Albedo-induced global warming impact of Conservation Reserve Program grasslands converted to annual and perennial bioenergy crops

Study sites are located within the northeastern part of the US Midwest Corn Belt in southwest Michigan at the Great Lakes Bioenergy Research Center of the Kellogg Biological Station's Long-term Ecological Research site (42°$27^{\prime}$N, 85°$19^{\prime}$W, 256 m asl) (figure 1). The region has a humid continental temperate climate. Mean annual air temperature is 10.2 °C and total annual precipitation averages 1005 mm with about half falling as snow (1981–2010) [26]. Total annual snowfall averages 1.3 m with three-quarters typically falling in December–February [26]. Soils are well-drained Typic Hapludalfs developed on glacial outwash [27] intermixed with loess [28].

Zoom In Zoom Out Reset image size

Four conventionally-tilled row-crop agricultural fields were set-aside in 1987 and planted to smooth brome grass—an introduced cool-season C₃ grass of Eurasian origin—for enrollment in the CRP. Three of the four fields were converted to no-till soybean in 2009; and to either no-till continuous corn, switchgrass or mixed native prairie from 2010 onwards (figure 1). Corn was planted in early May and harvested around mid-October each year (table S1 (available online at stacks.iop.org/ERL/16/084059/mmedia)). Switchgrass and mixed prairie were planted in 2010 and harvested annually around November following autumn senescence from 2011 onwards. Corn and switchgrass were fertilized at ∼180 and ∼56 kg N ha^−1 yr^−1, respectively. Restored prairie was not fertilized. The fourth field remained in smooth brome grass as a reference CRP grassland (CRP-Ref) and received no agronomic management. For details on land conversion and management see Abraha et al [29].

Incident and surface-reflected solar radiation were measured at all sites using four component net radiometers (CNR1, Kipp & Zonen, Delft, The Netherlands) placed ∼1.5 m above average canopy height. The radiation sensors were inter-calibrated among sites—for incident solar radiation on cloudless summer days, and for reflected solar radiation on cloudless winter days when surfaces were uniformly covered by snow—assuming that measurements for the respective conditions were the same across all sites. Data were logged every 30 min using Campbell CR5000 dataloggers (CR5000, Campbell Scientific Inc. Logan, UT, USA). Surface albedo (α) was computed as a ratio of the 30 min reflected to incident solar radiation between an hour after sunrise and an hour before sunset from October 2013 to September 2018. Precipitation and snow depth were retrieved from a weather station located ∼4 km from the study sites (https://lter.kbs.msu.edu/datatables).

Change in surface albedo (Δα) from converting CRP-Ref grassland to corn, switchgrass or restored prairie bioenergy crops was computed as:

$\begin{equation}\Delta \alpha = {\alpha _{{\text{bio}}}} - {\alpha _{{\text{ref}}}}\end{equation} \tag{ 1 }$

where α_bio is albedo of the bioenergy crops, and α_ref is albedo of the CRP-Ref grassland. Shortwave radiative forcing at the top of the atmosphere (${\text{RF}}_{\Delta \alpha }^{{\text{TOA}}},{ }\,{\text{W }}{{\text{m}}^{ - 2}}$) was then computed as [30]:

$\begin{equation}{\text{RF}}_{\Delta \alpha }^{{\text{TOA}}} = - \frac{1}{n}\sum^n_{n = 1} \left[ {{\text{S}}{{\text{W}}_{{\text{in}}}} \cdot {T_{\text{a}}} \cdot \Delta \alpha } \right]\end{equation} \tag{ 2 }$

where n is number of days, SW_in is the incident solar radiation on the surface, and T_a is the upward atmospheric transmittance—assumed equivalent to downward atmospheric transmittance and computed as a ratio of incident solar radiation to solar radiation at the top of the atmosphere (see supplementary material). Shortwave radiative forcing was converted into global warming impact (GWI_Δα ; kg CO₂-eq m^−2 yr^−1) [6–8, 31]:

$\begin{align}{\text{GW}}{{\text{I}}_{\Delta \alpha }}& = \frac{{{\text{RF}}_{\Delta \alpha }^{{\text{TOA}}}}}{{{ }{y_{{\text{C}}{{\text{O}}_{\text{2}}}}}\left( t \right)}} \cdot \frac{A}{{{A_{\text{E}}}}}\nonumber\\ & \quad \cdot\left( {\frac{{{\text{ln}}\left( 2 \right) \cdot {M_{{\text{CO2}}}} \cdot {m_{{\text{air}}}} \cdot {\text{C}}{{\text{O}}_{{\text{2ref}}}}}}{{\Delta {F_{2x}} \cdot {M_{{\text{air}}}}}}} \right) \cdot \frac{1}{{{\text{TH}}}}\end{align} \tag{ 3 }$

where A refers to the perturbed area (1 m²), A_E to the earth's surface area (5.1 × 10¹⁴ m²), ΔF_2x to the radiative forcing per doubling of current CO₂ concentration in the atmosphere (3.7 W m^−2), m_air to mass of the atmosphere (5.148 × 10¹⁵ Mg), M_air and M_CO2 to molecular weights of air (28.95 g mol^−1) and CO₂ (44.01 g mol^−1), respectively, CO_2ref to reference CO₂ concentration in the atmosphere (389 ppmv), TH to time horizon (100 years) and ${y_{{\text{C}}{{\text{O}}_{\text{2}}}}}\left( t \right)$ to the airborne CO₂ fraction that remains in the atmosphere at time t following a single pulse emission—derived from multi-model impulse response function analysis [32]; see supplementary material.

Data were analyzed in the statistical software R [33]. Linear mixed model fits using the nlme package [34] were used to analyze albedo for the growing season (April–September), non-growing season (October–March) and the entire crop year (October–September), with crop type as fixed effects and years as random effects. Mean albedos among seasons/years and sites were compared using Tukey's HSD test with p-values adjusted using Bonferroni correction. Treatment effects were considered significant at p < 0.05. The model discriminates significant differences among years; however, since our experimental design lacked replication, significant differences among crop treatments could be due to treatment effects as well as underlying spatial heterogeneity among sites.

Long-term observations (1981–2010) (figure 2(a)) for the region show snowfalls could occur from October to April, with highest snowfall in December–February [26]. Total annual snowfall for our sites was higher in 2014 (2.2 m), similar in 2015 (1.2 m) but lower in 2016–2018 (0.8, 0.7 and 1.0 m, respectively) compared to the long-term mean of 1.3 m (figure 2(b)) [26]. Number of days with snow depth of >100 mm on the ground, indicating snowfall accumulation, for our sites (2014–2018) are shown in figure 2(c) with a rank order (days): 2014 (76), 2015 (46), 2018 (24), 2016 (14) and 2017 (13) along with the long-term mean (31) [26].

Zoom In Zoom Out Reset image size

Changes in albedo from converting CRP-Ref grassland to corn showed similar average growing season albedo (α_gs) but higher averages of non-growing season (α_ngs) and annual (α_annual) albedos compared to those for the CRP-Ref grassland; conversion to switchgrass showed higher albedos than those for CRP-Ref grassland in all seasons; while conversion to restored prairie showed higher α_gs but similar α_ngs and α_annual compared to those for the CRP-Ref grassland (figures 3(a) and (b)).

Zoom In Zoom Out Reset image size

Average annual albedo during 2014–2018 for corn (0.30 ± 0.01; mean ± 1 standard error) was significantly higher than those for CRP-Ref grassland (0.26 ± 0.01; p< 0.001) and restored prairie (0.28 ± 0.01; p= 0.008), while α_annual for switchgrass (0.29 ± 0.01) was not statistically distinguishable from α_annual for corn or restored prairie, but was significantly higher than α_annual for CRP-Ref grassland (p< 0.001; figure 3(a)). The α_gs for switchgrass (0.21 ± 0.01) was significantly higher (p< 0.001) than α_gs for all other crops in the study; and for restored prairie (0.20 ± 0.01) than α_gs for CRP-Ref grassland and corn (0.19 ± 0.01; p< 0.001). The α_ngs for corn (0.42 ± 0.02) was significantly higher than α_ngs for all perennials (CRP-Ref grassland (0.35 ± 0.02; p< 0.001), restored prairie (0.36 ± 0.02; p< 0.001) and switchgrass (0.38 ± 0.02; p= 0.003)); and α_ngs for switchgrass was marginally higher (p< 0.046) than α_{n gs} for CRP-Ref grassland. For all crops, α_ngs was ∼two-fold higher than α_gs (p < 0.001). Annual and non-growing season cross-site albedos were highest in 2014 (p < 0.001) followed by 2015 (p < 0.001). Growing season cross-site albedos were similar among years except for significantly higher (p < 0.001) albedo in 2016 than in 2014, 2017 and 2018 (figure S1).

