Root carbohydrate storage in young saplings of an Amazonian tidal várzea forest before the onset of the wet season

Root starch and glucose content were measured for young saplings of 18 Amazonian tidal varzea tree species during a dry season. The pattern of carbohydrate storage depended on the type of plant involved and soil topography which is directly linked to flood regime. Most plants showed high root carbohydrate content at this point in the dry season, however, several typically flood-tolerant species (particularly palm trees) presented a low root carbohydrate content, suggesting a strategy of acquiring reserves during the wet season to survive the dry season, when depletion occurs. Plant survival in these flood-prone forests seems to be the result of more than only one adaptive mechanism.


Introduction
Flooding is a recurring phenomenon in many areas ofthe Amazon forest varying in nature, duration and depth. This has been described as one of the reasons that led to the present existence of a mosaic of habitats in the Amazonian floodplains (Junk , 1989). These distinct habitats were c\assified by Prance (1979) into seven groups. The habitat examined in the present work is what Prance called a tidal swamp forest, from the portuguese mata de várzea de estuário, which here will be called tidal várzea. These forests are often submitted twice daily to fresh water flooding backed up from tides. High tides temporarily block the flow of the rivers in the estuarine region and cause them to flood the adjacent forest.
Despite the economic relevance of the tidal várzea forests (rich in paIm trees, wood, and possessing a soil richer than most in the Amazon), few studies have attempted to investigate the effects of flooding on the plant species of these areas. U nfortunately, the shortage of ecophysiological information is also true for the rest of the Amazonian floodplains. Scarano and Crawford (1992) using Parkia pendula and Parkia discolor discussed flooding phenology and the concept of flood-tolerance. Junk (1989) related tree distribution with flood-tolerance for the Central Amazonian floodplains. Worbes (1985) studying the structural adaptations of trees to flooding in the Central Amazon, suggested that the adaptability of trees to flooding in the várzea forests is bound to be physiological rather than due to anatomical properties.
The ability to store carbohydrates in underground organs before the rainy season is one of the strategies that often guarantees survival for the plant during flooding (Steinmann & Brandle, 1984;Crawford, 1992), since anaerobic metabolism is costly in terms of carbohydrate consumption as compared with normal aerobic respiration (Crawford et alo 1989). There are few works which refer to carbohydrate balance in tropical plants. In an ear1y review by Kozlowski & Keller (1966) reference is made to some tropical studies. Figueiredo-Ribeiro et a!. (1986) studied reserve carbohydrates in the underground organs of Brazilian savanna (cerrado) plants. The present work uses the acquisition of carbohydrate before a potentially adverse flooding season as an initial parameter in comparing flood-tolerance in young saplings of various species of a tidal várzea forest in eastern Amazon.

Field work
The sampling of roots for carbohydrate analysis took place in Combu, a fluvial island of the river Guamá, near the city of Belém in the eastern part of the Brazilian Amazon . The area studied is a research work station of the Museu Paraense Emílio Goeldi . Root sampling was carried out on October 9th and 10th, 1991, at the characteristic peak of the dry season. At this time of the year flooding does not occur as often as during the rainy season (see Results).
A selection of uniform young saplings (based on height and number of leaves) for a same species, present in areas of high or low topography , was carried out for 18 tree species before subsequent root harvest (Table I). For taxonomic identification of the species studied, vegetative material of saplings was compared with that of identified adults. The main methodological difficulty encountered was to reach a uniformity of age for ali eighteen species studied.1t was possible, however, to estimate the age of the species studied as ranging from 1 to 3 years old. Species present in both high and low topography were sampled in both areas, producing a total of24 samples. For each sample, five replicates of one plant each were collected. It is important to highlight that the so-called high topography area is only 56 cm higher than the low topography area, which is, however, enough of a difference to provoke changes in physiognomy and species composition ofthe two areas (see Results). Root sampling in most cases was done simply by carefully pulling the whole plant as an intact core out ofthe soi!. When the soil presented resistance, plants were carefully dug out ofthe soil, making sure the roots were intact. After harvest, root and shoot length were measured, and the total number of leaves presented by each plant observed (data not shown). Subsequently the root system as a whole of each plant was deep frozen in liquid nitrogen, wrapped in alluminium foil and placed in incubators at below zero temperatures with silica ge!. At the end of a day of collection the material would be transported to the laboratories of CPATU-EMBRAPA in Belém where freeze drying took place. A week later, the root material was transported to The University of St.Andrews, Scotland, where carbohydrate analysis was carried out.

