The growing need in both producing renewable energy and food has led to the development of new systems such as agrivoltaic systems. Similar to agroforestry systems, agrivoltaics systems combine an understory of crop production with an overstory layer, casting spatial-temporal varying shade that can affect the yield of the understory crop, by either improving or reducing crop growth underneath, depending on type of crop, pedoclimatic context, photovoltaic (PV) structure and its implementation design. Current studies on agrivoltaic stations mainly evaluate global yield in areas under panels and in between panel and/or outside agrivoltaic system (ie. in full sun condition), and sometimes for various panels densities, but don’t consider shade gradient. The aim of this study was to assess field crops yields distribution around high biaxial PV installations that are trackers, with the spatial variations of shade, microclimate, soil parameters. This distribution was then used to simulate yields of different power plant designs. For this purpose, 2 types of crops (wheat and maize) were considered in 2022 and 2023, with a total of 21 trackers over 13 sites, located in Western France. Received radiation (RR) appeared as the most impacting variable of yield and biomass. Yields and biomass linearly decreased under decreasing RR, respectively by up to 45% and 40% locally with a 50% RR decrease, except for maize in 2022, probably due to exceptionally hot summer. The simulation designs showed limited global losses for both crops, only one, with the highest panel density, led to losses slightly greater than 10%.
Research Article
Effect of high biaxial solar tracker and its shade on crop yield, application to different design simulations
https://doi.org/10.21203/rs.3.rs-4848506/v1
This work is licensed under a CC BY 4.0 License
You are reading this latest preprint version
Agrivoltaic
yield
biomass
modelling
shade
Global energy demand is expected to rise from 13.5 terawatt in 2001 to 25.5 in 2022 and is likely to continue increasing due to economic and demographic growth (Global Electricity Consumption 2022 | Statista, n.d.; Lewis & Nocera, 2006) and the need to develop environmentally friendly energy. Among them, solar energy is the most abundant and available energy source (Johansson et al. 2012), but it requires large land area. Agricultural land offers a solution to this land-use challenge, with a world’s cultivated surface of 4.79 billion hectares in 2021 (FAO, 2023). Combining agricultural production with energy production to optimize the use of agricultural land, the concept behind agrivoltaism, could increase land productivity by 35–73% (Dupraz et al., 2011).
However, this production is highly dependent on climatic variations induced both by global warming (macroclimatic variation) and by the solar energy production system itself (microclimatic variation). As for macroclimatic scale, agriculture has effectively to face climate change with episodes of extreme heat and violent weather becoming more and more frequent, which would have effectively led to a 21% drop in global production in the 20th century (Ortiz-Bobea et al., 2021). Similar to the effects observed in agroforestry systems on a microclimatic scale (Jacobs et al., 2022; Kanzler et al., 2019), several studies showed that panels create a buffered microclimate underneath,, with lower temperatures during the day and warmer soil temperature at night (Barron-Gafford et al., 2016), higher humidity and limited evaporation during the day (Adeh et al., 2018; Amaducci et al., 2018; Armstrong et al., 2016), which could be beneficial to crop in a context of climate change. However, this possible side effect on crops requires to be assessed as it can easily lead to competition between plants for land and light and consequently, a reduction in food production. Contrary to agroforestry, agrivoltaic systems do not cause competition for soil nutrients.
As the concept of agrivoltaic and above all the associated research are fairly recent, crop yields in such systems have not been studied much, and the observed or simulated results are very divergent. This disparity in crop yields has biological and engineering origins. The former is due to the high variability of cropping systems, pedoclimatic contexts, interannual meteorological conditions, and environmental conditions with shading varying according to season, time of day and spatial location Dupraz, (2023). The latter is the result of highly variable PV systems in terms of design, height, panel density, position in the system, and panel distance and cardinal direction (Dupraz, 2023; Elamri et al., 2018; Marrou et al., 2013; Weselek et al., 2019). Thus, Weselek et al. (2021) showed divergent results from one crop type to another and from one year to another within the same field trial. Similarly, Amaducci et al. (2018) showed that corn yields under agrivoltaic systems increased under limiting water conditions but decreased slightly under non-limiting conditions thanks to irrigation. Today researches on agrivoltaic systems mostly compared yields under power plants, versus control yields obtained in full sun conditions (Adeh et al., 2018; Armstrong et al., 2016) or comparing yields for different panel densities (Dupraz et al., 2011; Gonocruz et al., 2021; Marrou et al., 2013). However, they do not investigate study yields heterogeneity within a power plant, nor under a continuous shade gradient. The latter approaches are of crucial importance as they should help optimise PV design for crop yields.
Indeed, on a more fine-grained level, several parameters linked to shading, including intensity and duration, can have an impact on crop yield, which itself depends on the stage of development involved (Fischer, 1985; Mcmaster et al., 1987; Savin & Slafer, 1991; Xu et al., 2016). In cereals, shading prior to anthesis mainly reduces ear density, number of spikelets, and number of kernels per ear, when post-anthesis shading primarily reduces grain weight, then the number of grains (Fischer, 1985; Mcmaster et al., 1987; Savin & Slafer, 1991). Light shading after anthesis tends to increase wheat yield by delaying leaf senescence and increasing the plant's capacity to photosynthesize and accumulate dry matter during grain filling. By contrast, high or moderate shade during this period reduces yields.
