Leaf Area, Light Transmission, Roots and Leaf Damage in Nine Tropical Plant Communities

Ewel, J., Gliessman, S., Amador, M., Benedict, F., Berish, C., BermfJdez, R., Brown, B., Martfnez, A., Miranda, R. and Price, N., 1982. Leaf area, light transmission, roots and leaf damage in nine tropical plant communities. Agro-Ecosystems, 7: 305--326. The vertical distribution of leaf area by species; transmission of photosynthetically active radiation; root biomass and fine-root surface area; and leaf damage were measured in nine tropical ecosystems: six in Costa Rica and three in Mexico. Ecosystems studied included monocultures of maize (young and mature) and sweet potato; year-old natural succession and vegetation designed to mimic succession; a 2.5-year-old mixture of three arborescent perennials (cacao, plantain, Cordia alliodora); 2.7Tear-old plantation of Gmelina arborea; coffee shaded by Erythrina poeppigiana; and an old, diverse wooded garden. Leaf area index ranged from 1.0 in young maize to 5.1 in natural succession and the gmelina plantation. The vertical distribution of leaves was most uniform in diverse ecosystems, and most clumped in species-poor ecosystems. Light transmission was inversely proportional to leaf area, and two dense-canopied monocultures (sweet potato and gmelina) were nearly as effective at light capture as were some of the more diverse ecosystems. Optical density of the canopy ranged from < 0.5 (35% transmission) in the young maize to > 2.0 (< 1% transmission) in the natural succession. Large roots (> 5 mm diameter) accounted for most root hiomass in the older ecosystems at a soil depth of 5--25 cm, and fine roots (< 5 mm diameter) were most important in the surface 5 cm in all ecosystems. The range of values for root biomass (39 to 422 g m -2 to a depth of 25 cm) were similar to the range of values for leaf biomass (33 to 345 g m-:) , and, with the exception of two monocultures, ecosystems with high leaf biomass also had high root biomass. The surface area of the fine roots was lower than leaf area, and ranged from 0.5 to > 2.0 m= m -2 of ground. Total root surface area increased with *Present address: Environmental Studies, University of California, Santa Cruz, CA 95064, U.S.A. 0304-3746/82/0000--0000/$02.75 © 1982 Elsevier Scientific Publishing Company 306 age and diversity, and the monocu l tu res -even those effect ive at light capture had low root surface area. Herbivore damage on leaves of 35 species ranged f rom <2 to > 16% of leaf area. Heavily damaged species con t r ibu ted less to total ecosys tem leaf area than did species damaged less than average. Ecosystem-level damage was not well correlated with age or diversity. Leaf damage in all ecosystems ranged f rom about 2 to 10% of leaf area, or < 2 to > 25 g m -~ o f ecosystem. Young monocul tu res do not necessarily capture less light, provide less soil cover, and experience more herbivory than older, more diverse ecosystems. However , root surface area (and therefore possible nutr ient-capture abili ty) is high only in ecosystems that are diverse or old, and this is an impor tan t design considerat ion for agroecosystems appropriate for the humid tropical lowlands. I N T R O D U C T I O N The efficiency of resource utilization and resistance to pest attack are two key issues in agriculture, especially as fertilizers and pesticides increase in cost. This is particularly true in the humid tropics, where year-round growth permits rapid pest and disease build-ups, high rainfall promotes nutrient leaching, and weeds invade aggressively. Structurally diverse multiple-crop tropical agroecosystems, such as those described by Wilken (1970), Dickinson (1972), and Gliessman et al. (1981), might reduce these problems more than the monocultures now often used. These diverse, structurally complex plant communities might have several desirable features: (1) Well developed surficial root mats and roots that extend well below the soil surface might impede nutrient loss. (2) A diverse habitat that houses predators and parasites, and reduces plant apparency, may ameliorate pest attack. (3) Unbroken canopies lessen the impact of high-intensity rains on the soil (Greenland 1977). (4) Dense foliage may reduce weed invasion. (5) Crop mixtures often yield more than the same area divided among monocultures of the component crops, and sometimes even "overyield", i.e., yield more than an equivalent area of a monoculture of the higheryielding component (Trenbath 1974). (6) Diversity may reduce the risk of total crop failure, should one or more components be devastated by pests or weather (Wilken 1975, Innis 1980). Are such benefits inevitable in complex, diverse agroecosystems? The results of many field and pot experiments comparing simple and complex systems, as well as explanations of the nature of the interactions involved, have been reviewed by Trenbath (1974), Harper (1977, Chapts. 