Abstract
Rangelands degraded by non-native annual grasses are commonly restored by integrating herbicide with fall seeding of native perennial grasses. Seeds are expected to remain dormant throughout winter and emerge in spring. However, restoration often fails due to annual grass seedlings out-competing seeded grasses because they emerge sooner and grow faster. At two sites in Montana, we tested eight seeding dates of Pseudoroegneria spicata to determine the effect on resistance to invasion by the annual grass Bromus tectorum. After one growing season, fall (November) and early spring (April) seeding resulted in larger P. spicata than later spring seeding (May) (829–1,180 tillers per m2 versus 349–561 tillers per m2) at one site, while seeding date had no effect on P. spicata at the second site. In the second season and after B. tectorum introduction, fall and early spring seeded P. spicata was more resistant to B. tectorum compared to late spring seeded. For example, B. tectorum was less dense (480 ± 101 tillers per m2) in P. spicata stands seeded 7 April than 12 May (943 ± 188 tillers per m2), and B. tectorum biomass was lower in fall-seeded stands (227 ± 81 g per m2) than stands seeded 12 May (533 ± 206 g per m2). Our results demonstrate that establishment and subsequent invasion resistance is enhanced by seeding native grasses from fall to early spring, presenting a wider timeframe for seeding than commonly practiced.
Restoration Recap
We seeded the native perennial grass Pseudoroegneria spicata in November 2015 and weekly from April to early May 2016 to determine which seeding date resulted in the best stand establishment and greatest potential to resist invasion by the invasive annual grass Bromus tectorum.
Pseudoroegneria spicata stands seeded in fall (November) and early spring (April) had higher plant density and biomass than those seeded later in spring (late April through May).
Bromus tectorum was less abundant in P. spicata stands seeded in fall and early spring than in stands seeded in late spring, suggesting larger perennial grasses can limit invasion of annual grasses.
Seeding P. spicata after 206 growing degree days (GDD, first week of May) is not advised as there was a trend of decreasing P. spicata and increasing B. tectorum abundance after this date.
Land managers may have a wide window of opportunity for seeding native grasses, ranging from a fall dormant seeding through early spring, but care should be taken to not seed too late in spring.
Degraded range and wildlands are difficult to revegetate due to the persistence of non-native, annual grasses or secondary invasions following control of the target annual grass (Pearson et al. 2016, Schantz et al. 2016). In habitats dominated by non-native, invasive annual grasses, their seedlings can outcompete native perennial grass seedlings due to early emergence of the annual grasses (Wainwright et al. 2012). Earlier emergence by these invasive winter annual grasses compared to native plants allows for primary access to water and nutrients (Mangla et al. 2011a, Dickson et al. 2012). These benefits are amplified after germination and emergence because ruderal invasive annual grasses grow faster than native perennial grasses, compounding preemptive use of essential resources (Grotkopp and Rejmánek 2007, James and Drenovsky 2007, James 2008). Being the first to fill a vacant ecological niche is known as a priority effect (Wainwright et al. 2012). Seed banks of lands infested by invasive annual grasses are often dominated by those same species, limiting the recruitment of native plants.
Strategically modifying the seeding date of native plants to mitigate seasonal priority effects of invasive annual grasses during revegetation can be an ecologically-based management approach for improving native species recruitment and establishment (James et al. 2011, Farrell and Fehmi 2018). For example, the native grass Pseudoroegneria spicata (bluebunch wheatgrass) suppressed the non-native, invasive annual grass Bromus tectorum (cheatgrass) in a greenhouse setting when seeded four weeks prior to seeding of B. tectorum (Orloff et al. 2013, Larson et al. 2018). Similar studies also support the effectiveness of seeding early to improve native grass establishment over invasive annual plants (Grman and Suding 2010, Boyd and James 2013). In the field, Boyd and Lemos (2015) attributed higher seedling survival of fall versus spring seeding to more established root systems from additional growing degree days (GDD). However, the success of direct seeding from fall versus spring seeding is still largely unknown in relation to its potential for resisting subsequent weed invasion.
