**Appendix D**

**Table A4.** Annual production costs for the 9 crops under full sunlight. Values adapted from [75,76].


(a) Considering a required harvest labor of 25 *labor units*/ha and a *labor unit* regulated price of 45 EUR/labor unit [77]; *labor unit* represents the work done by one worker in one day. (b) Considering a required harvest labor of 23 *labor units*/ha and a *labor unit* regulated price of 45 EUR/ labor unit. (c) Considering a required harvest labor of 20 *labor units*/ha and a *labor unit* regulated price of 45 EUR/ labor unit. (d) Since faba bean and forage maize are cultivated in sequential cropping, they are each ascribed with half the cost of land-property-tax and half the fixed component of irrigation cost.

### **Appendix E**

Uncertainty factors applied on literature references to obtain percentage yield variation under shading for each crop (high crop yield penalty).

### *Appendix E.1. Canola*

We took -20% yield value straightforwardly from Figure 1 [44], as the average of their experiments shading at flowering (2011) and shading at pod filling (2011). We disregarded yield value of shading at flowering (2010) because that year was extremely dry and we are analyzing irrigation farming.

### *Appendix E.2. Carrot*

We pay attention to the marketable yield column of Table 2 [45]. From the different shading nets listed there, we select white polyethylene (PE) as the closer to our APV-shed configuration (at first glance, one could think that due to monofaciality of our PV modules, black PE would be more similar, but the shading intensity decrease due to modules height plays a role). With respect to no-shade, the variation is of 9.6%, which we rounded to 10%.

### *Appendix E.3. Maize*

We paid attention to Table 2 [46], biomass of corn stover and Table 3, grain yield. To be conservative, for both tables we focused on the higher PV GCR. We took data from both tables because forage maize crop harvest is a mix of chopped stover, ears and grains. From Table 2 [46], we obtained −3% under shading, whereas from Table 3 [46], we obtained −3.6%. The average of both values is 3.3%. To be conservative, we applied an uncertainty factor of 2, which multiplied by 3.3 equals 6.6%, and finally, rounded to 7%. The reason underlying the uncertainty factor of 2 is that Sekiyama and Nagashima [46] experiments were conducted at latitude 35 ◦N, while our latitude is higher (37 ◦N).

### *Appendix E.4. Faba Bean*

Table 2 [47] shows higher yield under shading than under full sunlight. To be conservative, we assume zero variation with respect to full sunlight.

### *Appendix E.5. Melon*

In Figure A3 [48], we took marketable yields corresponding to control (full sunlight) and aluminet shading net, which to our understanding is more similar to our APV shading than the other two types of shading net categorized in Figure 3 [48]. The difference between them is approximately 8.4 t/ha, which divided by the control equals 16.5%, which we rounded to 17%.

### *Appendix E.6. Onion*

From Table 7 [49], we calculated an average yield variation of 2.3% between full sunlight and shading conditions. Then, we applied an uncertainty factor of 2.5, for a three-fold reason: First, the latitude of Khan et al. [49] experiment was tropical, unlike ours; second, their shade was not generated by an inert artificial screen, but by a plant canopy which entails a competition not only for sunlight, but also for soil nutrients. In third place, the onion yield (t/ha) reported [49] are much lower than the common in our area, most probably because spacing between plants was rather large. Finally, we calculated 2.3% · 2.5 = 5.8%, which we rounded to 6%.

### *Appendix E.7. Pea*

In Table 2 [50], we took the yield values of *lighter shading*—one layer of screen—which, to our understanding, reflects better the light conditions under our APV shed and compare them to the no-shade conditions. For the year 1973, we obtained 19.4%, whereas for 1974 we obtained 10.5%. The average of both is approximately 14.9%, which we rounded to 15%.
