Site Preparation and Planting Strategies to Improve Native Forb Establishment in Pasturelands ^†

Abstract

Increasing the diversity of native forbs in pasturelands can benefit insect pollinator populations, which have been declining widely. Establishing native forbs into existing pasturelands can be challenging, however, and information about effective planting strategies in these systems is lacking. In this study, we evaluated several planting strategies to improve native forb establishment. Two field experiments were conducted in Virginia, USA in 2021 and 2022. Experiment 1 evaluated how six herbicide treatments and tillage affected establishment success when forbs were planted in summer or fall. Experiment 2 investigated how different seeding rates from 2.2 to 56 kg/ha and pre-seeding cold stratification affected forb establishment. In experiment 1, treatments using Roundup/glyphosate and tillage resulted in the most forb establishment. Planting in summer improved establishment with Roundup/glyphosate application. In experiment 2, native forb plant establishment was positively associated with seeding rate (p < 0.001), with a rate of 56 kg/ha resulting in almost 3x more forbs compared to the lowest seeding rate. Cold stratification also increased target plant establishment (p < 0.01), but these effects were inconsistent among species. Our results suggest that effective native forb establishment can be achieved through intensive site preparation with Roundup/glyphosate or tillage to suppress vegetation and planting at rates no higher than 11 kg/ha.

1. Introduction

The southeastern USA has millions of hectares of low-diversity pastureland that, if enriched with more native forbs, could significantly boost plant diversity and possibly generate more ecosystem services such as pollination [1,2]. Most pasturelands are primarily dominated by non-native graminoids and legumes and are actively maintained to discourage non-leguminous forbs [3,4]. The wider use of non-native forbs in southeastern pasturelands has challenges since little is known about the propagation and establishment of these plants in pasture situations [4]. More information is needed about how to effectively establish native forbs in existing pasturelands to reduce the chance of stand failure or unpredictable outcomes in species composition [5,6].

Several methods have been developed to increase native plant diversity in pastures including sod removal [7], plug planting [8], and reduced grazing pressure to express a native seed and bud bank [9]. However, especially with large areas, biodiversity enrichment via inter-seeding is often the most cost-effective way to increase native forb diversity. When inter-seeding pastureland with native forbs, several factors must be taken into consideration including microsite availability, seeding rate, and seed conditioning [10]. Pasture vegetation inhibits microsite availability by reducing seed–soil contact and limiting resources through competition [11,12,13]. As such, the use of chemical vegetation suppression through herbicides or physical suppression using tillage can be effective tools to enhance native forb establishment.

As noted above, there are several avenues to establishing native forb stands, and previous studies have tested an array of viable methods. However, there is no clear “best” approach as there are often trade-offs in cost, labor, scalability, and efficacy. Topsoil removal, for example, can eliminate weed seeds in the soil, but it is impractical or impossible depending on the context, and may run counter to other restoration goals [11]. Herbicide treatments can be effective at eliminating weeds in an establishing forb stand, and, in many cases, this is a standard approach [14]. For example, a study conducted in Virginia, USA, compared glyphosate to tillage to suppress existing vegetation along with the use of post-emergent herbicides to suppress weed growth [15]. The authors found that glyphosate was superior to tillage for establishment, and, although there may be some benefit from post-emergent herbicides, the additional cost was viewed as prohibitive. Beran et al. [16] tested three different imidazolinone herbicides applied directly following seeding of the forbs to determine which species may be tolerant to herbicide application. Their findings indicate that some wildflower species can exhibit herbicide tolerance, and the benefits from herbicides are most pronounced when weed density is high. In another herbicide-related study, Ghajar et al. [17] found that imazapic was highly effective at limiting the emergence of weeds in native forb plantings, but the herbicide also suppressed some species of forbs, so should be used with caution.

