Final report on studies of nutrient cycling on white and black water areas m Amazonia

Studies were conducted near Manaus, Brazil in cooperation with INPA to try to establish how nutrient cycling influences the formation of black water and white water. The .;tudies measured the rate of decay of Caryocar villosum leaves on spodosol and oxisol terrestrial and aquatic sites when the leaves were untreated, and treated with a bacteriostat, o r insecticide or fungicide. It also measured litter, animal populations, and the elemental content of ten biologically important elements in soils and decomposing litter. Results show cor.siderable differences in the rates of decay and the agents and end products of decay which indicate that black water and white water forma... tion are closely tied to the rate and type of d~cay I.L'1d to basic soil types and their associated v~::ge­ tation, except for the eediments in white water.


lNTRODUCTION
lt was observed long ago that the areas of Amazonia which generate black water rlvers have white sands which have undergone podzolization.while those which produce white water rivers have oxisols from which sílica has been leached.The two processes.podzolization and laterization (producing oxisols in the new terminology) differ primarily in the pH of the leaching solution which reaches the soil, and the oxidation conditions.Spodosols are thought to be produced when the leachate from the litter into the soil is acid, thus making iron and aluminum readily soluble and hence leachable.Oxisols are produced when the leachate passing from the litter into the soil is more nearly alkaline so that sílica is solub.lizedand leached to deeper horizons.The extensive differences between black water N. Stark (* ) C. Holley ( *) and white water (Kiinge, 1967;Williams et ai., 1972) are thought to arise from differences originating with these two soil forming processes, and some differences of parent material .
The suspended material characteristic of white water is carried by these streams because of the nature of the clays that form on the oxisols which generate white water.However.clays from black water areas from the spodosols will turn the water milky white when these clays are soaked for a fong period of time .White water per se, often has black water in it and swampy areas on oxisols may give rise to black water.Also. the leaves of many types of tropical and temperate vegetation will produce black water of one type in acid conditions and black water of another type in alkaline conditions.
Since the precipitation which falls over forests on oxisols and spodosols is about the same in pH (4-5).the differences in the pH of the leaching solution must arise as a result of the chemical changes occurring on the leaf and bark surfaces during throughfall and stemflow, and within the litter itself during decomposition.The differences between the two major soil groups (oxisol and spodosol) in the tropics, and their resultant water chemistry have never been closely tied to the chemical processes which dominate litter decomposition.
This study attempted to determine whether black water is actually generated on land o r under water, and what the major decomposer groups were on spodosols and cxisols as well as in black and white waters in Amazonia.
Many studies have dea lt with the mysteries of black and white water (Brinkman, 1970  and personal communication; Fittkau, 1967;  Sioli et a/., 1969; Schmidt, 1970).In the last ten years, many studies of water quality have been conducted to try to describe the water of this huge network of rivers.Other studies have dealt with the soils (Sioli , 1966) and the sedimentary loads (Gibbs, 1967).For this reason, this study did not concentrate on water quality but on decompositio•n and nutrient cycling.
This research was sponsored by the Instituto Nacional" de Pesquisas da Amazônia (INPA) and was conducted from Manaus, Brazil.Some of the analyses were done at the School of Forestry, Llniversity of Montana, Missoula, Mor.tana.The study site for spodosols was west of Manacapuru on the Rio Manacapuru, while that for the oxisols was a wil of intermediate character on the new Punta Negra road out of Manaus, Brazil.
In this report, "igapó" refers to flooded areas with black water, while "varzea" refers to periodically flooded white water areas adjacent to oxisols."Campina" refers to a special type of forest (10 m) rich in epiphytes and growing on white sands (spodosols).
