A COMPARATIVE STUDY OF FOUR TREE SPECIES USED IN RIPARIAN FOREST RESTORATION ALONG URUGUAY RIVER , BRAZIL

O conhecimento silvicultural e ecologico sobre as especies arboreas sao fundamentais para o planejamento da restauracao ecologica em areas de alta diversidade. Este estudo apresenta uma comparacao entre quatro especies arboreas nativas da porcao media da bacia do rio Uruguai: Heliocarpus americanus L. (Malvaceae), Maclura tinctoria (L.) D. Don ex Steud. (Moraceae), Schinus terebinthifolius Raddi (Anacardiaceae) e Cordia trichotoma (Vell.) Arrab. ex Steud. (Boraginaceae). Foram obtidos dados sobre o crescimento inicial, a interceptacao de luz, a producao de serapilheira e o seu conteudo mineral. H. americanus apresentou maior crescimento em altura e menor razao entre altura e largura de copa. H. americanus e M. tinctoria tiveram a maior interceptacao da luz solar (> 94%) e a maior producao de serapilheira (879 ± 151 e 792 ± 164 g·m-2·ano-1, respectivamente). A menor producao de serapilheira, para o conjunto das quatro especies, ocorreu entre julho e setembro. H. americanus apresentou os maiores teores de K (1,13 %) na serapilheira, enquanto C. trichotoma, os maiores valores de Ca e Mg (6,35 e 2,02%, respectivamente). S. terebinthifolius exibiu a menor interceptacao da luz solar e os menores teores de nutrientes minerais na serapilheira.


INTRODUCTION
The forest cover of Rio Grande do Sul State, in southern Brazil, is highly reduced and fragmented as a function of forest exploitation and agricultural expansion during the past decades (BRENA et al., 2003;FUNDAÇÃO SOS MATA ATLÂNTICA/INPE, 2002).A great portion of the forest cover in northwestern Rio Grande do Sul State is constituted of secondary forest at different stages (GRANELL-PÉREZ et al., 1999).Riparian forests, considered permanent protection areas according to environmental legislation, present less than 20 % cover (SCHENKEL et al., 2003).
Elevated costs (GARAY, 2006) and weed invasion (YU, 2004) are among the greatest challenges to forest restoration in the tropics.Grasses (Brachiaria, Pennisetum, Cynodon, Sorghum) are among the weeds with the greatest invasive capacity, since they commonly prevent or arrest forest re-establishment in the middle Uruguay River Basin (authors' personal observation).
Regenerative agroforestry systems (VIVAN, 1998) represent a strategy for native forest restoration, and are presently considered a valid option by Brazilian environmental laws (CONAMA Nº 369/2006).Agroforestry systems are also considered a method to reduce the costs of forest restoration (MEDRADO, 2000) and could be more effective in restoring forest diversity rather than abandoning it to secondary succession (RODRIGUES, 2005).In the Uruguay River basin, Ghisleni et al. (2003) observed an advantageous effect of annual crop cultivation in the first year on the initial growth of several native tree species, when compared to abandonment after tree planting.
In mixed native forest plantings, fast growth and higher survival of the pioneer species is expected.Pioneer species are also expected to increase soil cover, humidity and organic matter production, facilitating the late successional species, which could present higher initial growth and survival under the shade provided by the pioneers (KAGEYAMA and GANDARA, 2000).
Litter production is also a significant contribution to restoration, since litter supports soil microorganisms, and is the principal pathway for cycling mineral nutrients (LANDSBERG and GOVER, 1997).Thus, high litter production contributes to soil fertility restoration in early secondary succession (EWEL, 1976).
In some situations, a mixed native tree planting can be the best method for forest rehabilitation (PARROTA and KNOWLES, 1999).Nonetheless, little is known about silviculture and ecological responses of Neotropical tree species, and this information gap makes planning of forest restoration difficult with this method.This paper presents information on the silvicultural development and ecological contribution of four native tree species associated with a regenerative agroforestry system, employed in the recovery of Riparian forest in the middle Uruguay River basin, Doutor Maurício Cardoso county, Rio Grande do Sul State, Brazil.The principal variables addressed are initial growth, light interception, litter fall, and mineral content of litter.The results are discussed in relation to restoration practices and the successional status of the species, according to the literature.
The native trees were planted in August 2002, with alternate lines of two species groups: pioneer and nonpioneer species, with a spacing of 2.5 m between lines and 2.0 m between plants.The species sequence in each group was established at random.Annual crops, including tomato, maize, cassava, sweet potato, papaya, beans, pumpkin and sweet pepper (Capsicum) were cultivated among the lines.
The especies evaluated were S. terebinthifolius, M. tinctoria, H. americanus and C. trichotoma.The first is considered a pioneer species (CARVALHO, 1994).The second is considered an early secondary species (PAULA et al, 2004), but was included in the group of the pioneer species in the planting model based on previous local observations.There is no consistent information in the literature about H. americanus.C. trichotoma is considered an early secondary species (CARVALHO, 1994).
The period of the study was between the 19 th and 30 th month after planting.

