Introduction

Woody plant invasion in grasslands and rangelands often creates a spatial mosaic of herbaceous species composition and production and soil nutrient distribution that is very different beneath than beyond woody canopies (Scholes and Archer 1997; Riginos et al. 2009). This effect is variable depending on the woody species involved, the nature of the grassland community before woody invasion, and a host of other abiotic and/or disturbance effects such as soil type, droughts, livestock grazing, and fire.

A common woody invader worldwide is mesquite (Prosopis spp.) (Van Auken 2000; Pasiecznik et al. 2001; Bustamante et al. 2006). Species of Prosopis have invaded regions of Australia, Africa, India, South America, Central America, and the south central and southwestern regions of the United States. Common observations in these regions are that, because Prosopis is a nitrogen-fixing legume, soil N is greater beneath than beyond the Prosopis canopy (Tiedemann and Klemmedson 1986; Franco-Pizana et al. 1995; Geesing et al. 2000; El-Keblawy and Al-Rawai 2007; Herrera-Arreola et al. 2007; Boutton and Liao 2010). Low densities of Prosopis can be beneficial to growth of some grass species beneath Prosopis canopies due to nutrient enrichment and ameliorating effects of the Prosopis canopy on air and soil temperature (Tiedemann and Klemmedson 1977; Scholes and Archer 1997; McClaran and Angell 2006). However, moderate to high Prosopis density and canopy cover reduces grass production, especially mid-sized C₄ grass species (Archer 1995; Ansley et al. 2004, 2013; Patnaik et al. 2017). Finally, most Prosopis species have a dimorphic root system composed of a deep taproot and lateral roots that can extend > 10 m beyond the canopy edge (Heitschmidt et al. 1988; Gibbens and Lenz 2001; Ansley et al. 2014), conferring a competitive advantage in drylands.

Honey mesquite (Prosopis glandulosa Torr., hereafter mesquite) is the dominant Prosopis species in the 368 000-km² Southern Great Plains (SGP) region of the United States (  Fig. S1 (mmc1.docx), available online at doi: 10.5061/dryad.nzs7h44rd). Mesquite expanded in range and density in the SGP after European settlement in the late 1800s due to grassland fire suppression, overgrazing of grass by cattle, and enhanced seed distribution by cattle via endozoochory (Van Auken 2000; Ansley et al. 2017).

In the SGP there exists a unique situation of mesquite subcanopy being dominated by a C₃ midgrass, Texas wintergrass (Nassella leucotricha), and intercanopy microsites containing a mixture of Texas wintergrass, C₄ short grasses, and remnant patches of C₄ midgrasses that are most sensitive to mesquite expansion (Ansley et al. 2013). In contrast, Prosopis invasion cases elsewhere have C₄ grass (Gibbens et al. 1996; El-Keblawy and Al-Rawai 2007; Wheeler et al. 2007; Riginos et al. 2009; Getachew et al. 2012; Ndhlovu et al. 2016), or C₃ subshrub species beneath Prosopis canopies (Archer 1990). Mesquite understory dominance by Texas wintergrass in the SGP has resulted in an alternate stable ecological state (Westoby et al. 1989), referred to as the “Prosopis/Nassella association” (Ansley et al. 2019), that is unlikely to transition back to what is assumed to have been C₄ midgrass dominance (Liao et al. 2006) without significant anthropogenic inputs to remove mesquite. The loss of C₄ midgrasses is significant because these high-producing grasses are critical for gallinaceous bird habitat (Tomecek et al. 2017), cattle forage (Ansley et al. 2013), and carbon sequestration (Conant and Paustian 2002).

Herbaceous production during mesquite expansion is different in the SGP than other Prosopis-dominated systems because of the presence of Texas wintergrass. In many regions, understory C₄ grass production declines to near zero as Prosopis cover increases (Archer 1995; El-Keblawy and Al-Rawai 2007; Mohamed et al. 2011; Patnaik et al. 2017). In contrast, in the SGP, as mesquite expansion increases and canopies coalesce, Texas wintergrass remains productive in subcanopy and the progressively smaller intercanopy areas transition from a mixture of C₄ midgrasses and short grasses to Texas wintergrass and C₄ short grasses; C₄ midgrasses largely disappear. Texas wintergrass can persist under high mesquite density by avoiding direct competition with mesquite for light and soil water by growing in early spring before mesquite leaves emerge and entering a quiescent physiological state in midsummer (Hicks et al. 1990; Simmons et al. 2008). Unlike C₄ midgrasses with deeper root systems, C₄ short grasses avoid competition with mesquite for soil moisture in intercanopy because their roots occupy soil layers above the 0.5–1.5 m depth where most of the first order mesquite lateral roots reside (Ansley et al. 2014), although some second-order branches extend to near the surface (Gibbens and Lenz 2001). C₄ short grasses are also better adapted than C₄ midgrasses to withstand heavy grazing (Wood and Blackburn 1984; Teague et al. 2011).