(Pequia) were dried at 1 ooo C. for 24 hours and packaged in 2 mm nylon mesh bags stitched with nylon line.This genus was se-Jected for testing since it has species in or near both sites.Three hundred and twenty bags were prepared with one dry leaf per bag and were numbered.These bags were subjected to the following treatments: A. Black Water -Terrestrial, Spodosol 5 lfags each month for four months.Control (distilled water) Fung icide (10 g Benlate, 10 g Oithane irl 5 L water) Bacteriocide (5 g streptomycin sulfate in 5 L water) lnsecticide ( 100 m "Oetefon" ( Thus eighty samp les were set out on each of the four sites (A-O) in late August.Once a month for four months, five samples were brought in from each test (80 bags per month) monthly.The bags were placed in a refrigerator at 2° C. until they could be examined under a dissecting ncope for the numbers of Arthropods and fungai hyphae pr esent in ten fields (0.5 em diameter) on the lower side of each leaf.Plans to run bacterial counts on a portion of these leaves failed, although a few counts were taken.The leaves were then dried at 100° C. for 24 hours and weighed to determine actual weight loss on a percentage basis.One gram of the dried leaf material was extracted for total cellulose content while another 1 gram sample was ground and homogenized for the determination of Ca, Cu, Fe, K, Mg, Mn, Na, and Zn using a Techtron AA-5 atomic absorption spectrophotometer and standard ashing procedures at 525° C. for two hours .The weights of dried leaf material used for cellulose determination were determined by volume so that changes in cellulose content relative to volume could be made.The cellulose content and elemental content of the leaves was determined on dried leaves before placing them in the field and at monthly int ervais after they were placed in the field.
The packets which remained in the field were retreated with the same strength of reagents in September and October, although there is little reason to believe that those samples actually placed under water retained these chemicals for a long enough period to have any effect on decomposition.The insecticide may have been somewhat effective because o f the oi ly nature of "Detefon ".
By studying the changes over time between leaves on land and in the water, and between leaves on spodosols and oxisols in terms of cellulose, elemental content, Arthropods, percent as h and fungi, it was hoped that a clearer understanding of the dominant factors in decomposition might be recognized, and hopefully provide an explanation for the differences in origin between black and white water.
Associated with the decompos ition study were mea~urements of soil temperature at 3 em, water temperature at 3 m, oxygen content (Winkler Method) and pH of the water at 3-5 m, pH of the soil at 0-5 and 20-25 em, pH of rainfall and thrufall and pH and elemental content of I itter leachate.The content of IN NH40Ac extractable Ca.Cu, Fe, K, Mg, Mn, Na and Zn from 0-5 em and 20-25 em so'l samples was measured using the atomic absorption spectrophotometer for the cations.Total nitrogen was determined by the modified microkjdahl procedure.These data were needed to characterize the soil chemistry of the main terrestrial study sites, and to determine how different these areas actually were in soil chemistry.Some "typical" oxisol and spodesol sites were also analyzed to see how representative the study sites actually were, and to compare soil nutrient content to tree height.
Final report on studies ...

RESULTS AND DISCUSSrON
Litter decomposition -lnitial studies An immediate problem arose with the decomposition study because the harvester ants on the "terra firme" sites quickly carried off ali but the leaves treated with insecticide.The insecticide packets had a smaller mesh (1 mm), but not enough to stop the ants since they cut through the heavier, tougher plastic of the 2 mm mesh bags.Ants were also found on leaves submerged under water.The position of each packet also proved to be criticai.Those packets which were in contact with the thin humus layer decomposed more rapidly than those placed on top of freshly fallen litter.The leaves used are quite high in Ca, N, P and other elements.lt is possible that the ants can detect high nutrient leveis in the leaves, and so.selected these dry leaves.
Observations suggest that the litter requires a "softening period" of severa!weeks or more on the surface before they readily attacked by either fungi or litter animais.This allows time for moistening, bacterial action, and settling into the moist microclimate of the lower litter.Counts of animal teces as an indication of animal activity on "new" and "old" leaves from the forest floor of spodosols showed O. 24 feces/mm 2 on new freshly fallen leaves and 1. 09 feces/mm 2 on old leaves which were softened and partly decomposed.Oxisol new litter fali had O. 59 feces/mm 2 while old leaves had 7. 26 feces/ mm 2 suggesting greater animal activity on oxisol litter compared to spodosol litter.Litter extractions also showed 20.5 animais (aquatic) per 250 cm 3 from "igapó" litter, 150/ 250 cm 3 from spodosol "terra firme" litter, 28.0/ 250 cm 3 from "Campina" I itter, 90.4/ 250 cm 3 from oxisol new leaf litter, 261.8/ 250 cm 3 from oxisol old leaf litter, and only 3.0/ 250 cm 3 from "Varzea " litter.