Most studies concerned agrivoltaic systems with low ground-mounted PV system. In such systems, shade is projected close to the photovoltaic system (Dupraz et al., 2011). Higher panels allow more diffuse radiations to reach soil and plants and generate a shadow moving quicker along the day. Other and higher systems also exist and had not been studied yet. This is the case of high biaxial solar tracker. With their height and small anchoring surface, along with a limited covering ratio, solar tracker can be adapted to all types of agricultural systems, with tractors passages or high solar needs plants. Furthermore, their biaxial moving capacity allow the system to track the sun and to produce more electricity than fixed panels for the same surface.
In our experiment, we considered the shade’s gradient all around a high biaxial solar tracker, on multiple sites located in western France (Fig. 1). The main aim of this study is to assess wheat and maize yields not only at a single focal point but at multiple points under the solar tracker, in order to capture the effect of changing shading conditions and related abiotic factors on plant growth. Based on the regression results obtained between crop yields and the most relevant and significant explanatory variables considered, we then simulated crop yields and plant biomass by testing different agrivoltaics plant designs with the aim of arriving at a spatial configuration of trackers that would ensure a loss of yield and biomass lower than that authorised by the regulations in force.
2.1 Experimental design and management
The OKwind solar tracker system is a high solar system developed by OKwind (Torcé, France). It consists of a plateau of 117m² of solar panel, mounted at 7m heigh, that can follow the sun during the day. It is implanted in soil with a concrete square block of 2.5x2.5m and 2m depth (figure 2).
The experimental sites were located in western France (Bretagne, Normandie and Pays-de-Loire). All sites are under a Cfb climate (name of the classification). Information for each site is available in complementary data (table 4). One to two solar trackers per site were studied. The field crop types considered for the two field years in 2022 and 2023 are those most frequently farmed in the study area, i.e. wheat and maize. In total, 24 trackers were studied, 6 in 2022 (2 for wheat and 4 for maize), and 18 in 2023 (9 for wheat and 9 for maize).
2.2 Sampling design
To assess solar tracker’s impact on crop yield, a sampling plan was built based on the study of tracker thrown shade and total radiation received (RR) by crop (figure 3) (Noirot-Cosson et al., 2022).(Noirot-Cosson et al., 2022). To grasp the shade gradient effect, the eight cardinal axes were considered, as well several distance from the solar tracker centre (5 meters, 10 meters, 15 meters, 20 meters, 25 meters (only in 2023) and 35 meters), leading to 48 studied points (40 in 2022). Maximum studied distance from the tracker was fixed at 35 meters, where received radiation is reduced by only 1% on crop cultural period.
2.3 Collected data
a. Field and crop data
Crop development and growth were monitored every two weeks from sowing to harvest. At crop maturity, aboveground biomass was sampled on 1m² for each sampling point and weighted. Grains and straw were separated, and crop grain were weighted.
Tractors wheel’s impact was noted if sampling point were less than a meter close to it. Wiring underground was noted for each site using field cartography of tracker implantation. These two parameters were each noted with a presence/absence notation, with 1 when a sampling point was impacted by wiring and tractor’s wheel, and 0 when not.
To explore tracker influence on crop yield, field related parameters were considered (table 1). For each wheat and maize sites, data about tracker age and field management were gathered. Sowing date, harvest date, and cultivated varieties were collected. Studied varieties consisted in 3 varieties for wheat (Chevignon, LG Absalon, and a mixt of different varieties) and 4 varieties in maize (Caroleen, Havelio KWS, LG31270, Amandeen).
b. Microclimatic and soil data
Microclimatic data and soil density were measured on 4 of the 9 wheat trackers studied. The latter measurements were focused on a limited number of sites. On these trackers, microclimatic data and soil density were collected at every sampling point, excepted those at 25 meters from the tracker. Soil temperature (°C) and humidity (%H) were measured using TMS4 (TOMST, Czech Republic) and air temperature (°C) and humidity (raw TMS soil moisture) were measured using EM300-TH (Milesight, Fujian, China). Once field mechanical treatment were over, microclimatic sensors were placed. Sensors were placed in early March for the TMS4 and in early April for the EM300-TH. All sensors were removed in July at harvest. Data was measured every 10 minutes by the TMS4 and every 15min for the EM300-TH. To ensure that air temperature was measured at canopy level, sensors were moved regularly to stay at canopy height. Soil compaction (kPa) was measured after harvest using a compaction meters Soil compaction tool (Spectrum Technologies, Inc., Aurora, Illinois, U.S.A) at the same sampling points than microclimatic data. Soil samples for soil texture and chemical properties analysis were taken at a 30cm depth for each studied point. Analyses were performed by an outside laboratory (Laboratoire d’analyses des sols (LAS), Arras, France). Quantities of peebles (<5mm) (g/kg), gravel (2-5mm) (g/kg), dirt (<2mm) (g/kg), total nitrogen (g/kg), total carbon (g/kg), total organic carbon (COT) (g/kg), C/N ratio, organic matter (g/kg), clay (0 to 0.002mm) (g/kg), silt (0.002 to 0.063mm) (g/kg) and sand (0.063 to 2mm) (g/kg) were analysed.
