Soil CO 2 Efflux Measurements by Alkali Absorption and Infrared Gas Analyzer in the Brazilian Semiarid Region

The CO2 emission from the soil surface, commonly referred to as soil CO2 efflux (ECO2) or soil respiration, is the sum of processes that include root respiration and microbial activity. Measuring this evolution is important to establish sustainable land use models and to estimate global fluxes of carbon, which affect climate change. Despite its importance, few measurements have been made in areas of the semiarid Brazilian Northeast region, and most of them were made using the alkali absorption method (AA), which can underestimate ECO2. Measurements using AA were compared to measurements using the infrared gas analyzer method (IRGA) over ten months (in rainy and dry seasons), during the day and night, in areas of Caatinga (xeric shrubland and thorn forest) and pasture in the Agreste region of the state of Pernambuco. The ECO2 measurements from AA varied little from night to day and throughout the year or in the rainy and dry seasons. However, those obtained from IRGA were higher in the rainy than in the dry season, but also without significant differences from day to night. The values of both methods were similar in the dry season, but in the rainy season they were higher with the IRGA. Therefore, AA seems to have little sensitivity to seasonal variations, in contrast with measurements from the IRGA, and it may underestimate soil ECO2 when it attains higher values. This result indicates that some of the soil ECO2 values determined in areas of the Brazilian semiarid region, and consequently annual C losses, may have been underestimated.


INTRODUCTION
The carbon cycle has received increased attention due to escalating concentrations of atmospheric carbon gas and its relationship with global warming and climatic changes (IPCC, 2014).Soil CO 2 efflux (ECO 2 ), also referred to as soil respiration, is an important component of the global carbon balance, returning about 80 × 10 9 tons of C to the atmosphere each year (Raich et al., 2002).Therefore, ECO 2 can have a large influence on global atmospheric CO 2 (Ryan and Law, 2005;Zheng et al., 2014) and the increase of ECO 2 may promote disturbing climatic changes (Pinto-Junior et al., 2009;Panosso et al., 2012).
The infrared gas analyzer (IRGA) measurement has been considered to be more precise but depends on the availability of relatively expensive equipment and, in general, is conducted over short periods of time (Rochette and Eller, 1991;Haynes and Gower, 1995).The AA methodology is cheaper since it requires only one simple piece of equipment (pH meter) and can integrate CO 2 evolution for many hours, which facilitates nocturnal measurements.These differences are probably the reason why AA has been the most commonly used methodology for ECO 2 measurement in the Northeast region of Brazil (Souto et al., 2009;Araujo et al., 2011;Ivo and Salcedo, 2012;Correia et al., 2015;Holanda et al., 2015).However, some reports point to possible overestimation of low ECO 2 fluxes and underestimation of high fluxes by the AA method (Janssens et al., 2000), although it can be reliably used in the intermediate flux range (Davidson et al., 2002).
Considering that both IRGA and AA methods have been used world-widely, our hypothesis is that they provide similar results when measuring ECO 2 under Caatinga or pasture and in different seasons in the Brazilian Northeastern semiarid region.Therefore, we aimed to compare ECO 2 measurements using the AA and IRGA methodologies applied to soils under Caatinga (xeric shrubland and thorn forest) and pasture vegetation in the Agreste region of the state of Pernambuco, Brazil.

MATERIALS AND METHODS
The experiment was conducted in two areas, one cultivated with the grass Brachiaria decumbens Stapf.and the other a neighboring Caatinga area, both located on the Riacho do Papagaio farm (8° 48' 34.2'' S; 36° 24' 29.3'' W and 702 m altitude) in the municipality of São João in the southern portion of the Agreste mesoregion of Pernambuco.Average annual rainfall is 782 mm, concentrated mainly from May to August (Silva et al., 2014).The climate is predominantly hot and humid, As' according to the Köppen classification system (Alvares et al., 2014), with dry summers and average temperatures ranging from 18.8 to 22.6 °C (Borges Júnior et al., 2012).The native vegetation is classified as hypo xerophilic Caatinga (Santos et al., 2012), and the soil in the experimental area is a sandy Neossolo Regolítico Eutrófico típico (Santos et al., 2012) or Entisol Arent (Soil Survey Staff, 2014).
The experimental area was deforested in the 1950s.Part of the area was not cropped, and the native Caatinga vegetation has been spontaneously regenerating ever since.Another part of the area was planted with Guatemala grass, followed by cassava and beans (1974-1980), corn (1981-1999), and Brachiaria grass (since 2000) (Almeida et al., 2015).In the pasture area, 24 points were marked for soil ECO 2 determination along two perpendicular transects in a cross shape (one with 14 and the other with 10 points); each point was 15 m distant from the next.In the Caatinga area, 12 points were marked along two winding trails.
