| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Soil Science, Univ. of Saskatchewan, Saskatoon, SK, S7N 0W0 Canada
b Dep. of Agronomy and Range Science, Univ. of California-Davis, Davis, CA 95616
* Corresponding author (cvankessel{at}ucdavis.edu)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: BNF, biological N fixation • 15NE, 15N-enriched • 15NA, 15N natural abundance • %Ndfa, percentage of N derived from the atmosphere
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Currently, the most common approach to estimate BNF in the field has been through the use of 15N isotope dilution techniques. The additional dilution of 15N in the legume as compared with the non-N2-fixing reference plant is used to calculate BNF. Two main approaches are used the 15NE approach and the 15NA approach. Using the 15NE approach enriched 15N-fertilizer is applied to both legume and reference crop and the differences in 15N dilution between the two crops are used to calculate N2 fixation. In contrast, the 15NA approach uses the naturally occurring differences in atom% 15N of soil available N and atmospheric N2 to estimate BNF. Again, differences in dilution of 15N between the legume and the reference crop are used to calculate BNF. There are a number of difficulties associated with both methods, which have been discussed in detail elsewhere (Witty, 1983; Shearer and Kohl, 1986; Danso et al., 1992). The 15NE and 15NA approaches for estimating BNF by field-grown legumes have been compared extensively with some researchers reporting that both methods provided similar estimates of N2 fixation (Shearer and Kohl, 1986; Bremer and van Kessel, 1990; Peoples and Herridge, 1990; Doughton et al., 1995).
Recently, the validity of the 15NA approach to estimate BNF under field conditions has been questioned (Handley and Scrimgeour, 1997). Handley and Scrimgeour (1997) stated that at natural abundance levels, such as would be encountered using the 15NA approach, 15/14N undergoes large fractionations relative to sample values due to a variety of biotic and abiotic processes, and thus can not be used as a tracer of N from source to sink. It follows that one can not infer that differences in isotopic signatures between legumes and reference crops are caused primarily by BNF. In contrast, using the 15NE approach, the 15N signature is significantly enriched above natural abundance levels, and the relative impact of isotope fractionations on the 15N signature of the plant is insignificant. Therefore, Handley and Scrimgeour (1997) have argued that the 15NA and 15NE approaches essentially reflect different processes. The authors cited two studies as experimental evidence (i.e., Androsoff et al., 1995; Stevenson et al., 1995).
In these two studies, pea (Pisum sativum L.) and a reference crop were grown across two landscapes and BNF was estimated by both the 15NA and 15NE approaches for nodes of a 10 m grid. Distances between the 15NE and 15NA microplots and distances between pea and the reference crop were 1.5 m. The size of the microplots was 1 m2. All samples were taken from the center of the microplots, and thus the distance between the plants sampled for the 15NA and 15NE approaches was 2.5 m. Although mean values for BNF across the whole field were similar in both studies, the results showed no significant correlation between the two approaches at the individual grid nodes. According to Handley and Scrimgeour (1997), failure to detect significant correlations between individual estimates of N2 fixation, and the consequent suggestion that the 15NA and 15NE methods measured different aspects of soil/plant N relations, "overturned much of published literature on comparing methods for estimating N2–fixation" (Handley and Scrimgeour, 1997).
Others strongly disagreed with this interpretation of the experimental data of the above-mentioned studies (Boddey et al., 2000). Boddey et al. (2000) concluded that the observed lack of correlation between the two approaches was probably caused by high spatial variability of the controlling factors for BNF at a 2.5 m distance. Androsoff et al. (1995) and Stevenson et al. (1995) also discussed the possibility that the observed lack of a significant correlation may have been caused by short-range variability. The possibility that the two approaches are measuring fundamentally different processes was raised, but considered unlikely.
To resolve this controversy, the spatial variability of BNF processes must be known. If BNF is spatially correlated at the 2.5-m scale, it can be concluded that the reported lack of significant correlation between 15NA and the 15NE estimates was due to methodological differences; the two approaches measure different processes. Alternatively, if the spatial correlation of BNF at the 2.5-m scale is insignificant, it can be concluded that the spatial variability in biotic and abiotic processes caused, in part, the differences between the two approaches. However, if this is the case, then we are still unable to determine if the two different approaches measure the same aspects of soil/plant N relations because, in practicality, they cannot be compared in situ due to the high degree of BNF spatial variability.