c. RR calculations
Shade cast by tracker was considered using simulation of total received radiation (RR) under each solar tracker. Sowing and harvest date were considered for each site, as well as shadow cast by nearby trackers. This allows us to integrate potential overlapping shade in the RR. RR was then calculated using the equation bellow (equation 1), based on Noirot-Cosson et al. (2022) .
Equation 1: Calculation of % RR
RR were considered: (i) equal to diffuse radiations when under a tracker shadow, and to the sum of direct and diffuse radiations otherwise, (ii) on plants implementation periods. Then, the total received radiations ratio (%RR) was calculated as in the equation below. Rdh stands for radiation received by plants (Wh.m-2) at a certain hour (h) and certain day (d), Rrefdh stands for Rdh in a location with no tracker shadow (Noirot-Cosson et al., 2022).
2.4 Statistical analysis
We first calculated yield relative ratio for each crop and site. Reference yield was calculated using sampling point for which RR was between 0.99 and 1 for each site, representing 18 plots over the 48, leading to 1 reference yield per site. Then, relative yield was calculated for each sampling point. Relative aboveground biomass ratio was calculated using the same method.
All statistical analysis was performed using the R software. Statistical analysis was done using Generalised Linear Mixted Model (GLMM). For crop yield, three GLMM were realised, two for wheat: one with microclimate and soil data (equation 2: pedoclimatic model), two without pedo-climatic data (equation 3: all sites model) for wheat and maize . The same models were used to study crop biomass (equation 4 and 5). Variables used for each model are available in complementary data (table 5).
Equation 2: lmer model used to for wheat yield study using microclimate data (pedoclimatic model)
lmer (Relative grain yield ~ Air humidity + Soil humidity + Air temperature + Soil temperature + Soil_ compaction (8cm depth) + Soil_ compaction (16cm depth) + Soil_ compaction (24cm depth) + Tractor wheel impact + RR + cultivated variety + wiring : tracker’s_age + cardinal axis : distance + reference yield + Peeble + Gravel + Dirt + N_total + C_total + COT + C/N + MO + Clay + Silt + Sand + (1|site))
Equation 3: lmer model used for wheat and maize yield study using all data (all sites model)
lmer (Relative grain yield ~ Tractor wheel impact + RR + cultivated variety + wiring : tracker’s_age + cardinal axis : distance + reference yield + (1|site))
Equation 4: lmer model used to for wheat biomass study using microclimate data (pedoclimatic model)
lmer (Relative aboveground biomass ~ Air humidity + Soil humidity + Air temperature + Soil temperature + Soil_ compaction (8cm depth) + Soil_ compaction (16cm depth) + Soil_ compaction (24cm depth) + Tractor wheel impact + RR + cultivated variety + wiring : tracker’s_age + cardinal axis : distance + reference yield + Peeble + Gravel + Dirt + N_total + C_total + COT + C/N + MO + Clay + Silt + Sand + (1|site))
Equation 5: lmer model used for wheat and maize yield study using all data (all sites model)
lmer (Relative aboveground Biomass ~ Tractor wheel impact + RR + cultivated variety + wiring : tracker’s_age + cardinal axis : distance + reference yield + (1|site))
The GLMM examines the effect of field management, cultivated variety and tracker age (independent variables) on crop yield or crop biomass (dependent variables), while considering the site as a random effect. This gave GLMM result for every site (all sites model). For the 4 wheat sites studied with microclimatic conditions and soil density, two separate GLMM were done including this data (pedoclimatic model) for yield and biomass. The GLMM allowed to determine the most impacting variable on crop yield and aboveground biomass for each culture. Once the most impacting variable were identified, linear regressions were traced, one per culture and per year.
2.5 Simulation of yield on different solar power plant design
To explore the effect of different agrivoltaic plant designs on crop grain and biomass yields , 3 theorical agrivoltaic fields were built. Fields with 20 trackers were considered as it represents a targeted dimension for such plants, and as, a rectangle form allows easier simulations and results interpretations. Different distances between PV trackers were tested (table 1). In the north-south axis, distance of 25 meters, 30 meters and 35 meters were tested, commonly set to avoid shadow from one tracker to the northern one. In the east-west axis, 40 meters, 50 meters and 58 meters were tested based on one or two passages of agricultural sprayer and with popular sprayer widths. The three tested designs are illustrated figure 4. For these three designs, RR was calculated for each m² based on 2022-2023 weather data from Torcé (35870), the Okwind headquarters location around which most of trackers were installed, 40km east from Rennes, and for both wheat and maize classical implementation period: from the 1st of October to the 1st of July for wheat and from the 15th of April to the 15th of September for Maize. Then, using the built regression between crop yields and RR, crop yields were spatially simulated, and the average field yields were calculated, using the linear regression previously traced. All linear regression were used, except the one for “maize 2022” for both yield and aboveground biomass simulation. An outer field area was considered and dimensioned on half of the inter PV tracker distance, in order to avoid as much as possible a border effect and to make results scalable to design with a different number of PV trackers.