8--11), Kass (1978), and Nair (1979}. Increased yield and lowered risk of complete crop failure are common, but not universal, benefits of species mixtures. Yield increases presumably result from increased efficiency of resource utilization (light, water, nutrients, CO2, and soil 02), but field data that 307 directly support this common assumption are few. Some yield increases may also reflect complementary effects among species, such as the maizebean-Rhizobium interactions described by Boucher (1979). Although the idea of reduced pest attack in diverse systems is appealing, both increased and decreased losses to pests can, in fact, occur in polycultures (Murdoch 1975, Litsinger and Moody 1976, Altieri et al. 1978). To find out if structurally complex ecosystems make better use of resources and experience less herbivory than do simple systems, nine varied agricultural and successional ecosystems were studied, ranging from simple to diverse, herbaceous to woody, short to tall, and young to old. We concentrated on a few study areas to reduce variation due solely to geographic variables; we used identical techniques at all sites; and we made all measurements in a short interval (2 weeks). Measurements included: leaf area index (LAI) by height and by species and optical density of the canopy, both indicators of a system's light-capture ability; root biomass by depth and diameter class, an indicator of ability to exploit root-zone resources; and leaf damage (caused primarily by herbivorous insects) by species, an indicator of resistance to pest attack. The study areas Six of the nine ecosystems were on the grounds of the Centro AgronSmico Tropical de InvestigaciSn y Ensefianza (CATIE), Turrialba, Costa Rica. CATIE is at latitude 9°51'N and occupies lands ranging from 600 to 660 m. Mean annual rainfall is about 2700 mm and the dry season extends from January through March. Three of the CATIE study sites were on the Colorado soil series (family: very fine, mixed, isohyperthermic) and the other three were on the Insti tuto series (family: fine, mixed, isohyperthermic); both soils are Inceptisols (Aguirre 1971). The other three study sites were in the state of Tabasco, Mexico. Two sites were on the grounds of the Colegio Superior de Agricultura Tropical (CSAT), located 23 km west of the town of C~rdenas (18 ° I 'N) . The other site was at the town of Masateupa (18 ° l l ' N ) . Annual rainfall averages about 2200 mm at CSAT and about 1600 mm at Masateupa. Both areas experience a dry season from mid-March to mid-May, and bimodal rainfall with maxima in June and September--October . Both areas are on a recent alluvial plain (elevation approximately 10 m). Most soils in the area are Entisols high in clays, principally of the LimSn and Nuevo series (Mejia 1978), and are subject to frequent water logging during the wet season. Each of the nine ecosystems studied is briefly described below. Systems 1--3 were at CATIE on the Inst i tuto soil; 4--6 were at CATIE on the Colorado soil; 7 and 8 were at CSAT on the LimSn soil; and 9 was at Masateupa on the Nuevo soil. 308 (1) Sweet potato This was a 48-day-old, 2000 m 2, unfertilized monocul ture of Ipomoea batatas {variety C-15). The plot was nearly weed-free, having been weeded 22 days after planting. The canopy was low, dense, and uniform. Insecticide had been applied three times. (2) Cacao-plantain-Co rdia This was a 2.5-year-old, 450 m 2 experimental planting of Theobroma cacao (varieties EET400 X SCA 12, UF29 X IMC27, and Catongo X Pound 12), at 3 X 3 m {1111 individuals per ha), Musa X paradisiaca (variety pelipita) at 3 X 3 m, and Cordia alliodora (a fast-growing native t imber tree) at 6 X 6 m (278 individuals per ha). The plot was weeded during its first year, both manually and using herbicides. No insecticide had been applied, and routine maintenance consisted of occasional light pruning of the Cordia and thinning of the plantain shoots. (3) Shaded coffee This was a 25-year-old, 2 ha planting of Coffea arabica at 1 X 2.3 m {4348 bushes per ha), with an overstory of Erythrina poeppigiana at a spacing of about 8.3 X 4.2 m (287 trees per ha). Fungicides are applied annually in April, June, and August, and herbicide is applied sparingly every 2 months. The coffee is pruned annually in March, and the Erythrina (a nitrogen-fixing legume) is pruned in January--February and July--August each year, but had not yet received its January--February pruning for 1980 at the time of our measurements. (4) Gmelina This was a 2.7-year-old, 0.8 ha planting of Gmelina arborea, a fast-growing pioneer tree, native of India and South-East Asia, that is planted in the humid tropics for t imber and paper pulp. This planting was in two blocks, one spaced at 2 X 1 m {5000 trees per ha) and the other at 2 X 3 m {1667 trees per ha). During its first year part of the gmelina was interplanted with maize and another part with beans (Phaseolus vulgaris). During that first year, only the maize-gmelina intercrop was weeded and received inputs of insecticide and herbicide. The gmelina were pruned at 1.5 and 2.5 years, and at the time of our measurements averaged about 10 m tall with trunks 14 cm in diameter. (5) Succession This vegetation was a diverse mixture of species that recolonized three 256 m 2 plots in a second-growth forest that had been felled 14 months and burned 11 months earlier (Ewel et al. 1981). The vegetation was about 5 m tall, and consisted of a dense mixture of vines, shrubs, large herbs and small trees. The three plots (all located within 1 ha} contained > 100 plant species. 