The main objective of this study was to determine the spatial correlation of BNF by field-grown chickpea at short distances. Chickpea and a reference crop were grown along a transect and sampled every 0.30 m. Estimates of BNF were derived using the 15NA approach. Geostatistical analysis were carried out to determine the spatial correlation of BNF estimates. Finally, a simulated study was carried out in order to re-evaluate the results of Androsoff et al. (1995) and Stevenson et al. (1995).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
A microscale study that consisted of a 33-m long transect was established to examine the spatial correlation of BNF. The transect was located on a gently sloping upper slope position with a maximum difference in elevation within the transect of 2 m.
Site Preparation and Soil/Plant Analysis
On 26 May 1996, kabuli chickpea (C. arietinum ‘Sanford’), inoculated with a peat-based, self-stick inoculant containing Rhizobium cicer (MicroBio Rhizogen Corp., Saskatoon, SK) was sown at a rate of 190 kg ha-1. Wheat (T. aestivum ‘Katepwa’) was hand seeded at a rate of 90 kg ha-1 1 wk after sowing chickpea for use as a reference crop for estimating BNF using the 15NA approach. The wheat was grown in a row at a distance of 0.15 m parallel to the chickpea row. The average distance between chickpea plants was 0.30 m. Therefore, a total of 110 samples were taken for each crop. A distance of 0.30 m between chickpea plants was considered the minimum distance, as it was assumed that plants take the majority of the nutrients from within a distance of 0.30 m. Plants grown at a distance less than 0.30 m would therefore explore a similar soil volume. At harvest, the aboveground portion of individual chickpea and wheat plants were collected, dried, separated into residue and grain. The grain was ground, ball-milled and analyzed for isotopic composition, using methods described in Stevenson and van Kessel (1997) on a 20-20 Mass Spectrometer interfaced with an ANCA-GSL sample converter (Europa Scientific, Crewe, UK). Analytical errors associated with this spectrometer measured using repeated measurements of a single standard were reported previously
(SE = 0.047
, 
2 = 0.2052, CV = 5.58%)
(Sutherland et al., 1991).
The proportion of N derived from the atmosphere via BNF (%Ndfa) in chickpea grain was calculated as reported by Shearer and Kohl (1986):
 | [1] |
15N is:
The value for c represents the 15N value of chickpea grown in an N-free medium. The c-value was determined by growing chickpea plants in Leonard jars in a growth chamber under the following conditions: 50% relative humidity, 20 °C during the day and 18 °C during the night. The photoperiod was a 16-h day and an 8-h night. The c-value for the grain of these plants was -1.71 ± 0.25. Atom% 15N of atmosphere is 0.3663%, which is equal to a 15N value of 0 (Mariotti, 1983).
Soil samples were taken at seeding time, and soil mineral N was measured using a Technicon AutoAnalyzer II System (Labtronics Inc., Tarrytown, NY). Gravimetric soil moisture content was determined using standard techniques (105 °C, 24 h).
Geostatistical Analysis
Spatial variability of BNF was assessed using geostatistical techniques (Deutsch and Journel, 1998). The scale of spatial correlation was expressed using variograms, and was calculated and modeled with the Variowin package (Pannatier, 1996). Parameters of the modeled variogram such as nugget (i.e., spatial variance at distances close to 0), sill (i.e., spatial variance at distances beyond spatial correlation), range (i.e., the range of spatial correlation) and relative nugget effect (i.e., the relative size of the nugget as compared with the sill) were used for characterization of spatial correlation. No interpolation of the results (e.g., using ordinary kriging) was necessary, because the measurements were taken at very short, regular intervals.
In order to reevaluate whether the results reported by Androsoff et al. (1995) and Stevenson et al. (1995) were likely due to high spatial variability of BNF or to a difference between the 15NA and 15NE approaches, a simulation study was conducted. Using the modeled variograms of %Ndfa of legume grain and residue, 10 simulated fields of 100 x 100 m (cells of 1 x 1 m) were generated for both variables using the Sequential Gaussian Simulation algorithm (Deutsch and Journel, 1998). The simulated fields reproduce the modeled variograms, and are therefore similar in spatial variability to the observed BNF data in the field. Similar to Androsoff et al. (1995), 60 sampling locations were randomly selected in the field. At each location, simulated cells that were spaced 1 m and 2 m apart were identified, and a correlation analysis was performed. We assumed that a poor correlation between the simulated values would suggest that spatial variability of BNF is too high to make any serious comparison between the 15NA and 15NE approaches. A strong spatial correlation of 15NA simulations at short distances would indicate that the poor correlation reported between 15NA and 15NE numbers approaches is probably because they reflect two different processes.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
15N values ranged from 1.5 to 4.4 with a mean value of 2.9. The mean 15N value was 8.4 for wheat. The values for %Ndfa for legume ranged from a minimum of 36% to a maximum of 70%, with a mean value of 55%. The values for the mean and median were similar.
