COMPARATIVE GROWTH OF ELEPHANT EAR TARO ( ALOCASIA MACRORRHIZA) AND GIANT SWAMP TARO ( CYRTOSPERMA MERKUSII) IN HAWAI‘I A THESIS SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAI‘I AT HILO IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE

Alocasia macrorrhiza and Cyrtosperma merkusii are root crops in the family Araceae that have the potential as animal feed in Hawai‘i. This research focused on growing C. merkusii and two varieties (Laufola and Faitama) of A. macrorrhiza to evaluate its growth, yield, and nutrient composition. A randomized complete block design was used to set up two growth trials in 2018 and 2019. Varieties were grown in pots in the first trial and directly on the ground in the second trial. Plant growth was measured weekly by plant height and leaf area of the main plants. Yield was measured by weights of the leaf blades, petioles and stems at harvest. Lateral plants and their weights were also measured. The yield data at harvest were statistically analyzed with a one-way ANOVA in PROC GLM and means were separated using a Post-hoc test, Least Significant Difference, at 5%. The influence of plant height, leaf area, number of leaves produced by main plants, number of lateral plants, and their total weight on yield were analyzed by Pearson's Correlation Coefficient. Wet chemistry analysis was performed for basic and mineral composition of the leaf blades, petioles, and stems harvested from the second growth trial. Growth and yield of plants in the second trial were generally superior to those in the first trial in which the Laufola variety had the highest growth increase in height and leaf area. The Laufola variety had the greatest average yield, in stem (54,896 kg/ha), petiole (99,647 kg/ha), and leaf blades (25,563 kg/ha) in the second growth trial. Plant height, leaf area, and number of leaves produced by the main plants had a strong positive influence with the yields. In the nutrient analysis, the leaf blades of the Faitama and C. merkusii variety had the ratios of Crude Protein-Acid Detergent Fiber-Neutral Detergent Fiber-Non Fiber Carbohydrates-Total Digestible Nutrients (CP-ADF-NDF-NFC-TDN) that could meet requirements or supplement the feedstuff

First and foremost, I would like to thank my committee chair Dr. Norman Arancon, for giving me the opportunity to conduct this master's research under his guidance, as well as providing me with employment to help and support my continuing education.Without his guidance this research would not have been successful.I owe many thanks to my committee members Dr. Sharadchandra Maratha and Dr.
Michael Shintaku for their invaluable time helping me to conduct the soil microbe analysis, and to identify the plant viral diseases in the field.Table 3.The moisture and the dry weights of the varieties grown in the second growth trial.
Basic and fiber analysis of yield of the varieties grown in the second growth trial.
Table 5. Mineral compositions of the yield of the varieties grown in the second growth trial.The edible portion of A. macrorrhiza and C. merkusii is produced above the ground.The stems are harvested as a staple food that supplies energy (Sakai 1983, Englberger 2008) primarily as digestible starch (Sakai 1983) for human consumption.Additionally, the leaves are important sources of protein, fiber, and minerals (Sakai 1983), making them an alternate food in the Asian region (Rashid and Daunicht 1979) during famine.The perennial growth of A. macrorrhiza and C. merkusii allows them to be harvested during off seasons for other common tropical crops such as C. esculenta and breadfruit (Sakai 1983).The crops have other desirable advantages in growth compared to the common varieties of C. esculenta and Xanthosoma including disease resistance, potential tolerance to pests, and fast recovery from environmental stress.In addition, it is commonly grown as an intercrop with yams, cassava, and coconuts.

