water availability, suggesting that water-stressed systems need more energy from ANPP to drive internal processes.

A model predicting ANPP of desert-shortgrass steppes was structured in terms of FSC, water availability, and temperature. The predictive power was found to be very highs and the model was successfully validated in two of three cases with an independent data set. A model predicting ANPP of forests was structured in terms of FSC, radiation, ASC, and temperature. The deviation of the observed ANPP relative to that calculated was 17%. Deviations from predicted values were highest for deciduous stands with high ANPP and low FSC.

Most relationships exhibited good correlations between ANPP and the various independent variables including both biotic, abiotic, and combinations of the two. However, in many instances the data tended to be grouped by ecosystem type, suggesting that variation in ANPP can be reduced if ecosystem type is an added independent variable. It was surprising to find that, with the limits of our data, differences in ANPP at the ecosystem level are not glaring, especially considering that soil factors were not included in our analyses. When considering the broad range of genotypes in each ecosystem, and the much broader genotypic range representing all ecosystems, the control that native ecosystems have over abiotic factors in producing ANPP is evident but not large.

Key words: net production models; photosynthetic efficiency; potential evapotranspiration; water a vailabilitv.

Analysis of net primary production and standing crop biomass of native ecosystems was a major focus of

I Manuscript received 13 April 1981: revised 19 February 1982; accepted 23 February 1982.

2 Present address: 14151 Castle Boulevard, Number 301, Silver Spring, Maryland 20904 USA.

3Send requests for reprints to the second author.

research in the United States during the recent Inter- national Biological Program. While many papers published before and during this program reported biomass composition of ecosystems, few extensively examined interrelationships of biomass structure within ecosystem types, and even fewer comparatively analyzed ecosystem types or biomes with regard to abiotic variables. The objectives of our paper are two-



February 1983 ABIOTIC CONTROL OF PRIMARY PRODUCTION 135

12 0 110 100 90 30 T4

HO PS N ~ ~ ~ ~ ~ ILS

OT

010 100 90 80

FIG. 1. Location of data collection sites 0 forest; 0 grassland, *U desert.

fold: to make such a comparative analysis and to explain aboveground net primary production in terms of standing crop biomass and corresponding abiotic variables.

Except for the models presented in Rosensweig (1968) and Lieth and Whittaker (1975), which relate production to precipitation and temperature, no work has been as wide in scope as this paper. The data were generated largely by United States/International Bio- logical Program (US/IBP) investigators and include coniferous and deciduous forests, hot and cold deserts, and several types of grasslands. These ecosystems represent a wide range of abiotic variations; the biotic variation, both qualitatively and quantitatively, is also extensive. In nearly all cases, biotic and abiotic data are taken from the same site and in the same year. We consider that this sampling greatly strengthens the relationships discussed here.

Our hypothesis concerned the existence of relationships that explain productivity of all systems studied. In particular, aboveground net primary production (ANPP), the energy source for ecosystem processes, was expected to be an orderly function of the abiotic environment. Response of primary producers to abiotic influences has been of major interest to ecologists, especially in the case of native ecosystems where dis- turbances have been minimal, and a near steady state may exist between the plant community and the environment.

The analysis techniques we used include tabular comparisons, regression analysis, and semi-empirical

models. The basic data were used to compute variables such as biomass accumulation, photosynthetic efficiency, and potential evapotranspiration (PET). PET combined with precipitation forms an index of water availability, which is the major abiotic variable used throughout this paper.

METHODS

Biotic data collection

Biotic data were collected from 30 sites in the con- tinental United States (Fig. 1): 15 forests, II grasslands, and 4 deserts. Year of data collection, major species, and publications containing detailed descrip- tions are listed in Tables I and 2. Biomass data included net primary production and standing crop, both aboveground and, when possible, belowground. Be- cause the site of photosynthesis is especially important in the functioning of ecosystems, data on foliar production and foliar standing crop were also included. When possible, age of the dominant species was determined. Methods of biotic data collection are discussed in Webb et al. (1978).

Abiotic data collection

All of the abiotic data in our analyses of these ecosystems were collected on-site during the same year the biotic data were collected. This feature is especially critical when analyzing systems very responsive to precipitation.

Climatological data obtained included air temperature, precipitation, vapor pressure, shortwave radia-



136 WARREN L. WEBB ET AL. Ecology, Vol. 64, No. I

tion, and wind speed. The 10 sites where such data were available year-round are listed in Table I. Char- acteristics of the meteorological stations are docu- mented by Nunn et al. (1972) for Pawnee, by Turner (1973) for Rock Valley, by Balph et al. (1974) for Cur- lew Valley, by Thames (1974) for Silverbell, and by Whitford (1975) for Jornada Bajada. Data for the H. J. Andrews Experimental Forest are from R. H. War- ing (personal communication) and for Coweeta and Hubbard Brook from A. C. Federer (personal communication). Data for Thompson are from M. Smith (persotial coflimiiunication). Climatological measure- ments were collected at the location where the biotic data were obtained except for the Noe Woods site in Wisconsin, for which such data were collected at the airport station in Madison.

Arid systems respond rapidly to a fluctuating environment; in our judgement weekly averages were the maximum time interval from which we could expect to obtain reasonable relationships between abiotic and biotic variables. Therefore we averaged the data to obtain weekly as well as annual mean values. Mean yearly values for maximum and minimum temperature were compared by averaging the daily maxima and minima. Vapor pressure deficits, maximum and minimum, are for daylight hours. Annual radiation, precipitation, and wind speed are expressed as yearly to- tals. Vapor pressure deficits for Thompson, the H. J. Andrews Experimental Forest, Coweeta, Noe Woods, and Pawnee were computed from the combination of hourly temperatures and dewpoint temperatures averaged over weekly intervals. For Hubbard Brook, Silverbell, Rock Valley, Jornada Bajada, and Curlew Valley, only daily summaries of humidity and temperature were available from which to compute daily maximum, minimum, and average vapor pressure.

Computation of potential evapotra1nspira tionl

Potential evapotranspiration (PET) is an estimator of the maximum water stress on an ecosystem. To use PET we modified Penman's (I1956) equation so that our weekly abiotic data could be used. To estimate net radiation, a major component in Penman's equation, we converted shortwave to net radiation by the method of Gay (1971):

r, k = hr8y -kk, , ( )

where r1, is net radiation and raSq is shortwave radiation, both in joules per square centimetre per minute. For forested systems, k, = 0.8439 and k. = 6738. For grassland and desert systems, k, = 0.6290 and k2

7583. The average weekly shortwave radiation during daylight hours is:

rUIt = R1,J(60.24 1)o7), (2)

where r-s, is in joules per square centimetre, R,,,. is the total weekly shortwave radiation in joules per square centimetre per week, and lo1 is the ratio of daylight

hours to the total daylength. Calculating the total weekly net radiation (R,,) during daylight hours,

R, = (60.24 ID * 7) (k IFS, - k2)

=klRR. - 10080 1J, 2 (3)

where R, and R8U are expressed in joules per square centimetre per week.

Penman's equation for calculating PET is:

PE~d = A/YR nd L 71.41.8

+ 0.35(VPD)(0.5 + 0.01 U)/[A/y + IIi,

where PETd is potential water loss in millimetres per day, Rnd is net radiation in joules per square centimetre per day, L is the latent heat of vaporization in joules per cubic centimetre, VPD is the average vapor pressure deficit of Hg (in millimetres), U is wind speed (in metres per day), and A/y is the ratio of the slope of the saturation vapor pressure curve and the psy- chrometric constant. Calculating PET on a weekly ba- SIS,

PET = (7)PET d

= 41.8- _A Rn A + y L

+ 2.45 A (VPD)(0.5 + 0.008877U), A + y

(4)

where PET is water (in millimetres per week). R7, is net radiation in joules per square centimetre per week, and U is wind speed in kilometres per week. Ex- pressed as functions of temperature,

A 0.5899 - 0.012307T, (5) A + y

and

L 595.24 -0.567 T,

where T is in degrees Celsius.

Computation of photosynthetic efficiency

The efficiency of the conversion of sunlight to biomass was calculated with the use of ANPP and shortwave radiation. This photosynthetic efficiency (PE), as it is often called, was calculated on an annual basis by dividing the energy content of ANPP by the shortwave radiation in the 400-700 nm band width. While PE is often calculated on the basis of growing season, it is difficult to do so for native ecosystems because (1) the growing season often cannot be defined as a single time interval, as in the case of deserts, and (2) some ecosystem processes contributing to production during the observed growth period may continue all year.

In addition to the standard calculation, PE was determined per unit of foliar standing crop (FSC) by di-




viding it by total FSC. The resulting parameter joined radiation to leaf biomass and gave a better measure of foliar conversion of radiant energy to biomass.

Desert-shortgrrass steppe model

Several models were formulated to relate abiotic and biotic variables to ANPP. The one that give the best fit for the desert-shortgrass steppe was

52

ANPP =A FSC 2

sinh[B max(O, PPT = PET)Ati], (6)

where ANPP is in grams per square metre per year, FSC is foliar standing crop in grams per square metre, 7i is weekly average temperature in degrees Celsius, sinh is hyperbolic sine (an average of two exponential terms), PPT is weekly precipitation (in millimetres), PET is potential evapotranspiration (in millimetres), Xti is the fraction of a year, and A and B are parameters derived from the data. The model has two distinct components. The input A FSC, computes the linear relationship between ANPP and FSC for all ecosystems. The second component shifts the calculated ANPP depending upon water availability (PPT - PET) and temperature. Parameter A has units of yr-1 and B has units of yr/mm.

