Analysis of the Effect of Concentrations of Four Whitening Products in Cover Transmissivity of Mediterranean Greenhouses

The present work analyses the traditional method of applying whitening products on Mediterranean greenhouses. Four commercial whitening products (agricultural solar protectors, ASPs), applied at four doses, were compared with a non-whitened cover. The traditional product “Blanco de España” with 99% calcium carbonate (CaCO3) and other three products with 97% CaCO3 that incorporate adhesives were tested. The use of adhesives in ASP did not influence the effect of the different products on the inside temperature, and at the same dose all four products show a similar behaviour. The findings support the maximum dose recommended by other authors of 0.50 kg L−1 (50/100), above which the transmissivity of the greenhouse cover decreases by over 50%. The effect of ASP on the transmissivity of the cover depends principally on the dose applied, but also on the climatic conditions (solar radiation, cloud cover, etc.) and on the time of year (solar elevation). The habitual use of a constant dose throughout the year does not seem to be the most adequate. Recommended doses should vary according to the time of year and the desired degree of transmissivity reduction. The adhesive components are shown to provide a high degree of protection against heavy rain. The study recommends a standardised method of ASP application, establishing a method that allows the grower to verify the concentration of the product that will remain on the greenhouse cover.


Introduction
The success of the greenhouses in the province of Almería (Spain) is founded on low-cost structures and a temperate climate that permit relatively high yields. However, at certain times of the year natural ventilation does not suffice to combat the high temperatures, and consequently 99% of growers whiten the greenhouse cover [1]. To do so they apply a mixture of water and micronized calcium carbonate ("Blanco de España"). Despite the importance of this technique in the climate control of Mediterranean greenhouses, few technical or scientific works have studied this topic. Transmissivity of greenhouse cover is one of the main parameters influencing the energy balance that determine inside temperature, that can vary along a crop season between 0.44 and 0.80, depending on whitening [2].

Measurement Equipment inside the Greenhouse
Temperature and relative air humidity were measured inside and outside the greenhouse by means of 13 CS215 sensors (Campbell Scientific Spain S.L., Barcelona, Spain) with accuracy for temperature of ±0.4 • C over 5-40 • C and for relative humidity of ±2% over 10%-90% RH. The sensors were protected from radiation inside a naturally aspirated box 41003-5 (Campbell Scientific Spain S.L., Barcelona, Spain). The data of humidity did not differ between the two sectors of the greenhouse, as experiments were conducted without crop. Solar radiation and PAR were measured inside and outside the greenhouse with three SP1110 pyranometers (Campbell Scientific Spain S.L.; sensitivity range of 350-1100 nm; accuracy of ±5%; Barcelona, Spain) and with three quantum sensors SKP215 (Skye Instruments Ltd, Llandrindod Wells, UK; sensitive to light between 400 nm and 700 nm wavelength; measurement range of 0-5 × 10 4 µmol m −2 s −1 ; accuracy ±5%). Net radiation was measured inside the greenhouse with two NR-Lite2 net radiometers (Kipp & Zonen B.V., Delft, The Netherlands; spectral response: 0 to 100 µm; measurement range of ±2000 W m −2 ; accuracy of ±5%). The data from all sensors were stored in five CR3000 microloggers (Campbell Scientific Spain S.L.) with a frequency of 1 Hz. Outside wind speed was measured at 10 m height with a Meteostation II (Hortimax S.L., Almería, Spain) incorporating a cup anemometer (measurement range of 0 to 40 m s −1 ; accuracy of ±5%) and a vane for wind direction (accuracy ±5 • ). The Meteostation II measurements were stored in an independent computer system once a minute. Figure 1 presents the location of the sensors in the experimental greenhouse. The greenhouse is fitted with three roof vents measuring 40×1 m 2 each (22.5×1 m 2 in sector 1 and 17.5×1 m 2 in sector 2), with the same orientation to the wind in each sector. The ventilation surface, i.e. surface area of the vent openings/greenhouse area, or SV/SA, was 11.25% for sector 1 and 10.81% for sector 2. The roof vents were fitted with insect-proof screens with a thread density of 10×20 threads cm -2 (36.0% porosity) and with the following geometric characteristics: thread density measured 9.6×20.3 threads cm -2 ; weft pore length 239.9 ± 18.5 µm; warp pore length 765.4 ± 27.1 µm; thread diameter 259.6 ± 19.1 µm; diameter of the inside pore circumference 241.9 ± 19.1 µm; mean pore area 0.182 ± 0.015 mm 2 ; corresponding with screen 3 discussed in López et al. [16].

Measurement Equipment inside the Greenhouse
Temperature and relative air humidity were measured inside and outside the greenhouse by means of 13 CS215 sensors (Campbell Scientific Spain S.L., Barcelona, Spain) with accuracy for temperature of ±0.4 °C over 5-40 °C and for relative humidity of ±2% over 10%-90% RH. The sensors were protected from radiation inside a naturally aspirated box 41003-5 (Campbell Scientific Spain S.L., Barcelona, Spain). The data of humidity did not differ between the two sectors of the greenhouse, as experiments were conducted without crop. Solar radiation and PAR were measured inside and outside the greenhouse with three SP1110 pyranometers (Campbell Scientific Spain S.L.; sensitivity range of 350-1100 nm; accuracy of ±5%; Barcelona, Spain) and with three quantum sensors SKP215 (Skye Instruments Ltd, Llandrindod Wells, UK; sensitive to light between 400 nm and 700 nm wavelength; measurement range of 0-5×10 4 µ mol m -2 s -1 ; accuracy ±5%). Net radiation was measured inside the greenhouse with two NR-Lite2 net radiometers (Kipp & Zonen B.V., Delft, The Netherlands; spectral response: 0 to 100 μm; measurement range of ±2000 W m -2 ; accuracy of ±5%). The data from all sensors were stored in five CR3000 microloggers (Campbell Scientific Spain S.L.) with a frequency of 1 Hz. Outside wind speed was measured at 10 m height with a Meteostation II (Hortimax S.L., Almería, Spain) incorporating a cup anemometer (measurement range of 0 to 40 m s -1 ; accuracy of ±5%) and a vane for wind direction (accuracy ±5°). The Meteostation II measurements were stored in an independent computer system once a minute. Figure 1 presents the location of the sensors in the experimental greenhouse.