Although effective, herbicides may be undesirable for other reasons. For example, a farm may be organic-certified and not permitted to use herbicides. Research also suggests that indirect exposure to herbicides can harm the same pollinators we are trying to support by establishing greater forb diversity in pasturelands [18,19]. Mechanical tillage represents a chemical-free alternative to herbicide, even though it may be less effective at reducing the density of weeds [20,21]. A recent study compared tillage, herbicide, and prescribed fire methods to establish native forbs and found that tillage yielded the greatest coverage of sown forbs and the lowest cover of weeds—even outperforming glyphosate application [17]. A study conducted in the UK compared different seed application methods using shallow tillage with turf stripping to suppress vegetation and restore biodiversity in a chalk grassland [22]. The authors found that both tillage and turf stripping were effective at restoring biodiversity, but the effectiveness depended on site characteristics. Another experiment in southern California (USA) compared tillage with solarization, herbicide, and mowing to restore native forb cover in old agricultural fields dominated by exotic grasses [23]. Results indicated a combination of treatments was most effective, but tillage worked well alone to eliminate exotic grasses while creating conditions ideal for sowing desirable native seeds.

An optimal seeding rate for native forbs should reflect a balance between seed cost and seeding rate to ensure the successful establishment of plants and efficient use of financial resources. Grassland restoration plantings carried out in the Midwest USA usually involve high seeding rates to improve the density and richness of sown plants in the resulting stand [24]. Planting native forbs at very high seeding rates can become prohibitively expensive, though, especially if the seed is purchased commercially. As an alternative, forbs can be sown at lower planting rates, particularly when seed germination is accelerated by stratifying, scarifying, or otherwise breaking seed dormancy prior to sowing [25]. Pre-treatment of forb seed using cold stratification can be an effective method to speed germination as many North American forb taxa have dormancy mechanisms that can be broken only with a period of moist cold stratification experienced between late fall and early spring [26]. Forb seeds can also be cold stratified artificially (e.g., through refrigeration) and sown in late spring to early summer to limit predation and pathogen risks in the seedbed [27].

The overall objective of this study was to evaluate how differences in site preparation, planting date, seeding rate, and seed treatment would influence native forb establishment in Virginia, USA. In experiment 1, we evaluated sown native forb density and richness across six herbicide treatments and one tillage treatment with native forbs sown either in summer or late fall. A second experiment focused on the effects of seeding rate and pre-seeding cold stratification on forb establishment. Collectively, these two experiments could add to guidelines for establishing native forbs in southeastern USA pasturelands, including for rehabilitation purposes or native grassland restoration.

2. Materials and Methods

2.1. Study Area

Experiments were conducted at two research farms located in the Ridge and Valley Province of Virginia, with one site in the Shenandoah Valley (Shenandoah Valley Agricultural Research and Extension Center [SVAREC]; 37°55′50.8″ N, 79°12′43.3″ W) and one in the New River Valley (Kentland Farm (37°11′45.0″ N, 80°34′46.2″ W). Both farms fall within the humid subtropical (Cfa) Köppen climate classification characterized by warm to hot summers and mild winters with precipitation evenly distributed throughout the year [28]. Annual average precipitation varies from 850 to 1270 mm, historically supporting broadleaf deciduous forests as well as woodlands, savannas, and open grasslands. Much of the study area has been heavily fragmented and modified, especially through fire suppression by five centuries of Euro-American settlement, with the area today forming a matrix of secondary forest, urban development, and agricultural land. Previous vegetation covering each experimental area was dominated by tall fescue (Schedonorus arundinaceus), orchardgrass (Dactylis glomerata), and white clover (Trifolium repens). The Kentland site soil was classified as Unison and Braddock loams. The SVAREC plots were represented by equal parts eroded Weikert–Berks channery silt loams and eroded Bookwood silt loam. No adjustments to soil fertility were made at either site. Previous soil tests analyzed at the Virginia Tech Soil Testing Lab indicated that fertility levels were likely adequate for wildflowers at both locations, with an average soil pH of 6.4 and P and K concentrations averaging 24 ppm and 51 ppm, respectively. Weather data were summarized from each study site (

Table 1. Weather data summarized from stations located near each study site. Data can be accessed from the following websites: Kentland—https://www.vaes.vt.edu/college-farm/weather/2024weather.html, SVAREC—https://wcc.sc.egov.usda.gov/nwcc/site?sitenum=2088&state=va (accessed on 10 January 2024). NA indicates no data.

