15-Yr Biomass Production in Semiarid Nebraska Sandhills Grasslands: Part 1—Plant Functional Group Analysis

ABSTRACT Semiarid grasslands of the Nebraska Sandhills provide critical ecosystem services and are an important forage resource for the local cattle industry. Over the past decades, warming and climate-related extremes have affected grassland production worldwide, which promotes the initiation of numerous grassland monitoring projects. Despite this, production trends for plant functional groups in the Sandhills regions in recent years have remained unknown. In this study, we analyzed plant biomass production of the Sandhills grasslands with a dataset collected over 15 yr from 2007 to 2021. Ungrazed total biomass and biomass of individual plant functional groups were assessed in grazing exclosures twice a year, in mid-June (for early season) and mid-August (for late season). This first paper reports our findings on total biomass and compositional changes of the three major plant functional groups, as well as trends in precipitation and temperature during the study period. A significant increasing trend (P < 0.05) was observed in temperature over time during the early season (April to mid-June), with a weak monotonic increasing trend (P= 0.07) during the full season (April to mid-August), whereas no significant pattern was reported for precipitation during the study, although it displayed complex within- and across-season patterns. The proportion of C3-grass biomass in total biomass increased (P < 0.05), while the proportion of C4-grass biomass decreased (P < 0.01). We did not observe any significant trends for forbs; however, the drought of 2012 resulted in up to a fivefold increase in the proportion of forb biomass the following year. These findings enhance our understanding of current patterns in grassland production and contribute to regional evidence on the response of plant functional groups to variability and extremes in intra-annual weather variables, which can improve our capability to perform adaptive grazing management in a similar semiarid grassland ecosystem.


