Twenty-three strawberry (Fragaria × ananassa) cultivars/accessions ( Table 1 ) were selected from the USDA-ARS-NCGR collection in Corvallis, OR, USA. All cultivars except one (‘Mara de Bois’) are short-day (SD) photoperiodic flowering types, due to the limited availability of off-patent long-day cultivars for public breeding. Rooted runner tips were planted in 396-mL black square containers with a coconut coir substrate (Finesse, Jiffy Group, Zwijndrecht, the Netherlands) and maintained under transplant growth conditions for 24 weeks in a walk-in growth chamber (Sankyo Frontier, Kashiwa, Chiba, Japan) at The Ohio State University (Columbus, OH, USA) before transferring to the production environment. Air temperature was maintained at a constant 20°C. An average photosynthetic photon flux density (PPFD) of 347 µmol m⁻² s⁻¹ (PAR, 400–700 nm) was provided with white LED light (FAHX30, Panasonic, Kadoma, Osaka, Japan) for a 16-h day/8-h night photoperiod. Locations of the plants in the growth chamber were randomized every other day to ensure even growing conditions. Plants were fertigated by hand every other day using municipal water with a Yamazaki strawberry formula containing (mg L^–1) 77 total N (70 NO₃-N and 7 NH₄-N), 21.4 P, 117.3 K, 40 Ca, 12.2 Mg, 16.1 S, 45.1 Cl, and micronutrients (Yamazaki, 1982). The pH and electrical conductivity (EC) of the nutrient solution were maintained at 6.5 and 1.0 dS m^–1.

All strawberry transplants were pruned down to three fully expanded leaves and one crown to standardize the transplant size from the long propagation time (24 weeks) before transplanting. Pruned plants were transplanted into 2.37-L black cylindrical plastic containers with the same coconut coir substrate (Finesse, Jiffy Group) covered with a 1-cm layer of rice hulls (PBH Nature’s Media Amendment, Riceland Foods, Stuttgart, AR, USA). The plants were grown in a 24 m² walk-in growth chamber (GH300, Conviron, Winnipeg, MB, Canada) at the Controlled Environment Food Production Research Complex (CEARC) at The Ohio State University ( Supplementary Figure S1A ). The plants were arranged in the growth chamber in a randomized complete block design with three blocks. All the cultivars had one plant in each block (n=3). Plants were arranged on table-top gutters with 7.1-7.7 plants m⁻² density. Border plants were placed on the gutters against the side walls and are excluded from the data collection.

A spatial average PPFD ± standard deviation of 332 ± 3 µmol m⁻² s⁻¹ at canopy level was provided with white LED (Gavita CT1930e, Hawthorne, Port Washington, NY, USA) mounted approximately 2.1 m above the canopy and measured using a quantum sensor (LI-190R sensor with a LI-250A meter, LI-COR Biosciences, Lincoln, NE, USA), and the light distribution is shown in Supplementary Figure S1B . Plants were first grown under a 13-h photoperiod (SD) for initial flower induction. During this period, many plants started showing semi-dormancy symptoms, and we extended the photoperiod to 16-h (LD) using photoperiodic lighting (FOCUS BR30, Focus LED, Deventer, the Netherlands) at 5 µmol m⁻² s⁻¹ PPFD with a 0.4 red-to-far-red photon flux ratio, providing a total photon flux density (400–800 nm) of 11 µmol m⁻² s⁻¹. The photoperiodic lighting was on for 13 h during SD and 16 h during LD. Daily light integral was 15.5 and 15.6 mol m⁻² d⁻¹ during SD and LD, respectively. The spectra with both white LED and photoperiodic lighting during the SD photoperiod and the extended photoperiod were measured using a spectroradiometer (PS-200, Apogee, Logan, UT, USA) and are shown in Supplementary Figure S2 . The timeline for photoperiod alternation was 12 weeks of SD, 8 weeks of LD, 5 weeks of SD, and 4 weeks of LD.

The air temperature during the 13-h photoperiod was set at 24°C, while the extended photoperiod and nighttime temperature was set at 12°C. The air temperature was measured at two locations at an approximate height of the plant canopy using Type-T thermocouples (gauge 36) and recorded with a CR23X datalogger (CR23X Micrologger, Campbell Scientific, Logan, UT, USA). The average day and extended-day/night temperatures over the entire production time were 24 ± 0.7°C and 12 ± 0.9°C, respectively. The relative humidity was set at 65%, except for a 6-h period during nighttime before daytime started when it was raised to 93% to prevent tip burn. The actual humidity was 63 ± 2% during the photoperiod and 89 ± 4% during the dark period (HMP60 sensor with a CR23X Micrologger, Campbell Scientific). The average air velocity at the plant canopy level measured perpendicularly to the air circulation direction using an anemometer (A004, Kanomax, Osaka, Japan) was 0.3 ± 0.09 m s^–1. The day (13 h) and night (11 h) concentrations of CO₂ in the growth chamber were 1000 ± 12 µmol mol⁻¹ and 402 ± 15 µmol mol⁻¹, respectively.