	Kentland
	Mean Air Temp. (°C)			Total Precipitation (mm)
Month	2021	2022	30 yr. mean	2021	2022	30 yr. mean
Jan.	0.90	−0.9	0.95	60.71	74.17	82.04
Feb.	1.86	2.60	2.35	88.65	82.80	71.88
Mar.	7.69	8.09	5.85	91.69	53.59	96.01
Apr.	11.00	11.93	10.70	45.97	53.85	95.76
May	15.42	17.44	15.10	43.69	140.21	113.54
June	21.01	21.50	18.95	82.04	12.45	108.46
July	22.34	23.21	20.85	64.77	115.82	106.93
Aug.	22.55	21.70	20.20	96.27	63.25	90.68
Sept.	18.24	17.50	17.15	120.40	64.52	87.63
Oct.	14.72	10.08	11.60	41.15	46.48	73.91
Nov.	4.45	7.48	6.35	22.35	48.77	72.39
Dec.	5.83	1.47	2.55	19.05	81.28	83.82
Mean/Total	12.17	11.83	11.05	776.73	837.18	1083.06
	SVAREC
	Mean Air Temp. (°C)			Total Precipitation (mm)
Month	2021	2022	30 yr. mean	2021	2022	30 yr. mean
Jan.	1.51	−0.80	2.1	44.45	63.25	80.01
Feb.	0.80	3.02	3.5	68.07	90.42	70.10
Mar.	8.51	8.12	7.15	72.39	61.21	89.40
April	12.45	11.68	12.2	49.53	71.37	94.23
May	15.86	16.58	16.55	80.77	139.95	104.39
June	20.81	20.84	20.6	81.53	85.60	121.92
July	23.50	22.95	22.65	47.75	128.78	104.14
Aug.	NA	21.43	22	NA	171.45	84.07
Sept.	NA	17.81	18.65	NA	79.50	93.21
Oct.	NA	10.20	13	NA	69.34	77.72
Nov.	NA	8.08	7.4	NA	96.27	82.04
Dec.	6.83	1.23	3.65	6.35	105.41	88.13
Mean/Total	-	11.76	12.45	-	1162.56	1089.36

). Overall, precipitation was lower in 2021 compared with 2022 at both sites and less than long-term means. No extended drought periods (e.g., greater than 30 days) were observed over the two years of the study, however. At the Kentland site, air temperatures were usually greater than the long-term mean values for most months.

2.2. Experiment 1

Experiment 1 evaluated how a suite of seven site preparation treatments would affect forb establishment at two planting times (summer, fall). Two experiments were established at each location, representing either a summer (June) or fall (November) planting time. A completely randomized block arrangement of treatments was used at each location. At each site, treatments were applied to 3 × 18 m plots (54 m²) replicated across four blocks. Site preparation treatments included a control, tillage with no herbicide, and six herbicide treatments (

Table 2. Site preparation treatments used applied to weaken or kill cool-season grasses in experiment 1.

Treatments	Herbicide Active Ingredient and Rate	General Action
Control/no vegetation suppression	---	---
Tillage/no herbicide	---	Physical vegetation suppression
Pastora (Envu Environmental Science, Cary, NC, USA)	Nicosulfuron at 7.36 g active ingredient (ai)/ha + metsulfuron methyl at 1.97 g ai/ha	Broad spectrum, selective broadleaf weed and grass control
Roundup (low rate) (Bayer Corp., Whippany, NJ, USA)	Glyphosate at 158 g acid equivalent (ae)/ha	Broad spectrum weed and grass suppression
Roundup (Bayer Corp., Whippany, NJ, USA)	Glyphosate at 2.54 kg ae/ha	Broad spectrum weed and grass control
Plateau (BASF Corp., Research Triangle Park, NC, USA)	Imazapic at 35.1 g ae/ha	Broad spectrum, selective broadleaf weed and grass control with residual effect
Cimarron Plus (Envu Environmental Science, Cary, NC, USA)	Metsulfuron methyl at 6.72 g ai/ha + chlorsulfuron at 2.10 g ai/ha	Broad spectrum, selective broadleaf weed and grass control with residual effect
Select Max (Valent USA, San Ramon, CA, USA)	Clethodim at 136 g ai/ha	Grass-selective postemergence herbicide

). Tillage plots were plowed twice with a disk plow to a depth of 15 cm. The herbicide treatment rates were intended to suppress existing vegetation such that it would allow forbs to establish (e.g., low-rate glyphosate) or eliminate existing vegetation. The lone exception was the full-rate glyphosate treatment, which was expected to kill all existing vegetation.