LIST OF FIGURES
Alocasia macrorrhiza (L.) G. Don Alocasia macrorrhiza (Figure 1.) is commonly known as giant taro as well as elephant ear taro due to the sagittate-shaped leaves.The classification of the varieties is based on the origin: A. macrorrhiza (L.) G. Don.var.macrorrhiza (Malaysia to Pacific regions) and A. macrorrhiza (L.) G. Don. Var. violacea (India to Malaysia;Bailey et al. 1976).These varieties are further categorized by the degree of acridity and coloration.For example, the commonly eaten varieties of the Malaysian-Pacific region have reduced acridity in comparison to the other varieties of Asia (Barrau 1961, Migvar 1968).The wild types are recorded to have higher acrid levels, hence they are harvested during famine and required thorough cooking (Sakai 1983).
These crops are best cultivated in well-drained soils where the precipitation is more than 200 mm/year; common in upland areas and higher elevations of the islands.Additionally, it grows in soils too wet for other crops and also in dry conditions that C. esculenta cannot withstand.The growth rate is reduced in temperatures below 10 0 C and in prolonged waterlogged conditions.However, A. macrorrhiza can withstand water stress and shade, and can be grown as an intercrop under the canopy layer (Sakai 1983).The lateral plants that develop through cormels are separated for the cultivation practices, but seed production via sexual reproduction is also possible.Harvest time and crop cycles vary widely from 9 to 48 months.The plants grow more than 2 m in height at maturity and the stem grows up to 1 m in length, ultimately weighing more than 20 kg (Standal 1979).
Cyrtosperma merkusii (Hassk.)Schott Cyrtosperma merkusii (Figure 2.) is commonly known as giant swamp taro, as the cultivation practices are based on freshwater swamps (Sakai 1983, Englberger 2008).There are more than 100 cultivars spread within the Pacific region, with high diversity within islands (Englberger et al. 2004, Levendusky 2006) where the identification of these cultivars requires expert skills due to the wide variation in plant size, leaf shape and size, time taken for maturity, petiole spininess, and the color of the leaves and stems (Plucknett 1977).
The native region for these plants is in the tropics where plant heights can reach 6 m at harvest time (Sakai 1983, Englberger 2008).They can be grown in rainfed areas up to the elevations of 150 m where prolonged soil moisture is maintained (Foliaki et al. 1990, Plucknett 1977).It is one of the most common staple food plants thriving in harsh environments of Pacific atolls where low rainfall and high salinity in sandy soils are prevalent and the agricultural resources are limited (Mahoney 1960b, Pollock 1992, Englberger 2008).Cyrtosperma merkusii contains higher fiber content than C. esculenta (Dignan 2004) and other antioxidants such as carotenoids (Englberger 2008).Lateral plants (cormels) produced by the main stem as well as fertile seeds are used for propagation.The crop growth cycle can vary from 1 to 4 years in different cultivars, and the length of a single stem can grow more than 1 m, weighing 22 kg or more (Mahonet 1960a, Pollock 1992, Englberger 2008).
Geographically located as an isolated island range with a volcanic origin, there is a wide variability in the soils and the prevailing rainfall that impacts plant growth in Hawai'i.Nearly 1.3 million acres used for pasture and rangeland are marginal lands receiving low rainfall and existing soil characteristics make it difficult to grow high quality forage for livestock (The Livestock Industry in Hawaii, 1985).Even conventional crops require thorough land preparation for commercial-scale production.Only a few commercially grown crops are known to survive and grow in places where the soil is minimal with inhospitable climate, and varieties of giant taro and swamp taro have been identified among those few crops.The cost of the local meat and dairy products are relatively high as the expenses go for importing animal feeds (The Livestock Industry in Hawaii, 1985).Even the dairy industry is heavily dependent on imported feeds such as cereal grains, protein and mineral supplements from outside of the state and prices of the grains are nearly $60 to $90 (per 1 ton) higher than what is paid on the mainland (The Livestock Industry in Hawaii, 1985).Therefore, it is one of the major limiting factors in the local scale production of livestock.
Grown as ornamental plants, these cultivars are not popular food crops (Sakai 1983) or animal feed in the commercial market in Hawai'i.Nevertheless, these underutilized stem crops hold significant potential in crop production of the future in embracing climate change, land scarcity, and sustaining food security.The adaptation to different growing conditions, crop growth cycle, and the disease resistance make it suitable to grow under minimum resources (Plucknett 1977, Sakai 1983) Chapter 2. Materials and methods