Since arid systems are very responsive to rainfall, which falls infrequently and somewhat randomly, the use of weekly time resolution was necessary, as was the stipulation that production data and abiotic data be from the same site and in the same year. Temper- ature was included in the model since the average annual temperature range was large, between 7.20 and 21.40C. FSC was the primary biotic variable, both because of the strong relationship found between it and ANPP and because ecophysiological field studies com- monly relate abiotic variables to plant production through leaf biomass or leaf surface area. The major variable, water deficit, was closely coupled to temperature. This coupling was based on the recognized relationship between the temperature needed to drive chemical reactions and the water needed to allow them to proceed at their maximum rates. The temperature function followed the Q,( rule which allows a rate doubling for each 10C increment. Light intensity was excluded from the model because its annual variation per site was minimal and because the coefficient of variation for all sites was low, 8%.

Parameters A and B were derived from daily temperatures, daily precipitation, and computed weekly PET combined with peak foliar biomass and annual net production. Values for A and B were obtained by calculating ANPP with a computer program which it- erated paired values for A and B consistent with the measured weekly abiotic variables and annual biomass variables of FSC and ANPP. This procedure was fol-

lowed for each site and for each year on a given site; it yielded a series of relationships between A and B, one relationship for each site and for each year.

Forest inodel

The forest systems listed in Table 2 represent di- verse aboveground biotic characteristics which occur over a wide range of environmental conditions and geographical areas. Though it may be optimistic to expect that the variance in ANPP can be adequately resolved by a single model, there is sufficient information now available to expect a first-generation model. Several forms were tried with the available data; the one that gave the best results related ANPP to FSC, aboveground standing crop (ASC), shortwave radiation (RAD), and annual average temperature (T):

ANPP = [A' FSC RAD - B' ASC]27"1'' (7)

Conventionally, ANPP equals gross photosynthesis (an input) minus autotrophic respiration (a loss), so that

Input= [2T11 -FSC RAD]A', (8)

where T is annual average temperature in degrees Cel- sius, FSC is peak foliar standing crop in grams per square metre, RAD is total annual shortwave radiation in joules per square metre, and A' is a parameter (measured in joules).

Loss = 2T110 ASC B', (9)

where ASC is aboveground standing crop in grams per square metre. Parameter B' has units of yr-1.

The input term incorporates radiation multiplied by FSC, implying that a linear relationship exists between the input and radiation. While such a relationship is not realistic for single leaves which do not utilize two- thirds of the incident radiation, it is realistic when foliar biomass approaches a maximum so that <10% of the incident radiation penetrates to the forest floor. The second term in the model (B' ASC) is propor- tional to the aboveground biomass and is analogous to a respirational cost for maintenance and production processes. Both forms of the model are multiplied by a temperature correction which allows for a doubling of both input and loss terms for each 1?O change in annual temperature. The base temperature is 10?, the same as for the desert model. Because the forest systems in the study are not significantly water stressed, no correction term is applied for this factor.

Although the model is conceptually pleasing, in the most rigorous sense it must be considered empirical because it is not possible to obtain theoretical parameter values when the data do not include belowground productivity. The model form is, however, independent of stand age so that systems do not have to be mature to be incorporated into the results.




TABLE 1. Summary of annual abiotic characteristics for sampled forest, grassland, and desert ecosystems.

Potential evapo- Temperature Vapor pressure Shortwave Wind

latitttde/ Precip- trans (IC) deficit (mm Hg) radiation speed

Ecosystem longitude itation piration*

type Site naute Year (degrees) (mm/yr) (mm/yr) Mean Max. Min. Mean Max. Min. (J Cm-2 yr -) (km/yr)

Forest Thompson. Washington 1973 47/122 1433 629 9.2 13.6 5.1 3.4 4.4 1.2 406 710 38 000

Andrews Forest. Oregon 1975 44/122 2592 370 7.3 13.7 2.3 2.4 4.9 0.3 344 850 61)00

Coweeta. North Carolina 1972 35/83 1924 701 12.1 18.4 8.1 4.7 7.9 1.5 493 240 1900

Hubbard Brook, New Hampshire 1972 44/72 (430 603 5.5 10.8 0.1 3.3 6.1 0.0 438 806 29 400

Noe Woods, Wisconsin 1971 43/89 639 1014 7.8 14.2 1.5 2.7 8.5 -0.3 549 670 141 400

Grassland Pawnee, Colorado 1974 41/105 241 1338 111.0 18.7 1.3 9.6 14.1 3.1 591 470 87 100

)esert 1972 315 1052 7.6 15.4 0.0 6.8 7.0 2.1 644 970 62 500

Curlew Valley, 1973 261 1049 7.2 14.4 -0.1 7.1 10.6 3.8 671 730 53 3)0

Utah 1974 42/113 204 1088 8.3 16.1 0.5 6.9 10.6 3.3 667 550 61 ()0)

Silverhbell 1972 344 1820 21.4 28.4 14.2 18.1 26.5 10.0 713 250 66 700

Arizona 1973 32/112 21)9 1928 20.7 27.4 14.0 18.4 26.0 11.1 748 660 76 300

1972 311 1192 18.5 25.3 11.6 10.8 16.5 5.0 648 160 53700

Jornada Bajada. 1973 237 1287 18.4 25.2 11.5 12.3 18.2 6.3 687 200 49 000

New Mexico 1974 33/107 394 1340 18.6 25.2 11.9 13.4 19.1 7.7 659 140 51 600

Rock Valley. 1972 116 1905 16.3 22.5 9.9 12.1 18.3 5.8 789 650 100 900

Nevada 1973 37/116 213 1978 18.6 24.8 12.1 14.4 21.5 7.3 763 220 104 400

Computed weekly with modified Penman (1956) equation.

RESULTS AND DISCUSSION

Abiotic data

To provide a perspective for results that appear later in the paper we first give a descriptive analysis of the abiotic and biotic features of the sites. The sampled sites cover an extremely wide range of precipitation and temperature (Table 1). Annual precipitation ranged from 116 mm in 1972 at the Rock Valley desert site to 2592 mm in 1975 at the H. J. Andrews Experimental Forest, a 20-fold difference. The mean annual temperature in 1972 varied from 5.50C at Hubbard Brook to 21.4? at Silverbell, a four-fold difference. The precipitation gradient is closely related to ecosystem type. In general, forest sites have high annual precipitation, while the deserts and grasslands are characterized by lower values. The average annual precipitation for the forest sites listed in Table I is 1606 mm, while for deserts and the grassland site the average is 259 mm, 84% less.

The annual temperature gradient is not well correlated with ecosystem type. For example, Curlew Val- ley, a cold desert, has a mean annual temperature of 7.7?, while Andrews Forest and Noe Woods have nearly the same mean annual temperatures, 7.3? and 7.8?. In general, warmer temperatures are associated with deserts and forests, but in this group of sites differences are evident. For example, Coweeta, a forested site, has an annual temperature of 12?, but all deserts reported have values above 16?.

PET averages 663 mm/yr for forests and 1452 mm/ yr for the deserts and shortgrass steppes. For the water- limited systems, annual precipitation for a given site may be quite variable, but annual PET is relatively constant. For example, precipitation at Rock Valley was 116 mm/yr in 1972 but nearly doubled to 213 mm the following year. Annual PET, however, varied only 4% between these 2 yr. Data for the other desert sites are similar. It appears that water-stressed systems exist in environments with variable precipitation but a constantly high evaporative demand. Thus, parameters such as annual temperature, vapor pressure deficit, and radiation, while variable among sites, are relatively constant among years for a given site.

Annual shortwave radiation varies >50% among the sampled sites. The lowest value, 344 850 J cm-2 yr-1, was reported from the Andrews Forest, whereas the highest, 789 650 J cm-2 yr-1, was reported from Rock Valley. Radiation was significantly lower on the Jor- nada Bajada desert than on Silverbell or Rock Valley deserts, which are at comparable latitudes.

For the forested systems listed in Table 1, there is almost an inverse linear relationship between annual precipitation and annual shortwave radiation. On the Andrews Forest, precipitation in 1975 was 2592 mm, while total shortwave radiation was only 344 850 J/cm. In contrast, radiation at Noe Woods was greater (549 670 J/cm2), whereas precipitation was corre- spondingly lower (639 mm/yr). During much of the year, Andrews Forest regularly has cloud cover with




FOREST

200 ANDREWS

1974 -75

U X L- FOREST ',100 100 HUBBARD BROOKO

RASSLAND COLD DESERT X HT L ESER

? ~~~~ ~ ~~~~~~197210 _100EPAWNEE a 100 CURLEALLEY

100 SILVER BELL

50 - - 1973

AN. DEC. JAN. DEC. JAN.DEC50.

1OU COWEETA lOG NOE WOODS 0 19721971OO~~

50 50- -1 BAJADA I A JEE~~aflhlI B~ 11973

E 50L- 0 C

0 .0 GRASSLAND COLD DESERT o-0

-=100 -IOU Fl 100 -U PAWNEE - CURLEW VALLEY ROCK VALLEY .1 1974 1 9 1973 B B

B50 ..E 50-5 50-

WQ. Wa. 0.0. 0.0.

JAN. DEC. JAN. DEC. JAN. DEC. MONTH MONTH MONTH

FIG. 2. Seasonal patterns of precipitation (PPT) and potential evapotranspiration (PET) at selected study sites. PET computed weekly with Penman's (1956) equation. Shading indicates that PET is greater than PPT.

attendant precipitation. The other forest sites receiv- ing high precipitation also have significantly reduced radiation.

