Final Report for GW15-006
Project Information
Biological soil crusts (biocrusts) provide many ecosystem services to rangelands, but their role in grass establishment is unclear. We conducted a series of experiments to determine the extent to which nativity and seed attributes (e.g., size, presence/absence of awns) influence grass establishment outcomes on biocrusted and non-crusted soils and whether a restored biocrust community can confer resistance to invasion by non-native grasses. The experiments were conducted in contrasting bioclimatic regions to assess the robustness of our findings. Placing seeds in cracks of biocrusts facilitated seed-soil contact and resulted in high germination/emergence in semi-controlled environments. This supports the notion that species with small seeds and or seeds with small or no appendages may be better suited to exploit cracks in biocrusts. In addition, native species with small seeds emerged at higher rates than native species with larger seeds on intact biocrusts in a Sonoran Desert semi-controlled environment. Larger seeds also had longer awns, but contrary to predictions, removing awns decreased emergence. In field settings, higher rates of germination and early-establishment on broken biocrusts may be due to enhanced seed capture/retention, seed burial, or biocrust nutrient release.
Introduction
Biocrusts (biological soil crusts) generally cover open spaces on soils in rangelands of the American West. Communities of cyanobacteria, lichens, mosses, and other organisms that live on and in the soil surface, biocrusts contribute to ecosystem services such as soil stability, water infiltration, and carbon-nitrogen and fixation (Belnap 2003). Biocrust microtopography, lichen, and bryophyte abundance increase with decreasing potential evapotranspiration (Belnap 2003). In hot deserts, biocrusts typically have up to 3cm of microtopography (Belnap 2003) with up to 5cm of relief observed in the Mojave Desert. (Williams et al. 2012). In cold deserts, biocrusts are pinnacled and have up to 15cm of relief, likely due to freezing of soils (Belnap and Gardner 1993). The microtopography generated by biocrusts may affect vascular plant composition via its influence on seed capture (relatively low on smooth biocrusts; higher on biocrusts with more microtopography (Belnap et al. 2003).
Biocrusts have variable effects on native plant germination and establishment (Belnap et al. 2003). Observational field studies and experiments in controlled environments often show a negative effect of biocrusts on the germination and abundance of cheatgrass (Bromus tectorum) and other annual exotic grasses (Belnap et al. 2003). Biocrust and cheatgrass abundance in the Great Basin are negatively correlated (Peterson 2013). Conversely, in the Mojave Desert Bromus spp. density is higher on biocrusts compared to bare soil (DeFalco et al. 2001) suggesting that biocrust microtopography and species composition may affect establishment outcomes. Some Great Basin biocrusts can inhibit cheatgrass establishment in greenhouse (Howell 1998) and growth chamber (Deines et al. 2007) settings. However, the effect depends on biocrust community composition with some lichens inhibiting establishment and others being neutral (Deines et al. 2007). Disturbance of biocrusts can increase Bromus spp. germination (Kaltenecker et al. 1999, Belnap et al. 2003).
Several authors speculate that the species-specific outcomes are influenced by seed morphology (Zhang and Belnap 2015), seed size (Briggs and Morgan 2011, Kitajima 2007) and biocrust species composition (Deines et al. 2007). Along these lines, we seek to quantify how biocrust characteristics (dominant species, microtopography, integrity) and seed morphology (size, appendages) interact to determine grass germination and establishment.
Our goal was to determine the extent to which nativity and seed attributes (e.g., size, presence/absence of awns) influence grass establishment on biocrusted and non-crusted soils and whether a restored biocrust community can confer resistance to future invasions. Our objectives were to: 1) quantify biocrust influence on native vs. non-native grass germination and establishment; 2) determine if grass seed morphology affects establishment on biocrusts; 3) determine if biocrust type and integrity affect grass establishment; and 4) quantify the effect of biocrust restoration on establishment of native grasses and reinvasion of non-native grasses (cheatgrass (Bromus tectorum), buffelgrass (Pennisetum ciliare), and red brome (Bromus rubens)).