Native forb seed for both experiments was sourced from Ernst Conservation Seed (Meadsville, PA, USA), and twenty species were selected based on commercial seed availability and flowering phenology to promote nectar availability throughout summer (

Table 3. Information on the forb species mix used for experiment 1.

Species	Code	Family	Seeds kg−1	Pure Live Seed (PLS)	Seeding Rate (kg ha−1)
Agastache foeniculum	AGAFOE	Lamiaceae	3,080,000	0.99	0.03
Chamaecrista fasciculata	CHAFAS	Fabaceae	143,000	0.97	0.78
Coreopsis lanceolata	CORLAN	Asteraceae	486,200	0.83	0.22
Dalea candida	DALCAN	Fabaceae	611,600	0.75	0.22
Dalea purpurea	DALPUR	Fabaceae	660,000	0.91	0.11
Desmanthus illinoiensis	DESILL	Fabaceae	187,000	0.59	0.56
Desmodium canadense	DESCAN	Fabaceae	159,500	0.93	0.11
Desmodium paniculatum	DEAPAN	Fabaceae	440,000	0.96	0.22
Echinacea purpurea	ECHPUR	Asteraceae	254,460	0.91	0.44
Gaillardia pulchella	GAIPUL	Asteraceae	506,000	0.91	0.22
Helianthus maximiliani	HELMAX	Asteraceae	431,992	0.79	0.22
Heliopsis helianthoides	HELHEL	Asteraceae	224,400	0.92	0.44
Lespedeza capitata	LESCAP	Fabaceae	605,000	0.95	0.22
Lespedeza virginica	LESVIR	Fabaceae	352,000	0.98	0.33
Ratibida columnifera	RATCOL	Asteraceae	1,478,400	0.89	0.11
Ratibida pinnata	RATPIN	Asteraceae	978,942	0.95	0.11
Rudbeckia hirta	RUDHIR	Asteraceae	3,466,672	0.89	0.03
Silphium perfoliatum	SILPER	Asteraceae	220,000	0.90	0.44
Silphium terebinthinaceum	SILTER	Asteraceae	37,400	0.78	2.90
Solidago canadensis	SOLCAN	Asteraceae	10,120,000	0.57	0.01

). The native forb seeding rate was designed to deliver 11 seeds/species/m², on a pure live seed basis (PLS). Summer-sown plots received site preparation treatments in late May and early June 2020, with herbicide treatments applied with a tractor-mounted sprayer using a 6-nozzle hand-held boom with 11002XR TeeJet nozzles with 46 cm spacing calibrated to deliver 140 L per hectare. Plots were sown with native forb seed mix in mid-June using a Dew Drop seed drill (Little Sioux Prairie Co., Spencer, IA, USA) with a target seed rate of 7.6 kg/ha⁻¹. For the fall sowing treatment, the seed was planted using the same equipment in mid-November 2020. The plots assigned to the tillage treatment were tilled using the same equipment several days before planting occurred in November. Since no wildflowers were expected to emerge late in the growing season, herbicide application treatments for the fall planting were carried out the following spring in April 2021. Due to difficulty in sourcing Silphium terebinthinaceum for the fall planting, this species was dropped from the fall seed mix, resulting in a lower seeding rate (4.8 kg/ha⁻¹), with a total of 8.1 kg of cracked corn as a carrier. Cracked corn was added to help the flow of seed through the Dew Drop drill. The corn was not needed for the spring planting because the heavy Silphium terebinthinaceum seeds aided the seed flow. Seeding rates for all other species in the fall seed mix remained unchanged. No post-planting management was imposed on these plots after planting.