Preliminary studies also included fungai counts as the number of fungai hyphae/mm 2 • The new leaves from the spodosols averaged O. 647 hyphae/mm 2 , while the old leaves averaged 0.981 hyphae/mm 2 • The "igapó" had on ly O. 25 hyphae/ mm 2 • Some of which could have been on the leaf when it fell into the water.Oxisol new leaf litter had O. 815 hyphae/mm2 while old litter had O. 923 hyphae/ mm 2 ; suggesting in most cases that older leaves are more readily attacked by fungí.
Another study showed an average pH of leaf surface scrapings from black water litter of 3. 45, while the corresponding terrestrial litter showed a pH of 5. 1 O. Spodosol ter-restri~l litter had a su rface pH of 4.85 while the aquatic litter, " Varzea", had 5.40.
The pH of surface leaf scrapings seemed important because of the large number of bacteria cultured from the surface of aquatic litter in some trial dilutions.The black water "igapó" I itter with a pH o f 3. 45 and an oxygen levei of from O. 22 to O mg/ 1 would defintely favor anaerobic, acidic bacteria while the higher pH range ot the "varzea" I itter (5.40) would favor a different bacterial flora.The abundance of bacteria in both aquatic litter sites strongly suggests a bacterial dom:nated decomposition pattern resulting in a high output of organic parti cu lates.Areas of the "varzea" which h ave white water during high water have been observed to turn to black water as the water goes down and side channels stagnate.Much of the ditferences between black and white water may be explained by the rate of flow of the w ater, its sed iment load and settling rate, the pH and oxygen content, what types of organisms are the dominant decomposers, and what their final metabolic end product is chem ically.

Decomposition of Caryocar villosum leaves
Table 1 shows the weight tosses as percent of the original dry weight for the leaves used in this study .Although only one species was used, it should give an indication of the general types of organisms which dominate decomposition on spodosol and oxisol terrestrial and aquatic sites.
The corttrol leaves showed rapid weight loss after one month to ants on the spodosol site (Fig. 1).The inundated leaves in t he "igapó" (b lack water) lost about 50% of their dry

54-
weight in the first two months, and then leveled off with little additional weight loss .This pattern suggests bacterial decomposition with little or no succession when the bacterial substrate (p .. esumably an organic) is used up, the decomposition slows because few other organisms are able to take over, once the organics are depleted."Igapó" leaves under water become thinner with little evidence of chewing or breaking.However, the season of heavy larva!infestations did not come until the end of the study.Mayfly and midge larvae may be very important at some times of the year.The build-up of Fe, Ca, Mn, P, K and Mg in "igapó" litter with time (Table 2) strong ly suggests that this is what occurs.Once the water recedes, the animal and fungai decomposition pattern can return using carbohydrate from softened new litter fali.H:gh N content in the litter would certainly encourage more rapid microbial colonization once the pH is raised and oxygen returns to the system.Contrai (untreated) litter on oxisols decomposed in about 2. 5 months with heavy litter animal activity and considerable root-leaf contact suggesting some "direct nutrient cycling" (Went and Stark, 1968).
Ali leaves placed in the white water "varzea" were taken by humans.
When bacteriostat treated leaves were exposed on the spodosol site, ants quickly destroyed these leaves which are rich in nutrients (Fig. 1).The "ig3pó" l;tter probably did not retain the bacteriostatic agent, and leaf decomposition closely resembled that of the contrais from the "igapó".lt is possible that the pH of the "igapó" bottom sediments becomes too low after months of stagnation and flooding so that the active bacteria may be inhibited or slowed by their own acid (pH 3. 45).Leaves placed on the oxisol site decomposed in about four months when treated with bacteriostatic agcnts suggesting that bacteria may be essential in the initial "softening" process on land, and to a lesser extent in I ater decomposition.
