Contents lists available at ScienceDirect

Agricultural and Forest Meteorology

journal homepage: www.elsevier.com/locate/agrformet

Native shrubland and managed buffelgrass savanna in drylands:Implications for ecosystem carbon and water fluxes

César Hinojo-Hinojoa, Alejandro E. Castellanosa,⁎, Travis Huxmanb, Julio C. Rodriguezc,Rodrigo Vargasd, José R. Romo-Leóna, Joel A. Biedermane

a Departamento de Investigaciones Científicas y Tecnológicas, Universidad de Sonora, Hermosillo, Sonora 83000, Méxicob Ecology and Evolutionary Biology, Center for Environmental Biology, University of California-Irvine, Irvine, CA 92629, USAc Departamento de Agricultura y Ganadería, Universidad de Sonora, Hermosillo, Sonora 83000, MéxicodDepartment of Plant and Soil Sciences, University of Delaware, Newark, Delaware 19716, USAe Southwest Watershed Research Center, Agricultural Research Service, Tucson, AZ 85719, USA

A R T I C L E I N F O

Keywords:BuffelgrassCenchrus ciliarisEddy covarianceLand cover changeSonoran DesertCarbon and water fluxesGrass encroachment

A B S T R A C T

Land cover and land-use change (LCLUC) between woody- and grass-dominated ecosystems in drylands compriseone of the largest uncertainties in the land CO2 sink. This is especially true for the widespread transition fromshrublands to grasslands/savannas caused by the establishment of exotic C4 grass species for grazing or throughbiological invasion of these species, where information about its impacts on ecosystem CO2 fluxes is limited. Forstudying this, we used three years of eddy covariance measurements of net ecosystem production (NEP), grossprimary production (GPP), ecosystem respiration (Reco) and evapotranspiration (ET) over a Sonoran Desertshrubland and an adjacent grazing savanna of buffelgrass (Cenchrus ciliaris L.), established 35 years ago. Atmonthly, seasonal and annual time scales, we assessed whether between-site differences in CO2 fluxes wererelated to differences in ecosystem water use, measured as the fraction ET to precipitation, water use efficiency(WUE, i.e. the ratio between GPP and ET) and/or to the relation between Reco and GPP. Although the savannahad higher WUE than the shrubland, its summer NEP was limited by water use, due to limitations in leaf areaindex and likely rooting patterns. Conversely, the savanna had higher NEP than the shrubland during fall tospring due to increased WUE; possibly due to activity of buffelgrass and remaining woody species using(summer) water from deeper soil layers. However, differences across these seasons compensated each other atthe inter-annual scale, and both sites had comparable net carbon sinks over the three-year study period. Furtherstudies are needed to understand and reduce the uncertainty associated with carbon and water fluxes associatedwith LCLUC in dryland ecosystems.

1. Introduction

Dryland ecosystems have a key role in the trend and interannualvariability of the global land carbon sink (Ahlstrom et al., 2015; Poulteret al., 2014). These globally distributed ecosystems have large temporaland spatial variability in net ecosystem production (NEP), and conse-quently in CO2 uptake by gross primary production (GPP) and emis-sions to the atmosphere by ecosystem respiration (Reco) (Ahlstromet al., 2015; Biederman et al., 2016, 2017). Furthermore, extensiveland-cover and land-use change (LCLUC) has been documented acrossdrylands (Safriel et al., 2005), which in turn directly influence themagnitude of these fluxes (Le Quéré et al., 2016; Baldocchi, 2008). Intemperate forest ecosystems, which have traditionally have been more

studied than drylands, LCLUC typically results in net CO2 emissions tothe atmosphere due to biomass loss and decomposition (Le Quéré et al.2016). Changes in the net carbon balance in drylands following LCLUCare still unclear and represent one of the largest uncertainties in theglobal carbon balance (Hayes et al., 2018; King et al., 2007).

Woody shrubs and grasses are dominant in drylands (Safriel et al.,2005). The transitions from woody- to grass-dominated ecosystems andvice versa are recognized as the most widespread LCLUC across dry-lands (Barger et al., 2011; Marshall et al., 2012; Sala and Maestre,2014). The gradual change over decades from grass- to woody-domi-nated ecosystems is referred as woody encroachment, which is regu-lated by interactions among climate, vegetation, and managementacross drylands (Van Auken, 2009). In the opposite direction, a change

https://doi.org/10.1016/j.agrformet.2019.01.030Received 6 March 2018; Received in revised form 14 November 2018; Accepted 21 January 2019

⁎ Corresponding author.E-mail address: acastell@guaymas.uson.mx (A.E. Castellanos).

Agricultural and Forest Meteorology 268 (2019) 269–278

Available online 30 January 20190168-1923/ © 2019 Elsevier B.V. All rights reserved.

T

from woody- (e.g., shrublands) to grass-dominated ecosystems couldoccur by the anthropogenic introduction of highly productive anddrought-resistant African/Asian C4 grass species to increase forageproduction (Castellanos et al., 2002, 2010; Marshall et al., 2012;Williams and Baruch, 2000). Vegetation structure may be dramaticallychanged within a season through land clearing of native vegetation andseeding of grass or more gradually through invasion of exotic grasses(Arriaga et al., 2004; Williams and Baruch, 2000). These bidirectionalLCLUC transitions could have important consequences for regional-to-global carbon budgets, but large uncertainties remain (Barger et al.,2011; Houghton et al., 2012 Jackson et al., 2002; Pacala et al., 2001).Woody encroachment has been more extensively studied (Petrie et al.,2015; Scott et al., 2006, 2014), and because of the lack of information,larger uncertainties exist for the shrubland-to-grassland transition andparticularly the consequences for whole-ecosystem stocks and annualfluxes (Bradley et al., 2006; Prater et al., 2006).

Woody encroachment increases whole-ecosystem carbon fluxes(NEP, Reco, GPP) in drylands because access to deeper soil andgroundwater sources supports longer growing seasons (Scott et al.,2006, 2014; Petrie et al., 2015), and because encroaching woody spe-cies can physiologically outperform native grass species under a widerange of drought conditions (Barron-Gafford et al., 2012; Throop et al.,2012). Changes are expected in evapotranspiration (ET), the water fluxof available soil moisture (SM) that remains after infiltration fromprecipitation (P), and losses to runoff (R) and drainage (D) beyond therooting zone (Fig. 1). ET approximates the available water that drivescarbon fluxes (Biederman et al., 2016), and ET/P represents the fractionof water used at the ecosystem scale. Two relations are particularlyimportant and explain the variation of NEP in drylands, which can be

modified by land cover: the relation of GPP with ET, and the relation ofReco with GPP (Biederman et al., 2016). ET/GPP describes how avail-able water is used for photosynthesis, reflecting the ecosystem water-use efficiency. Meanwhile, Reco/GPP reflects how carbon availabilitydrives ecosystem respiration, and is sensitive to changes in vegetation,carbon pools and allocation caused by LCLUC or disturbances(Baldocchi, 2008; Biederman et al., 2016). In this context, C4 grasslandshave higher water-use efficiency than woody-encroached ecosystemsdominated by C3 plants (Emmerich, 2007; Kurc and Small, 2007; Scottet al., 2014). This advantage is counteracted by an increased Reco/GPP,which is likely related to the high proportion of root biomass in grasses(Scott et al., 2014; Petrie et al., 2015). The balance between these re-lationships further contributes to explain the lower NEP of grasslandswhen compared to woody-encroached sites. It is unclear how a LCLUCin the opposite direction (i.e., shrublands converted into C4 grasslands)influences NEP in dryland ecosystems. We could expect diminished NEPby reversing the typical results of woody encroachment. Howeverexotic grass species have historically been selected for high productivityunder limiting abiotic conditions (Cox et al., 1988) and could lead to adifferent outcome.