Hypothesis 1 (Objectives 1 and 2): The influence of biocrusts on grass germination and establishment (positive, neutral or negative) varies according to seed characteristics. Specifically, (a) biocrusts will reduce recruitment of plants whose seeds have large awns or appendages compared to smoother seeds; (b) biocrusts will reduce recruitment of large-seeded species compared to small-seeded species; and (c) biocrusts will reduce recruitment of exotic grasses compared to native grasses.
Hypothesis 2 (Objective 3): Grass germination and establishment varies with biocrust type and integrity. Specifically, (a) lichen/moss biocrusts are a more effective barrier than cyanobacterial biocrusts when of comparable roughness; (b) biocrusts form a physical, not biological barrier to seeds; and (c) intact biocrusts are a more effective barrier against exotic grasses than broken biocrusts.
Hypothesis 3 (Objective 4): Restoration of biocrust communities following removal of non-native grasses (a) decreases the probability of their reinvasion and (b) increases or has a neutral effect on re-establishment of native grasses.
Cooperators
Research
We conducted experiments in field and semi-controlled settings on the Colorado Plateau and in the Sonoran Desert. Procurement of additional funds from a USDA NIFA grant (leveraged in part from this project) allowed us to expand our experiments to Sonoran Desert sites, investigate additional invasive and native grass species, and add the experiment in a semi-controlled environment. We used 7 native and 3 non-native grasses, a mix of warm- and cool-season grasses, seeds with a range of sizes, and manipulated seeds by clipping awns or bristles to address our Objectives 1 and 2 of determining the extent to which seed attributes influence germination and establishment outcomes. We investigated cyanobacteria-dominated and lichen-dominated biocrusts and simulated trampling of biocrusts to determine if the integrity of biocrusts influence grass germination and establishment outcomes (Objective 3). We mechanically controlled cheatgrass (Colorado Plateau) or buffelgrass (Sonoran Desert) and restored biocrust communities to address Objective 4.
Study Areas
In the Sonoran Desert, we conducted Experiment 1 in a hoop house at the University of Arizona Campus Agricultural Center in Tucson, Arizona, USA. We conducted Experiments 2-3 at three sites (~880m elevation) on the Santa Rita Experimental Range (SRER) south of Tucson and these sites were also the source of materials for Experiment 1. Soils on the SRER sites were gravelly sandy loam; vegetation was desert scrub dominated by Prosopis, Acacia, Opuntia, and Cylindropuntia spp. with native grasses (Aristida, Digitaria, Bouteloua spp.) present. Each site contained several patches of non-native buffelgrass.
On the Colorado Plateau, we conducted Experiment 1 in a hoop house at The Nature Conservancy Canyonlands Research Center (Figure 1) at the Dugout Ranch. We conducted Experiments 2-3 in at three sites south of Moab, Utah near Hatch Point (~1780-1900m elevation). A fourth site nearby was the source of materials for Experiment 1. Soils on three Colorado Plateau field sites were sandy loam whereas the fourth site (for Experiment 1) had sandy loam to loamy sand soils. Vegetation on the three field sites was largely native grasses (Bouteloua, Stipa, Hesperostipa, Sporobolous spp.) with patches of cheatgrass and patches of sagebrush (Artemisia spp.) and fourwing saltbush (Atriplex canescens).
Figure 1. Experiment 1 in a hoop house at The Nature Conservancy Canyonlands Research Center.
Experimental Species
We used 7 native and 3 non-native grasses that were a mix of warm- and cool-season species and represented a range of seed sizes in our experiments (Table 1). We obtained seeds from a variety of commercial seed sources, but utilized hand-collected cheatgrass and red brome seeds from sites near Moab, Utah and Tucson, Arizona respectively. We manipulated half of the seeds for a given species by clipping awns or bristles. Buffelgrass was cold scarified for 1 week at ~5°C (USDA 2011), needle-and-thread was cold scarified for 120 days at ~4°C (Ogle et al. 2006), and sand dropseed was mechanically scarified (Jackson 1928).