During revegetation of invasive annual grass-infested areas in the semi-arid western United States, seeding of native perennial grasses like P. spicata often occurs in the fall, following a summer or fall herbicide application aimed at controlling annual grasses. Seeds are expected to remain dormant throughout winter and emerge in the spring at the earliest opportune time (Boyd and James 2013). However, fall dormant seeding has its limitations. First, fall dormant seedings expose seeds to predators and pathogens for several months before the seeds are expected to germinate, and some research suggests that seeds of native species may be more susceptible to predation than those of non-native species (Lucero and Callaway 2018). Furthermore, unanticipated germination prior to spring can put seedlings at risk of failure during winter freeze-thaw cycles (James et al. 2012). Revegetation failure can also result from the slow-growing nature of native perennial grasses, which do not typically reach reproductive maturity in the first growing season in contrast to quickly-growing annual grasses that complete their life cycle in less than a year (Schantz et al. 2016). Finally, though fall dormant seedings are common practice (Monsen and Stevens 2004), they eliminate the option of a spring herbicide application to control spring-emerging invasive grasses because such an application would also injure seedlings of the seeded species.
Delaying seeding until spring, as an alternative to seeding in the fall, has had varied success in terms of density and biomass of seeded perennial grasses (Schantz et al. 2015, 2016). Density of surviving seedlings from spring seeding can be similar or greater than that of fall seeding (Boyd and Lemos 2015). Spring seeding on a site where annual grass propagule pressure was low (between 150– 1,500 seeds per m2) and irrigation was available during the first growing season, resulted in satisfactory perennial grass establishment (Schantz et al. 2016). However, in semi-arid climates where irrigation is not available, delaying seeding until spring can be risky. Seedlings emerge, but may not grow enough, particularly in terms of root length, to survive seasonal drought (Schantz et al. 2016). Late spring-seeded perennial grasses also experience fewer GDD and accumulated precipitation than fall or early spring seedings (Boyd and Lemos 2015). Addressing how late in the spring seeding can occur while still resulting in satisfactory native perennial grass establishment is necessary to improve revegetation of invasive annual grass-infested range and wildlands.
The first objective of this study was to examine whether fall or early to late spring seeding resulted in higher establishment of the native perennial grass P. spicata over two growing seasons. Consistent with previous studies (Boyd and James 2013, Boyd and Lemos 2015), we hypothesized perennial grass density and biomass to be highest for fall and early spring-seeded cohorts when compared to late spring-seeded cohorts. We used P. spicata for this study because it is a large-statured, native bunchgrass that is common across mountain and foothill rangelands, is valued as a forage species for livestock and wildlife (Stubbendieck et al. 1997), and in Montana it is often one of the last native perennial grasses to persist in areas invaded by non-native species (J. Mangold, Montana State University, pers. comm.). Several cultivars of P. spicata have demonstrated successful establishment in other studies (Sheley et al. 2001, Sheley et al. 2006, Jacobs et al. 2006, Rinella et al. 2012, Mangold et al. 2015).
The second objective addressed whether P. spicata seeding date affected suppression of the invasive annual grass B. tectorum when it was introduced to the plots at the end of P. spicata’s first growing season. We hypothesized B. tectorum biomass and density would be lowest where P. spicata was seeded in fall and early spring when compared to late spring seeding, due to the predicted increase in density and biomass of P. spicata plants during the first growing season.
Methods
This study occurred at two field sites which were located at the Arthur H. Post Research Farm (45°40′38.01″ N 111° 9′19.49″ W) and Fort Ellis Research Farm (45°40′13.19″ N 110°58′48.13″ W), hereafter referred to as Post Farm and Fort Ellis, respectively, near Bozeman, Montana. Elevation at Post Farm and Fort Ellis are 1,450 m and 1,484 m, respectively. In 2016 and 2017, the total annual precipitation at Post Farm was 32.5 cm and 37.7 cm, and annual minimum/maximum temperatures were 1.21°C/14.8°C and 0.23°C/13.8°C, respectively (Table 1, NOAA 2018). In 2016 and 2017, the total annual precipitation at Fort Ellis was 46.7 cm and 60.7 cm, and minimum/maximum temperatures were 1.06°C/13.8°C and –0.1°C/13.0°C, respectively (Table 1, Weather Underground 2018).