Leaves treated with fungic:de were also attractive to the ants and disappeared within one month after exposure.The leaves placed in the "igapó" decomposed rapidly for the first month and then leveled off in weight loss as c;id the controls.Fungicide treated leaves in the "igapó" lost slightly less weight (7%) than C:id the contrais suggesting a possible minar role of aquatic fungi in decomposition.The leaves placed on the oxisol site lost nearly ali their weight in four months when treated with fungicide (Table 1).The fungicide retarded weight loss by one month suggesting that fungi are relatively important decomposers on the oxisols.The slowed weight loss during the second month is the result of severe drying.Nutrient loss from ali •treatments on the oxisol site were minimal tending to change little from month to month which is characteristic of a decolllposer which "eats" whole parts and does not selectively choose only one substrate within the leaf.There are insufficient data on cellulose content for lnterpretation.
final report on studies ...
Figure 1 shows that decomposition on ali sites.aquatic and terrestrial.was gradual and nearly identical from month to month for insecticide treated leaves.The leaves did not disappear in four months time and had lost only between 50 and 75% of their original dry weight.The data and observations suggest that litter animais (mites.collembola, insect larvae) are extremely important on the oxisols and nearly as important on the spodosol terrestrial sites.In the water, midges and mayfly larvae were found indicating some importance in the aquatic habitat of aquatic insects.lt is too bad that bacterial counts could not have been made to establish the relative importance of aquatic insects and bacteria in decomposition under water.The abundance of bacterial slimes strongly suggests an aquatic decomposition pattern dominated by bacteria; with insects of lesser importance.
The leaves on the black water "terra firme " site lost sl ight ly more weight in the fourth month than did the other leaves.This difference could be attributed to control of fungai eating insects which allowed greater fungai growth .Data to date indicate that fungi are extremely important on the spodosols as decomposers, and particularly mycorrh izal fungi.

Elemental Content of Vegetation
Only one species of leaves, Caryocar villosum, was selected for study because of the complexity of using the natural mixed litter.lf mixed litter is used in decomposition studies, then it is essential to use exactly the same weight of each type of leaf, including the same weight of petiole versus blade .This handling of litter is extremely difficult and time-consuming .
Unfortunately, the harvester ants were attracted to the control leaf packets on the spodoso l site and they completely destroyed these samples after the first month.The rapid disappearance of litter as a resu lt of ants is an indication of the importance of these Arthropods to decomposition.They also destroyed the entire fungicide study w ;thin the first month.People removed ali of the samples placed in the water of the "varzea" si te within the first month.Drying on the "terra firme" sites (spodosol and oxisol) occurred in Sept ember because of the dry season.The data cannot be extrapolated to other sites or times of year .Control of funga i or bacterial growth has limited effect if the season is too dry to allow their growth.Full data for four months is ava ilable from the black water "igapó " innundated site.
Table 2 shows the P'g/ g of ten biologically important elements , percent ash, and total cations for the two study sites, three treatments, and innundated versus "terra firme" sites.
In general, Ca decreased slightly w ith time in the decaying leaves in the "igapó" site under inunllation (Fig. 4).lt should be remembered that these data are on a dry weight basis and the carbon content was continually changing resulting in periodic comparisons 56-which are not truly comparable.Magnesium showed no significant changes in the leaves with time and remained quite low (245-1800 p.g/ g. lron increased in the black water "igapó" leaves regardless of treatment (Fig. 4).The "igapó" si te may be considered as an inundated control decomposition test since the chemicals used in the treatments were quickly leached away.The aquatic samples were treated so that they would begin the study in the same condition as the terrestrial samples.Manganese increased considerably in t he leaves in the " igapó" site as did iron.These two elements are not read ily used by organisms in high concentrations , so they are probabl y se lected aga inst by the decomposers.The least change of ali treatments was in the insecticide treated samples suggesting that the chemical used limited ali forms of decomposition.This chemical is oily and tends to stick to surfaces.Copper and zinc changed slightly in the leaves over the four months with a slight tendency to increase during the fourth month (Table 2).