Study Site, Experimental design, and Agronomic practices
The experiment was set up at the University of Hawai'i at Hilo Agricultural Farm Young cormel shoots (suckers) were separated from the main plants and established on the field.The tops of the plants (leaf blades) were removed to reduce evapotranspiration.The experiment was laid out into a randomized block design with 4 blocks and each single block (7.5 m x 4.5 m) consisted of 4 replications with 4 plants per replication.The taro varieties were considered as the treatments.The allocated space for one plant was 1.5 m x 0.9 m (5 ft x 3 ft) (Rashid and Daunicht 1979, Sakai 1983, Foliaki et al. 1990).The experimental area (16.8 m x 10.6 m) was covered by a black weed mat to suppress weed growth leaving holes for transplants.
The experimental area was surrounded by the C. merkusii (Pula'a) variety.
The plants were transplanted in polystyrene pots (3.8 l of volume) and established on the ground in the first growth trial (GT 1) by digging holes with a diameter of 40 cm and a depth of 30 cm.The bottom portion of the pots was removed allowing the roots to extend to the ground as plants develop.The growth media for the plants consisted of Hawai'ian black cinder and Pro-Mix BX substrate mixed in a 1:1 ratio.Pro-Mix BX was an all-purpose growing media with the composition of 77% -85% Canadian sphagnum peat moss, dolomite, and calcite limestone with adjusted pH, arbuscular mycorrhizal fungi (Rhizophagus irregularis), perlite, vermiculite, and wetting agent.A slow-release fertilizer (NPK 16 -16 -16) used in the experiment was applied at a rate of 89.6 kg/ha per month (Srivastava 1972).The study site was manually irrigated once every two days.
The plants were established directly in the ground in the second trial (GT 2) in August 2019 on a site with cane-washed soil from the Hamakua Coast.A soil analysis was conducted prior to the establishment of plants to determine the soil characteristics, including the nutrients.
All the varieties were provided with the same conditions of nutrients and irrigation.The plants were positioned in the same-sized holes as in the first trial with a diameter of 40 cm and a depth of 30 cm.The fertilizer ] recommendation for the experimental design was 280 kg/ha for each nitrogen, phosphorus, and potassium (Foliaki et al. 1990), applied at 3, 5, and 7 months after planting (Foliaki et al. 1990).Irrigation for the study site was an automated drip irrigation system.The irrigation frequency was set up based on winter and summer seasons on the island.The field was irrigated three days per week during the winter season and four days per week during the summer season, twice a day for 15 minutes per application.

Growth
Plant height and the leaf area of the tallest expanded leaf (Foliaki et al. 1990, Lewu et al. 2017) were measured every 7 days.The plant height was taken from the ground level of the stem of the plant to the petiole attachment point of the leaf blade and to the tip of the tallest leaf (Paul et al. 2015).The length of the leaf blade was taken by measuring the distance between the tip of the leaf blade to the petiole attachment point (Lewu et al. 2017).In addition, the widest width (breadth) above the petiole attachment point of the leaf blade was also measured (Lewu et al.

2017
).The leaf area was calculated using these measurements (length x width x ¾ of lengthbreadth ratio) (Montgomery 1911, Paul et al. 2015).In addition, other growth characteristics such as the number of leaves produced by main plants, number of lateral plants, and their total weight were also measured at the end of the harvest.

The total yields of the varieties
The plants were harvested 11 months after planting.The fresh weights of the stem, petioles and leaf blades of main plants were taken as yield.The total yield of each variety was calculated per hectare at harvest.

Statistical analysis
The yield (leaf and stem) of the varieties in GT 1 and GT 2 were examined using Analysis of Variance (ANOVA -Proc GLM) separately and the mean differences were tested using Fisher's Least Significant Difference (LSD) at 5% probability level, in SAS (Version 9.2).
The correlation coefficient of five growth components (average leaf area, average plant height, number of leaves produced by main plants, number of lateral plants, and the total weight of the lateral plants) of GT 2 was analyzed to find the influence on the yield in RStudio (Version 1.2.5033).Multiple regression analysis was also conducted for these characteristics to determine the further influence on yield.

Sample preparation and nutrient analysis
The nutrient analysis was done only for the varieties grown in GT 2. The plant tissues of the leaf blade, petiole, and stem were collected (approximately 200 g to 300 g) from the main plants, and recorded upon the collection.The leaf blades were randomly selected among young to mature leaves and cut into pieces.The petiole samples were collected along the vertical length above the stem attachment up to leaf blade attachment.The stems were separated into halves and a corer was used to collect the samples along the vertical length.The samples were dried in VWR Gravity Convection ovens in paper bags at 60 0 C until dry.The moisture content was estimated by taking the difference of fresh and dry weights.After drying, these samples were grounded separately using electric food grinders for further analysis.
The ground tissue samples were integrated, and three composite samples (leaf blade, petiole and stems tissues) were prepared per each variety.The integrated samples were submitted to the Dairy One Cooperative Inc. Forage Lab in Ithaca, NY, for basic analysis of forage fibers, protein, minerals, and energy levels.Wet chemistry analysis was performed by standard procedures.The feed ingredients of the analysis expressed the percentages of crude protein (CP), acid detergent fiber (ADF), neutral detergent fiber (NDF), non-fiber carbohydrates (NFC), and total digestible nutrients (TDN).The hemicellulose content was estimated by taking the difference of TDN and ADF.Moreover, the macrominerals such as calcium (Ca), phosphorus (P), magnesium (Mg), potassium (K), and sodium (Na) were given as percentages.The microminerals such as iron (Fe), copper (Cu), zinc (Zn), manganese (Mn), and molybdenum (Mo) were given in mg per 100g of sample.