Weekly precipitation and evapotranspiration are particularly useful for comparing ecosystems. The difference between weekly precipitation and PET can be interpreted as an index of the effectiveness of precipitation and of the relative water stress under which the biota of a particular ecosystem is existing. As one would expect, there is a gradient in water deficits from forests to grasslands to deserts (Fig. 2). When the deficits are calculated weekly, the deciduous forest at Hubbard Brook and the pine plantation at Coweeta have the lowest annual water deficits: 153 and 188 mm, respectively. The annual deficit at the Pseudotsuga rnenziesii site on the Andrews Forest is 291 mm, and the deficit at Noe Woods is 659 mm. The explanation for the greater deficits at Andrews Forest than at either Hubbard Brook or Coweeta relates to the seasonal distribution of precipitation. Precipitation is evenly distributed throughout the year at both Hubbard Brook and Coweeta, but at Andrews Forest there is an excess of PET between May and October. Very little precipitation occurs during the summer months. Moisture conditions of the forest sites can be summarized, then, as generally favorable but with intervening drought periods.

At the grassland and desert sites, in contrast, drought prevails, but there are short periods of favorable soil- water conditions. Deficits at these sites range from 924

mm at Curlew Valley to 1821 mm at Rock Valley. The large deficit at Rock Valley is partially the result of the concentration of precipitation during the fall and winter months when PET is low and of the almost complete absence of precipitation during the rest of the year. The relatively low water deficit for Jornada Bajada is explained by the midsummer peak in precipitation, which lowers both PET and the water deficit.

Biotic (a1ta

The data represent an extremely wide range of biotic values in terms of both standing crop and net primary production (Table 2). Others have also presented such data (e.g., Westlake 1963). Aboveground standing crop (ASC) is highest for the Wildcat Mountain forested site in Oregon (88 200 g/m2) and lowest for the Ale grassland site in Washington (70 g/m2), a difference greater than three orders of magnitude. Similarly, net primary production varies from 3279 to 16 g-m-2 yr-1; the most productive site was a young Pinuis taeda stand at Re- search Triangle Park, North Carolina, while the least productive was a hot-desert site, Rock Valley, Ne- vada, during a dry year. This difference in net primary productivity is greater than two orders of magnitude.

Annual foliage production was lowest for Rock Val- ley (10 g-m-2 yr-1) and =50 times higher for the pine site at Triangle Park (561 g-m-2 yr-1). Peak FSC, however, was highest for a hemlock forest near Otis, Oregon (2100 g/m2). Except for another coniferous stand, Wildcat Mountain, FSC on all the sites was



14() WARREN L. WEBB ET AL. Ecology, Vol. 64, No. I

TABLE 2. Net primary production and standing crop data from the 30 forest, grassland, and desert sites sampled in the United States. M = data missing or not collected.

Aboveground Belowground Foliar

Net Pro- primary Net duc- Ap- produc- produc- tion Stand- prox-

Year tion Standing tion Standing (g ing mate Ecosystem data (g m cr - Clop (g m-2 - crop m 2 crop age

type Dominant species Site name` taken yr 1) (g/m2) yr-) (g/m2) yl 1) (g/m2) (yr) References

Coniferous '.Seudoixug.sugi nnclziexii Thompson, 1974 NI 994 NI 93 36 104 9 Turner and Long (1975) forest Washington 1974 1150 12 650 555 NI 210 499 22

1974 987 15 299 539 N1 314 621 30 1974 Ni 17 154 NI 3298 199 910 36 1974 1027 19 657 428 NI 223 827 42 1974 876 29 532 447 NI 228 1074 73 1974 830 34 811 443 NI 214 1288 95

Pfvw(lotmixu 11nclzie'vii Andrews Forest, 1972 797 83 855 290 13 900 272 1'203 450 Grier and Logan (1977) Oregon

/luga heterophvlla Otis. Oregon NI 3070 19 270 N1 N 272 2 100 26 Fujimori ( 1971)

/isluga heterophv I lla Cascade Hd., NI 1030 87 500 N1 NI NI 790 1 10 Fujimori et al. (1976) Picc .sitchens.is Oregon

Psculllomisga Illenziexsii, Blue River, Oregon M 1270 66 900 NI NI NI 1110 100 Fujimori et al. (1976) /Ua .ll ierophl./la

Ahiexs procera Wildcat Mountain NI 1300 88 200 NI NI NI 1750 1 15 Fujimori et al. (1976) P.s ludot~islga (nenziesvii Oregon

P/it/us uacda Triangle Park. 1970 NI 6776 N1 1798 506 911 13 Kinerson et al. (1977) North Carolina 1971 2852 7947 1042 2008 523 941 14

1972 2983 9193 1065 2 173 561 101( 15 1973 3041 10 310 1139 2342 534 961 16 1974 3279 1 1 792 1363 2620 543 977 17

Piiu.s sirohus Coweeta no. 1. 1972 135() 6960 300 1500 363 839 15

North Carolina

Deciduous Acer saccharumn. Hubbard Brook. 1966 890 15 700 165 3142 290 290 1(0 Whittaker et al. (1974) forest Betula l// eghanictisis. New Hampshire

Liriodendroll illipi/fi'ra Oak Ridge, Tennes- 1971 650 14 000 720 3320 372 372 NI

see

Quercu) s alba. Noe Woods. Wis- 1971 788 26 525 NI 6603 435 435 13() Lawson et al. (1972) (. elitilla, consin PrunuA Serolitina

Quercus 1lba, Brookhaven. New 1964 864 6384 333 3631 351 412 43 Whittaker and Q. Co(Cuilul, York Woodwell ( 1969) Pinu11.s rigid

Quercus abIh, Nakoma, Wiscon- 1971 716 12 936 N1 N1 NI 259 NI Lawson et al. (1972) (9. velutina sin

Quercuts sit//lat, Norman, Oklahoma 1971 1 192 21 957 NI NI NI 476 NI Johnson and Risser (9. marilaldica (1974)

Qucrcul s pri/u. Coweeta no. 18, 1973 876 14 037 NI 3065 561 561 60

(Q. c((j(ilC(l North Carolina

Car*va Klabra

Grassland Agropyron spicaluilm, Ale, Washington 1971 82 70 NI N1 82 70 NI Sims and Singh (1978) Aremisila tridentaa 1972 1 14 80 NI M 1 14 80 NI Sims et al. ( 1978)

fevsiluc scabrelia, Bison, Montana 1970 272 220 N1 NI 272 220 NI Sims and Singh (1978)

t-I. idalo/wetsi Sims et al. (1978)

l'esica idahllocisis. Bridger. Montana 1970 168 145 NI MI 168 145 NI Sims et al. (1978) Agrop/vron sulhwund.un101111 1972 330 300 471 1700 3 30 300 NI

Atgropvro stnithii Cottonwood, 1970 212 160 516 900 212 160 NI Sims et al. (1979) BachIloe dacIloi&e.s South Dakota 1971 255 210 821 1700 255 2'10 NI

1972 279 195 304 1850 279 195 NI

Stil)(i Colm .Dll(l Dickinson. South 1970 351 280 932 2200 351 28() M Sims and Singh (1978) Agr Vo/pI .mlltiihii Dakota Sims et al. ( 1978)

Anidropogon .soparius. Hays, Kansas 1970 363 225 1062 1950 363 225 NM Sims and Singh (1978) A. .vradjli Sims et al. ( 1978)

Boutelolow eiop)oda~ Jornadla Bajadla 197() 134 137 91 240) 134 137 NI Sims andl Singh (1978) New Mexico 1971 125 137 254 215 125 137 MN Sims et a1l. (1978)

1972 86( 110( 96 180) 186 110) N1



February 1983 ABIOTIC CONTROL OF PRIMARY PROI)UCTION 141

TABLE 2. Continued.

Aboveground BeIoWgFoLnd Foliir

Net Pro- primary Net duc- Ap- prodLIC- produLC- tion Stand- prox-

Year tion Standing tion Standing (g ing imate Ecosystem data (g m 2 crop (ginm-2 crop m 2 crop age

type Dominant species Site name' taken yr ') (g/ m2) yr ') (g nm2) yr ) (g/ m) (yr) References

Anidropogoni scoparius, Osage, Oklahoma 1970 331 280 602 1400 331 2,80 M Sims and Singh (1978) Sorghastrum 1t1italns 1971 416 345 431 1600 416 345 M Sims et al. (1978)

1972 290 275 592 850 290 275 M

Boute/loa grac/is Pantex, Texas 1970 155 260 417 1200 155 260 M Sims and Singh (1978) 1971 289 240 606 490 289 240 M Sims et al. ( 1978) 1972 327 245 876 1000 327 245 M

Bouteloa grac/ilis, Pawnee, Colorado 1970 182 162 512 1366 182 182 M Sims and Singh (1978) O)puttia po/vacatitha 1971 114 119 378 558 114 114 M Sims et al. ( 1978)

1972 166 176 250 690 166 166 M

1973 121 190 110 760 121 121 M

1975 99 131 250 829 99 99 M

Brownis nollis San Joaquin, 1973 457 457 582 757 457 457 1 Sims and Singh (1978) California 1974 457 457 451 623 457 457 1 Sims et al. (1978)

1975 343 343 361 543 343 343 1

Cold desert Artemnisia tridletntata, Curlew Valley, 1971 M 564 M 2086 M M 16 Balph (1973) Atrip/(ex 5ontjfrti/ia Utah 1972 M 521 M 2249 M M 16

1973 149 505 1177 3436 95 124 16 1974 70 409 552 2240 51 74 16

Hot desert Ambrosia d(Ie/toi(da, Silverbell, Arizona 1972 66 116 M M 24 50 M Thames (1973) Larrea dii'aricata,

0111()/ a tesota

Lairra diivaricawt Jornada Bajada, 1971 161 396 M 548 14 26 M

Yc(c elata, New Mexico 1972 2 9 !2 M M 86 43 M

Prosopis glandhulosa 1973 129 M M NI 51 48 M

1974 1(0 M M M 37 51 M

Ambrosia dum-nosa, Rock Valley, 1971 16 93 M M 10 M M Turner (1973) Larrea aifaricrata, Nevada 1972 19 100 M M II M NI

1973 125 M M 509 24 M NI

1974 20 120 M 600 M M M

See Table 1 for state code.