Buffelgrass (Cenchrus ciliaris L.) is one of the main C4 exotic grassspecies used for converting xeric shrublands into more productivegrasslands/savannas for cattle grazing over America and Australia (Coxet al., 1988; Marshall et al., 2012). Notably, the Sonoran Desert ofMexico and the US may be one of the region's most heavily impacted bythese practices (Bravo-Peña et al., 2010; Bravo-Peña and Castellanos,2013; Castellanos et al., 2002, 2010; Marshall et al., 2012). This per-ennial bunchgrass has one of the highest photosynthesis rates(Castellanos et al., 2010; Larcher, 2014) and can physiologically out-perform dominant C3 woody species from the Sonoran Desert(Castellanos et al., 2010) enabling exotic buffelgrass savannas to besubstantial carbon sinks (Hinojo-Hinojo et al., 2016). However, remotesensing estimates suggest that buffelgrass-dominated ecosystems haveless above—ground biomass production than Sonoran Desert shrub-lands (Bravo-Peña and Castellanos, 2013; Franklin et al., 2006; Franklinand Molina-Freaner, 2010). Thus, research is needed to contribute toour understanding of shrubland—to—grassland transitions across dry-land ecosystems.

In this study we compare three years of CO2 and water fluxes over ashrubland and an adjacent ecosystem converted to exotic buffelgrasssavanna ˜35 years ago to address the following questions: 1) Does thechange from shrubland into an exotic buffelgrass savanna modifiescarbon fluxes (e.g., NEP, GPP and Reco) at seasonal and annual scales?;and if so, 2) are changes related to responses in water use (ET/P), wateruse efficiency (GEP/ET) and/or the relation between Reco/GPP? Wehypothesize that the high-productivity potential and water-use effi-ciency of buffelgrass will compensate with a high Reco/GPP and lowerwater use in the savanna; which is characterized by a shallower rootingstructure and allows this ecosystem to sustain similar or higher NEPthan in the shrubland. Our ultimate aim is to contribute to the discus-sion about the bidirectional changes of wood-to-grass transitions acrossglobally important drylands ecosystems.

2. Materials and methods

2.1. Study sites

Two adjacent sites representing a native shrubland and a buffelgrasssavanna were selected for installing eddy covariance (EC) systems tomeasure ecosystem scale carbon dioxide (CO2), water and energyfluxes. These sites are part of the Mexican eddy covariance network(MexFlux, Vargas et al., 2013) and are located in Sonora, Mexico,within the Southeastern portion of the Sonoran Desert, a heavily im-pacted region by LCLUC for buffelgrass since its introduction to Mexicoin 1960′s (Bravo-Peña and Castellanos, 2013; Castellanos et al., 2002,2010 Franklin et al., 2006). Both sites are located in El Churi, a private

Fig. 1. Conceptual diagram of the main ecosystem CO2 and water fluxes in anative shrubland and one changed to exotic buffelgrass savanna, and the mainstructural characteristics which may influence the magnitude of those fluxes.GPP: gross primary production, Reco: ecosystem respiration, ET:Evapotranspiration, P: precipitation, R: runoff, D: drainage (For interpretationof the references to colour in this figure legend, the reader is referred to the webversion of this article).

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

270

ranch near La Colorada, Sonora, where rotational management of cattlegrazing is the prevalent activity throughout the year. The sites are se-parated by 1.8 km, thus both are exposed to the same regional en-vironmental conditions but with different vegetation cover.

The shrubland site (located at 28° 41′ 53.60″ N, 110° 32′ 20.59″ W,with 450m elevation; Fig. 2) has a Sonoran Desert scrub vegetationcharacteristic of the Plains of Sonora (Brown, 1994). The vegetationcover is composed of 14% trees (Olneya tesota, Fouquieria macdougalii,Ipomoea arborescens, in descending order of importance) and 28%shrubs (Jatropha cardiophylla, Encelia farinosa, Mimosa laxiflora, Lyciumberlandieri) (Celaya-Michel et al., 2015). During the study period, in-terspaces between trees and shrubs were mostly covered with summer-active herbs, grasses and litterfall. Vines can also have some coverduring summer (about 10%). Mean vegetation height is nearly 1m, buta few of the tallest trees can grow up to 7m. The EC tower is located atthe center of homogeneous vegetation for more than 1 km in everydirection (Fig. 2, Footprint analysis in Supplementary material, FigureS1).

The buffelgrass savanna site (located at 28° 42′ 40.32″ N, 110° 32′58.14″ W, with 399m elevation; Fig. 2) was a former Sonoran Desertscrub converted to savanna approximately 35 years ago through se-lective land clearing and re-cleared periodically, with only few of thedominant tree species left (J. Dueñas, oral communication, June 2009),and some shrub and other species that have partially recolonized thesite since. The vegetation cover was 32% buffelgrass (Cenchrus ciliaris),3% trees (Olneya tesota and Prosopis velutina), and 7% shrubs (mostlyMimosa laxiflora, and Jatropha cardiophylla). During the study period,interspaces between plants had a cover of summer active herbs, grassesand litterfall. Mean vegetation height is about 0.5 m, but a few treesgrow up to 4m.

Climate for these sites (30-year averages from nearby station 26268,San Jose de Pima, Servicio Meteorológico Nacional) is warm-semi-arid,with a mean annual precipitation of 476mm, 70% of which occursduring the summer monsoon as thunderstorms. Mean annual tem-perature is 22.8 °C, with mean maximum in the hottest month (June) of40.1 °C, and a mean minimum of 5.4 °C in the coldest month (January).Most plant species from both sites are deciduous with the major period

of photosynthetic activity between July and September (period ofmonsoon rains), but some activity can occur during other seasons givensufficient precipitation (Hinojo-Hinojo et al., 2016). The species Olneyatesota, Prosopis velutina, Encelia farinosa and Lycium berlandieri can re-tain their leaves during most of the year when sufficient soil moistureremains in their rooting zone (Castellanos et al., 2016). Terrain is flatwithin the tower footprints with an average slope< 2% (Fig. 2). Soilsare calcic regosol and haplic phaeozem with loamy sand texture with0.6–2.6 % of organic matter content (Celaya-Michel et al., 2015).

2.2. Eddy covariance and meteorological measurements

The instrumentation installed in both sites allowed us to measurethe net ecosystem CO2 exchange (NEE), evapotranspiration (ET), andenergy fluxes between the ecosystems and the atmosphere using theeddy covariance technique (Aubinet et al., 2012) and to monitor theenvironmental conditions under which these fluxes occur. In this studywe used six site—years of EC comparative data (from March 2013 to theend of 2015). At the buffelgrass site, the EC system on a 6m-tall towerconsisted of a sonic anemometer (CSAT3, Campbell Scientific, Logan,UT, USA) for measuring 3D components of wind velocity and sonictemperature and an open path CO2/H2O gas analyzer (Li-7500,LI−COR, Lincoln, NE, USA), both sampling at 10 Hz. Also at the top ofthe tower, net radiation (NR Lite, Kipp & Zonen, Delft, The Nether-lands) and air temperature and relative humidity (HMP45C, VaisalaInc., Vantaa, Finland) were measured every minute. All other en-vironmental sensors had a sampling rate of 15min including a raingauge (TR-525USW-R, Texas Electronics, Dallas, TX, USA) and twopairs of soil heat flux plates (HFP01-L50, Hukseflux, Delft, The Neth-erlands) located at 5 cm depth under buffelgrass cover and beneathopen spaces (with annual herbs cover). All data were sampled andstored using a data-micrologger (CR3000, Campbell Scientific, Logan,UT, USA).

At the shrubland site, the EC system on a 9m tall tower had a 3Dsonic anemometer (Wind Master Pro, Gill Instruments) and an openpath CO2/H2O gas analyzer (Li-7500 A, LI−COR, Lincoln, NE, USA) atthe top, both taking measurements at 10 Hz. All other sensors measured

Fig. 2. Location of the study area within the Sonoran Desert and of the eddy covariance (EC) towers in the landscape. See Figure S1 on supplementary material forfurther details on footprint analysis.