Table 1: Experimental designs characteristics, with North-South distance and East-West distance between each tracker, as well as tracker density for a theorical field for each design.
|
|
North-South distance |
East-West distance |
Tracker density |
|
Design 1 |
25 m |
40 m |
10 |
|
Design 2 |
30 m |
50 m |
6.66 |
|
Design 3 |
35 m |
58 m |
4.93 |
2.6 Method summary (figure 5)
3.1 Data description
Yield mean for all wheat sites was around 1029g/m² in 2023 and 1319g/m² in 2022. For maize, grain yield mean was 2504g/m² in 2023 and 837g/m² in 2022. Wheat yield difference between the maximum and minimum within a site is around 2% for most sites, with two sites having a difference between maximum and minimum of 6% and 8%. Yield difference between maximum and minimum is more important for maize, most sites showed a difference of around 10%, with one site having a 20% difference.
Aboveground biomass results in wheat follow the same tendency as grains yield. Mean wheat aboveground biomass was around 2059g/m² in 2023 and 2304g/m² in 2022. Difference between maximum and minimum being around 2% as well for those two exceptional sites showing a difference of 6% and 8%. On the contrary, tendencies for maize aboveground biomass did not follow the same trend as grain yield. Mean maize aboveground biomass was around 9089g/m² in 2023 and 3839g/m² in 2022. Difference between maximum and minimum is smaller than for grain yield, with this difference being around 2% for all sites, except for one having an 8% difference.
3.2 GLMM result
Table 2: GLMM results on grains and biomass yields for wheat (2 models: pedoclimatic model and all sites model) and maize (1 model: all sites model). P-value is compiled for each GLM model and variable. P-value < 0.05 : significant. (NA : variable not studied in GLMM model ; RR: received radiation ; COT: Total organic carbon ; Soil compaction 8, 16 and 24 : soil compaction at 8, 18 and 24 cm in depth respectively ; Age: tracker and underground wiring age ; Card axis dist.: cardinal axis and distance ; Wheel impact : tractor wheel impact ; Variety: cultivated variety)
|
|
GRAIN YIELD |
ABOVEGROUND BIOMASS |
||||
|
Model |
Wheat pedoclimatic model |
Wheat all sites model |
Maize all sites model |
Wheat pedoclimatic model |
Wheat all sites model |
Maize all sites model |
|
Variables |
|
|
|
|
|
|
|
Air humidity |
0.836055 |
NA |
NA |
0.49652 |
NA |
NA |
|
Soil humidity |
0.318522 |
NA |
NA |
0.44311 |
NA |
NA |
|
Air temperature |
0.867178 |
NA |
NA |
0.60768 |
NA |
NA |
|
Soil temperature |
0.995359 |
NA |
NA |
0.99456 |
NA |
NA |
|
RR |
1.15-09 |
<2.2-16 |
6.742e-09 |
1.343-05 |
5.789-12 |
4.912-10 |
|
Peeble |
0.825865 |
NA |
NA |
0.0994 |
NA |
NA |
|
Gravel |
0.837631 |
NA |
NA |
0.09511 |
NA |
NA |
|
Dirt |
0.834376 |
NA |
NA |
0.09169 |
NA |
NA |
|
Total Nitrogen |
0.001182 |
NA |
NA |
0.05529 |
NA |
NA |
|
Total Carbon |
0.889780 |
NA |
NA |
0.9999 |
NA |
NA |
|
COT |
0.249342 |
NA |
NA |
0.29457 |
NA |
NA |
|
C/N |
0.005398 |
NA |
NA |
0.09463 |
NA |
NA |
|
Organic matter |
0.456803 |
NA |
NA |
0.40442 |
NA |
NA |
|
Clay |
0.048998 |
NA |
NA |
0.40739 |
NA |
NA |
|
Silt |
0.023935 |
NA |
NA |
0.01994 |
NA |
NA |
|
Sand |
0 |
NA |
NA |
0 |
NA |
NA |
|
Soil compaction 8 |
0.584291 |
NA |
NA |
0.36502 |
NA |
NA |
|
Soil compaction 16 |
0.758211 |
NA |
NA |
0.76200 |
NA |
NA |
|
Soil compaction 24 |
0.138802 |
NA |
NA |
0.20014 |
NA |
NA |
|
Age |
0.55121 |
0.002792 |
0.06225 |
0.22855 |
0.05049 |
0.2430 |
|
Card axis dist. |
0.023527 |
0.062235 |
0.97109 |
0.49576 |
0.88156 |
0.2702 |
|
Wheel impact |
0.057469 |
0.417705 |
NA |
0.91119 |
0.78898 |
NA |
|
Variety |
0.356628 |
0.589827 |
0.06109 |
0.08102 |
0.73734 |
4.963-8 |
|
Reference yield |
0.105206 |
0.814792 |
0.80256 |
0.19885 |
0.86449 |
0.9975 |
For grain yield, both GLMM model on wheat data showed satisfying Rc² of 0.54 (equation 2 and 3). Maize all sites model GLMM (equation 3) accuracy was lower, with a Rc² of 0.2 for grain yield data. For aboveground biomass analysis, GLMM model showed Rc² of 0.36 for the pedoclimatic model (equation 5), 0.43 for all sites model in wheat (equation 6) and 0.18 in all sites model in maize (equation 6). Also, all three GLM models showed high values of residual deviance making them unsuitable for data prediction.
