Assessment of different LED lighting systems for indoor living walls

Assessment of different LED lighting systems for indoor living walls 1 Maria P. Kaltsidia, Rafael Fernández-Cañerob, Luis Pérez-Urrestarazua,* 2 a Urban greening and Biosystems Engineering research group. Area of Agro-Forestry 3 Engineering. Universidad de Sevilla. ETSIA Ctra. Utrera km.1, 41013. Seville. Spain. 4 E-mail addresses: [email protected], [email protected] 5 b Urban greening and Biosystems Engineering research group. Department of Agro- 6 Forestry Sciences. Universidad de Sevilla. ETSIA Ctra. Utrera km.1, 41013. Seville. 7 Spain. E-mail address: raf[email protected] 8 *Corresponding author: Telephone number: +34 954486480; E-mail address: 9 [email protected] 10 11 Abstract 12 Building-integrated vegetation systems, such as living walls (LW), are becoming 13 common tools for improving the sustainability of cities as well as an aesthetic resource. 14 When used indoors, LW usually require a lighting system to ensure both an adequate 15 plant development and a correct appearance. In this study, six commercial LED lighting 16 systems are tested in order to assess their suitability for the proper performance of LW. 17 The LW monitored were composed of two plant species (Soleirolia soleirolii and 18 Spathiphyllum wallisii) frequently used in indoor LW. All the lamps tested (Aster and 19 Dahlia of Ignia Green, Logar CMH, CLH and Forum of Lledó) proved to be apt for 20 their use to light LW (except for the case of CF-UT01 of Panda Grow), as they showed 21 a favourable performance in terms of plant development, with few differences between 22 them in biomass production and green cover. The tested Aster (Ignia Green) and Logar 23 CMH (Lledó) lamp models were not efficient for long distances between the vegetation 24 and the light source. Despite these results, as illumination is one of the factors that 25 determines the indoor ambience, aesthetics and viewers’ preferences were also studied. 26 According to the observers' perception, the Dahlia model (Ignia Green) was preferred 27 by 54.4 % of the respondents, while the rest of the lamps were preferred less. 28 Keywords: vertical greening system, ornamental lighting, plant development, urban 29 greening, viewer’s perception 30 Nomenclature 31 Symbol Units ADW: Aerial Dry Weight g plant-1 AFW: Aerial Fresh Weight g plant-1 CRI: Colour Rendering Index -- ET: Evapotranspiration l d-1 LED: Light-Emitting Diodes -- LW: Living Wall (s) -- PAR: Photosynthetically Active Radiation -- PPFD: Photosynthetic Photon Flux Density mol m-2 s-1 RDW: Root Dry Weight g plant-1 RFW: Root Fresh Weight g plant-1 RH: Relative Humidity % Soleirolia: Soleirolia soleirolii -- Spathiphyllum: Spathiphyllum wallisii -- SPAD: relative measure of chlorophyll content -- T: Temperature ºC TDW: Total Dry Weight g plant-1 TFW: Total (whole-plant) Fresh Weight g plant-1 LA: mean Leaf Area cm2leave-1 32 33 34 Introduction 35 Nowadays, the inclusion of vegetation in the built environment in the form of green 36 roofs and vertical greening systems is spreading. They are usually located outdoors, but 37 in the case of living walls (LW), indoor installations are becoming frequent, given the 38 multiple benefits which they offer, improving indoor air quality (particles and VOC 39 retention), environmental conditions (temperature and humidity levels), acoustics and 40 wellbeing (Gunawardena and Steemers, 2019; Moya et al., 2019). However, when 41 plants are grown inside a building, one of the main constraints is the light that they 42 receive. The available natural light in indoor environments is frequently not sufficient, 43 thus auxiliary artificial lighting is often required for adequate plant growth and 44 development (Tan et al., 2017). 45 Selecting the proper lighting system for indoor plant growth is a demanding process that 46 requires an accurate prior study. It should ensure certain characteristics in terms of 47 intensity (the amount of light received by the vegetation) and quality (the spectral 48 composition of the light source) (GOTO, 2003). In the case of LW, regulating the 49 intensity is even more complicated, given that the lamps are usually located in the 50 ceiling, so the lighting is not uniform over the entire vertical surface. In terms of 51 quality, not only obtaining an effective spectral range is essential but also ensuring that 52 the LW have a proper appearance (Egea et al., 2014). 