Carbohydrate analysis
Enzymatic analysis was used to determine starch and D-glucose content in the samples . Roots were reduced to a fine powder, and 100 mg ofthis was treated with 5.0 ml hydrochloric acid / 20.0 ml dimethylsulphoxide to solubilize the starch and the Dglucose present. Subsequently the solutions were left for 30 minutes in a 60°C water bath, before being filtered through sterilized muslin. The pH of the extracts was adjusted to 4.5 and the total volume increased to 100 ml by addition of distilled water. Boehringer-Mannheim kits for biochemical analysis of starch were used to perform the enzymatic analysis in a Pye Unicam ultra-violet spectrophotometer. Starch was hydrolised to D-glucose with the addition of amyloglucosidase. An enzyme suspension consisting ofhexokinase and glucose-6-phosphate dehydrogenase was used in Dglucose and starch determination . The measurements were made at room temperature using glass cuvettes (I cm light path) and a 340 nm wavelength. Details of the methodology are described in the instruction leaflet which accompanies the kits (Boehringer & Mannheim 1987). Table I. List of species studied and topography of the site where they were selected for root harvest and subsequent carbohydrate analysis .
Species studied high low area are a 1) PaIm species

Results
The gradient of only 56 em between high and low várzea is enough to establish a distinct flood regime for each area. During the wet season the main difference is in regard to duration of flooding: the high are a is flooded twice daily and the low are a is tlooded permanently. For most of the dry season, the low are as are tlooded once or twice daily and the high areas are flooded twice a month or less, depending on rainfall.
At the time roots were harvested for the present work, the low areas were tlooded once daily and the high areas had been already ca. 3 months free oftlooding, possibly due to an uncommonly severe dry season in 1991. In this context, the results can be seen as retlecting the response of plants to dry soils (high area) and to soils alternately waterlogged (low areas).
The saplings presently studied showed differential response to soil water saturation and topography, in terms of carbohydrate storage. Dividing the species studied into three groupsi) paim trees; ii) leguminous trees ; and iü) other tree speciesprovided the pattem seen in Figure 1. Paim species in the low topography areas show only 20% of the root starch and 41 % of the root glucose presented in the higher areas. Leguminous species showed similar amounts of both carbohydrates for high and low area. The other tree species, however, showed 35 % more starch and 28 % more glucose in the lower than in the higher areas.
Examination within each of these plant groups shows that more variation in the pattem of carbohydrate storage can be detected.   (le g ) and other dicotyledonous tree species (oth) , at two distinct topographical areas (higlz and low) in an Amazonian tidal várzea forest, before the onset of the wet season. Flooding is longer and more freqllent for the lower areas. Bars indicate standard error. n=20 for palms, legllminollS trees and other dicots at the lower topography and for palms at the higher topography (4 species in each grollp , each with 5 plants sampled). n=15 for leguminolls trees at the higher topography (3 species, each with 5 plants sampled). n=25 for other dicots at the higher topography (5 species, each with 5 plants sampled).
Paim species -Euterpe oleracea and Astrocaryum l1lurumuru are species of common occurrence in high and in low areas of the tidal várzea forests. Figure 2 shows that the carbohydrate stored in the roots of both species was reduced irrespective of topography, if compared with palms predominantly present in high areas (Maximiliana maripa and Socratea exhorriza). Reduced quantities were equally found in Raplúa taedigera and Bactris sp , which are characteristic of low areas. Similarly, Maximiliana maripa and Socratea exhorriza presented considerably higher starch:glucose ratio than their low and high/low are a counterparts. Leguminous species - Figure 3 shows that for the leguminous species the intra and interspecific differences were more pronounced than in the case of the paIm saplings. lnga nobilis andPentaclethra macr%ba are present in both topographies but showed, nevertheless , opposite patterns of storage. lnga nobilis presented 63 % lower starch leveis in the low are a compared with the high area, whereas Pentaclethra macroloba showed 49 % higher starch leveIs in the low are a compared with the high areas. Another lnga species,lnga edulis, which is predominantly oflow areas, had however 34% more starch than lnga nobilis of high area and 76% more than lnga nobilis of low area, probably characterising adaptation to flooding regimes. An opposite case of interspecific variation within the same genus is observed for Pithecellobium. Pithecellobium cauliflorum, characteristic of high areas showed 87 % higher starch leveIs than Pithecellobium latifolium, its characteristically 10w area counterpart. Glucose was stored in similar amounts for ali the leguminous species sampled. For this reason those species with higher starch leveIs showed higher starch:glucose ratio.
Other dicotyledonous species -In this group only Campa guianensis (Meliaceae) is commonly present in both high and low areas. This species seems to use different strategies for each area: in the low topography it has twice the starch and half the glucose stored in high areas (Figure 4). The starch:glucose ratio is 2: 1 in the high are a and 8: 1 in the low area. Figure 5 indicates that the predominantly low area species (Matisia pamensis, Virola surinamensis andPseudobombax munguba) were general-Iy richer in starch and glucose than the characteristical1y high area species (Guarea kunthiana, Sterculia speciosa, Protium sp. and Eschweilem coriacea). Pseudobombax munguba had the highest starch leveIs among the species studied (303.25 mg.gdw· I ). The starch:glucose ratio was often similar among the species in this group, with the exception oflow area Virola surinamensis which ratio was high compared to the others (15:1).