Introduction
Grasslands and savannas cover 25% of the global terrestrial ecosystem and contribute to around 35% of the total terrestrial ✩ This research was funded by McIntire Stennis Fund (project 1017851) from the US Dept of Agriculture National Institute of Food and Agriculture and Nebraska state-aided funds to University of Nebraska−Lincoln Research, Extension, and Education centers located in North Platte and Scottsbluff, Nebraska.
E-mail address: yshi18@unl.edu(Y.Shi).# Current working address Amazon.com,Inc. 10550 NE 10th St, Bellevue, WA 98004, USA.Work was done prior to joining Amazon.net primary production ( Chapin et al. 2011 ;Scholtz and Twidwell 2022 ).Grasslands provide critical ecosystem services including provisioning (e.g., livestock and forage production), supporting (e.g., biodiversity, carbon sequestration, and water and nutrient cycling), regulating (e.g., climate), and culturing (e.g., spiritual, recreational, aesthetic) ( Zhao et al. 2020 ).At 50 0 0 0 km 2 , semiarid grasslands of Nebraska Sandhills in the Central Great Plains of the United States comprise the largest stabilized sand dune formation in the Western Hemisphere and one of the most intact temperate grasslands in the world ( Eggemeyer et al. 2008 ;Scholtz and Twidwell 2022 ).They are dominated by C 4 or warm-season grasses, with a strong presence of C 3 or cool-season grasses, forbs, and shrubs, being a habitat of around 720 plant species ( Kaul 1998 ).In addition to the ecological significance, livestock grazing occurs on 95% of the Nebraska Sandhills.This is an important forage resource for the local beef cattle industry that contributes approximately 12 billion dollars to Nebraska's economy each year, or nearly half of the agriculture receipts of the state ( Aiken et al. 2014 ).
Changes in climate and weather patterns can threaten beef production because of the high uncertainties in grassland forage production under variable precipitation and temperature scenarios ( Mangan et al. 2004 ;Reeves et al. 2014 ;Klemm et al. 2020a ).Driven by land use change and projected rise in temperature and summer droughts anticipated during the 21st century over the Great Plains region, several long-term grassland monitoring programs and grassland studies have been initiated in this region ( Bradford et al. 2006 ;Twine and Kucharik 2009 ;Al-Yaari et al. 2020 ;Briske et al. 2021 ).In the mixed-grass prairies of Nebraska Sandhills that occupy a transitional region of the Great Plains moving from semiarid in the west to subhumid in the east, several years of consecutive drought and temperature extremes are not uncommon.Extremes in this region are expected to be exacerbated with climate change during the 21st century ( Mangan et al. 2004 ;Nicholson and Swinehart 2005 ;Awada et al. 2012 ).Data used in this study are part of an ongoing long-term monitoring of plant production in the Nebraska Sandhills grasslands with a purpose of better understanding forage biomass production and its response to intra-annual and interannual weather variability ( Poděbradská et al. 2019 ;Stephenson et al. 2013Stephenson et al. , 2019 ) ).
As is widely acknowledged, producers and researchers have long been aware of variability in grassland production ( Lieth 1973 ;Briggs and Knapp 1995 ;Knapp and Smith 2001 ).Since the 21st century, because of applications of long-term and large-extend monitoring enabled by advanced tools and technologies such as remote sensing, there has been mounting evidence indicating variations in grassland production across various regions and over years ( Humphreys et al. 2006 ;Höglind et al. 2013 ;Zhu et al. 2016 ;Wang et al. 2020 ).In the Great Plains regions, for example, the aboveground net primary productivity often varies 40% between years due to climatic variability in mean annual temperature and mean annual precipitation ( Bradford et al. 2006 ;Reeves et al. 2021 ).In addition, spatial variations in grassland forage production are projected for the Great Plains region, with frequent forage surpluses in the Northern Great Plains, while in the Central and Southern Great Plains, forage deficits will occur more often ( Briske et al. 2021 ;Klemm et al. 2020b ).Recent evidence in the Nebraska Sandhills of the central Great Plains showed the Sandhills grassland production can vary significantly, up to threefold, between wet and dry years ( Poděbradská et al. 2019 ).
Variations in grassland production are collective responses of plant communities to climate variability ( Belesky and Malinowski 2016 ;Liu et al. 2018 ;Wilson et al. 2018 ).Assessing production trends for individual plant functional groups is crucial for better understanding and predicting the impacts of weather variables on the plant communities.Derner et al. (2008) reported that increases in forage production of three semiarid rangelands of the Great Plains were largely attributed to the greater C 3 production in each of rangelands, which were sensitive to increasing spring precipitation.Similarly, shifts in the abundance of C 4 over C 3 plant species are linked to summer rainfall patterns (i.e., above-average summer rainfall favored C 4 -grass growth and benefited competitive advantage of C 4 over C 3 ), resulting in total grass production increases because of the increase of C 4 production during summer ( Von Fischer et al. 2008 ;Ansley et al. 2019 ).Other weather conditions can lead to different responses of plant functional groups (e.g., increased CO 2 levels and warming can stimulate dry matter production in C 3 plants more than C 4 plants) ( Augustine et al. 2018 ;Havrilla et al. 2023 ), which can result in fluctuations and variations in total grassland forage production between years and across regions.To enhance our understanding, ongoing monitoring of grassland biomass production should not only focus on total production but also consider different plant functional groups.
Nevertheless, little is known regarding the production dynamics of plant functional groups in the Sandhills grasslands over the past decades.In this paper, we thus aim to assess the long-term semiarid grasslands biomass production in the Sandhills, including changes and trends in biomass production of combined and individual plant functional groups and their contributions to the total biomass.We analyzed plant biomass production with a 15yr monitoring dataset collected at the UNL Gudmundsen Sandhills Laboratory (GSL) from 2007 to 2021.Total biomass and biomass of six plant functional groups (i.e., C 3 grasses, C 4 grasses, forbs, shrubs, sedges, and native annual grass species) were assessed in grazing exclosures twice a year in mid-June (for the early season) and mid-August (for the late season).Specifically, this paper investigated the 15-yr 1) patterns of growing-season precipitation and temperature; 2) trends in biomass production of combined and individual plant functional groups; and 3) trends in biomass proportion of plant functional groups in the total grassland biomass.We expected a temperature warming trend during the study period, and we anticipated that the responses of biomass and biomass proportion among various plant functional groups would vary over a 15-yr period.This paper contributes evidence on the responses of plant functional groups to variability and extremes in intraannual weather variables, helps advance our understanding of climate effects on different plant functional groups in semiarid grasslands, offers ecological insights into ecosystem dynamics and longterm biomass trends, and has potential applications in policy and land management, enhancing our ability to address climate-related changes in this critical ecosystem.

Gudmundsen Sandhills Laboratory
Gudmundsen Sandhills Laboratory (GSL) at the University of Nebraska-Lincoln covers an area of approximately 5 145 ha centrally located within the Sandhills ecoregion ( Figure 1 ).The GSL is characterized as an upland Sandhills grassland with a mean elevation of 1 082 m.The climate is semiarid continental, with 75−80% of annual total precipitation falls during the growing season between April and September ( Hendrickson et al. 20 0 0 ).The mean annual temperature is 8.4 °C with a mean minimum temperature of -14 °C in January and a mean maximum temperature of 31 °C in July.Soils are rolling Valentine fine sand (54%) mixed with other Valentine fine sand (mesic Typic Ustipsamments) and deep, loose, sandy, well-drained soils.The vegetation at the GSL consists of a mixture of predominately native C 4 and C 3 grasses, forbs, and prairie shrubs ( Kaul 1998 ).Important species include prairie sandreed (Calamovilfa longifolia), little bluestem (Schizachyrium scoparium), switchgrass (Panicum virgatum), needle-and-thread grass (Hesperostipa comata), stiff sunflower (Helianthus pauciflorus), and leadplant (Amorpha canescens).
The study area comprises an approximately 5-ha study site of upland sandhills range within a larger 77-ha pasture (i.e., Branding Pen pasture) that was grazed by cow-calf pairs at moderate stocking rates (a mean of 1.51 animal unit months • ha −1 ) (see Figure 1 b).This 5-ha area is a representative sand site that shares common characteristics with the GSL and the central Sandhills in terms of soil properties (rolling Valentine fine sand), slopes (9−24%), and species compositions ( SSURGO and Soil Survey Staff 2023 ).