53 Artificial lighting technologies have been used in crop production for many years, with 54 incandescent, fluorescent or high-intensity discharge lamps having been those most 55 employed. However, the advance of solid-state lighting using light-emitting diodes 56 (LEDs), with a great technical development in the last years and an important cost 57 reduction, has displaced the other types of lamps. LEDs show several advantages such 58 as a much longer lifespan and producing a high luminous flux with a low radiant heat 59 output (Morrow, 2008; Yeh and Chung, 2009). This makes them more competitive in 60 energy efficiency and economic terms (Singh et al., 2015). 61 LEDs also have the ability to emit in a controlled spectral composition (Olle and 62 Viršile, 2013), which is an advantage when growing plants. Given that LEDs emit in a 63 very narrow spectrum (20-40 nm), the specific peak absorption bands of chlorophyll can 64 be targeted. This improves the use of energy as most emitted light can be used for 65 photosynthesis. Precisely, that is the basis of commercial LED grow lights, which 66 mainly emit in the blue and red regions. Nevertheless, they give plants an unnatural 67 appearance due to their colour (red/blue), so they are not so apt for aesthetical purposes, 68 including LW lighting. In addition, some studies indicate that a better plant growth is 69 achieved when using a broader spectrum with additional wavelengths (Kim et al., 70 2006). This makes white light more adequate. In order to obtain white LEDs, blue LEDs 71 are usually coated with phosphor. Though this makes them less efficient than the single- 72 wave-peak LEDs, the visualisation of plants greatly improves (Massa et al., 2008). 73 In artificial lighting, the term white light refers to light formed by a mixture of colours. 74 However, not all whites are the same, since they depend on the colours that compose 75 them. In this sense, a white with a higher proportion of red will favour a "warmer" 76 lighting and a white with a higher proportion of blues will give a "cooler" appearance. 77 Colour temperature is used to classify the different types of white light and to facilitate 78 comparison with "full spectrum" sunlight (Morrow, 2008). This concept refers to the 79 type of light that a black body radiates when heated to a specific temperature, so that the 80 higher the colour temperature, the colder the light source. For instance, at 2,000-3,000 81 K, the colour of the light will look white yellow; at 4,000 K, neutral white, and at 5,000- 82 7,000 K, cold white. Shaw (2018) suggested that colour temperature has an effect on the 83 growth of hydroponic lettuce seedlings, as plants under 6,000 K lights grew more than 84 under 3,000 K. However, even when two light sources have the same colour 85 temperature, the surfaces can be seen in different colours, given that two lights that 86 appear to produce the same white may be the result of different wavelength mixes. For 87 this reason, the concept of colour rendering is used to elucidate the similarity between 88 the natural colour of an object (that is, in daylight conditions) and its colour under 89 artificial lighting. Based on this concept, the colour rendering index (CRI) classifies 90 light sources according to their colour rendering properties: the higher the CRI, the 91 closer it is to natural colour. 92 LED lighting in horticultural production has been widely addressed (Islam et al., 2012; 93 Massa et al., 2008; Morrow, 2008; Olle and Viršile, 2013; Samuoliene et al., 2013; 94 Singh et al., 2015), but it has not been studied when it is used with an ornamental 95 purpose (as is the case of LW illumination). Only Tan et al. (2017) and Egea et al., 96 (2014) have addressed this topic. The former quantified the impact of growth light 97 provision on indoor greenery and the light compensation point of two ornamental 98 species. The latter analysed different artificial lighting systems for LW, but in their 99 study LEDs were not contemplated. 100 The main objective of the current study was to assess the adaptation of six different 101 commercial LED lamps (five of which were not specifically designed for plant growth) 102 for the lighting of indoor LW. Both the performance and correct development of the 103 vegetation under each lamp and its appearance were taken into consideration. The study 104 was completed with an analysis of public preferences. 