Chapter 3. Results
The population was recovered from the top-cutting nearly 30 days after planting (DAP).
Thus, growth data was taken starting 30 DAP and repeated every 7 days.

Average plant height
There was a significant difference between the growth (height) of the varieties in GT 2 and GT 1. GT 2 exhibited a significantly higher growth in all the varieties compared to GT 1 (Figure 3.).The average height of the Laufola showed a rapid increase with a greatest recorded height of 238.9 cm (Figure 3.) followed by Faitama with the second-largest growth (205 cm) in comparison to equivalent varieties (Laufola and Faitama) in GT 1.However, the growth of the

Average leaf area
There was a significant difference between the growth (LA) of the varieties in GT 2 and GT 1.In GT 2, all the varieties showed greater average LA values than their equivalent varieties in GT 1 (Figure 4.).The Laufola had the greatest average leaf area (GT 2 -0.59 m 2 ) followed by
The average leaf blade yield of Laufola and the Faitama varieties significantly vary from the Pula'a and Control in GT 2 similar to the stem and petiole yields (Figure 7.).However, there was no significant difference in average leaf blade yield of Laufola and Faitama in GT 1 (Figure 7.a.).

The total yields of the varieties
The greatest yield per hectare was in petioles in both GT 1 and GT 2. The yields of stem, petiole and leaf blade of GT 2 were significantly larger than those in GT1.Laufola variety in GT 2 had a significantly higher yield in each stem (54,896 kg/ha), petiole (99,647 kg/ha), and leaf weights (25,563 kg/ha; Table 1) than the Faitama variety in GT 2 (Table 1).In addition, the Pula'a variety grown in GT 2 had lower yields compared to GT 1. Furthermore, the average weights of Control were relatively lower in GT 1 than in GT 2 (Table 1).

Correlation Analysis
The correlation analysis between yield and the five other growth characteristics are shown in Table 2. Number of leaves produced by the main plant had a significantly positive correlation with the stem, petiole, and leaf blade weights.Leaf area of the varieties had a strong positive correlation on the stem, petiole and leaf blade weights.Moreover, the average plant height had a notable positive correlation with the stem, petiole and leaf blade weights.
Additionally, it had a strong positive correlation with other growth components such as number of leaves produced and leaf area.The average plant height was highly correlated with the total weights of the lateral plants and had a negative correlation with the number of lateral plants produced.There was a significantly negative correlation between the number of lateral plants produced and the production of the stem, petiole, and the number of leaves generated by the main plant (Table 2).In addition, the number of lateral plants produced by the main plant had a strong negative correlation with the leaf blade weight and the leaf area.

Multiple linear regression
Multiple regression analysis was done between the yield and other components such as the number of leaves, leaf area, number of lateral plantlets, and the total weights of the lateral plantlets.The leaf area showed a significant positive influence on the stem (F5, 58= 74.56, R 2 = 0.85, p <0.001) and the leaf yields (F5, 58= 204.5, R 2 = 0.94, p <0.001).The yield components such as the number of leaves, number of plantlets, and the total weight of the lateral plants did not show a significant influence on the stem and the leaf yield increase.

Basic and fiber analyses of yield
Table 3 illustrates the moisture content and dry weights of all the varieties in GT 2. The petioles had the greatest moisture content (more than 83%) among the yields in Laufola, Faitama, and Pula'a varieties.In addition, the stems had the greatest dry weights (13% -34%) compared to petioles and leaf blades.The greatest CP contents were recorded in leaf blades of all the varieties (22.6% to 24.8%).The stem tissues had the lowest contents of the ADF (2.8% to 6.9%) and NDF (6.9% -22%) in all the varieties (Table 4.).The leaves contained a high NDF (35.7% -44.4%).The ADF contents of the leaves of all the varieties were also relatively higher than the petioles except in petioles of Pula'a (ADF = 36%, NDF = 43.9%).The Pula'a petioles and leaves had significantly similar amounts of ADF and NDF contents (Table 4).NFC amounts define the non-fiber carbohydrates and the neutral detergent soluble carbohydrates of the animal feed.The NFC values ranged from 20.7% to 31.1% in leaf blades.The stem tissues of the varieties had the overall greatest NFC values (Table 4).The stems had higher TDN (68% -77%) than the other yield components of the varieties.The Laufola and Faitama varieties had 68% to 70% of TDN in petioles and stems.The leaf blades had comparatively lower TDN values to the stems and the petioles.Hemicellulose content in the leaf blades and the stems were nearly the same and comparatively lower amounts presented in petioles.