Weather data
Monthly precipitation and temperature at GSL were collected from an on-site weather station (USW0 0 094079) (see Figure 1 b).This weather station is part of the US Climate Reference Network (USCRN) that measures high-quality micrometeorological data ( https://www.ncei.noaa.gov/access/crn/ ) ( Diamond et al. 2013 ).The raw weather data of this study were accessed at the National Oceanic and Atmospheric Administration (NOAA) website ( https://www.ncdc.noaa.gov/cdo-web/).The highest monthly precipitation was observed between May and August, and the highest monthly mean air temperature occurred in July and August ( Figure 2 ).These weather patterns are consistent with long-term trends of the larger Sandhills region ( Awada et al. 2012 ).
We calculated precipitation accumulation (PreAcc) and growing degree days accumulation (GDDAcc) by using the daily precipitation and temperature data achieved from the on-site weather station.According to the timing of plant biomass data collection, three time periods for the accumulation were used in this study (Fig. S1, available online at …) (i.e., the early-season growing period, April to mid-June) from the 90th to 165th day of the year, late-season growing period (i.e., between mid-June to mid-August) from the 166th to 225th day of the year, and the full growing period (i.e., April to mid-August) from the 90th to 225th day of the year.The calculation of daily GDD was shown as follows: where T max and T min represent the maximum and minimum daily temperature, respectively.T base represents the base temperature for the growth of plants and crops to occur.In this study, T base was 10 °C due to the mixed grasses on the grasslands of Sandhills ( Zhou et al. 2003 ).

Plant biomass measurements
Thirty fenced exclosures (1.2 × 1.2 m 2 area and 1.2 m height) were randomly distributed across the 5-ha study area before cattle entry each year to prevent cattle grazing (Fig. S2, available online at …).These fenced exclosures were moved approximately 5 m every year to account for the grazed working landscape within the Sandhills.Biomass sampling from these exclosures occurred in mid-June for early season, or C 3 peak standing biomass (namely June-harvested biomass in the rest of the paper), and in mid-August for late-season, or C 4 peak standing biomass (namely August-harvested biomass in the rest of the paper), from 2007 to 2021.All plant materials were manually clipped to ground level within a 0.25 m 2 (100 × 25 cm) quadrat placed on opposite sides of each exclosure for the mid-June and mid-August harvests, respectively.All clipped plant materials were separated into six functional groups: C 3 grasses, C 4 grasses, forbs, shrubs, sedges, and native annual grass species.The forb functional group contains perennial forbs and annual forbs together.According to a recent survey, there is only one native annual grass present on the site, which is 6-wk fescue (Vulpia octoflora).Separated plant materials were placed in paper bags and oven-dried at 60 °C until a con-stant weight was reached ( ∼48 hr).Intra-annual biomass changes ( Biomass ) from mid-June to mid-August were calculated as: where Biomass August−harv ested and Biomass June −harv ested stand for biomass harvested in mid-August and in mid-June, respectively.
In addition to ungrazed biomass measured from exclosures, we simply plotted vegetation conditions (i.e., greenness) for the 5-ha study area, entire GSL, and the larger Sandhills ecoregion from 2007 to 2021 by using the Normalized Difference Vegetation Index (NDVI) ( Rouse et al. 1974 ).The NDVI used in this study is a standard vegetation index product from satellite-based Moderate Resolution Imaging Spectroradiometer (MODIS) imagery at a spatial resolution of 250 m, produced on 16-d intervals.This study accessed the MODIS Vegetation Index Products (ImageCollection: MODIS/061/MOD13Q1) by using Google Earth Engine and plotted the time-series NDVI for the three areas by averaging pixels of each of these areas.The calculation of NDVI was shown as follows: where, NIR and RED are the surface reflectance of near-infrared (NIR) and red (RED) bands of MODIS.The standard MODIS NDVI product (ImageCollection: MODIS/061/MOD13Q1) used in this study was accessed by using Google Earth Engine.
The NDVI calculated from satellite remote sensing can be considered a response metric of vegetation condition at a canopy level, which is well correlated with vegetation cover, leaf area index, biomass, and plant phenology ( Rigge et al. 2013 ;Gong et al. 2015 ;Kern et al. 2020 ).Change in NDVI can be caused by environment (e.g., precipitation and temperature) and land uses (e.g., grazing) ( Tieszen et al. 1997 ;Yang et al. 1997 ;Hartman et al. 2020 ).On the basis of the grazing schedules of the 15 yr (Table S1, available online at …), a mean NDVI value of the 5-ha study area on the 129th day of the 15 yr was calculated as the pregrazing NDVI peak of each year.This primarily corresponds to the C 3 -grass production and growth in the cool season.NDVI values after this date (129th day of the year) were considered as a combined results of biomass production and grazing.

