Drought eliminates transplant size benefits in the rare endemic Oregon semaphoregrass (Pleuropogon oregonus)

MATERIALS AND METHODS

This study was performed at the Oregon State University Eastern Oregon Agricultural Research Center (43.517497 N, 119.0220011 W), located at 1260 m ASL approximately 7 km SE of Burns, Oregon. In December 2021, we obtained 65 Oregon semaphoregrass tillers from Burns Paiute Tribal DNR stock and planted them into 13.2 l (3.5 gal) pots filled with the same soil mixture they originally grew in at the BPT DNR greenhouse (Metro-Mix 840 potting mix, Sun Gro Horticulture, Agawam, Massachusetts). These tillers were watered roughly twice weekly under high output red/blue LED lights (800 mmol m²/s photosynthetic photon flux density) with a 12 h daily photoperiod in a temperature-controlled (22–25 °C [71.6–77 °F]) grow room. By April 2022 we had propagated more than 700 tillers. Plants were moved on 1 May 2022 to outside the grow room to acclimate to external light and temperature conditions. They were brought back inside for the night to avoid any freezing temperatures.

On 11 May 2022, we selected tiller clumps from healthy plants, cut them to an even height (8–10 cm [3.2–4 in]), removed and cleaned away soil, and separated them into 10 large (10–12 tillers) and 10 small (5–7 tillers) clumps. We transplanted these into larger pots (28.5 cm diameter × 36 cm height [11.4 × 14.4 in; 18.9 l [5 gal]) filled with soil to 5 cm (2 in) below the upper edge with the same soil media in which the original tillers were grown. These pots were moved into a small, ventilated greenhouse structure for the duration of the experiment. To ensure plugs comprised tillers of similar condition, a subset of 5 large and 5 small clumps were scanned on a calibrated flatbed scanner (Epson 1300), with scans analyzed for average tiller height and rooting depth (WinCAM 2022a, Regent Instruments, Quebec City, Quebec, Canada) and for total leaf and root areas (WinRhizo 2022b, Regent Instruments, Quebec City, Quebec). Aboveground and belowground portions were then separated and dried at 48 °C (118.5 °F) for at least 72 h to determine aboveground and belowground biomass and root:shoot ratios.

In the greenhouse, plants were watered daily to a water level 2 cm (0.8 in) from the top of the pot edge until 24 May. On 25 May, half of the pots from each transplant size were randomly assigned into well-watered control or drought treatment groups (n = 5 for each size and treatment combination). Drought treatment consisted of withholding watering until we observed total aboveground senescence, which was attained by 24 Aug, 92 d after the last watering. At this point, all pots were re-watered to saturation.

• Download figure
• Open in new tab
• Download powerpoint

Oregon semaphoregrass plug following a 2-wk establishment period in the greenhouse pots.

Starting 18 May, weekly live tiller count and volumetric soil moisture (θ_soil; mol H₂O mol⁻¹ soil) were measured for all pots. Original transplant tillers were marked with small paper-clip tags at transplanting and counted to establish baseline tiller counts. We used a handheld time-domain-reflectometry (TDR) soil moisture probe (HydroSense HSII, Campbell Scientific, Logan, Utah) to measure θ_soil across 0–12 and 0–20 cm (0–4.8 and 0–8 in) soil depths. Saturating soil conditions at the initial sampling periods and throughout the study for well-watered controls exceeded instrument resolution and were recorded as 52.2 mol mol⁻¹, 0.1 mol mol⁻¹ above the HSII maximum resolution. We did not have micrometeorological instrumentation available to monitor environmental conditions within the greenhouse. To provide a general temperature context over the course of the study, we obtained daily maximum and minimum air temperature data from a nearby US Bureau of Reclamation AgriMet station located approximately 400 m from the greenhouse (USBR 2024).