Monthly average albedos across years (α_m) were significantly higher during the non-growing season (December–March) than during the growing season, with the lowest α_m in early autumn (September–October) for all cropping systems (figure 4). The highest α_m for all perennials occurred in February and for corn in January. Corn had significantly higher α_m than that for CRP-Ref grassland in August–September and December–February, but significantly lower α_m in May–July, and statistically similar α_m for all other months. Switchgrass had significantly higher α_m than that for CRP-Ref grassland in April–October, but statistically similar albedos for all other months. Restored prairie showed the same statistical pattern as for switchgrass except for similar α_m to that of CRP-Ref grassland in May.

Zoom In Zoom Out Reset image size

Converting CRP-Ref grassland to corn, switchgrass or restored prairie fields caused, on average, cooling of the local climate (i.e. equivalent to net CO₂ uptake) in all seasons, except for neutral effect during the growing season at the corn field. Converting CRP-Ref grasslands to corn, switchgrass and restored prairie resulted in average annual albedo-induced GWI (GWI_{Δα-annual}; mean ± 1 se) of −0.31 ± 0.05, −0.52 ± 0.04 and −0.23 ± 0.04 MgCO₂-eq ha^−1 yr^−1; average growing season albedo-induced GWI (GWI_Δα-gs) of −0.02 ± 0.05, −0.77 ± 0.05 and −0.28 ± 0.04 MgCO₂-eq ha^−1 gs^−1; and average non-growing season albedo-induced GWI (GWI_Δα-ngs) of −0.61 ± 0.09, −0.27 ± 0.05, and −0.18 ± 0.07 MgCO₂-eq ha^−1 ngs^−1, respectively (figure 5).

Zoom In Zoom Out Reset image size

Switchgrass and restored prairie fields had significantly higher α_gs, whereas corn had statistically similar α_gs compared to that of the CRP-Ref grassland they replaced (figures 3(a) and (b)). The α_gs difference among the crop fields was likely due to leaf characteristics, crop morphology, as well as phenology and management practices such as timing of planting (for corn), green-up in spring (perennials), leaf area accumulation in summer, and eventual senescence and harvest. The perennial grasses usually green-up in April and close their canopies in May. The leaf area for switchgrass and restored prairie increases at a rate faster than that for CRP-Ref grassland, which explains their higher albedo during the growing season (figure 4). In contrast, corn is planted in early May and the ground remains relatively exposed. Hence, the corn field absorbs more solar radiation resulting in lower albedo than the perennial fields, at least until its canopy fully develops [35]. Corn had, however, significantly higher α_m in August–September than those for CRP-Ref grassland and restored prairie fields (figure 4).

Corn and switchgrass fields had significantly higher α_ngs, whereas the restored prairie field had statistically similar α_ngs compared to that of the CRP-Ref grassland they replaced (figures 3(a) and (b)). The α_ngs difference between the bioenergy crops and CRP-Ref grassland was likely due to over-winter canopy structure resulting from field management. The bioenergy crops were harvested at the end of each season and snowfalls on these fields create highly reflective white surfaces. In contrast, the CRP-Ref grassland was not harvested, with senesced plants standing throughout winter, at least partially masking snow cover and resulting in lower albedo than that of the harvested fields. For the unharvested CRP-Ref grassland to have similar albedo to those of the harvested fields, the snow had to accumulate deep enough to uniformly cover the plant stands. This likely depended on the amount and frequency of snowfall. Occasional, small snowfalls resulted in higher albedo of the harvested fields than that of the unharvested CRP-Ref grassland.

Corn also had significantly higher α_ngs than the harvested perennials. This was likely because the perennials remain unharvested for some time after corn harvest (table S1), which usually occurs before the onset of snow. Any snowfall after corn harvest but before harvest of the perennials results in higher albedo of the corn field than that of the perennial fields for reasons explained above. Even after all fields were harvested, corn may have higher albedo than the perennial fields due to differences in litter accumulation. The lower ∼10 cm of the perennial stem stalks were left behind following harvest. Some litter may also escape harvest (for example, when harvesting lodged grasses). In contrast, the corn fields were mowed and the stover partially harvested leaving less litter on the ground. This results in greater litter accumulation at the perennial fields than at the corn fields, which could potentially mask shallow snow layers resulting in lower albedo than at the corn fields. In agreement with this, Miller et al [21] reported lower albedo of switchgrass than that of corn when snow was present on the ground but overall similar α_ngs. The corn and perennial fields would likely have similar albedo once the snow depth is higher than any standing stem stalks and litter in the perennial fields.