Because mesquite resprouts vigorously after top-killing treatments such as prescribed fire or mechanical chaining, many resource managers recognize the long-term importance of root-killing treatments (Bovey and Whisenant 1991; Ansley and Castellano 2006). Aerial application of herbicides is currently the lowest-cost treatment for root-killing mesquite on large land areas and requires only a single application every 20–25 yr (Ansley et al. 2004). While several studies have quantified changes in herbaceous production following mesquite treatments in the SGP (McDaniel et al. 1982; Bedunah and Sosebee 1984), none have recorded responses within intercanopy and subcanopy microsites and in concert with soil N measurements. Differences in C₃ and C₄ grass composition and potentially different soil N amounts between these microsites in the SGP provides a unique post-treatment dynamic compared with other Prosopis systems that needs further investigation to broaden our understanding of Prosopis invasion and management strategies in temperate climates (Bustamante et al. 2006; Riginos et al. 2009).

Some brush treatment studies have found a high forb response after treatment (Fulbright 2004). This usually occurs following mechanical woody plant treatments that disturb the soil and create favorable environments for forb seed germination (Bozzo et al. 1992; Kunst et al. 2012) or prescribed fire that temporarily increases bare ground (Engle et al. 1998; Fuhlendorf and Engle 2004). Forbs are not a prominent component of the Prosopis/Nassella system of the current study (Ansley and Castellano 2006; Teague et al. 2014), except for occasional rainfall-related outbreaks of annual forbs (Stanford et al. 2008).

Our objectives were to compare production of grass and forb functional groups and available soil N amounts within intercanopy and mesquite subcanopy microsites in the first 2 yr after a non–soil-disturbing mesquite root-killing treatment. We hypothesized that there would be an increase in Texas wintergrass production in soil N-enriched subcanopy to the exclusion of other grass or forb species. In addition, we expected an increase in intercanopy production of C₃ and C₄ grass species due to the loss of competition from mesquite via lateral roots that extend into intercanopy areas. We hypothesized that any short-term forb responses would occur in intercanopy but not subcanopy due to Texas wintergrass dominance in subcanopy. In addition, since our brush treatment did not disturb the soil, we expected forb production to be low relative to grass production.

Methods

Research was conducted on the Smith-Walker Experimental Ranch in north central Texas (34°01′52″N; 99°15′00″E; elevation 372 m), which is near the north-south midpoint in the SGP (see   Fig. S1 (mmc1.docx); USDA-NRCS 2006). Mean annual rainfall (30 yr; 1981–2010) is 710 mm with peaks in June (108 mm) and September (80 mm). Mean annual air temperature is 17.1°C, and average daily monthly air temperatures range from 35.9°C in July to –2.4°C in January (NOAA-NCDC 2019). Soils are fine, mixed, superactive, thermic Typic Paleustalfs of the Wichita series and fine, mixed, superactive, thermic Vertic Paleustolls of the Tillman series; both are 1- to 2-m deep clay loams on 1–3% slopes (USDA-NRCS 2019a). Ecological site description is clay loam R078CY096TX (USDA-NRCS 2019b).

Vegetation consists of a woody overstory of multistemmed 3–5 m tall, 30- to 40-yr-old honey mesquite, 40–60% canopy cover, and a herbaceous mixture of C₃ and C₄ perennial grasses. The primary C₃ midgrass is the bunchgrass Texas wintergrass (Nassella leucotricha [Trin. and Rupr.] Pohl.). Common C₄ midgrasses are sideoats grama (Bouteloua curtipendula [Michx.] Torr.), vine mesquite (Hopia obtusa [Kunth] Zuloaga & Monroe), and sand dropseed (Sporobolus cryptandrus [Torr.] A. Gray). Common C₄ short grasses are buffalograss (Buchloe dactyloides [Nutt.] Engelm.), curly mesquite (Hilaria belangeri [Steud.] Nash), and the bunchgrass form of blue grama (Bouteloua gracilis [Kunth.] Lag. ex Griffiths). C₃ annual grasses include Japanese brome (Bromus japonicus Thunb. ex Murray) and little barley (Hordeum pusillum Nutt.). Perennial forbs include western ragweed (Ambrosia psilostachya DC.) and silverleaf nightshade (Solanum elaeagnifolium Cav.). Annual forbs include annual broomweed (Amphiachyris dracunculoides DC.) and marestail (Conyza canadensis), [L.] Cronquist) (Hatch and Pluhar 1993; Stubbendieck et al. 2017; USDA-NRCS 2020). Before and during the study period, cattle freely grazed at a moderate stocking rate of 6–8 ha · animal unit^–1 · yr^–1.