| Table 1. Grass species and associated characteristics used in experiments. | ||||||
|
Species |
Common name |
Nativity |
Season |
Lemma length (mm) |
Awn/bristle length (mm) |
Experiments |
|
Aristida purpurea |
purple threeawn |
Native |
Warm |
10-12 |
35-45 |
All |
|
Aristida purpurea var. longiseta |
red threeawn |
Native |
Warm |
12-15 |
40-80 |
#1 |
|
Bouteloua gracilis |
blue grama |
Native |
Warm |
4-6 |
1-3 |
#1 |
|
Bromus rubens |
red brome |
Non-native |
Cool |
10-17 |
15-25 |
All, Sonoran Desert only |
|
Bromus tectorum |
cheatgrass |
Non-native |
Cool |
9-16 |
10-25 |
All, Colorado Plateau only |
|
Hesperostipa comata |
needle-and-thread |
Native |
Cool |
8-12 |
120-150 |
All |
|
Muhlenbergia porteri |
bush muhly |
Native |
Warm |
3-4 |
5-10 |
#1 |
|
Pennisetum ciliare |
buffelgrass |
Non-native |
Warm |
5 |
<14 |
All, Sonoran Desert only |
|
Sporobolus cryptandrus |
sand dropseed |
Native |
Warm |
1-2 |
0 |
#1 |
|
Vulpia octoflora |
sixweeks fescue |
Native |
Cool |
4-5 |
1-6 |
#1 |
Experiment 1 – Emergence in a semi-controlled environment
Within each study area, we constructed a hoop house such that natural precipitation was excluded but ambient temperature and wind conditions were somewhat modified. We filled 10cm square pots with nursery foam and vermiculite, leaving room for a ~1cm veneer of soil from the respective study area (passed through 2mm sieve and homogenized). Pots were topped with a cyanobacteria or lichen biocrust (harvested from study area) or with additional bare soil. We applied 20mL of polyacrylamide (PAM; Tramfloc 1038 prepared according to manufacturer’s instructions) to three-quarters of the bare soil pots two days prior to each experimental trial. PAM crusts simulated the physical barrier of biocrusts but lacked the biological attributes of biocrusts. The biocrust or PAM crusts had one of three surface conditions: intact, cracked, or broken. We disrupted the crusts with a paint scraper such that fragments were generally ≤1cm2 in area and to a depth of ~1cm. Twenty seeds (intact or manipulated) of a given study species were sprinkled into each pot. Among pots receiving the cracked treatment, seeds were placed into the cracks. We replicated each species × seed manipulation × surface type × surface condition combination five times. To aid in species identification, we seeded a single species into a “grid” with all surface type × surface condition combinations resulting in a block design with block and species confounded. However, we treat the data as a randomized block design in analysis. We watered the pots with a graduated cylinder (year 1) or hand-held sprayer with a fine mist (year 2) to approximate a 6mm rainfall event. We utilized a variety of watering regimes to maximize germination; but generally, pots were watered for 3 (year 1) or 4 (year 2) consecutive days to approximate rainfall known to germinate buffelgrass (Ward et al. 2006) or every other day. We recorded germination/emergence (radicle or cotyledon visible) daily for two weeks and removed the seeds upon emergence. In the Sonoran Desert, we conducted the experiments in July-August 2015, December 2015-February 2016, July-August 2016, and November 2016-January 2017. On the Colorado Plateau, we conducted the experiments in August-September 2015, March-April 2016, and August-September 2016.
Experiment 2 – Establishment in a field environment
Within each of the three sites within each study area, we evaluated the potential of well-developed biocrusts to limit invasion by non-native grasses and studied the influence of well-developed biocrusts. We arranged biocrusts, PAM crusts (as described above) and bare soil is a series of contiguous grids composed of 10x10cm patches. We transplanted cyanobacteria-dominated biocrusts and mixed-lichen biocrusts obtained from a location within each study area. A paint scraper was used to simulate trampling in half of the biocrust and PAM patches to a depth of 1cm with most fragments <1cm2 area. We placed 20 seeds of one of the grass species used in the field experiments (Table 1) with seeds, either intact or manipulated (awns/bristles clipped), onto each 10x10cm patch. We tracked seedling establishment weekly for four weeks. To aid in species identification, we seeded a single species into a “grid” with all surface type × surface condition combinations resulting in a block design with block and species confounded. However, we treat the data as a randomized block design in analysis. In the Sonoran Desert, we conducted the establishment experiments in July-August 2015, December 2015-February 2016, July-August 2016, and November 2016-January 2017. On the Colorado Plateau, we conducted the establishment experiments in August-October 2015, March-May 2016, and August-September 2016.