Macrominerals of the yield
The leaf blades exhibited higher levels of macrominerals such as Ca, P, and Mg in comparison to the petioles and stems (Table 5).The highest values of Ca were found in Faitama (2.4%) and Laufola (2.0%).The P content was also high in leaf blades with the highest value recorded in Control (0.5%).The percentage of Mg was the same (0.4%) in leaf blades of Faitama, Laufola, and Pula'a (Table 5).However, the stems of the Faitama also showed a comparatively high level of Mg content (0.4%).The K levels vary in yields.The highest K values were found in petioles, leaf blades of Control, and petioles of Pula'a, respectively (Table 5).The Na percentage was higher in the petioles in comparison with leaf blades and the stems except in the Control (leaf blades -0.03%) (Table 5).

Microminerals of the yield
Microminerals are shown in mg per 100 g of the samples (Table 5).Leaf blades had higher Fe contents (8.9 -14.8 mg/100 g) in comparison to the petioles and the stems.The amounts of the Zn, Cu, and Mn were higher in leaf blades than in the petioles and stems.

Chapter 4. Discussion
The average plant height and LA were relatively high in all the varieties grown in GT 2 in comparison to the GT 1 except the Pula'a variety.In addition to the change of experimental plots of GT 1 and GT 2 within the farm, there were also changes in the frequency of irrigation and fertilization.The irrigation frequency and the amount of water received were higher in GT 2 compared to GT 1.The fresh weights of the corms and their dry matter content could decrease when reducing the irrigation water levels as a result of water stress, especially in crops such as C.
esculenta (Abd El-Aal et al. 2019).The amount of fertilizers received by the individual plants was different at the end of the crop time for both GTs (GT 1-89.6 kg/ha per month, GT 2 -280 kg/ha in 3, 5, 7 months after planting).However, the difference in the irrigation frequency and the amount of water received could be crucial in absorbing the applied nutrients to the soil in each GT.Wang et al. (2017) showed that the application of agronomic practices such as irrigation, and fertilizer led to enhanced nutrient uptake with increased soil moisture levels.The plants had limited space to develop roots and utilize the resources from the surrounding soil in the plants of GT 1 grown in pots.Evolving as understory perennial shrubs, the varieties of the Faitama and Laufola require large spaces.Gye-Bin et al. (2021) reported that increase in the size of containers enhanced yields of potatoes (Solanum tuberosum L.).The two Alocasia varieties Laufola (height -230 cm, LA -0.6 m 2 ), and Faitama (height -205 cm, LA -0.3 m 2 ) in GT 2 had the highest growth followed by the C. esculenta (GT 2 -125 cm, LA -0.11 m 2 ) among the two trials.The Laufola (height -99 cm, LA -0.2 m 2 ) and Faitama (height -78 cm, LA -0.1 m 2 ) grown in GT 1 had average growths lower than C. esculenta grown in GT 2. The height of the Laufola can reach more than 240 cm of height while Faitama can reach more than 150 cm with a harvesting time of a year (Foliaki et al. 1990).In addition, the growth in height can reach up to 500 cm within the 18 -24 months of harvesting time (Sakai 1983).The average height of Pula'a varieties in GT 1 was relatively higher than that of GT 2. During the harvest, it was observed that the roots of the Faitama and Laufola extended underneath the weed mat invading the spaces given to Pula'a varieties in GT 2. However, Pula'a variety in both GT 1 and GT 2 displayed a significantly lower growth in comparison to the Laufola and Faitama varieties.Certain cultivars of C. merkusii from the island nations can have a short harvesting time which begins from 6 -12 months after the initial planting (Sakai 1983).However, the majority of the varieties in the Pacific region require 2 years of growth to reach more than 2 m of height in wet marshes and the best growth occurred in waterlogged conditions (Plucknette 1977).In the C. esculenta variety, the growth (height -31.3 cm, LA -0.02 m 2 ) was the lowest in GT 1.One of the possible reasons for the lower growth was the severe outbreak of taro leaf blight in the field, infecting more than 50% of the plants in 205 DAP.The leaves of C. esculenta usually live for more than a month but could get destroyed after infection (Nelson et al. 2011).The highest average growth of the C. esculenta was observed in GT 2 where the root growth was not limited.However, the growth increment was lower in comparison to wetland cultivation.In addition, taro leaf blight infested the field in 107 DAP (Figure 3.) infecting nearly 50% of the C. esculenta population and spreading to the Faitama and Laufola varieties in GT 2. The infestation was not as severe as in C.
esculenta of GT 1, showing the symptoms on the margins of the leaf blades closer to the ground level.However, GT 2 had been occasionally treated with a fungicide to prevent further spread within the field.All the varieties grown in GT 2 had a relatively higher LA increase than their equivalent varieties grown in GT 1. Plant height increase was correlated to leaf area and the increase in height was directly proportional to the increase in leaf area (Harrington et al. 1997).
The varieties of Alocasia generated the greatest yields per hectare in both growth trials.