We estimated total canopy growth (cm²/pot) from Normalized Difference Vegetation Index (NDVI) images taken with a Canon SX260 HS digital camera modified to capture NDVI wavelength bandwidths as part of the WinCAM software package (Regent Instruments). Images were taken at 1- to 2-wk intervals throughout the study. NDVI images were taken at nadir (that is, directly overhead) against a blue drop cloth background with a standard area reference square in each image to correct for variation in nadir heights. Prior to image analysis, images were processed in Photoshop to remove background soil, shadows, pot edges, and inclusion of leaves from any neighboring plants. Exclusion pixel classifications were generated for each exclusion class, then removed automatically by the software. Images were then visually inspected and any remaining excluded conditions still apparent were removed by manual delineation. The processed images were then analyzed for total canopy area (NDVI_total) and total area of maximum NDVI (NDVI > 0.7, indicating healthy green tissue) with WinCAM 2022a software (Regent Instruments). From these, we calculated the proportion of healthy canopy as %NDVI_0.7 = 100 × (1-(NDVI_total-NDVI_0.7)/ NDVI_total)).

• Download figure
• Open in new tab
• Download powerpoint

Photo highlights the uniform rooting depth of the Oregon semaphoregrass. This material was later hand-separated into small (5–7 tiller) and large (10–12 tiller) plugs for transplant.

After re-watering after the drought period, we withheld water for 4 wk for all plants to determine if previous drought exposure reduced the ability of these plants to tolerate subsequent soil water deficit. We then watered the plants roughly every week until freezing conditions (mid-November 2023) were sufficient to freeze the soil, at which point we stopped watering. Upon spring thaw during April 2023, the plants were re-watered. Then, for any surviving plants that flowered, we counted the number of seed heads produced per pot and used a one-way ANOVA (Statistix v.8.0, Analytical Software, Tallahassee, Florida) to test for differences in flowering between well-watered and previously droughted plants 1 y after the start of the experiment.

Within the drought treated pots, we used split-plot repeated measures analysis of variance (RM-ANOVA; Statistix v.8.0, Analytical Software, Tallahassee, Florida) to test for differences between large and small transplant clump soil, live tiller count, NDVI_total, and %NDVI_0.7 over the experimental drought period and for the 4-wk period following rewatering. Clump size (large vs small) was the whole plot between treatment effect, using the size-by-replicate interaction as the whole plot F-test error term. Week of sampling and the size-by-time interaction were the subplot within treatment effects, using the size-by-time-by-replicate interaction effect as the F-test error term. Post hoc means tests were made using least significant difference (LSD), with an a-adjusted P < 0.05 considered significant. As we were solely interested in the effects of declining water availability on the growth and performance of different-sized planting cohorts, we did not directly statistically compare well-watered to drought treatments. However, to establish the broader context of how much drought impacted these plants, for each watering and size combination, we summed all live tillers and %NDVI_0.7 areas and calculated the summed differences between well-watered and droughted clump cohorts for each appropriate sampling date.

• Download figure
• Open in new tab
• Download powerpoint

Close-up of an Oregon semaphoregrass seed head.

• Download figure
• Open in new tab
• Download powerpoint

Overhead view of Oregon semaphoregrass pots after the completion of the droughted and well-watered treatments.

RESULTS

Small and large clumps did not differ in tiller height, rooting depth, or dry biomass-based root:shoot ratios prior to transplanting (Table 1). Small and large clumps did not experience significant differences in θ_soil across 12 cm (4.8 in) (F_1,10 = 4.01; P > 0.05) or 20 cm (8 in) profiles (F_1,10 = 1.28; P > 0.05), with no significant size-by-time interaction as soils dried through the course of the experiment (Figure 1a). As expected, larger clumps produced significantly more live tillers (13.8 tillers pot-1 ±0.74 SE) than small clumps (9.8 tillers pot-1 ±0.58 SE; LSD < 0.05) pooled over the entire experiment, but there was a strong size-by-time interaction in live tiller counts (Table 2). Both large and small clumps consistently increased live tiller numbers over the first 9 wk following the last watering (Figure 1b), even as air temperatures increased markedly (Figure 2). When θ_soil fell below 10 mol mol^-1 by wk 10, live tiller numbers began to decline (Figure 1b), a period that coincided with the highest air temperatures (Figure 2). After this, tiller counts were indistinguishable between large and small clump treatments by week 11, after which they remained similar, even after re-watering established a prolonged period of favorable soil moisture conditions (Figure 1a, 1b) and air temperatures moderated (Figure 2).