105 106 Materials and methods 107 Experimental setup and tests performed 108 The study was performed at the Urban Greening Laboratory of the School of 109 Agricultural Engineering of the University of Seville (Seville, Spain), with no natural 110 light. Six different types of lamps were tested in this study and two experiments were 111 carried out. Five of the lamps were conventional white LED lamps (4000 K) while one 112 (C) was a commercial Grow-LED lamp specially designed for plant cultivation. Table 1 113 presents the main characteristics of each lamp and Figure 1 shows the relative emission 114 intensity spectrum, when available. The first experiment involved lamps A to C and was 115 conducted over the period mid-May to end-July 2018 (68 days). During this period, the 116 daily mean room temperature and relative humidity were 24.9±0.7ºC and 68± 5 %, 117 respectively. Lamps D, E and F were tested in a second experiment from mid-February 118 to end-April (70 days). In this case, the daily room temperature was 22.4 ± 0.6 ºC and 119 the relative humidity was 56 ± 7 %. 120 121 Table 1. LED lamps used in the study and their characteristics. The different letters (A, B, C, D, E, F) refer to different lamp type treatments and 122 the different numbers (1 or 2) refer to module closer (1) or farther (2) to the light source. 123 Lamp Model Projector Curves Dimensions Manufacturer LW module Power (W) Flux (lm) CRI Beam angle (º) Colour temperature (K) Type of light Aster Ignia Green (Girona, Spain) A1 40 2.575 >90? 36º 3.700 White Logar CMH Superflood Lledó (Madrid, Spain) B1 35 2.650 >90 31º 4.000 White CF-UT01 NA Panda Grow (Shenzhen, China) C1 100 5.000 NA 120º NA Blue/red Dahlia Ignia Green (Girona, Spain) D1-D2 110 7.950 >90? 97º 3.700 White Logar CLH Superflood Lledó (Madrid, Spain) E1-E2 48 3.300 >90 *41º 4.000 White Foru m Lledó (Madrid, Spain) F1-F2 83 7.350 >80 68º 4.000 White * Due to its small beam angle, two identical lamps of this model were placed at the same spot with different angles pointing at the centre of each 124 of the two modules. NA: Not available 125 126 Figure 1. Relative emission intensity (%) spectrum for a) Lledo, Forum lamp b) Lledo, 127 CMH, CLH lamps and c) Ignia Green, Aster and Dahlia. (Graphs courtesy Lledo and 128 Ignia green, images modified) 129 130 In the first test, only one lamp per LW module was placed at a distance of 1 m from the 131 wall where the LW modules were installed, pointing at the centre of each LW module 132 (100º) (Figure 2). In the second test, as the light intensity provided by the lamps was 133 adequate at a higher distance, a second LW module was added right below the existing 134 ones to test the capacity of these lamps to light a higher LW up. D and F lamps were 135 pointing between the two LW modules at a distance of 1 m from the wall and with a 136 120º inclination angle. E1 was pointing at the centre of the upper LW module (100º), 137 0.80 m apart from the LW module surface. E2 was angled to face the centre of the lower 138 module (140º), at a distance of 1.50 m from it. The 3 phase electrified rails of the lamps 139 B, E, F and the lamps A, C, D were attached in a metallic base 0.50 m from the ceiling. 140 Thus, all lamps were placed just in front of the middle of the upper LW module. The 141 inclination angles were determined by doing a simulation using the professional 142 DIALux evo lighting design software (DIAL, Lüdenscheid, Germany) for professional 143 light planning, to optimise their illumination. The different LW modules were separated 144 from each other using opaque black plastic curtains and a constant photoperiod of 14 145 hours per day was provided during both trials. 146 147 148 32.3 mol m-2 s-1 for D2, E2 and F2, respectively). Module C received very poor values 280 (7 mol m-2 s-1 in average). 281 Table 2. Mean Photosynthetic Photon Flux Density values (mol m-2 s-1) for all lamps 282 (A to F) in the upper (1) and lower (2) modules at three different heights (Up, Mid, 283 Down) within each module. 284 A B C D E F 1 Up 28.9 13.3 7.3 62.6 58.6 58.0 Mid 55.2 43.7 7.7 78.7 109.9 88.8 Down 23.0 19.8 5.9 74.2 78.9 34.0 2 Up 1.9 2.9 2.2 38.3 73.8 44.1 Mid 0.6 0.7 1.2 26.0 52.7 32.3 Down 0.4 0.3 0.8 19.2 19.9 20.4 285 Both the illuminance received and the PPFD depend, among other factors, on the 286 distance to the light source. Figure 4 shows the different values of these two factors 287 according to the distance from the LW to the different lamps tested. In the first metre, 288 the values severely decrease, while this decrease is observed to be less intense as the 289 distance increases.. 290 291 Figure 4. Illuminance (left) and Photosynthetic Photon Flux Density (right) at different 292 distances from the light source for each lamp. 293 Figure 5 represents the relation between the measured values of illuminance vs the 294 PPFD for the different lamps, hence obtaining the conversion equations between both 295 factors, which are distinct for each lamp. Lamp A exhibited a good relation, comparing 296 to the rest lamps, where then minimum illuminance of 420 lx corresponds to 5.8 mol 297 m-2 s-1 and a 1048 lx corresponds to 13.2 mol m-2 s-1. Lamp C presented the most 298 elevated PPFD value (22.2 mol m-2 s-1) in 1136 lx, though, to be achieved, a short 299 distance of 0.5 m is required (Figure 4). Lamp D had the highest PPFD value (94.8 300 mol m-2 s-1) when illuminance reaches 7204 lx. Lamp E showed a good relation 301 between PPFD and illuminance. 