The total nitrogen content of the leaves in ali surviving tests tended to increase by the second month and then decreased through the fourth month (Fig. 5).The absolute leveis of nitrogen probably reflect the nitrogen of the decomposers and the ir waste products , so that it is impossible to separate completely the protoplasm of the decomposer from that of the substrate.Nitrogen leveis were high in the litter ranging from 28,900 to 49,140 p.g/ g.Litter with these leve is of nitrogen are ideal sources of nitrogen needed for plant grow th which suggests that direct nutrient cycling can be of great value to growing vegetation.Tne insecticide treated samples had the lowest leveis of nitrogen suggesting that the animal populations normally enrich the litter considerably with t heir feces.The insecticide treated leaves had leveis of nitrogen somewhat lower than that of the controls.
Potassium remained low and stable in the litter samples regardless of treatment (290-1000 p.g/g, Table 2).

FUNGlCIDE
cide test (3110 to 5370 1-'g/g).Oxisol ("terra firme") and spodosol ("igapó") sites showed the greatest difference in litter phosphon.s in the bacteriostat test (Fig. 5).Midge larvae were seen to be quite active on the "igapó"  litter.These organisms may contribute some phosphorus to the decaying leaves.The treatment did not eliminate bacteria which are important in the aquatic decomposition sites and are also high in phosphorus.Data on the percent ash content (Fg.2) tended to increase with time for ali treatments.The ash content of the contrais was 1 .8 to 12.62% (Table 2, Fig. 2).The ash content of the bacteriostat tests on the oxisols increased BACTERIOSTAT ... ~-----------x------------
.... Fe :. tent.The as h content of leaves treated with insecticides changed from 2. 70 to 14.94% on the spodosol, from 3 .20 ( decreased to 2. 72 in the second month, then i ncreased to 7. 16%, and from 3. 92% for the black water " igapó n site to 13.14%. The initial levei of ash in the untreated, unexposed leaves was 3. 31 to 3. 36% (Table 2), so that many tests showed a slight decline in as h content during the first month .This decrease in ash content is not easy to explain without cellulose data.In nearly ali cases, the individual elements increased in concentration on a dry weight basis (1-"9/ g) after one month 's exposure.This is probably the result of remova ! of cellulose and hydrocarbons leaving relatively more of each element behind.
Total cations were calculated in rg/g for the eight cations measured as an indication of the changes brought about by the treatments and exposure time (Table 2).The total cation content varied from 8355 to 28,772 ,u.g/g.The controls which were untreated had 8,418 to 10,876 p.g/g whi le the unexposed and untreated leaves had 6628 to 6723 ,u.g/g of total cations at the start of the tests.The total cation content increased most drastically in the " igapó" sites regardless of treatment (Fig. 3).The total cations changed little on samples placed on oxisols and those on terra firme spodosols.Decomposition on the "igapó" sites appears to be selective removing some materiais and not others as is the case with bacterial and some types of fungai decay where a specific substrate is required.When the insects were not controlled on terra firme sites, the decomposition appeared to be less selective with the disappearance of whole leaf segments and without a selective concentration of total of specific cations.The oxisol sites appear to be strongly dependent on litter animais for decomposition with a lesser dependence of spodoso l litter on animal decomposers.litter from the bottom of black water streams had high ash contents (28.26 to 28.48%) and low total cations (2544 to 2753 60-,u.g/g,Tab le 2).Those data suggest that some element not measured in the eight cations is present in great quantities in the black water areas.Silicon is one element which may be very high in the decomposing "igapó" litter, but was not measured here.Similar discrepancies exist between other sites where total cations and percent ash were measured, probably for the same reason.In addition, these plants were digested by dry ashing at 525° C. which means that e lements such as sílica will not go into solution, while others such as nitrogen will volatize during ashing .For this reason, total nitrogen was determined from a wet digestion of O_ 1 g samples by the microkjeldahl procedure.Percent as h and total cations cannot be expected to agree under these circumstances.
Other "igapó" sites showed 10,394 to 10,562 ,u.g/ g of total cations, and 16.39 to 17.53% ash (Table 2).Campina Creek litter (inundated) was low in both percent ash (7.58-7.62%) and total cations (1434-1457 ,,gfg) which is a direct reflection of the low e I ementa I content of "campina" vegetation and soil in general.
