Research Article |
Corresponding author: Dean E. Pearson ( dean.pearson@usda.gov ) Academic editor: Bruce Osborne
© 2024 Dean E. Pearson, Yvette K. Ortega, Ylva Lekberg.
This is an open access article distributed under the terms of the CC0 Public Domain Dedication.
Citation:
Pearson DE, Ortega YK, Lekberg Y (2024) Invaders break assembly rules to beat the natives: how cheatgrass cheats. NeoBiota 96: 299-324. https://doi.org/10.3897/neobiota.96.129679
|
Understanding how some introduced plants achieve invasive status while most simply become naturalized is a fundamental question in invasion ecology. Traditional approaches comparing native and introduced plants have linked ruderal traits such as annual life history, high fecundity, and rapid growth rates to invasiveness. However, they do not explain why other introduced species bearing similar traits fail to become invasive, possibly because generic comparisons ignore local processes that drive community assembly. Herein, we contrasted native and introduced annuals in the context of local successional processes to elucidate how introduced annual bromes like cheatgrass (Bromus tectorum) overtake perennial grasslands in the intermountain western United States. We created disturbed plots and seeded them first with annuals representing natives, naturalized species, or invasive bromes. We then seeded plots with native perennial community dominants to examine how the different annuals influenced succession. Native annuals established transient populations that facilitated perennial establishment compared to unseeded controls, enabling the shift to perennial dominance. Naturalized annuals mirrored the natives, but invasive annuals maintained robust populations at high biomass that inhibited perennial establishment and impeded succession. Mechanistically, invasive annuals reduced soil moisture and elevated plant biomass, litter, and soil N. However, only litter abundance correlated with perennial seedling recruitment across treatments. Overall, litter showed a unimodal relationship wherein lower litter abundance associated with native and naturalized annuals appeared to facilitate perennial seedling establishment while higher litter levels generated by invasives appeared to suppress perennial establishment and inhibit succession. Additional experiments provided little support for the roles of pathogen spill-over or plant-soil feedbacks favoring the introduced bromes. The domination of perennial grasslands by annual bromes may be driven by litter buildup that allows these introduced plants to break local succession rules by acting as both early seral and climax species. Traits like litter accumulation may strongly influence invasion outcomes but are indistinguishable using trait comparisons lacking community context.
Annual bromes, community assembly, ecosystem engineering, extended phenotype, grass-fire cycle, invasive plants, litter, naturalized plants, pathogen spillover, plant-soil feedback, succession
Some introduced species cause inordinate ecological and economic damage (
Succession theory offers the longest-standing framework for explaining plant community assembly (
If we apply this succession framework as a benchmark for understanding the rules by which native plant communities assemble (sensu
Mechanistically, introduced plants can impact communities through a variety of processes (see
Within western North American grasslands, cheatgrass (Bromus tectorum) and other annual bromes (Bromus spp.) are notorious invaders that have overtaken vast regions of perennial grasslands and shrublands (
We explored the above ideas within Intermountain Grasslands of the western United States that are susceptible to takeover by introduced annual brome grasses (
We established our main experiment within a fenced space at Diettert Gardens on the University of Montana campus, Missoula, MT, USA in 2014 (46.841981, -113.992030). The site reflects abiotic conditions found in the adjacent intermountain bluebunch wheatgrass habitat (
We prepared the site by watering for two weeks to initiate seed germination and purge the seedbank in August 2015. We then applied Roundup® to kill actively growing plants in September. In October, after the herbicide had degraded, we delineated 100 0.7 × 0.7 m (0.5-m2) plots with 0.8-m spacing (separated by weed cloth to maintain plot integrity) in a rectangular array and simulated disturbance in each plot by digging the soil to a depth of ~15 cm and removing plant biomass. We then sowed each plot with 1000 seeds of one of nine focal annual species, with ten replicate plots per species randomly assigned to represent the following treatments: 1) natives (Collinsia parviflora, Clarkia pulchella, or Plantago patagonica), 2) naturalized taxa (Alyssum alyssoides, Arenaria serpyllifolia, Filago arvensis, or Veronica verna), and 3) invasives (Bromus japonicus or B. tectorum). The remaining 10 plots were left unseeded to represent controls. In September 2016, after one growing season, we seeded all plots with three species of native perennial grasses (Festuca idahoensis, Koeleria macrantha, and Pseudoroegneria spicata) and three species of native perennial forbs (Achillea millefolium, Balsamorhiza sagittata, and Lupinus sericeus), all representing climax species that dominate undisturbed grasslands in this system (seeding rates were 100, 150, and 200 seeds per plot for large, medium, and small-seeded species, respectively, to reflect natural variation in fecundity [sensu