Unlike tiller counts, total canopy development (NDVI_total) did not significantly differ between large and small clump treatments and only changed significantly over time with soil drying, with no significant size-by-time interaction (Figure 1c; Table 2). Canopy development was consistent with tiller count, increasing even as θ_soil declined, attaining peak values of approximately 225 cm²/pot at wk 10, after which it declined sharply (Figure 1a, 1c). The proportion of NDVI_total comprising the healthiest tissue (%NDVI_0.7) did show a significant size-by-time interaction (Table 2). Over the first 2 sampling periods after cessation of watering, %NDVI0.7 was similar between large and small clump pots but then increased to a greater degree in large clump pots compared to small clump pots until peak canopy development was attained by wk 10 (Figure 1d). After this, %NDVI_0.7 did not differ between large and small clumps as %NDVI and θ_soil declined to its lowest levels (Figure 1d). While NDVI_total increased in response to watering (Figure 1c), %NDVI_0.7 remained at peakdrought low levels (Figure 1d).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Average (a) θ_soil, (b) live tiller, (c) NDVI_total-derived canopy area, and (d) proportion of canopy with NDVI_0.7 (%NDVI_0.7) of different-sized transplant clumps of Oregon semaphoregrass (Pleuropogon oregonus) over a 92-d dry down and 1 mo recovery period. Each point is the mean of 5 independent measurements; error bars are 1 SE of the mean. * indicates significant differences between large and small transplant clumps at that time (LSD < 0.05).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Daily minimum and maximum air temperatures at the USBR AgriMet meteorological monitoring station located near the study location.

Both large and small clump well-watered controls produced slightly but consistently more (approximately 1 to 11) total live tillers than their droughted counterparts over the first 7 wk of the dry down (Figure 3a). Starting in wk 8, the total number of live tillers declined steeply compared to well-watered controls, reaching 643 fewer across large clumps and 621 fewer for small clump pots by the end of the drought (Figure 3a). Droughted large clumps had lower total NDVI₀ ₇ areas compared to well-watered controls over the first 6 wk of soil drying (11.5 to 0.3 cm² less) (Figure 3b), while droughted small clumps had about 1.1 to 14.1 cm² greater %NDVI_0.7 areas compared to well-watered controls over this period (Figure 3b). Ten wk after withholding water, total NDVI_0.7 of large clumps was markedly higher than that produced by well-watered counterparts (approximately +95 cm²) (Figure 3b), whereas small clumps had much lower NDVI_0.7 areas as compared to large clumps (Figure 3b). After this, NDVI_0.7 in both types was well below levels attained by well-watered pots and showed limited but marked recovery upon re-watering (Figure 3b).

Post-winter and 1 y after the start of the experiment, of the pots with surviving live plants, 3 of the 4 drought-treated pots produced flowering culms, compared to all 11 well-watered pots having plants with flowering culms. The drought-treated plants that did flower produced significantly more seed heads (approximately 2.6 times more) compared to plants well-watered the previous year (F_1,13 = 10.3; P < 0.05) (Figure 4).

DISCUSSION

Even as soil moisture levels declined, larger plant clumps consistently had more live tillers than smaller plant clumps, partially supporting our overall hypothesis. Live tiller count is a net effect of new tiller production and tiller mortality, and our data do not account for these 2 processes. As a result, we do not know to what extent live tiller counts early on reflect greater tiller production, reduced tiller mortality, or some combination thereof. The marked declines in live tiller count with drought more likely reflects expression of tiller mortality, since it is likely that very low soil moisture at these times limited tiller initiation and growth (Cone and others 1995). A more detailed, demographic approach would be needed to discriminate between the effect of these 2 processes, but they do show that planting larger clumps maintained a larger net positive effect in tiller number compared to smaller plugs, even as soil moisture levels declined. Larger plugs did not attain larger canopies, as indicated by similar NDVI_total, but, while total canopy development was similar, larger clumps developed canopies with a larger proportion of very healthy tissue (%NDVI_0.7) compared to smaller counterparts (Figure 1c, d). This finding suggests the greater number of tillers in larger plugs facilitated growth and were better able to more extensively acquire soil resources as water became more limiting (Frescher and others 2018).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Differences of drought-treated Oregon semaphoregrass from well-watered controls in total summed (a) live tiller count and (b) NDVI_0.7 canopy area pooled across all replicate pots.