Experimental Design
Data were taken in the months of July, August, September and October 2014 (Table 1). Sector 1 of the greenhouse without ASP was used as control. In sector 2, four ASP of different characteristics were applied: the traditional product "Blanco de España" (ASP BE ), and three products that incorporate adhesives, Flex (ASP F ), SuperFlex (ASP SF ) and Special Pepper (ASP SP ), all of which are commercial products (Indaloblanc S.L., Almería, Spain). Three concentrations of each product were tested [kg of product/l of water]: 0.125 kg L −1 (25/200), 0.25 kg L −1 (25/100), 0.50 kg L −1 (50/100); a fourth concentration of 0.08 kg L −1 (25/300) was tested for the product ASP BE . The manufacturers recommend a dose of 0.25 kg L −1 (25/100), though a work published after the experiments were carried out found that the mean dose applied in the province of Almería is 40/100 [1], an intermediate value between 25/100 and 50/100 tested here. Application of the product involves consuming approximately 0.1 l of mixture per m 2 of the greenhouse cover, which implies the following approximate quantities of product: 8.3, 12.5, 25.0 and 50.0 g m −2 (for the four concentrations tested). The traditional method of applying these products follows three steps: (i) the product is mixed according to the dose (kg/l) in a container of large capacity; (ii) one worker operates the hydraulic pump to apply the mixture through a hose; (iii) a second worker holding the hose (without a regulated nipple) walks over the greenhouse roof, applying the quantity of product that he considers suitable. There is no technical control of the real quantity of product applied to the greenhouse cover. Rather it all depends on the skill and knowhow of the worker.
According to the technical data supplied by the products' distributors, the traditional product "Blanco de España" ASP BE consists of over 99% calcium carbonate (CaCO 3 ), whereas the other three products (ASP F , ASP SF and ASP SP ) have about 97% calcium carbonate (CaCO 3 ). The traditional product ASP BE incorporates less than 1% of other elements, without adhesive substances. The others three products use unidentified adhesives and elements in proportions less than 3%. The manufacturer did not supply data on the precise compositions of additives, but indicate that ASP SF presents a higher resistance to weather elements like rain with an approximate durability of 3 to 5 months, whereas for ASP F and ASP SP it is about 3 months.
All the products are soluble in cold water, with a mean and maximum particle diameters of 2.8 µm and 33 µm, respectively. Each dose of each product was tested over 5 days: the product was applied early in the morning on the first day and data were taken on the second, third and fourth days (used as the three repetitions for statistical analyses); on the fifth day, the cover was cleaned and the following dose was applied, commencing a new test cycle. The different concentrations of each product were tested in three consecutive days, to allow a minimum of 720 data when analysed statistically values of transmissivity in the interval 12-16 h. These three days can be considered as different replications ( Figure 2) of each treatment (a product with a concentration). The climatic conditions outside the greenhouse on the days when data were recorded are presented in Table 1. For the first tests carried out with the product ASP BE (from July 19 to August 4, 2014) no inside temperature data are available due to a malfunction of the sensors, and so a second set of tests was carried out with this product in late September and early October. Tests were carried out at a time of year when crops are usually transplanted, i.e. when the cooling effect of crop evapotranspiration is low. As no crop was present in the greenhouse, the effect of the products was quantified in the most extreme conditions possible, simulating the situation when a crop is transplanted.
In spring 2015, an experiment was carried out to determine the resistance to rain of the traditional adhesive-free product ASP BE and of ASP F , which includes adhesive. On March 27 2015, these products were applied in sectors 1 and 2, respectively, of the experimental greenhouse at a concentration of 0.25 kg L −1 (25/100). From April to July the transmissivity of the greenhouse cover was determined in both sectors at different times in order to evaluate the effect of precipitation and time on the two products. Transmissivity was determined by measuring PAR outside the greenhouse and inside each sector. An HD2302.0 photo-radiometer (Delta OHM S.R.L., Padua, Italia) was used, equipped with an LP 471 PAR probe (sensitive to light between 400 nm and 700 nm wavelength; measurement range of 0.01 to 104 µmol m −2 s −1 ; accuracy <5%), to measure the photon flow in the PAR range.

Statistical Analysis
We have carried out regression analyses to compare the different variables for statistically significant relationships (p-value<0.05) using Statgraphics®Centurion 18 v 18.1 (Statgraphics Technologies, Inc., The Plains, VA, USA). The different transmissivity and inside air temperature in both compartments of the experimental greenhouse (with and without whitening) were examined using an analysis of variance (p-value < 0.05), comparing mean values using Fisher's least significant difference (LSD) approach. When there was a difference statistically significant between the standard deviations, the parametric analysis was not viable by means of an analysis of variance. For parameters with different variance, we made a non-parametric analysis with the Friedman test, appropriate when each row represents a block (the date of measurement), using box-and-whisker plot [17].

