A comparison of the forest soils in the Peruvian Amazon : Terra firme , palm , white sand and igapó

The Amazonian rainforest is critical to our shared human future. Therefore, we need to understand the structure and function of their forests and soils are a critical part of that investigation. Towards that end, soil samples were taken in eight different forests, both non-flooded (terra firme-low forests, terra firme-high forests, white sand-varillal forests, white sand-chamizal forests and palm forests) and flooded by black-water (high restinga, low restinga and tahaumpa forests), at two locations in Loreto Province, Peru and analyzed for soil pH, soil organic matter, and soil nitrogen (N), phosphorus (P), and potassium (K). Results showed that: (1) soil pH of the non-flooded forests was very similar to flooded forests, but flooded forests became more basic as flooding increased in duration; (2) soil organic matter was lowest in the two non-flooded terra firme forests and increased as flooding increased in duration; (3) N was lowest in the palm forest, P was lowest in terra firme-low terrace forest and K was lowest in the terra firme-high terrace forest; (4) N decreased sharply as flooding duration increased, both P and K increased while (5) for some non-flooded forests there was a correspondence between soil fertility and floristic similarity. In conclusion, while flooding decreases pH and N, it increases soil organic matter, P and K.


INTRODUCTION
The structure and function of tropic ecosystems is critical to global issues such as biodiversity and climate change (Keller et al., 2001) and thus to our shared human future.A more complete understanding of tropical soils, for example, may clarify their influence on biogeochemical cycles and improve predictions of current and future tropical vegetation.Vegetation and soil interact in various ways, where plants create the soil and the soil influences the plants (Sanchez 1976, Wambeke, 1992).Particularly in the tropics, plant species composition of rainforests alters soil characteristics (e.g., tropical legumes affecting soil nitrogen (N) levels: Rhodes et al., 1998), and the ranges and boundaries of plant communities correlate with levels and forms of soil nutrients (Walter, 1973).This may be especially true in the environmentally and socially important Western Amazon where topographic relief is low and other environmental drivers, such as temperature and precipitation, may have less variation compared to other tropical areas (Myster, 2009).Indeed, studies have found strong correspondences between soil characteristics and both non-flooded (Honorio, 2006) and flooded (Junk, 1997) forest composition in Western Amazonia.
Soils in the Amazon are largely Ferralsols and Acrisols but Plinthosols, Gleysols, Cambisols, Leptosols, Arenosols, Fluvisols, Regosols, Lixisols, Podzols, Alisols, Histosols and Nitisols are common (Quesada et al., 2011).Within the Western Amazon, non-flooded forests have acidic, clayey-loamy shallow soils rich in superficial organic matter, which they share with non-flooded rainforests found throughout the rest of the Neotropics (Pitman et al., 2001).The flooded forests organize by water quality and along a flooding duration and maximum water height gradient (Whittaker, 1975) which strongly affects their soils and the distribution and abundance of their plant species (Junk, 1997).For example, the blackwater floodplains (creating igapó forests) investigated here have more acidic soils than those of the non-flooded forests with less concentrations of phosphorus (P) and potassium (K) exchangeable cations (Honorio, 2006).Flooding can also reduce N availability, which is relatively available in non-flooded tropical forests (Neill et al., 1999).
The objective of this study is to provide a better understanding of Amazon soils, and the difference between non-flooded and flooded forests.Towards that aim, soils in five non-flooded forests and in three flooded forests of the Western Amazon were sampled and the following three hypotheses were tested: 1. Terra firme forest soils will be higher in organic matter and NPK than soils in palm forest and in white sand forest, but soil pH will be similar among these nonflooded forest types (Honorio, 2006;Carmo and Cerri, 2007;Koutika et al., 1999).2. Black-water (igapó) forests will increase in pH and organic matter as duration of flooding increases but NPK will decrease (Quesada et al., 2011;Forsberg, 1984;Koschorreck and Darwich, 2003).3. Non-flooded forests will have greater pH, organic matter and NPK compared to flooded igapó forests (Quesada et al., 2011).