Any advantage gained by planting larger clumps quickly degraded and was lost after an apparent θ_soil threshold near 8 mol mol^-1 in our soils was met, even as temperatures consistently exceeded 35 °C (95 °F) (Figure 2). Up to this point, large clumps produced more tillers and had canopies with greater %NDVI_0.7 than did smaller plantings (Figure 1) and produced larger total areas of NDVI_0.7 than did their well-watered counterparts, while smaller clumps had total NDVI₀ ₇ areas decline well below well-watered counterparts (Figure 3). We should point out that θ_soil varies with soil texture and organic content, and that these strongly affect the relationship between θ_soil and water potential (y), the feature that drives plant–water relations (Biliske 2001). In addition, our study cannot discriminate between the effects of low soil moisture and high temperature extremes (Figure 2). Thus, this threshold likely depends on the soil characteristics media and air temperature conditions. We should note the commercial soil mixture is similar to the native wet meadow soils in Logan Valley where BPT DNR has successfully established semaphoregrass populations (Copeland and others 2023), but it may be dissimilar to natural population locations and may or may not reflect the soil conditions of the many introduction sites where outplantings of the species have been unsuccessful (Gisler and Meinke 2003; Groberg 2013). Additionally, our greenhouse did not have any active climate control beyond windows and vents to facilitate air movement. Prevailing external summer climate conditions likely determined the rate of θ_soil decline, and hence how quickly critical soil y was met. Regardless, there was a distinct θ_soil threshold and reached very low levels over an extended period. These conditions were severe enough to eliminate any advantages the large clumps had attained over smaller clumps and imposed a severe restriction on the ability of plants from either sized planting to recover after drought was ameliorated.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Seed head production of droughted and well-watered Oregon semaphoregrass after the winter and thaw, 1 y after start of the experiment. Each bar is the mean of 4 (droughted) and 11 (well-watered) independent measurements; error bars are 1 SE of the mean; letters indicate significant means differences (LSD < 0.05).

Our results suggest area estimations based on total NDVI may not be as useful for monitoring success of a rare plant like semaphoregrass, as different sized transplants did not produce distinct differences between NDVI_total based canopy area estimations (Figure 1). Narrowing to a specific NDVI strength, such as NDVI_0.7, may be more useful, as tracing this proportion showed larger transplant clumps produced canopies with a larger vigorous proportion of tissue (Figure 1), while total NDVI_0.7 area (Figure 3) could be used to identify a point when taking active steps to intervene against developing water deficit would be effective after transplanting.

In conclusion, this study demonstrated that Oregon semaphoregrass is resilient to severe water deficit. Even after these severe soil moisture conditions, it still managed to produce new tillers and healthy tissue. In addition, exposure to drought triggered an increase in sexual reproduction the following year in drought-exposed plants (Figure 4). Given that the grass spreads rhizomatously, and often vigorously, this increased sexual reproductive potential might eventually constitute a legacy effect of severe drought by increasing seedbank numbers rather than relying on vegetative growth during the growing season following drought. Moreover, our findings suggest the BPT DNR practice of extensively watering transplanted semaphoregrass immediately upon outplanting is an important aspect of their success in establishing Oregon semaphoregrass stands that spread beyond their original planting locations even when under prevailing dry conditions (Copeland and others 2023). However, droughted plants were severely limited in their ability to recover. Continued observations after watering showed that live tiller counts and %NDVI_0.7 never recovered to pre-drought levels, even in a prolonged period under more favorable soil water conditions (Figure 1). Our results also show that NDVI as a “greenness” metric could be useful in identifying plantings that would most benefit from additional watering in the immediate establishment phase to sustain the positive consequences of larger plantings and to impart and facilitate persistence through seasonal soil drying. This method supports general observations that larger plant material deployment enhances long-term re-introduction success (Bellis and others 2024). The effectiveness of this approach could be paired with estimates of population growth rate, such as a change in tillers from planting through the beginning of the second growing season. This would establish management practices that determine differences that lead to growth, not simply persistence, of introduced populations that are essential to long-term success (Godefroid and others 2011). In the case of Oregon semaphoregrass, this study offers a helpful quantitative indicator of how differences in size translate functionally in terms of thriving and increasing under at least one important environmental stress: drought.