< 50% of the maximum value for Otis. Jornada Ba- jada had FSC values as low as 26 g/m2, or =1/100 of the maximum.

ANPP related to FSC

Aboveground net primary production (ANPP) was found to be positively related to peak foliar standing crop (FSC; Fig. 3). The equation for this relationship, in log-transformed form, is:

In ANPP = 0.76 + 0.93 In FSC. (10)

The estimated intercept is small (22.2 g.m-2 yr-1), and is not significantly different from zero. The estimated slope coefficient is significantly different from zero; the coefficient of determination is 0.79. The anti-log of this equation is:

ANPP = 2.23 FSCt93. (11)

When the parameters in Eq. 11 were used as first ap- proximations, nonlinear fitting to the curve resulted in a similar expression:

ANPP = 5.0 FSC83. (12)

While Eq. 12 gives slightly higher values for ANPP, both Eqs. 11 and 12 are somewhat curvilinear downward over the range of the data. A linear model fitted to the data gave a positive intercept of 82 g m-2. yr- for ANPP, which was not significantly different from zero, and a slope coefficient of 1.41 for ANPP/FSC, which was significantly different from zero. The coefficient of determination was 0.58. Gholz (1979) found a linear relationship between ANPP and leaf surface area for stands with ANPP between 30 and 1400 g m-2yr l.

Eq. 12 was evaluated in terms of ecosystem type by computing its slope (dANPP/dFSC) for the average foliar biomass of each type (Table 3). The result, ANPP/ FSC, was compared to the average value, which im- plies a linear and positive relationship between ANPP and FSC for each ecosystem type. Correlation coef- ficients between ANPP and FSC were included to indicate the relationship between ANPP and FSC for each biome.

The general model (Eq. 12) indicates that primary production per unit of foliage is higher for systems




100007 - - -

2 r 0 794 In ANPP= 076+ 093 In FSC

1000 .**+* -

C\j 0

E ~~~~~~0

100_ * 0

0

*CONIFEROUS FORESTS * DECIDUOUS FORESTS

o GRASSLANDS * DESERTS

10I D 100 1000 10000

FSC (g/m2)

FIG. 3. Relationship between aboveground net primary production (ANPP) and peak foliar standing crop (FSC) of forests, grasslands, and desert sites.

with low foliar biomass. Comparison of the data from the four ecosystem types revealed that desert systems have the highest ratio of ANPP/FSC, 2.3. ANPP/FSC for Jornada Bajada is the highest in this study. Includ- ing these data resulted in a negative correlation between ANPP and FSC for desert systems, while the remainder of the ecosystem types have positive correlations for this relationship. Without the Joranda Ba- jada data, the desert systems average 1.3 for ANPP/ FSC.

There is a strong correlation between ANPP and FSC for the grasslands, 1.21 (Table 3). For most tem- perate grasslands, ANPP/FSC would be expected to be 1.0. The presence of other life forms increases the value of the ratio above unity. Although the grassland data seemingly form the core of the relationship between ANPP and FSC, there was significant positive correlation when grassland data were deleted (r2 = .50). The intercept value was large (320 g 2yr 1), but not significantly different from zero.

Data from coniferous forests include both young vigorous stands (e.g., Triangle Park, North Carolina) and mature stands. Deciduous forests (for which no young stands were represented in the data) have an average ANPP/FSC value of 2.22. When young stands are excluded from coniferous forests, the average value for ANPP/FSC is 0.84, or about one-third the corresponding value for deciduous forests. This comparison seems especially important because all included systems are mature, and the difference between the two systems is significant. Since deciduous systems produce more with less leaf biomass, and since complete leaf turnover occurs each year, the implications for faster nutrient cycling are significant.

Zavitkovski et al. (1974), using their data and that

TABLE 3. Aboveground net primary production (ANPP) related to peak foliar standing crop (FSC) from Eq. 12, and average ANPP/FSC for each ecosystem type.

d Average Corre- Number (ANPP)* ANPP lationt

of coef- Ecosystem type sites/yr d(FSC) FSC ficient

Desert 7 2.50 2.33 -.17 Grassland 25 2.02 1.21 .93 Deciduous forest 7 1.80 2.22 .21 Coniferous forest 25 1.55 1.64 .40

* First derivative, taken from Eq. 12. t Correlation between ANPP and FSC for each ecosystem

type.

from the literature, found deciduous systems to produce 3.41 g/g of foliar biomass. They also found that coniferous systems were less productive, with an average ANPP/FSC of 1.6, but their data for coniferous systems had a significant intercept value greater than the one reported for the deciduous systems. The systems would be more nearly the same if the regression intercepts were forced through zero.

While all models relating FSC to ANPP were either linear or slightly curvilinear downward, none was curvilinear upward. Another possibility exists besides the models discussed. A piecewise linear (or curvilinear) model is possible with different parameters for different ecosystem types. The data we have are not well enough correlated within an ecosystem type to make such a test. It is especially difficult to do so when the climatic state of the system is not known. The data in Table 3 suggest the order of efficiency to be deserts >

deciduous forests > coniferous forests > grasslands, but the differences are not statistically significant. Whittaker and Niering (1975), in an altitudinal gradient study, reported that visual trend lines fitted to their data suggested that desert grasslands produce more per unit of leaf area than do coniferous forests.

Photosynthetic efficiency

Although photosynthetic efficiency of natural ecosystems normally is assumed to be low, this assumption may hinge on the techniques used and the time period over which the calculation is made. The time period used here is yearly, as this is the only basis possible for comparing systems when the growing season cannot be determined. Furthermore, an estimate of the incident photosynthetically active radiation (PAR) rather than actual intercepted light is used, as is aboveground production rather than moles of 02

evolved or CO2 taken up. As such, the calculations used here will result in the lowest possible measure of efficiency for those techniques and time periods. Jor- don (1971) has calculated efficiencies for a variety of ecosystems.

Annual PAR varied from a low of 1 546 000 kJ/m2 at




TABLE 4. Annual photosynthetic efficiency by ecosystem type for aboveground net primary production (ANPP) and total net primary production (TNPP). Both values are also expressed relative to the foliar standing crop (FSC). Dashes indicate insufficient data.

Photosynthetic efficiency (%)t

PARt ANPP/FSC TNPP/FSC Ecosystem type and site* (kJ.m-2 yr-') ANPP TNPP (x 10-3)? (X 10-3)?

Coniferous forest Thompson, Washington 1 894 000 1.04 1.56 1.42 2.10 Andrews Forest, Oregon" 1 546 000 1.04 1.42 0.86 1.18 Otis, Oregon" 2 127 400 2.93 0.49 Cascade Head, Oregon" 2 127 400 0.98 1.24 Blue River, Oregon 1 553 700 1.66 0.89 Triangle Park, North Carolina 2 344 900 2.65 3.65 2.73 3.78 Coweeta no. 1, North Carolina 2 200 000 1.23 1.51 1.5 2.7

Deciduous forest Hubbard Brook, New Hampshire" 1 972 600 0.82 0.97 2.8 3.3 Noe Woods, Wisconsin 2 472 900 0.58 1.3 Nakoma, Wisconsin 2 689 600 0.53 2.0 Norman, Oklahoma 2 689 800 0.81 1.7 Coweeta no. 18, North Carolina 2 220 000 0.72 1.3 Oak Ridge, Tennessee 2 313 600 0.57 1.5 Brookhaven, New York 2 319 500 0.76 1.8

Grassland Pawnee, Colorado 2 661 600 0.037 0.62

Desert Curlew Valley, Utah (1973) 3 023 100 0.095 0.76 Curlew Valley, Utah (1974) 3 005 300 0.045 0.61 Jornada Bajada, New Mexico (1972) 2 916 700 0.195 4.5 Jornada Bajada, New Mexico (1973) 3 092 400 0.081 1.7 Jornada Bajada, New Mexico (1974) 2 966 100 0.066 1.3 Silverbell, Arizona 3 209 600 0.039 0.78 Rock Valley, Nevada (1972) 3 553 400 0.01 Rock Valley, Nevada (1973) 3 434 700 0.07

* See Table 1 for state code. t Calculated with photosynthetically active radiation (PAR) equal to 0.45 incident shortwave radiation. t Calculated with mean energy content for vegetative type in: coniferous forest 20 273J/g, deciduous forest 18 183J/g,

grassland 16 720J/g, and desert 19 437J/g (Larcher 1975). ? Tabled values should be divided by 103. " Radiation and biotic data not taken in the same year.

Andrews Forest, Oregon, to a high of 3 553 400 kJ/m2 at Rock Valley, Arizona, more than a two-fold difference (Table 4). In general, forested sites have significantly lower annual PAR than do desert sites. Sites of comparable latitude have large differences in radiation, e.g., Curlew, Utah, at latitude 420N has an annual PAR of 3 023 100 kJ/m2, while at Andrews Forest PAR is 50% of this value. The former is a high-elevation cold desert, and the latter a mature coniferous forest.