Experiment 3
We mechanically removed buffelgrass (Sonoran Desert) or cheatgrass (Colorado Plateau) from one 100m2 subplot within each field site. Within each subplot, we applied restoration treatments to 0.16m2 areas generally arranged in a split-plot (0.8x0.8m, 0.64m2) design. Within each subplot applied restoration treatments to 0.16m2 areas: inoculum ratios of 1:1 (area collected : area applied), 1:5, 1:10, PAM, control (soil applied) or control (no treatment). Each restoration treatment was replicated three times within each study site. We collected inoculum from within each study site and sieved (2mm) and homogenized the inoculum prior to application. As the biocrusts were re-growing, we counted and carefully removed any cheatgrass or buffelgrass seedlings within each restoration area. After at least 7 months of biocrust regrowth, we placed 20 seeds, intact or manipulated (awns/bristles clipped), of one of the grass species used in the field experiments (Table 1) into a 10x10cm patch. We tracked seedling establishment weekly for four weeks. At the start of each germination test, we estimated biocrust development using an index developed for the Colorado Plateau (Belnap et al. 2008). In the Sonoran Desert, we conducted the establishment experiments in July-August 2016 and November 2016-January 2017. On the Colorado Plateau, we conducted the establishment experiments in August-September 2016. During analysis, we removed one site from each desert due to differences in the split-plot design (Colorado Plateau) and flooding (Sonoran Desert).
Data Analyses
We utilized linear mixed-models to analyze our germination (emergence) data from Experiment 1 and establishment data from Experiments 2 and 3. Analyses were conducted in RStudio version 1.0.44 running R version 3.3.2 with packages nlme and gmodels. For models for Experiment 1, species, seed manipulation (intact or awns/bristles removed), surface type (bare soil, PAM crust, cyanobacteria-dominated biocrust, and lichen-dominated biocrust), surface condition (intact, cracked, or broken), were the main effects in the analysis (without interactions) and we included pot, grid, trial run, and side of hoop house as random effects. For Experiment 2, species, seed manipulation (intact or awns/bristles removed), surface type (bare soil, PAM crust, cyanobacteria-dominated biocrust, and lichen-dominated biocrust), and surface disturbance (intact or broken) were main effects with site, block, and sub-site included as random effects. For Experiment 3, restoration treatment was the main effect with site, block, and sub-site included as random effects.
Experiment 1 – Emergence in a semi-controlled environment
Preliminary analysis of germination/emergence in the Sonoran Desert (all species, all trials from 2015-2017 combined) show that seed placement, surface condition, soil surface type, and species influenced emergence rates. The native needle-and-thread had very low seed viability and was excluded from these analyses. Mean (± SE) emergence was significantly higher (p<0.01) when seeds were placed in cracks of cyanobacteria (54% ± 4) or lichen biocrusts (55% ± 4) compared to seeds placed on the surface of intact cyanobacteria (8% ± 1), lichen biocrusts (10% ± 1) or bare soil (33% ± 3). There were no differences between seeds placed in the cracks of cyanobacteria, lichen, or synthetic polyacrylamide (PAM) crusts. Placing seeds in cracks of biocrusts facilitated seed-soil contact and resulted in high germination/emergence. This supports the notion that small seeds and seeds with small or no appendages may be more likely to enter cracks and hence be favored on biocrusts (Zhang et al. 2016). Emergence was significantly higher (p<0.05) for broken cyanobacteria biocrusts (21% ± 2) and lichen biocrusts (32% ± 4) compared to intact biocrusts, whereas emergence on broken lichen biocrusts was comparable to that occurring on bare soil and emergence on broken cyanobacteria biocrusts was significantly less than on bare soil. The increased germination on broken biocrusts compared to intact biocrusts may be due to improved seed-soil contact and/or nutrient release (Belnap et al. 2003, Zhang et al. 2016). This was somewhat in line with our prediction that intact biocrusts would be more of a barrier to exotic grasses than broken biocrusts. However, we also saw the trend with native grasses. Contrary to our predictions, native (10% ± 1) and non-native (7% ± 2) grasses emerged at similar rates on intact biocrusts in the Sonoran Desert. Among the native grasses, small seeds (lemma and awn lengths < 10mm) were more likely to emerge on intact biocrusts (14% ± 1) compared to large seeds (lemma lengths ≥ 10mm and awn lengths ≥ 30mm; 6% ± 1). This supported our prediction that small-seeded grasses would be favored on biocrusts. Removing the awn(s) decreased emergence for the native grasses on intact biocrusts (after accounting for potential effects of manipulating the seed), which was contrary to expectations. The removal of the awns may have altered water absorption (Zhang et al. 2016).