Litter from oxisol soils showed 9,264 to 9,402 ,u.g/g total cations and 9.13 to 9.18 % ash, a reflection of a slightly richer site, but one which is poor from severe leaching (Table 2)."Varzea" litter had very high total cations (13,134 to 13,786 ,u.g/ g) and high percent ash (68.23 to 68. 7%) indicating a nutrient rich river bottom litter capable of re leasing large amounts of cations and anions, and enriching t he soil when the waters subsided (Table 2).The "varzea" litter was noticeably low in nitrogen (5,740 ,u.g/g) compared to 14,000 ,u.g/ g for "igapó" litter.High leveis of Mg, Cu, Fe, K and Zn were found in the "varzea" litter (inundated).lt is interesting that the soil also showed relatively high leveis of Mn, Mg, K, and Zn, but low leve is of Fe.The balance of Fe in the decomposing litter cou ld well influence what kinds of microorganisms can and will c. -- In general, the leveis of elements in the "terra firme" litter agrea w ith those from a former st udy on spodosols (Stark, 1971).The Caryocar leaves represent on ly one species wh ile i.he litt er analyzed in 1971 represents a conglomerate of many species w it h generally lower Cu, Fe, M n, Zn, Ca, Mg , N, and P.
The elementa l analyses of litter from land and water f rom white and black water areas has shown some distinctive d.fference in elemental content and decomposition pattern.Unfortu nately, the data on bacteria l act ivity in t hese sites could not be detc mnined.Studies of microb ial ATP would provide the by information needed to interpret these data.
Figs. 6-10 show the appearance of leaves of Caryocar after different treatments and exposure ti me in the field.Unfortunate!y, there are no photographs of the contrais.Leaves treated wit h fungicide show considerable frag mentation after fou r months in the black water "igapó •• but t he leaves were noticeably thi n.There is no reason to be lieve that the t reatm ent persisted.Leaves treated with fungicide on "terra firme" (oxisol) showed evidence of chewing and fragmentation after tw o months (Fig. 6).
Leaves treated w ith bact eriostatic agent and placed on " terra firme" (spodosol) were near ly completely decomposed after three months and were heavily infi ltrated with roots and fungi (Fig. 7).lt is probable that contra i of bacteria allowed rapid fungai growth .
Leaves treated with bacteriostatic agent and placed in black water did not show a strong effect of the chemical because of leaching, but t he leaves were fragmented on the edges (Fig s .7, 8).After four months 111 black w ater, t he ..l eaves were heavily fragmented.Bacteri ostat treated I e aves on oxisol " terra firme" showed prog ressive fragmentation, insect damage, and root actitivy (Fig. 8) .

-
When insectic ide was used, the "terra firme" spodosol leaves did not show extensive insect damage.lnsects were not prom:nent during t he montbs of this study, and the leaves pl aced in black water and treated with insecticide showed l ittle decomposition and almost no insect damage (Fig. 9).Leaves placed on oxi sol "terra fi rme " w ere heavi ly fragmented after four month's exposure.Roots and fungi appear to have baen important during the fourth month (Fi g. 1 O).but t here w as little evidence of insect damage.

Elemental Content of Soils
Table 3 shows the elemental content of various spodosols and ox isols in p.g/ g .The first t en soils are associated with black water rivers and are extremely low in Ca , K, Mg , Mn , N and Zn compared to "varzea" (flooded white areas) sites or oxiso ls.Some exceptions are the nitrogen leveis of surface sai ls on lowand spodosols, and the Mg leveis of the " Campina" forest .Areas which are rich in calc ium are t he "varzea inundated muck ", and "varzea " bank sediments wh ich were not recently flooded (Table 3).In general, the oxisol soils were quite low in Ca (10-26 p.g/g ) even when earth worms were present.Black water "igapó" [drowned forest) had 26 p.g ca/g at the surface, and on ly 15 p.g ca/ g at 20-25 em depth.Spodosols in general ranged f rom 9. 5 to 26 f .A.Q/ g for IN NH40AC extractable Ca .These leveis of Ca are the same to slightly lower than those report ed from Brazil spodosols in 1971 [St ark, 1971).