The experiment followed a completely randomized design with two treatments (the AA and IRGA methods) and 24 replicates in the pasture area and 12 in the Caatinga area.
At each point, ECO 2 measurements were made with IRGA between 6:00 and 11:00 a.m. and between 6:00 and 11:00 p.m. once a month from April 2012 to January 2013.The period from April to August 2012 corresponded to the rainy season (total rainfall of 160 mm) and from September 2012 to January 2013 to the dry season (39 mm rainfall).It is important to note that 2012 was a drought year; therefore, rains during the rainy season were below average.
The IRGA apparatus (Licor LI-6400-09) has a gas retention chamber of 991 cm 3 , covering a soil surface area of 71.6 cm 2 , an infrared irradiator, and a measurement chamber, also described as an optical path and filter plus a detector.The infrared signal traverses the measurement chamber, which is filled with the sampled gas, and it is measured by the detector.The CO 2 emission is calculated by the linear regression of the increase in CO 2 concentration inside the chamber along the measurement period (Davidson et al., 2002).Before beginning measurement, the CO 2 concentration near the soil surface was registered (about 350 μmol mol -1 ), and this value was introduced in the software system of the apparatus to function as a reference value.At the beginning of each measurement, part of the gas chamber was inserted into the soil, and the CO 2 concentration inside the chamber was reduced to 10 μmol mol -1 , flushing the gas through a mixture of calcium oxide and sodium hydroxide.After this reduction, the subsequent CO 2 concentration increases inside the chamber, due to soil emission, were measured every 2.5 seconds, for a total period of 90 seconds, at which time the concentration inside the chamber was approximately 10 μmol mol -1 above the reference CO 2 concentration value.After this period, the LI-6400-09 software system calculated the linear regression between the CO 2 emission and its concentration inside the chamber, and the emission was assumed to have the value registered when the concentration inside the chamber was equal to the concentration of the reference value.
On the same days and in the periods when the IRGA measurements were made, AA measurements of soil ECO 2 were made at points close to those marked for the IRGA measurements.The CO 2 evolved from an area of 390.57cm 2 was absorbed by a 0.5 mol L -1 KOH solution placed in a 100-mL beaker with a 4.2 cm diameter opening (13.9 cm 2 ) under a cylindrical bucket with a 22.3 cm diameter opening and 21 cm height, which was inserted 2 to 3 cm inside the soil to avoid direct gas exchange with the atmosphere.Measurements were made separately during the day (from 6:00 a.m. to 6:00 p.m.) and during the night (from 6:00 p.m. to 6:00 a.m.), integrating the emissions for each 12-hour period.During each measurement, three control measurements were made, placing beakers with 10 mL KOH inside buckets that were hermetically sealed.After each measurement period, the KOH solution was transferred to hermetically sealed flasks to avoid exchange with the atmosphere, and the flasks were taken to the laboratory, where CO 2 was determined by titration with 0.1 mol L -1 HCl using 1 % phenolphthalein and methyl orange as pH indicators.The ECO 2 value (mg CO 2 m -2 h -1 ) was converted to μmol CO 2 m -2 s -1 through multiplication by 0.0063.
In addition to the ECO 2 measurements, ten soil samples were also collected in each area (pasture and Caatinga) from the 0.00-0.20 m surface layer to determine some Rev Bras Cienc Solo 2018;42:e0160563 soil properties: total organic carbon (TOC), texture, soil bulk density (SD), total porosity (TP), and macro-and micro-porosity.These determinations were made following the methodologies described by Donagema et al. (2011).The data were subjected to analysis of variance (F-test), and the averages compared by the Tukey test at the 5 % probability level.The identity test proposed by Leite and Oliveira (2002) was used to test if ECO 2 data obtained by the AA and IRGA methodologies were similar.Since the aim of the study was to compare the two methodologies, data from pasture were not compared to Caatinga, but only data from one methodology were compared to data from the other methodology in each area separately.These comparisons were made considering the whole period (April 2012 to January 2013), the rainy (April to August 2012) and the dry (August 2012 to January 2013) periods, and the diurnal and nocturnal periods.