2.3. Experiment 2

To elucidate the influence of seeding rate and pre-seeding cold stratification on native forb establishment, our second experiment tested four seeding rates (2.2, 11, 28, 56 kg ha⁻¹; equivalent to 2, 10, 25, and 50 lbs. acre⁻¹) using either stratified or non-stratified seeds. We used a nested block design with three replicates per site. As in experiment 1, planting rates were formulated to equalize the number of pure live seeds planted per species in the mix. The area within each 6 × 3.8 m plot was divided into eight 1 m² subplots, two for each seeding rate, which were located next to each other, separated by a 10 cm buffer, adapted from the methodology of Groves and Brudvig [29]. The experimental sites were prepared by applying glyphosate at 2.54 kg ae/ha, tilled using a 1.5 m wide tractor-mounted Land Pride rototiller, and sprayed again using the same rate prior to planting. The stratified seeds were mixed with moist peat moss and refrigerated for four weeks at 4.4 °C prior to sowing. Seeds were sown between June 13 and 15th in 2021, and the same experiment was repeated in 2022 over the same dates. The seed mix for experiment 2 contained 16 species: 13 forbs and 3 grasses (

Table 4. Species list for mix planted in experiment 2.

Species	Code	Family	Seeds kg−1
Bouteloua curtipendula	BOUCUR	Poaceae	349,800
Coreopsis lanceolata	CORLAN	Asteraceae	486,200
Dalea purpurea	DALPUR	Fabaceae	660,000
Desmanthus illinoensis	DESILL	Fabaceae	187,000
Eryngium yuccifolium	ERYYUC	Apiaceae	391,600
Heliopsis helianthoides	HELHEL	Asteraceae	224,400
Lespedeza virginica	LESVIR	Fabaceae	352,000
Pycnanthemum virginianum	PYCVIR	Lamiaceae	8,518,400
Rudbeckia hirta	RUDHIR	Asteraceae	3,466,672
Schizachyrium scoparium	SCHSCO	Poaceae	530,200
Senna hebecarpa	SENHEB	Fabaceae	44,000
Solidago juncea	SOLJUN	Asteraceae	5,583,600
Symphyotrichum lateriflorum	SYMLAT	Asteraceae	1,760,000
Tradescantia ohiensis	TRAOHI	Commelinaceae	281,600
Tridens flavus	TRIFLA	Poaceae	1,023,000
Zizia aurea	ZIZAUR	Apiaceae	378,400

). The species choice reflected a wide range of taxa for deriving broader conclusions, with species falling into three broad functional groups: grasses, forbs, and legumes. Species were also selected based on being native to eastern North America.

Seeds were mixed with sand and hand broadcast over the plots. Non-stratified seed was mixed with peat moss at the time of sowing to eliminate any effects of peat moss independent of cold stratification. Plots were lightly raked to incorporate seeds into the soil. When all plants in the plot reached an average canopy height of 37 cm, the plot was cut back to 11 cm to reduce seedling competition with ruderal plants. At the end of the 2021 growing season, we observed that Heliopsis helianthoides was aggressively dominating the experimental plots, particularly at SVAREC. We adjusted the seed mix and reduced the H. helianthoides seed content by 25% for the experiment that was repeated in 2022. Other species in the mix were planted at the same rate and not increased to compensate for the lower H. helianthoides component.

2.4. Data Collection

Vegetation establishment data for experiment 1 were collected between Sep and Nov 2021. Data for experiment 2 was collected between Sep and Nov 2022. For experiment 1, we measured vegetation by laying a transect along the length of each treatment plot and placing 1 m² quadrats at 3.0, 7.6, and 12.0 m. Within quadrats, the percent ground cover for each species, sown and non-sown, was estimated, and the native forb species richness calculated. For sown species, stem density was counted across the entire treatment plot. In experiment 2, a 1 m² quadrat was placed over each subplot, and stem density was counted within the quadrat for each sown species. For both experiments, seedling establishment was considered successful if more than 5 forbs/m² could be observed [25].

2.5. Analysis

For experiment 1, linear and generalized linear mixed-effect regressions were used to evaluate the effects of the herbicide and tillage treatments (8 levels including control) and season sown (fall, summer) on native forb establishment, grass cover (%), and non-target vegetation cover (%). A regression approach was used, rather than ANOVA, mainly because we were more interested in how multiple independent variables might have affected the dependent variable(s) in question and less interested in detecting specific differences among treatment groups. Sites were used as replicates in the analysis. Random, intercept-varying effects were employed to account for the effects of treatment blocks nested within sites. Statistical analyses were performed using R Statistical Software version 4.2.2 (R Core Team R: A Language and Environment for Statistical Computing 2022). The lme4 package for mixed-effect regressions with lmerTest was used to generate p-values for fixed effects based on Satterthwaite approximation. Models assumed a Poisson distribution for stem counts and species richness and a Gaussian distribution for percent cover of grasses and non-target vegetation.