Copper in the spodosols ranged from O. 8 to 2. 5 JJ.g/ g (Table 3) which is about th e same as t hose values from an earlier study from the same general areas (St ark, 1971).The best soils studied ("varzea") showed only 3 .0 p.g/ g for Cu ind icating generall y widespread low leveis for this element.
Extractab le iron content ranged from O. 8 to 22 .5 f .A.Q/ g for the spodosols , with not over Final report on s tudies .. .31 p.g/g from the best "varzea" sites (Table 3).The leve is of iron on spodosols are comparable to those from earlier studies .Extractable iron on spodosols is comparab le to that obtained from earlier studies.Extractable iron is generally sufficient from al i areas studied except the "campina" forest on sandstone and the black sand spodosols.
Potassium was generally low from the spodosol areas ranging from 5. 5 to 74 p.g/g.The highest values were from the "campina" forest (74 p.g/g) and the "varzea" sites (85 p.g/ g, Table 3).In 1971, values for K f rom this general area ranged from 12 to 320 p.g/g (Stark, 1971).In both stud ies, lowland spodosols which periodically flood had the highest leve is of K.
Nitrogen leveis were variable as can be expected for surface soi ls (Table 3).lnundadated sites on the black water "igapó" had only 252 ~-'-g /g for nitrogen whereas surface samples from lowland spodoso ls showed 3164 p.gN/ g of soil.The oxisol surface soi ls did not show significantly higher leveis of nitrogen over the spodoso!"s, except for the "terra firme" pri mary forest with trees to 50 m height.Oxisols generally had higher leveis of nitrogen at 20-25 em depth.but most sites had acceptable leveis of nitrogen by temperate zone standards.
Because of heavy leaching and the high solubility of sodium salts, this element usually does not tell a great deal about tropical soils.The inundated "igapó" site had 7 p.g/ g while the inundat~d varzea soils had 28-31 ,~LgNa/ g of soil , and the dry site (not recently flooded) "varzea" sites had 37 ,u.g/g for this element (Table 3).Campina Creek mud had the highest 66-Na (4 1 .31'-9/g).except for lowland spodosol sites (62 ,~Lg/g).These data suggest that some of t he inundated sites are co llecting or concentrating Na whi le others are losing this element.The leveis of Na found here are comparab le to those reported earlier (Stark , 1971).
Zinc was low (0 . 5 to 1.1 p.g/g, Table 3) on nearly ali soils except the "varzea" (2 .3,u.g/ g).lf the surface soils are arranged in the order of total extractable cations from lowest to highest (Tab le 4), the spodosols (w:th the exception of one agricultura!oxisol) ali f ali in the lower tota l extractable cation range (44 .8-130.6 ,u.g/ g. These are exceptiona lly low leveis of extractab le cations.A coniferous forest soi l wil l range from 2975 , u.g/ g of extractable cations in the surface soi I.Where tree he ight data are ava;lable, the average height tends to increase with the higher leve is of extractab le cations, except for the " camp ina" sites.lhe vegetation on the " campina" sites is low (8-1 O m), and the extractable cations are also low (80.2-130.6 1'-9/g).The depth of undecomposed litter in the "campina" is generally low (under 2 em), indicating possibly lower production and greater dependence on the soil nutrients, or less direct nutrient cycl ing from organic litter t o living roots.This is consistent with the dry nature of the campina.The highest total extractable cations from surface soi I come from the "varzea" which was not flooded in August (2290 .9 p.g/g, Tab le 3).
The total extractable cation content of the subsurface soils is generally low (35.3-128.5 JLQ! g, Table 3).Where tree height is known , t he height correlates more closely with the total extractable cations from the subsurface soi l than from the surface so il.The "campina" subsoils appear to be higher in total extractable cations than many of the other sites , inc!uding the "varzea", but average tree height is extremely low (under 1 O m) suggesting that some other factor is limiting to growth, possibly t he anions N02 -or P04 -• Nitrate is low for soil 28 on sandstone (392-630 1' -9/ g, Table 3).Organic litter and tree height - -----------------------------------------------do not appear to be closely related from these limited data, but more information is definit ely needed.For surface soils, the deeper litter tends to be on the poorest soils suggesting a strong dependence on litter.These soils have relatively small forests of 20-30 m high.