The efficiency with which radiant energy is converted into dry matter is indicated in Table 4. On a land surface area basis, forest systems are more efficient than deserts or the shortgrass steppe sites listed. The most efficient system was a young hemlock stand in Otis, Oregon, which converted 2.93% of its shortwave radiation to aboveground biomass. This conversion was more than two orders of magnitude higher than that at the Rock Valley, Nevada, desert site, where the efficiency was 0.01% in 1972, a dry year.

When coniferous and deciduous forests were compared, the annual mean photosynthetic efficiency was

1.69% for all coniferous systems reported and only 0.689% for deciduous systems. Included in the average for the coniferous systems were three young stands with high productivity (Coweeta no. 1, Triangle Park, and Otis). No two deciduous stands reported are comparable in stand age. When data from the young conifer stands were excluded, the annual efficiency was 1. 11% or slightly less than twice that of deciduous systems. When the efficiency of the deciduous systems was reported only for the period of foliation, the efficiency was 1.31%, more comparable to coniferous systems.

Another way to compare photosynthetic efficiency of ecosystems is to express it per unit of foliage. This measure is more realistic when comparing photosynthetic efficiency of different types of vegetation. It is noteworthy that when the photosynthetic efficiency of the productive Piniis taeda stand at Triangle Park, North Carolina, was expressed on an ANPP basis, the value was 2.6%, while that of a productive hemlock stand near Otis, Oregon, was a comparable 2.9%.




TABLE 5. Theoretical photosynthetic efficiencies for forest sites in comparison with actual production. Theoretical values assume 10% of photosynthetically active radiation (PAR) is converted to dry mass and 1 g of dry mass is equivalent to 20 273 J.

Actual dry mass

Theoretical PAR ANPP dry mass gain Theoretical

dry mass Forest type and site* (kJ-m -2yr') (g m2 yr) (g m2 yr) (%)

Coniferous Andrews Forest, Oregon 1 546 000 797 7630 10 Thompson, Washington 1 894 000 974t 9340 10 Otis, Oregon 2 127 400 3070 10 490 29 Cascade Head, Oregon 2 127 400 1030 10 490 10 Blue River, Oregon 1 553 700 1270 7630 17 Coweeta no. 1, North Carolina 2 220 000 1350 1110 12 Triangle Park, North Carolina 2 344 900 2890 11 560 25

Deciduous Hubbard Brook, New Hampshire 1 972 600 890 9730 9 Noe Woods, Wisconsin 2 472 900 788 12 200 6 Nakoma, Wisconsin 2 689 600 716 12 200 6 Norman, Oklahoma 2 689 800 1192 13 300 9 Coweeta no. 18, North Carolina 2 220 000 876 11 000 8 Oak Ridge, Tennessee 2 313 600 650 11 400 5 Brookhaven, New York 2 319 500 864 11 400 8

* See Table 1 for state code. t Average of five sites.

However, when expressed per unit of foliage, the efficiency of pine was 2.7 x 10-39%/g of foliage, while that of hemlock dropped to 0.49 x 10-39%/g. Although the latter is the lowest efficiency per unit of foliage reported, this stand of Tsuga heterophylla also has the highest foliar standing crop reported (Table 2). The highest efficiency per unit of foliage for a single year was reported at Jornada Bajada in 1972: 4.5 x 10-317%/ g of foliage. As pointed out previously, this system, while considered a desert, has a midsummer rainy season (Fig. 2).

While it is instructive to compare PE of sites and biomes, it is more useful to relate PE to its theoretical maximum efficiency. The theoretical maximum photosynthetic efficiency that can be achieved is usually considered to be between 10 and 12%; most measure- ments have been made on Chlorella for short time periods. Wassinke (1959), however, reports values >10-12% for higher plants.

If we use 10% as the upper limit of PAR conversion to carbohydrate, then the theoretical maximum gain in mass of forest sites, independent of species or vegetation density, varies from 7.6 x 103 g.m-2 yr-1 at Andrews Forest to 1.33 x 104 g.m-2 yr-1 at Norman (Table 5). The deciduous stands have a slightly higher maximum than do coniferous ones. The upper limit does not include whole-plant respiration; thus, comparison with such a limit is only a convenient reference point. When this comparison is made, ANPP of coniferous and deciduous forests varies from 5 to 9% of the theoretical maximum (Table 5). The young stands

are the most efficient, varying from 12 to 25% of the theoretical limit, the deciduous stands have slightly lower efficiencies than do the coniferous ones.

Estimates of gross photosynthesis at three forest sites are available for comparison with the upper theoretical limit. Such estimates have been obtained from labo- rious measurement of CO2 exchange at Triangle Park, North Carolina; Brookhaven, New York; and Oak Ridge, Tennessee. These estimates were made in the conventional manner by adding respiration during the dark period to net CO2 exchange measured in the light period. Actual CO2 uptake relative to the theoretical maximum was 71% for Triangle Park, 22% for Brook- haven, and 38% for Oak Ridge, all on an annual basis. Thus, the young Pinus taeda stand in North Carolina approaches the theoretical upper limit of radiant energy capture. The other two stands have significantly lower efficiencies, perhaps because of lack of foliage during a large part of the year. Foliage of the pine stand at Triangle Park amounts to 1000 g/m2, which is not excessively high for conifers, but the temperature pattern in eastern North Carolina undoubtedly allows for year-round photosynthesis.

Bray (1961) calculated efficiency of a forest stand to be 7.9%. He calculated the minimum conversion efficiency and concluded it was greater for shorter periods during the year. Botkin and Malone (1968) calculated efficiencies as high as 10% for a short period of active growth. The longer the period, the lower the efficiency they computed. They used the harvest technique and measured the light actually intercepted by the vege-




100000 A

ANDREWS

\2-0.82

0NOE

\ THOMPSON

( HUBBARD BROOK

COWE E TA 10000 18 1.0

SILVERBELL (72)

0 AsC *PAWNEE (74)

* ANPP/ASC CURLEW *JORNADA(71) (73)0

EWATER SURPLUS ROCK VALLEY

,' (7 2) 'CURLEW

2 /\(74)

10- r-:0 71// \ 0.10

COWEETA 8 / \

HUBBARD BROOK , OCURLEW (73) / CURLEW (74)CE)JORNADA (71)

/ NOE ,'THOMPSON

C'J PAWNEE (7)z EM/ \0 ROCK -7

/ANDREWS SlLVERBELL 72)0ALLEY(7 U

v 1001 --- LEBL(2 I

00

-2000 -1000 0 1000 2000 > n ()

700 0 I I 07

B

600/ 0.6

500 - /05

400 - /ANPP -0 34 + 0.00067 (PET- PPT) 0.4

300 - - 0.3

ASC - 900e-000177(PET-PPT)

200 _ 02

0

100 _0.1

o I l I l l o0 0 1000 2000

PET - PPT (mm/ yr)

FIG. 4. Relationship between annual water availability (PET - PPT) and aboveground standing crop (ASC), as well as the ratio of aboveground net primary production (ANPP) to ASC, for deserts, grasslands, and forests (A) and for water- stressed systems only (B). PET computed weekly with Pen- man's (1956) equation. Year of measurement given in paren- theses following site name in Part A.

tation. Yocum et al. (1964) measured CO, fluxes in a corn stand and calculated efficiencies as high as 40% of the theoretical maximum for a short period during the day (5.%l on absolute basis). This efficiency decreased when calculated on a daily basis to 3.2%.

For deserts and shortgrass steppes, the theoretical limit is higher than for forests, ranging from a maximum at Rock Valley, Nevada, of 1.82 x 104 g.m-2

yr-1 to a low at Pawnee, Colorado, of 1.59 x 104 g m- 2.yr-1. Actual efficiency relative to the theoretical maximum varies from a low at Rock Valley of 0. 1%

to a high at Jornada Bajada, New Mexico, of 1.7%. Caldwell et al. (1977) found annual carbon gain in the Curlew Valley, Utah, area to be =235 g.m-2 yr-1, which on a dry mass basis converts to =670 g m-2.

yr-1. The theoretical maximum averages 1.57 x 104 g

m-2 yr-1; the gross photosynthesis relative to the theoretical upper limit is 4%, much lower than comparable measures of 71, 38, and 22% measured in forests.

The results show that photosynthetic efficiency is lower with increased water deficit. Although incident radiation and, consequently, potential efficiency are higher in deserts, the density of vegetation is low, and efficiencies are also low. When efficiency is expressed relative to peak FSC, there is generally a convergence of efficiencies for all systems. Such a method of measuring efficiency per unit of leaf biomass is more ex- pressive of species adaptation relative to abiotic parameters other than radiation. Before this adjustment, the differences in efficiency in the sampled areas were large, encompassing more than two orders of magnitude. The differences when efficiencies are expressed relative to leaf biomass are reduced to less than one order of magnitude. On this basis, deciduous systems are more efficient than mature coniferous systems, while desert systems tend to be least efficient. How- ever, the Jornada Bajada site has an average efficiency comparable to the Pilius taeda stand at Triangle Park, the highest reported.