Average spring (March–May), summer (June–August), fall (September–November), and total precipitation (cm) and average minimum and maximum temperatures (°C) for spring, summer, fall and annual at Post Farm and Fort Ellis across a 30-year period and years 2016 and 2017. *30-year averages (1981–2010) were based on the Western Regional Climate Center (WRCC) monthly data for this region and were only available at the Post Farm site
Prior to setting up the study, existing vegetation was manually removed, and the fallow fields were tilled. A completely randomized split-plot design was used to test eight seeding dates of P. spicata (whole plots, 1 m × 2 m with additional 0.5 m buffer) and two levels (absent and present) of B. tectorum competition (split-plots, 1 m × 1 m). Seeding treatments were replicated four times at each site for a total of 32 main plots and 64 split-plots per site.
The eight seeding date treatments were 8 November 2015 (fall) and seven weekly spring seeding dates from 1 April 2016 to 12 May 2016 (early to late spring, Table 2). Seeds of P. spicata (Goldar cultivar) were hand broadcasted evenly on the soil surface at 667 seeds per m2 (following Orloff et al. 2013) for each treatment at each site. Seeds were produced and obtained from Bruce Seed Farm near Townsend, Montana, approximately 100 km from the study sites. June of the first growing season was drier than the 30-year mean precipitation (Western Regional Climate Center 2019) at the Post Farm: we applied the equivalent of 1.27 cm of precipitation in June 2016 to address this deficit. All vegetation other than P. spicata was removed by hand throughout the first growing season. In September 2016, after the first growing season and when plants were in a reproductive state, P. spicata density (tillers per m2) was measured across whole plots (1 m × 2 m).
Seeding dates and cumulative growing degree days (GDD = sum of average daily temperature °C — temperature base 4.4°C between 1 March 2016 to 12 May 2016) for Pseudoroegneria spicata at Post Farm (PF) and Fort Ellis (FE) sites.
After the first growing season, B. tectorum seeds were sown into competition split-plots on 1 September 2016 at 1,112 seeds per m2 (following Orloff et al. 2013), using seeds collected from several sites near Bozeman, Montana. In both years, only target species, P. spicata and B. tectorum, were grown and all other species were removed by hand until data were collected from 23 June 2017–10 July 2017. Three randomly-placed 50 cm × 20 cm frames (Daubenmire 1959) per split-plot were used to sample density (tillers per m2) and aboveground biomass of P. spicata and B. tectorum. Biomass was clipped to 2 cm above ground level, dried at 37°C for 48 hours, and weighed to the nearest 0.1 gram.
An analysis of variance (ANOVA) was used on a general linear model with a Quasi-Poisson distribution to detect differences in P. spicata density (tillers per m2) by seeding date for each site for objective one. One replication of the Spring 5 seeding data at Post Farm was removed from analysis because the plots received run-off irrigation from an adjacent field and produced abnormally high P. spicata densities. In both years, P. spicata density across main plots was used to determine establishment for all seeding cohorts, whereas in the second year P. spicata density in the no competition plots were evaluated. Tukey pairwise comparisons were used to determine differences across seeding date treatments for each site.
For objective two, the effect of seeding date on P. spicata and B. tectorum was analyzed using ANOVA models. Specifically, an ANOVA on a linear mixed effects model was used to determine differences in P. spicata density and biomass, where fixed effects were seeding and competition treatments and site, and random effects accounted for plot-to-plot variation. When analyzing B. tectorum density, an ANOVA was used on a general linear model to determine seeding date treatment effects across sites. Data were over-dispersed, so a Quasi-Poisson distribution was used. When analyzing B. tectorum biomass, an ANOVA was used on a linear model to determine seeding date treatment effects across sites. Tukey pairwise comparisons were used to determine differences between seeding groups and competition levels.
For all analyses, all models were explored for normality, independence, and constant variance assumptions and median values are shared. Graphical interpretations and analyses were conducted using R Software 3.5.1 (R Core Team 2018), specifically dplyr, lme4, emmeans, and ggplot2 packages (Bates et al. 2015, Lenth et al. 2018, Wickham 2016, Wickham et al. 2018).