Herbaceous and soil nutrient responses were measured within two treatments, untreated and mesquite sprayed with root-killing herbicide (hereafter “treated”). A previous extension-oriented demonstration study established two 4-ha plots each of three different mesquite-killing clopyralid-based foliar herbicide treatments, all of which were applied on July 10, 2014. For the current study, which was conceived after these treatments were applied, we randomly selected four plots for the “treated” treatment, with the common response being that mesquite canopy cover was reduced to near zero, and there was a high percentage of mesquite with complete aboveground mortality (“top-kill”) and apparent mortality (“root-kill”) (formal evaluation at 2 yr post treatment found 74.5% [s.e. ± 6.5] top-kill and 66% [± 6.2] root-kill). In addition, these plots were distributed such that each plot had an adjacent ∼4-ha area of untreated mesquite. There is no indication in the literature that these herbicide mixtures (  Table S2 (mmc1.docx), available online at …) adversely affect grass or forb production. Thus, we considered the four treated plots as functionally equivalent replicates with respect to the elimination of mesquite competition.

Each pair of plots (treated, untreated) was identified as a block. Within each treatment in each block, 3 patches of 10–20 mesquite trees were randomly selected as sample areas. Within each of the three sample areas, herbaceous and soil measurements were made beneath the canopy (hereafter “subcanopy”) of a single large (> 3 m tall; canopy radius > 3 m) mesquite tree and in an adjacent intercanopy area between mesquites during 2015 and 2016 (yr 1 and 2 post mesquite treatment). Subcanopy and intercanopy sites were termed “microsites.” Total number of sample areas each year was 48 (2 treatments × 4 blocks per treatment × 2 microsites per treatment within a block × 3 samples per microsite). The N value for each of the four treatment-microsite combinations (intercanopy-untreated, intercanopy-treated, subcanopy-untreated, subcanopy-treated) was 12.

Herbaceous production was measured within 1-m wide × 2-m long × 1.5-m tall wire cages made of sheep fence that were randomly located in each microsite. Cages in intercanopy were located at least 2 m from the outside edge of any mesquite canopy. Cages in the subcanopy were located within the 2-m-wide band occurring at least 0.5 m away from the outer basal stems and 0.5 m inside the canopy edge of each live mesquite tree in untreated plots, or beneath standing dead stems in treated plots. During sampling, all herbaceous material was clipped to within 1–2 cm of ground level inside a 0.125-m² quadrat frame randomly placed in each cage, separated by species, bagged, dried at 60°C until dry, and weighed. Two clip samples were taken at different positions within each cage, one in early summer (late May or June) and the other in fall (late September or October), to account for different timing of peak production of C₃ and C₄ species. Herbaceous litter was collected in each clip sample by gathering any dead herbaceous material that was disconnected from standing grass or forbs, or determined visually to not be part of the current year's production. After oven drying, soil particles were removed from litter using a 2-mm mesh screen. All cages were moved to a different location within 2 m of the previous location in yr 2. Herbaceous data were grouped into functional groups: C₃ midgrasses (C3M), C₃ annual grasses (C3A), C₄ short grasses (C4S), C₄ midgrasses (C4M), perennial forbs, cool-season annual forbs, warm-season annual forbs, and litter (LIT). Perennial grass (PGR) production was determined by adding C₃ midgrass, C₄ short grass, and C₄ midgrass values. Total forbs (FRB) included all forb subgroups. Total herbaceous (THB) production was determined by adding all grass and forb values.