Discussion
The root carbohydrate storage of the saplings studied varied according to soil water saturation, topography and type of plant involved. The different patterns of carbohydrate storage presented by different groups of plants may retlect the existence of distinct adaptive strategies of survival. These variations in response could also be observed within the groups, intra and interspecifically.  Cattânio et aI. (in preparation) show that the low topographic areas of Combu Island have higher density and lower species diversity than the higher areas, being characteristically populated by paIm trees. This fact suggests a higher degree of adaptation to flooding by paim trees than by any other plant group in the forest. However, this was not reflected in the root food storage of the paim saplings prior to flooding. Setter et a1.(1987) say that plants with shoot in air and roots deprived of oxygen, usually have higher concentrations of carbohydrate than totally aerobic plants. The increase in carbohydrate leveis due to reduced use in growth, generally exceeds the depletion of carbohydrate that resuIts from Iow photosynthesis, often due to high stomatal resistance and acceIerated breakdown of carbohydrates in anaerobic conditions. NevertheIess, apart from presenting Iower amounts of root carbohydrate than the other groups of plants (leguminous and others), the paIms in the Iow waterlogged areas actually showed Iess carbohydrate stored than the high dry ones. This fact is particularly unexpected if one considers that at the time when root harvest was made, plants should be gathering reserves for the following wet season, when they become permanently flooded. This result could be explained by one or more of the following hypotheses: i) carbohydrate reserves could be stored in aerial roots or in the shoots and allocated to the underground roots when needed. Translocation of carbohydrates to roots is often reduced by flooding (Kozlowski & Pallardy, 1984) and is a major limiting factor in determining the length of anaerobic life (Brandle & Crawford 1987); ii) the venti1ation system of these plants could be such that they do not need to previous1y store food to survive long periods of flooding, being able to rely on currently produced carbohydrates throughout the year. Worbes (1986) however, studying plants of seasonal várzea and igapó, did not find air transport systems for the ninety tree species he investigated, although it is relevant to mention that this survey did not include palms.
iii) these species could be gathering carbohydrate during the wet season for consumption during the dry season, rather than the opposite. Keel & Prance (1979) studying igapó flooded forests suggested that drought may represent more of an impairment to survival than flooding to the local vegetation. The same could possibly be true for the case of the tidal várzeas.
If any of the hypotheses above are true, reduced carbohydrate leveIs in the roots of the waterlogged paIm saplings would not mean maladaptation.  The leguminous plants showed considerable intra and interspecific difference. Inga nobilis, although present in both high and low areas is possibly not as adapted to the lower areas as Inga edulis, as seen by the carbohydrate figures. Indeed, the latter appears predominantly in the low waterlogged areas. Pentaclethra macroloba seems to be well adapted to low topographies, where the individuaIs show considerably more reserves than in high areas. The results with the species of the genus Pithecellobium were rather surprising. Pithecel/obiwn cauliflorul11, a predominantly high area species, had 87% higher starch leveIs than Pithecel/obium latifolium, its characteristically low area species. The hypotheses formulated for the palms above are also possibly applied for Pithecellobium latifolium. This species is characterised by producing pneumatophores which should enhance internaI aeration .
Among the other dicotyledonous species, Carapa guianensis, commonly seen in both high and low areas, is probably the most curious case. Although expectedly showing a higher total carbohydrate in the low areas, this species showed a differentiation in the type of carbohydrate accumulated. The low area plants showed increased starch and reduced glucose compared with the high are a roots. Starch reserves are important in avoiding shortage offood reserves for growth when favorable environmental conditions induce a greater food consumption than can be replaced by current production of photosynthates (Ericsson & Persson 1980). Glucose and soluble sugars in general mightplay a role in preserving ultrastructure and survival ofroot tips (Setteret a!. 1987).
The ability of some tidal várzea plants to store carbohydrates in their root systems during the growing season, therefore, appears to be one ofthe adaptations that accounts to their success in surviving long periods of flooding. Nevertheless, as seen in the cases of the highly flood-adapted palms and the leguminous species Pithecel/obium latifolium, food storage is either unnecessary during the dry season, suggesting the existence of other adaptive mechanisms to flooding , or, conversely, food storage takes place during the wet season to allow the plants to survive the dry season. However, future work analysing carbohydrate reserves and depletion during a wet season shall be necessary in order to have a complete view of the role of carbohydrate storage in plant survival in tidal várzea forests.