Results and Discussion
The aim of this work was to know about the effect on transmissivity of using products composed mainly of CaCO 3 at different concentration for whitening Mediterranean greenhouse roofs. The results obtained were statistically analyzed to verify the influence of different products on reduction of cover transmissivity and to compare different dose of each product. For a better understanding the results were divided and presented under four subsections. In Section 3.1 (Transmissivity of the cover with Agricultural Solar Protector without adhesives) we analyse differences in behavior of the use of traditional product for whitening Mediterranean greenhouse roofs ASP BE on two test periods, July-August and September-October. Transmissivity data for ASP BE are compared to the other products, ASP F , ASP SF and ASP SP incorporating adhesives in Section 3.2 (Transmissivity of the cover with Agricultural Solar Protector with adhesives). Section 3.3. (Effect of climatic conditions on the transmissivity of the cover with Agricultural Solar Protectors) show the effect of rain on the transmissivity of the whitened greenhouse cover, comparing the products ASP BE (without adhesive) and ASP F (with adhesive). Finally, a global analysis of the four products on the temperature inside the greenhouse is presented in Section 3.4 (Greenhouse temperature influenced by the cover with Agricultural Solar Protector).
3.1. Transmissivity of the Cover with Traditional Agricultural Solar Protector without Adhesives Figure 2a shows the level of solar radiation outside and inside the greenhouse for the experiment carried out with ASP BE at a concentration of 0.50 kg L −1 on August 2-4, 2014. Irrespective of the dose applied, the use of this product has been seen to reduce fluctuations in the intensity of solar ( Figure 2a) and PAR radiation inside the greenhouse, confirming the findings of Baille et al. [11]. This is beneficial for the crop, since the radiation levels received will remain stable throughout the day. The results of Baille et al. [11] showed that application of the product on the greenhouse cover reduced both the difference in temperature between crop leaves and the surrounding air and "the canopy-to-air vapour pressure deficit", while increasing the crop transpiration rate, which mitigated the previously observed fluctuations in this parameter the day after application. the greenhouse is presented in Subsection 3.4 (Greenhouse temperature influenced by the cover with Agricultural Solar Protector). Figure 2a shows the level of solar radiation outside and inside the greenhouse for the experiment carried out with ASPBE at a concentration of 0.50 kg l -1 on August 2-4, 2014. Irrespective of the dose applied, the use of this product has been seen to reduce fluctuations in the intensity of solar ( Figure  2a) and PAR radiation inside the greenhouse, confirming the findings of Baille et al. [11]. This is beneficial for the crop, since the radiation levels received will remain stable throughout the day. The results of Baille et al. [11] showed that application of the product on the greenhouse cover reduced both the difference in temperature between crop leaves and the surrounding air and "the canopy-toair vapour pressure deficit", while increasing the crop transpiration rate, which mitigated the previously observed fluctuations in this parameter the day after application. The values of transmissivity of the greenhouse cover fluctuate less when ASPBE is applied ( Figure  2b). The combined effects of reduction of fluctuation of the mean inside radiation and of the calculated transmissivity are likely due to the increase in the proportion of diffuse radiation when ASP is used [13], as diffuse radiation is less sensitive to the presence of obstacles including the greenhouse structure itself and any greenhouse equipment [11]. Indeed, Baille et al. [11] found less fluctuation in the values of mean inside radiation and of transmissivity of the cover with ASP than without it (mean values of 0.31 and 0.62, respectively, from 9:00 to 19:00). Figure 2b illustrates the sharp fall in the transmissivity of the greenhouse cover in sector 1 in periods when the withdrawn shading mesh affected the radiation sensors (between 11:00 and 11:30, and 17:00 and 17:30, approximately). Table 2 presents the values of transmissivity of the greenhouse cover to solar radiation, τs (Rs,i/Rs,o), and PAR, τPAR (RPAR,i/RPAR,o), for each dose of product applied. Transmissivity was analysded between 12:00 and 16:00 h, obtaining the average value at the interval of 4 hours around the time when the sun is shining vertically (local time 14:30 h). For the climatic conditions of the experiments, the transmissivity of the cover to total radiation and PAR can be obtained from a power regression equation based on the dose applied (Figures 3a and 3b).

Transmissivity of the Cover with Traditional Agricultural Solar Protector without Adhesives
The power regression equations presented in Figure 3 are only valid for concentrations of ASPBE between 0.08 and 0.50 kg l -1 ; for concentrations close to 0 these fits are not valid, as the values obtained would tend to infinity. Figures 3c and 3d show the fits to obtain the ratio τs,2/τs,1 as a function of the dose of ASPBE applied. The power regression equations presented in Figures 3c and 3d would be valid to estimate the effect on transmissivity of any type of greenhouse cover as a function of the concentration of ASPBE applied under similar climatic conditions to those of these experiments.
For all doses analysed, transmissivity of the whitened cover with ASPBE was statistically lower than transmissivity of the cover without whitening ( Table 2). We can also observe a reduction statistically significant of transmissivity when the dose of whitening increased (Table 2). Furthermore, transmissivity of the un-whitened cover show a statistically significant variation along The values of transmissivity of the greenhouse cover fluctuate less when ASP BE is applied ( Figure 2b). The combined effects of reduction of fluctuation of the mean inside radiation and of the calculated transmissivity are likely due to the increase in the proportion of diffuse radiation when ASP is used [13], as diffuse radiation is less sensitive to the presence of obstacles including the greenhouse structure itself and any greenhouse equipment [11]. Indeed, Baille et al. [11] found less fluctuation in the values of mean inside radiation and of transmissivity of the cover with ASP than without it (mean values of 0.31 and 0.62, respectively, from 9:00 to 19:00). Figure 2b illustrates the sharp fall in the transmissivity of the greenhouse cover in sector 1 in periods when the withdrawn shading mesh affected the radiation sensors (between 11:00 and 11:30, and 17:00 and 17:30, approximately). Table 2 presents the values of transmissivity of the greenhouse cover to solar radiation, τ s (R s,i /R s,o ), and PAR, τ PAR (R PAR,i /R PAR,o ), for each dose of product applied. Transmissivity was analysded between 12:00 and 16:00 h, obtaining the average value at the interval of 4 hours around the time when the sun is shining vertically (local time 14:30 h). For the climatic conditions of the experiments, the transmissivity of the cover to total radiation and PAR can be obtained from a power regression equation based on the dose applied (Figure 3a,b).