302 303 Figure 5. Relation between illuminance (lx) and Photosynthetic Photon Flux Density. 304 Finally, an illuminance simulation of both tests was performed in DIALux evo (Figure 305 6), showing a very similar pattern of lux levels to that depicted in Figure 3. The Pearson 306 correlation coefficients results (0.95, 0.98, 0.92, 0.89, 0.95, 0.88, 0.79, 0.77 and 0.98 for 307 modules A, B, C, D1, D2, E1, E2, F1 and F2, respectively) exhibited that the correlation 308 between the simulations (Figure 6) and the actual measured values (Figure 3) was high, 309 being slightly inferior for the lower modules.). 310 311 312 313 Figure 6. Simulation of illuminance levels for Test 1 (up) and Test 2 (down) using 314 DIALux evo software and Pearson correlation coefficients (r) between the simulation 315 and the measured illuminance values (lx). 316 Temperature and water consumption 317 The evolution of the temperature (T) close to each module is depicted, for both tests, in 318 Figure 7. Variations in T were within 5ºC even between tests. The average T of test 1 319 and test 2 differed by 3ºC. During test 2, a difference of 1ºC on average was observed 320 between the upper and lower modules except for D1 and D2 which did not differ. RH 321 ranged between 50 % and 70 %. The average values were higher for the first test. In the 322 second test, the RH was lower in the upper modules compared to the lower ones. 323 324 Figure 7. Evolution of the mean daily temperature near each living wall module during 325 both tests 326 The average daily water consumption ranged between 1 and 1.5 L m-2 d-1 (Figure 8), 327 resulting in more water consumed in module D2 (50.4 L) compared to B (35.2 L). 328 Statistically significant differences (F = 2.834198; P-value = 0.00617977) in the average 329 daily water consumption values were observed. 330 331 332 Figure 8. Water consumption in the different living wall modules: (a) Cumulative 333 evolution during the tests (L) and (b) mean daily values (L m-2 d-1). Different letters at 334 the bottom of the bars indicate significant differences following Duncan’s multiple 335 range test (P< 0.05) 336 337 Vegetation performance 338 Plant biomass produced in each of the LW modules was calculated at the end of the 339 tests. Both fresh and dry weights per plant were measured for the aerial and root parts. 340 Total leaf area (TLA) was also obtained only for Spathiphyllum. 341 In the case of Spathiphyllum (Table 3), differences in fresh weight were more 342 significant in the aerial part,while significant differences were exhibited only in the root 343 system of module A. Module A had the higher fresh weights, while E2 presented the 344 lowest. No differences were observed in fresh weight within modules lighted by lamps 345 D, E and F. However, looking into their dry weights, the only significant difference 346 occurred in the aerial part between E1 and D2. Even though no significant differences 347 between upper and lower modules were observed, dry biomass in lower modules was 348 82.2 % of the average observed in the upper ones. Plants in module D2 had the lowest 349 dry biomass, being 57 % of the obtained in module A, which produced the highest value 350 (significantly different to the rest, excepting modules B and E1). There were no 351 significant differences in leaf area. 352 353 Table 3. Weights and leaf area of Spathiphyllum plants. TFW: total fresh weight; RFW: 354 root fresh weight; AFW: aerial fresh weight; TDW: total dry weight; RDW: root dry 355 weight; ADW: aerial dry weight; LA: mean leaf area. 356 Measured variables LW module P-value A B C D1 E1 F1 D2 E2 F2 TFW (g plant-1) 170.5a 134.9b 112.8bc 94.4cd 102.0cd 97.8cd 88.7cd 82.5d 93.5cd 0.0000 RFW (g plant-1) 43.50a 30.04b 21.30b 22.53b 26.82b 26.94b 23.18b 26.08b 24.43b 0.0005 AFW (g plant-1) 126.9a 104.82b 91.47bc 71.84d 75.22cd 70.91d 65.53d 56.41d 69.05d 0.0000 TDW (g plant-1) 14.89a 12.29ab 10.71bc 11.29bc 12.53ab 10.94bc 8.50c 9.95bc 10.12bc 0.0150 RDW (g plant-1) 3.94a 2.65abc 1.41c 2.60abc 3.55ab 2.92abc 2.05abc 3.06abc 2.39abc 0.0478 ADW (g plant-1) 10.94a 9.64ab 9.31ab 8.70bc 8.98abc 8.02bcd 6.44d 6.89cd 7.73abc 0.0014 ADW / RDW 2.78 3.64 6.60 3.35 2.53 2.75 3.14 2.25 3.23 - TFW / TDW 11.5 11.0 10.5 8.4 8.1 8.9 10.4 8.3 9.2 - LA (cm2leave-1) 15.73bc 14.27c 14.70bc 15.07bc 14.10c 13.28c 17.53b 14.20c 13.13c 0.0768 For each row, mean values followed by different letters indicate significant differences 357 following Duncan’s multiple range test (P< 0.05) and each value is the mean of six 358 replicates (n=6) per experimental unit (A, B, C, D1, E1, F1, D2, E2, and F2). 