Data analysis
Coefficient of variation (CV) across seasons was calculated at a 15-yr scale (i.e., using the 15 datapoints from 2007 to 2021) to quantify the interannual variations in precipitation and temperature during the study period.Three seasonal CVs were calculated for early-, late-, and full-season growing periods, respectively, and both for PreAcc and GDDAcc.In addition, correlation between PreAcc and GDDAcc of each growing period was investigated by using a simple linear correlation analysis.The calculation of CV at a 15-yr scale was by: where CV 15 means the coefficient of variation at a 15-yr scale throughout the study period.σ and μ represent the standard deviation and the mean of using the 15 datapoints from 2007 to 2021, respectively.The Mann-Kendall Test (MK test) was used to analyze trends of weather and grassland biomass production throughout the 15yr period.The MK test is a nonparametric test, which assesses whether a variable presents a monotonic upward (increasing) or downward (decreasing) trend over time ( Mann 1945 ).Compared with parametric methods (e.g., linear regression), the MK test does not require data being tested to meet assumption of normality ( Opiyo et al. 2014 ).The MK test is preferred for analyzing time-series data in environmental and nature resource studies such as rangeland monitoring because it can avoid statistical issues caused by data skewness ( Guido et al. 2014 ;Zimmer et al. 2021 ;Kleinhesselink et al. 2023 ).The null hypothesis (H 0 ) in this study was that no trends were presented according to the data (H 0 : P ≥ 0.05).H 0 was rejected when the significance level was < 0.05, which meant that the data exhibited a significant trend ( P < 0.05).

Overview from weather variability and NDVI plot
Over the 15-yr study period, no significant trends were found in PreAcc in the early-, late-, and full-season growing periods ( P > 0.05), but precipitation showed complex within-and acrossseason patterns ( Table 1 , Figure 2 a, and Figure 3 a ).On the other hand, a significant increasing trend was found in GDDAcc for the early-season growing period ( P < 0.05) (see Table 1 ).In addition, GDDAcc in the full-season growing period presented a weak monotonic trend (increasing trend, P = 0.07), which needs more data from continuous monitoring to validate.Negative correlations ( n = 15) were found between PreAcc and GDDAcc for each growing period: r = −0.43 ( P > 0.05) in the early-season growing period, r = −0.64 ( P < 0.05) in the late-season growing period, and r = −0.77( P < 0.001) in the full-season growing period (see Figure 3 c).
Throughout the 15 yr, larger intra-annual (within-season) variability in PreAcc occurred during the late-season growing period from mid-June to mid-August (CV 15 = 40.9%)compared with the early-season (CV 15 = 27.8%), or the full-season (CV 15 = 28.0%)growing periods (see Table 1 ).The larger CV 15 in PreAcc during the late-season growing period was attributed to the extreme deficit of precipitation in 2012 (the least dark green area in Figure 3 a).In 2012, almost 72% of the total precipitation of the growing season from April to September (85 mm) was received in the early-season growing period from April to mid-June (61 mm) (Fig. S3, available online at …).Compared with the extreme low precipitation during the late-season growing period in 2012, the wettest late-season growing period was observed in 2011 (280.5 mm), followed by 2018 (189.6 mm) and 2009 (181.9 mm).The wettest early-season growing period was observed in 2010 (248.3 mm), followed by 2015 (241.8 mm), 2014 (239.8 mm), 2019 (220.1 mm), 2018 (219.6 mm), and 2007 (206.1 mm) (see Figure 3 a), with PreAcc of > 200 mm.The wettest full-season growing period was observed in 2011 with the highest PreAcc of 471.2 mm, which was mainly attributed to the greatest PreAcc (280.5 mm) received during the late-season growing period compared with other years (see Figure 3 a).
In contrast to PreAcc, GDDAcc exhibited relatively small intra-annual variations among the three growing periods (see Table 1 and Fig. S4, available online at …); CV 15 was 19.5%, 9.0%, and 10.5% for early-, late-, and full-season growing periods, respectively.Over the 15 yr, substantial GDDAcc peaks were observed in 2018 (GDDAcc = 377.8)and 2012 (GDDAcc = 356.6)for the early-season growing period, followed by 2021 (GDDAcc = 329.5)and 2007 (GDDAcc = 299.5).For the full-season growing period, the greatest GDDAcc peak was found in 2012 (GDDAcc = 1282.4),which was mainly attributed to the greatest GDDAcc of 913.8 during the late-season period.The minimum was recorded in 2009 (GDDAcc = 832.4)for the late-and full-season growing periods (see Figure 3 b).
Throughout the 15 yr, a high degree of consistency of NDVI among the three spatial scales of 5-ha study area, the entire 5145ha GSL, and the Sandhills ecoregion were found ( Figure 4 a ).This could be supportive evidence to resemble our monitoring results and findings from the 5-ha study site to the entire GSL and broader Sandhills ecoregions.A mean NDVI value of 0.35 was calculated on the 129th day of the year for the 15-yr period, as a pregrazing NDVI peak before grazing of each year for this study.The exact date that the NDVI reached 0.35 differed from year to year, though.We observed in some years it occurred earlier than the 129th day (e.g., in 2007, 2012, 2015, 2016, 2017, and 2021), and in some other years it occurred later, such as in 2011 when the NDVI reached 0.35 on the 145th day of the year.

