Woody Plant Encroachment and Ecosystem Biogeochemistry

Changes in soils and microclimate accompanying long-term heavy grazing may have shifted the balance in favor of N2-fixing or evergreen woody plants which are better adapted than grasses to nutrient-poor soils and warmer, drier microenvironments. The establishment of trees and shrubs would have been further augmented by grazing-induced reductions in herbaceous competition and fire (Archer, 1995a). In addition, the woody plants at La Copita are highly unpalatable, and browsing by wildlife or cattle is minimal. However, fruits of the dominant tree invader (Prosopis glandulosa) are readily consumed by livestock, which disperse large numbers of viable seeds into grasslands (Brown and Archer, 1987). Thus, heavy, continuous, and preferential grazing of grasses by livestock has promoted woody plant encroachment via numerous direct and indirect effects (Archer, 1994). As woody communities develop in grazed grasslands, plant and soil C and N pool sizes and flux rates change as described in the following sections.

3.3.1 Plant Carbon Pools

Quantitative changes in woody plant cover at La Copita are depicted in Fig. 4 and 5. To ascertain the effects of these vegetation changes on plant carbon stocks, we linked CENTURY with a plant succession model developed for La Copita (Scanlan and Archer, 1991). We initiated woody plant encroachment in the late 1800s on a heavily grazed, fire-free landscape in which SOC content had been reduced by grazing (Fig. 3). The landscape, consisting of a

65 r

—B— Landscape 1 -#■- Landscape 2 ...q Landscape 3

FIGURE 4 Changes in total woody plant cover on three replicated landscapes at the La Copita site in southern Texas. See Figure 5 for spatial pattern of change in various patch types on Landscape 1 (Archer and Boutton, unpublished).

—B— Landscape 1 -#■- Landscape 2 ...q Landscape 3

FIGURE 4 Changes in total woody plant cover on three replicated landscapes at the La Copita site in southern Texas. See Figure 5 for spatial pattern of change in various patch types on Landscape 1 (Archer and Boutton, unpublished).

1976

FIGURE 5 Landscape-scale changes in herbaceous and woody plant community cover from 1950 to 1990 in upland (herbaceous, discrete cluster, and grove) and lowland (woodland) plant communities at the La Copita site in southern Texas (Archer and Boutton, unpublished). Values given below dates are total hectares of woody cover (cluster + grove + woodland) for the 11.06 ha "pixel." See Figure 4 (Landscape 1) for changes in percentage woody cover.

1976

FIGURE 5 Landscape-scale changes in herbaceous and woody plant community cover from 1950 to 1990 in upland (herbaceous, discrete cluster, and grove) and lowland (woodland) plant communities at the La Copita site in southern Texas (Archer and Boutton, unpublished). Values given below dates are total hectares of woody cover (cluster + grove + woodland) for the 11.06 ha "pixel." See Figure 4 (Landscape 1) for changes in percentage woody cover.

sandy loam upland and a clay loam intermittent drainage, was populated with woody plants based on rainfall-driven transition probabilities. Grassland and forest CENTURY subroutines were applied, as appropriate, to compute plant carbon and SOC (0-20 cm) stocks in wooded and herbaceous landscape elements. At each time step, plant carbon and SOC were estimated by summing across the entire landscape (upland plus lowland vegetation patch types). Results suggest that the development of the present-day savanna parkland-woodland complex has increased plant carbon stocks 10-fold over that which would be present had the "pristine" grassland vegetation been maintained on the site (Fig. 6). Part of that increase is attributable to an increase in aboveground net pri mary productivity (Table 2) and part of it represents the decline in tissue turnover which occurs when herbaceous vegetation is replaced by woody vegetation. These results are conservative in that CENTURY simulations include root C mass only in the top 20 cm of soil. Biomass distributions of woody plant roots at La Copita (Watts, 1993; Boutton et al, 1998; 1999; Midwood etal, 1998; Gill and Burke, 1999) are typical of those of other dryland tree/shrub systems (Jackson et al, 1996; Canadell et al, 1996) where, relative to grasslands, there is substantially greater mass at deeper depths where turnover and decomposition are likely to be reduced. The fact that fluctuations in monthly woody plant root biomass in upper soil horizons exceeded monthly foliar litter inputs by one to