Plants were drip-fertigated daily using Jack’s Strawberry Part A (JR Peters Inc, Allentown, PA, USA) and calcium nitrate (Yara, Oslo, Norway), containing (mg L^–1) 80 N (72 NO₃-N and 8 NH₄-N), 24 P, 121 K, 44 Ca, 13 Mg, 50 S, 10 Cl, and micronutrients. The amount of nutrient solution that plants received was monitored three times a week and adjusted as needed to maintain a 0.1-0.2 drain-to-drip solution volume ratio. The pH and EC of the nutrient solutions were 6.7 ± 0.1 and 1.0 ± 0.1 dS m^–1. The pH and EC of the drain solutions were 6.8 ± 0.1 and 1.0 ± 0.2 dS m^–1. During production, runners and old leaves were removed weekly. The plants were pollinated by hand, a hand-held leaf blower (CMCBL710, Craftsman, Towson, MD, USA) to create mechanical shaking, and a bumble bee micro-colony of seven worker bees in the growth chamber.

To evaluate fruit earliness and productivity, time to first flower (d) and time to first harvest (d) from transplanting were recorded for each plant, and the flower-to-fruit harvest time was calculated. Fruit was deemed harvestable when the redness reached the proximal end row of achenes. Fruit was harvested at least three times a week, and the fresh weight of each harvest and the numbers of fruit were recorded. Average fruit size (g) was calculated as the total fruit fresh weight divided by the total number of fruit. The harvest goal for each plant was 200 g in order to provide enough fruit for fruit sensory analysis in a separate experiment. The number of weeks from the first harvest to when the harvest goal was reached was recorded. Average weekly yield (g) was calculated as the total fruit weight divided by the number of weeks from the week of the first harvest to the week when the harvest goal was reached. Maximum weekly yield (g) was defined as the highest yield produced in one week.

After the strawberry plants were transplanted and moved into the production environment, every month for five months (23 weeks), the plant height (cm) from the top of the substrate to the highest point of the plant shoot was measured using a ruler, and pictures of the plant’s projected canopy were taken. Images of the plant canopies were analyzed in a custom-made image analyzer operated with a Python program (v. 3.8) using the OpenCV library (v. 4.5.4) following an approach described by Kim and van Iersel (2023). Specifically, plant objects and backgrounds in the images were separated using intensity-based thresholding with the saturation channel in the HSV color model. The number of pixels in the plant objects was counted and converted into length (cm) and area (cm²) based on a reference scale. Plant sensitivity to dormancy-inducing SD photoperiod was quantified by the percent change in plant height (%), which was calculated as the percent difference between the average plant height under SD and LD conditions: (height_LD – height_SD)/height_SD, where height_SD was determined using plant height measured after plants were grown five and nine weeks in SD environment, and height_LD was determined using plant height measured after plants were grown seven weeks in LD environment.

To evaluate fruit morphology, three representative fruits, selected for representative color and fruit shape, were collected from each plant. The calyx was separated from the fruit, and the fruit was dissected into two halves lengthwise along the widest side. The calyx and the exterior and the interior of the fruit were scanned using a desktop scanner (CanoScan LiDE 210, Canon, Tokyo, Japan) covered with a black cloth. The scanned RGB images were analyzed using the Python program described above to determine calyx area (cm²), fruit diameter (cm), fruit length (cm), and fruit interior area (cm²). Fruit diameter and length were the maximum width and length of the cross-section, respectively, and the fruit diameter-to-length ratio was calculated. The calyx-to-fruit area ratio was calculated by dividing the calyx area with fruit interior area. The fruit interior redness and fruit exterior redness were quantified using the normalized difference anthocyanin index (NDAI) reported by Kim and van Iersel (2023). Specifically, the pixel intensities of the red and green channels of the scanned images (I_red and I_green) were used to calculate the NDAI values of the fruit objects using the equation: NDAI = (I_red − I_green)/(I_red + I_green). The pixel intensities of the images were adjusted using a standard gray card (50% reflection at all color channels) that has a 50% pixel intensity in each color channel.