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

271

at a sampling rate of 1min. Also at the top, the tower has a netradiometer (NR Lite2, Kipp & Zonen, Delft, Netherlands), and at 6mheight an air temperature and relative humidity probe (HMP-155,Vaisala Inc., Vantaa, Finland). Rainfall was measured with a rain gauge(TR-525M, Texas Electronics, Dallas, TX, USA). Soil heat flux wasmeasured (HFP01SC, Hukseflux, Delft, The Netherlands) at 5 cm depthunder tree, shrub and open space cover. All 1-minute frequency datawas sampled with a datalogger (Xlite 9210, Sutron, Sterling, VA, USA)and 1min and 10 Hz frequency data are stored in the interface unit (Li-7550, LI−COR, Lincoln, NE, USA) of the gas analyzer.

2.3. Data processing

The high frequency (10 Hz) raw data processing and flux calculationwere performed on Eddy Pro software (v. 4-5, LICOR Biosciences) on30-minute blocks. Displacement height and roughness length settingsused on Eddy Pro were set to be a function of vegetation height at thesites (see Study sites section). Prior to flux calculation, the followingprocessing was performed on raw data: despiking and other statisticalquality tests (following Vickers and Mahrt, 1997), time lags compen-sation due to instrument separation by maximizing covariance, anddouble axis rotation for wind components (Wilczak et al., 2001). Angleof attack correction (Nakai and Shimoyama, 2012) was performed onthe shrubland anemometer data to compensate for errors in wind ve-locity data due to instrument design, whereas the anemometer used atthe buffelgrass savanna is designed to minimize such errors, and thiscorrection was not used. Turbulent fluxes of CO2, ET, sensible (H) andlatent heat (LE) were calculated using the vertical component of windvelocity, CO2 and water vapor molar densities, and sonic temperature.These fluxes were corrected for frequency spectral attenuations(Moncrieff et al., 1997 and 2004), humidity effects on sonic tempera-ture (van Dijk et al., 2004), and air density fluctuations (Webb et al.,1980).

As a quality control, we discarded data under the following condi-tions: 1) when data failed tests for steady state and sufficiently devel-oped turbulence assumptions of the EC technique (Mauder and Foken,2004), 2) fluxes measured below a friction velocity threshold to avoidkeeping underestimated nighttime flux data at conditions of low airmixing, 3) during rain events, 4) when bird feces occluded the gasanalyzer lenses dropping CO2 concentration below 350 ppm (whichcaused substantial overestimation of CO2 fluxes, personal observation).Friction velocity thresholds for the shrubland were 0.15–0.13m/s whilesavanna thresholds were 0.1 m/s for all years. Quality control andsystem failure resulted in an annual NEE gap fraction of 37–51% forshrubland, and 40–51 % for the savanna, depending on the year. Theestimation of friction velocity thresholds, gap-filling and NEE fluxpartitioning into Reco and GPP were performed according to Reichsteinet al. (2005) using the online tool available at www.bgc-jena.mpg.de/MDIwork/eddyproc/. We followed the ecosystem-centered nomen-clature where Net Ecosystem Production (NEP = - NEE) positive valuesindicate net carbon uptake by the ecosystem and negative values in-dicating net carbon emission to the atmosphere (Chapin et al., 2006).

Monthly, summer and annual flux sums were obtained by adding upthrough time the half-hourly gap-filled data. We estimated uncertaintyin flux sums derived from gapfilling processing (obtained from theonline tool, following Reichstein et al., 2005) and from random errorsand long gaps (following Richardson and Hollinger, 2007). We ana-lyzed the energy balance closure for the sites as a way to assess theaccuracy of the eddy covariance measurements (Supplementary mate-rial, Figure S2).

2.4. Comparison of carbon and water fluxes between sites

To answer our first research question, we compared seasonal andannual CO2 fluxes between the study sites. For this, a t-test was per-formed to compare mean daily flux values for the period of interest

(either a single season or a single year). Given that the daily meanmultiplied by the number of days equals the sum for a given period, anybetween-site difference detected at daily means could be interpreted asa potentially significant difference that influence flux sums. For theseasonal flux comparison, we split each year in two seasons; (a) summerfrom July to September, where most of the annual precipitation and thehighest flux rates occur, and (b) the rest of the year including data fromJanuary to June and October to December where rainy events are un-common and flux rates tend to be low.

2.5. Drivers of ecosystem carbon flux differences between-sites

Our second research question was about identifying whether be-tween-site differences in ecosystem carbon fluxes were related to dif-ferences in the amount of water used by the ecosystem; in terms ofwater use efficiency or in the relation Reco/GPP. Both monthly ET andseasonal ET/P were used as a measure of water use by the ecosystem, asboth have been shown to represent well the available water that drivesCO2 fluxes (Biederman et al., 2016, 2017). Water-use efficiency wasassessed at the monthly scale in two ways: (a) as the linear relationshipof GPP against ET, which gives an estimate of realized water-use effi-ciency at the ecosystem scale (WUE) (Emmerich, 2007); and (b) as thelinear relationship of the product of GPP and vapor pressure deficit(VPD) against ET, which is an estimate of inherent water use efficiency(WUEi) and better reflects changes in plant physiology by controllingfor the effect of VPD (Beer et al., 2009). The linear relationship ofmonthly Reco against GPP was also assessed. To compare how sitesdiffered in these relationships, we fit the data to a statistical modelevaluating the effects of the independent variable (either ET or GPP),site, and independent variable*site interaction on the dependent vari-able. The site effect gives an estimate of mean displacement of thedependent variable between sites, when all other effects are controlled.The interaction effect gives an estimate of the amount of between-sitechange in slopes of the independent vs dependent variables. We inter-pret any statistically significant effect of site and/or interaction as asignificant change in WUE, WUEi or Reco/GPP.

2.6. Vegetation phenology

We followed vegetation phenology through remote sensing data andwith field observations to assess whether changes in flux dynamics andrelations could be related to changes in canopy development or phe-nology of dominant species. For remote sensing data, we used timeseries of 8-day composites of leaf area index (LAI) derived from MODIS(Moderate-Resolution Imaging Spectroradiometer, MOD15A2, collec-tion 5) for the same dates as flux measurements obtained from thewebpage of Distributed Active Archive Center for BiogeochemicalDynamics (http://daac.ornl.gov/MODIS/). Linear regression analysiswas used to assess how LAI responded to seasonal rainfall and how LAIinfluenced CO2 fluxes at both sites. For field observations of leaf phe-nology, we qualitatively censed the presence of green leaves on 10 in-dividuals of the dominant species (> 0.5% cover) during visits to thesites every three to four weeks throughout the study period. Any givenspecies was considered to have green leaves when they were the mostabundant in more than six individuals. Leaf phenology was combined insome of our analyses placing special attention to whether buffelgrassphenology could help explain the observed differences in fluxes andflux relations between sites. All statistical analyses were performed inJMP version 9.0.1 (SAS Institute, Inc, 2010).

3. Results

3.1. Temporal dynamics of meteorological conditions and ecosystem fluxes

Marked differences in precipitation amounts and seasonal timingcharacterized the three years of study (Fig. 3a–c). In 2013, annual

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

272

rainfall (253mm) was below the long-term average, mostly dominated(86%) by summer rains (Fig. 3a), which also had the coldest winter ofthe three years (Fig. 3d). In contrast, total annual precipitation in 2014was 412mm with a rainy summer but dry winter and spring (Fig. 3b–c),and in 2015, a wet year (505mm), rains were spread throughout theyear with above-average precipitation in winter and spring but below-average in summer (Fig. 3c). Temperature patterns were similar foryears 2014 and 2015 (Fig. 3e-f), which drove daily mean air VPDhighest during late spring (4 kPa), and 1–3 kPa for the rest of the sea-sons (Fig. 3g–i). Thus, the three years allowed us to compare CO2 andwater fluxes over a wide range of conditions for the two sites, withweekly and monthly values providing a first approach for comparisons.