RESULTS AND DISCUSSION
Soils from the two areas were not significantly different regarding their particle size fractions, both were classified as sandy.They were also similar regarding soil bulk density (SD), total porosity (TP), macroporosity (Ma), and microporosity (Mi).However, the Caatinga soil had a higher total organic carbon (TOC) content than the pasture soil (Table 1).Both contents were within the range usually found in sandy soils of semiarid regions (Santos et al., 2012;Medeiros et al., 2015) and the higher Caatinga TOC was not enough to cause a different pore size distribution compared to that from pasture soil.As expected for sandy soils (Santos et al., 2012;Silva et al., 2014;Almeida et al., 2015), macropores predominated over micropores.Texture can influence soil CO 2 movement because it alters porosity (Bouma and Bryla, 2000); more macropores favor soil respiration due to greater penetration of air.
Monthly variations in soil ECO 2 values obtained from the AA methodology were small, both for diurnal and nocturnal measurements and for Caatinga and pasture, with average values around 1.05 μmol CO 2 m -2 s 1 (Figure 1).Values obtained from the IRGA methodology were also similar for the diurnal and nocturnal periods and Caatinga and pasture areas, but they differed significantly between the rainy and the dry seasons, both in the Caatinga and pasture areas (Figure 1 and Table 2).
Differences between CO 2 evolution during the day and the night may be caused by lower night temperatures, which would result in higher relative humidity and, consequently, higher soil microbial activity (Souto et al., 2009;Holanda et al., 2015).However, no differences were found between day and night in either of the two methods (Table 2).It is possible that, in this region, even at night, soil temperatures remain high enough not to cause effects different from those observed during the day.In addition, soil in the areas has a sandy texture, and its high macroporosity (Table 1) may lead to high diffusion and aeration.SD: soil bulk density; TP: total porosity; Ma: macro-porosity; Mi: micro-porosity; TOC: total organic carbon.Sand, silt, clay, SD, TP, Ma, Mi, and TOC were determined following the methodologies described by Donagema et al. (2011).Averages followed by the same letter in the column are not significantly different by the Tukey test at the 5 % probability level.
Rev Bras Cienc Solo 2018;42:e0160563 These results confirm those of Correia et al. (2015) who measured ECO 2 using the AA method in Caatinga and pasture areas in the semiarid region of the state of Paraíba under climatic and soil conditions similar to those of the present study.Several authors (Souto et al., 2009;Correia et al., 2015;Holanda et al., 2015) report that high soil temperatures limit microbial activity in Caatinga soils because they reduce microbial populations and, consequently, reduce the intensity of organic residue decomposition and soil ECO 2 .
Soil ECO 2 values determined by IRGA were two to four times higher than those by AA when measured during the rainy season, from April to August 2012, whereas in the dry season (September 2012 to January 2013), values obtained from both methods were similar (Table 2).Considering that ECO 2 varies as a function of climate and soil variables, such as rainfall, air temperature, and soil moisture and temperature (Davidson et al., 2000;Deng et al., 2012;Chen et al., 2014), it was expected that ECO 2 values measured by AA would change significantly from the wet to the dry season, as they did when measured by IRGA (Figure 1 and Table 2).However, the average values measured by AA were only 10 % higher in the rainy than in the dry season, whereas those measured by IRGA were more than 260 % higher in the rainy season.Therefore, the IRGA methodology is more sensitive to seasonal variations than the AA methodology.Martins et al. ( 2010) stated that higher water availability during the rainy season should result in higher biological activity.Therefore, higher ECO 2 should also have been registered during the rainy season in this area in the Agreste region of the state of Pernambuco, which was only observed using the IRGA methodology (Figure 1 and Table 2).
Measurements performed in other Caatinga areas (Souto et al., 2009;Correia et al., 2015;Holanda et al., 2015), all of them using the AA methodology, registered small (0.28 µmol CO 2 m -2 s -1 ) and high amplitudes (0.75 µmol CO 2 m -2 s -1 ) between maximum (1.00 to 1.32 µmol CO 2 m -2 s -1 ) and minimum values (0.57 to 0.72 µmol CO 2 m -2 s -1 ).It is not possible to know if the similarity of ECO 2 values throughout the year, reported by some of these articles, was due to absence of environmental variables that substantially altered root respiration and soil microbial activity, or if it was caused by a limitation of the methodology.Nonetheless, in this study, it is more likely that rainfall variation was enough to explain the high variations in ECO 2 obtained from the IRGA methodology.