359 360 Table 4 shows the biomass production for Soleirolia plants. In this case, a much lower 361 weight per plant was obtained in module C (especially regarding the aerial part), 362 followed by F2.The total dry weight of plants in module C was 35 % of that obtained in 363 D1 and E1. Plants grown in lower modules had, on average, 66 % of the dry weight of 364 the plants in the upper modules. However, lamps D and F showed significant 365 differences between the upper and lower modules only due to the root part, and no 366 differences were found for lamp E. Precisely, lamp F was the one with a lower biomass 367 production in the lower modules, as the average total dry weigh of plants in module F2 368 was 57 % of that observed in E2 (though no statistically significant differences were 369 found between both). 370 371 Table 4. Weights determined for Soleirolia plants. TFW: total fresh weight; RFW: root 372 fresh weight; AFW: aerial fresh weight; TDW: total dry weight; RDW: root dry weight; 373 ADW: aerial dry weight. 374 Measured variables LW module P-value A B C D1 E1 F1 D2 E2 F2 TFW (g plant-1) 65.3bcd 61.0cde 36.1e 92.2ab 104.1a 70.2bcd 77.4abcd 84.4abc 51.4de 0.0001 RFW (g plant-1) 10.5e 10.3e 11.6de 29.4ab 32.7a 24.4bc 19.1cd 21.2bc 8.1e 0.0000 AFW (g plant-1) 54.8ab 50.6ab 24.5c 62.8ab 71.4a 45.8bc 58.2ab 63.2ab 43.4bc 0.004 TDW (g plant-1) 9.73ab 8.83ab 3.93d 11.14a 11.20a 7.75abc 7.00bcd 8.12ab 4.67cd 0.0000 RDW (g plant-1) 1.52cd 1.28cd 1.32cd 3.92a 3.73ab 2.59bc 1.74cd 1.74cd 0.74d 0.0000 ADW (g plant-1) 8.21a 7.56ab 2.62d 7.22ab 7.47ab 5.16bcd 5.26ab 6.38abc 3.92cd 0.0002 ADW / RDW 5.40 5.91 1.98 1.84 2.00 1.99 3.02 3.67 5.30 - TFW / TDW 6.7 6.9 9.2 8.3 9.3 9.1 11.1 10.4 11.0 - For each row, mean values followed by different letters indicate significant differences 375 following Duncan’s multiple range test (P< 0.05) and each value is the mean of seven 376 replicates(n=7) per experimental unit (A, B, C, D1, E1, F1, D2, E2, and F2).. 377 378 The evolution of the green cover expressed by the % of the LW module covered by 379 vegetation is shown in Figure 9. The vegetation initially covered around 28 % of the 380 LW modules and differences were already appreciated from the first week after 381 planting. In general, the upper modules showed a higher green cover, exceeding 80 % of 382 the LW module covered by vegetation at the end of the test in A, B and D1. E1 and F1 383 reached 79 % and 73 %, respectively. Module C, however, presented a much lower 384 coverage (64 %), similar to that obtained in the lower modules of the second test (67 %, 385 65 % and 71 % for D2, E2 and F2, respectively). 386 387 Figure 9. Evolution of the green cover (GC, %) in the different living wall modules 388 The number of Spathiphyllum white flowers in each LW module is shown in Figure 10 389 on a weekly basis. There was a big difference between tests, but not as much between 390 the lamps used. In the first one, the average number of flowers was 11, 18 and 12 for 391 modules A, B and C, respectively. In contrast, an average of 43, 50, 45, 46, 44 and 43 392 flowers were observed in D1, E1, F1, D2, E2 and F2, respectively. 393 394 Figure 10. Evolution of the number of Spathiphyllum white flowers in the different 395 modules 396 397 Table 5 shows the mean NDVI values obtained at the middle and end of each test. All 398 the values ranged between 0.68 (C and E1) and 0.91 (D2). After four weeks since 399 planting, all the values were fairly similar, though C already showed the lowest NDVI 400 value. Modules A, B, C and F2 maintained or a slightly increased NDVI at the end of 401 the tests. However, the NDVI decreased in D1, E1 and F1, showing lower values than 402 the rest of the modules (even C). Conversely, the NDVI was considerably higher for D2 403 and E2 at the end of the test. Only module B did not show significant differences 404 between weeks 4 and 10. 405 406 Table 5. Mean Normalized Difference Vegetation Index (NDVI) values taken for each 407 living wall module four and ten weeks after planting 408 Module Week A B C D1 E1 F1 D2 E2 F2 4 0.75de* 0.79c 0.68f* 0.82ab* 0.77cd* 0.79bc* 0.83a* 0.74e* 0.79c* 10 0.77d 0.79c 0.71e 0.69ef 0.68f 0.70e 0.91a 0.84b 0.82b Different letters in a row show statistically significant differences among the treatments 409 of each week (week 4th and week 10th) and the asterisk (*) indicates the statistically 410 significant differences between the treatments in both weeks (e.g. module A week 4 411 compared to module A week 10). 