Plant biomass production
The total grassland biomass measured in exclosures for the 15yr period averaged 1 209.3 ± 283.6 kg •ha −1 for the early season (June-harvested) and 2 097.8 ± 359.7 kg •ha −1 for the late season (August-harvested), with an average annual increase of 888.6 ± 419.6 kg •ha −1 from mid-June to mid-August ( Figure 5 ).Overall, the total grassland biomass did not exhibit any trend over the 15 yr irrespective of seasonality (i.e., mid-June or mid-August) ( Table 2 ).However, for plant functional groups, C 4 grasses exhibited a decreasing trend in both June-and August-harvested biomass ( P < 0.05), while shrubs showed significant increasing trend in Juneharvested biomass ( P < 0.05).Except for C 4 grasses and shrubs, no significant trends were observed in the biomass production of the other four plant functional groups (i.e., C 3 grasses, forbs, sedges, and native annual grass species) ( P > 0.05).Throughout the 15-yr study, a higher CV 15 for total biomass or biomass of the six plant functional groups was found in the late-season growing period, compared with the other two growing periods (see Table 2 ).
There were 10 yr where the June-harvested total biomass was > 15-yr mean from 2007 to 2021 (see Figure 5 , dash line in red ).The lowest total biomass harvested in mid-June was recorded in 2013 (470.7 kg •ha −1 ), the year after an extreme drought in 2012 (see Figure 5 ), while the highest was recorded in 2007 (1 558.1 kg •ha −1 ), followed by 2018 (1 549.2 kg •ha −1 ) and 2012 (1 477.3 kg •ha −1 ).For the August-harvested total biomass, the highest amount was observed in 2011 with 2 728.1 kg •ha −1 , followed by 2 671.1 kg •ha −1 in 2010 and 2 369.9 kg •ha −1 in 2018, while the lowest amount was 1 526.9 kg •ha −1 in 2012.In addition to 2010, 2011,  and 2018, the August-harvested total biomass in 2009, 2013, 2015, and 2016 was greater than its 15-yr mean (see Figure 5 ).It is notable that although 2013 had the lowest total biomass in mid-June (June-harvested) (see Figure 5 ), it had the largest biomass increase from mid-June to mid-August (1 748.3 kg •ha −1 ) over the study period (see Figure 5 and 6 ).In contrast, the smallest increment observed from June to August amounted to 49.6 kg •ha −1 in 2012 (see Figure 5 and Figure 6 ), a year of extreme drought.
As for the biomass change from mid-June to mid-August ( Biomass ), no significant trends were found in total biomass and biomass of individual plant functional group, except for a potential weak monotonic decreasing trend in C 4 -grass biomass ( P = 0.06) (see Table 2 ).The Biomass of total biomass was significantly correlated with Biomass of C 4 grasses ( r = 0.71, P < 0.01, see Figure 6 a), while Biomass of C 3 grasses showed moderate correlation with Biomass of total biomass but was not significant ( r = 0.42, P > 0.05, see Figure 6 b).Correlation between Biomass of forb and Biomass of total biomass was 0.61 ( P < 0.05, see Figure 6 c) over the 15 yr.Nevertheless, it declined to 0.26 ( P > 0.05, see Figure 6 c) after removing the 2013 datapoint, which indicates the significant effects of 2013 on forb biomass.In addition, we found Biomass in the other three plant functional groups (i.e., shrubs, sedges, and native annual grass species) varied from year to year but showed minor, nonsignificant impacts on the Biomass in total biomass from mid-June to mid-August (see Fig. 6 d−f).

Relative contributions of plant functional groups
Overall, C 4 , C 3 grasses, and forbs constituted approximately 90% of total biomass irrespective of seasonality (i.e., either in mid-June or mid-August) ( Figure 7 a ).In mid-June, C 3 grasses accounted the most for an average of 41.0% of the June-harvested total biomass, followed by an average of 37.3% in C 4 grasses, 10.4% in forbs, 7.5% in sedges, and 2.2% and 1.6% in shrubs and annual grass species, respectively (see Figure 7 a).In mid-August, C 4 grasses exceeded the contribution of C 3 grasses, contributing to 46.5% of the August-harvested total biomass, while the percentage of C 3 grasses declined to 31.9% of the total biomass (see Figure 7 a).Forb biomass contribution to total biomass increased slightly from 10.4% of June-harvested to 14.2% of August-harvested on average, mainly because of the considerable increasing percentage in 2013 (see Figure 7 a).In 2013, the percent of total biomass of forbs considerably exceeded (i.e., ∼4 × in mid-June and ∼5 × in mid-August) the recorded average for the study (see Figure 7 a).As a result, the relative biomass contributions of forbs in 2013 (40% in mid-June and 58.3% in mid-August) was greater than the relative contributions of C 4 (32.8% in mid-June and 27.1% in mid-August) and C 3 grasses (23.3% in mid-June and 11.6% in mid-August) (see Figure 7 b).These results indicate that forb biomass was highly variable and responded to the extreme weather event during the drought of 2012 and the following year.
During the study duration, we noted a consistent decline in the biomass percentage of C 4 grasses in both mid-June and mid-August across successive years ( P < 0.01).Conversely, there was a concur-  rent rise in the biomass percentage of C 3 grasses ( P < 0.05) for the same growing periods ( Table 3 and see Figure 7 b).However, there were no significant trends detected for forbs ( P > 0.05).Additionally, no other significant trends were identified for the changes in biomass contribution from mid-June to mid-August among the three functional groups ( P > 0.05, see Table 3 ).