TABLE 2 Contrasts in Aboveground Net Primary Production (ANPP), Soil Physical Properties, Organic Carbon and Total Nitrogen Pools (0- to 10-cm depth), and Fluxes in Soils Associated with Woody Plant and Grazed Grassland Communities in a Sandy Loam Upland Landscape at the La Copita Research Area in Southern Texas

Community Type

Parameter

Herbaceous

TABLE 2 Contrasts in Aboveground Net Primary Production (ANPP), Soil Physical Properties, Organic Carbon and Total Nitrogen Pools (0- to 10-cm depth), and Fluxes in Soils Associated with Woody Plant and Grazed Grassland Communities in a Sandy Loam Upland Landscape at the La Copita Research Area in Southern Texas

Parameter

Herbaceous

Woody Plant

ANPP (Mgha-1 year-1)"

1.9

-3.4

5.1

- 6.0

Bulk density (g cm" !)J

1.4

± 0.01

1.1

± 0.04

% Clay'

20

± 0.7

20

± 1

Fine roots (g m ~:)J

100

-175

400-

- 700

Coarse roots (g m-2)'1

100

- 400

400-

- 1,100

Organic C'1, %

0.84

- 0.05

2.2

± 0.23

gni-2

1165

± 67

2352

± 276

Potential C mineralization

(mg C kg"1 soil day-1)1'

7.3

± 5.7

15.5

± 6.8

Soil respiration (mg CO; m-2 year-1)"

611

± 83

730

± 67

Qu) values for in situ soil respiration1.

1.2

1.4-

- 2.7

Total N'1, %

0.07

± 0.00

0.18

± 0.02

gm-2

91

± 6

192

± 20

N mineralization'1, g N m "2 year-1

6

± 1

22

± 2

MgNg-1 year-1

42

± 5

200

± 18

NO flux (ng NO-N cm-2 h-!)''

Dry soil

0.2

± 0.07

2.8

± 0.25

Wet soil

1.1

± 0.11

16.2

"Pristine" Grassland

- Heavily Grazed Grassland

I | Grass + Woody Patches

Woody Plant

Note. Maximum and minimum monthly values for samples obtained over an annual cycle are shown for root-standing crop (coarse roots >0.1-mm diameter); a range is presented for ANPP and Q,0. All other values are means ± SE.

' Hibbard, 1995. b McCulley, 1998. 'Cole et al. 1996.

two orders of magnitude (Table 2) suggests that belowground inputs of organic matter drive changes in soil physical and chemical properties subsequent to woody plant establishment in grasslands. These substantial fluctuations in woody plant root biomass suggest a high turnover, which is consistent with detailed observations on woody plants in other systems (Eissenstat and Yanai, 1997; Hendricks et al., 1997). In addition, turnover of grass roots may be slower than has been generally assumed (Milchunas et al., 1992). Thus, increases in aboveground and belowground net primary productivities may accompany woody plant encroachment into grasslands and foster C and N accumulation.

As an independent test of the reconstruction in Figure 6, we quantified aboveground plant carbon density in patches representing the dominant community types at La Copita. This was accomplished using allometric relationships and belt-transect surveys (Northup et al., 1996). Plant carbon density was then multiplied by community area measured on aerial photographs (1950, 1976, 1990) to obtain a community-level estimate. Estimates for each community type were then summed to obtain landscape-scale estimates. For patches representing various woody and herbaceous community types, CENTURY estimates of above-ground carbon density were lower than field-based estimates (Table 3), further suggesting that model estimates were conservative. Aboveground carbon density differed substantially among

1900 1910 1920 1930 1940 1950 1960 1970 1980 1993 Year

FIGURE 6 Modeled changes in whole-landscape (upland + lowland and all patch types therein) plant carbon density (aboveground + roots to 20 cm) accompanying succession from grassland to savanna parkland/ woodland (from Hibbard, 1995). Dashed line depicts steady-state SOC expected for a lightly grazed grassland landscape (upland + lowland communities pooled) with fire at 10-year intervals and no woody plants; solid line depicts steady-state plant carbon density for heavily, continuously grazed grassland landscape with no fire and no woody plants. Changes in woody plant abundance on each soil type were directed by a succession model (Scanlan and Archer, 1991); subsequent changes in plant carbon stocks were then assessed with a biogeochemistry model (CENTURY; Par-ton et til., 1994). See Figure 7 for validation results.

the three landscapes inventoried in 1950 (Fig. 7), primarily reflecting differences in woody plant cover on this date (Fig. 4). By 1990, woody cover and carbon density were comparable on the three landscapes. The CENTURY-succession model estimates of aboveground carbon density for an "average" landscape in 1950, 1976, and 1990 closely approximated those obtained from the field-historical aerial photo approach.