To measure the concentration of fruit total soluble solids (TSS, or Brix) and TA, an additional 50 g fruit sample was collected from each plant, frozen at −20°C, and stored at −50°C. The samples were thawed at 4°C, homogenized, and centrifuged at 2490 × g for 10 min. The TSS concentration of the supernatant was analyzed using a digital refractometer (PR-32 alpha, ATAGO, Tokyo, Japan). For TA, 1 mL of the supernatant was diluted to 20 mL with distilled water and gradually titrated with 0.1 N NaOH to pH 8.2, and the pH change was closely monitored using a pH meter (accumet AB150, Thermo Fisher Scientific, Waltham, MA, USA). TA (g L⁻¹) was determined using citric acid as the representative acid, and the Brix-to-TA ratio was calculated.

Fruit firmness (N) was measured with an additional three fruits as sub-samples using a texture analyzer (TA.XTPlus, Stable Micro Systems Texture Technologies, Scarsdale, NY, USA) fitted with a 4 mm probe. The fruits were harvested into a tray with ice and transported immediately to the lab on ice for firmness measurements to eliminate changes of fruit texture. Each fruit was dissected into two halves, and the top 25% of the halved fruit was penetrated at a speed of 1.7 mm s⁻¹. The maximum gram force developed during the test was recorded and converted to Newton (N).

Statistical analyses of plant phenotypic data, hierarchical cluster analysis, and correlation analysis were conducted in R (v. 4.3.2) (R Core Team, 2023). All phenotypic data was analyzed using ANOVA-protected mean separation using Fisher’s least significant difference (LSD) with the R package “agricolae”. For fruit color, shape, firmness analyses, three representative fruits per plant were used as sub-samples. Hierarchical clustering was performed using Ward’s method, and four groups of strawberry cultivars were categorized based on the results of the analysis. The average value in each category was calculated using the mean of each cultivar ± standard errors across the cultivars. Correlation analysis was conducted using Pearson correlation matrix analysis.

Under the indoor growing conditions used in this study, all plants grew well and produced fruit after transplanting. Based on the average weekly yield (200-g yield divided by total number of production weeks) and maximum weekly yield (the maximum yield of a week), we classified the 23 cultivars into four categories of productivity levels using cluster analysis. These categories are 1) high average and maximum yield, 2) medium productivity with higher average yield, 3) medium productivity with higher maximum yield, and 4) low average and maximum yield ( Table 2 ). ‘Benton’, ‘Bountiful’, ‘Earliglow’, ‘Hood’, ‘Shuksan’, and ‘Tamella’ had high average and maximum yield, producing an average weekly yield of 43.0 g and a maximum weekly yield of 92.6 g per plant ( Table 2 ). Among cultivars with medium productivity, ‘Chandler’, ‘Dover’, ‘Mara des Bois’, ‘Marshall’, ‘Miyazaki’, ‘Perle de Prague’, and ‘Sweet Sunrise’ had a relatively higher average weekly yield of 37.4 g and relatively lower maximum weekly yield of 66.7 g ( Table 2 ). ‘Honeoye’, ‘Melody’, ‘Nyohou’, ‘Puget Beauty’, and ‘Tangi’ displayed medium productivity with a relatively lower average weekly yield of 23.9 g and a relatively higher maximum weekly yield of 79.9 g per plant ( Table 2 ). Low average and maximum yields were seen in ‘Elsanta’, ‘NW 90054-37’, ‘ORUS 1267-236’, ‘ORUS 740-7’, and ‘Redchief’, resulting in an average weekly yield of 24.3 g and a maximum weekly yield of 54.5 g per plant ( Table 2 ). When grown under greenhouse conditions, ‘Chandler’, ‘Elsanta’, and ‘Honeoye’ showed similar average weekly yield as in our indoor environment (Fletcher et al., 2002; Paparozzi et al., 2018). Different from our results, when cultivated under open-field conditions in Oregon, USA, ‘Mara des Bois’ and ‘Miyazaki’ had lower yield than ‘ORUS 740-7′ and ‘ORUS 1267-236′, and the yield of ‘NW90054-37′ was higher than ‘Tamella’ (Hummer et al., 2022).