Ejec t ot water avl ,ilabilitv oil ASC anda ANPPIASC

When water availability was defined as annual potential evapotranspiration (PET) minus precipitation (PPT), it was found to increase with a decrease in aboveground standing crop (Fig. 4A). PET, calculated weekly, has been shown accurately to predict actual evaporation from wet surfaces. Whittaker and Niering (1975) reported a similar biomass increase with a decrease in their moisture index. The data in Fig. 4A are divided between forests, where there is a water surplus such that PET - PPT < 0, and grasslands and deserts, where there is a water deficit. The forest systems reported have ASC values > 20 000 g/m2; An- drews Forest, for example, has an ASC of 83 855 g/m2. These systems can be considered "mature," or their state is such that yearly ecosystem production is low. Without this assumption, the analysis would be jeopardized because forest systems would be increasing in ASC even when the relationship established in Fig. 4A assumes that the upper limit in ASC has been achieved. Deserts and shortgrass steppes are also ap- proaching maturity, but with ASC values much less than forests; deserts have ASCs between 100 and 500 g/m2. Precipitation in desert systems seldom matches PET, even at weekly intervals. Annual water deficits reported here varied between <1000 mm at Curlew and 1978 mm at Rock Valley, but there are no data for the midrange of water availabilities where PET and PPT are nearly equal.




The model relating water availability to ASC is not linear when all the data are considered. The expression, from the data in Fig. 4A, is:

ASC = 2590e -000177(PET-PPT) (13)

This is the antilog of a linear regression equation fitted to a log transformation of the data (r2 = .82) and represents the best fit to the combined data on forests and desert-shortgrass steppes. Inspection of Fig. 4A shows a much greater variance about the regression line in the region of the forest data than in that of desert data.

Fig. 4B shows the relationship between water deficit and ASC for the desert-shortgrass steppe systems only. While the same exponential model is used as for the forest-desert data, there is a 25% reduction in the scal- ing parameter:

ASC = I900e -0.00177(PET-PPT) (14)

A linear model can also be fitted to these data but cannot be extrapolated to include the forest systems. For example, the linear model would underpredict ASC by a factor of 35 for PET - PPT = -2000. Thus, the exponential model is apparently the correct one and is very general, as its ASC and water availability values encompass more than three orders of magnitude.

Fig. 4A also shows that ANPP per unit of ASC increased exponentially with increased water deficit. The model encompasses nearly two orders of magnitude, with ANPP/ASC ranging from 0.01 to 0.57. As noted, the forest systems shown are mature, the ANPP/ASC varies from near 0.01 at the Andrews Forest to 0.06 for a hardwood stand at Coweeta. Desert systems are more efficient, from 0.14 at Curlew Valley, Utah, to 0.57 at Silverbell, Arizona. There is a greater variance in the data from the forest systems than in data from the desert systems. The model describing both data sets is:

AN PP = 0. 093 e 0.00084(PET-PPT) (15) ASC

This is the antilog of a linear model of transformed data (r2 = .71). When the water-stressed systems are examined (Fig. 4B), a linear relationship is more ap- propriate over the water deficit range of 900-1800 mm of water. Notice that one data point (Rock Valley) is not included in the relationship. Over this restricted water deficit range, the exponential model is nearly linear (Fig. 4B) and fits the data nearly as well as the linear model, but this linear model cannot be extrapolated to include the forest systems.

It is clear that the size of an ecosystem aboveground is closely related to the abiotic environment, in this case, to the water availability or annual sum of weekly potential evapotranspiration minus precipitation. Thus, the size of the system can be viewed as an adaptation to the abiotic environment, or to state it differently, the mix of species on a site is the set that maximizes the aboveground standing crop with respect to the en-

vironment. It would be inappropriate to suggest that community composition is an adaptation to water availability. It is just as likely that unrecorded environmental extremes such as low temperature or extreme drought operate to determine the species on each site. Since "mature" systems were used in the analysis, however, it does appear that the upper limit of the standing crop is strongly related to water availability.

The use of water deficit, or water surplus, to compare forest systems is somewhat hazardous because other factors cause large variances in ANPP. The fact that the size of the forest systems at maturity is more than one order of magnitude greater than predicted from linear extrapolation of water deficit data for desert systems indicates that other factors also promote the growth of standing material. If the standing material could be segregated into living and nonliving cells, the relationship between living material and water deficit might differ from that observed between ANPP and water deficit. Even in the sapwood of forest trees, only 7-10% of the cell material respires. The only function of much of the standing crop is to provide structural support to other living parts as they obtain organic constituents from the forest environment.

Species in dry environments produce more ANPP per unit of ASC as PET - PPT increases, perhaps because of the amount of nonmetabolizing tissue in such species. In desert environments, the increase is nearly linear. Since these systems can change in size from year to year as climate varies (Table 2), it may be that the data are showing only climatic responses. Data from Silverbell and Pawnee show production only for years that were dry in comparison to the long-term average PPT, and fit the linear response. Data from Curlew Valley, however, show greater production and standing crop in 1973 than in 1974, although moisture regime was similar during the two years. There is a suggestion, then, that the linear increase in ANPP/ ASC with increasing water deficit is not due to yearly variation in moisture. The alternative is that primary producers require more energy for internal processes as the environment becomes more harsh.

Desert-shortgrass steppe model

As noted previously, the desert-shortgrass steppe model was formulated as:

52

ANPP =A FSC 2 Tip t

sinh[B max(0, PPT - PET)Ati],

where the variables are as defined in Methods. Values for A and B were obtained with data from Curlew Valley, Silverbell, and Pawnee. Results of solving the equation for A and B with these data are plotted in Fig. 5. Note that there is a small region in the X, Y plane where the curves for A vs. B intersect. Outside this region, there is no convergence, and while the




1 w T - I-lf

6- PAWNEE (74) SILVE RELL

72 CURLEW

C RLEW (74)

4 4-

4 3-

2-

01l I I 0 001 0 01 0 1 1 0 10

PARAMETER B

FIG. 5. Relationships between input and loss parameters A and B (described in Methods) for three sites used in pa- rameterizing the desert model. Note region of intersection.

uniqueness of this region was not tested, intersection apparently occurs only here. Accordingly, values of A = 1.8 and B = 0.1 were chosen from this region.

There was very good agreement between predicted and observed ANPP at the three sites used to parameterize the model (Fig. 6A). Observed ANPP at Paw- nee, a shortgrass steppe, during 1974 was 59 g/m2, the exact value predicted by the model. Pawnee is a high- elevation site with relatively severe winters, warm summers, and precipitation occurring during thunder- storms (Fig. 2); its average temperature was 10.0? (Ta- ble 1). In contrast, Silverbell, where predicted ANPP was 15% higher than the observed, has an average temperature of 210; its summers are warm, and all precipitation occurs as rain.

The third site, Curlew Valley, is classed as a cold desert; annual average temperature for 1972-1974 was 7.7?. Precipitation occurs as snow in the winter and occasionally as rain in the summer. Seasonal PET at Curlew Valley is lower than at Silverbell and appears similar to that at Pawnee, although the annual PET at Curlew Valley was =30% lower. ANPP was predicted correctly for Curlew Valley for two successive years, 1973 and 1974 (Fig. 6A). During these 2 yr, ANPP declined from 149 to 70 gm -2 yr-'.

A test of the model was made with an independent data set from Jornada Bajada, New Mexico. This hot- desert site has an annual average temperature of 18.5? (Table 1), and although its latitude is similar to that of Silverbell, its annual PET is about one-third less. For 1974, the observed ANPP at Jornada Bajada was 101 gm m-2yr 1, whereas the predicted value was 127 g m-2 yr-1 (Fig. 6A). For 1973, the predicted value was 100 g m-2 yr- , while the actual ANPP was 129 g m-2 yr-'. For 1972. however, the predicted value of 1 10 gm -2. yr- was much lower than the actual ANPP (292 g M-2. yr- 1). Thus, in two of three cases the mod-

300

52 T, /10 ANPP = 18 FSC 2 sinh[0. I ma. (O,PPT-PET) a j

* FOREST

o GRASSLAND

200 - DESERT

J CURLEW (73) JORNADA (741

0 -(73) JORNADA 172)

SILVERBELL (72) *JORNADA (73) *

CURLEW (74)

IE EPAWNEE (74

al- CL1

< 0 100 200 3 0

0 @J B 'OTIS

() ANPP-[(77 10 RAD FSC)- (2 47 - ASC)] 2T/I

CL 3000 -

*TRIANGLE PARK

2000_

COWEETA I

NORMAN* LUE RIVER

ANDREWS 0 COWEETA 18 1000- FO REST

_ 0 OCASCADE HEAD OAK RIDGE BROOKHAVEN

*NOE WOODS

/NAKOMA

* HUBBARD BROOK

0f I I i 0 1000 2000 3000 4000

OBSERVED ANPP (g M Yr)

FIG. 6. Relationship between predicted and observed aboveground net primary production (ANPP) for water- stressed deserts and grasslands (A) and for nonstressed forests (B). Diagonal line represents a perfect fit between predicted and observed values.

el was within 25% of observed ANPP, but for the single high value the model badly underpredicted. More data are needed to better examine this apparent discrepancy, but the results of the validation are reason for optimism.

Although ANPP has been shown to be related to peak FSC (Fig. 3), the correlation between ANPP and FSC was not positive for the desert systems reported here. A much better basis for predicting ANPP was established by adding weekly temperature, precipitation, and potential evapotranspiration to the model, an addition hereafter called the "environmental parameter." Parameter A in the model scales ANPP to be 1.8 times FSC. In most cases this is an overprediction, and the environmental parameter, which quantifies the current year's environment, often adjusts ANPP to a lower value. The environmental parameters in Table 6 differ by a factor of three and greatly increase pre-




dictive power. Since these values are calculated from weekly data, they emphasize the importance of weekly events in regulating production in these systems.