In the first growing season after seeding perennials, we documented recruitment by counting seedlings in each plot twice during the main germination window, May-June 2017. We carefully removed perennial seedlings after the final count and then reseeded the plots with the same species in September 2017, following the prior year’s methods. This approach allowed us to quantify perennial recruitment over two growing seasons. In spring 2018, we counted perennial seedlings as done previously, but this time we left plants to grow. To represent recruitment per plot and year in analyses, we took the maximum count across the two surveys for each sown perennial species and summed these across species. Late in the growing season of 2019, we harvested all plots for aboveground biomass. Biomass of live focal annuals, sown perennials and litter (dead material from past years) was separated and dried for 48 hours at 62 °C before weighing.
To track populations of individual focal annuals in the years preceding harvest, 2016–2018, we visually estimated cover of both seeded and colonizing species in each plot (to the nearest 1% if < 10%, and nearest 5% if > 10%) aided by a frame marked in 1% increments. For examination of factors influencing perennial recruitment, we considered total focal annual cover summed across all functionally similar species per plot. For native, naturalized, and control treatments, this metric included focal native and naturalized annuals directly sown in the plots and those colonizing from other plots, as justified by comparison of perennial recruitment among plots established with these taxa (Suppl. material
To understand how focal annuals might affect abiotic conditions, we measured soil properties in each study plot in 2018. We extracted four soil cores (2.5 cm diameter x 10 cm deep) from each plot in mid-May. Soil cores were pooled by plot and sieved through a 2 mm sieve before lab analyses. We used a subset of each fresh soil sample to quantify available N (NO3− and NH4+) via KCl extraction (
The invasive annuals B. tectorum and B. japonicus host the seed pathogen Pyrenophora semeniperda which can spillover onto and suppress germination of native grasses (
We initiated two PSF experiments in the greenhouse. The first experiment was designed to evaluate feedbacks of individual focal annuals on themselves and the second was designed to assay feedbacks from the focal invasive, B. tectorum, on the native perennials sown in our field experiment. This second experiment focused on B. tectorum to simplify logistics. Because we were interested in both abiotic and biotic PSFs, we followed the approach outlined in
In round 1 of the first experiment, each focal annual species sown in the field experiment was grown for three months (Nov-Jan) in 650 ml pots pots filled to approximately 500 ml with sieved (3 mm) soil collected to 10 cm deep from the experimental site and mixed with heat-treated sand and Turface (2:1:1, v:v:v). In round 2 of this experiment, seeds of the same species were planted in conspecific-trained soil (8 replicates per species in each round of the experiment for n = 72 pots per round). In both rounds, we grew three plants per plot after weeding out extra plants from the initial seeding effort and watered with tap water as needed. At the midpoint of each round, all pots received 20 mL of a 0.5 g/L 20-2-20 (N-P-K, Peters Professional fertilizer, JR Peters, Inc., Allentown, PA, USA) solution to address nutrient limitation indicated by yellowing of leaves. At the end of each round, we harvested shoots, dried them for 48 hours at 62 °C, and weighed them. PSFs in this experiment were calculated by comparing biomass between rounds 1 and 2. To address possible differences in environmental conditions between rounds, we also grew each species in an inert medium (1:1 mix of Black Gold® Seedling Mix and Miracle Gro® Seed Starting Potting Mix) during each round (n = 3 replicates/species). In the second experiment, we first trained soils in round 1 by growing either B. tectorum or non-Bromus annuals (each of seven focal species) in pots for three months using the protocol described for the first experiment. In round 2, each of the 6 perennial species sown in our field experiment was grown either in the B. tectorum-trained soils (8 replicates/perennial species for 48 total pots) or in soils trained by non-Bromus annuals (7 replicates/perennial species comprised of 1 pot per non-Bromus species for 42 total pots). A single perennial plant was grown per pot after weeding out extra plants from the initial seeding effort, and plants were watered as needed with tap water. After two months, we harvested shoot biomass as described above. To evaluate PSF in this experiment, we compared biomass of perennials grown in soil trained by B. tectorum vs. by non-Bromus annuals. In both experiments, pots were not root-bound at the end of either round, and the few plants that died were excluded from analyses (n = 4 and n = 8 plants from each experiment, respectively).