Growth evaluation
The attributes measured were maximum height (h) and mean geometric crown width (lm).The mean geometric width was obtained from two transverse measures.Ratio h/lm was calculated for each plant from measurements made in the 19 th month after planting (n varied from 15-20 for each species).

Light interception
PAR light interception was measured between 10 and 11 am, at a distance of 1.5 m from the tree and at 0.2 m of height, on the south side (n = 10 trees).The data were obtained in May, 2004, with a LD-209 Instrutherm © luximeter.Each value, including total sun light, was obtained in triplicate.

Litterfall
Litter was gathered monthly with two 0.5 x 0.5 m traps per plant, with a total of 42 traps.The traps were at 0.5 m of height above the soil and at a distance of 1.5 m from the tree, with a 1.0 mm 2 mesh nylon screen.The plants were selected at random, and samples were obtained between February 2004 and January 2005, on 11 dates: March 22, 2004;April 24, 2004;May 29, 2004;June 08, 2004;August 10, 2004;September 11, 2004;October 08, 2004;November 17, 2004;December 25, 2004;January 27, 2005;February 25, 2005.For comparison between the species, the individual sampled values were added to obtain an annual value for each trap.The material was dried in an oven at 50 ºC prior to weight determination.

Mineral nutrient contents in the litter
The mineral nutrient contents (N, P, K, Ca, and Mg) were determined from samples from the annually mixed and homogenized pool.The chemical procedures were carried out at the UFRGS Soil Laboratory (Porto Alegre, Brazil).
The annual mineral input of each species was estimated by multiplying the mineral concentration by the annual litter production.

Statistical procedure
A completely randomized statistical design was adopted.The data were analyzed through ANOVA and the Tukey's test.The h/lm ratio data were transformed according to . The percentage mineral concentration data were transformed according to the arcsen procedure (ZAR, 1999).

Growth analysis
H. americanus, C. trichotoma and S. terebinthifolius presented the greatest growth in height.The highest crown width values were observed in H. americanus, M. tinctoria and S. terebinthifolius.H. americanus and M. tinctoria presented the lowest values of the h/lm ratio (Table 1).
S. terebinthifolius is classified as a pioneer species and C. trichotoma as a secondary species (CARVALHO, 1994).However, despite its succession classification, C. trichotoma has a high initial growth potential.On the other hand, the higher value of the h/lm ratio high apical dominance, which is a feature of early secondary species (BUDOWSKY, 1965) and could be an adaptation to escape from shaded understory conditions (POORTER et al., 2005).
The high values of crown width and low values of h/lm ratio of H. americanus and M. tinctoria are typical features of pioneer species and could favor the establishment and rehabilitation of the forest (BARBOSA, 2000).However, since M. tinctoria is a long-lived species (CARVALHO, 1994) must be included in restoration with caution: excessive or exclusive use of long-lived pioneers could arrest the succession and re-establishment of late successional species (MARTINEZ-GRAZA and HOWE, 2003).
The height growth of C. trichotoma was appreciably greater than in the data summarized by Carvalho (1994).The height growth of S. terebinthifolius was also significant, among the greatest data reported by that author for this species.