To compare models with varying complexity, we generated AIC scores corrected for small sample sizes (AICc) using the AICcmodavg package. We compared models, including a null model (intercept only), establishment treatment, season sown, treatment + season, and treatment × season. Selected models were further evaluated using Tukey’s honestly significant difference (HSD) tests to test for pairwise differences between treatment levels. Model performance was reported using R², calculated using the performance package [30].

For experiment 2, linear mixed-effect regression was used to evaluate the effects of seeding rate (2.2, 11, 28, 56 kg ha⁻¹) and pre-seeding stratification (yes/no) on target plant stem count and species richness per m². These data consisted of larger numbers than the site preparation experiment and approximated a Gaussian distribution. We developed models including null [intercept only], seed rate, stratification, seed rate + stratification, and seed rate × stratification.

3. Results

3.1. Experiment 1—Herbicides and Tillage

Establishment success was poor in experiment 1 as forb density did not exceed 1 plant/m² in any of the treatments. Of the 20 forbs sown, less than half were observed during sampling, with the most observed forbs including Silphium perfoliatum, Heliopsis helianthoides, Ratibida pinnata, Solidago canadensis, and Echinacea purpurea, which accounted for 73.5% of all observed sown forbs (Figure 1).

Results from linear mixed-effect regressions are shown in

Table 5. Results from linear mixed-effect regressions showing parameter estimates for the effects of site preparation treatment and planting date (PD) on native forb stem density and species richness.

Model	Parameters	Estimate	SE	Z	p-Value
Forb Stem Density (stems per m2)	Control	−0.405	0.645	−0.445	0.6550
	Cimmaron	−1.297	0.911	−2.011	0.0440
	Pastora	1.203	0.656	1.832	0.0660
	Plateau	3.319	0.586	5.661	<0.0001
	Glyphosate (full rate)	3.270	0.586	5.572	<0.0001
	Glyphosate (half rate)	3.075	0.589	5.219	<0.0001
	Select Max	3.877	0.582	6.662	<0.0001
	Tillage	3.898	0.581	6.699	<0.0001
	Planting Date (PD)	2.119	0.609	3.477	0.0005
	PD × Cimmaron	0.821	0.353	2.328	0.0199
	PD × Pastora	−1.092	0.712	−1.531	0.1250
	PD × Plateau	−4.459	0.712	−6.255	<0.0001
	PD × Glyphosate (full rate)	−1.034	0.623	−1.659	0.0971
	PD × Glyphosate (low rate)	−3.349	0.662	−5.053	<0.0001
	PD × Select Max	−4.899	0.699	−7.004	<0.0001
	PD × Tillage	−3.146	0.630	−4.993	<0.0001
Forb Species Richness	Control	−0.405	0.907	−0.447	0.6550
	Cimmaron	−1.149	0.612	−1.875	0.0600
	Pastora	0.287	0.759	0.379	0.7040
	Plateau	2.397	0.599	3.998	<0.0001
	Glyphosate (full rate)	2.197	0.605	3.630	0.0002
	Glyphosate (low rate)	2.159	0.606	3.561	0.0003
	Select Max	2.427	0.599	4.053	<0.0001
	Tillage	2.456	0.598	4.106	<0.0001
	Planting Date (PD)	1.299	0.647	2.006	0.0440
	PD × Cimmaron	0.150	0.368	0.408	0.6833
	PD × Pastora	0.147	0.851	0.173	0.8620
	PD × Plateau	−3.004	0.783	−3.832	0.0001
	PD × Glyphosate (full rate)	−0.606	0.688	−0.880	0.378
	PD × Glyphosate (half rate)	−2.254	0.746	−3.022	0.0025
	PD × Select Max	−3.216	0.804	−4.000	<0.0001
	PD × Tillage	−1.910	0.707	−2.702	0.0069

. While many treatment-by-planting-date interactions were noted, some general trends can be summarized. Summer-sown plots treated with the full rate glyphosate showed the greatest forb stem density (200 stems/m²) and species richness (14) compared with other treatment × season combinations (Figure 1). Forb establishment was more consistent across most fall-sown treatments. Plots that were either tilled (stem density, 134/m²) or had received Select herbicide (stem density, 116/m²) also established well and had comparable levels of sown species richness (Figure 1). Cimarron/metsulfuron and Pastora/nicosulfuron herbicide treatments were ineffective in facilitating forb establishment regardless of planting date. The full-rate Roundup/glyphosate treatment was most effective (p < 0.05) at reducing the existing grass vegetation originally present in plots (Figure 2).