For our field experiment, we used generalized linear models (GLMs) in SAS (PROC GLIMMIX,
We also evaluated the relationship between recruitment of native perennials and conditions in plots to explore potential mechanisms governing succession in our field experiment. To do so, we used GLMs with perennial seedling counts from 2018 (the cohort linked to sampling of soil properties and to final biomass measures in the subsequent year) as the response fit with a negative binomial distribution. Based on observed differences among treatments, we opted to construct separate models for 1) plots established with “non-invasive” annuals (native or naturalized treatments) or eventually colonized by these species (controls), and 2) invasive annual plots. This allowed us to take a simple approach to examining whether mechanisms of perennial establishment might differ for communities dominated by either non-invasive or invasive annuals. Model covariates represented measured biotic and abiotic conditions from 2018, and all were considered in the same multivariate model to isolate the independent influence of each covariate (i.e., when variation attributable to other covariates was accounted for). These covariates were total focal annual cover, litter cover, soil moisture, and soil NH4+ and NO3− (remaining soil properties showed minimal differences among treatments; see Results). We also included an additional measure of litter abundance, litter biomass, in models because litter properties differed between non-invasive/control and invasive annual plots. For plots dominated by non-invasive species, litter cover captured within-year variation in litter abundance (e.g., litter cover and biomass measured in 2019 were significantly correlated: r = 0.12, P = 0.002), which was largely two-dimensional. However, in invasive plots, old plant material accumulated in three dimensions and litter cover did not suffice to capture variation in litter quantity among plots (e.g., litter cover and biomass measured in 2019 were not significantly correlated: r < 0.1, P = 0.18; see Results). Therefore, in addition to including litter cover in models (as measured in 2018 to align with seedling counts), we also included litter biomass (only measured in 2019). We screened for potential multicollinearity issues by testing whether model covariates were highly correlated in bivariate space (r > 0.9;
For the greenhouse experiment designed to compare PSFs of individual focal annuals among native, naturalized and invasive taxa, we treated shoot biomass (mean of three plants per pot) as the response in a generalized linear mixed model in SAS (PROC GLIMMIX,
Focal annuals established rapidly and formed monocultures in plots where sown in the first year after seeding, 2016 (Fig.
Abundance of seeded and colonizing focal annual species by treatment. Given is mean (+ 1 SE) cover per focal species seeded into experimental plots to represent native, naturalized, or invasive annuals, 2016–2018. Plots were disturbed and purged of plants prior to seeding in fall 2015, and control plots were not seeded. Sown native and naturalized annuals were allowed to colonize plots where unseeded, and mean (+ 1 SE) cover per species across this set of colonizing taxa is given for comparison to seeded species cover. Within-year patterns for each variable were evaluated with post-hoc comparisons when the treatment effect was significant (P < 0.05), and means that do not share letters (seeded species: lower case, colonizing species: upper case) are significantly different.
Invasive annuals suppressed recruitment of sown native perennials in both 2017 (F3,96 = 15.3, P < 0.001) and 2018 (F3,96 = 24.1, P < 0.001; Fig.