Results
Pseudoroegneria spicata Density and Biomass
Seeding treatments affected P. spicata density after one growing season, but the effect depended on site (F7,110 = 2.759, p = 0.011). At Fort Ellis, Spring 2 resulted in 1,180 ± 169 tillers per m2 (estimate ± standard error), which was similar to Fall (1,009 ± 176 tillers per m2), Spring 1 (1,038 ± 122 tillers per m2), Spring 3 (856 ± 180 tillers per m2), and Spring 5 (829 ± 223 tillers per m2) (Figure 1). Spring 4, Spring 6, and Spring 7 had similarly low densities at 472 ± 72 tillers per m2, 561 ± 134 tillers per m2, and 349 ± 63 tillers per m2, respectively. At Post Farm, there was no effect of seeding treatment on P. spicata density. Densities ranged from 1,054 ± 47 tillers per m2 in Spring 4 to 1,706 ± 207 tillers per m2 in Spring 1 (Figure 1).
Pseudoroegneria spicata density (tillers per m2 ± SE) after one growing season (2016) at two sites in southwestern Montana. Seeding dates are Fall (11 November 2015), S1 (1 April 2016), S2 (7 April 2016), S3 (13 April 2016), S4 (21 April 2016), S5 (29 April 2016), S6 (5 May 2016), S7 (12 May 2016). Similar letters indicate no difference in density across treatments within a site (α = 0.05).
After two growing seasons, seeding treatment affected density of P. spicata (F7,47 = 3.344, p = 0.005) (Table 3). Spring 7, which resulted in 1,221 ± 158 tillers per m2, was lower than Spring 1 and Spring 2 (1,988 ± 145 and 1,895 ± 110 tillers per m2, respectively) (Figure 2). Density of P. spicata was otherwise similar across seeding treatments and ranged from 1,221 to 1,988 tillers per m2.
Pseudoroegneria spicata and Bromus tectorum density (tillers per m2 ± SE) across seeding treatments after two growing seasons (2017) at two sites in southwestern Montana. Seeding dates are Fall (11 November 2015), S1 (1 April 2016), S2 (7 April 2016), S3 (13 April 2016), S4 (21 April 2016), S5 (29 April 2016), S6 (5 May 2016), S7 (12 May 2016). Similar letters indicate no difference in a species’ density across treatments (α = 0.05).
Analysis of variance (ANOVA) table for Pseudoroegneria spicata density (tillers per m2) and biomass (g per m2) after two growing seasons (2017). Competition indicates presence or absence of Bromus tectorum.
Density of P. spicata was also affected by competition from B. tectorum, where the effect varied by site (F1,54 = 23.63, p < 0.001) (Table 3). At Fort Ellis, the no competition treatment (control) yielded higher density of P. spicata (1,735 ± 122 tillers per m2) than the B. tectorum competition treatment (1,161 ± 107 tillers per m2), while B. tectorum competition had no effect on P. spicata density at Post Farm (mean = 1,748 ± 55 per m2).
After two growing seasons, biomass of P. spicata was affected by seeding treatment (F7,47 = 3.423, p = 0.005) (Table 3). Biomass was lower in seeding treatments Spring 7 and Spring 4 (560 ± 88.9 g per m2 and 578 ± 74.6 g per m2, respectively) compared to Fall (888 ± 77.9 g per m2) (Figure 3). Otherwise, biomass was similar across most seeding treatments (Figure 3).
Pseudoroegneria spicata and Bromus tectorum biomass (g per m2 ± SE) across seeding treatments after two growing seasons (2017) at two sites in southwestern Montana. Seeding dates are Fall (11 November 2015), S1 (1 April 2016), S2 (7 April 2016), S3 (13 April 2016), S4 (21 April 2016), S5 (29 April 2016), S6 (5 May 2016), S7 (12 May 2016). Similar letters indicate no difference in a species’ biomass across treatments (α = 0.05).