The environment may change production in several ways. Two ways evident here are through a change in FSC and a change in photosynthetic efficiency per unit of FSC. For Curlew Valley, FSC was reduced from 124 g/m2 in 1973 to 74 g/m2 in 1974, while precipitation varied from 315 mm in 1972 to 261 mm in 1973 and 204 mm in 1974. The environmental parameter also decreased from 1973 to 1974, with a 50% reduction in ANPP. FSC at Jornada Bajada had much less fluctuation during 1972-1974 (43, 48, and 51 g/m2) largely because there was little fluctuation in precipitation (31 1, 237, and 394 mm/yr). The environmental parameters for Jornada Bajada are all greater than unity and are double those for the other sites. Thus, production is apparently more dependent on environmental modifi- cation of ANPP than on annual changes in FSC. In- corporation of FSC into the model allows for any car- ry-over of productive potential from a preceding year when ANPP is increased because of better growth conditions. In an earlier publication, Webb et al. (1978) demonstrated that such an effect was very likely in shortgrass steppes and cold deserts.

The model is formulated without explicit reference to species and provides good predictive power, although plant communities on each site are quite distinct. It should be noted, however, that the inclusion of FSC in the model does not mean that all productivity is directly or indirectly controlled by the model formulation of the abiotic environment. Abiotic factors not included in the model may be related to species adaptation to a site, and the occurrence of a plant community with a given FSC may have an adaptive significance not related to productivity. The model would not show this significance directly.

Forest model

A standard multiple linear regression technique was used to fit the data from seven coniferous, six deciduous, and one mixed conifer-deciduous stand to the forest model, the equation for which was as follows:

ANPP = [A' RAD FSC - B' ASC]2 7'/,

where all variables are as previously defined. Plotting of predicted and observed ANPP values is shown in Fig. 6B. The coefficient of determination was 0.88. The two parameter values for the input and loss terms in the model were 7.7 x 10-7 and 2.47 x 10-3, respectively. The average error relative to the observed ANPP was 17%. Another model, one with an added constant term, gave better predictive power, but the constant term was essentially an average ANPP for all the sites and had limited biological meaning.

The premise on which the model is based is that ANPP of forest systems not significantly water stressed

TABLE 6. ANPP is predicted by the desert-shortgrass steppe model and its relationship to FSC and the environmental paramenter.

Environ- mental Predicted

FSC param- ANPPt Site and year (g/m2) eter* (g-m-2 yr-1)

Curlew Valley 1973 124 0.67 149 Curlew Valley 1974 74 0.53 70 Silverbell 1972 50 0.96 86 Pawnee 1974 59 0.56 59 Jornada Bajada 1972 43 1.42 110t Jornada Bajada 1973 48 1.16 100 Jornada Bajada 1974 51 1.38 127

* Combined effects of weekly temperature, precipitation, and potential evapotranspiration as parameterized in the model.

t According to the model, ANPP is calculated by multi- plying 1.8 x FSC x the environmental paramenter.

t Model underpredicts ANPP by -50%.

is a function of two biotic variables, FSC and ASC, and two abiotic variables, total yearly shortwave radiation and average annual temperature. The forest model, like the desert model, does not explicitly iden- tify species as a variable, but there is nothing in the model which rules out species adaptation to the site. However, the model formulation strongly suggests that, once an ecosystem is developing, yearly production is largely controlled by the abiotic environment. This control results because radiation and average temperature do not vary greatly for a given site, foliar biomass is nearly constant after canopy closure, and ASC increases only a small percentage each year.

Although the model predicts ANPP for both coniferous and deciduous forests (Fig. 6B), the average relative error was 12% for the coniferous systems and 31% for the deciduous ones. In general, the model underpredicts for deciduous systems that have a low FSC. For example, Hubbard Brook with a peak FSC of 282 g.m-2 yr-1 and Nakoma with one of 259 g.m-2

yr-1 both have predicted ANPP values significantly less than the actual. No data from coniferous systems with low FSC values are available for comparative purposes. It is likely that a low FSC invalidates the assumption of linearity between gross photosynthesis and radiation, leading to an underprediction of ANPP. A correction term applied to FSC required an evaluation of two additional parameters and significantly increased the predicted ANPP for systems with a low FSC. However, the increase in the coefficient of determination was only from 0.88 to 0.90.

Validity of the model structure and the parameter values was estimated by utilizing data from the literature, chosen when possible to represent extremes or conditions outside the original data. Crow (1978) published data from three second-growth hardwood stands in Wisconsin. The average temperature was 5.50, and




TABLE 7. Parameters A and B for the forest model as derived from direct computation based on data from sites discussed in Kinerson et al. (1977) and Kira et al. (1964).

Model parameters Relativet

Dominant species Site* A B error

Pinits tae(ld Triangle Park, North Carolina 2.35 x 10-6 10.9 x 10-2 0.11 LiriodendIron tidipiferra Oak Ridge, Tennessee 3.78 x 10-6 6.55 x 10-2 -0.43 Quercus-Pinus Brookhaven, New York 2.06 x 10-6 4.96 x 10-2 -0.01 Tropical Thailand 1.62 x 10-6 4.16 x 10-2 -0.36

Average 2.45 x 10-6 6.64 x 10-2

* See Table I for state code. t Error between observed ANPP and that predicted by the forest model with A = 7.7 x 10-7 and B = 2.47 x 10-3.

leaf biomass averaged 297 g/m2. The observed ANPP for these three stands was 831, 745, and 755 g.m-2 yr-1, while the predicted aboveground production was 312, 373, and 501 gm-2 yr-1. The underprediction by the model occurs here as it did with the original data. In this case however the stands were young, and ASC averaged 104 g/m2, considerably less than in the original data. In a simulation test of the model, the annual temperature was doubled, resulting in predicted ANPP values of 553, 661, 888 g m-2 yr-'. The relative error is 33, 11, and - 18%. This consistent underprediction of stands with low foliar biomass and low temperatures shows that these stands produce more ANPP per unit of foliar biomass than do the average. While the trend is consistent, it is difficult to assign a cause- effect relationship. Results from Fig. 3 showing the relationship between ANPP and FSC confirm that deciduous forests in general are more productive per unit of foliage than are conifers.

Kira et al. (1964) obtained data from a tropical rain forest on the Malay Peninsula, latitude 7035'N, with production of 2850 g m-2 yr-', standing crop of 29 400 g/m2, and foliar biomass of 789 g/m2. Predicted ANPP was 3876 gm2yr'. They judged ecosystem production to be nil and the system to be in equilibrium with its environment. The model predicts 35% more production than was measured. While the biotic parameters are not extreme in terms of the stands used for the model, the average temperature of 22? is greatly above that in the original data set and contributes markedly to the predicted production. It is noteworthy that this tropical stand is not producing a net increment in biomass but is producing as much annually as the vigorous young conifer stands near Otis, Oregon, and Triangle Park, North Carolina. If none of the energy in the tropical system increases the biomass increment, considerably more energy is used for internal processes than was the case in the stands reported in Table 2.

For an A/nuts rubra stand, Zavitkovski and Stevens (1972) reported an average ANPP of 1708 g.m-2 yr-' over 20 yr. If we use an FSC value of 600 g/m2, a maximum value for annual litterfall, the model can be tested. Zavitkovski and Stevens also reported a max-

imum ASC 24 900 g/m2. Computation of ANPP from the model is 1018 g/m2, well below the 1708 g.m-2. yr-1 they reported. Zavitkovski and Stevens also reported other stands in the same locality in Oregon to have ANPP values ranging from 2600 to 3000 g m-2.

yr-1. The model clearly underpredicts for deciduous stands that are highly productive. Art and Marks (1972) report an ANPP of 3420 g.m-2-yr-1 and an ASC of 30 531 g/m2 for an 18-yr-old stand of Pinus radiata, the fastest-growing conifer in New Zealand. The forest model predicts an ANPP for this stand of 3162 g.m-2

yr-1 or =8% below that measured. The two parameter values for input and loss in the

forest model are published here for the first time; consequently, there are no comparable values in the literature. Accordingly, A and B parameter values were computed directly from data in Kinerson et al. (1977) on a Pinus taeda stand at Triangle Park, North Car- olina, a Liriodendron tulipifera stand at Oak Ridge, Tennessee, and an oak-pine stand at Brookhaven, New York, as well as from data in Kira et al. (1964) on a tropical rain forest in Thailand (Table 7). Since estimates of both the input and loss terms for these sites are available, direct computation was possible.

Values from the literature for the input term are =4 times the input parameter A of the model, while the value of B, the loss parameter, from the literature is z35 times that predicted by the model. The differ-

ences are large, perhaps because of the empirical na- ture of the parameters, but also because belowground production is included in the values from the literature and the input parameter must necessarily be greater. Furthermore, the ASC of these four stands is less than the average for the stands included in the model. The error term expressing actual ANPP relative to that predicted from the model is included in Table 7. Al- though parameter values vary between the model and direct calculation, the prediction is reasonable in two of four cases.

Considering the diversity in plant form, environment, phenology, and techniques for measuring CO2 exchange, the A parameters in Table 7 appear to be consistent; the average is 2.45 x 10-6 with a standard deviation of 0.93 x 10-6. None of the systems is sig-




nificantly water stressed, and evaluation of the parameter seems closely related to annual environmental variables coupled with peak foliar biomass. All stands represented in Table 7 have closed crowns, and FSC is very likely near its maximum and stable amount. The fact that the parameter values computed directly on the basis of the model structure are stable illus- trates the stability of the model input term. A large variance would have signaled instability. The differences between the computed parameter values and those used in the model result from differences in total net primary production. The stability of the indepen- dently calculated values is encouraging because the validation data represent coniferous, deciduous, mixed conifer-dedicuous, and tropical forest systems.