The overall germination/emergence rates in the Colorado Plateau semi-controlled environment experiments were lower than those in the Sonoran Desert, and germination on bare soil was lower. Preliminary analysis of cheatgrass germination/emergence (all trials from 2015-2016 combined, including spring 2016 [see below]) show emergence was significantly higher (p<0.01) when seeds were placed in cracks of cyanobacteria (55% ± 8) and lichen biocrusts (47% ± 7) compared to all other combinations. As predicted, cheatgrass emergence on broken lichen biocrusts (21% ± 3) was higher than on intact lichen biocrusts (11% ± 1). Cheatgrass emergence on bare soil (3% ± 0.4) was statistically comparable to that on intact cyanobacteria biocrusts (8% ± 1). Poor seed germination limited the strength of our intended comparisons to native grasses. Our single trail with native grasses in spring 2016 occurred when freezing temperatures occurred most nights. Two of the six native species had only a single seed germinate; in the other four species only 4% of the viable seed germinated. Although germination rates were low overall, there appeared to be differences among native species for preferred surfaces. For example, sixweeks fescue emerged most frequently on broken cyanobacteria biocrusts (37% ± 12), whereas blue grama emergence was highest among seeds placed in biocrust cracks.
Experiment 2 – Establishment in a field environment
In the Sonoran Desert, we studied the establishment of three native grasses and two non-native grasses (Table 1). Overall, germination and establishment rates were low, with no establishment of the native needle-and-thread grass. Among the other four grasses, establishment of each species was highest on broken/trampled biocrusts. The native purple threeawn (Aristida purpurea) and sixweeks fescue were most likely to establish (p<0.01) in broken lichen biocrusts. The non-native red brome (Bromus rubens) was most likely to establish on broken crusts (cyanobacteria, lichen or PAM) compared to intact crusts (p<0.05). Trends for non-native buffelgrass were less clear, but establishment was highest in broken lichen biocrusts. These differences confirmed our prediction that the barrier posed by biocrusts dissipates with disturbance to the biocrusts. The greater establishment on broken biocrusts may be a result of increased seed-soil contact, seed burial, nutrient release and/or increased seed retention (Belnap et al. 2003, Zhang et al. 2016).
We report the results of cheatgrass establishment only from the fall of 2015 trial because insufficient rainfall during other periods caused experiments to fail. Establishment of cheatgrass on the Colorado Plateau varied significantly with surface type (p=0.02) and surface disturbance (p<0.01) with cheatgrass being more likely to establish in broken crusts (16% ± 2) than on intact crusts (4 ± 1%). Awn presence or absence had no effect on establishment rates (5% ± 2 vs. 4% ± 1 for seed with intact awns vs. awns removed). Cheatgrass was less likely to establish on intact cyanobacteria (3% ± 1) or lichen biocrusts (6% ± 2) compared to disturbed cyanobacteria (15% ± 4), lichen biocrusts (16% ± 3), soil stabilizing polyacrylamide gel (PAM) crusts (12% ± 2 intact, 24% ± 5 broken), or bare soil (18% ± 4). However, cheatgrass establishment on bare soil and PAM crusts were statistically comparable. Greater establishment on broken vs. intact biocrusts follows our prediction and may be a result of increased seed-soil contact and/or increased seed retention. In the spring of 2016, we conducted establishment experiments on cheatgrass and native grasses on the Colorado Plateau. However, insufficient rainfall caused experiments to fail (only 86 of 2,520 cheatgrass seeds produced seedlings and only three of 7,560 native seeds produced in seedlings). Insufficient rainfall also affected our experiments in the fall of 2016.