3.3.2. Nonmethane Hydrocarbon Fluxes

On a regional basis, shifts from grass to woody plant domination have the potential to influence biophysical aspects of land-atmosphere interactions, such as albedo, évapotranspiration, boundary layer conditions, and dust loading (e.g., Bryant et al.,

TABLE 3 Observed and Predicted Aboveground Carbon Density in Patches Representing Tree/Shrub and Grassland Communities at La Copita, Texas

Carbon Density (kg iyT

TABLE 3 Observed and Predicted Aboveground Carbon Density in Patches Representing Tree/Shrub and Grassland Communities at La Copita, Texas

Carbon Density (kg iyT

Topoedaphic Setting

Patch Type

Field Estimate

Model Estimate

Sandy loam upland

Cluster

2.9 ï 0.4

2.2

Grove

6.3 ± 0.8

4.0

Grassland'1

0.05 ± 0.00

0.04

Clay loam lowland

Woodland

5.8 ± 0.8

4.5

Note. Observed data (means ± SE) are based on belt transects and plant size-biomass relationships for woody communities (Archer and Boutton, unpublished) and on clipped plots in grasslands (Archer, unpublished). Predicted values are CENTURY estimates for 100-year-old patches (Hibbard, 1995).

'Peak aboveground biomass.

Landscape: I II III I II III I II III

1950 1976 1990

FIGURE 7 Changes in aboveground plant carbon density on three landscapes at the La Copita from 1950 to 1990 (Archer and Boutton, unpublished). Patch/soil-specific field estimates of plant carbon density (Northup et al., 1996, McMurtry, Nelson and Archer, unpublished) were multiplied by patch area as measured in aerial photographs to generate whole-landscape estimates. Dashed lines denote predictions from linked CENTURY-succession model (Hibbard, 1995). See Figure 4 for changes in woody cover on the three landscapes.

1990; Pilke and Avissar, 1990; Graetz, 1991). These changes in vegetation may also influence atmospheric oxidizing capacity, aerosol burden, and radiative properties by affecting emissions of non-methane hydrocarbons (NMFlCs) such as terpenes, isoprene, and other aromatics (Fehsenfeld et al., 1992). There are many sources of atmospheric NMHCs, but > 90% of the global annual emission is from vegetation (Guenther et al., 1995). NMHC emissions are therefore highly dependent on species composition as constrained by environmental conditions which influence plant physiology and production. The high temperatures and solar radiation fluxes associated with subtropical and tropical grasslands and savannas make these geographically extensive bioclimatic regions large potential sources of biogenic NMHC emissions. However, grasses are typically low emitters of NMHCs, whereas emissions from trees and shrubs in forest systems are highly variable, with some species being low emitters and other species being high emitters.

We hypothesized that foliar emissions of NMHCs in woody plants would be positively correlated with leaf longevity and inversely related to photosynthetic capacity. Plants characterized by low photosynthetic capacities and slow growth rates (e.g., evergreens) depend on extended leaf longevities to achieve a positive carbon balance. Preferential allocation to secondary compounds such as terpenes would help ensure foliage longevity by reducing levels of herbivory. Species with low photosynthetic capacities and high levels of secondary compounds should also dominate under-story environments where low light levels preclude high growth rates and where plants are more accessible to browsers. In contrast, species selected for competitive ability would have high photosynthetic rates, high growth rates, and high rates of tissue turnover (e.g., deciduous shrubs). Allocation to secondary compounds that deter herbivory would be of lower priority since leaf longevity is less critical to realizing a positive return on foliar investments. Such plants would be expected to preferentially allocate resources such as nitrogen to the carboxylating enzyme and productive tissues rather than to structural tissues or secondary compounds such as terpenes, and would thus be low NMHC emitters (or isoprene emitters, since isoprene is not known to be associated with defense; Coley et al., 1985).