The B parameter in Table 7 is also stable. The coefficient of variation is 45%, slightly greater than for parameter A. The model value is -4% of the average value computed from the independent data. The model coefficient of 2.47 x 10-3 represents an average loss per unit of ASC for the stands used to parameterize the model. Since many stands used to derive the model had much greater biomass than those in Table 7, the model parameter value underestimates the computed value. This may underscore the unreliability of computing an average loss term per unit of biomass rather than per unit of the living system.

CONCLUSIONS

The analyses support our contention that differences in ANPP are closely related to differences in abiotic variation but that community composition also plays a role in determining ANPP. They demonstrate that ASC is related to water deficit and further that ANPP per unit of ASC increases as water availability decreases. While the analyses were not conclusive, they suggest that more severely water-stressed systems require more ANPP per unit of ASC for internal ecosystem processes. Production per unit of FSC was characterized by a single relationship for all data, but it could not be shown conclusively that ecosystem types were related to ANPP/FSC. It was shown, however, that deciduous forests produce more per unit of FSC than do coniferous forests when the comparison is made for mature ecosystems.

When production is linked to solar radiation, it is clear that forest sites are more productive than water- limited sites, which have less dense vegetation for light interception. When photosynthetic efficiency is nor- malized per unit of FSC, the differences fade, and a desert system turns out to be as as efficient in con- verting radiant energy to biomass as does a highly productive young coniferous forest. Without the normal- ization of FSC, it can be easily shown that ecosystem type is correlated with efficiency of radiant energy capture.

The success of the desert and forest models shows

the strong relationships that can be derived for predicting ANPP when biotic and abiotic variables are combined. While the models were reasonably suc- cessful, there was enough discrepancy in the validation to give a hint at differences resulting from ecosystem types. The desert model, for example, could not account for the very high productivity of the Jor- nada Bajada site, and the forest model better predicted ANPP of coniferous forests than it did deciduous ones. On balance, it appears that variation in predicted ANPP of native ecosystems can be reduced when ecosystem type is included as an independent variable.

ACKNOWLE DGMENTS

The authors thank the many people who collected and syn- thesized the data reported in this paper, most of whom were associated directly with the United States/International Bi- ological Program. This study was funded by National Science Foundation Grant DEB 74-20955 AOl to Oregon State Uni- versity.

LITERATURE CITED

Art, H. W., and P. L. Marks. 1971. A summary table of biomass and net annual primary production in forest ecosystems of the world. Pages 3-240 ill H. Young, editor. Forest biomass studies. International Union of Forest Re- search Organizations Working Group on Growth and Yield, Gainesville, Florida, USA.

Balph, D. F. 1973. Curlew Valley validation site report. Research Memorandum 73-1, United States/International Biological Program, Desert Biome, Logan, Utah, USA.

Balph, D. F., R. S. Shinn, R. D. Anderson, and C. Gist. 1974. Curlew Valley validation site report. Research Memorandum 74-1, United States/International Biological Program, Desert Biome, Logan, Utah, USA.

Botkin, D. B., and C. R. Malone. 1968. Efficiency of net primary production based on light intercepted during the growing season. Ecology 49:438-444.

Bray, J. R. 1961. An estimate of a minimum quantum yield of photosynthesis based on ecological data. Plant Physi- ology 36:371-373.

Caldwell, M., R. S. White, R. T. Moore, and L. B. Camp. 1977. Carbon balance, productivity, and water use of cold winter desert shrub communities dominated by C3 and C4 species. Oecologia (Berlin) 29:275-300.

Crow, T. R. 1978. Biomass and production in three contig- uous forests in northern Wisconsin. Ecology 59:265-273.

Fujimori, T. 1971. Primary production of a young Tsiuga heteroph'I1a stand and some speculations about biomass of forest communities on the Oregon Coast. Research Pa- per PNW-123, United States Forest Service, Pacific North- west Forest and Range Experiment Station, Portland, Or- egon, USA.

Fujimori, T., S. Kawanabe, H. Saitu, C. Grier, and T. Shidei. 1976. Biomass and primary production in forests of three major vegetational zones of the Northwestern United States. Journal of the Japanese Forestry Society 59:360-373.

Gay, L. W. 1971. The regression of net radiation upon solar radiation. Archiv fuer Meteorologie Geophysik und Bio- klimatologie Serie B 13:1-14.

Gholz, H. L. 1979. Limits on aboveground net primary production, leaf area, and biomass in vegetational zones of the Pacific Northwest. Dissertation. Oregon State Uni- versity, Corvallis, Oregon, USA.

Grier, C. C., and R. S. Logan. 1977. Old growth Pseudlo- tsuga menziesii of a western Oregon watershed: biomass distribution and production budgets. Ecological Mono- graphs 47:373-400.




Johnson, F. L., and P. G. Risser. 1974. Biomass, annual net primary production, and dynamics of six mineral ele- ments in a post oak-blackjack oak forest. Ecology 55:1246- 1258.

Jordan, C. F. 1971. Productivity of a tropical forest and its relation to a world pattern of energy storage. Journal of Ecology 59:127-142.

Kinerson, R., C. W. Ralston, and C. G. Wells. 1977. Car- bon cycling in loblolly pine plantations. Oecologia (Berlin) 29:1-10.

Kira, T., 0. Husato, Y. Kyoji, and 0. Kazuhike. 1964. Pri- mary production by a tropical rain forest of Southern Thai- land. Botanical Magazine 77:428-429.

Larcher, W. 1975. Physiological plant ecology. Springer- Verlag, Berlin, West Germany.

Lawson, G. J., G. Cottam, and 0. L. Loucks. 1972. Struc- ture and annual productivity of two watersheds in the Lake Wingra Basin. Research Memorandum 72-98, United States/ International Biological Program, Eastern Deciduous For- est Biome, Madison, Wisconsin, USA.

Leith, H., and R. Whittaker, editors. 1975. Primary productivity of the biosphere. Springer-Verlag, New York, New York, USA.

Nunn, J. R., C. F. Becker, G. Twitchell, R. D. Burman, and J. D. Arijo. 1972. Meteorological instrumentation data and data analysis for 1971. Technical Report 180, United States/International Biological Program, Fort Collins, Col- orado, USA.

Penman, H. L. 1956. Estimating evaporation. Transactions of the American Geophysical Union 37:43-50.

Rosensweig, M. L. 1968. Net primary productivity of ter- restrial communities: Prediction from climatological data. American Naturalist 102:67-74.

Sims, P. L., and J. S. Singh. 1978. The structure and function of ten Western North American grasslands. III. Net primary production, turnover and efficiencies of energy capture in water use. Journal of Ecology 66:573-597.

Sims, P. L., J. S. Singh, and W. K. Lauenroth. 1978. Com- parative structure and function of ten western North Amer- ican grasslands. I. Abiotic and vegetational characteristics. Journal of Ecology 66:251-285.

Thames, J. L. 1973. Tucson basin validation site report. Research Memorandum 73-3, United States/International Biological Program, Desert Biome, Tucson, Arizona. USA.

1974. Tucson basin validation site report. Research Memorandum 74-3, United States/International Biological Program, Desert Biome, Tucson, Arizona, USA.

Turner, F. B. 1973. Rock Valley validation site report. Re- search Memorandum 73-2, United States/International Bi- ological Program, Desert Biome, Tucson, Arizona, USA.

Turner, J., and J. N. Long. 1975. Accumulation of organic matter in a series of Douglas-fir stands. Canadian Journal of Forest Research 5:681-690.

Wassink, E. C. 1959. Efficiency of light energy conversion in plant growth. Plant Physiology 34:356-361.

Webb, W. L., S. Szarek, W. Lauenroth, R. Kinerson, and M. Smith. 1978. Primary productivity and water use in native forest, grassland, and desert ecosystems. Ecology 59:1239-1247.

Westlake, D. F. 1963. Comparisons of plant productivity. Biological Review 38:385-425.

Whitford, W. G. 1975. Jornada validation site report. Re- search Memorandum 75-4, United States/International Bi- ological Program, Desert Biome, Tempe, Arizona, USA.

Whittaker, R. H., F. H. Bormann, G. E. Likens, and T. G. Siccama. 1974. The Hubbard Brook ecosystem study; forest biomass and production. Ecological Monographs 44: 233-254.

Whittaker, R. H., and W. A. Niering. 1975. Vegetation of the Santa Catalina Mountains, Arizona. V. Biomass, production and diversity along the elevational gradient. Ecol- ogy 56:771-790.

Whittaker, R. H., and G. M. Woodwell. 1969. Structure, production, and diversity of the oak-pine forest at Brook- haven, N.Y. Journal of Ecology 57:155-174.

Yocum, C. S., L. Allen, and E. R. Lemon. 1964. Photo- synthesis under field conditions. VI. Solar radiation balance and photosynthetic efficiency. Agronomy Journal 56: 289-53.

Zavitkovski, J., J. G. Isbrands, and T. R. Crow. 1974. Ap- plication of growth analysis in forest biomass studies. Pages 196-226 ill C. P. P Reid and G. H. Fechner, editors. Pro- ceedings of the third North American Forest Biology Workshop, College of Forestry and Natural Resources, Colorado State University, Ft. Collins. Colorado, USA.

Zavitkovski, J., and L. D. Stevens. 1972. Primary productivity of red alder ecosystems. Ecology 53:235-242.



Primary Production and Abiotic Controls in Forests, Grasslands, and ...

Documents

Transcript of Primary Production and Abiotic Controls in Forests, Grasslands, and ...