GLMM results showed that crop yields were mostly driven by RR in all tested models, on grain as well as aboveground biomass (table 2). The combined effect of wiring and tracker’s age (related to a supposed impact of tracker installation on soils) significantly impacted grain yields (p = 0.003) but only for wheat all sites model (equation 3). In the pedoclimatic model (equation 2), the variables RR and in a lower manner cardinal axis:distance, as well as some soil parameters (total nitrogen, C/N, clay and silt) were the only significant variables for grain yields prediction. For wheat aboveground biomass, the pedoclimatic model (equation 5) showed a significant influence of RR, and a smaller influence of silt and cultivated variety. These results suggested weak or unsignificant effects on crop yields and on aboveground biomass of the supposed soil impacts of tracker installation despite the relatively young age installations (mean age of almost 3 years) for both crops. They showed a strong impact of tracker shadow directly through RR while the putative indirect effects through air and soil temperature and humidity were not observable. For maize, the all sites GLMM models showed also significant impact of cultivated variety on aboveground biomass, but not on crop yield, suggesting a site effect stronger than for wheat.
3.3 Linear regression result
Linear regressions (figure 6, A and B) between RR and yield showed R² (between 0.38 and 0.55) and similar trend, except for “maize 2022” characterized by high dispersion of data and a flat trend. Linear regression between aboveground biomass and RR showed R² between 0.12 and 0.37, with “maize 2022” having flat trend as well, though less important than for maize yield.
The “maize 2022” specificity might be mostly due to the heat wave of summer 2022 which drastically reduced maize grain yields (1244.69g/m² of average maize yield in 2022 compared to 3125.67g/m² whereas average maize yield in 2023 was on average for all sites and plots 3125.67g/m²). This heat wave could have impacted flowering by causing flower abortion, thus reducing yield, whereas aboveground biomass was still impacted by shade probably due to crop height reduction. During this summer 2022, the impact of tracker shadow was less detrimental for crop yields probably because also preventing the crop from too much radiation, heat or evapotranspiration. These 2022-2023 maize yields differences highlighted the dependence of climatic and meteorological contexts for one PV system effects on crops. In Weselek et al. (2021), differences in winter wheat yields between agrivoltaic system and control, were inversed from on year to another. In 2017, winter wheat yield decreased by 19% under agrivoltaic, and in 2018 increased by 2,7%, the latter year being a special climatic year, with a hot and dry summer. Under these special conditions, radiation reduction under agrivoltaic system may not have been limiting for crop yield, but draught was, especially under full sun conditions.
Linear regression showed that grain yields loss around 45% when RR is decreased by 50%, except for “maize 2022”. For aboveground biomass, losses were less important being around 40% when RR is decreased by 50%. These results are consistent with other studies carried out on wheat and maize in reduced incident light conditions and shade where the level of yield losses increased with the level of shade (Artru et al., 2017; Dufour et al., 2013; Li et al., 2010; Mu et al., 2010; Ramos-Fuentes et al., 2023) notably in an agrivoltaic context (Dupraz, 2023). However, the intensity of crop response to shade depends on the intensity and applied period of shade (Artru et al., 2017; Mu et al., 2010). Here, the impact of shade has been considered in relation to crop yield, without concern for the period of shading, nor for grain quality and protein content. The linear relationship between yield and RR and aboveground biomass and RR contrast with result obtain by Laub et al. in their meta-analysis (2022), though the decreased in yield with solar radiation reduction’s relationship remain similar.
3.4 Application to different tracker configuration
Simulation results according to different design are presented in figure 7 and table 3. As the density of PV panels increased, yield losses also increased. For trackers density of 10 (design 1), 4.93 (design 2) and 6.66 (design 3), the simulated yield losses for wheat were 11%, 6% and 7% respectively, and for maize, they were 14%, 7% and 10% respectively. The simulations on aboveground wheat biomass show almost identical results to those of the simulations on wheat yields. However, for maize, the simulated aboveground biomass losses were less than the yield losses, with results of 7%, 4%, and 3% for designs 1, 2, and 3, respectively. Laub et al. (2022) also found that maize grain yields decrease more sharply than wheat yields with decreasing solar radiations. A similar reduction in grain yield was observed in European climate agroforestry systems for wheat (Ivezić et al., 2021) and maize (Baier et al., 2023).