Our observations that crop fields have higher winter albedo than grasslands agree with satellite-derived albedo observations by Zhao and Jackson [13], who found zonally averaged white-sky albedo around mid-January across North America over 45° N–60° N to be 0.57 for croplands and 0.50 for grasslands. Average albedo for January (2014–2018) at our sites ranged from 0.61 for corn to 0.51 for the CRP-Ref grassland (figure 4).

Converting CRP-Ref grassland to perennial and corn bioenergy crops caused local climate cooling on an annual scale, as indicated by negative GWI_{Δα-annual} (figure 5). For conversion to perennials, the cooling was higher during the growing season (74% for switchgrass and 61% for restored prairie) than during the non-growing season, whereas for conversion to corn, the cooling solely occurred during the non-growing season (97%) with near-neutral warming during the growing season. Negative and positive Δα between corn and the CRP-Ref grassland fields early and late during the growing season, respectively, canceled out to yield near-neutral warming (figure 4). The non-growing season Δα between perennial bioenergy crops and CRP-Ref grassland was higher than that of the growing season, albeit with higher variation (figure 3(b)). However, shortwave radiative forcing per unit Δα (equation (2)) is higher during the growing season due to the higher incident solar radiation. In general, the perennials bring albedo-induced cooling in the summer—counteracting warming, which could be vital in extreme hot weather [36], while the albedo-induced cooling provided by the corn field occurs in winter when it is already cold enough for snow to persist.

Converting CRP-Ref grassland to switchgrass, restored prairie and corn resulted in GWI_{Δα-annual} of −0.52 ± 0.04, −0.23 ± 0.04 and −0.31 ± 0.05 Mg CO₂-eq ha^−1 yr^−1, respectively. The corresponding average annual biogeochemical GWIs that accounted for GHG fluxes, farming practices, agronomic inputs and fossil fuel offsets over eight years on the same fields were −1.5, −0.3 and 9.5 Mg CO₂-eq ha^−1 yr^−1, respectively (figure 6) [19, 37]. Comparing the GWI_{Δα-annual} results of the present study with the biogeochemical GWI estimates of Abraha et al [19] indicates that the GWI_{Δα-annual} benefit (cooling) provided by the switchgrass and restored prairie fields were ∼35% and ∼78% of the biogeochemical mitigation, respectively, whereas for corn field, the benefits offset ∼3.3% of the GHG emissions incurred over the eight-year period (figure 6). Not including the fossil fuel offsets, the GWI_{Δα-annual} benefits reduced the annual net emissions (biogeochemical and biogeophysical) by 36%, 20%, and 2.2% for switchgrass, restored prairie and corn fields, respectively, over the eight-year period.

Zoom In Zoom Out Reset image size

Our study suggests that the ∼5.8 Mha CRP lands that have been converted to annual crops—primarily corn—since 2007 in the US (USDA-FSA, 2020) had GWI_Δα benefit of ∼−1.8 Mt CO₂-eq yr^−1, assuming all the CRP lands were in smooth brome grass with similar climate and snowfall. If the ∼5.8 Mha CRP lands had been converted to switchgrass or restored prairie bioenergy crops instead of corn, the GWI_{Δα-annual} benefit would have been ∼−3.0 Mt CO₂-eq yr^−1 (almost two-fold higher than that of the corn field) or ∼−1.3 Mt CO₂-eq yr^−1 (greater than two thirds of that of the corn field), respectively. In comparison, the biogeochemical GWI from same land conversions suggest an emission of ∼55.1 Mt CO₂-eq yr^−1 for corn; and uptake of ∼−8.6 and ∼−1.7 Mt CO₂-eq yr^−1 for switchgrass and restored prairie fields, respectively, for the eight years following conversion [19, 37]. It should be noted that CRP conversion to agricultural production results in very high biogeochemical emissions in the initial years of conversion, which decline over time and may eventually switch to GHG uptake as in the case of the perennials [19, 38]. Hence, following such conversions, biogeochemical GWIs show large inter-annual variations and may change direction from net GHG emission to net uptake over the lifetime of the crop while GWI_Δα tend to be relatively constant and unidirectional.