The power regression equations presented in Figure 3 are only valid for concentrations of ASP BE between 0.08 and 0.50 kg L −1 ; for concentrations close to 0 these fits are not valid, as the values obtained would tend to infinity. Figure 3c,d show the fits to obtain the ratio τ s,2 /τ s,1 as a function of the dose of ASP BE applied. The power regression equations presented in Figure 3c,d would be valid to estimate the effect on transmissivity of any type of greenhouse cover as a function of the concentration of ASP BE applied under similar climatic conditions to those of these experiments.
For all doses analysed, transmissivity of the whitened cover with ASP BE was statistically lower than transmissivity of the cover without whitening (Table 2). We can also observe a reduction statistically significant of transmissivity when the dose of whitening increased (Table 2). Furthermore, transmissivity of the un-whitened cover show a statistically significant variation along the year. At the end of July, transmissivity increased with the day of year (DOY), as we can observe in Table 2. the year. At the end of July, transmissivity increased with the day of year (DOY), as we can observe in Table 2.  The dose of 0.50 kg l -1 (50/100) could be recommended as the maximum concentration, respecting the limit of 50% reduction in transmissivity recommended by Baille et al. [11]. In the . Mean values of the ratio τ s,2 /τ s,1 (c) y τ PAR,2 /τ PAR,1 (d). Subscript: 1, sector 1 (without ASP BE ); 2, sector 2 (with ASP BE ).
, 10:00 to 19:00; ♦, 12:00 to 16:00. The dose of 0.50 kg L −1 (50/100) could be recommended as the maximum concentration, respecting the limit of 50% reduction in transmissivity recommended by Baille et al. [11]. In the present study, with this dose, the values of transmissivity of the cover were around 0.30, which is similar to the results obtained by the cited authors with a much lower concentration of the product, 0.08 kg L −1 (8/100). This discrepancy may be mainly due to: (i) the traditional method of applying the product, which is imprecise and unreliable, and as a result the amount of product that is finally applied to the cover will depend on the skill of the worker to a great extent; and (ii) the use of different types of greenhouse cover, namely a three-layer co-extrusion greenhouse film (PE-EVA-PE) of 0.2 mm in the present study and a glass roof in the case of Baille et al. [11].
Baille et al. [11] found that the ratio of net to solar irradiance measured above a well-developed crop of roses was not significantly different before and after whitening, with R n,i /R s,i values of 0.70 before application of the product and 0.73 afterwards. In the present study the greenhouse was empty, i.e. in similar conditions to a greenhouse with a recently transplanted crop, and in this case R n,i /R s,i was slightly lower with ASP BE (R n,1 /R s,1 ) than without it (R n,2 /R s,2 ), as Table 2 illustrates. ASP BE appears to reduce the amount of direct solar radiation entering the greenhouse, but it increases the proportion of diffuse radiation inside the greenhouse, which influences the lower receiver of the net radiation sensor. Were a crop present, maybe all this radiation would be recorded by the sensor, and no difference would be observed in the R n,i /R s,i ratios between sectors, as occurred in the above-mentioned study.
PAR is presented as µmolm −2 s −1 , and in order to compare it with the values of solar/total radiation obtained with a pyranometer (Wm −2 ) it can be multiplied by a factor of 4.57 (in µmolm −2 s −1 )/(W m −2 ) [18] or 4.6 [19], the former of which was chosen. One drawback of using the traditional product ASP BE is that it slightly reduced the proportion of PAR vs. total radiation (R PAR /R s ) inside the greenhouse (Table 2), which contrasts with the findings of Kittas et al. [9], who recorded a slight increase in this proportion. This type of product is 99% calcium carbonate (CaCO 3 ), but other compounds should be sought to act selectively depending on the wavelength of the radiation.
In short, the use of ASP BE , applied in the traditional fashion, led to a marked reduction in the transmissivity of the greenhouse cover. On the downside, it also appeared to reduce slightly the proportion of net radiation (though it should be noted that there was no crop in the greenhouse) and the proportion of PAR with respect to mean total radiation. The reduction in transmissivity has been seen to be statistically related to the dose applied, although the values of transmissivity of the greenhouse cover below a certain dose of product also depend on the prevalent conditions of solar radiation and elevation (see Section 3.3). It should also be remarked that the doses recommended by manufacturers are difficult to adhere to, since the product application method precludes verification of the final number of grams of product per m 2 of roof.
Due to technical problems, no inside temperature data were available for the experiments carried out in summer with ASP BE , and so it was decided to repeat the experiments in early October omitting the lowest concentration of the product, 0.08 kg L −1 . Soriano et al. [20] carried out a laboratory study on how the angle of incidence of solar radiation affected the transmissivity of several samples of glass, finding that the transmissivity was greatest when radiation was perpendicular to the glass. Transmissivity decreased with the angle of incidence, though the decrease was not marked until the angle reached 50-60 • with respect to the perpendicular; Mashonjowa et al. [15] obtained similar results. Given these findings, it might be expected that the effect of ASP BE on transmissivity of the greenhouse cover would differ between the experiments carried out in summer and autumn. Furthermore, in Mediterranean greenhouses ASP BE is usually only applied on the roof, not on the sides, and so the effect of the product might be expected to increase with solar elevation. In addition to the effect of the angle of incidence of the radiation, the level of diffuse radiation will affect the transmissivity values calculated, leading to differences depending on whether the sky is clear or overcast. Table 3 presents the transmissivity data obtained for the experiments carried out in September-October, and the values in sector 1 without ASP are higher than those in summer (Table 2). Moreover, the transmissivity of the cover for the same dose of product was significantly higher than that recorded in summer (Tables 2 and 3). The transmissivity to solar radiation of the cover without ASP BE was 8% (10:00-19:00) and 10% (12:00-16:00) greater in autumn than in summer. When ASP BE was applied, between 12:00 and 16:00 transmissivity to solar radiation was 18% ([ASP BE ] = 0.125 kg L −1 ), 19% ([ASP BE ] = 0.25 kg L −1 ) and 20% ([ASP BE ] = 0.125 kg L −1 ) greater in the autumn experiments (Tables 2 and 3). As in the first experiment, transmissivity of whitened cover decreased (with statistical significance) when the dose of ASP BE increased. A statistical difference was also observed between whitened cover with ASP BE and un-whitened cover (Table 3). However, in autumn transmittance of the cover without whitening reduced along the date, inversely to that observed in summer (Table 2). This difference was not statistically significant for transmissivity around the time of maximum outside solar radiation (12:00-16:00). This would appear to contradict the findings of other works [15,20], since solar elevation is greater in summer than in autumn, suggesting that transmissivity should also be greater. However, the mean angles of incidence of solar radiation on the greenhouse cover have been calculated (Section 3.3), and they are below 50-60 • , the margin in which reduction in transmissivity becomes more marked. On the other hand, in autumn the degree of solar elevation is lower and so a greater proportion of total radiation in the greenhouse will enter through the sides, which will affect transmissivity values calculated. Finally, in autumn there is a greater probability of overcast skies, conditions in which the proportion of diffuse radiation is greater, which will contribute to higher transmissivity values calculated in autumn than in summer.