Discussion
The findings of this study have enhanced our comprehension of both the seasonal fluctuations and the extended patterns in total biomass and the biomass of specific plant functional groups, both within and across growing seasons, in the semiarid Nebraska Sand-hills.Over the 15-yr study period, the temperature of the earlyseason growing period from April to mid-June exhibited an increasing trend ( P < 0.05), but no significant patterns were found in precipitation (see Table 1 ).Our detailed plant functional group analysis demonstrated that the plant biomass production in the semiarid Sandhills grasslands was mainly modulated by the productivity of three plant functional groups (i.e., C 4 , C 3 grasses, and forbs, see Figures 5 −7 ).This aligns well with established knowledge that the Sandhills grasslands are primarily dominated by C 4 grasses, with a notable presence of C 3 grasses, followed by forbs ( Kaul 1998 ;Mazis et al. 2021 ).Another significant contribution concerning functional groups is the increasing relative contribution of C 3 -grass biomass to the annual total biomass during the study period ( P < 0.05), while conversely, the relative contribution of C 4grass biomass has been decreasing ( P < 0.01).
The increasing relative contribution of C 3 -grass biomass in our study (see Figure 7 b) may be additional regional evidence of an earlier start of growing season and extended growing period in spring ( Linderholm 2006 ;Piao et al. 2019 ).Over the 15 yr, temperature (i.e., GDDAcc) in the Sandhills increased, particularly in the early-season growing period (see Table 1 ), which likely promotes earlier growth of C 3 grasses and leads to the increase in its relative biomass contribution.C 3 grasses exhibit earlier growth compared with C 4 grasses, typically in March and April, in response to rise in spring temperatures ( Sage and Kubien 2007 ).In addition, more precipitation in winter-spring period can lead to an earlier start of C 3 plants' growth and development ( Tieszen et al. 1997 ;Knapp et al. 2020 ).The earlier attainment of the pregrazing NDVI peak in some years (see Figure 4 b), specifically a mean NDVI of 0.35 calculated using 15 yr of NDVI data on the 129th day of the year in our study, may suggest the early-season greening of C 3 plants in response to an early-season warming or wetter early-season scenario (see Table 1 ).Other factors that may contribute to vegetation cover change include rising CO 2 concertation and nitrogen deposition from year to year associated with early warming, which also favor the growth of C 3 over C 4 grasses ( Ehleringer et al. 1997 ;Morgan et al. 2011 ;Msanne et al. 2017 ), resulting in the increase in relative contribution of C 3 -grass biomass and decline in relative contribution of C 4 -grass biomass.
The declines observed in both C 4 -grass biomass and its relative contribution in our study (as indicated in Tables 2 and 3 and Figure 7 b) showed agreement with prior research findings, suggesting that production of C 4 species tend to decrease under conditions of elevated temperatures and reduced precipitation as the later part of the growing season progresses ( Sage and Kubien 2007 ;Ansley et al. 2019 ;Maschler et al. 2022 ).For example, during the extreme droughts in the 1930s, the losses of plant cover of C 4 species (e.g., little bluestem) were reported at over 75% in Nebraska ( Weaver and Albertson 1936 ), which led to severe decline in plant production ( Peters et al. 2020 ;Burruss et al. 2023 ).On the other hand, warm-season (i.e., July) precipitation is a determinant for the productivity of C 4 grasses ( Epstein et al. 1997 ;Fay et al. 2003 ;Knapp et al. 2020 ).Precipitation pattern and timing in the summer are greatly correlated with grassland productivity in a C 4 -dominant tallgrass prairie in the Great Plains ( Yang et al. 1998 ;Chen et al. 2019 ).
The trend of increasing relative contribution of C 3 -grass biomass in C 4 -dominant grasslands may accompany shifts in species composition ( Guretzky et al. 2016 ).Similar results of increased ratios of C 3 to C 4 grass biomass have been reported in experiments simulating the 1930s' drought effects ( Knapp et al. 2020 ;Peters et al. 2020 ;Havrilla et al. 2023 ).In addition, a 7-yr experiment (2007−2013) in the western Great Plains reported significant species-compositional shifts of increasing C 3 graminoids, accompanied with an average increase of 38% in forage production under simulated warming and elevated carbon dioxide (CO 2 ) con-certation ( Augustine et al. 2018 ).Nevertheless, this overcompetition of C 3 on C 4 grasses may raise the risk of reducing total grassland production, considering the high correlation between seasonal biomass change in C 4 -grass and total biomass in the Sandhills grasslands (see Figure 6 a).To prevent any undesirable plant community change or grassland biomass losses in the native C 4 grasslands of Sandhills, conservation of the native C 4 grasslands may be a priority for maintaining ecological functions under projected warming climate in the future.
The Sandhills region experienced an extreme drought in 2012 with the lowest precipitation and continuous high temperature during the growing season in this study (see Figure 2 −4 ).This drought was described as the most severe seasonal drought (during May to August) across the contiguous United States in the past 117 yr ( Hoerling et al. 2014 ), severely affecting yields across the United States ( Mallya et al. 2013 ).Our results showed that the biomass changes in 2012 and 2013 substantially deviated from the clusters of other years (see Figure 6 ).When removing the 2013 datapoint, correlation between Biomass of forb and Biomass of total biomass declined from 0.61 ( P < 0.05) to 0.26 ( P > 0.05) (see Figure 6 c), which indicates the significant effects of 2013 on forb biomass.The vegetation greenness (i.e., the lower value of NDVI in 2012 and 2013 growing seasons) could be an evidence/response for the biomass variability throughout the 15-yr period (see Figure 4 ).Accordingly, the drought of 2012 and its legacy effects in 2013 increased interannual and intra-annual biomass variability ( Griffin-Nolan et al. 2018 ).Previous studies also reported the legacy effects of drought on forbs (e.g., greater seedling establishment of forbs in 2013 after the drought) ( Hartman 2015 ;Andrade et al. 2022 ).The remarkable increase in forb biomass in 2013 was attributed to the annual forbs such as native annual sunflower ( Helianthus annuus L.) and lamb's quarters ( Chenopodium album L.), which contributed most to the increased forb production in this region, with a lesser contribution of Russian thistle ( Salsola tragus L.) in 2013 ( Hartman 2015 ).In addition, annual forbs can become productive due to the decreased competition with other perennial plants such as grasses under drought and increases in available soil nitrates associated with drier conditions ( Hartman 2015 ).However, annual forb production in our study has returned to levels close to the mean over the 15 yr since 2014, indicating a relatively rapid recovery ability of the Sandhills grasslands.
Precipitation and temperature were inversely correlated in the semiarid Sandhills grasslands during our study period (see Figure 3 c), suggesting that there could be a cumulative effect of less precipitation and higher temperatures on biomass production in the Sandhills.The inverse correlation between precipitation and temperature in the semiarid Sandhills grasslands are consistent with the anticipated drier scenario projected for the future ( Garnett 2009 ;IPCC 2021 ), which can continue to affect the biomass variability across and within seasons ( Hartman et al. 2020 ;Reeves et al. 2021 ).Hence the understanding of the climate impacts on the grasslands in the Nebraska Sandhills is meaningful for informing grazing management as these types of droughts may become more frequent in the future ( Klemm et al. 2020b ;Briske et al. 2021 ;Burruss et al. 2023 ).Experiments that simulate the impacts of variable weather on plant productivity, relative contribution to total production, and species composition at different scales can provide improved knowledge on resistance and enhance our ability to adapt to weather variability on semiarid grasslands ( Baldocchi et al. 2018 ;Poděbradská et al. 2019 ;Briske et al. 2021 ).
Our study site was selected to resemble upland sandhills ecological areas across the GSL and the broader central Sandhills region.Our observations of consistent NDVI patterns among the three spatial scales (i.e., 5-ha study site, entire GSL, and central Sandhills) through satellite remote sensing over the 15-yr period indicated similar grassland responses to precipitation and temper-ature (see Figure 4 ).Remote sensing serves as a valuable tool in landscape ecology as it enables ecologists to effectively balance the level of detail and coverage for the area of interest by selecting appropriate spatial resolutions of imagery ( Wiens 1989 ;Marceau and Hay 1999 ).This approach allows for the examination of scale-dependent patterns and processes occurring within the landscape ( Frazier et al. 2023 ;Markham et al. 2023 ).In addition, methods like multilevel modeling and meta-analysis provide a hierarchical perspective to understand the processes and patterns that are interacting within and across different scales ( Wu 1999 ;Garson 2013 ;Hassan et al. 2022 ).A multiple-scale perspective can be beneficial for landscape management, planning, and conservation tasks ( Baskent et al. 2020 ;Dorrough et al. 2020 ;Liao et al. 2023 ).For example, decisions of grazing management for GSL can be more adaptive and site specific when including finer-scale level information of specific plant community on a site and broaderscale level information of climate patterns.In our ongoing, longterm grassland monitoring project at this site, our future monitoring effort s can encompass a wide range of dat a collection, spanning from individual or population-level data, such as plant species counts, to larger biome-scale observations, such as linking remote sensing and other weather variables across larger areas ( Noss 1990 ;Marceau and Hay 1999 ;Dorrough et al. 2020 ;Briske et al. 2021 ).