Early flowering, early harvest, and short ripening time can help accelerate the production process of strawberries. These are valuable traits for indoor farm production as they can reduce operational costs during the non-production time (Hiwasa-Tanase and Ezura, 2016). Our results showed that without receiving chilling treatment, ‘Dover’, ‘Honeoye’, ‘Mara des Bois’, ‘Melody’, ‘Puget Beauty’, and ‘Redchief’ flowered in less than 45 days after transplanting with SD photoperiod, while ‘Hood’, ‘Miyazaki’, and ‘Shuksan’ took more than 90 days to flower ( Figure 1A ). The earliest fruit harvest was seen in ‘Mara des Bois’ in less than 70 days after transplanting ( Figure 1B ), likely because it is a long-day strawberry. As long-day strawberries do not require SD conditions for flowering, flower induction and initiation likely occurred during the vegetative growing stage with LD conditions. SD strawberries ‘Dover’, ‘Honeoye’, ‘Melody’, ‘Puget Beauty’, and ‘Redchief’ produced first harvest in less than 80 days after transplanting, while ‘Miyazaki’ and ‘Shuksan’ took more than 130 days to produce the first harvestable fruit ( Figure 1B ). In addition to cultivar and production environment, the development of flowers and fruit in strawberry can also be affected by pre-transplant chilling (Ariza et al., 2015). Typically, strawberry transplants are treated with chilling to promote early flowering and enhance yield, and chilling requirements are variable among different cultivars (Moreira et al., 2022). Even though none of the strawberry plants in this study received chilling, all of them flowered and produced fruit with SD treatment. ‘Hood’ and ‘Shuksan’ produced fruit relatively late in this study, likely because they are typically cultivated in the Pacific Northwest, including Oregon, Washington, and British Columbia (Hokansan and Finn, 2000). Therefore, chilling may promote earlier flowering and fruit production in these cultivars. Further studies may focus on evaluating the interaction between photoperiod and chilling on the earliness of strawberry production.

Flower-to-fruit harvest time of the 23 cultivars ranged from 28 to 42 days ( Figure 2 ). This flower-to-fruit harvest time for ‘Elsanta’ grown under open-field conditions at five different sites in Europe for three production cycles was 29–34 days (Krüger et al., 2012), similar to our observations in the indoor environment. However, Hummer et al. (2022) evaluated ‘Mara des Bois’, ‘Marshall’, ‘Melody’, ‘Miyazaki’, ‘NW 90054-37′, ‘Nyohou’, ‘ORUS 1267-236’, ‘ORUS 740-7’, ‘Puget Beauty’, and ‘Tamella’ under open-field conditions in Oregon, USA, and these cultivars had longer flower-to-fruit harvest time (35–62 days) than what we observed in this study. Based on these comparisons, time required for fruit development and ripening in strawberry varies among genotypes, while environmental factors, such as temperature, have been shown to affect strawberry fruit development and ripening (Krüger et al., 2012; Rosati et al., 1988).

Compared to open fields, indoor farms generally have less available cultivation space. To maximize the use of space without affecting plant growth, it is important to consider plant size when designing an indoor farm with vertical growing structure (Hiwasa-Tanase and Ezura, 2016). Maximum plant height must be considered when deciding the height of growing shelves and the level of lights. Maximum projected canopy area can be used to optimize cultivar-specific planting density, improving space use efficiency. Based on these morphological parameters, we divided the 23 strawberry cultivars in this study into four plant size categories: large, medium, small, and extra small ( Table 6 ). The large plants were ‘Miyazaki’, ‘Nyohou’, and ‘Shuksan’ with an average shoot height of 31.7 cm and an average maximum projected canopy area of 1609.3 cm² ( Table 6 ). Medium size plants included ‘Benton’, ‘Chandler’, ‘Dover’, ‘Hood’, ‘Marshall’, ‘Sweet Sunrise’, and ‘Tangi’, producing an average shoot height of 26.1 cm and an average maximum projected canopy area of 1333.5 cm² ( Table 6 ). Small plants, including ‘Bountiful’, ‘Earliglow’, ‘Elsanta’, ‘Honeoye’, ‘Melody’, ‘NW 90054-37’, ‘ORUS 1267-236’, and ‘Perle de Prague’ showed an average shoot height of 23.0 cm and an average maximum projected canopy area of 910.3 cm² ( Table 6 ). ‘Mara des Bois’, ‘ORUS 740-7’, ‘Puget Beauty’, ‘Redchief’, and ‘Tamella’ were classified as extra small plants with an average shoot height of 19.0 cm and an average maximum projected canopy area of 623.7 cm² ( Table 6 ). Pictures of representative plant canopies showed variation in plant spread among the 23 cultivars ( Figure 6 ). Among the 23 cultivars, ‘Mara des Bois’ and ‘Tamella’ had high average weekly yields relative to their extra small projected canopy areas ( Table 2 and 6 ).