3.2. Experiment 2—Planting Rate and Cold Stratification

The establishment of native forbs was more successful in experiment 2, as evidenced by the higher forb density, averaging 88 stems/m² in the treatment plots (Figure 3). The most abundant species included Heliopsis helianthoides, Rudbeckia hirta, Coreopsis lanceolata, Tridens flavus, Desmanthus illinoiensis, Bouteloua curtpiendula, Senna herbecarpa, and Zizia aurea. The species composition of plots also differed considerably in the experiment planted in 2021 compared with the 2022 experiment. Two grasses, Tridens flavus and Bouteloua curtpiendula, were particularly more abundant in 2022 compared with 2021. Of the forbs, Rudbeckia hirta, Desmanthus illinoiensis, and Symphyotrichum lateriflorum were more abundant in 2022 as well. Overall, forb stem density was positively associated with seeding rate and was also greater with cold stratification (p < 0.004) (

Table 6. Results from linear mixed-effect regressions showing parameter estimates for the associations between seeding rate, cold stratification, and target plant density and richness.

Model	Parameters	Estimates	SE	df	t Value	p-Value
Forb Stem Density (stems per m2)	Seeding rate	1.524	0.289	40	1.357	<0.0001
	Cold stratification	36.08	11.85	40	3.046	0.0040
Forb Species Richness	Seeding rate	0.004	0.009	40	0.415	0.6810
	Cold stratification	0.041	0.408	40	0.102	0.9190

). A seeding rate of 56 kg/ha resulted in almost three times as many target plants compared with the lowest seeding rate. Forb species richness was unaffected by seeding rate (p = 0.68) or cold stratification (p = 0.92).

4. Discussion

The purpose of this study was to evaluate methods for establishing native forbs in pastures to increase plant diversity, with the overarching goal of benefiting insect pollinators in the southern USA. We found that suppressing vegetation with a full rate of Roundup/glyphosate and planting in summer resulted in the greatest native forb establishment. For fall planting, forb establishment was favored by tillage or application of a grass-selective herbicide (Select Max). We also found that successful native forb establishment increased with greater seeding rates and with cold stratification for summer-planted stands.

The differential timing of herbicide application between fall and summer planting treatments may have affected the establishment of native forbs in our study. Summer-sown plots had herbicides applied in late May 2020, and fall-sown plots were sprayed in early April 2021, at a time when tall fescue is more vulnerable to glyphosate [31]. Select Max is effective at controlling summer annual grass weeds in our region, and these are particularly problematic in young native plant stands [32]. The spring application of Select Max may have killed summer annual grass seedlings that were just emerging in those fall-planted treatments, and this suppression may have allowed more native forbs to establish later that summer.

Additionally, certain herbicides have residual herbicidal activity in the soil and can potentially suppress emerging forbs. Some of the herbicides with residual activity used in this experiment include metsulfuron (component of Pastora and Cimarron) and Plateau/imazapic. We would expect summer-planted plots to be more affected by residual herbicide effects because the herbicides were applied just before many of the forb species would be emerging in early summer. For the fall-sown treatment, the herbicides were applied in April several months before most sown wildflowers would have emerged. However, Pastora/nicosulfuron and Cimarron Plus/metsulfuron showed the opposite trend, indicating that the persistence of Cimmaron/metsulfuron was likely not a concern. Plateau/imazapic has a greater persistence in soil and may remain active up to 120 days, versus ~30 days for Cimmaron Plus [33], and this may account for the greater forb species establishment in fall-sown plots compared to summer-sown plots. However, species sensitivity to each herbicide also must be considered. Plateau/imazapic is labeled for the establishment of certain forb species seeded in our study, including Coreopsis lanceolata, Dalea candida, Dalea purpurea, Echinacea purpurea, Ratibida columnifera, and Rudbeckia hirta. Although these forb species may carry some tolerance to Plateau/imazapic herbicide, other work has found these responses to be variable [17].