The Laufola variety in GT 2 produced 54,896 kg/ha (Table 1) of stem yield within the 11 month harvesting time and the amounts were more than twice those of the previous study given in Foliaki et al. (1990) (Laufola -24,900 kg/ha per year).The varieties grown for more than 18 months could produce 201,752 kg/ha of stem yield (Migvar 1968) which was higher than the stem yield production of Laufola in GT 2. The stem yield of Faitama (33,023 kg/ha) in this study was also higher than in Foliaki et al., (1990) (Faitama 20,506 kg/ha per year).Consequently, there was a highly significant difference in the yields (stem, petioles, and leaf blades; Figure 5., 6. & 7.) of the varieties grown in GT 1 and GT 2 at the end of the 11 month harvest period.
However, there were no records available to compare the leaf production (leaf blades and petioles) harvested as yield.The yield of Pula'a was significantly lower than the yield of the other varieties, and the yield of GT 1 was higher than the GT 2.Moreover, the stem yields of Pula'a (GT 1 -1,556 kg/ha, GT 2 -3,370 kg/ha) were extremely low in comparison to the lowest average values from Micronesia (10,000 kg/ha per year) when grown in waterlogged conditions.
In this work, the five growth characteristics of average plant height, average leaf area, number of leaves produced by main plants, number of lateral plants, and their total weight were analyzed for correlation coefficient to show their influence on the yields.The characteristics such as the number of leaves of the main plant, leaf area, and plant height had significant positive correlations to the yields of stem, petiole, and leaf blade weights (Table 2).The same trend was reported by Harrington et al., (1997) who showed that the height was highly correlated to the increase in the number of leaves and leaf area.The increase in the plant height resulted in the expansion and development of the canopy that facilitates solar radiation reception, production of energy and food partitioning.Paul et al., (2015) also reported that leaf size characteristic such as leaf area was highly correlated with the above-ground yield (stem, petiole, and leaf blade weight) that was considered as biomass and Harrington et al., (1997) showed similar trends in the growth trial reported here.In addition, the number of lateral plants produced had a significant negative correlation with yields and number of leaves, and leaf area of the main plants.The varieties such as Faitama (Foliaki et al. 1990) and C. esculenta naturally produce a large number of lateral plants.Therefore, different varieties also vary in generating certain growth characteristics such as the number of leaves per plant, number of lateral plantlets, and size of the stem yield (Figure 8.).
The lateral plants could negatively influence the yield due to nutrient partitioning.The improved varieties of C. esculenta that produced high yields had few lateral plants (Rao et al. 2010).
Therefore, maintaining fewer lateral plants in an area could improve yield in the main plants.
The yields of all the varieties had high moisture contents (Table 3).The remaining dry weights after removing the moisture component were low in yields.Dry matter content was important in determining the fiber and mineral composition of the tissues (Henning et al. 1996).
The nutrient requirements for feed vary depending on the animal species, age, and stage of lactation.The nutrient analyses were compared with the nutrient requirement for lactating cows as they required relatively high nutrient availability in their feeds.When considering the CP, ADF, and NDF in the tissues, it is desirable to have a 20-30-40 ratio of CP-ADF-NDF in forage or feedstuff given to the lactating cows (Henning et al. 1996) and the values of ADF and NDF are indirectly proportional to the quality of the forages (Saha et al. 2010).The leaf blades of the Faitama and Pula'a varieties fall into this desirable ratio whereas the Laufola variety had relatively low percentages of ADF (24%) and NDF (36%).None of the stem tissues of the varieties satisfied this ratio.Petioles of the varieties also did not fall into this ratio except for Pula'a which had sufficient amounts of ADF (36%) and NDF (44%).However, combining the Faitama and Laufola petioles, stems and leaf blades could satisfy the ratio of CP, ADF, and NDF.
When considering the NFCs, high-quality forages supposedly have high NFC values (NFC >20% desired) where the values were directly proportional to the quality (Saha et al. 2010).The stems and the petioles of Laufola and Faitama varieties had NFC amounts higher than 60% whereas the values with the lowest percentage in leaf blades were still above the desired amounts (20%).Pula'a variety also consisted of sufficient NFC amounts (leaf blades -23%, Petioles 35%, and stem -72%).Laufola and Faitama petioles and stems had more than 68% TDN which was relatively closer to the desired TDN levels of the dairy cattle (large breeds of 680 kg in Early Lactation -TDN -78%; Nutrient Requirement of Dairy Cattle 2001).The TDN content of the Pula'a (TDN -61%) leaf blades and the petioles were comparatively lower than the stems (TDN -71%; Table 4).However, these values were higher than the content of alfalfa (fresh forage at late vegetative stage -TDN 63%; National Research Council 1984); a common forage of the animal.TDN content of the yield was important as it indicated the energy availability of the feed for the growth and metabolism of the animal (Weiss 1998).Nevertheless, the content of TDN could be changed based on the maturity of the forages (Nutrient Requirement of Dairy Cattle