Important differences were detected in monthly fluxes and 8-dayLAI dynamics between sites (Fig. 4). Except for the summer, buffelgrasssavanna had higher monthly CO2 fluxes at other times during the threeyears (Fig. 4g–o), especially for GPP and NEP, but these were not ex-clusively related to buffelgrass activity in 2013 and 2014 (Fig. 4a–b).On-site phenological observations showed that bufflegrass had greenleaves mainly during summer in 2013 and 2014, but rain patterns in2015 allowed this species to maintain green leaves during seasons ofthe year that otherwise they would not have been active already(Fig. 4a–c). The shrubland had higher CO2 fluxes, ET and LAI than thebuffelgrass savanna early in the summer seasons of 2014 and 2015, butwere similar and highly variable for both sites during 2013 (Fig. 4).

3.2. Annual and growing season comparisons of ecosystem fluxes

Annual CO2 fluxes ranged from being carbon neutral to a sinkduring the years of study. Carbon and water fluxes for the years 2013 to2015 ranged from being carbon neutral in the shrubland in 2013 (t-testagainst 0: t= 0.0544, P=0.3524) to a sink in the buffelgrass savanna(year 2013). Both sites were carbon sinks during 2014 and 2015 and nostatistically significant differences were found between-sites for bothyears. Flux uncertainties for annual NEP estimates were±65 to 68 g Cm−2 for the shrubland and±66 to 78 g C m−2 for the buffelgrass sa-vanna. When summed over the three years, both sites had similar NEP(393 and 414 g C m−2 year-1 for shrubland and savanna, respectively).Reco at the savanna was 75–90 g C m−2 year-1 higher than the shrublandduring 2013 and 2015, but GPP at the savanna compensated for thosedifferences. Sites had similar ET during 2013, but the savanna hadlower ET than the shrubland during 2014 and 2015 and lower ET/Pratios, which were evident both in the total annual and during thesummer (Table 2).

The importance of summer fluxes to the total annual was alwayshigher, although their magnitude and annual sums were different

between years and sites. During the years with minimal rains in winterand spring (2013 and 2014), summer contribution to the total annualNEP was over a hundred percent in the shrubland, but only 59–89 % inthe buffelgrass savanna (Table 2). In those same years, summer con-tributed with 68–86 % of the yearly fluxes of GPP, Reco and ET at bothsites, with about 7–18 % higher contribution of carbon fluxes and 1–7% higher ET in the shrubland than the buffelgrass savanna. However,during the more widespread rainfalls of 2015 (with rains in winter,spring and summer), summer contribution to annual flux totals was 46to 56% (ET and NEP respectively) in the buffelgrass savanna, and 46(ET) to 72% (NEP) in the native shrubland.

Leaf area index was related to summer precipitation. We found 1.6times higher slope between summer rainfall and LAI for the shrublandsite than for the buffelgrass savanna (Fig. 6a), but no differences in theslope of the relationship between LAI and NEP were found (Fig. 6b).

3.3. Comparison of water use, water use efficiency and Reco/GPP

Lower ET/P was always found at the buffelgrass savanna site.Annual ET/P was lower during the rainy years (2014 and 2015) andevery summer season during our period of study in the buffelgrass sa-vanna (Table 2). The buffelgrass savanna had also higher WUEi than theshrubland, either when buffelgrass was active or inactive (Fig. 5a). Thesavanna assimilated about 2.4 g C mm−1 H2O month−1 and 188 g ChPa and had higher GPP*VPD for a given ET (marginally significantinteraction effect and significant site effect in Table 1) than theshrubland. Our estimates indicate that the shrubland could only hadhigher NEP than the savanna if it diminished 5.5 mm month−1 ET ormore (y=-5.8+1.06x, Fig. 7), which only happened during summermonths and few times in other seasons.

4. Discussion

4.1. Comparison with current understanding of reverse LCLUC transitionsfrom woody- to grass- dominated ecosystems

Our study supports the hypothesis that a LCLUC from shrubland tosavanna dominated by a highly productive exotic C4 grass may result inan ecosystem that is as productive as the shrubland. This contradicts theprediction from the reverse transition, woody encroachment, wherewoody-dominated ecosystems tend to have higher NEP than grass-dominated ecosystems (Barger et al., 2011; Scott et al., 2006, 2014;Petrie et al., 2015). Our findings are supported by an experimentalstudy that found increasing carbon pools in shrublands invaded byexotic grasses (Wolkovich et al., 2010). Here, we discuss mechanisms

Fig. 3. Annual course of monthly rainfall (a–c), mean daily airtemperature (d–f), and vapor pressure deficit (g–i) at theshrubland (black symbols) and buffelgrass savanna sites (graysymbols) during the study period. In a–c, bars representmonthly rainfall during the study period and lines representlong-term average monthly rainfall. Vertical dotted lines arethe limits between seasons (i.e. Winter, Spring, Summer andFall).

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

273

and feedbacks operating in both ecosystems that could explain theobserved pattern and possible implications of bidirectional woody-grassland transitions on dryland ecosystem carbon fluxes.

When compared, the buffelgrass savanna had either higher or si-milar annual NEP than the shrubland in the same year but summed overthe study period (3 years) the sites were net carbon sinks and ended upwith closely similar NEP. This contrasts with higher NEP expected fromwoody-dominated compared to grass-dominated ecosystems attributedto woody plants having access to deeper water sources, allowing higherET/P ratio, longer growing seasons, and lower Reco/GPP (Huxmanet al., 2005; Petrie et al., 2015; Scott et al., 2006, 2014). These expected

changes did not operate in the same way at our sites. Growing seasonlength was fairly similar between sites (as suggested by time series ofNEP and GPP, Fig. 4) despite increased water availability at theshrubland (ET/P from Table 2), probably because vegetation at bothsites is dominated by deciduous and highly seasonal species (Turneret al., 1995). Furthermore, the relationship between GPP and Reco didnot differ between sites (Fig. 5b). Although Reco was higher at the sa-vanna than shrubland during some years (2013, 2015), GPP throughoutthose years compensated such increases at both sites (Fig. 5b), whichagrees with our hypothesis that the high productivity potential ofbuffelgrass would compensate the typical high Reco/GPP relationcommonly reported for grass-dominated ecosystems. Alternatively, sitesdiffered in their seasonal contributions to annual fluxes, making theshrubland NEP more dependent on the summer season than the buf-felgrass savanna (Table 2, Fig. 7).

4.2. Water use and water use efficiency as drivers of seasonal differences inecosystem fluxes

Although both sites had similar NEP sums over the three years, theunderlying seasonal fluxes and water-carbon relationships differedamong sites, specifically water use (ET/P, Table 2), and water-use ef-ficiency (GPP*VPD/ET, Fig. 5). The buffelgrass savanna had higherWUEi than the shrubland throughout the year, supporting our hy-pothesis that at least part of this increase in WUE is due to the C4

physiology of buffelgrass, as has been found for other C4 dominated

Fig. 4. Annual course of a–c) leaf area index, d–f) monthly evapotranspiration (ET), g–i) net ecosystem production (NEP), j-l) ecosystem respiration (Reco) and m–o)gross primary production (GPP) for the shrubland (black symbols) and buffelgrass savana (grey symbols) sites. Horizontal grey bars in a–c) indicate periods of greenleaf phenology of buffelgrass. Vertical dotted lines as in Fig. 3.

Fig. 5. a) Intrinsic water-use efficiency and b) Reco/GPP relationship for theshrubland (black symbols-full line) and buffelgrass savanna (open symbols-dotted line) sites. Star symbols represent months when buffelgrass was active,and lines are linear regressions for each site. Labels as in Fig. 4 and text.