Among all three designs, design 1 showed more important yield losses, up to 14% in maize, which is way above the zero yield losses policy of the French agrivoltaic law, and its 10% losses tolerance, and AFNOR label (French Law, Republique Française 2023; AFNOR 2021). Here, the outer field area was considered and dimensioned on half of the inter PV tracker distance, to avoid as much as possible a border effect, and making the different designs comparable. It differed from the method to determine the considered field area for crop yield calculations edited in the French agrivoltaic decree which considered a surface delimited by the outer panels border. However, the latter method rather fit with large and dense solar systems, with low border effect, where the studied system with trackers was small and with low covering ratio.
Table 3: Percentage of remaining yield and remaining biomass for each experimental simulation design for wheat and maize
|
|
Grain yield |
Aboveground biomass |
||
|
Wheat |
Maize |
Wheat |
Maize |
|
|
Design 1 |
89% |
86% |
89% |
93% |
|
Design 2 |
93% |
90% |
92% |
96% |
|
Design 3 |
94% |
93% |
94% |
97% |
The aim of this study was to empirically investigate the influence of agrivoltaic high biaxial solar tracking on wheat and maize yields and above-ground biomass. Unlike previous studies which focused on the comparison of crop yields under agrivoltaic solar panel versus crop yield obtained in full sun, taking into account or not different panel densities, our study is innovative and more in-depth because it takes into account the heterogeneities of yield and above-ground biomass within the same power plant, i.e. in a continuous shade gradient. We have effectively measured the effect of these trackers by taking into account variables associated either directly with the trackers such as shading, the resulting microclimate, soil compaction, or indirectly such as the age of the tracker, the crop variety and weather conditions.
Our results showed that the reduction of radiations received by plants was the most impacting factor on crop yield and aboveground biomass, when microclimate and soil were less impacting. It highlighted that this impact is dependent on weather and light conditions during crop growth for example. They also allowed to explore the effect of high biaxial solar tracker on different agrivoltaic power plant design on crop yield and to determine the acceptable tracker density for crop and energy production, with less than 10% of losses.
The results obtained will be strengthened and deepened by carrying a study exploring the ecophysiological mechanisms which link the agrivoltaic environment to agricultural yields. Other parameters regarding photosynthesis efficiency or physiological response of plants to shade should be investigated in future studies.
Declarations
Conflicts of interest/Competing interests :
Three of the authors (Clementine Inghels, Paul-Emile Noirot-Cosson and Tanguy Riou) work for the OKwind group that is financing the research project.
Ethics approval :
Not applicable
Consent to participate :
Not applicable.
Consent for publication :
Not applicable.
Funding :
This research is funded by the OKwind group.
Author Contribution
Conceptualization: C.I, PE.NC, A.G, T.K ; Data curation: C.I, T.R; Formal analysis: C.I ; Investigation: C.I ; Methodology: PE.NC, A.G, T.K ; Project administration: C.I ; Resources: C.I, PE.NC, A.G, T.K ; Supervision: C.I, PE.NC, A.G, T.K ; Visualization: C.I ; Writing – original draft preparation: C.I ; Writing – review & editing: C.I, PE.NC, A.G, T.K
Data Availability
Data sets generated during the current study are available from the corresponding author on reasonable request.