Biomass of P. spicata was affected by competition treatment, and the effect varied by site (F1,54 = 9.731, p = 0.003) (Table 3). At Fort Ellis, the no competition treatment yielded higher P. spicata biomass (687 ± 57.2 g per m2) than the B. tectorum competition treatment (455 ± 51.8 g per m2). Competition from B. tectorum had no effect on P. spicata biomass at Post Farm (mean = 859 ± 31.2 g per m2).
Bromus tectorum Density and Biomass
There was a difference in B. tectorum density due to the main effects of site (F1,54 = 173.2, p < 0.001) and seeding treatment (F7,55 = 4.500, p < 0.001; Table 4). Fort Ellis had higher density of B. tectorum (1,019 ± 59 tillers per m2) than Post Farm (356 ± 29 tillers per m2). Averaged across sites, B. tectorum density was higher in Spring 7 (943 ± 188 tillers per m2) than Fall (553 ± 88 tillers per m2) and Spring 2 (480 ± 101 tillers per m2) (Figure 2). All other seeding treatments were similar to each other.
Biomass showed a similar pattern to density: there was a difference in B. tectorum biomass across the main effects of site (F1,47 = 158.0, p < 0.001) and seeding treatment (F7,47 = 2.397, p = 0.035; Table 4). Fort Ellis had higher B. tectorum biomass (636 ± 53.4 g per m2) than Post Farm (26.3 ± 6.0 g per m2). Averaged across sites, B. tectorum biomass was higher in Spring 7 (533 ± 206 g per m2) than Fall (227 ± 80.7 g per m2) (Figure 3). All other seeding treatments were similar to each other.
Discussion
Fall (i.e., November) and early spring (i.e., April) seeding generally resulted in greater establishment of P. spicata after one and two growing seasons than late spring seeding, as has been observed in other studies (Boyd and Davies 2012, Boyd and James 2013, Rinella and James 2017). The effect of seeding date was more pronouced in the first year at Fort Ellis, with the exception of Spring 5 at both sites. We believe the Spring 5 seeding cohort benefited from 2.2 cm of snow that fell from 23 April to 28 April 2016 (NOAA 2018) across both sites. The legacy of seeding date was less pronounced in the second year, as would be expected, though greater establishment of P. spicata in the fall (i.e., biomass) and early spring (i.e., density) than late spring was evident across the two sites. Our results generally support the first hypothesis and previous studies, indicating fall and early spring seeding produce larger perennial grasses than late spring seeding dates (James et al. 2012, Boyd and Lemos 2015). We recognize that other native grasses may differ in optimal seeding time, but given that P. spicata is arguably one of the most important native grasses on western rangeland, our study provides important information on best practices for its establishment.
The second hypothesis that B. tectorum density and biomass was expected to be lowest where P. spicata was highest in fall and early spring seeding was also generally supported. In particular, the high density of P. spicata in the Spring 2 treatment (i.e., seeded 7 April) had the greatest competitive effect on B. tectorum density. Pseudoroegneria spicata biomass also was greatest in the Fall as compared to late spring (i.e., seeded 21 April and 12 May) seeding which subsequently led to lower B. tectorum biomass. This study also indicates that seeding after the first week of May (above 206 GDD; Table 2) is not ideal as there was a trend of decreasing P. spicata abundance and subsequent increase in B. tectorum abundance at the latest spring seeding date. This trend was consistent across both sites, suggesting results found here may hold true across locations that have similar climatic conditions as southwestern Montana.
Across both sites there was no competition by seeding date interaction which suggests once P. spicata is established, the effect of competition from B. tectorum is lessened. We did not measure resource availability (e.g., soil moisture or nutrient concentrations), and we acknowledge that competition in fallow and tilled fields used in this study is not like that in impacted rangelands. Furthermore, establishment of P. spicata was higher than that seen in studies that occurred in range or wildlands infested with B. tectorum or other invasive plants (Whitson and Koch 1998, Sheley et al. 2001, Mangold et al. 2015). However, our results support that P. spicata, once established, can be highly competitive against secondary invasion (Rinella et al. 2012, USDA-NRCS 2012). Providing seeded species with a seasonal priority effect can shift competitive dynamics in a positive direction (Orloff et al. 2013), and along with other research suggests large and robust perennial grasses can limit invasion by annual grasses (Whitson and Koch 1998, Humphrey and Schupp 2004, James et al. 2008, Davies and Johnson 2017, Young et al. 2005).