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

274

grasslands when compared to C3 dominated shrublands (Ehleringer andMonson, 1993; Emmerich, 2007; Scott et al., 2014, 2015). However, thesavanna had lower NEP and ET than shrubland during most summermonths (Fig. 7) despite having higher WUE, which suggests waterlimitation. Two biotic mechanisms appear to be limiting water use andNEP at the savanna during summer. First, the conversion of a shrublandinto a buffelgrass savanna reduces the cover of most plant functionaltypes, especially woody species with deep roots (Saucedo-Monarqueet al., 1997; Castellanos et al., 2002, 2010; Franklin and Molina-Freaner, 2010). At the savanna site, deep-rooted woody species coverless than 10%, and buffelgrass and annual plants are known to haveroots limited to the top 60 cm of soil (Forseth et al., 1984; Mnif andChaieb, 2009). Thus, most plant cover at the savanna is dependent onshallow soil water availability. In the shrubland, at least 30% of itscover is deep-rooted woody plants, and there is an important propor-tion of herbs and shrubs with shallow roots, thus allowing this eco-system to better exploit water throughout the soil profile duringsummer and thus allowing higher summer transpiration and associatedGPP (Castellanos et al., 2016; Celaya-Michel et al., 2015 Huxman et al.,2005). However, the small proportion (˜10%) of deep-rooted woodyplants remaining in the savanna may benefit at other seasons, from thedeep water that buffelgrass was not able to utilize during the summer

rainy season.We found that LAI is another mechanism that may be limiting NEP

in the savanna. LAI was similar for both sites in the low range of pre-cipitation, but differences increased as summer rainfall increased, withshrubland LAI exceeding that of the buffelgrass savanna by up to 60%in wet years (Fig. 6). Knapp et al. (2008) reported similar patterns andsuggested that grasses may have an intrinsic LAI limitation due to theirarchitecture, with meristematic disposition only close to or below thesoil surface, while the branching and meristem arrangement of shrubsallow for greater canopy deployment and higher possible values of LAI.Because LAI is one of the parameters that greatly determine CO2 andwater fluxes over ecosystems (Allen et al., 1998; Bonan, 1993), thesedifferences in canopy responses contribute to explain most between-sitedifferences on summer fluxes. Taking both limitations into account,when the shrubland increase ET by more than 5mmmonth−1 above thesavanna, the effect of ET over NEP in the shrubland can surpass theadvantage of higher WUE of the buffelgrass savanna (Fig. 7).

During the seasons of fall, winter and spring, the buffelgrass sa-vanna had higher NEP than the shrubland, mostly related to increasedGPP as a consequence of increased WUE (Figs. 4 and 5). The higher GPPand WUE could be attributed to the physiology of C4 photosynthesiswhen buffelgrass was active (e.g. winter and spring of 2015) in

Table 1Results of the statistical comparison of monthly water use efficiency (WUE), intrinsic water use efficiency (WUEi) and the relationship of Reco/GPP the shrubland andthe buffelgrass savanna site.

R2 Intercept Independent variable effect Site (savanna) effect Independent variable × Site (savanna) effect

WUE (GPP vs ET) 0.90 −5.91 1.95**** 7.28* 0.02WUEi (GPP*VPD vs ET) 0.90 25.86 31.42**** 189.17**** 2.43*Reco vs GPP 0.91 9.60**** 0.65**** 0.41 0.02

*p < 0.1 (marginally significant).**p < 0.05.***p < 0.01.****p < 0.001.When no significant interaction effect is found, the independent variable effect corresponds to the overall slope.When interaction effect is statistically significant, the independent variable effect corresponds to that of shrubland site and the independent variable x Site effectcorrespond to the amount of change in the slope of the savanna in relation to that of the shrubland.The Site (savanna) effect corresponds to the difference of the means of the dependent variable of the savanna in relation to that of the shrubland when the otherfactors are controlled.

Table 2Annual and summer fluxes and seasonal contributions for the study years at shrubland and buffelgrass savanna sites. P values are for between-site t-test comparisonfor the same period.

Annual sums Summer sums Summer contribution

2013* 2014 2015 2013 2014 2015 2013 2014 2015

Net ecosystem production (g C m−2)Shrubland 16 183 193 36 193 139 222 105 72Savanna 75 140 199 44 125 112 59 89 56p 0.029 0.318 0.843 0.772 0.069 0.215Ecosystem respiration (g C m−2)Shrubland 326 561 674 261 448 337 80 80 50Savanna 402 557 763 283 406 368 70 73 48p 0.017 0.938 0.026 0.085 0.005 0.012Gross primary production (g C m−2)Shrubland 343 744 867 298 641 476 87 86 55Savanna 477 697 962 327 531 480 68 76 50p 0.005 0.579 0.127 0.383 0.017 0.870Evapotranspiration (mm)Shrubland 236 400 582 194 324 268 82 81 46Savanna 246 321 519 185 257 240 75 80 46p 0.722 0.044 0.047 0.555 < 0.001 0.095Evapotranspiration / PrecipitationShrubland 0.94 0.97 1.15 0.9 0.84 1.15Savanna 0.97 0.78 1.03 0.85 0.66 1.03

* Data for winter 2013 is missing at the shrubland site. Thus, annual flux sums for 2013 at both sites were based on data from March 6 through the rest of the year.Omitting winter data may cause and error of around ± 7 g C m−2 or less as this is the flux that we have registered during the winters at both sites for 2013 and 2014.

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

275

agreement with our hypothesis that buffelgrass activity would enhanceWUE. Unexpectedly, WUE in the savanna was also higher during fall,winter and spring when buffelgrass was not active (e.g. winter andspring of 2013 and 2014). We suggest that water infiltration into deepersoil layers and subsequent use by native woody species that retain theirleaves during most of the year, such as Prosopis and Olneya, increasedGPP and thus WUE at the savanna. In agreement with this, water ac-cumulation in soil at 1.5–2m depth was found at the same savanna siteafter rainy summer and fall seasons, in which a decrease in soil watercontent at these depths during the drier months until the following

summer gives evidence of its use (Castellanos et al., 2016; Celaya-Michel et al., 2015). Deep soil water accumulation could be expected insites invaded with exotic grasses, as suggested by modeling studies ofwater balance in drylands (Wilcox et al., 2012). This evidence suggeststhat the high WUE and biological limitations at the savanna, becausethe buffelgrass C4 physiology, cause a lower ET/P during summer rainyseasons and years (e.g., 2014 and 2015), and results in the observedwater accumulation that feedbacks to the activity at the savanna duringdry fall, winter and spring seasons (Fig. 8).

5. Concluding remarks

There is a long way ahead the scientific community to understandhow LCLUC impacts NEP when woody shrublands change toC4—grass—dominated ecosystems. The high productivity, photosynth-esis rates, resource-use efficiency, and other traits of C4 exotic grass-es—dominated ecosystems, can make important differences in theirimpact and feedbacks of these transitions from those reported in woodyencroachment studies (Huxman et al., 2005; Petrie et al., 2015; Scottet al., 2014; Wolkovich et al., 2010). By using the EC techniques overadjacent shrubland and buffelgrass savanna sites, we documented thatsuch LCLUC altered the temporal dynamics of carbon fluxes, and wereable to link these differences to changes in water use and WUE. Therepresentation of photosynthetic pathway, rooting pattern, and differ-ential canopy responses were the main factors responsible for suchchanges (Fig. 8), all of these characteristics greatly modified by LCLUCfrom shrublands to grass-dominated ecosystems. However, our findingssuggest that the effects of temporal changes in water use and WUE onNEP compensated in time, either between seasons (e.g., 2014) or be-tween years (e.g., 2013 vs 2014), to a similar NEP over the years despitethe strong biotic, and vegetation-specific structural and functional dif-ferences in land cover. Further studies on LCLUC from woody-domi-nated to grass-dominated ecosystems should focus on understandinghow the impact of biotic traits (photosynthetic pathway, rooting

Fig. 6. a) Peak summer leaf area index response to seasonalprecipitation and b) the effect of leaf area index on net eco-system production. Black symbols are for the shrubland andgray symbols for the buffelgrass savanna. Numbers in a) areyears for each pair of points. Arrow is the six year average ofsummer precipitation in our sites (2009–2015).

Fig. 7. Between-site difference in evapotranspiration as a driver of the differ-ence in net ecosystem production for shrubland buffelgrass savanna sites.Closed circles are values for summer months, and open circles are for monthsfrom the other seasons.