|
|
15N values for wheat and chickpea were between 4 and 6 m. The relative nugget effect for the 15N values for wheat was 30.6%, whereas the nugget effect for the 15N values for chickpea was around 61%. The %Ndfa variogram showed a large relative nugget effects of 71% (Fig. 2)
. The large nugget effect, combined with the short range, indicates that most of the spatial correlation is confined to distances shorter than the sampling interval, that is, 0.30 m. The variogram of mineral N shows similar features, with a nugget effect of 68%, and a range of 3.1 m.
|
|
The modeled variogram for the %Ndfa values generated for the simulated fields (Fig. 3) was similar to that which was generated from the field sampled materials (Table 2), indicating that the simulated fields are realistic representations of spatial variability of BNF. Resampling the 10 simulations and calculating experimental variograms would therefore result in variograms identical to Fig. 2c. Scatterplots of simulated values of %Ndfa sampled at distances of 1 m (Fig. 4a) and 2 m (Fig. 4b) were developed. The r2 for these %Ndfa values were 0.12 and 0.02, respectively, indicating little or no correlation at both distances. Results for the other simulated fields (not depicted) were similar, with no r2 values larger than 0.14.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The more narrow range in %Ndfa values found here for chickpea may therefore be attributed to the limited extent of the sample area in this study (i.e., the transect encompassed only a portion of a knoll/upper level position). In the previous studies, the highest and lowest values for BNF were observed for pea grown on the knolls/upper level and footslopes/lower level landform elements, respectively. Medium values for %Ndfa were observed in areas classified as shoulders. Species differences (pea versus chickpea) may also play a role. However, other studies reported that chickpea is able to derive a high percentage of its N from BNF (Doughton et al., 1995; Unkovich and Pate, 2000).
Spatial Variability
The nugget effect for 15N of wheat was less than half of the nugget effect observed for 15N of chickpea (Table 2). This lower nugget effect for 15N in wheat reflects either lower short-range (i.e., shorter than 0.30 m) spatial variability or lower measurement errors. Since measurement errors associated with 15N in wheat and chickpeas should be similar, we believe that the relatively high nugget effect observed for 15N of chickpea is due to larger short-range variability for 15N in legumes. Moreover, because BNF is an extra source of N for legumes and this extra N source is associated with a higher spatial variability of 15N, the variability likely resulted from the complex environmental factors that control BNF and the different 15N values for soil N and atmospheric N2.
The isotopic signature of 15N natural abundance in a plant is dependent on numerous processes which can occur simultaneously and whose cumulative effect can lead to an increase, no change, or a decrease in the 15N natural abundance of the plant. Although Sutherland et al. (1991) reported no spatial correlation in grain 15N of wheat grown under dryland conditions in 3 out of 4 transects, under irrigated conditions there was clear spatial correlation (Sutherland et al., 1993). It was postulated that the rate of denitrification varied significantly across the irrigated field, thereby considerably changing the isotopic signature of the available soil N pool and hence the isotopic signature in the plant.
In this study, the much larger nugget effect of 15N in the legume compared with 15N in the reference crop indicates that the observed variability in %Ndfa values was more related to processes changing legume 15N values than those affecting wheat 15N values. This suggests that N2 fixation is controlled by highly variable, localized environmental factors. There are a large number of such biotic and abiotic factors (Graham and Vance, 2000). The lack of significant spatial correlation of %Ndfa values was in sharp contrast to the observed spatial correlation for soil variables total soil C and N, and soil moisture (Table 2). It is not likely that any of these variables had a defining impact on levels of BNF. Biological N fixation estimates from shoots showed a similar high variability as the presented grain data (not depicted). Therefore, it is not likely that any internal discriminating process in the crop would lead to higher spatial variability in the grain than in the whole plant.
The lack of a significant correlation between estimates based on the 15NA and 15NE approach in previously conducted landscape studies (Androsoff et al., 1995; Stevenson et al., 1995) has led to disagreements in the interpretations of the results. Handley and Scrimgeour (1997) concluded that the lack of a significant correlation between the two approaches is a strong indication that the two approaches are controlled by different sets of plant/soil processes and, therefore, BNF estimates derived using the two different methods can not be compared. Boddey et al. (2000) argued that the lack of a significant correlation between methods is merely caused by high spatial variability in controlling environmental variables, and that the two approaches are essentially measuring the same process.
