Leaf Photosynthesis, Nitrogen Use Eciency, and Relationships with Growth and Yield in Teff

Background: Teff (Eragrostis tef (Zucc.) Trotter) is an important grain crop, but a paucity of research means that the mean yield is below 1.5 t ha −1 . Its high genetic diversity has not been exploited to improve its yield potential. Previous efforts at increasing yield were based entirely on phenotypic, morphologic, and agronomic merits. However, photosynthetic capacity has been neglected, so its possible contributions to yield improvements remain unexplored. Results: We grew 15 teff cultivars in a controlled environment to explore variations in photosynthetic capacity and nitrogen use eciency (NUE), and examined the relationships of gas exchange and NUE before anthesis with growth and yield attributes. Substantial differences were found in leaf photosynthetic rate (19 ± 9 μmol m −2 s −1 , mean ± SD), stomatal conductance (0.11 ± 0.09 mol m −2 s −1 ), and transpiration rate (2.4 ± 1.3 mmol m −2 s −1 ). The cultivars differed signicantly (P < 0.05) in both NUE (190 ± 227 g m −2 ) and photosynthetic NUE (59 ± 34 µmol g −1 s −1 ). On average, the plants partitioned 29% of N to leaf, 28% to panicle, and 13% to stem before anthesis. Yield and biomass production were closely associated with photosynthetic capacity and NUE. Clustering of the cultivars on the basis of photosynthesis, NUE, growth, and yield indicates wide variations in photosynthetic capacity and NUE in the wider teff gene pool that should be exploited. Conclusion: We conclude that leaf photosynthesis and NUE are positively related to yield and biomass production in teff.

Teff straw is a preferred feed for livestock (Assefa et al., 2001b;Assefa et al., 2011;Teklu and Tefera, 2005). Annually, 3 × 10 6 ha of land produces 5.2 × 10 6 t of teff grain in Ethiopia (CSA, 2018). Teff is the main ingredient in the traditional at bread called "injera", and is used in bread, pancakes, porridge, and alcoholic beverages in Ethiopia. Teff is grown mainly as a forage crop outside Ethiopia, but demand for the grain is expanding worldwide owing to its nutritional bene ts, mainly as it is gluten free and is rich in lysine, iron, and calcium (Baye, 2014;Mengesha, 1966;Paff and Asseng, 2018).
Although teff is a high-value crop, it has received little research attention ( Crop yield is a function of photosynthetic performance, yet there is no information on variations in the photosynthetic capacity and NUE of teff, governing factors, or potential contributions to yield improvement. Despite the large genetic diversity present in teff, potential variations in photosynthetic capacity and NUE are not well understood. Thus, yields could be increased through enhancing NUE, photosynthetic capacity, or both. Therefore, exploring and characterizing variations in photosynthetic capacity and related traits in teff genetic resources is an important task for improving yields. The objective of this study was to explore potential variations in photosynthetic capacity and NUE and their possible association with yield and biomass traits of teff. We grew 15 teff cultivars obtained from the Japanese Genetic Resources Center, National Agriculture and Food Research Organization (NARO), in a glasshouse. As this is the rst study to analyze variations in photosynthetic capacity and NUE in teff, knowledge gained from this work can be used to reveal variability of such traits within the wider teff gene pool and to improve yields via genetic and agronomic manipulations.

Results And Discussion
Leaf photosynthesis Cultivars had substantial variations in net photosynthetic rate (P n ), with signi cant differences (P < 0.01).
Overall, cultivars with better photosynthetic capacity had better growth and yield.
The variation in P n among cultivars indicates the presence of natural diversity in photosynthetic capacity within the teff gene pool. The observed variations in growth and physiological traits might be contributing factors, as crop biomass production is governed mainly by photosynthetic capacity (Driever et al., 2014;Mae, 1997;Makino, 2011;Raines, 2011). Until now, no attempt has been made to understand variations in leaf photosynthetic traits and the association with yield and biomass of teff. We found signi cant variability in leaf P n , g s , and T. P n at ambient CO 2 (393 µmol mol -1 ) under saturated light (1000 µmol s −1 m −2 ) ranged from 1.5 to 37.8 μmol m −2 s −1 . The photosynthetic rate of the improved cultivar 'DZ-01-354' ranged from 27 to 31.8 μmol m −2 s −1 at a range of temperatures (Kebede et al., 1989). In contrast, that of 'DZ-01-196' was <4 μmol m −2 s −1 (Tekalign, 2007). The disparity might be due to the collection of measurements at an early growth stage and the di culty of gas exchange measurements under nonuniform eld conditions. Compared with C 3 crops, the P n values of teff are slightly higher than ranges of Despite the complexity, the lower photosynthetic capacity of teff may be attributable to reduced photosynthetic apparatus due to the smaller plant size, lower rubisco kinetics, less e cient CO 2 xation mechanism, and less e cient chloroplast electron transport systems than in maize, sorghum, and Our results show that leaf P n is associated more strongly with BY than with GY. Biomass production is a direct function of photosynthesis, but yield formation is a more complex process that also depends on the e ciency of biomass conversion to grain yield, as in wheat ( found more variability if we had included more teff cultivars and wild relatives in this study, but the export of teff genetic resources from Ethiopia is currently banned.