Conclusions
This paper presented a detailed investigation on 15-yr (2007 to 2021) trends in total biomass, biomass of different plant functional groups, and relative plant functional group contributions in the semiarid Sandhills grasslands, providing a recent understanding of the patterns of precipitation and temperature that may drive grassland plant biomass variability.Our long-term monitoring dataset demonstrated that the Sandhill grassland biomass production is modulated through the productivity of three main plant functional groups (i.e., C 4 , C 3 grasses, and forbs).For example, the intraseason biomass change from mid-June to mid-August of total production was highly correlated ( r = 0.71, P < 0.05) to the change in C 4 grasses.We observed the relative biomass contribution of C 3 grasses to the total biomass has increased from 2007 to 2021.In contrast to the increasing trends in C 3 grasses, C 4 grasses exhibited decreasing trends throughout the 15-yr period.The summer drought in 2012 reduced grassland production not only in the 2012 late-season period but also in the 2013 early-season period due to the loss of biomass in grasses.But the increase in predominantly native annual forbs during this 2012−2013 period partially offset the total plant biomass decline.In 2014, grass production recovered rapidly, which suggests the resilience of Sandhills grasslands.In summary, our results revealed trends and changes in grassland production for different plant function groups in the Sandhills over a 15-yr period and contributed regional evidence to responses of plant functional groups as affected by climate variability and extremes.The further quantification of weather effects on the plant biomass production during this period will support adaptive grazing strategy development for grassland management in the anticipated changing climate scenario.