Recruitment of native perennials and final community composition by treatment. Native perennials were seeded into plots representing native, naturalized, or invasive annuals, and controls, and their mean abundance (+ 1 SE) was measured by a seedling recruitment in 2017 and 2018, and b biomass relative to that of focal annuals at the end of the study in 2019. Sown native and naturalized annuals were allowed to colonize plots where they were not seeded and are included in focal annual biomass in all cases except the invasive treatment, where they were a minor component (Fig.
This pattern of recruitment translated to marked differences in community composition in 2019 as measured by final plot biomass, with invasive annuals impeding succession toward native perennial dominance (Fig.
Both biotic and abiotic conditions differed among treatments, potentially influencing patterns of native perennial recruitment. Total focal annual cover accounted for the combined abundance of those annual species seeded into plots and those sown native and naturalized taxa colonizing from other plots for all but the invasive treatment, where colonization was negligible. Total focal annual cover was greatest in the invasive species treatment in two of three years despite the boost given to remaining treatments by colonizing taxa, and it was generally lowest in the control treatment due to the lack of initial seeding (Fig.
Total focal annual cover and litter in treatments representing native, naturalized, or invasive annuals, and controls. Given is mean (+ SE) a total cover of focal annual species, 2016–2018 b litter cover, 2016-2018 and c litter biomass at the end of the experiment in 2019. Sown native and naturalized annuals were allowed to colonize plots where they were not seeded and are included in total focal species cover in all cases except the invasive treatment, where they were a minor component (Fig.
Litter cover was limited to trace levels (<1%) across treatments in 2016, the first year after annuals were seeded, but in subsequent years, treatments differed markedly (Fig.
Invasive annuals altered soil conditions, as measured in 2018. NH4+ concentration was > 50% higher on average in invasive annual plots relative to those established with native or naturalized annuals (F3,94 = 6.4, P < 0.001), and NO3− concentration was on average > 75% higher (F3,93 = 7.8, P < 0.001; Fig.
Soil conditions in treatments representing native, naturalized, or invasive annuals, and controls. Given is mean (+ SE) a available ammonium (NH4+) content b available nitrate (NO3−) content, and c soil moisture (volumetric water content); as measured in 2018. For each variable, means that do not share letters are significantly different, as evaluated with post-hoc comparisons when the treatment effect was significant (P < 0.05).
To consider how these changes in biotic and abiotic conditions might influence patterns of native perennial recruitment, we modeled the relationship between seedling counts from 2018 and measured covariates (Suppl. material
Relationships between recruitment of perennial seedlings in 2018 and litter abundance. Litter abundance was measured by a litter cover for treatments representing native or naturalized annuals and controls, b litter biomass for the invasive annual treatment, and c a principal component combining both litter metrics to visualize the unimodal relationship across all treatments (this relationship was also significant when we ran a parallel model using 2017 data: F1,97 = 17.5, P < 0.001). Photo shows a perennial forb (Balsamorhiza sagittata) seedling emerging amidst sparce litter in a naturalized plot. Note that recruitment was modeled with a negative binomial distribution in all cases and associated predicted relationships have been back-transformed to the original scale.
To depict the linkage between native perennial recruitment and litter abundance across all treatments, we combined litter cover and litter biomass measures into a principal component and treated this as covariate in a model that also included a quadratic term to account for the observed shift in pattern at low vs high litter abundance. This exercise showed a significant unimodal relationship wherein perennial seedling recruitment increased at lower litter abundances represented by non-invasive annual and control treatments but decreased at higher litter abundances represented by the invasive annual treatment (F1,97 = 21.4, P < 0.001; Fig.