309 (6) Mimic of succession This vegetation had identical site location, history, and plot size as the succession, bu t consisted of species planted by the investigators, rather than those introduced through natural processes of dispersal and coppicing. It contained about 40 species, including both economic and non-economic species. This mimic is being developed as part o f a larger project designed to test the feasibility of building agroecosystems that imitate the structure and function of natural successional ecosystems. At the time of this s tudy the mimic vegetation was about the same height but less dense than the natural successional vegetation it was designed to imitate. (7) Young maize This was a 2-month-old, 0.12 ha test planting (local variety criollo blanco) on a site that had been cultivated continuously, using machinery, since forest clearing 12--15 years earlier. Beans and maize had been planted in rotation on the site each June and December, respectively, for the three previous years. Planting was in furrows 0.92 m apart, With 4--5 seeds deposited every meter, giving an initial density of about 44,000--50,000 plants per ha. The site was fertilized with 80-40-40. Weeds were controlled with a preemergence herbicide followed by mechanical cultivation three weeks after planting. (8) Old maize This was a 3.5-month-old, 0.5 ha planting (of the same variety) established by students using traditional farming methods of the region. The soil was prepared 2 years earlier with machinery, bu t had since been left fallow; the vegetation was cleared with machetes prior to planting. Planting was done by placing 4--5 seeds in lO--12-cm-deep uncovered holes, about 1 m apart. This resulted in an initial density of 40,000--50,000 plants per ha. A light application of urea was made at planting and again at 30 days, but no insecticides or herbicides were used. The plot was weeded by machete 30 days after planting. At the time of sampling the maize was 3--4 m tall, the ears were well formed yet still green, and the weed cover was dense. (9) Wooded garden (huerto familiar) This plot, of ca. 0.5 ha, was more than 40 years old, and contained a diverse mixture of useful plants, ranging from timber trees (e.g., Cedrela mexicana, Colubrina sp.) and fruit trees (avocado, coconut, mango) to lightdemanding herbs (e.g., tomato, chili). It was typical o f such gardens in Tabasco, as described by Gliessman et al. (1981). The structure and floristic composit ion of these multi-layered gardens are carefully manipulated by their owners, who harvest both cash crops and products for home consumption from them. The most conspicuous species in the plot sampled were coconut, cacao and coffee, but a cursory inventory revealed the presence of 38 species. 310 METHODS L e a f a r e a The leaf area index (LAI), defined as the area of leaf (one side only) per unit area of ground, was measured in each agroecosystem using a plumbbob method. Using this method, the unit ground area is reduced to a point, and each measurement consists of the number of leaves touching a thin cord, marked at 25 cm intervals and extended vertically through the vegetation. Each intersection of cord and leaf was recorded by species and height above the ground. This method underestimates true LAI by an amount dependent on leaf angle (deviation from horizontal). In most of the systems, sampling locations were chosen using randomly determined paired coordinates. In the natural succession and the mimic it was necessary to work from existing trails, and points were chosen randomly along the trails. At each of nine sampling locations, six (two at the ninth location) LAI measurements were made, giving 50 determinations per ecosystem. The six points at which the cord was extended through the vegetation were evenly spaced about the sampling location, at a distance of 1--2 m. In the wooded garden, the height of the vegetation created special problems. The plot was divided into nine subplots of equal size. Within each subplot, the measurements were made from a tall tree, with the cord extended vertically 3 m from the tree trunk. In a few cases it was necessary to estimate the number of leaves above the top of the cord. Ligh t transmission Photosynthetically active radiation ( P A R = 400--700 nm) was measured using a LiCor meter (LI-185A) coupled to two sensors (LiCor 190S) via a switching box. One sensor was placed in a fixed position in a clearing; the other was placed on the soil surface in the ecosystem, leveled, read, and immediately moved to a new position. One hundred near-simultaneous pairs of readings (clearing/under