Fig. 8. Summary of seasonal differences in water use, wateruse efficiency and CO2 fluxes between shrubland and exoticbuffelgrass savanna. P: precipitation, ET: evapotranspiration,GPP: gross primary production, Reco: ecosystem respiration,NEP: net ecosystem production. The symbols “-” and “+” in-dicate that savanna has a lower (or higher, respectively)magnitude of that process. Double symbols indicate the pro-cesses that are considered the main drivers of differences inNEP between sites.

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

276

pattern, differential LAI canopy responses) affect water use, WUE andNEP over ecological gradients such as rainfall (amount and season-ality), vegetation (over different kinds of woody dominated ecosys-tems), and management conditions (low and high cover of exoticgrasses, overgrazed, non-overgrazed). Such studies will help to reducethe uncertainty associated with LCLUC between woody- and grass-dominated ecosystems and their impacts on carbon processes.

Acknowledgments

This manuscript was much improved by the insightful suggestionsfrom the editor (C.K. Thomas) and two anonymous reviewers. Wegreatly appreciate support from CONACYT (CB61865, INF2012/1-188387) to AEC and CB223525 to AEC and JRRL and Universidad deSonora. CHH acknowledges CONACYT and Posgrado en Biociencias fora PhD scholarship. RV acknowledges support from NASA CarbonMonitoring Systems (80NSSC18K0173). We thank H. Celaya-Michel, A.Ibarra, J. Llano and J.E. López-Piña for field and lab support. We spe-cially thank and deeply appreciate Mr. L. Sierra for allowing us to in-stall the eddy covariance towers and carry out this study in his ranchproperty, and to J. Dueñas for initial support in the field.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in theonline version, at doi:https://doi.org/10.1016/j.agrformet.2019.01.030.

References

Ahlstrom, A., Raupach, M.R., Schurgers, G., Benjamin, S., Arneth, A., Jung, M.,Reichstein, M., Canadell, J.G., Friedlingstein, P., Jain, A.K., Kato, E., Poulter, B.,Sitch, S., Stocker, B.D., Viovy, N., Wang, Y.P., Wiltshire, A., Zaehle, S., Zeng, N.,2015. The dominant role of semi-arid ecosystem in the trend and variability of theland CO2 sink. Science 80- (348), pp. 895–899.

Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration – Guidelinesfor Computing Crop Water Requirements. FAO Irrigation and Drainage Paper 56.FAO, Rome 300 pp.

Arriaga, L., Castellanos, A.E., Moreno, E., Alarcón, J., 2004. Potential ecological dis-tribution of alien invasive species and risk assessment: a case study for buffel grass inarid regions of Mexico. Conserv. Biol. 18, 1504–1514.

Aubinet, M., Vesala, T., Papale, D., 2012. Eddy Covariance: a Practical Guide toMeasurement and Data Analysis. Springer Science & Business Media.

Baldocchi, D.D., 2008. “Breathing” of the terrestrial biosphere: lessons learned from aglobal network of carbon dioxide flux measurement systems. Aust. J. Bot. 56, 1–26.https://doi.org/10.1071/BT07151.

Barger, N.N., Archer, S.R., Campbell, J.L., Huang, C.Y., Morton, J.A., Knapp, A.K., 2011.Woody plant proliferation in North American drylands: a synthesis of impacts onecosystem carbon balance. J. Geophys. Res. Biogeosci. 116, 1–17. https://doi.org/10.1029/2010JG001506.

Barron-Gafford, G.A., Scott, R.L., Jenerette, G.D., Hamerlynck, E.P., Huxman, T.E., 2012.Temperature and precipitation controls over leaf- and ecosystem-level CO2 flux alonga woody plant encroachment gradient. Glob. Change Biol. 18, 1389–1400. https://doi.org/10.1111/j.1365-2486.2011.02599.x.

Beer, C., Ciais, P., Reichstein, M., Baldocchi, D., Law, B.E., Papale, D., Soussana, J.-F.,Ammann, C., Buchmann, N., Frank, D., Gianelle, D., Janssens, I.A., Knohl, A.,Köstner, B., Moors, E., Roupsard, O., Verbeeck, H., Vesala, T., Williams, C.A.,Wohlfahrt, G., 2009. Temporal and among-site variability of inherent water use ef-ficiency at the ecosystem level. Glob. Biogeochem. Cycles 23, GB2018. https://doi.org/10.1029/2008GB003233.

Biederman, J.A., Scott, R.L., Goulden, M.L., Vargas, R., Litvak, M.E., Kolb, T.E., Yepez,E.A., Oechel, W.C., Blanken, P.D., Bell, T.W., Garatuza-Payan, J., Maurer, G.E., Dore,S., Burns, S.P., 2016. Terrestrial carbon balance in a drier world: the effects of wateravailability in southwestern North America. Glob. Change Biol. 22, 1867–1879.https://doi.org/10.1111/gcb.13222.

Biederman, J.A., Scott, R.L., Bell, T.W., Bowling, D.R., Dore, S., Garatuza-Payán, J., Kolb,T.E., Krishnan, P., Krofcheck, D.J., Litvak, M.E., Maurer, G.E., Meyers, T.P., Oechel,W.C., Papuga, S.A., Ponce-Campos, G.E., Rodriguez, J.C., Smith, W.K., Vargas, R.,Watts, C.J., Yepez, E.A., Goulden, M.L., 2017. CO2 exchange and evapotranspirationacross dryland ecosystems of southwestern North America. Glob. Change Biol. 10,4204–4221.

Bonan, G.B., 1993. Importance of leaf area index and forest type when estimating pho-tosynthesis in boreal forests. Remote Sens. Environ. 43, 303–314.

Bradley, B.A., Houghton, R.A., Mustard, J.F., Hamburg, S.P., 2006. Invasive grass reducesaboveground carbon stocks in shrublands of the Western US. Glob. Change Biol. 12,1815–1822. https://doi.org/10.1111/j.1365-2486.2006.01232.x.

Bravo-Peña, L.C., Castellanos, A.E., 2013. Tendencias del índice de la diferencia nor-malizada de la vegetación (NDVI) en el estado de sonora. Implicaciones potencialessobre el sector pecuario en el contexto del cambio climático. In: Sanchez-Flores, E.,Díaz-Caravantes, R.E. (Eds.), Aplicaciones de Percepción Remota y Análisis Espacialen la Evaluación del Uso del Territorio. Universidad Autónoma de Ciudad Juárez, Cd.Juárez, pp. 245–283.

Bravo-Peña, L.C., Doode-Matsumoto, S., Castellanos-Villegas, A.E., Espejel-Carbajal, I.,2010. Políticas rurales y pérdida de cobertura vegetal. Elementos para reformularinstrumentos de fomento agropecuario relacionados con la apertura de praderasganaderas. Región y Sociedad 22, 3–35.

Brown, D.E., 1994. Biotic Communities: Southwestern United States and NorthwesternMexico. University of Utah Press, Salt Lake City.

Castellanos, A.E., Celaya-Michel, H., Rodríguez, J.C., Wilcox, B.P., 2016. Ecohydrologicalchanges in semiarid ecosystems transformed from shrubland to buffelgrass savanna.Ecohydrology 9, 1663–1674. https://doi.org/10.1002/eco.1756.

Castellanos, A.E., Yanes, G., Valdez-Zamudio, D., 2002. Drought-tolerant exotic buffel-grass and desertification. In: Tellman, B. (Ed.), Weeds Across Borders. Arizona-SonoraDesert Museum, Tucson, pp. 99–112.

Castellanos, A.E., Bravo, L.C., Koch, G.W., Llano, J., López, D., Méndez, R., Rodriguez,J.C., Romo, J.R., Sisk, T.D., Yanes-Arvayo, G., 2010. Impactos ecológicos por el usodel terreno en el funcionamiento de ecosistemas áridos y semiáridos. In: Molina-Freaner, F.E., Van-Devender, T.R. (Eds.), Diversidad Biológica de Sonora. UNAM,Mexico, DF, pp. 157–186.