For the dominant native perennial grass, P. spicata, emergence of marked seeds planted into field plots in fall 2018 was reduced by > 25% in invasive annual vs other treatments (F3,61 = 10.0, P < 0.001), with no differences among the latter (Suppl. material
In the greenhouse experiment designed to evaluate PSFs of individual focal annuals, growth across all species in round 2 was 33% greater than in round 1 (F1,68 = 94.1, P < 0.001) while growth in inert soil did not differ between rounds overall (F1,16 = 0.7, P = 0.41), indicating that plants tended to generate soil conditions that favored subsequent growth. However, this effect on focal annual growth did not differ among native, naturalized, and invasive annuals (round x annual type: F2,68 = 0.7, P = 0.41; Suppl. material
Explaining how some introduced plants become problematic pests is a central question in invasion ecology. Here, we explored this question by contrasting the roles that native, naturalized, and invasive annuals play in the context of local succession rules in perennial grasslands. We found that invasive annual bromes, including both cheatgrass and Japanese brome behaved quite differently from native and naturalized annuals. Native and naturalized annuals both developed ephemeral populations and facilitated establishment of the climax perennial natives, transitioning the community from early seral annuals to perennial dominance, consistent with succession theory. In contrast, the invasive bromes established robust populations that strongly inhibited recruitment of native perennials, thereby impeding succession. In essence, the invasive annuals acted as both early seral and climax species, thereby breaking local assembly “rules” for succession. As we discuss below, framing our experiment in the context of local successional processes allowed us to identify traits like litter accumulation that may facilitate invasiveness but are not readily apparent using traditional trait comparisons.
In establishing the baseline for our system, we found that native annuals exhibited two functional behaviors that transitioned the community toward climax. First, they failed to maintain community dominance where sown, readily ceding space to other species (Fig.
In evaluating plausible explanations for these patterns, we examined covariates in the succession experiment and conducted additional experiments testing for pathogen spillover and PSFs (Fig.
Potential mechanisms allowing invasive annual bromes to inhibit grassland succession, as explored in our study. Summarized are 1) significant effects of invasive bromes (Bromus japonicus and B. tectorum) on biotic and abiotic factors, measured relative to treatments established with native and naturalized annuals, and 2) primary linkages between studied factors and recruitment of sown native perennials revealed through multivariable modeling and secondary experiments testing for pathogen spillover and PSFs (greenhouse only). Signs indicate direction of effects. Note that pathways are not necessarily independent. Collective effects of invasive annual bromes, including excessive litter buildup, prevented the transition to native climax perennials and maintained communities in a novel seral-climax state (see Discussion for details).
Of course, the link between litter and perennial recruitment is potentially confounded with litter chemistry since the species producing the litter differed among treatments. In terms of litter quality, many introduced plants have higher leaf N (low C:N ratio) than co-occurring natives, which has been linked to increased litter decomposition rates and elevated soil N (
Numerous studies have linked annual bromes to elevated soil N (
Despite the potential for PSFs to strongly influence both succession and invasion processes, we found no evidence in our greenhouse experiments that PSFs differentially affected growth of Bromus relative to other annuals or that growth of native perennials was suppressed in soil trained by cheatgrass vs. other annuals. Whereas we might expect all the annuals to have negative feedbacks as predicted in the context of succession (
Pathogen spillover from annual bromes by black fingers of death (Pyrenophora semeniperda) may periodically suppress native grass establishment in invaded stands (
Cheatgrass is the most notorious invasive plant in the western United States (
Traditional approaches to understanding how some introduced species become problematic pests have compared traits among invasive, naturalized, and native species without reference to the local assembly rules that define recipient communities. Such studies have broadly linked invasive species to traits like high fecundity and rapid growth (Pysek and Richardson 2007;
We thank Morgan McLeod for conducting soil analyses and Alex Aromin, Linden Beegle, Ashley Braae, Hillary Cimino, and Hilary Schultz for valuable field support. The University of Montana provided access to Diettert Gardens and greenhouses.
The authors have declared that no competing interests exist.
No ethical statement was reported.
DEP and YKO were supported by the Rocky Mountain Research Station, USDA Forest Service. MPG Ranch provided support for YL.
DEP and YKO initiated the project and developed and implemented the field experiment. YL led in developing the plant-soil-feedback experiments, soil nutrient sampling, and evaluation of seed pathogens. YKO analyzed the data. All authors contributed to the writing.
Data is available in Dryad: https://doi.org/10.5061/dryad.9ghx3fft6.
Additional results and study design information for field and greenhouse experiments
Data type: docx