This variation in transmissivity of the greenhouse cover, an in the effect of applying ASP BE , at different times of year (differences in solar elevation and the level of solar radiation) makes it difficult to compare the different ASP tested in the present work. It also makes it difficult for the manufacturers to suggest a recommended dose, since on the one hand the method of application would have to be standardised to ensure that the correct amount of product was applied to the greenhouse roof. On the other hand, the manufacturers' recommendations should take into account different climatic conditions (time of year, level of radiation, etc.).

Transmissivity of the Cover with Agricultural Solar Protector with Adhesives
This product was tested in the first weeks of August, with high levels of solar radiation and outside temperature. The experiments using a concentration of 0.25 kg L −1 (25/100) took place on overcast days, which affected the results: the transmissivity values obtained were higher than those for the concentration of 0.125 kg L −1 (25/200) ( Table 4). This may be due to the influence of the cloudy skies (a greater proportion of diffuse radiation) and to the traditional method of application, which makes it impossible to verify the exact quantity of product retained on the cover. Table 4. Mean values of transmissivity of the greenhouse cover in sector 1 (without ASP) and sector 2 (with ASP) for the products with adhesive. DOY, day of year; [ASP], concentration of product in kg L −1 ; τ s , transmissivity to solar radiation; τ PAR , transmissivity to PAR; subscript: 1, sector 1 (without ASP); 2, sector 2 (with ASP). Products: ASP F , flex; ASP SF , superflex; ASP SP , special pepper. As for the standard product ASP BE (Tables 2 and 3), the transmissivity of the cover with whitening product using adhesives in tis compositions (ASP F , ASP SF and ASP SP ) was statically lower that the un-whitened cover, for all the doses tested (Table 4). In general, the increase in the dose produced a reduction (statically significant) of the transmissivity (Table 4). However, differences statistically significant between the two lower doses (0.125 and 0.25 kg L −1 ) changed in function of the date and the weather conditions (cloudy and sunny days).

DOY
The reduction of transmissivity with respect to the cover without whitening was statistically greater (lower values of the ratio τ s , 2 /τ s,1 ) whit the higher concentration of 0.50 kg L −1 (50/100) of the products ASP F and ASP SF than whit the others two doses or than whit the others products ASP BE and ASP SP (Table 5). Table 5. Mean values of ratio τ s,2 /τ s,1 of transmissivity to solar radiation of the greenhouse cover in sector 2 τ s,2 (with ASP) and sector 1 τ s,1 (without ASP) for each concentration [ASP] in kg L −1 .

[ASP]
ASP With ASP F at the dose of 0.50 kg L −1 , a far greater decrease in transmissivity was observed (τ s , 2 /τ s,1 = 0.33 between 12:00 and 16:00) than with ASP BE (τ s,2 /τ s,1 = 0.48 in summer and 0.57 autumn) (Table 5). However, for the concentration of 0.125 kg L −1 the difference between ASP F (τ s,2 /τ s,1 = 0.73 between 12:00 and 16:00) and ASP BE (τ s,2 /τ s,1 = 0.66 in summer and 0.68 in autumn) was to the contrary, i.e. the decrease in transmissivity was greater with ASP BE . When comparing the results of these two products important factors should be taken into account: (i) the experiments were carried out on different days under similar but not identical climatic conditions; (ii) the traditional method of applying the products does not ensure that the same amount of product was applied per m 2 of greenhouse cover in each replication or test, even though the dose kg L −1 was the same. The results do indicate, however, that the presence of adhesives in the product (less than 3%) clearly increases the effect of the product on the transmissivity of the cover.
The difference between the products ASP F and ASP SF lies in the quantity of adhesive components they incorporate. Although the manufacturers declined to provide specific data, it is known that ASP SF has the greater adhesive content. These tests were carried out using concentrations of 0.125 kg L −1 and 0.50 kg L −1 on sunny days, and of 0.25 kg L −1 in partly cloudy conditions. For this product, the ratio τ s,2 /τ s,1 was similar at concentrations of 0.125 kg L −1 and 0.25 kg L −1 (Table 5), possibly due to the partially cloudy sky during the test for the latter concentration, which might explain the reduced effect of ASP SF on the transmissivity of the cover. In comparison with the results obtained for the traditional product ASP BE , there appear to be no statistical differences in the values of the ratio τ s,2 /τ s,1 (Tables 2 and 4). Between 12:00 and 16:00 the ratio τ s,2 /τ s,1 reaches similar values at a concentration of of 0.25 kg L −1 for ASP SF (0.58) and for ASP BE (0.56) in summer). Only at the highest concentration tested for ASP SF (0.50 kg L −1 ) was a greater difference observed in the ratio τ s,2 /τ s,1 (0.48 for ASP BE in summer and 0.57 for ASP BE in autumn and 0.41 for ASP SF ). As occurs with ASP F , with the product ASP SF (which in theory contains a greater quantity of adhesives) the results provide no clear indication that the adhesive clearly increases the effect of the product on the transmissivity of the greenhouse cover.