We hypothesized that fall plantings might show better establishment compared with summer-planted treatments since fall-sown seeds received a natural cold stratification in the soil while overwintering. A high proportion (50–90%) of wild plants produce seed that is dormant upon maturity, and cold stratification improves germination for many North American forbs [34,35]. Despite this, we did not find consistent evidence that fall planting dates resulted in superior native forb establishment. In a related study, Kubesch et al. [32] evaluated native forbs that were planted either in summer or late fall and found similar establishment success regardless of timing. These results suggest that the effectiveness of fall planting to break seed dormancy and improve the establishment of native forbs may be variable and potentially obfuscated by other factors such as weather variation or unpredictable seedling mortality [6].

Overall, forb establishment in experiment 1 was less successful than in experiment 2, and this result may have been caused by several factors, including a lack of intensive site preparation, a low seeding rate, or insufficient maintenance (e.g., weeding by hand). Experiment 2 involved a more comprehensive site preparation that included two full-rate glyphosate applications and tillage. Plots were also mowed to reduce competition with ruderal plants. We are unable to parse which of these management practices made the greatest difference between our two experiments, but the more intensive management used in experiment 2 probably contributed to better establishment success.

As predicted, higher seeding rates resulted in greater and more consistent target plant establishment in experiment 2. This result conforms to Madison et al. [31], who found in a survey of 20 tallgrass prairie managers that 50% used seeding rates at or greater than 11 kg/ha to help ensure establishment success. At an 11 kg/ha seed rate, our plots exceeded the threshold for successful establishment at >5 forbs/m² [36], regardless of cold stratification. The higher seeding rates likely allow for establishing stands to compensate for some mortality loss due to predation [37] or stochastic events [38].

Cold stratification improved forb establishment, but this effect was more distinct in year 2. The magnitude of cold stratification effects also was species-specific, with some species (e.g., Tridens flavus) being much more abundant when artificially stratified prior to sowing. For other species, the four-week stratification period may not have been long enough to break dormancy; for example, Zizia aurea and Eryngium yuccifolium were more abundant after plots had undergone a period of natural cold stratification. We also note that some non-dormant species such as Dalea purpurea began germinating while being cold stratified, which likely led to increased mortality. Our findings support other research showing the effectiveness of cold stratification can vary depending on species and other environmental conditions. For example, Bratcher et al. [34] evaluated germination responses of five native forbs and found that cold stratification increased the germination potential of all species, but most responded differently to the length of stratification. Chen et al. [39] evaluated how cold stratification and soil moisture affected the seed germination of approximately 20 native species from alpine and desert environments and found that while cold stratification can improve germination, it may depend more on how the species may respond to variation in soil moisture. Overall, despite some variability across species, our results suggest that cold stratification could help improve establishment success and this positive effect may be more valuable when native forbs are sown at lower seeding rates [40].

5. Conclusions

The successful establishment of native forbs to enhance the diversity of pasturelands depends on the effective suppression of existing vegetation. In this study, glyphosate herbicide and tillage proved most effective at suppressing vegetation to allow a reasonable establishment of native forbs. Although we did not examine multiple combinations of vegetation suppression in this study, our results suggest that light tillage combined with timely glyphosate application might be an optimal strategy for vegetation suppression—particularly if tillage can be done early enough to allow weeds from the soil seedbank to emerge and then be eliminated by a subsequent glyphosate application. It is also notable that, of the vegetation suppression methods we tested, glyphosate and tillage are the most inexpensive options.

Results from the other experiment summarized in this paper found that higher seeding rates usually produced better forb establishment up to rates of 56 kg/ha. Cold stratification also improved establishment success in some cases, but this positive effect was inconsistent between years and among species. Although higher seeding rates can provide more consistent establishment success, native forb seed can be very expensive—often exceeding USD 500/ha depending on the species mix. With such high seed costs, planting large areas of pastureland at high seeding rates quickly becomes prohibitively expensive. Our study suggests that comparatively low seeding rates also can be effective, especially after intensive site preparation prior to planting. Overall, we believe a cost-effective approach to establishing native forb stands in existing pasturelands might employ planting at rates no higher than 11 kg/ha following light tillage and multiple glyphosate applications to suppress vegetation.