Study sites
The first study site was the Area de Conservación Regional Comunal de Tamshiyacu-Tahuayo (ACRCTT) (www.perujungle.com;Gottdenker and Bodmer, 1998) located in Loreto Province, 80 miles southeast of Iquitos, Peru (~2°S, 75°W) at an elevation of 106 m.The reserve is part of one of the largest (270, 654 ha) protected area in the Amazon.ACRCTT is classified as a wet lowland tropical rainforest of high diversity (Daly and Prance, 1989).It is comprised of low, seasonally inundated river basins of the upper Amazon and named after two of the major rivers (the Tahuayo and the Tamshiyacu) which form boundaries to the north and west and create large fringing floodplains (varzéa; Junk, 1997).The substrate of these forests is composed of alluvial and fluvial Holocene sediments from the eastern slopes of the Myster 131 Andes.Annual precipitation ranges from 2400 to 3000 mm per year, and the average temperature is relatively steady at 26°C.The rainforest itself has distinct communities defined by their flooding regime with the rainy season between November and April (Kalliola et al., 1991).These include black-water (igapó) forest which floods 1 to 3 months per year (high restinga), black-water forest which floods 4 to 6 months of the year (low restinga), and black-water forest which floods at least 6 months of the year (tahuampa).Maximum level of flooding is correlated with flooding duration (author's personal observation).Black-water is local forest runoff containing humic substances, such as tannins, leached from litter, making the water acidic and low in nutrients (Myster, 2009).
The second study site was the Centro de Investigacion de Jenaro Herrera (CIJH (Myster, 2009(Myster, , 2015) ) operated by the Instituto de Investigaciónes de la Amazónia Peruana (www.iiap.org)located 2.5 km from the town of Jenaro Herrera, 200 km south of Iquitos on the east margin of the Ucayali River in Loreto Province, Peru (4° 54' S, 73° 40' W) (Honorio, 2006).The mean annual temperature is 26.0°C with a range between 25.1 and 26.5°C.The mean annual rainfall is 2724 mm with two dry seasons: The more severe between June and September and the less severe between December and March.Here the non-flooded forests have soils that are highly weathered, acidic, nutrient-poor, and clayey-loamy (oxisols) (Spichiger et al., 1996).Dominating the non-flooded portions of CIJH are low terrace broad leaf forest (terra firme-low) and high terrace broad leaf forest (terra firme-high) with scattered patches of white sand-varillal (with dense growth and trees up to 20 m tall), white sand-chamizal (with shrubs and scattered trees less than 8 m tall), and low terrace palm (dominated by palms such as Mauritia flexuosa, Oenocarpus bataua, Euterpe precatoria and Socratea exorrhiza).Common genera in the high and low terrace broad leaf terra firme forests include Eshweilera, Pouteria, Oenocarpus, Miconia, and Protium, in white sand-varillal Pachira, Haploclathra, and Macrolobium, and in white sand-chamizal Caraipa, Pachira, Macrolobium, Calophyllum, Haploclathra, and Platycarpum (Honorio, 2006).Terra firme-high forest is located adjacent to rivers (Pires and Prance, 1985) but is rarely flooded.Terra firme-low is not located on ridges adjacent to rivers and has clay soils.Palm swamps (Montufar and Pintaud, 2006) occur in depressions or low-lying patches with poor drainage.White sand soils consist mainly of quartz with individual trees having slender boles and roots concentrated at the soil surface (Klinge et al., 1990).These forests are divided into varillal (dense with straight and thin trees between 10-20 m high) and chamizal (shrubs 3 m high and scattered trees <8 m high) (Honorio, 2006).The two study sites are located close together and this, along with the low relief of the Amazon and the small seasonal and yearly variation in basic environmental conditions, facilitates comparison of its soils.