412 413 536 Figure 13. Relationship between the total dry weight (TDW) of Spathiphyllum (SP) and 537 Soleirolia (SO), green cover (GC) and the mean daily light integral (DLI). The dotted 538 lines denote the regression lines for each group of values. 539 540 The proposed optimum DLI value for Spathiphyllum is 4 mol m-2d-1(Faust, 2001), so 541 following this recommendation, only E1 received an adequate DLI, being close in D1, 542 but this did not have an influence on significant differences in the dry mass per 543 Spathiphyllum plant obtained (for instance module A showed the highest dry biomass 544 only receiving 1.8 mol m-2d-1).No proposed DLI values were found for Soleirolia, 545 though Yue (2004) suggested a quite wide PAR scope for the growth of Soleirolia, in 546 the range of 8.5 to 299 μmol m-2 s-1. In any case, the differences in plant development 547 between the lamp treatments found in our study, higher for Soleirolia than for 548 Spathiphyllum, suggest that the former seems to be more sensitive to DLI variations. 549 A higher DLI can also increase flowering (Currey and Erwin, 2011; Oh et al., 2009). In 550 our study, this did not happen as DLI for modules A and B were similar to D2 and F2 551 but there were far fewer flowers in the former. In this case, the mean daily temperature 552 might have been a key factor. According to Meng and Runkle (2014), the mean daily 553 temperature and the DLI can interact to influence the flowering time of various 554 ornamental crops. Also, the previous growing conditions in the nursery before the 555 transplant for the trials might have affected them as the differences in temperature 556 between tests 1 and 2 were low (3-4 ºC), being higher for the first one (when, precisely, 557 higher temperatures are supposed to induce flowering (Blanchard et al., 2011)). 558 The PPFD measured in our study was in general much higher than that reported by Egea 559 et al.(2014) (excluding lamp C). Biomass production in the present study was also 560 higher, especially for Soleirolia plants, except for module C, which produced similar 561 values to those observed by Egea et al.(2014). 562 The use of LED lamps also had implications on the water consumption. For instance, 563 the daily water volume consumed was slightly higher in the lower modules than in the 564 upper ones (for the same lamp) though the differences were not statistically significant. 565 In contrast, the results provided by Egea et al. (2014)denoted a bigger influence of the 566 type of lamp and the distance to the light source, as the heat produced by the lamps was 567 an issue. In fact, the water consumed in that study ranged between 2.1 and 5 L m-2 d-1, 568 while in the present work the values were between 1 and 1.5 L m-2 d-1. 569 As LW have a marked ornamental purpose, the healthy appearance of the plants and a 570 good vegetation cover are rather more important than the growth of the plants. In this 571 regard, even when there were few significant differences found in the generated plant 572 biomass, the vegetation cover was higher in the modules close to the light source. 573 Conversely, for lamps D, E and F (with a higher light intensity), the appearance of the 574 plants in the modules closer to the lamp became worse with the course of time 575 (especially in Soleirolia). 576 In this regard, it is interesting to note that in terms of the NDVI and the SPAD, those 577 modules specifically receiving a lower PPFD showed higher values. Receiving an 578 excessive luminous flux sometimes results in a decrease in the chlorophyll content of 579 leaves and vice versa (Dibenedetto, 1991; Zhang et al., 2016). Krause and Winter 580 (1996) even reported a certain photoinhibition of photosynthesis in species growing in a 581 Tropical forest when subjected to a highlight intensity exposure. Differences in the 582 NDVI can be associated with changes in pigment composition and protective 583 mechanisms against excess light (Mielke and Schaffer, 2010). 584 In spite of this, the participants in the perception analyses preferred lamps D and F over 585 the rest. The colour composition and temperature often have an influence on these 586 decisions (Jost-Boissard et al., 2009), but it seems that the lamps producing a 587 homogeneous distribution of light were also preferred over those creating a beam of 588 light. 