Experiment 3 – Establishment on restored biocrusts
In the Sonoran Desert, restoration treatment did not have a significant effect on the level of biocrust development. Similarly, restoration treatment did not have a significant effect on the number of buffelgrass seedlings establishing when placed in the field 7 months after the restoration treatments were applied.
On the Colorado Plateau, restoration treatment was significant (p<0.05) with the level of development in fall 2016 higher for subplots with a 1:1 biocrust inoculation ratio compared to all other restoration treatments. This follows previous restoration research (J. Belnap, pers. comm.). All other restoration treatments were statistically similar. Restoration treatments with a 1:1 inoculation ratio also stimulated germination of cheatgrass from the seedbank (summed across all visits) compared to other restoration treatments (p<0.01). This may have been due to cheatgrass seed in the inoculum material or due to added soil cover and nutrients in the 1:1 ratio plots.
Research Outcomes
Education and Outreach
Participation Summary:
McIntyre, C.L., Archer, S., and J. Belnap. 2015. Do biocrusts differentially influence native and non-native grass establishment? 13th Biennial Conference of Science & Management on the Colorado Plateau & Southwest Region, Flagstaff, Arizona. October 5-8, 2015. (Talk)
McIntyre, C.L., and S. Archer. 2015. Do biocrusts differentially influence native and non-native grass establishment? Annual Research Insights in Semiarid Ecosystems (RISE) Symposium, Tucson, Arizona. October 17, 2015. (Poster)
McIntyre, C.L., Archer, S.R., Belnap, J. 2016. Establishment of native and non-native grasses on biocrusts in two North American deserts. Achievement Rewards for College Scientists Foundation-Phoenix 41st Annual Meeting, April 29, 2016 at the Phoenix Country Club, Phoenix, AZ. (Poster)
McIntyre, C.L., Archer, S., and J. Belnap. 2016. Influence of biocrusts on grass germination and establishment in two North American deserts. Third International Workshop on Biological Soil Crusts, Moab, Utah. September 26-30, 2016. (Talk)
McIntyre, C.L., Archer, S., and J. Belnap. 2017 Influence of biological soil crusts on grass germination and establishment of native and non-native grasses. 70th Annual Society for Range Management Meeting, Technical Training & Trade Show, St. George, Utah. January 29 – February 2, 2017. (Poster)
In addition to presenting a paper, the graduate student was an invited panelist on a discussion of biocrusts and restoration at Third International Workshop on Biological Soil Crusts in Moab, Utah. The workshop brought biocrust researchers from around the world together with land management professionals from agencies and non-governmental organizations. The graduate student also presented on the ecological role of biocrusts at a Santa Rita Experimental Range Discovery Saturday event. These weekly public Discovery Saturday events are attended by a divers citizenry in southern Arizona. The biocrust presentation was well attended (~30 individuals) and well-received by the attendees. The graduate student on the project also gave an invited presentation to the Buffelgrass Working Group in Tucson, Arizona on November 18, 2015. The Buffelgrass Working Group consists of local, state, and federal managers and scientists, university scientists, and non-profit scientists and citizen-advocates who coordinate management of buffelgrass in the Tucson Basin under the umbrella of the then Southern Arizona Buffelgrass Coordination Center (now under the Arizona-Sonora Desert Museum).
The graduate student on the project gave an informal overview of the project to the Bureau of Land Management Canyon Country staff (Moab and Monticello Field Offices) during a meeting at The Nature Conservancy’s Canyonlands Research Center near Moab and Monticello, Utah. In addition, the student has had numerous interactions with the managers/producers of the Dugout Ranch, home of the Canyonlands Research Center.
Information Products
- Poster for Annual Research Insights in Semiarid Ecosystems (RISE) Symposium, 2015
- Poster for 70th Annual Society for Range Management Meeting, Technical Training & Trade Show, 2017
- Narrated slideshow (video) on biocrust-grass germination research
- Factsheet on biocrusts in Arizona
- Website on biocrust-grass germination research