Celaya-Michel, H., García-Oliva, F., Rodríguez, J.C., Castellanos-Villegas, A.E., 2015.Cambios en el almacenamiento de nitrógeno y agua en el suelo de un matorraldesértico transformado a sabana de buffel (Pennisetum cialiare (L.) Link). TerraLatinoam. 33, 79–93.

Chapin, F.S., Woodwell, G.M., Randerson, J.T., Rastetter, E.B., Lovett, G.M., Baldocchi,D.D., Clark, Da., Harmon, M.E., Schimel, D.S., Valentini, R., Wirth, C., Aber, J.D.,Cole, J.J., Goulden, M.L., Harden, J.W., Heimann, M., Howarth, R.W., Matson, P.A.,McGuire, A.D., Melillo, J.M., Mooney, H.A., Neff, J.C., Houghton, R.A., Pace, M.L.,Ryan, M.G., Running, S.W., Sala, O.E., Schlesinger, W.H., Schulze, E.-D., 2006.Reconciling carbon-cycle concepts, terminology, and methods. Ecosystems 9,1041–1050. https://doi.org/10.1007/s10021-005-0105-7.

Cox, J.R., Martin-R, M.H., Ibarra-F, F.A., Fourie, J.H., Rethman, N.F.G., Wilcox, D.G.,1988. The influence of climate and soils on the distribution of four African grasses. J.Range Manag. 41, 127–139. https://doi.org/10.2307/3898948.

Ehleringer, J.R., Monson, R.K., 1993. Evolutionary and ecological aspects of photo-synthetic pathway variation. Annu. Rev. Ecol. Syst. 24, 411–439.

Emmerich, W.E., 2007. Ecosystem water use efficiency in a semiarid shrubland andgrassland community. Rangel. Ecol. Manag. 60, 464–470.

Forseth, I.N., Ehleringer, J.R., Werk, K.S., Cook, C.S., 1984. Field water relations of so-noran desert annuals. Ecology 65, 1436–1444.

Franklin, K., Molina-Freaner, F., 2010. Consequences of buffelgrass pasture developmentfor primary productivity, perennial plant richness, and vegetation structure in thedrylands of Sonora, Mexico. Conserv. Biol. 24, 1664–1673. https://doi.org/10.1111/j.1523-1739.2010.01540.x.

Franklin, K.A., Lyons, K., Nagler, P.L., Lampkin, D., Glenn, E.P., Molina-Freaner, F.,Markow, T., Huete, A.R., 2006. Buffelgrass (Pennisetum ciliare) land conversion andproductivity in the plains of Sonora, Mexico. Biol. Conserv. 127, 62–71. https://doi.org/10.1016/j.biocon.2005.07.018.

Hayes, D.J., Vargas, R., Alin, S.R., Conant, R.T., Hutyra, L.R., Jacobson, A.R., Kurz, W.A.,Liu, S., McGuire, A.D., Poulter, B., Woodall, C.W., 2018. Chapter 2: the NorthAmerican carbon budget. In: Cavallaro, N., Shrestha, G., Birdsey, R., Mayes, M.A.,Najjar, R.G., Reed, S.C., Romero-Lankao, P., Zhu, Z. (Eds.), Second State of theCarbon Cycle Report (SOCCR2): A Sustained Assessment Report. U.S. Global ChangeResearch Program, Washington, DC, USA, pp. 71–108. https://doi.org/10.7930/SOCCR2.2018.

Hinojo-Hinojo, C., Castellanos, A.E., Rodriguez, J.C., Delgado-Balbuena, J., Romo-León,J.R., Celaya-Michel, H., Huxman, T.E., 2016. Carbon and water fluxes in an exoticbuffelgrass savanna. Rangel. Ecol. Manag. 69, 334–341.

Houghton, R.A., House, J.I., Pongratz, J., van der Werf, G.R., DeFries, R.S., Hansen, M.C.,Le Quéré, C., Ramankutty, N., 2012. Carbon emissions from land use and land-coverchange. Biogeosciences 9, 5125–5142. https://doi.org/10.5194/bg-9-5125-2012.

Huxman, T.E., Wilcox, B.P., Breshears, D.D., Scott, R.L., Snyder, K.A., Small, E.E., Hultine,K., Pockman, W.T., Jackson, R.B., 2005. Ecohydrological implications of woody plantencroachment. Ecology 86, 308–319.

Jackson, R.B., Banner, J.L., Jobbágy, E.G., Pockman, W.T., Wall, D.H., 2002. Ecosystemcarbon loss with woody plant invasion of grasslands. Nature 418, 623–626. https://doi.org/10.1038/nature00952.

King, A.W., Dilling, L., Zimmerman, G.P., Fairman, D.M., Houghton, R.A., Marland, G.,Rose, A.Z., Wilbanks, T.J., 2007. The First State of the Carbon Cycle Report (SOCCR):The North American Carbon Budget and Implications for the Global Carbon Cycle. AReport by the U.S. Climate Change Science Program and the Subcommittee on GlobalChange Research. National Oceanic and Atmospheric Administration. NationalClimatic Data Center, Asheville NC, USA 242 pp.

Knapp, A.K., Briggs, J.M., Collins, S.L., Archer, S.R., Bret-Harte, M.S., Ewers, B.E., Peters,D.P., Young, D.R., Shaver, G.R., Pendall, E., Cleary, M.B., 2008. Shrub encroachmentin North American grasslands: shifts in growth form dominance rapidly alters controlof ecosystem carbon inputs. Glob. Change Biol. 14, 615–623. https://doi.org/10.1111/j.1365-2486.2007.01512.x.

Kurc, S.A., Small, E.E., 2007. Soil moisture variations and ecosystem-scale fluxes of waterand carbon in semiarid grassland and shrubland. Water Resour. Res. 43, 1–13.https://doi.org/10.1029/2006WR005011.

Larcher, W., 2014. Physiological Plant Ecology: Ecophysiology and Stress Physiology ofFunctional Groups, 5th ed. Springer, Germany.

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

277

Le Quéré, C., Andrew, R.M., Canadell, J.G., Sitch, S., Korsbakken, J.I., et al., 2016. Globalcarbon budget 2016. Earth Syst. Sci. Data 8, 605–649. https://doi.org/10.5194/essd-8-605-2016.

Marshall, V.M., Lewis, M.M., Ostendorf, B., 2012. Buffel grass (Cenchrus ciliaris) as aninvader and threat to biodiversity in arid environments: a review. J. Arid Environ. 78,1–12. https://doi.org/10.1016/j.jaridenv.2011.11.005.

Mauder, M., Foken, T., 2004. Documentation and Instruction Manual of the EddyCovariance Software Package TK2. Universität Bayreuth, AbteilungMikrometeorologie, Bayreuth.

Mnif, L., Chaieb, M., 2009. Root growth and morphology of four provenances of a per-ennial grass (Cenchrus ciliaris L.) in rhizotron chamber. Acta Bot. Gall. 156, 273–282.https://doi.org/10.1080/12538078.2009.10516157.

Moncrieff, J.B., Massheder, J.M., de Bruin, H., Elbers, J., Friborg, T., Heusinkveld, B.,Kabat, P., Scott, S., Soegaard, H., Verhoef, A., 1997. A system to measure surfacefluxes of momentum, sensible heat, water vapour and carbon dioxide. J. Hydrol.(Amst) 188–189, 589–611.

Moncrieff, J., Clement, R., Finnigan, J., Meyers, T., 2004. Averaging, detrending, andfiltering of eddy covariance time series. In: Lee, X., Massman, W., Law, B. (Eds.),Handbook of Micrometeorology. Kluwer Academic, Dordrecht, pp. 7–31.

Nakai, T., Shimoyama, K., 2012. Ultrasonic anemometer angle of attack errors underturbulent conditions. Agric. For. Meteorol. 162–163, 14–26. https://doi.org/10.1016/j.agrformet.2012.04.004.