N distribution and NUE
We quanti ed N contents of leaf (LN), panicle (PN), stem (SN), and seed (SDN) of teff cultivars and determined their relationship with photosynthetic and growth characteristics ( Table 2). We found signi cant variation in LN, SN, and SDN (but not PN) among cultivars. The means were 1.4 ± 0.67 mg LN g −1 , 1.4 ± 0.54 mg PN g −1 , 0.81 ± 0.50 mg SN g −1 , and 2.1 ± 0.55 mg SDN g −1 . On average, the N distribution ratios were 29% to leaf, 28% to panicle, 13% to stem, and 30% to seed at maturity. Differences in leaf chlorophyll content (Chl) among cultivars were highly signi cant. Chl ranged from 26.8 μg cm -2 in cv. 3711 to 49.3 μg cm -2 in cv. 3704, with a mean of 33.8 μg cm -2 . There was a strong correlation (R 2 = 0.72, P < 0.05) between LN and Chl. The variation among cultivars in N distribution ratios among plant parts might be attributable to substantial variation in NUE and P-NUE. Importantly, both BY and P-NUE were positively correlated with NUE, showing that changes in BY and GY are associated with changes in NUE (Fig. 4).
Plant N content is crucial to photosynthesis and to yield-and biomass-making processes, which vary with Leaf N status has been used in deciding N management and predicting yields of most cereal crops. Therefore, the dynamics of leaf N at different crop stages is not only a potential input for future N management strategies, but also a good indicator of yield performance in teff. As expected, our results show that Chl was closely associated with leaf N content. We derived the value of Chl from SPAD readings with a general empirical formula (Cerovic et al., 2012), because SPAD measurement is not calibrated for teff. Quanti cation of Chl would be more accurate if SPAD readings could be calibrated against actual extracted Chl. We expect that the correlation between leaf N and chlorophyll content would be stronger in this case. Mathematical relationships have been established between leaf N, Chl, and SPAD reading and have been used for predicting N requirements of several crops species (Wood et al., 1993), so this should be possible in teff.
However, we found no direct relationship between leaf N and P n , g s , and T, even though leaf N is one of . As we sampled the same leaf during leaf N quanti cation at the same time as measuring gas exchange, this could have contributed to the weak correlation between leaf N and P n . We recommend that further studies investigate natural variation in leaf N and NUE among teff cultivars and their interactions with N supply.

Growth performance
Plant height (PH), biomass (BY), and grain yield (GY) varied widely among cultivars. The mean PH increment was ~7 cm every 10 days. On average, cultivars height increment rate was 0.7 cm day -1 after planting (Fig. 5), but differences in BY and GY were signi cant (P < 0.05). The means were 69 ± 62 g m −2 for GY and 572 ± 444 g m −2 for BY. Both BY and GY were highest in cv. 3704 and lowest in cv. 3708 (Fig.   6). There was a strong correlation (R 2 = 0.94; P < 0.01) between BY and GY, as previously reported (Assefa et al., 2015); Dargo et al. (2016). Although not statically signi cant, days to emergency for cultivars was ranged from 4 to 6 days; days to heading ranged from 53 to 60 days; and days to physiological maturity ranged from138 to 157 days. Better GY is a function of enhanced crop growth that leads to better biomass production. In contrast, vigorous biomass growth may cause lodging, particularly at high N application, and thus yield reduction (Habtegebrial et al., 2007). Teff is sensitive to root and stem lodging (Habtegebrial et al., 2007). However, lodging was controlled in our study. Similarly, cultivars differed widely in their dry matter allocations to leaves, panicles (signi cantly), and stems before anthesis ( Table  2). The mean dry weight distributions per plant were 0.25 ± 0.05 g (14%) to leaf, 0.57 ± 0.33 g (35%) to panicle, and 0.77 ± 0.26 g (51%) to stem. Limited data is available on dry matter distribution of teff; however, the stem partitioning index was reported as 0.44 (Bediye et al. (1996). Dry matter allocation depends on crop or cultivar type, growth stage, soil nutrient availability, and moisture availability (Marcelis, 1996). Overall, our results show that allocating a higher proportion of dry matter resulted in greater GY. Similarly, performances in BY and GY were closely associated with photosynthetic capacity and NUE.