To test these hypotheses, we screened plant species representing the major growth forms at La Copita for NMHC emissions. As expected, grasses had low NMHC emission rates and several common woody species had high emission rates (Guenther et al., 1999). Flowever, there was little evidence of emissions being consistently related to woody plant taxonomy, growth form, or functional group. As a result, generalizations regarding NMHC emissions spectra for tree/shrub species assemblages in other systems do not appear feasible.

To determine if biogenic NMHC emissions have been altered as a result of the change in land cover from grass to woody plant domination at La Copita, a vegetation change model (Scanlan and Archer, 1991) was then linked with a model which predicted NMHC emissions as a function of foliar density, leaf temperature, and photosynthetic photon flux density as modulated by ambient temperature, cloud cover, precipitation, relative humidity, and wind speed (Guenther et al., 1995; Guenther, 1997).

Linkage of the biogenic emissions model with the plant succession model indicated that land cover change since the early 1800s has elicited a threefold increase in isoprene emissions (Fig. 8). This increase reflected changes in vegetation composition and increases in foliar density. Model predictions of current NMHC emissions were within 20% of those measured by a tower flux system. Detailed field measurements on two common shrub species indicated that isoprene emission increased exponentially with increases in leaf temperature from 20 to 40°C and were not suppressed by drought stress. Accordingly, the model predicted that under a projected 2X-C02 climate, present-day biogenic NMHC emissions would double.

These estimates of changes in NMHC emissions associated with the conversion of grassland to woodland are in accordance with estimates in other ecosystems. For example, Klinger et al. (1998) documented a fourfold increase in total terpenoid emissions per unit foliar mass along a savanna to woodland transect in Central Africa. These changes in NMHC emissions associated with vegetation change in subtropical Texas and tropical Africa also mirror those reported for temperate forest (Martin and Guenther, 1995). Together, these results indicate the magnitude of change in NMFIC emissions that could occur when climate and vegetation composition are altered. The importance of these increases in NMHC emissions is magnified at La Copita, as they occur in conjunction with elevated nitric oxide (NO) emissions from shrub-modified soils (Table 2; see Sec. 3.3.4 for elaboration).

Why are vegetation-induced increases in NMHC of concern? Biogenic hydrocarbons play an important role in generating pollutants such as 0(, CO, and organic peroxides, while influencing hydroxyl radical (OH-) chemistry to reduce atmospheric oxida-

Historic (1800's) Grassland / Savanna

Isoprene Flux [mg C nr2 tr1]

modeled = 0.72 measured = 0.54

Present Day Savanna Parkland / Thorn Woodland

Land Cover Change t Grazing 4. Browsing 4 Fire

modeled measured

modeled = 0.72 measured = 0.54

modeled measured

FIGURE 8 Changes in nonmethane hydrocarbon (isoprene) emissions predicted to accompany a shift from savanna grassland to a savanna woodland at the La Copita site in southern Texas (based on Guenther et al., 1999). Predictions from a coupled succession-NMHC emission model are compared with values measured from flux towers. The "measured" values shown for the historic landscape are from a tower located in a savanna grassland landscape with low woody cover.

tion capacity and increase the residence time of greenhouse gases. It has been estimated that to meet current air quality standards for tropospheric ozone, anthropogenic hydrocarbon emissions would have to be reduced by only 30% in the absence of natural isoprene emissions, but by 70% in the presence of them (Monson et al., 1991). Changes in NMHC-NO emissions associated with regional conversion of grassland to shrubland may therefore constitute a "moving baseline" from which to gauge tropospheric ozone production triggered by emissions from automobiles or industrial sources.

3.3.3 Soil C and N Pools

Once established, woody plants alter soils and microclimate in their immediate vicinity to affect both pool sizes and flux rates of nutrients. The result is the formation of "islands of fertility," a phenomenon which has been widely quantified in drylands (see Charley and West, 1975; Schlesinger et al., 1990; Scholes and Archer, 1997; special issue of Biogeochemistry 42 (1/2) 1998). Three general mechanisms have been proposed to account for this (e.g., Virginia, 1986): (1) woody plants act as nutrient pumps, drawing nutrients from deep soil horizons and laterally from areas beyond the canopy, depositing them beneath the canopy via stem flow, litterfall, and canopy leaching; (2) tall, aerodynamically rough woody plant canopies trap nutrient-laclen atmospheric dust that rain washes off the leaves and into the subcanopy soil; and (3) woody plants may serve as focal points attracting roosting birds, insects, and mammals seeking food, shade, or cover. These animals may enrich the soil via defecation and burrowing. For these reasons, soil carbon and nitrogen pools should increase subsequent to woody plant colonization in grazed grasslands.