vegetation) were made in each ecosystem. Sampling points were selected by twice extending a 25 m tape along the ecosystem floor (at an oblique angle to plant rows, where present) and placing the sensor along it at 50 cm intervals. Each pair of readings was converted to a percent light transmission and optical density (OD), where: OD = logw ( P A R c / P A R v ) , P A R c = light in clearing, and P A R v = light beneath vegetation. Optical density was calculated as it is linearly related to absorbance plus reflectance by the vegetation. All statistical comparisons were made using optical density, rather than percent transmission. R o o t s Twelve root samples were taken from each of two depths (0--5 cm in311 cluding litter, and 5--25 cm) in each ecosystem. Sample points were determined using randomly selected coordinates. Shallow roots were sampled by driving a sharp edged, 5.5 cm diameter bulk density sampler into the soil. In all six Costa Rica ecosystems, deeper roots (5--25 cm) were sampled with a long, 4.2 cm diameter corer, designed by D. Santantonio of Oregon State University. In Mexico, deeper roots were sampled in the young maize and wooded garden using a sharpened, 4.6 cm diameter pipe, and in the old maize by excavating 5 X 5 cm cores. Each soil core was soaked briefly in water containing a dispersing agent (household detergent), then placed in a tank containing an overflow spout. Water was circulated into the bo t tom of the tank, and the flotsam, including roots, was collected on a 0.5 mm sieve placed beneath the overflow spout. Roots were separated from all other flotsam, then separated into six diameter classes (in mm): < 1, 1 to < 2, 2 to < 5, 5 to < 10, 10 to < 20. Live and dead roots were not distinguished, but no obviously dead large roots were encountered. Roots in the smallest size class were rewashed to ensure removal of all external mineral matter, including soil particles bound to the roots by mycorrhizae. The clean root samples were dried to constant weight at 70°C and weighed t o 0.0005 g. Roo t mass for the three smallest diameter classes was converted to surface area using linear regressions developed by C. Berish (unpublished) based on 120 root samples taken from the successional vegetation and mimic (plus other Sites) in Costa Rica. The length: biomass regression equations (L = length (mm), M = dry mass (g), and subscripts indicate diameter classes (mm), were: L< 1 = 10103 M< 1 L 1_2 = 1392 M1--2 L 2 5 = 341M2--5 Lengths were then converted to surface area by assuming that all roots were cylinders having a diameter equal to the midpoint of the diameter class. Leaf damage In each ecosystem, 1 to 11 dominant species were chosen for measurement of leaf damage.. Dominance was determined by the LAI measurements, and the species selected accounted for 60 to 99% (~ = 79) of the total LAI of each ecosystem. Damage was measured on a total of 35 species. For each species for which damage was assessed, individual plants and/or branches were chosen arbitrarily (if individuals were few) or randomly (most species). Once a branch or plant was selected, leaves were chosen randomly from among all potentially acceptable leaves using the method described by Ward (1974). Any fully expanded leaf that had not been damaged by human activity was potentially acceptable, regardless of the amount of damage it had experienced. No more than two leaves were collected per plant, except 312 for species having few individuals, where an effort was made to include more than five individuals of varying ages, sizes, and locations. Thirty leaves per species were collected in this manner. With a few exceptions, damage was measured for complete leaf blades, including all leaflets in the case of compound leaves. For eleven large-leaved species, fractions of leaves were either randomly or systematically chosen fo r damage measurements. A clear plastic sheet was laid over each leaf, and damage traced and filled in with a permanent black marking pen. Two kinds of damage were distinguished, and traced onto separate plastic sheets. All missing tissue, plus damage that left only a transparent layer of leaf tissue, was recorded as holes. All other damage, including that caused by leaf miners, piercingsucking insects, pit feeders, viruses, etc., plus the necrotic tissue around holes, was recorded as brown spots. Area determinations were made with a portable area meter (LiCor LI-3000) equipped with a high speed belt drive. This instrument measures area to the nearest 1 mm 2 and is accurate to within 1%. For each leaf, three area measurements were made: holes (H), brown spots (B), and residual area of the leaf (R). Percent of leaf area damaged was then calculated as: Damage = [(H + B)/ (R + H)] × 100 All leaves of each species from the same ecosystem were pooled, oven dried to constant weight at 70 ° C, and weighed to 0.05 g. The leaf specific mass (mass per unit area of leaf) of each species was then calculated so that damage could be expressed on a mass basis as well as an area basis.