Of the products tested, ASP SP contains the largest amount of adhesives. For this product tests were carried out at concentrations of 0.125 kg L −1 and 0.25 kg L −1 on mainly sunny days, while at the concentration of 0.50 kg L −1 on the last two days of testing the sky was rather overcast. Comparison of the results obtained for ASP SP , tested in September, with those obtained in September/October for ASP BE does not highlight any great differences (Tables 3 and 4). The lowest ratio τ s,2 /τ s,1 at concentrations of 0.125 kg L −1 was obtained for the ASP SP , and the highest for the ASP F with a statistical significant difference. For the higher dose of 0.50 kg L −1 , we can observe the inverse effect, with the ASP SP producing the greatest value of the ratio τ s,2 /τ s,1 ., and the ASP F the lowest (Table 5). This result confirm the difficulty to predict the behavior of the different whitening products. At greater doses, the product with most adhesive component can allow a better adherence to the plastic cover, requiring less quantity of product to cover the roof, resulting in a greater transmissivity that product with a lower adherence. However, at low doses the effect of the different type of adhesives could affect to the greenhouse transmissivity.

Effect of Climatic Conditions on the Transmissivity of the Cover with Agricultural Solar Protectors
In short, Figure 4 illustrates that there were no notable differences between the capacity of the four products tested to reduce the transmissivity of the greenhouse cover (τ s,2 /τ s,1 ) at low concentrations (0.125 kg L −1 and 0.25 kg L −1 ). Bearing in mind that CaCO 3 constitutes 97-99% of the products, and that a maximum of 3% is composed of adhesives, we can state that the addition of this amount of adhesive does not noticeably alter the products' effect on the transmissivity of the greenhouse cover for low doses. Considering that all 4 products behave in a similar fashion at the same concentration, we can obtain a setting curve with which to estimate the ratio τ s,2 /τ s,1 as a function of the dose applied [kg L −1 ].
Statistical analyses have been carried out considering all the products (ASP BE , ASP F , ASP SF , ASP SP ) as the same ASP, in order to determine which of the parameters measured bear a significant influence on the values of transmissivity of the greenhouse cover with and without ASP.
The curved roof of the experimental greenhouse means that the angle of incidence of the radiation from the cover varies from practically 0 • to 90 • according to the position of the sun and the part of the roof considered. Considering a mean roof angle of 23.1 • (calculated as the mean value of 50 different points in the roof), the angle of incidence of solar radiation α c for the southern slope of the cover at the time of maximum solar elevation would vary between 20.3 • and 37.3 • (for the experiments from July 19-21 and from October 10-12, 2014, respectively), and between 23.2 • and 50.7 • for the northern slope of the cover for the same experimental periods. For an angle of incidence where 0 • corresponds to a perpendicular incidence of solar radiation and a value of 90 • corresponds to incidence parallel to the cover. These mean angles of incidence do not reach 50-60 • , beyond which Soriano et al. [20] found that transmissivity decreased significantly. where 0° corresponds to a perpendicular incidence of solar radiation and a value of 90° corresponds to incidence parallel to the cover. These mean angles of incidence do not reach 50-60°, beyond which Soriano et al. [20] found that transmissivity decreased significantly. The angle of incidence αc (14 h) obtained for the northern slope at 14 h, around the time of maximum solar elevation, increased along the period of tests avec the DOY, producing a variation of transmissivity τs,1 (Figure 5a). The influence of this angle in the cover transmissivity for the greenhouse without whitening (Figure 5b) can be represented by a statistically significate regression as (R 2 = 0.85; p-value < 0.0001): The angle of incidence α c (14 h) obtained for the northern slope at 14 h, around the time of maximum solar elevation, increased along the period of tests avec the DOY, producing a variation of transmissivity τ s,1 (Figure 5a). The influence of this angle in the cover transmissivity for the greenhouse without whitening (Figure 5b) can be represented by a statistically significate regression as (R 2 = 0.85; p-value < 0.0001): The angle of incidence αc (14 h) obtained for the northern slope at 14 h, around the time of maximum solar elevation, increased along the period of tests avec the DOY, producing a variation of transmissivity τs,1 (Figure 5a). The influence of this angle in the cover transmissivity for the greenhouse without whitening (Figure 5b) can be represented by a statistically significate regression as (R 2 = 0.85; p-value < 0.0001): Data analysis from all the tests carried out from July to October shows that there is a statistically significant correlation (p-value<0.01) between the transmissivity of the greenhouse cover in sector 1 (without ASP) (τs,1), the maximum daytime solar elevation (γmax) and solar radiation. Analysis of the period from 10:00 to 19:00 provides the following equation (R 2 = 0.54; p-value = 0.0068): Data analysis from all the tests carried out from July to October shows that there is a statistically significant correlation (p-value<0.01) between the transmissivity of the greenhouse cover in sector 1 (without ASP) (τ s,1 ), the maximum daytime solar elevation (γ max ) and solar radiation. Analysis of the period from 10:00 to 19:00 provides the following equation (R 2 = 0.54; p-value = 0.0068): Omitting solar elevation from (2), since the angles of incidence of solar radiation do not reach those beyond which Soriano et al. [20] found a sharp fall in transmissivity, provides the following equation with a lower p-value (R 2 = 0.53; p-value = 0.0014): The transmissivity of the cover without ASP increases as solar radiation decreases, which may be due to the proportion of diffuse radiation on the days in which the level of radiation is lower (overcast days and/or autumn days). Between 12:00 and 16:00 the following equation is obtained (R 2 = 0.36; p-value = 0.0137): Given the relationship between the transmissivity of the cover without ASP and the levels of outside radiation (cloud, diffuse radiation), it appears logical to suppose that the effect of applying any ASP product on the greenhouse cover will depend on, among other factors, solar radiation and the concentration or dose of the product [kg L −1 ]. Between 10:00 and 19:00 the following equation is obtained (R 2 = 0.58; p-value = 0.0038): This fit improves on the value of R 2 = 0.49 obtained when only the concentration of the product is considered (Figure 6a). The same fit, for the period between 12:00 and 16:00, would be (R 2 = 0.58; p-value = 0.0037): The effect that ASP has in reducing the transmissivity of the greenhouse cover (τ s,2 /τ s,1 ), decreases on days with low levels of outside radiation (days that are overcast and with a higher level of diffuse radiation) and increases with the dose of product applied. The values of R 2 obtained in the different fits are low due to other factors on which this value depends but which are not included in the analysis, such as the variability in the concentration of product applied to the covering as a result of the method of application. However, the p-values below 0.05 indicate a statistically significant relationship between the variables included in the statistical analysis.