- Adeh, E. H., Selker, J. S., & Higgins, C. W. (2018). Remarkable agrivoltaic influence on soil moisture, micrometeorology and water-use efficiency. PLoS ONE, 13(11). https://doi.org/10.1371/journal.pone.0203256
- Amaducci, S., Yin, X., & Colauzzi, M. (2018). Agrivoltaic systems to optimise land use for electric energy production. Applied Energy, 220, 545–561. https://doi.org/10.1016/j.apenergy.2018.03.081
- Armstrong, A., Ostle, N. J., & Whitaker, J. (2016). Solar park microclimate and vegetation management effects on grassland carbon cycling. Environmental Research Letters, 11(7). https://doi.org/10.1088/1748-9326/11/7/074016
- Artru, S., Garré, S., Dupraz, C., Hiel, M. P., Blitz-Frayret, C., & Lassois, L. (2017). Impact of spatio-temporal shade dynamics on wheat growth and yield, perspectives for temperate agroforestry. European Journal of Agronomy, 82, 60–70. https://doi.org/10.1016/j.eja.2016.10.004
- Baier, C., Gross, A., Thevs, N., & Glaser, B. (2023). Effects of agroforestry on grain yield of maize (Zea mays L.)—A global meta-analysis. Frontiers in Sustainable Food Systems, 7, 1167686. https://doi.org/10.3389/FSUFS.2023.1167686/BIBTEX
- Barron-Gafford, G. A., Minor, R. L., Allen, N. A., Cronin, A. D., Brooks, A. E., & Pavao-Zuckerman, M. A. (2016). The Photovoltaic Heat Island Effect: Larger solar power plants increase local temperatures. Scientific Reports, 6. https://doi.org/10.1038/SREP35070
- Dufour, L., Metay, A., Talbot, G., & Dupraz, C. (2013). Assessing light competition for cereal production in temperate agroforestry systems using experimentation and crop modelling. Journal of Agronomy and Crop Science, 199(3), 217–227. https://doi.org/10.1111/jac.12008
- Dupraz, C. (2023). Assessment of the Ground Coverage Ratio of AgriVoltaic systems as a proxy for potential crop productivity. https://doi.org/10.21203/RS.3.RS-3030967/V1
- Dupraz, C., Marrou, H., Talbot, G., Dufour, L., Nogier, A., & Ferard, Y. (2011). Combining solar photovoltaic panels and food crops for optimising land use: Towards new agrivoltaic schemes. Renewable Energy, 36(10), 2725–2732. https://doi.org/10.1016/j.renene.2011.03.005
- Elamri, Y., Cheviron, B., Lopez, J. M., Dejean, C., & Belaud, G. (2018). Water budget and crop modelling for agrivoltaic systems: Application to irrigated lettuces. Agricultural Water Management, 208, 440–453. https://doi.org/10.1016/j.agwat.2018.07.001
- Fischer, R. A. (1985). Number of kernels in wheat crops and the influence of solar radiation and temperature. The Journal of Agricultural Science, 105(2), 447–461. https://doi.org/10.1017/S0021859600056495
- Global electricity consumption 2022 | Statista. (n.d.). Retrieved April 10, 2024, from https://www.statista.com/statistics/280704/world-power-consumption/
- Gonocruz, R. A., Nakamura, R., Yoshino, K., Homma, M., Doi, T., Yoshida, Y., & Tani, A. (2021). Analysis of the Rice Yield under an Agrivoltaic System: A Case Study in Japan. https://doi.org/10.3390/environments8070065
- Ivezić, V., Yu, Y., & Werf, W. van der. (2021). Crop Yields in European Agroforestry Systems: A Meta-Analysis. Frontiers in Sustainable Food Systems, 5, 606631. https://doi.org/10.3389/FSUFS.2021.606631/BIBTEX
- Jacobs, S. R., Webber, H., Niether, W., Grahmann, K., Lüttschwager, D., Schwartz, C., Breuer, L., & Bellingrath-Kimura, S. D. (2022). Modification of the microclimate and water balance through the integration of trees into temperate cropping systems. Agricultural and Forest Meteorology, 323, 109065. https://doi.org/10.1016/J.AGRFORMET.2022.109065
- Kanzler, M., Böhm, C., Mirck, J., Schmitt, D., & Veste, M. (2019). Microclimate effects on evaporation and winter wheat (Triticum aestivum L.) yield within a temperate agroforestry system. Agroforestry Systems, 93(5), 1821–1841. https://doi.org/10.1007/S10457-018-0289-4
- Kemp, D., & Whingwiri, E. (1980). Effect of Tiller Removal and Shading on Spikelet Development and Yield Components of the Main Shoot of Wheat and on the Sugar Concentration of the Ear and Flag Leaf. Functional Plant Biology, 7(5), 501–510. https://doi.org/10.1071/PP9800501
- Land use statistics and indicators Global, regional and country trends Land use statistics and indicators Global, regional and country trends 1990–2019 FAOSTAT Analytical Brief 28 LAND USE AND LAND USE INDICATORS BACKGROUND. (2019).