589 Conclusions 590 When artificial lighting is required for indoor greenery, selecting the most efficient 591 lamps is very important, as the wrong choice may be crucial for the survival of a green 592 wall. All the commercial LED lamps tested in this study, except for lamp C which was 593 precisely the one designed for crop production, are apt for LW lighting. However, their 594 placement (the distance from the LW, the beam angle, the lamp orientation) should be 595 based on the lamp characteristics and plays an important role in obtaining a proper 596 result. Energy consumption should also be considered, as some lamps use the energy 597 more efficiently to produce light in the spectrum which is more usable by the plants. 598 Lastly, the visual quality of the light in terms of producing a natural appearance of the 599 vegetation is important in order to be pleasant for observers. 600 601 Acknowledgments 602 We would like to thank the spin-off company Terapia Urbana for providing the 603 materials to build the living wall prototypes. We are also grateful to the companies 604 Lledó and Ignia Green who lent us the lamps for the tests. Finally, our thanks to David 605 Sevillano, Patricio Espinosa and Ángela Morales, who helped us with the maintenance 606 of the LW modules and the measurements; Christina Mitsi, who helped with the 607 statistical analysis, and Antonio Franco Salas, who participated in installing the lamps 608 and living wall modules. 609 References 610 Blanchard, M.G., Runkle, E.S., Frantz, J.M., 2011. Energy-efficient greenhouse 611 production of Petunia and Tagetes by manipulation of temperature and 612 photosynthetic daily light integral. Acta Hortic. 893, 857–864. 613 https://doi.org/10.17660/ActaHortic.2011.893.94 614 Chen, C., 2005. Fluorescent lighting distribution for plant micropropagation. Biosyst. 615 Eng. 90, 295–306. https://doi.org/10.1016/j.biosystemseng.2004.10.005 616 Christopher Brickell, 2011. American Horticultural Society, Encyclopedia of Plants and 617 Flowers, First full. ed. DK Publishing, 375 Hudson Street, New York, New York 618 10014, United States. 619 Currey, C.J., Erwin, J.E., 2011. Photosynthetic daily light integral impacts growth and 620 flowering of several kalanchoe species. Horttechnology 21, 98–102. 621 Dibenedetto, A.H., 1991. Light environment effects on chlorophyll content in 622 Aglaonema commutatum . J. Hortic. Sci. 66, 283–289. 623 https://doi.org/10.1080/00221589.1991.11516155 624 Egea, G., Pérez-Urrestarazu, L., González-Pérez, J., Antonio, F.-S., Rafael, F.-C., 2014. 625 Lighting systems evaluation for indoor living walls. Urban For. Urban Green. 13, 626 475–483. https://doi.org/10.1016/j.ufug.2014.04.009 627 Fausey, B.A., Heins, R.D., Cameron, A.C., 2005. Daily Light Integral Affects 628 Flowering and Quality of Greenhouse-grown Achillea, Gaura, and Lavandula. 629 HortScience 40, 114–118. https://doi.org/10.21273/HORTSCI.40.1.114 630 Faust, J.E., 2001. Light, in: Hamrick, D. (Ed.), Ball Redbook: Crop Production. Ball 631 Publishing, Batavia, IL., p. 71−84. 632 GOTO, E., 2003. Effects of Light Quality on Growth of Crop Plants under Artificial 633 Lighting. Environ. Control Biol. 41, 121–132. 634 https://doi.org/10.2525/ecb1963.41.121 635 Gunawardena, K., Steemers, K., 2019. Living walls in indoor environments. Build. 636 Environ. 148, 478–487. https://doi.org/10.1016/J.BUILDENV.2018.11.014 637 Islam, M.A., Kuwar, G., Clarke, J.L., Blystad, D.R., Gislerød, H.R., Olsen, J.E., Torre, 638 S., 2012. Artificial light from light emitting diodes (LEDs) with a high portion of 639 blue light results in shorter poinsettias compared to high pressure sodium (HPS) 640 lamps. Sci. Hortic. (Amsterdam). 147, 136–143. 641 https://doi.org/10.1016/j.scienta.2012.08.034 642 Jost-Boissard, S., Fontoynont, M., Blanc-Gonnet, J., 2009. Perceived lighting quality of 643 LED sources for the presentation of fruit and vegetables. J. Mod. Opt. 56, 1420– 644 1432. https://doi.org/10.1080/09500340903056550 645 Kim, H.H., Wheeler, R.M., Sager, J.C., Gains, G.D., Naikane, J.H., 2006. Evaluation of 646 lettuce growth using supplemental green light with red and blue light-emitting 647 diodes in a controlled environment—A review of research at Kennedy Space 648 Center. Acta Hortic. 711, 111–120. 649 https://doi.org/10.17660/ActaHortic.2006.711.11 650 Krause, G.H., Winter, K., 1996. Photoinhibition of Photosynthesis in Plants Growing in 651 Natural Tropical Forest Gaps. A Chlorophyll Fluorescence Study. Bot. Acta 109, 652 456–462. https://doi.org/10.1111/j.1438-8677.1996.tb00598.x 653 Massa, G.D., Kim, H.-H., Wheeler, R.M., Mitchell, C.A., 2008. Plant Productivity in 654 Response to LED Lighting. HortScience 43, 1951–1956. 