Our results show a high short-range variability in BNF, as measured by the 15NA approach. As a consequence of the high short range spatial variability of BNF, correlations between different methods of estimating BNF are unlikely to be detected under field conditions due to physical constraints associated with field experimentation. Indeed, under simulated conditions deemed realistic, we observed little or no spatial correlation between 15NA values at separation distances of only 1 to 2 m (Fig. 3 and 4). We, therefore, agree with part of the interpretation made by Boddey et al. (2000); experiments similar to those of Androsoff et al. (1995) and Stevenson et al. (1995) would have resulted in no significant correlations between the 15NA and the 15NE approach, solely due to high spatial variability of BNF.
It is noteworthy that most experiments dealing with BNF estimation used a classical experimental design, and were therefore ill-designed to detect this high spatial variability. The foundations of classical design (randomization and replication) are specifically meant to neutralize any spatial variability from the observed processes. This can be illustrated by the results from Androsoff et al. (1995) and Stevenson et al. (1995), where the averages of the 15NE and 15NA estimates were practically similar, and the obvious discrepancies between the methods became apparent only when spatial variability was taken into account.
Our results, however, cannot lead to the rejection of the interpretation made by Handley and Scrimgeour (1997). For example, the observed variability in the 15N values in the chickpea also might have been caused by species specific differences in isotopic discrimination during N accumulation, reallocation within the plant or N volatilization (Handley and Scrimgeour, 1997). Alternatively, the 15N signatures of plant-available NO-3 and NH+4 in the soil are known to be variable (see Handley and Scrimgeour, 1997 and references therein) and the reference crop and the legume may accumulate NO-3 and NH+4 in different ratios, therefore, the observed difference in the isotopic signature between the two species may not have been due soley to variations in BNF.
The lack of a strong correlation between the 15NA and the 15NE approach in the earlier landscape studies and the observed high spatial variability of %Ndfa in this study may also be of concern when classical research experiments are conducted (i.e., smaller, randomized complete block designs). If the isotopic signature in the plant is characterized by a very high short-range variation, a reference plant may be too far away to reflect the isotopic signature of the available soil N that the legume has access to, even when relatively small plot experiments are conducted. In many small plot experiments, the distance between the legume and the reference crop is larger than the distance of 2.5 m used in the two landscape studies. Reducing this distance to 0.30 m, which can be considered the minimum distance between plants in the field, may not solve the problem, as most of our observed variation in %Ndfa showed no spatial structure, even at that scale. As a consequence, we can conclude that spatial variability in BNF is high, and thus estimates of BNF are expected to vary on a point by point basis irrespective of the method used to measure N2 fixation (i.e., 15NA vs. 15NE). Moreover, the variability in BNF is large enough to explain the lack of correlation between different methods of assessing BNF as reported in previous studies (i.e., Androsoff et al., 1995; Stevenson et al., 1995). However, because the spatial variability of BNF is high, we can not conclude that lack of correlation between methods used to estimate BNF observed in previous studies is necessarily an indication that the 15NA and the 15NE are measuring different processes and thus can not be compared. Similarly, having observed the high levels of variability in BNF, we can not conclude that the 15NA approach should not be used for estimating BNF in situ, as was suggested by Handley and Scrimgeour (1997). Although this conclusion may indeed be correct, controlled experiments are needed to verify their assertion.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
15N spatial correlation was very weak. Although samples were collected at 0.3-m spacing, there was a high nugget effect and spatial correlation was confined to ranges of 3 to 5 m. Similarly, %Ndfa showed a weak spatial correlation. No significant correlation was found at distances of 1 to 2 m. If such high spatial variability of BNF is common, as it seems to be from other studies, a direct comparison between estimates for BNF based on the 15NE and 15NA approaches under field conditions can not be made with sufficient confidence. Previous observations that the 15NA and the 15NE approach are not correlated in field studies may have been due to the high degree of spatial variability of BNF or may have been due to fundamental differences in the processes that the two methods estimate; our current understanding of these two approaches fails to distinguish between these two possibilities. |
ACKNOWLEDGMENTS |
|---|
Received for publication February 13, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. L. Walley, G. W. Clayton, P. R. Miller, P. M. Carr, and G. P. Lafond Nitrogen Economy of Pulse Crop Production in the Northern Great Plains Agron. J., November 6, 2007; 99(6): 1710 - 1718. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. P. Russelle and A. S. Birr Large-Scale Assessment of Symbiotic Dinitrogen Fixation by Crops: Soybean and Alfalfa in the Mississippi River Basin Agron. J., November 1, 2004; 96(6): 1754 - 1760. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