At La Copita, surficial (0-10 cm) soils associated with woody plants known to have encroached over the past century have a lower bulk density, contain more root biomass, have higher concentrations of SOC and total N, and have greater rates of respiration and N mineralization than soils associated with the remaining grazed grassland communities (Table 2). As the continuity of woody plant cover increases through time, the landscape-scale soil nutrient pools and fluxes would be expected to increase and become more homogeneously distributed. Accordingly, the linked CENTURY—succession model exercise (see Section 3.3.1) predicted that by 1950, landscape-scale SOC had returned to levels which would have occurred had the "pristine" grasslands been maintained on the site (Fig. 9). By the early 1990s, landscape-scale SOC levels were about 10% higher than those expected for the "pristine" grassland, and about 30% higher than those for a heavily grazed grassland not experiencing woody plant encroachment. Forward model projections suggest SOC aggradation will continue for several hundred years, reaching equilibrium levels three times those of the present-day grazed grassland communities.

While the "island of fertility" phenomenon has been widely recognized, little is known of the rates of nutrient enrichment in tree-dominated patches. Total C and N in soil under Acacia Senegal and Balanites aegyptiaca tree canopies were positively correlated with tree girth (r2 = 0.62 and 0.71, respectively; Bernhard-Reversat, 1982), indicating net accumulation with time of woody plant occupancy of a patch. In temperate old fields undergoing forest succession, carbon storage increased 40% in plant + soil pools over 40 years (Johnston et al, 1996). At La Copita, soil C and N were quantified under Prosopis glandulosa trees whose age was determined by annual ring counts. Soil organic carbon storage (top 20 cm of soil) increased linearly with tree stem age at

"Pristine" grassland Heavily grazed grassland Grass + woody patches

•at,—

=->

tMfl

JSs

8 ,

*

1900 1910 1920 1930 1940 1950 1960 1970 Year

1900 1910 1920 1930 1940 1950 1960 1970 Year y

1980

1993

FIGURE 9 Changes in soil organic carbon (SOC; 0-20 cm) predicted to accompany woody plant encroachment into a grazed landscape consisting of a sandy loam uplands and clay loam intermittent drainages at the La Copita site (from Hibbard, 1995). Dashed line depicts steady-state SOC expected for a lightly grazed grassland landscape (upland + lowland communities pooled) with fire at 10-year intervals and no woody plants; solid line depicts steady-state SOC for heavily, continuously grazed grassland landscape with no fire and no woody plants (see Fig. 3). Bars denote SOC summed across the entire landscape and include both grassland and woody plant communities. Note that by 1950, SOC levels had increased to a level comparable to that of the "pristine" grassland (cross-hatched bar). Changes in woody plant abundance on each soil type were directed by a succession model (Scanlan and Archer, 1991); subsequent changes in soil carbon were then assessed with a biogeochemistry model (CENTURY; Parton et <//., 1994).

rates ranging from 11.8 to 21.5 g C m 2 year-1 in sandy loam uplands woody patch types to 47.2 g C m-2 year-1 in moister, clay loam woodland patches (Table 4). Rates of total N accumulation (top 20 cm of soil) ranged from 1.9 to 2.7 g N m-2 year-1 in sandy uplands and averaged 4.6 g N m-2 year-1 in clay loam lowlands. However, woody plant age explained only 21-68% of the variation in soil C and N sequestration rates. These low r2 values may indicate that tree stem ages do not accurately reflect plant ages, possibly due to past disturbance and subsequent vegetative regeneration of woody cover. Low r2's may also indicate that factors un related to time of tree occupancy influence soil C and N under tree canopies. Such factors may include small-scale heterogeneity associated with large mammal or bird defecation, soil mixing by small mammals and arthropods, or patch-specific differences in the species composition, productivity, and rate of development of the understory community.

Modeling experiments allowed us to control for factors that might cause variation in field-based estimates of woody plant age-SOC relationships. Model estimates of SOC accumulation were comparable to field estimates for upland patch types and substantially lower than field estimates for lowland patch types (Table 4). Model estimates of soil N accumulation were substantially lower than field estimates, especially in lowlands. Given that woody patch age explained only 26-68% of the variance in soil C and N content, our field estimates of accumulation rates cannot be taken as definitive. Model results underestimated field observations, especially for N. Reliability of model estimates of soil carbon could likely be improved with a better understanding of how turnover of the substantial root mass (Table 2) might differ among patch types. Model estimates of soil N are likely constrained by lack of information on inputs associated with N2 fixation, atmospheric N deposition, translocation between uplands and lowlands, and root turnover.