Biogeochemical GWIs have highlighted climate benefit potentials of perennial bioenergy crops over annual cropping systems—mainly corn [16, 18, 19, 39]. Converting tilled corn-soybean rotations to switchgrass and restored prairie bioenergy crops provided biogeochemical GWI benefits of −3.7 ± 0.7 and −4.6 ± 0.7 Mg CO₂-eq ha^−1 yr^−1, respectively, whereas converting to no-till continuous corn bioenergy crop was near-neutral (−0.2 ± 0.6 Mg CO₂-eq ha^−1 yr^−1) over eight years [19, 37]. Our present study suggests that if the CRP-Ref grasslands were converted to switchgrass instead of corn, the GWI_Δα benefits, on average, would provide an additional −0.22 ± 0.06 Mg CO₂-eq ha^−1 yr^−1 savings, while conversion to restored prairie instead of corn would be near-neutral (0.08 ± 0.06 Mg CO₂-eq ha^−1 yr^−1). This suggests that converting no-till corn to switchgrass or restored prairie fields results in GWI_{Δα-annual} of −0.22 ± 0.06 or 0.08 ± 0.06 Mg CO₂-eq ha^−1 yr^−1, respectively. However, the conversions to switchgrass and restored prairie fields indicated cooling during the growing season (GWI_Δα-gs; −0.75 and −0.26 Mg CO₂-eq ha^−1 gs^−1, respectively) but warming during the non-growing season (GWI_Δα-ngs; 0.31 and 0.41 Mg CO₂-eq ha^−1 ngs^−1, respectively). For corn to switchgrass conversions, Caiazzo et al [12] and Georgescu et al [10] reported GWI_Δα-gs of −0.52 and −0.68 Mg CO₂-eq ha^−1 gs^−1, respectively—similar to our GWI_Δα-gs of −0.75 Mg CO₂-eq ha^−1 gs^−1. Miller et al [21] also reported, for corn to switchgrass conversions, a warming in winter—in agreement with ours—although exact figures were not provided. In general, our findings of albedo-induced cooling benefit and near-neutral GWI_{Δα-annual} from converting no-till corn bioenergy crop to switchgrass and restored prairie bioenergy crops, respectively, combined with the relatively higher biogeochemical climate benefits due to same conversions indicate greater climate benefits of perennials over the corn bioenergy crop.

Biogeochemical impacts also indicate that climate benefits are substantially higher when corn bioenergy croplands rather than CRP grasslands are converted to perennial bioenergy crops [19, 38]. Our findings suggest that converting CRP-Ref grasslands rather than existing corn bioenergy croplands to either switchgrass or restored prairie bioenergy crop results in GWI_{Δα-annual} benefits of ∼−0.3 Mg CO₂-eq ha^−1 yr^−1—although almost all the benefits occur during the non-growing season. However, the biogeochemical GWI of converting CRP-Ref grasslands rather than existing corn bioenergy croplands to switchgrass and restored prairie bioenergy crops suggest GHG emissions of 2.2 and 4.3 Mg CO₂-eq ha^−1 yr^−1, respectively, over eight years [19, 37]. These reinforce the notion that climate benefits are higher when perennial bioenergy crops replace corn bioenergy croplands rather than conservation lands. It also demonstrates the overall higher climate benefits of perennial bioenergy crops (e.g. switchgrass and restored prairie) than that for corn e.g. [16, 18, 19].

A few additional caveats warrant mention: (a) Our results apply to settings with seasonal snow cover and would differ markedly if there were little or no snow accumulation; (b) albedo-induced GWIs considered in this study are for mature perennial stands, and may differ from albedo during conversion, planting and establishment of perennial fields; (c) GWI_Δα in this study is for 100 year horizon, which assumes fields will remain unconverted for that long and shorter (e.g. 20 year) time horizons may enhance the cooling impact proportionately; and (d) the CRP has been discouraging planting of smooth brome grass in recent years in favor of diverse native grassland species that enhance wildlife habitat, although in humid temperate climates other grassland assemblages that might be used instead of smooth brome grass may have similar albedo as long as they are not harvested.