The pH of surface spodosols averaged 3 • 65 with 3. 98 at 20-25 em whi I e the pH of the "igapó" was 5. 45 and 5. 35 respectively, and the oxisol sites were 3. 84 and 4. 28 respectively.The "varzea" pH was 6. 58.Sai ls with earth worms (3.8) were generally in the middle range of extractable cations for the surface and high in extractable cations in the subsoil.
Although this study was not able to sample adequately to fully describe the re lations of vegetat:on to soils and nutrient cycling, the data do indicate some interesting trends.Future studies should include tree height, biomass per hectare, extractable cations, anions, organic littAr depth, root mat (depth, and location), and studies of direct and indirect nutrient cycling to fully describe nutrient cycling differences from spodosols and oxisols.
The spodosol terra firme litter appears to be decomposed mainly by fungi (possibly mycorrhiza l fungi) and litter anima is.This litter also requires a softening period before rapid decay begins.Once the leaves are properly softened, decay is rapid and complete within a month or so.Ants are extremely important decomposers on spodosols.
The richest soi/s are the "varzea" soils as is well known.There is some relationship between tree height and fertility, but litter depth complicates this relationship .
Caryocar leaves changed in elemental content and percent ash as it decays tending to increase in concentration with time.These leaves are rich in N, P, and Ca."Varzea" litter is one of the richest types of natural litter types in Amazonia, except in nitrogen.The "Campina" litter is the poorest of forest litters in nutrient content.

CONCLUSIONS
This brief study cannot hope to have amassed enough data to explain ali of the differences in nutrient cycling on black and white water soils.lt h as been shown that a factor of prime importance is the rate of decomposition and the dominant decomposer groups and their waste products.Studies on the nature of the leaching solution which falls through the forest canopy and contacts the litter on the forest floor were cancelled because of time.Such studies should !)econtinued to fully explain nutrient cycling since it is probably the pH and chemistry of the leaves and their microenvironments which finally determines which organisms will dominate the decomposition of litter, and in what sequence.Nutrient cycling appears to be strongly related to type and rate of decomposition, and less to soil types, although white water depends on a clay to provide the sediment load.Decomposition occurs in an acid, anaerobic environment which after a time may stagnate further decay.The decay end' products are particulate organics which give the water its calor and these can form under water or either oxisols or spodosols if the aeration, pH and currents are suitable.The pH of the water coming from the land may also be important to determining the dom:nant decomposers.The build-up of cations in this litter suggests selective decomposition by bacteria.
Oxisol terrestrial sites appear to be dominated by litter animais which require a period of softening of the leaves by moisture and bacteria before they beco me abundant.Mycorrhizal fungi and certainly free-living fungi are also important decomposers on oxisols.
The nutrient loss pattern suggests decomposers that "eat" chunks of leaf rather than selective decay of one fraction of the leaf.This "eating" reduces ali elements proportionately.

Fig. 9 .Fig
Fig. 9 .Nos. 265, 270.Leaves of Caryocar villosum treated with insecticide and left in the black water of the "igapó" for one and two m.onths r espectively .N°s .285, 299 leaves treated simílarly, but exposed to oxisol "terra firme" for one and four months respectively.
Figure 4. Changes in the average Ca, Mg, and Fe content in Caryocar villosum leaves with treatment and time.Ca t-.. -t Spodosol Mg --.. --Fe A-.. -A Final report on studies ...

Table 2 .
Average elemental content, total cations, and percent ash of Caryocar villosum leaves after different lengths ot exposure on terrestrial and aquatic sites associated with oxisols and spodosols.

Table 3 .
Leveis ot IN NHPAc extractable and total cations found in spodosols and oxisols from Brazil, and total nitrogen.