Soil samples (2-cm diameter) at 0- to 15-cm and 15- to 30-cm depth increments were collected during each herbaceous sampling period at each intercanopy (1 m north of each intercanopy clip cage) and subcanopy (1 m away from the outer basal stems and outside of the clip cages) microsite. Due to limitations in the number of laboratory samples that could be analyzed, soils from each of the three sample areas in each treatment and microsite in each block were composited (N = 4 instead of 12). Samples were mailed to the US Department of Agriculture–Agriculture Research Service Grassland Soil and Water Research Laboratory (808 E Blackland Rd, Temple, TX 76502) and analyzed for soil inorganic N (SINN; NH₄N + NO₃-N) and water-extractable organic N (WEON) via the Haney Soil Test procedure (Haney et al. 2006, 2010). This test uses water and an organic acid extractant, H³A (including citric, malic, and oxalic acids), which are designed to mimic plant root exudates to estimate plant-available nutrients. WEON represents the portion of organic N that is available to plants over the course of a growing season (Bavougian et al. 2019; Ward Lab 2020). For herbaceous and soil data, values from both sample periods each year were averaged.

To quantify differences in shading by mesquite canopies before and after treatment, photosynthetic photon flux density (PPFD) was measured with a Li-Cor LI-191SA Quantum sensor at ground level in intercanopy and subcanopy microsites in untreated and treated plots during a cloud-free midday period (1 050 to 1 240 h) in July 2015. Similar measurements were made in four additional plots that had similar-sized mesquite, were sprayed with the same treatments in 2013, and were located within each block that included the 2014 treated and untreated plots. These plots were not part of the herbaceous and soil response data but provided an example of subcanopy PPFD at 2 yr post treatment to account for deterioration of standing dead mesquite stems from yr 1 to yr 2. This provided a chronosequenced comparison of shade cast by untreated, 1-yr post-treated, and 2-yr post-treated mesquite in subcanopy and untreated and treated intercanopy microsites. Twenty-four PPFD measurements were made during the midday period (4 blocks × 3 treatments/block × 2 microsites/treatment; N = 4).

Results

Precipitation and PPFD

Annual precipitation was near normal during the 2014 spray yr (yr 0) and 54% and 30% above normal in 2015 (yr 1) and 2016 (yr 2), respectively (Fig. 1). Growing season precipitation (April–September) was 11%, 59%, and 38% above normal in 2014, 2015, and 2016, respectively. May 2015 had > 3 × the normal precipitation. Precipitation in fall 2014 (October–December) before yr 1 was 125.9 mm, close to average (125.8 mm). In contrast, precipitation in fall 2015 before yr 2 was 106% above normal at 259.8 mm.

Fig. 1.

Monthly and annual precipitation recorded at the study site, 2014–2016, with data collection in 2015 (yr 1) and 2016 (yr 2). Gray bars are monthly totals, and solid symbols connected by lines are 30-yr mean monthly values. Values below year label are annual precipitation totals (mean annual total 710 mm). Mesquite spray treatment was in early July 2014.

Untreated mesquite canopies reduced subcanopy PPFD by 60.3% compared with untreated intercanopy (Fig. 2). The shading effect of standing dead treated mesquite canopies decreased from 1–2 yr post treatment (32.4–19.5%) as stems deteriorated. Untreated mesquite reduced subcanopy PPFD by 44.5% and 52.6% compared with treated mesquite at 1 and 2 yr post treatment, respectively. PPFD was similar in all intercanopy microsites.

Discussion

The increase in PGR production following root-killing of mesquite in the first yr post treatment came mainly from the C₃ midgrass, Texas wintergrass, in both microsites. The percentage increase in PGR and C₃ midgrass production in treated subcanopy was similar due to PGR being mostly composed of C₃ midgrass. In contrast, the percentage increase in PGR production in treated intercanopy (73%) was less than the increase in C₃ midgrass production (106%) because of the low response of C₄ grasses to treatment in yr 1. Other studies in the SGP with a similar Prosopis/Nassella association found increases in Texas wintergrass production following mesquite treatment (McDaniel et al. 1982; Laxson et al. 1997) but did not contrast production between microsites. Increased litter in treated microsites in yr 2 was due in part to increased yr 1 C₃ midgrass production, as well as greater precipitation in yr 1 compared with the previous yr.

Fig. 2.

A and B, Midday photosynthetic photon flux density (PPFD) at ground level on July 17, 2015 and for 1- and 2-yr post-treatment microsites. Data for untreated intercanopy and untreated subcanopy are duplicated in each “chronosequenced” year. Vertical lines are standard error (n = 4). Actual PPFD values are shown above each bar. Percentage values represent the percent reduction in PPFD between 2 means connected by a red line. Means with similar letters within each year are not different at P ≤ 0.05.