Plant heights of ‘Mara des Bois’, ‘Marshall’, ‘Melody’, ‘Miyazaki’, ‘NW 90054-37′, ‘Nyohou’, ‘ORUS 1267-236’, ‘ORUS 740-7’, ‘Puget Beauty’, and ‘Tamella’ were also measured under field conditions (Hummer et al., 2022). ‘Marshall’, ‘NW 90054-37’, ‘Nyohou’, ‘ORUS 1267-236’, ‘ORUS 740-7’, ‘Puget Beauty’, and ‘Tamella’ displayed similar plant heights in an open-field environment as they were in the provided indoor environment. However, the plants of ‘Mara des Bois’, ‘Melody’, and ‘Miyazaki’ in this study were 2–5 times as tall as those grown in an open-field environment (Hummer et al., 2022). It is thus very important to consider variations of strawberry morphology when changing growing environments/systems.

Understanding the relationship among productivity, fruit quality and plant morphological characteristics of 23 strawberry cultivars can provide information to assist in the improvement of strawberry production. Our results showed that the amount of time to produce the first flower and first harvest from transplanting correlated positively with plant height and projected canopy area ( Table 7 ), indicating that larger plants took a longer time to produce flowers and fruit. This was likely because the plants allocated more photoassimilates to support vegetative growth than promoting the development of reproductive organs. Even though we did not directly measure plant biomass, plant size-related parameters (e.g., canopy area and plant height) have been shown to be highly correlated with strawberry plant dry biomass (Abd-Elrahman et al., 2020; Guan et al., 2020). A similar correlation between vegetative growth and earliness of production was observed in an open-field study with nine strawberry genotypes (Shaw, 1993). A phylogenetic comparative study analyzing 93 perennial herbs from the temperate regions of the Northern Hemisphere also suggests that plant height correlates positively with flower onset time (Bolmgren and Cowan, 2008). These results demonstrated a competitive relationship between plant vegetative growth and reproductive development, although no correlation was identified between plant size and productivity in this study (data not shown).

A negative correlation was identified between average fruit size and Brix ( Table 7 ), similar to what was seen in apple (Malus × domestica) and guava (Psidium guajava), indicating that larger fruit has lower soluble sugar content (De Salvador et al., 2006; Thaipong and Boonprakob, 2006). Reducing irrigation in tomato cultivation reduces fruit size but increases the content of total soluble solids (Chen et al., 2014; Wang and Xing, 2017). However, due to the different compositions of sugars in strawberry cultivars, further studies focusing on the analysis of specific soluble sugars and fruit sensory evaluation can help us better understand the relationships between fruit size, fruit sweetness, and consumer perception. Interestingly, there was no correlation between strawberry productivity parameters (average weekly yield and maximum weekly yield) and any of the fruit quality traits characterized in this study (data not shown), challenging the commonly held belief in the negative relationship between crop yield and quality (Cockerton et al., 2021). Future research in improving controlled environment strawberry production can focus on the enhancement of fruit quality without compromising crop productivity.

The majority of the cultivars evaluated in this study required SD conditions to induce flowering. However, SD conditions also induce semi-dormancy in strawberry plants. Semi-dormancy is a state of plant when plant growth is strongly inhibited but not completely ceased (Sønsteby and Heide, 2011). The symptoms of semi-dormancy include stunted growth, compacted shoot canopy, and reduced flowering and fruit productivity (Bodson and Verhoeven, 2005; Sønsteby and Heide, 2011, Sønsteby and Heide, 2021). After prolonged exposure to SD in this study, all cultivars of strawberry plants showed typical semi-dormancy symptoms. To recover plant vigor and productivity, LD extension lighting was introduced 12 weeks after transplanting. Plant sensitivity to dormancy-inducing SD photoperiod was quantified based on the percentage difference of plant height under SD and LD conditions ( Table 8 ). Twenty-three strawberry cultivars were classified into four different categories using cluster analysis: extra high, high, medium, and low sensitivity to dormancy-inducing SD ( Table 8 ). The cultivars that were most sensitive to SD photoperiod were ‘Benton’, ‘Earliglow’, ‘Elsanta’, ‘Honeoye’, ‘Hood’, and ‘Miyazaki’ ( Table 8 ). ‘Chandler’, ‘Dover’, ‘Mara des Bois’, and ‘Perle de Prague’ showed the lowest sensitivity to SD ( Table 8 ). Solutions to prevent semi-dormancy symptoms include alternation of SD and LD photoperiods (as used in this study), chilling, and cultivation of cultivars insensitive to a dormancy-inducing SD environment (Hytönen and Kurokura, 2020; Melke, 2015; Vince-Prue et al., 1976). Our classification of the 23 strawberry cultivars based on their sensitivity to dormancy-inducing SD photoperiod can provide information to assist in production management in strawberry indoor farming.