3.3.4 Soil C and N Dynamics

Increases in the C and N pools of soils associated with woody plant communities developing on grazed grasslands at La Copita have been accompanied by increases in soil respiration, N mineralization, and nitric oxide (NO) emissions (Table 2). The increase in NO fluxes accompanying expansion of woody plants into grasslands at La Copita is noteworthy. Nitric oxide plays several critical roles in atmospheric chemistry by contributing to acid rain and by catalyzing the formation of photochemical smog and tropospheric ozone. The latter is potentially accentuated in the La Copita setting, since NO and hydrocarbon emissions (see Sec. 3.3.2) are concomitantly elevated subsequent to woody plant establishment.

The quality and quantity of organic matter inputs interact to drive soil metabolic activity (Zak et al., 1994). Hence, annual soil

TABLE 4 Estimated Rates of Organic Carbon and Total Nitrogen Accumulation in Soils (0- to 20-cm Depth) Developing Beneath Woody Plants Establishing on a Former Grassland

Location

Soil Texture

Patch Type

g C m"

"2 year 1

g N m"

~2 year

Field

Modeled

Field

Modeled

Upland

Sandy loam

Shrub Cluster

21.5

18.5

2.67

1.15

(r2 =

0.26)

(r2 =

0.45)

Grove

11.8

10.5

1.90

0.87

(r =

0.21)

(r -

0.51)

Lowland

Clay loam

Woodland

47.2

13.1

4.64

0.58

(r =

0.57)

(r =

0.68)

Note. Field data are from linear correlations between patch age (determined by dendrochronology) and soil C and N mass (Boutton and Archer, unpublished). Model estimates are from CENTURY simulations (Hibbard, 1995). Descriptions of contrasting woody patch types can be found in Archer (1995b).

Note. Field data are from linear correlations between patch age (determined by dendrochronology) and soil C and N mass (Boutton and Archer, unpublished). Model estimates are from CENTURY simulations (Hibbard, 1995). Descriptions of contrasting woody patch types can be found in Archer (1995b).

respiration rates are positively correlated with net primary productivity (Raich and Schlesinger, 1992). The elevated carbon fluxes observed with the development of woody communities in semiarid La Copita grasslands may reflect increased root (Table 2) and leaf biomass inputs and enhancement of soil moisture beneath woody plant canopies (via concentration of rainfall from stem flow, hydraulic lift, and/or reduced evaporation). Together, these biotic and abiotic factors may interact to stimulate microbial activity relative to that in grass-dominated soils. In fact, microbial biomass in woody communities is comparable to or higher than that in grassland communities at La Copita (McCulley, 1998). However, experimental irrigation, which alleviated plant water stress, enhanced photosynthesis (McMurtry, 1997), and increased soil respiration, elicited a decrease in soil microbial biomass. This suggests the elevated soil respiration observed in woody plant communities at La Copita may be a consequence more of changes in root biomass (Table 2) and respiration than of changes in microbial biomass and activity.

To estimate landscape-scale changes in soil CO, flux, we multiplied patch/soil-specific estimates of annual soil respiration (McCulley, 1998) by patch area. We then computed changes in patch area with a succession model (Scanlan and Archer, 1991). Landscape-scale soil respiration (kg C ha-1 year-1) is projected to have increased from 6687 (200 YBP) to 7377 (1990s) to 7602 (200 years in future)(Table 5). This represents a 10.3% increase with the transition from historic grassland savanna to the present-day savanna parkland-thorn woodland complex, with an additional 3% increase occurring if the present savanna parkland progresses to woodland. If mean annual temperatures increase as projected in general circulation models, further increases in soil respiration would be expected (all other factors being equal). Indeed, Q10 values for soil respiration in woody plant communities (1.4, 2.7, and 2.3 in cluster, grove, and woodland types, respectively) exceed those of grazed grasslands (1.2) at La Copita (McCulley, 1998). This suggests that if future temperature changes occur, the importance of recent and projected future vegetation changes on soil respiration will be further magnified. For example, the magnitude of increase in soil respiration from past grassland savanna with MAT of 22.4°C to future woodland with MAT of 28.4°C would be 22.5% (= 6687-8197 kg C ha-1 year-1; based on Raich and Schlesinger, 1992) to 99.3% (= 6687-13,328 kg C ha-1 year-1; based on McCulley, 1998)(Table 5). Potential changes in the amount, seasonality, and effectiveness of rainfall would have important, but as yet unknown, effects on these projections.