Figure 1
Figure 1.a, Nebraska Sandhills ecoregion (NAD_1983) and b, 5-ha study site within a larger grazed pasture of the Gudmundsen Sandhills Laboratory of the University of Nebraska−Lincoln (Projected on NAD_1983_UTM_Zone_14N).

Figure 2
Figure 2. a, Monthly cumulative precipitation (mm) and b, average temperature ( °C) between 2007 and 2021 from an on-site weather station at the Gudmundsen Sandhills Laboratory (GSL) (USW0 0 094079, south of GSL).The growing season mean was calculated on the basis of daily data from the 90th day of the year (roughly April 1) to the 270th day of year (end of September).The original dataset used here was accessed 05/02/2022.

Figure 3 .
Figure 3. a, Seasonal precipitation accumulation (PreAcc).b, Growing degree day accumulation (GDDAcc) across 15 yr and correlations between PreAcc and GDDAcc among the three growing periods by using c, 15-yr data points.The three growing periods include early-season growing period from April to mid-June (90th to 165th day of the year), late-season growing period from mid-June to mid-August (166th to 225th day of the year), and full-season growing period from April to mid-August (90th to 225th day of the year).c , * * * significance level of P < 0.001 and * significance level of P < 0.05.

Figure 4
Figure 4. a , Normalized Difference Vegetation Index (NDVI) (i.e., vegetation greenness) derived from satellite-based Moderate Resolution Imaging Spectroradiometer (MODIS) imagery from 2007 to 2021 for three spatial scales of 5-ha study site, 5 145-ha GSL, and the Sandhills ecoregions.b, Estimated day of the year in the 5-ha study site when the NDVI in each year reached the 15-yr mean threshold of 0.35 before grazing.In b, the orange dashed line indicates the mean NDVI of 0.35 that was calculated on the basis of the NDVI on the 129th day of the year for 15 yr, as a mean peak NDVI before the grazing period of the study site.A dark solid line perpendicular to the orange dashed line with a number represents the approximate date the NDVI reached the threshold of 0.35 each year.

Figure 5 .
Figure 5. Ungrazed total plant biomass as a sum of the six plant functional groups from 2007 to 2021.The six plant functional groups include C 4 and C 3 grasses, forbs, sedges, shrubs, and annual species.The vertical bar indicates the standard deviation.

Figure 6 .
Figure 6.Correlation between seasonal biomass changes in each of the six plant functional groups and amount of increase in total ungrazed biomass production across 15 yr ( n = 15) for a, C 4 grasses, b, C 3 grasses, c, forbs, d, shrubs, e, sedges, f, and annual species.In c, for forb, an embedded figure showed the decline of correlation ( r, from 0.61 to 0.26) and its change in significance level (from significant, P < 0.05, to nonsignificant, P > 0.05), when removing the 2013 datapoint, which indicates the significant effects of 2013 on forb biomass.* Significance level of P < 0.05.* * Significance level of P < 0.01.

Figure 7 .
Figure 7. Relative biomass contribution of individual plant functional group in total production across a, 15 yr and b, year-to-year trends in the biomass percentage of C 3 grasses, C 4 grasses, and forbs in the total biomass production during 2007 to 2021, respectively, in mid-June and mid-August.In a, diamond marks represent outliers in the indicated year.Trends for C 3 and C 4 grasses are significant ( P < 0.05).

Table 2
Results of Mann-Kendall test for trends of the total biomass and biomass of individual plant functional group in the Sandhills grasslands during 2007 to 2021.1Coefficient of variation at a 15-yr scale for the entire study period. 2Significance level of P < 0.05.

Table 3
Results of Mann-Kendall test for trends of relative biomass contributions for C 4 and C 3 grasses and forbs from 2007 to 2021.