Traits association
We analyzed associations among 27 growth, yield, and photosynthetic traits. Among growth and yield traits, total plant dry biomass, total plant N content, panicle weight, panicle N content, and NUE explained most of the variation among cultivars (Fig. 7). Among physiological parameters, P n and T contributed most to the overall variation. Principal component analysis clustered the cultivars into three groups: group 1, cvv. 3704, 3729, 3715, 3712, and 3707; group 2, cvv. 3706, 3727, 3716, 3728, 3710, 3708, and 3703; and group 3, cvv. 3709, 3711, and 3713 (Fig. 7). This clustering indicates that NUE and photosynthesis-related parameters are good descriptors for evaluating teff genetic resources in addition to agronomic criteria. Genetic mapping using diversity array technology clustered cultivars into three groups also: group 1, cvv. 3703, 3713, 3729, 3707, and 3709; group 2, cvv. 3727, 3712, 3716, 3728, 3715, 3708, and 3710; and group 3, cvv. 3711, 3704, and 3706 (Fig. 8). Although the groupings do not match, the results show the presence of variation among the teff cultivars evaluated here. As most of these cultivars were collected from teff improvement elds, they were already selected by breeders for their yield-related merits. However, selection of teff cultivars has never been based on NUE and photosynthetic traits. Therefore, it is to be expected that their genetic differences would not t the variation observed here among variables related to their photosynthetic and NUE performance. As crop growth and development is a complex process governed by multitudinous genes, it is unlikely to always match genetic and phenotypic observations related to yield and physiology. This is why, for example, improvements in photosynthetic capacity of wheat and rice have been possible without changes in genetic makeup (Makino, 2011). Despite the wide diversity in agronomic and morphologic features of teff, little is supported by genetic information (Assefa et al., 2015).

Conclusion
The presence of signi cant variation in leaf photosynthetic rate, NUE, and growth performance among the 15 cultivars evaluated here reveals the potential variability that could be utilized from the huge diversity in teff genetic resources. The signi cant associations with biomass and yield show that using photosynthetic rate and NUE offers promise for increasing yield. Developing N-e cient cultivars will greatly help in minimizing the trend to increasing N fertilizer application and thus lodging, which reduces yield. Since we evaluated few cultivars compared with the thousands available, we recommend largescale characterization and identi cation of important phenotypic and genotypic traits related to higher photosynthetic rate and NUE.

Planting materials and growing conditions
The 15 teff cultivars were obtained from the Genetic Resources Center, NARO, Japan. They originated from Ethiopia and were imported into Japan in 1972 (see Table 1 for passport and accession number data). Formal collection and identi cation of the cultivars were made by NARO. In addition, voucher specimen of this material has been deposited in a publicly available. Further information can be accessed from the website: https://www.gene.affrc.go.jp/databases-plant_search_en.php . Teff is a tropical short-day plant that requires 11-13 h of sunlight and a temperature range of 20-22 °C (Kebede et al., 1989). Hence, it is di cult to grow teff outdoors in Japan, where daylength and temperatures uctuate throughout the year. So we conducted a greenhouse experiment at the Arid Land Research Center of Tottori University. The glasshouse was equipped with ceramic metal halide lamps with an average intensity of 98.6 µmol m −2 s −1 . Supplementary lighting was supplied every day at 05:00-07:00 and 16:00-18:00 to meet the average daylength and temperature requirements. During the experiment, the average daily minimum, mean, and maximum temperatures inside the greenhouse were 23.0, 23.4, and 30.9 °C. Seedlings were raised in plastic trays and watered every day. They were transplanted at 30 days (19 January 2018) into rectangular pots with dimensions of 75 cm (length) × 25 cm (width) × 55 cm (depth). Pots were lled with slightly acidic soil that contained 4.18 mg total C g −1 , 0.23 mg total N g −1 , 0.15 mg available P g −1 , 0.215 mg K g −1 , 0.22 mg Cu kg −1 , 13.00 mg Fe kg −1 , 12.8 mg Mn kg −1 , and 0.27 mg Zn kg −1 . Each pot had six hills 5 cm apart and two rows 25 cm apart, with two seedlings per hill. Seedlings were planted at a depth of ~3 mm. Each pot received a total of ~10 g of superphosphate (7.7% P; all at planting) and 5 g of ammonium phosphate (21% N; half at planting, half at booting stage). Pots were watered every other day; a water depth of ~5 mm was kept to maintain eld capacity. Plants were harvested on 14 June 2018, when >80% of the stands had reached physiological maturity.