3.3.5 Soils as Sources and Sinks

Elevated fluxes of C and N from plants and soils following grass-land-to-woodland conversion at La Copita suggest a potential for augmenting greenhouse gas accumulation and altering tropos-pheric chemistry, particularly if woody plant encroachment has been geographically widespread (as suggested in Table 1). However, as noted in Sec. 3.3.3, organic C and N have accumulated in soils of developing woody plant communities at La Copita, despite elevated fluxes and higher turnover rates. This indicates that inputs have exceeded outputs and that soils and vegetation at La

TABLE 5 Projected Landscape-Scale Changes in Annual Soil Respiration (SR; kg C ha-1 year ') Accompanying Succession from an Open Savanna/Grassland to Woodland and Potential Changes in Mean Annual Temperature

Landscape-Scale Soil Respiration (kg C ha-1 year-1)

Future Thorn

Mean Annual Past Grassland Present Savanna Woodland

Temperature (°C) (200 YBP)'1 Parkland/Woodland Complex (200 YAP)1,

A. Based on MAT/MAP regression in Raich and Schlesinger (1992)

22.4 6687 7377 7602

25.4 6948 7666 7899

28.4 7209 7954 8197

25.4 8083 9938 10,465

28.4 9480 12,499 13,328

Note. Patch- (grass and various woody communities) and soil-specific SR rates measured monthly over an annual cycle at La Copita ( McCulley, 1998) were multiplied by the area of respective community types (Scanlan and Archer, 1991 j. Effects of mean annual temperature change (MAT, °C) on SR were estimated from (A) equations in Raich and Schlesinger (1992); for La Copita (MAT — 22.4°C and MAP — 720 mm) a 3 and 6°C increase in MAT would produce a 3.9 and 7.8% increase, respectively, in soil respiration; and (B) Ql0 values of in situ, community-specific soil respiration from McCulley (1998). Estimates are probably conservative, as respiration rates used in computations were measured during a below-normal rainfall year. ''YBP, years before present. h YAP, years after present.

Copita have been functioning as C and N sinks over the past century. A variety of factors might interact to account for the observed increases in soil C and N pools:

• The trees and shrubs which have displaced grasses may be more productive abovcground and bclowground and hence deliver more organic matter into soils (see root biomass and ANPP in Table 2).

• 1,eaves of leguminous and nonleguminous woody plants at La Copita have higher [N] than grasses (2-4% vs < 1%; Archer, unpublished). However, woody plants in these landscapes arc seldom browsed by livestock or wildlife, suggesting high concentrations of secondary compounds. This could result in a significant litter quality X quantity interaction, whereby

•• a large fraction of the foliar biomass produced by trees and shrubs goes into the soil pool and directly as litter rather than through the herbivory pathway, and • • a larger fraction of foliar biomass inputs from woody plants may be resistant to decomposition.

• Woody litter inputs and the coarser, more lignified roots of shrubs would promote C and N accumulation compared to that of grass roots and shoots.

• Shading by tree/shrub canopies reduces soil temperatures relative to those in grassland (Archer, 1995b), thus constraining potential mineralization (Q,0 effect).

• Nitrogen accumulation is potentially a consequence of N2 fixation by leguminous shrubs common to the site (P. glan-dulosa and several Acacia spp.) and/or the uptake and lateral translocation of N from grassland patches. While nodula-tion has been induced in controlled environments and observed under field conditions at the La Copita site (Zitzer et al, 1996) and elsewhere (Virginia et ai, 1986; Johnson and Mayeux, 1990), methodological constraints have prevented quantification of N2 fixation (Handley and Scrim-geour, 1997; Liao et al., 1999). Root distribution studies (Watts, 1993) discount the lateral foraging hypothesis.