As ASP BE contains no adhesive additives, on rainy days the greenhouse cover gets "washed". For the concentrations of 0.25 and 0.50 kg L −1 of the 6-day autumn experiments, the first three days were relatively clear, whereas the last three were cloudy with occasional precipitation and much lower levels of outside radiation ( Table 1). As the days passed, the effect of the high atmospheric humidity, the morning dew and the showers led to a sharp fall in the effect of the product, with a concomitant increase in the transmissivity of the greenhouse cover. Comparison of the first three sunny days with the last three cloudy ones (Figure 6a) shows increases in transmissivity to solar radiation between 12:00 and 16:00 of 27% ([ASP BE ] = 0.25 kg L −1 ) and 30% ([ASP BE ] = 0.50 kg L −1 ), while the increases in transmissivity to PAR for the same concentrations of product were 24% and 23%, respectively. However, these increases can be attributed in part to the increase in diffuse radiation. Figure 6b illustrates that during the three cloudy days with showers the greenhouse cover is not completely washed, since the values of the ratio τ s,2 /τ s,1 do not reach 1.  Figure 7 presents the values of transmissivity (to PAR radiation) of the greenhouse cover in sector 1 with ASPBE and in sector 2 with ASPF in spring 2015 for an initial concentration of 0.25 kg l -1 (25/100), together with the values of precipitation recorded at the Almería airport weather station (Almería, Spain). Transmissivity for ASPF was 0.41 six days after application, increasing to 0.48 after 24 days, several of which were rainy. In the following months, the transmissivity remained at around the same value. For ASPBE, on the other hand, transmissivity was 0.42 after six days, increasing to 0.73 after 24 days in the same meteorological conditions. This valued decreased slightly, possibly due to the accumulation of dirt on the greenhouse cover and the varied climatic conditions. The heavy rainfall at the start of the experiment washed off the ASPBE almost completely. At the conclusion of the experiment the transmissivity value for sector 1 was 0.61, similar to those in sector 1 without ASP recorded during the experiments in 2014 (Tables 2 and 3).  Figure 7 presents the values of transmissivity (to PAR radiation) of the greenhouse cover in sector 1 with ASP BE and in sector 2 with ASP F in spring 2015 for an initial concentration of 0.25 kg L −1 (25/100), together with the values of precipitation recorded at the Almería airport weather station (Almería, Spain). Transmissivity for ASP F was 0.41 six days after application, increasing to 0.48 after 24 days, several of which were rainy. In the following months, the transmissivity remained at around the same value. For ASP BE , on the other hand, transmissivity was 0.42 after six days, increasing to 0.73 after 24 days in the same meteorological conditions. This valued decreased slightly, possibly due to the accumulation of dirt on the greenhouse cover and the varied climatic conditions. The heavy rainfall at the start of the experiment washed off the ASP BE almost completely. At the conclusion of the experiment the transmissivity value for sector 1 was 0.61, similar to those in sector 1 without ASP recorded during the experiments in 2014 (Tables 2 and 3). sector 1 with ASPBE and in sector 2 with ASPF in spring 2015 for an initial concentration of 0.25 kg l -1 (25/100), together with the values of precipitation recorded at the Almería airport weather station (Almería, Spain). Transmissivity for ASPF was 0.41 six days after application, increasing to 0.48 after 24 days, several of which were rainy. In the following months, the transmissivity remained at around the same value. For ASPBE, on the other hand, transmissivity was 0.42 after six days, increasing to 0.73 after 24 days in the same meteorological conditions. This valued decreased slightly, possibly due to the accumulation of dirt on the greenhouse cover and the varied climatic conditions. The heavy rainfall at the start of the experiment washed off the ASPBE almost completely. At the conclusion of the experiment the transmissivity value for sector 1 was 0.61, similar to those in sector 1 without ASP recorded during the experiments in 2014 (Tables 2 and 3). . Transmissivity (to PAR radiation) of the greenhouse cover in sector 1 with ASPBE (×) and in sector 2 with ASPF (◊). Initial concentration of the product applied 25/100 (0.25 kg l -1 ). Accumulated rainfall according to data from the Almería airport weather station (-•-). . Transmissivity (to PAR radiation) of the greenhouse cover in sector 1 with ASP BE (×) and in sector 2 with ASP F (♦). Initial concentration of the product applied 25/100 (0.25 kg L −1 ). Accumulated rainfall according to data from the Almería airport weather station (-•-).