Gas exchange measurements
Plants were grown in a uniform environment and treated equally to avoid the in uence of the external environment during leaf gas exchange measurements. Gas exchange was measured with a Li-Cor LI 6400 portable photosynthesis system. Using an infrared gas-exchange system with a 2-cm 2 leaf chamber, we measured fully mature ag leaves during 08:00-12:00 week before anthesis (heading).
During measurement, the photosynthetic photon ux density (PPFD) was kept at 1000 μmol photons m −2 s −2 by an LED light source with 10% blue and 90% red light. Throughout the measurements, the concentration of CO 2 was 393.2 ± 3.6 μmol mol −1 . The relative humidity was set at ~40% to stabilize the vapor pressure density within the chamber, and the temperature was held at 29.3 ± 3.2 °C. Data were recorded under steady state, 20 min after leaves were inserted into the chamber. Net photosynthetic rate (P n ), stomatal conductance (g s ), and transpiration rate (T) were calculated as described by (Evans and Santiago, 2014). P n and T were calculated from the uxes of H 2 O and CO 2 between leaf and atmosphere, leaf temperature, air ow rate, atmospheric pressure, irradiance, and leaf area in the chamber.

Yield and yield attributes
Four plants were randomly harvested from each cultivar before anthesis, and leaves, stems, and panicles were detached and weighed. Detached leaves were scanned on an HP Laserjet scanner and area was calculated in ImageJ software (Schneider et al., 2012). The mean of the four plants was recorded.
Leaves, stems, and panicles were oven-dried at 80 °C for 72 h.
Panicles were harvested at full physiological maturity. After air drying for a week, they were threshed, and grain yield was determined. The aboveground dry biomass weight was also determined after oven-drying at 80 °C for 72 h. Grain yield (GY) and aboveground biomass yield (BY) per plant were calculated as ratios to total weight and total number of plants of each cultivar.

Plant N quanti cation and NUE
Oven-dried samples were ground for N and C quanti cation. N and C contents of 500 mg each of leaf, stem, and panicle were determined by combustion in a JM1000CN Macro Corder, J-SCIENCE GROUP. NUE and photosynthetic NUE (P-NUE) were determined as: where GY is grain yield per plant; TPNup is total plant N uptake (N uptake by leaf + panicle + stem); P n is net photosynthetic rate; and LN is leaf N content.
Chlorophyll contents (Chl) of ag, primary, and secondary leaves were estimated with a SPAD-502 Plus chlorophyll meter (Konica Minolta) before anthesis. SPAD readings were converted to surface-based speci c units (μg cm -2 ) as (Cerovic et al., 2012):

Genotyping
In order to con rm and examine the genetic diversity of the 15 teff cultivars, fresh leaves were collected from one-month-old seedlings and genomic DNA was extracted as illustrated by Edet  Authors' contributions "FM is the major contributor in running the experiment and writing the manuscript. AT supervised the study, analyzed and interpreted the data. WT supervised the study, analyze and interpreted the data. MS supervised the study, analyzed and interpreted the data. NH supervised the study, analyzed and interpreted the data. MB involved in acquiring the study materials, data collection and analysis. EA supervised the study, analyzed and interpreted the data. MT supervised the study, analyze and interpreted the data. YG helped to analyze the genetic analysis, interpreting the result and writing the manuscript". All authors have read and approved the manuscript.  Means with the same letter are not signi cantly different at 1% or 5% level. CV (%), Coe cient of variation.