• La Copita is within ca. 70 km of a major oil refinery center (Corpus Christi, TX) and atmospheric N deposition has likely been significant over the past 50-75 years (e.g., Holland et al., 1999). Increased N availability may have promoted woody plant expansion (e.g., Kochy, 1999) by alleviating grass-woody plant competition for soil N and by promoting growth of woody plants more than that of grazed grasses. This, in turn, may have translated into greater organic C and N inputs into soils associated with woody plants.

3.3.6 An Uncertain Future

Prosopis glandalosa currently dominates the overstory in upland and lowland woody plant communities. Depending on patch type, it constitutes 40-90% of the above-ground biomass (Archer and Boutton, unpublished) and 30-70% of the coarse root (>l-mm diameter) biomass (Watts, 1993). As such, the dynamics of P. glan dalosa must be a primary driver of changes in plant and soil C and N stocks at La Copita. Future increases in landscape nutrient pools and fluxes will reflect a combination of (a) continued growth of P. glandulosa and associated shrubs in existing woody plant communities and (b) expansion of woody plants into the remaining grasslands.

How likely is continued expansion? That may depend on land management practices. Relaxation of grazing pressure could enable grass biomass to accumulate and fire (prescribed or natural) to occur. Together, these could retard expansion and growth of woody plants. However, the La Copita appears to have crossed a threshold, whereby soils, seed banks, and vegetative regenerative characteristics are such that reductions in grazing pressure may be of little consequence (Archer, 1996). Relaxation of grazing would influence woody plant establishment in grassland primarily through its influences on the fire regime (Archer, 1995a; Brown and Archer, 1999). However, the remaining herbaceous clearings are small and discontinuously distributed. Hence, even if fine fuels were to accumulate, fires would be highly localized. Such fires might prevent future encroachment into remaining grassland clearings but would not likely convert woody plant communities to grassland, since the trees and shrubs at La Copita quickly regenerate by sprouting after disturbance (Scanlan, 1988; Flinn et al, 1992). Expenses for clearing woody vegetation via mechanical or chemical treatments are prohibitive and generally not cost-effective, especially since the effects of the treatments are relatively short-lived. Thus, the likelihood of continued woody plant dominance is high, even with aggressive land management practices which might favor grasses.

The succession model which simulates the expansion of woody plants into remaining grasslands (Scanlan and Archer, 1991) projects that with heavy grazing and no fire, woody cover will continue to increase until the landscape goes to nearly complete canopy closure. This assumption has been substantiated by field data which indicate extension of lateral roots beyond woody canopies is minimal (Watts, 1993). Hence, there is little opportunity for between-cluster root competition and density-dependent regulation. As a result, tree/shrub densities may continue to increase until all available herbaceous clearings have been occupied and canopy cover is nearly continuous. Accordingly, woody patches on contrasting upland soils and woody patches on uplands that border woody communities of lowlands have grown and coalesced from the 1940s through the 1990s (Archer et al, 1988; Stokes, 1999). However, recent studies suggest that La Copita landscapes may be reaching their carrying capacity for woody plants, owing to topoedaphic constraints (Stroh, 1995; Stokes, 1999). If this is the case, future changes in C and N pools will occur only with growth of plants in existing woody communities. Only time will tell if this is indeed the case.

As the current population of the dominant P. glandulosa ages, growth and biomass accumulation rates should slow, unless other woody species compensate. The understory shrubs that colonize beneath the Prosopis canopy subsequent to its establishment in grasslands slow Prosopis growth and seed production, hasten its mortality (Barnes and Archer, 1998) and prevent its reestablishment (Archer, 1995b). Thus, it appears that P. glandulosa will not be a component of future woodlands on this landscape. Assessments to date suggest that over the short-term, loss of Prosopis will not adversely affect understory shrub productivity or soil C and N pools (Hibbard, 1995; Barnes and Archer, 1996). However, none of the associated woody species appears to have the genetic potential to achieve the size of mature Prosopis plants, either above-or below-ground. Thus, there may be less potential for carbon storage once Prosopis is lost from the system, unless the remaining understory species compensate by increasing their productivity. In addition, the carbon currently stored in Prosopis biomass would be lost via death and decomposition, albeit rather slowly. It would be interesting to explore these scenarios with a linked CENTURY-succession model. Unfortunately, we know little of the productivity of the understory shrubs. Further, the maximum age of P. glandulosa is unknown and we have little basis on which to prescribe mortality from the present-day population.

0 0

Post a comment