Greenhouse Temperature is Influenced by the Cover with Agricultural Solar Protector
The use of whitening produced a statistically significant reduction of the temperature inside the greenhouse (Table 6) when outside mean temperature was greater than 28.5 ºC (with the exception of the ASP SP at 0.50 kg L −1 ). Whitening is traditionally used in Almeria at the end of summer and at the end of the winter, when new crops are transplanted in the greenhouse. When outside temperature begin to decrease, growers remove the whitening from cover washing it with water. When outside temperature was lower than 28.5 ºC, the whitening did not produce a significant effect in inside temperature (Table 6) whereas transmissivity to PAR radiation of the whitened cover was reduced (Tables 2-4). Figure 8a illustrates that as the concentration of product applied increases there is a slight increase in temperature difference between the greenhouse sectors, although the trend is not clear due to the intrinsic variability as a result of the application method. However, it is clear that as the ratio τ s,2 /τ s,1 decreases, the temperature difference between sectors increases (Figure 8b). The use of whitening produced a statistically significant reduction of the temperature inside the greenhouse (Table 6) when outside mean temperature was greater than 28.5 ºC (with the exception of the ASPSP at 0.50 kg l -1 ). Whitening is traditionally used in Almeria at the end of summer and at the end of the winter, when new crops are transplanted in the greenhouse. When outside temperature begin to decrease, growers remove the whitening from cover washing it with water. When outside temperature was lower than 28.5 ºC, the whitening did not produce a significant effect in inside temperature (Table 6) whereas transmissivity to PAR radiation of the whitened cover was reduced (Tables 2-4). Figure 8a illustrates that as the concentration of product applied increases there is a slight increase in temperature difference between the greenhouse sectors, although the trend is not clear due to the intrinsic variability as a result of the application method. However, it is clear that as the ratio τs,2/τs,1 decreases, the temperature difference between sectors increases (Figure 8b). A global analysis has been carried out considering all the products as one. It has been determined that there is a statistically significant relationship (p-value<0.05) between the temperature difference between the two sectors of the greenhouse on the one hand and outside solar radiation and the ratio τs,2/τs,1 on the other; the temperature difference increases with the former and decreases with the A global analysis has been carried out considering all the products as one. It has been determined that there is a statistically significant relationship (p-value<0.05) between the temperature difference between the two sectors of the greenhouse on the one hand and outside solar radiation and the ratio τ s,2 /τ s,1 on the other; the temperature difference increases with the former and decreases with the latter. For the period 12:00-16:00 the following equation is obtained (R 2 = 0.09; p-value<0.0001): The values of temperature difference between sector 1 (without ASP) and sector 2 (with ASP) are well below the 4.4 • C reported by Baille et al. [11], whose experiments were in a greenhouse with a crop and the transpiration rate was higher in the sector with ASP. However, the maximum temperature differences recorded between the two sectors at the hottest time of day, for the concentration of product recommended by the manufacturer in this province (25/100) was 4.2 • C for ASP BE , 3.9 • C for ASP F , 5.0 • C for ASP SF and 2.0 • C for ASP SP . Although no great differences were observed from 12:00 to 16:00 in the mean temperature values between sectors (Table 6), with ASP the maximum temperature decreases considerably inside the greenhouse without crop. This finding may prove of interest, as the conditions are similar to those of a recently transplanted crop, when plants are more sensitive to temperature extremes.
Application of ASP does affect the heterogeneity of temperature inside the greenhouse. The difference between the mean temperatures recorded by the "warmest" and "coldest" sensors (∆T max,1 and ∆T max,2 ) has been estimated for three days from 12:00 to 16:00, and it was always higher in sector 1 without ASP than in sector 2 with ASP for all four products tested ( Table 6). The ratio σ ∆T i,o /∆T i,o proposed by Kittas et al. [21] has also been estimated; the greater the value of this ratio, the greater the temperature heterogeneity inside the greenhouse. Table 6 shows that this ratio decreases in the sector where ASP is applied in 10 of the 12 experiments. Table 6. Mean outside air temperature T 0 [ • C]; mean temperatures inside sector 1 (without ASP) T 1 and sector 2 (with ASP) T 2 [ • C]; ]; maximum difference between the mean temperatures inside sectors 1 and 2 ∆T 1,2 max [ • C]; temperature difference between sector 2 (with ASP) and outside ∆T 2,o [ • C]; maximum difference between the mean temperatures recorded by the different sensors in sectors 1 and 2, ∆T max,1 and ∆T max,2 [ • C]; ratio for the heterogeneity of temperature distribution inside the greenhouse σ ∆T i,o /∆T i,o ,. Values for the time period 12:00-16:00. [ASP] T 0 T 1

Conclusions
As final conclusions, four agricultural solar protectors (ASPs) have been tested: "Blanco de España" (ASP BE ), the product traditionally used in the province of Almería, and three other commercial products that incorporate adhesives. The presence of the adhesive does not appear to influence the effect of the different products on the temperature inside the greenhouse, as all four products behave in a similar fashion at the same concentrations. The present findings support the maximum dose of product recommended by other authors: 0.50 kgL −1 (50/100), above which the transmissivity of the greenhouse cover produces a statistically significant decrease of over 50%. The effect of ASP on transmissivity of the greenhouse cover depends mainly on the dose applied, but also on the climatic conditions (solar radiation, cloud cover, etc.) and the time of year (solar elevation). This makes it difficult to recommend a single dose of product to growers. Different doses should be recommended depending on the time of year and the desired reduction in transmissivity. One of the products containing adhesives (ASP F ) has been shown to remain on the greenhouse cover after periods of heavy rain, while the non-adhesive product traditionally used (ASP BE ) is washed away. The method of application of ASP should be standardised in order to establish a means of applying a given concentration of product in gm −2 of cover. The traditional method of application establishes a dose (in kgL −1 ), but the amount of product that finally remains on the cover is impossible to determine as it is applied manually. Funding: This work has been funded by the Spanish Ministry of Economy and Competitiveness and the European Regional Development Fund (ERDF) by means of research grant AGL2015-68050-R.