Table 1

Pr > F values in main model analysis of effects of treatment (T), microsite (M), and year (Y) and their interactions on annual production of C₃ midgrass (C3M), C₃ annual grass (C3A), C₄ short grass (C4S), C₄ midgrass (C4M), perennial grass (PGR), forbs (FRB), total herbaceous (THB), and litter (LIT). Values in bold and gray shade are P ≤ 0.05.

Table 2

Pr > F values in main model analysis of effects of treatment (T), microsite (M), and year (Y), and their interactions on soil inorganic N (SINN) and water-extractable organic N (WEON) at 0- to 15-cm and 15- to 30-cm soil depths. Values in bold and gray shade are P ≤ 0.05.

While most Prosopis-dominated systems worldwide have a C₄ grass understory, there are examples where C₃ grasses have displaced C₄ grasses beneath Prosopis or other woody legume species (Stuart-Hill and Tainton 1989; Rossi and Villagra 2003; Prober et al. 2005; Rauber et al. 2014). In contrast, in the subtropical region of southern Texas, P. glandulosa serves as a nucleus tree, facilitating the recruitment of secondary C₃ shrubs instead of C₃ grasses beneath its canopy (Archer 1990; Franco-Pizana et al. 1995) in what was C₄ grassland before mesquite invasion (Liao et al. 2006). Bustamante et al. (2006) proposed that with conspecific mesquite and similar annual precipitation in north and south Texas, the reason why C₃ subshrubs are not associated with north Texas mesquite may be due to colder winter temperatures.

Fig. 3.

A–H, Production of grass functional groups at 1-yr (2015) and 2-yr (2016) postmesquite treatment (TRT) compared with untreated (UNT) intercanopy and subcanopy microsites. Vertical lines are standard error (n = 12). Means with similar letters within each panel are not different at P ≤ 0.05. An asterisk in 2016 panels indicates a significant (P ≤ 0.05) difference between 2015 and 2016 within that treatment-microsite.

Regarding soil N, studies in southern Texas (Franco-Pizana et al. 1995; Geesing et al. 2000; McCulley et al. 2004), Arizona (Tiedemann and Klemmedson 1986; Wheeler et al. 2007), and northern Mexico (Herrera-Arreola et al. 2007) found that soil total N and/or inorganic N were greater in Prosopis subcanopy than intercanopy microsites. We found similar responses with SINN in two of the four treatment-microsite-year combinations, and it was consistent across both soil depth increments. However, greater SINN in subcanopy occurred in treated plots in yr 1 but in untreated in yr 2, so no conclusions regarding treatment effect can be drawn.

Fig. 4.

A–F, Perennial grass (PGR) and forb production (annual + perennial) and litter amount at 1-yr and 2-yr postmesquite treatment (TRT) compared with untreated (UNT) in intercanopy and subcanopy. Error bars and mean comparison symbols are described in Fig. 3 caption.

SINN significantly (P ≤ 0.05) increased from yr 1 to yr 2 in three of the four treatment-microsite combinations in both depth increments. In the other instance, untreated intercanopy, SINN increased numerically from yr 1 to yr 2, but the difference was not significant. We are uncertain as to the factors that caused SINN to increase in yr 2. Untreated intercanopy would likely have the lowest probability for the increase in SINN to be related to mesquite. The increase in SINN in the subcanopy microsites, both treated and untreated, could be related to N-fixation by mesquite. The yr 2 increase in SINN in treated intercanopy could have come from the decomposition of dead mesquite lateral roots, but other factors may have also been involved.

Soil WEON significantly decreased from yr 1 to yr 2 in five of the eight treatment-microsite-depth combinations. SINN increased from yr 1 to yr 2 in four of those five situations, but it is unknown if those two trends are related. More sample dates than used here are needed to correlate WEON with SINN.

C₃ midgrasses in untreated and treated plots in yr 1, as well as C₃ annual grasses in treated plots in yr 2, were the only functional groups that had greater production in subcanopy than intercanopy microsites within a particular treatment. However, in only one of these instances (C₃ midgrasses in treated plots) did greater grass production coincide with greater SINN in subcanopy compared with intercanopy. Other factors besides enhanced soil N must have been responsible for the increased C₃ grass production in subcanopy in the other two situations. Greater C₃ midgrass production in untreated subcanopy than untreated intercanopy may have been due to greater competition from C₄ grasses in intercanopy, or more moderate temperatures beneath the mesquite canopy that favored C₃ midgrasses (Tiedemann and Klemmedson 1977; Simmons et al. 2008). Enhanced C₃ annual grass production in treated subcanopy in yr 2 may have been the result of more propagules in subcanopy than intercanopy to begin with, a wet winter that stimulated germination, and enhanced PPFD from mesquite canopy removal.