Considering the extreme difference in water availability usually observed between the rainy and the dry season in the Northeastern semi-arid region, and the effect that this availability has on root and soil microbial activities (Souto et al., 2009), large differences in ECO 2 are expected.Therefore, a doubt arises as to whether AA measurements during the rainy season of the semiarid region were not underestimated.In other Brazilian regions where measurements were also made with the AA methodology (Assis Júnior et al., 2003;Valentini et al., 2015), including forest and Cerrado (Brazilian tropical savanna) areas, there are no cases of low amplitude in ECO 2 values.The highest values were at least 50 % higher than the lowest, but were as much as eight times higher.
Since no ECO 2 measurements made with IRGA in the Northeast region were found in the literature beyond those of the present study, comparisons are not possible.However, measurements in forest, Cerrado, and agricultural areas in other Brazilian regions (Sotta et al., 2004;Valentini et al., 2008;Panosso et al., 2009;Pinto-Junior et al., 2009;Zanchi et al., 2012) report higher amplitudes (2.51 to 6.16 µmol CO 2 m -2 s -1 ) between maximum (3.96 to 10.51 µmol CO 2 m -2 s -1 ) and minimum (1.45 to 4.35 µmol CO 2 m -2 s -1 ) values than those measured with the AA methodology.The smallest difference (about 50 %) occurred in a sugarcane field in Mato Grosso do Sul (Moitinho et al., 2013); all other maximum values were two to five times higher than the minimum values.These findings suggest that the IRGA methodology has higher sensitivity in detecting ECO 2 differences than the AA methodology.However, these are comparisons derived from few studies and under different climate and soil conditions, and it cannot be confirmed that the AA methodology was not adequate under the conditions in which it was used.
Several studies point to advantages in using the AA methodology: Assis Júnior et al. (2003) and Souto et al. (2009) highlighted that it is a simple, low cost, and sensitive method that can be applied simultaneously at many points and that can integrate CO 2 evolutions over many hours (even 24 hours), therefore reducing spatial and temporal variability.If it is not possible to use an IRGA, which requires equipment that is expensive, has high maintenance costs, and demands a specially trained operator, use of the AA methodology is justified.However, many authors, especially those working in other countries, have claimed that the AA methodology can underestimate ECO 2 values (Haynes and Gower, 1995;Jensen et al., 1996;Yim et al., 2003).
There are several explanations for these underestimations.The underestimations might result from the measuring chamber remaining in the same spot for a long time, which could disturb the micro-climate, demands the daily temperature amplitude (Jensen et al., 1996).However, Hendry et al. (2001) performed measurements for periods longer than 24 hours and reported values similar to those obtained from other methodologies.Absorption of CO 2 by the alkaline solution could also decrease over the measurement period due to low diffusion (Freijer and Bouten, 1991), the CO 2 concentration inside the chamber decreasing the diffusion gradient between the soil and air inside the chamber, reducing the flux absorbed by the alkaline solution.Under laboratory conditions, Freijer and Bouten (1991) observed that static absorption was not sufficient to react with all the respired CO 2 , due to lack of contact between the solution and the entire CO 2 volume inside the chamber.One or more of these problems probably occurred in the present study since the ECO 2 values obtained by the AA methodology were underestimated and did not reflect the changes that happened from the dry to the rainy season.However, it is not possible to know for sure if those problems occurred during the present study.Considering that the methodology was applied in a way similar to other studies using AA, similar problems could have occurred in these other studies, which were not detected due to the absence of comparisons with other methodologies.
The IRGA methodology has been criticized because it does not integrate CO 2 fluxes over long time periods, as the AA methodology does, since measurements last only a few minutes (Davidson et al., 2002).To extend these periods, it would be necessary to perform a large number of measurements (Panosso et al., 2012).According to Perez-Quezada et al. (2016), this is possible, and despite the high cost of the equipment, the IRGA methodology has a better cost/benefit ratio than others.

Table 1 .
Soil physicalproperties and total organic carbon in areas of Caatinga and pasture, in the municipality of São João, Pernambuco, Brazil Figure 1.Average soil ECO 2 values, in diurnal and nocturnal periods, measured by infrared gas analyzer and alkali absorption methodologies, in areas of Caatinga and pasture, in the municipality of São João, Pernambuco, Brazil

Table 2 .
Leite and Oliveira (2002)alysis, according to the methodology proposed byLeite and Oliveira (2002), comparing infrared gas analyzer (IRGA) and alkali absorption (AA) methodologies for ECO 2 measurements in areas of Caatinga and pasture, in the municipality of São João, : significant at the 5 % probability level; Yj ≠ Y1: the AA methodology differs from the IRGA methodology. *