2001).
The mineral components consist of macrominerals and microminerals are important not only for maintaining health, but also the growth and development, reproduction and milk production.Macrominerals that should be in a feed include Ca, Mg, P, K, Na, and Cl, and S.
Among the macrominerals only Ca, Mg, P, K, and Na were given in the yield samples (Table 5).
Furthermore, five microminerals Fe, Zn, Cu, Mn, and Mo were detected in the analysis.The standard nutrient requirements (Ca -0.58%, Mg -0.40%, P -0.26%,K -3.00%, Na -0.10%; Nutrient Requirement of Dairy Cattle 2001) of dairy cattle were considered in comparing the existing nutrient composition of the yield.The Ca in leaves and petioles of Faitama, Laufola and Pula'a were twice as high as the required amounts.Mg levels in the yields were within the maximum tolerance levels required by the Beef Cattles.K percentages were within the tolerance levels except in the C. esculenta leaf blades (4%), petioles (5%) and Pula'a petioles (4%).Na percentages were within the preferred levels in Faitama stems and Laufola stems.In general, the stems consisted of relatively lower levels of macrominerals and were within the range of the maximum requirements.The leaves had relatively higher amounts of microminerals, and studies showed the presence of macrominerals as a defense mechanism for the herbivory and damage by the insects (Aqueel and Leather 2011;Chesnais et al. 2016).Leaf blades had nearly three times the Fe contents (Fe > 9 mg/100 g) compared to the petioles and stems.The Zn contents of the yield were relatively higher than the maximum recommended levels (3 mg/100 g) for dairy cattle.The leaf blades of Laufola, Faitama and C. esculenta had relatively higher amounts of Cu (Cu > 5 mg/100 g) than the petioles and stems.The Zn and Mn levels in Pula'a leaf blades exceeds the maximum tolerance levels (Zn -50 mg/100g, Mn -100 mg/100g; Nutrient Requirement of Dairy Cattle 2001).In general, the leaf blades had significantly higher amounts

Chapter 5. Conclusion
Faitama and Laufola varieties have a reassuring potential for dry land agricultural systems with lower requirements for fertilizers and water.Increasing the duration of the growth cycle (up to 18 months) could result in significantly higher stem production.However, the stem production after 11 months is already sufficient to meet the market demand.In addition, the leaf yield (petiole and leaf blades) individually provides sufficient growth with the same crop time and meets the dietary requirements of the animal feed.The growth parameters (plant height, leaf area and number of lateral plants) play an important role in determining the yield.Leaf blades, petioles and stems have sufficient amounts of dietary fibers, proteins, energy and mineral components essential for the diet of the livestock.Though the stem and petioles lack the preferable ratios and amounts of some minerals, combining them with leaf blades could be satisfactory.The Pula'a variety did not perform well in the dry land agricultural systems though the yield components meet the dietary nutrients of the forages.Further research will be required to evaluate Faitama, Laufola and determine the palatability as animal feed.
Table 1.Fresh yield (kg/ha) of the varieties grown for 11 months.The moisture content was estimated by taking the difference of fresh and dry weights of the yields.
Table 4. Basic and fiber analysis of yield of the varieties in the second growth trial.
All the components were given on a DM basis.The moisture content was completely removed.
Table 5. Mineral compositions of the yield of the varieties in the second growth trial.

I
am also very grateful to Dr. Jesse Eiben for his unwavering guidance during the early stages of this research.Special acknowledgement goes to the USDA for funding this research, and the Pacific Basin Agricultural Research Center (PBARC) United States Department of Agriculture (USDA) center in Hilo for supplying the planting materials.
Figure 1.Fully grown Colocasia esculenta (Control), Pula'a, Laufola, and Faitama (from left to right; in the second row behind the author) at harvest (11 months) in GT 2. The plants in the front row are all Pula'a.