- Lewis, N. S., & Nocera, D. G. (2006). Powering the planet: Chemical challenges in solar energy utilization. Proceedings of the National Academy of Sciences of the United States of America, 103(43), 15729–15735. https://doi.org/10.1073/PNAS.0603395103
- Li, H., Jiang, D., Wollenweber, B., Dai, T., & Cao, W. (2010). Effects of shading on morphology, physiology and grain yield of winter wheat. European Journal of Agronomy, 33(4), 267–275. https://doi.org/10.1016/j.eja.2010.07.002
- Marrou, H., Guilioni, L., Dufour, L., Dupraz, C., & Wery, J. (2013). Microclimate under agrivoltaic systems: Is crop growth rate affected in the partial shade of solar panels? Agricultural and Forest Meteorology, 177, 117–132. https://doi.org/10.1016/j.agrformet.2013.04.012
- McMaster, G. S., Morgan, J. A., & Willis, W. O. (1987). Effects of Shading on Winter Wheat Yield, Spike Characteristics, and Carbohydrate Allocation1. Crop Science, 27(5), 967–973. https://doi.org/10.2135/CROPSCI1987.0011183X002700050030X
- Mu, H., Jiang, D., Wollenweber, B., Dai, T., Jing, Q., & Cao, W. (2010). Long-term Low Radiation Decreases Leaf Photosynthesis, Photochemical Efficiency and Grain Yield in Winter Wheat. Journal of Agronomy and Crop Science, 196(1), 38–47. https://doi.org/10.1111/J.1439-037X.2009.00394.X
- Noirot-Cosson, P. E., Riou, T., & Bugny, Y. (2022). Toward assessing photovoltaic trackers effects on annual crops growth and building optimized agrivoltaics systems based on annual crops. AIP Conference Proceedings, 2635(1). https://doi.org/10.1063/5.0103326/2830641
- Ortiz-Bobea, A., Ault, T. R., Carrillo, C. M., Chambers, R. G., & Lobell, D. B. (2021). Anthropogenic climate change has slowed global agricultural productivity growth. Nature Climate Change, 11(4), 306–312. https://doi.org/10.1038/s41558-021-01000-1
- Ramos-Fuentes, I. A., Elamri, Y., Cheviron, B., Dejean, C., Belaud, G., & Fumey, D. (2023). Effects of shade and deficit irrigation on maize growth and development in fixed and dynamic AgriVoltaic systems. Agricultural Water Management, 280, 108187. https://doi.org/10.1016/J.AGWAT.2023.108187
- Rogner, H., Dusseault, M. B., Rogner, H.-H., Authors Roberto Aguilera, L. F., Archer, C. L., Bertani Enel Green Power SpA, R., Bhattacharya, I. S., Gagnon HydroQu é bec, L., Helmut Haberl, C., Hoogwijk, M., Johnson, A., Rogner, M. L., Wagner, H., Yakushev Gazprom, V., Contributing Authors Doug Arent, R. J., Bryden, I., Krausmann, F., Odell, P., Schillings Ali Shafi ei, C., & Zou, J. (n.d.). Chapter 7-Energy Resources and Potentials. In Global Energy Assessment-Toward a Sustainable Future Land Use, climate change and biodiversity in cultural landscapes: Assessing feedbacks and promoting land-use strategies towards a viable future [LUBIO] View project Hydrogen Production-PEC View project Energy Resources and Potentials Convening Lead Author (CLA). https://doi.org/10.13140/RG.2.1.3049.8724
- Savin, R., & Slafer, G. A. (1991). Shading effects on the yield of an Argentinian wheat cultivar. The Journal of Agricultural Science, 116(1), 1–7. https://doi.org/10.1017/S0021859600076085
- Weselek, A., Bauerle, A., Zikeli, S., Lewandowski, I., & Högy, P. (2021). Effects on crop development, yields and chemical composition of celeriac (Apium graveolens L. var. rapaceum) cultivated underneath an agrivoltaic system. Agronomy, 11(4). https://doi.org/10.3390/agronomy11040733
- Weselek, A., Ehmann, A., Zikeli, S., Lewandowski, I., Schindele, S., & Högy, P. (2019). Agrophotovoltaic systems: applications, challenges, and opportunities. A review. In Agronomy for Sustainable Development (Vol. 39, Issue 4). Springer-Verlag France. https://doi.org/10.1007/s13593-019-0581-3
- Xu, C. long, Tao, H. bin, Wang, P., & Wang, Z. lin. (2016). Slight shading after anthesis increases photosynthetic productivity and grain yield of winter wheat (Triticum aestivum L.) due to the delaying of leaf senescence. Journal of Integrative Agriculture, 15(1), 63–75. https://doi.org/10.1016/S2095-3119(15)61047-4
- Total number of words (including references): 6784 words
- Total number of words (excluding references): 5228 words
Table 4 is not available with this version.
Table 5: Data collected for each GLMM model
|
Variable |
Wheat microclimate trackers (n = 4) |
Wheat all trackers (n = 11) |
Maize all trackers (n = 13) |
|
Air humidity |
X |
NA |
NA |
|
Soil humidity |
X |
NA |
NA |
|
Air temperature |
X |
NA |
NA |
|
Soil temperature |
X |
NA |
NA |
|
RR |
X |
X |
X |
|
Peeble |
X |
NA |
NA |
|
Gravel |
X |
NA |
NA |
|
Dirt |
X |
NA |
NA |
|
Total Nitrogen |
X |
NA |
NA |
|
Total Carbon |
X |
NA |
NA |
|
Total organic carbon (COT) |
X |
NA |
NA |
|
C/N |
X |
NA |
NA |
|
Organic matter |
X |
NA |
NA |
|
Clay |
X |
NA |
NA |
|
Silt |
X |
NA |
NA |
|
Sand |
X |
NA |
NA |
|
Soil compaction (8cm depth) |
X |
NA |
NA |
|
Soil compaction (16cm depth) |
X |
NA |
NA |
|
Soil compaction (24cm depth) |
X |
NA |
NA |
|
Wiring : tracker’s age |
X |
X |
X |
|
Cardinal axis : distance |
X |
X |
X |
|
Tractor wheel impact |
X |
X |
NA |
|
Cultivated variety |
X |
X |
X |
|
Reference yield |
X |
X |
X |
No competing interests reported.