C₄ short-grass production was greater in intercanopy than subcanopy but did not increase in treated plots within each microsite in either year. Before treatment, mesquite canopy cover was high enough (40–60%) to reduce the size of intercanopy patches sufficiently to enable Texas wintergrass to establish beyond the mesquite canopy edge. At 1 yr post treatment, Texas wintergrass may have been able to exploit mesquite mortality in intercanopy more effectively than C₄ short-grasses. By yr 2, enhanced annual forb production may have limited C₄ short-grass production in treated intercanopy.

Fig. 5.

A and B, Total herbaceous (THB) production comprising all herbaceous functional groups at 1 and 2 yr post mesquite treatment (TRT) compared with untreated (UNT) in intercanopy and subcanopy. Error bars and mean comparison symbols are for THB only and are described in the Fig. 3 caption. A. Forbs indicates annual forbs; P. Forbs, perennial forbs; C4M, C₄ midgrasses; C4S, C₄ short grasses; C3A, C₃ annual grasses; C3M, C₃ midgrasses.

Fig. 6.

A–D, Soil inorganic N (SINN) and water-extractable organic N (WEON) at two depths in untreated (U) and treated (T) intercanopy (INT) and subcanopy (SC) in yr 1 (2015) and yr 2 (2016). Vertical lines are standard error (n = 4). Means with similar letters within each soil depth and year are not different at P ≤ 0.05. Asterisk in yr 2 indicates a significant (P ≤ 0.05) difference between years in a particular treatment-microsite combination.

Low C₄ midgrass production in subcanopy in yr 1 may have been a legacy effect due to the loss of C₄ midgrass propagules from prolonged dominance by Texas wintergrass, shading by mesquite, or continuous livestock grazing that caused greater detrimental effects to C₄ midgrasses (Wood and Blackburn 1984; Teague et al. 2011). Ansley et al. (2019) found that C₄ midgrass production did not begin to displace Texas wintergrass production in intercanopy until 3 yr after a mesquite top-killing treatment. Several studies have found an allelopathic effect of Prosopis juliflora leaf exudates on germination of C₄ grasses (Al-Humaid and Warrag 1998; Getachew et al. 2012). It is not known if P. glandulosa in the SGP inhibits herbaceous production via allelopathy. If so, this may have contributed to low C₄ grass production in subcanopy in yr 1 but did not inhibit C₃ Texas wintergrass growth. Kaur et al. (2012) suggested that Prosopis allelopathy is more likely to occur when Prosopis is introduced as an exotic (e.g., P. juliflora native to Venezuela introduced to India). Because P. glandulosa is native to the SGP, it may have less of an allelopathic effect on native grasses.

Ecological and Management Implications

Our study revealed that post-treatment perennial grass production may differ between intercanopy and subcanopy microsites, especially if the woody overstory is an N-fixing legume that may facilitate C₃ over C₄ grass production beneath its canopy. Models that predict landscape scale post treatment herbaceous responses in such systems could incorporate these differences based on pretreatment woody canopy cover. Composition of total herbaceous production varied considerably in yr 2, yet mesquite removal yielded similar herbaceous production levels each year and in each microsite. Because both years had above-average growing season precipitation, site production potential may have been achieved with available propagules (i.e., without higher-producing C₄ midgrasses). There was no strong association between soil inorganic N and increased herbaceous production when compared across all treatments and microsites. This was due to low herbaceous production beneath untreated mesquite canopies where soil inorganic N was highest and high production in treated intercanopy where inorganic N was lowest. Other factors, such as increased light from mesquite canopy removal and precipitation patterns that stimulated annual forb and grass germination, had greater effect on total herbaceous production.

From a management perspective, although this study measured responses during only the first 2 yr after treatment, it revealed an important principle that grass forage production following brush control can deviate markedly from predicted results under certain conditions. In the system studied here (dense mesquite with a Texas wintergrass understory), research and models predict an increase in C₃ Texas wintergrass production the first few years after a mesquite root-killing treatment, followed by a replacement of Texas wintergrass by more productive C₄ grasses (Teague et al. 2001; Ansley et al. 2019). The high forb production that displaced Texas wintergrass production in yr 2 was unexpected, especially with a brush treatment that did not disturb the soil or remove litter. Responses may have been linked to late fall and early spring rainfall patterns before the yr 2 growing season. While forbs are critical for certain ecosystem services, an awareness of such a possible response is important if the business model depends on increased livestock production to offset brush treatment cost.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.