Despite evidence suggesting precipitation is a key factor influencing revegetation success in semi-arid regions (Hardegree et al. 2013, Hardegree et al. 2016), wetter conditions at Fort Ellis compared to Post Farm did not have a substantial impact on P. spicata establishment and competitive dynamics with B. tectorum. In fact, abundance of P. spicata was lower at Fort Ellis than Post Farm; and, perhaps accordingly, abundance of B. tectorum was higher at Fort Ellis. Other revegetation studies have failed to find a beneficial effect of additional moisture on seedling success. For example, experimental observations in Oregon found no effect of additional watering on final density and biomass of P. spicata that was seeded fall (Boyd and James 2013) or spring (James and Svejcar 2010). However, it should be noted that in our study there was more variability at Fort Ellis in the response of P. spicata density to seeding date during the first year of the study, and there was an effect of the competition treatment during the second year of the study. Of further note, annual precipitation at Fort Ellis during our study was higher than the 30-year average precipitation and within the recommended precipitation zone (at least 30.5 cm) for the cultivar of P. spicata used in this study (USDA-NRCS 2012). Of the recommended USDA cultivars of P. spicata, “Goldar,” which was used in this study, is well adapted for rapid growth under dry conditions at or above 30.5 cm of annual rainfall above 1,000 m elevation (USDA-NRCS 2012).
One of the foremost risks to seeding a degraded site with native species is competition from invasive seedlings (Ray-Mukherjee et al. 2011). Decreasing seeds of undesirable species in the seed bank to a level that allows native species to establish (less than 1,500 seeds per m2; Schantz et al. 2016) can prove difficult when natural seed rain of B. tectorum can be as high as 12,000 seeds per m2 (Humphrey and Schupp 2001). Applying management tactics that eliminate or reduce seed production, like herbicides or targeted grazing, over multiple years or using herbicides with multiple years of annual grass control (Sebastian et al. 2016, Sebastian et al. 2017a, Sebastian et al. 2017b) can reduce the seed bank of invasive annual grasses and minimize competition with seeded species. Continued practice of fall dormant seeding at the recommended rate (approximately 17 pounds per acre for P. spicata ) (Sheley et al. 2008) is supported by this study, but practitioners should consider early spring seeding as an option as well. Our findings suggest that land managers can delay seeding until early spring without compromising native grass establishment. Spring seeding offers the option of applying a non-selective, non-persistent herbicide like glyphosate to control B. tectorum seedlings that emerged over the winter and into early spring (Kyser et al. 2012, Kyser et al. 2013), eliminating competition for at least one growing season. If seeding is delayed until spring, though, seeding after the first week of May (or GDD greater than 206) is not recommended as it could result in decreased establishment.
Our results show that providing native perennial grasses with a seasonal priority effect can improve their resistance to invasion by weedy annual grasses. The success of varied seeding dates indicates a wide range of opportunity available to land managers when seeding native grasses with fall (November) and early spring (April) seeding producing favorable results. However, we also found a trend of decreasing P. spicata, as well as increasing B. tectorum, when cohorts were seeded after 206 GDD. Further studies could help consolidate the optimum range of P. spicata seeding dates for different regions across the West. In addition, testing the effect of seeding date for other important native grass species of mountain and foothill rangelands in the West is also encouraged.
Wood pewee. Chambers W. and R. Chambers. 1881. Encyclopaedia - A Dictionary of Universal Knowledge for the People. Philadelphia, PA: J.B. Lippincott & Co. The Florida Center for Instructional Technology, fcit.usf.edu.
Acknowledgments
Funding for this project was provided by Montana Noxious Weed Trust Fund, and L.J.R. and J.M.M. are supported by NIFA-USDA Hatch MONB000363 and MONB00359, respectively.
This open access article is distributed under the terms of the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0) and is freely available online at:http://jhr.uwpress.org