Pacala, S.W., Hurtt, G.C., Baker, D., Peylin, P., Houghton, R.A., Birdsey, R.A., Heath, L.,Sundquist, E.T., Stallard, R.F., Ciais, P., Moorcroft, P., Caspersen, J.P., Shevliakova,E., Moore, B., Kohlmaier, G., Holland, E., Gloor, M., Harmon, M.E., Fan, S.-M.,Sarmiento, J.L., Goodale, C.L., Schimel, D., Field, C.B., 2001. Consistent land- andatmosphere-based U.S. Carbon sink estimates. Science 80- (292), 2316–2320.

Petrie, M.D., Collins, S.L., Swann, A.M., Ford, P.L., Litvak, M.E., 2015. Grassland toshrubland state transitions enhance carbon sequestration in the northern ChihuahuanDesert. Glob. Chang. Biol. 21, 1226–1235. https://doi.org/10.1111/gcb.12743.

Poulter, B., Frank, D., Ciais, P., Myneni, R.B., Andela, N., Bi, J., Broquet, G., Canadell,J.G., Chevallier, F., Liu, Y.Y., Running, S.W., Sitch, S., van der Werf, G.R., 2014.Contribution of semi-arid ecosystems to interannual variability of the global carboncycle. Nature 509, 600–603. https://doi.org/10.1038/nature13376.

Prater, M.R., Obrist, D., Arnone, J.A., DeLucia, E.H., 2006. Net carbon exchange andevapotranspiration in postfire and intact sagebrush communities in the Great Basin.Oecologia 146, 595–607. https://doi.org/10.1007/s00442-005-0231-0.

Reichstein, M., Falge, E., Baldocchi, D., Papale, D., Aubinet, M., Berbigier, P., Bernhofer,C., Buchmann, N., Gilmanov, T., Granier, A., Grunwald, T., Havrankova, K.,Ilvesniemi, H., Janous, D., Knohl, A., Laurila, T., Lohila, A., Loustau, D., Matteucci,G., Meyers, T., Miglietta, F., Ourcival, J.-M., Pumpanen, J., Rambal, S., Rotenberg, E.,Sanz, M., Tenhunen, J., Seufert, G., Vaccari, F., Vesala, T., Yakir, D., Valentini, R.,2005. On the separation of net ecosystem exchange into assimilation and ecosystemrespiration: review and improved algorithm. Glob. Change Biol. 11, 1424–1439.https://doi.org/10.1111/j.1365-2486.2005.001002.x.

Richardson, A.D., Hollinger, D.Y., 2007. A method to estimate the additional uncertaintyin gap-filled NEE resulting from long gaps in the CO2 flux record. Agric. For.Meteorol. 147, 199–208. https://doi.org/10.1016/j.agrformet.2007.06.004.

Safriel, U., Adeel, Z., Niemeijer, D., Puigdefabregas, J., White, R., Lal, R., Winslow, M.,Ziedler, J., Prince, S., Archer, E., King, C., Shapiro, B., Wessels, K., Nielsen, T.,Portnov, B., Reshef, I., Thonell, J., Lachman, E., Mcnab, D., El-kassas, R.E.M.,Ezcurra, E., 2005. Dryland systems. In: In: Hassan, R., Scholes, R., Ash, N. (Eds.),Ecosystems and Human Well-Being: Current State and Trends, vol. 1. Island Press,

Washington, DC, pp. 623–662.Sala, O.E., Maestre, F.T., 2014. Grass-woodland transitions: determinants and con-

sequences for ecosystem functioning and provisioning of services. J. Ecol. 102,1357–1362. https://doi.org/10.1111/1365-2745.12326.

Saucedo-Monarque, E., García-Moya, E., Castellanos-Villegas, A.E., Flores-Flores, J.L.,1997. La riqueza, una variable de respuesta de la vegetación a la introducción delzacate buffel. Agrociencia 31, 83–90.

Scott, R.L., Huxman, T.E., Williams, D.G., Goodrich, D.C., 2006. Ecohydrological impactsof woody plant encroachment: seasonal patterns of water and carbon dioxide ex-change within a semiarid riparian environment. Glob. Change Biol. 12, 311–324.https://doi.org/10.1111/j.1365-2486.2005.01093.x.

Scott, R.L., Huxman, T.E., Barron-Gafford, G.A., Janerette, G.D., Young, J.M.,Hamerlynck, E.P., 2014. When vegetation change alters ecosystem water availability.Glob. Change Biol. 20, 2198–2210. https://doi.org/10.1111/gcb.12511.

Scott, R.L., Biederman, J.A., Hamerlynck, E.P., Barron-Gafford, G.A., 2015. The carbonbalance pivot point of southwestern U.S. Semiarid ecosystems: insights from the 21st

century drought. J. Geophys. Res. Biogeosci. 120, 2612–2624. https://doi.org/10.1002/2015JG003181.

Throop, H.L., Reichmann, L.G., Sala, O.E., Archer, S.R., 2012. Response of dominant grassand shrub species to water manipulation: an ecophysiological basis for shrub invasionin a Chihuahuan Desert Grassland. Oecologia 169, 373–383. https://doi.org/10.1007/s00442-011-2217-4.

Turner, R.M., Bowers, J.E., Burgess, T.L., 1995. Sonoran Desert Plants: An EcologicalAtlas. The University of Arizona Press, Tucson, AZ.

Van Auken, O.W., 2009. Causes and consequences of woody plant encroachment intowestern North American grasslands. J. Environ. Manage. 90, 2931–2942. https://doi.org/10.1016/j.jenvman.2009.04.023.

van Dijk, A., Moene, A.F., de Bruin, H.A.R., 2004. The Principles of Surface Flux Physics:Theory, Practice and Description of the ECPACK Library, Internal Report 2004/1.Wageningen, The Nederlands. .

Vargas, R., Yépez, E.A., Andrade, J.L., Ángeles, G., Arredondo, T., Castellanos, A.E.,Delgado-Balbuena, J., Garatuza-Payán, J., González del Castillo, E., Oechel, W.,Rodríguez, J.C., Sánchez-Azofeifa, A., Velasco, E., Vivoni, E.R., Watts, C., 2013.Progress and opportunities for monitoring greenhouse gases fluxes in Mexican eco-systems: the MexFlux network. Atmsfera 26, 325–336. https://doi.org/10.1016/S0187-6236(13)71079-8.

Vickers, D., Mahrt, L., 1997. Quality control and flux sampling problems for tower andaircraft data. J. Atmos. Ocean. Technol. 14, 512–526.

Webb, E.K., Pearman, G.I., Leuning, R., 1980. Correction of flux measurements for densityeffects due to heat and water vapour transfer. Q. J. R. Meteorol. Soc. 106, 85–100.

Wilcox, B.P., Turnbull, L., Young, M.H., Williams, C.J., Ravi, S., Seyfried, M.S., Bowling,D.R., Scott, R.L., Germino, M.J., Caldwell, T.G., Wainwright, J., 2012. Invasion ofshrublands by exotic grasses: ecohydrological consequences in cold versus warmdeserts. Ecohydrology 5, 160–173. https://doi.org/10.1002/eco.

Wilczak, J.M., Oncley, S.P., Stage, S.A., 2001. Sonic anemometer tilt correction algo-rithms. Boundary-Layer Meteorol. 99, 127–150.

Williams, D.G., Baruch, Z., 2000. African grass invasion in the Americas: ecosystemconsequences and the role of ecophysiology. Biol. Invasions 2, 123–140.

Wolkovich, E.M., Lipson, D.A., Virginia, R.A., Cottingham, K.L., Bolger, D.T., 2010. Grassinvasion causes rapid increases in ecosystem carbon and nitrogen storage in asemiarid shrubland. Glob. Change Biol. 16, 1351–1365. https://doi.org/10.1111/j.1365-2486.2009.02001.x.

C. Hinojo-Hinojo et al. Agricultural and Forest Meteorology 268 (2019) 269–278

278