655 Mattson, N.S., Erwin, J.E., 2005. The impact of photoperiod and irradiance on 656 flowering of several herbaceous ornamentals. Sci. Hortic. (Amsterdam). 104, 275– 657 292. https://doi.org/10.1016/j.scienta.2004.08.018 658 McDonald, J.H., 2014. Handbook of Biological Statistics, 3rd ed. Sparky House. 659 Meng, Q., Runkle, E.S., 2014. Controlling Flowering of Photoperiodic Ornamental 660 Crops with Light-emitting Diode Lamps: A Coordinated Grower Trial. 661 Horttechnology 24, 702–711. https://doi.org/10.21273/HORTTECH.24.6.702 662 Mielke, M.S., Schaffer, B., 2010. Leaf gas exchange, chlorophyll fluorescence and 663 pigment indexes of Eugenia uniflora L. in response to changes in light intensity 664 and soil flooding. Tree Physiol. 30, 45–55. https://doi.org/10.1093/treephys/tpp095 665 Morrow, R.C., 2008. LED lighting in horticulture. HortScience 43, 1947–1950. 666 https://doi.org/10.21273/HORTSCI.43.7.1947 667 Moya, T.A., Van Den Dobbelsteen, A., Ottelé, M., Bluyssen, P.M., 2019. A review of 668 green systems within the indoor environment. Indoor Built Environ. 28, 298–309. 669 https://doi.org/10.1177/1420326X18783042 670 Oh, W., Cheon, I.H., Kim, K.S., Runkle, E.S., 2009. Photosynthetic daily light integral 671 influences flowering time and crop characteristics of Cyclamen persicum. 672 HortScience 44, 341–344. 673 Olle, M., Viršile, A., 2013. The effects of light-emitting diode lighting on greenhouse 674 plant growth and quality. Agric. Food Sci. 22, 223–234. 675 https://doi.org/10.23986/afsci.7897 676 Park, Y., Runkle, E.S., 2018. Spectral effects of light-emitting diodes on plant growth, 677 visual color quality, and photosynthetic photon efficacy: White versus blue plus 678 red radiation. PLoS One 13, e0202386. 679 https://doi.org/10.1371/journal.pone.0202386 680 Pérez-Urrestarazu, L., Fernández-Cañero, R., Campos-Navarro, P., Sousa-Ortega, C., 681 Egea, G., 2019. Assessment of perlite, expanded clay and pumice as substrates for 682 living walls. Sci. Hortic. (Amsterdam). 254, 48–54. 683 https://doi.org/10.1016/j.scienta.2019.04.078 684 Rueden, C.T., Schindelin, J., Hiner, M.C., DeZonia, B.E., Walter, A.E., Arena, E.T., 685 Eliceiri, K.W., 2017. ImageJ2: ImageJ for the next generation of scientific image 686 data. BMC Bioinformatics 18, 529. https://doi.org/10.1186/s12859-017-1934-z 687 Samuoliene, G., Brazaityte, A., Sirtautas, R., Viršile, A., Sakalauskaite, J., 688 Sakalauskiene, S., Duchovskis, P., 2013. LED illumination affects bioactive 689 compounds in romaine baby leaf lettuce. J. Sci. Food Agric. 93, 3286–3291. 690 https://doi.org/10.1002/jsfa.6173 691 Shaw, J., 2018. LED Colour Temperature and its Effect on the Growth of Hydroponic 692 Lettuce Seedlings. Young Res. 2, 164–171. 693 Singh, D., Basu, C., Meinhardt-Wollweber, M., Roth, B., 2015. LEDs for energy 694 efficient greenhouse lighting. Renew. Sustain. Energy Rev. 49, 139–147. 695 https://doi.org/10.1016/J.RSER.2015.04.117 696 Tan, C.L., Wong, N.H., Tan, P.Y., Ismail, M., Wee, L.Y., 2017. Growth light provision 697 for indoor greenery: A case study. Energy Build. 144, 207–217. 698 https://doi.org/10.1016/j.enbuild.2017.03.044 699 Thiel, S., Döhring, T., Köfferlein, M., Kosak, A., Martin, P., Seidlitz, H.K., 1996. A 700 Phytotron for Plant Stress Research: How Far Can Artificial Lighting Compare to 701 Natural Sunlight? J. Plant Physiol. 148, 456–463. https://doi.org/10.1016/S0176- 702 1617(96)80279-3 703 Torres, A.P., Lopez, R.G., 2010. Measuring Daily Light Integral in a Greenhouse. 704 Purdue Ext. HO-238-W, 1–7. 705 Turvey, C.G., Mclaurin, M.K., 2012. Applicability of the Normalized Difference 706 Vegetation Index (NDVI) in Index-Based Crop Insurance Design. Weather. Clim. 707 Soc. 4, 271–284. https://doi.org/10.1175/WCAS-D-11-00059.1 708 Warner, R.M., Erwin, J.E., 2005. Prolonged High Temperature Exposure and Daily 709 Light Integral Impact Growth and Flowering of Five Herbaceous Ornamental 710 Species. J. Am. Soc. Hortic. Sci. jashs 130, 319–325. 711 https://doi.org/10.21273/JASHS.130.3.319 712 Yeh, N., Chung, J.-P., 2009. High-brightness LEDs—Energy efficient lighting sources 713 and their potential in indoor plant cultivation. Renew. Sustain. Energy Rev. 13, 714 2175–2180. https://doi.org/10.1016/J.RSER.2009.01.027 715 Yue, H., 2004. Radiation (PAR) scope of indoor grouth of Soleirolia soleirolii. Chinese 716 J. Ecol. 03. 717 Zhang, H., Zhong, H., Wang, J., Sui, X., Xu, N., 2016. Adaptive changes in chlorophyll 718 content and photosynthetic features to low light in Physocarpus amurensis Maxim 719 and Physocarpus opulifolius “Diabolo.” PeerJ 2016. 720 https://doi.org/10.7717/peerj.2125 721 722