Light interception
Shading capacity is a major factor for forest ecosystem recovery.H. americanus and M. tinctoria intercepted more than 94 % of sun light (Table 2).C. trichotoma has an intermediate position intercepting nearly 90 % of sun light (Table 2).This could be related to the low crown width of this species (Table 1).On the other hand, as it is a deciduous species, this value probably presents a reduction in shading during winter (July to September). S. terebinthifolius showed the lowest light interception rate, in spite of the intermediate value of crown width, suggesting a low leaf area index, which in turn could be related to the small leaf size.The data indicates a minor contribution to soil protection by this species.These results should be considered with caution, however, as they represent only one period of measurements.
H. americanus presented the highest litter production in November through January and also in April, reaching 127 g/m² in December.M. tinctoria presented a similar phenology of litter production, with a maximum in December (191 g/m²).
The deciduous C. trichotoma presented a litter production peak between May and June, followed by a noticeable reduction.
S. terebinthifolius presented the lowest annual litter production; nonetheless, it was the species with a more regular phenology, and also a superior litter production in October (117 g/m²), a month with low production by the other species (Figure 2).This feature indicates a complementary contribution obtained by adding the species to the system, which could be significant in the spring time, a period with high germination and growth of weeds.
In contrast to the majority of native forests in Southern Brazil, which presents higher litterfall during August and September (KÖNIG et al., 2002;ARATO et al., 2003;VITAL et al., 2004;SANTOS and VÁLIO, 2002;PAGANO and DURIGAN, 2000;SOUZA and DAVIDE, 2001), the set of species studied showed the lowest litterfall in this period (Figure 2

Mineral nutrients in the litter
H. americanus and M. tinctoria presented the highest K content in the litter.Also, together with C. trichotoma, they presented the highest N content (Table 3).Palma et al. (2000) observed a Ca content of 4.5 % for M. tinctoria, an extremely high value when compared to other forest species (GONÇALVES et al., 1992).Nevertheless, the values here reported are even higher (Table 3).The high Ca content could be related to the capacity of Moraceae to accumulate Ca in the form of carbonate (cystoliths) and oxalate crystals (WU et al., 2006).The species with highest Ca and Mg concentration in the litter was C. trichotoma (Table 3), which also present cystoliths.
H. americanus presented the highest amount of estimated annual N, P and K returned to the soil through litter.On the other hand, C. trichotoma had the highest values in annual estimated amounts of Ca and Mg in the litter (Table 4).The return of Ca is extremely important in increasing the exchangeable Ca as well as pH maintenance in the topsoil (VANLAUWE et al., 2005).

Successional status and restoration
The growth, litterfall, light interception values for M. tinctoria indicate a pioneer status for this species (but see PAULA et al. 2004), in spite of its longevity (CARVALHO, 1994).Longevity, however, does not necessarily exclude the concept of pioneer, and some authors admit the existence of long-lived pioneers in tropical forests (WEBB et al., 1972;SWAINE and HALL, 1983).Litter decomposition could lead to a significant reduction in litter cover on the soil.Giacomini (2006) demonstrated that 50 % of the litter was decomposed in 240 days in Ijuí-RS, near the site of the present study.Since litterfall is variable along the year, in that same experiment (a monospecific plantation) the accumulated litter on the soil showed variations of 70 % in less than a year (COELHO and BORGES, 2006).S. terebinthifolius presents significantly lower light interception and annual litterfall.Also, its contribution to the mineral cycling is less.The great contribution of this species could be the litterfall increase during spring time.Additionally, as a zoochorous species, it is an attraction for frugivorous birds and could aid in the colonization by other forest species (REIS et al. 2003).

CONCLUSIONS
H. americanus presented the best performance in relation to this set of goals, with this highest maximum height, crown width, light interception, litterfall and N, P and K transfer to the topsoil.
M. tinctoria also showed significant growth, litterfall, light interception, and a high contribution to Ca cycling.
C. trichotoma showed a significant contribution to litterfall and especially to Ca and Mg cycling.Notwithstanding its low contribution to shading can be explained at least partially by its smaller crown width.
The set of species studied showed a significant reduction in litter fall between July and October, which could favor weed invasion.As Southern Brazil forests present a high litterfall in this period, a search for complementary early successional species to occupy this gap in litter production is recommended.