Figure 2 .
Figure 2. Fully grown Cyrtosperma merkusii (Pula'a) main plant with lateral plants at harvest in GT 2.

Figure 3 .
Figure 3. Average plant height (cm) of four varieties taken at every seven days from (a) growth trial 1 and (b) growth trial 2 for eleven months.Error bars represent ± SE every seven days.

Figure 4 .Figure 5 .
Figure 4. Average leaf area (m 2 ) of four varieties was taken every seven days from growth (a) growth trial 1 and (b) growth trial 2 for eleven months.Error bars represent ± SE every seven days.

Figure 6 .
Figure 6.Average weights of petiole yields of the varieties in a) growth trial 1, 2) growth trial 2 at 95% CI.Error bars represent ± SE.

Figure 7 .Figure 8 .
Figure 7. Average weights of leaf blade yields of the varieties in a) growth trial 1, 2) growth trial 2 at 95% CI.Error bars represent ± SE.
. Hence, A. macrorrhiza and C. merkusii are promising varieties as potential resources for the challenging future.Taking these concerns into consideration, this research was focused on the growth of two varieties of A. macrorrhiza: Faitama, Laufola, and one variety of C. merkusii: Pula'a in dryland agriculture to evaluate their growth under the same environmental variables.Research Objectives a) Evaluate the growth and yield of A. macrorrhiza, and C. merkusii in dryland agriculture under the same environmental variabilities.b) Identify the yield-influencing growth parameters.c) Evaluate and compare the nutritional content and assess the potential of those varieties to be given as animal feed.
Laboratory (location -19.653°N, 155.050°W).The area is subjected to two seasonal variations: summer season (average rainfall -273.4 mm; Climate of Hawai'i 2014) from May to October and winter season (average rainfall -328.4 mm; Climate of Hawai'i 2014) in October to April (Giambelluca et al. 2014, National Weath er Service NOAA).The prevailing average annual air temperature is 22 -24 0 C (Giambelluca et al. 2014).Four taro varieties: Cyrtosperma merkusii (Pula'a), Colocasia esculenta (Control), and two varieties of A. macrorrhiza: Laufola, Faitama, were received from the Pacific Basin Agricultural Research Center (PBARC), United States Department of Agriculture (USDA) center in Hilo, Hawai'i on August 31, 2018.These plants were grown in pots under greenhouse conditions for three months before conducting the growth trials (GT 1 and GT 2).
Pula'a variety was significantly inferior to the A. macrorrhiza varieties (Figure3.)with the highest average values (79 cm) shown in GT 1.The average height of the Control (C.esculenta) decreased significantly at the end of the crop time in the GT 1.However, there was a gradual increase in height in Control (C.esculenta), which was greater than Pula'a variety in the GT 2 (Figure3.).

Faitama
variety (GT 2 -0.32 m 2 ) (Figure 4.b.).The Laufola in GT 1 (0.22 m 2 ; Figure 4.a.) and Control (0.11 m 2 ; Figure 4.b.) in GT 2, had the second and third highest values, respectively after the Alocasia varieties grown in GT 2. The Pula'a showed inferior growth in comparison to the Alocasia varieties in both GT 1 and GT 2 with the highest recorded value (0.07 m 2 at 289 DAP) shown in GT 2 (Figure 4.b.).The Control in GT 1 had the lowest growth that gradually declined at 177 DAP (Figure 4.a).

F3, 60
Figure 7.) yields of the growth trials.The average stem and petiole yields of Laufola and the n = 16.Varieties were grown at 1.5 m x 0.9 m (5 ft x 3 ft) per individual plant.Colocasia esculenta planted as the Control.

Figure 1 .
Figure 1.Fully grown Colocasia esculenta (Control), Pula'a, Laufola, and Faitama (from left to right; in the second row behind the author) at harvest (11 months) in GT 2. The plants in the front row are all Pula'a.

Figure 2 .
Figure 2. Fully grown Cyrtosperma merkusii (Pula'a) main plant with lateral plants at harvest in GT 2.

Figure 3 .
Figure 3. Average plant height (cm) of four varieties taken at every seven days from (a) growth trial 1 and (b) growth trial 2 for eleven months.Error bars represent ± SE every seven days.

Figure 4 .
Figure 4. Average leaf area (m 2 ) of four varieties was taken every seven days from growth (a) growth trial 1 and (b) growth trial 2 for eleven months.Error bars represent ± SE every seven days.

Table 2 .
Pearson correlation coefficient among the growth characteristics and yield of the varieties.

Table 3 .
The moisture and the dry weights of the varieties. a