Soil sampling
In May 2009, a 100 m transect was established in each of the 16 study forests found at the two study sites ACRCTT (two high restinga forests, two low restinga forests, two tahaumpa forests) and CIJH (two terra firme-low forests, two terra firme-high forests, two white sand-varillal forests, two white sand-chamizal forests, two palm forests).Along each transect five microsites were selected randomly among the ten of 10 m intervals apart.A soil sample was taken at each microsite by inserting a standard soil-sampling probe Forests are underwater more months per year as you read High restinga to Tahuampa.
of 1 cm diameter into the soil to a depth of 10 cm, removing the soil without the litter and reinserting the probe until the standard soil sample bag was filled.The soil was room dried, stored in a cool place, and taken to the Soil, Water, and Forage Analytical Laboratory (SWFAL) (www.soiltesting.okstate.edu)on the campus of Oklahoma State University for analysis.Soil samples were first dried at 60°C overnight and ground to pass a 2 mm sieve.Then soil organic matter (SOM) and total nitrogen (N) were determined using a LECO Truspec dry combustion carbon analyzer (Nelson and Sommers, 1996).Soil pH was measured by glass electrode in a 1:1 soil: water suspension and Sikora buffer solution, respectively (Sims, 1996;Sikora, 2006).Finally, plant available phosphorus (P) and potassium (K) were extracted using Mehlich 3 solution (Mehlich, 1984).Then P and K in the extract were quantified by a Spectro CirOs ICP spectrometer (Soltanpour et al., 1996;personal communication by Dr. Hailing Zhang, Director SWFAL).All three soil nutrients (N, P, K) are expressed in parts per million or ppm and 80 samples were analyzed.Means and standard errors for each soil chemical parameter and forest-type were determined.Data was first graphed to ensure normal distribution.Data was then used for one-way analysis of variance (ANOVA) (SAS, 1985) tests for each soil parameter between the non-flooded forests (ten forests with 50 total samples) and the flooded forests (six forests with 30 total samples).

RESULTS
Soil of the non-flooded forests was acidic and their pH values were very similar, ranging only between means of 3.4 and 4.25 (Table 1).Soil organic matter (SOM) however, varied widely among these forests with terra firme forests having the lowest, ranging from a mean of 6.22 to a mean of 8.89, and the other forests having a range from a mean of 46.65 to a mean of 73.40 (Table 1).N, P and K also varied greatly among these forest-types.N was lowest in the palm forest (mean = 80.5 ppm) and in the white sand forests (72.0 and 71.0 ppm), P was lowest in the terra firme forests (18.0 and 23.5 ppm) and K was also lowest in the terra firme forests (91.5 and 63.5 ppm) (Table 1).
The soil of the flooded forests was also acidic but became more basic with increased flooding.The SOM also increased with flooding (Table 2).Both P and K increased with length of flooding period, but N decreased sharply (Table 2).Flooded forest showed a trend of being more basic than the terra firme non-flooded forests (Table 3).The SOM was significantly greater (five times as much) in the flooded forests.Nitrogen decreased significantly as flooding increased in flooded forests, but P and K increased significantly (Table 3).

DISCUSSION
All three hypotheses were generally accepted.Nonflooded forests varied much less in pH than in the other soil parameters where the most common terra firme forests were lowest in SOM, P and K, and palm was lowest in N. The range of variation was less within the two terra firme forests compared to the range within the two white sand forests, which may indicate that vegetation has a controlling effect on nutrient availability in those forests.The variation of all soil parameters for the flooded forests corresponded well to the flooding gradient where pH, SOM, P and K all increased with a longer inundation period, but N decreased.Comparing the two most common non-flooded forests (terra firmelow and terra firme-high) with the flooded forests showed that most trends were significant: SOM, P, and K One-way analysis of variance results are given as an F statistic and its corresponding P-value.*Different levels of significant difference: "*", significant at 0.05 < P < 0.01 and "**"significant 0.01 < P < 0.001.
increased with flooded while N decreased.Soil pH and SOM levels were within ranges sampled at other Amazon forests both flooded and non-flooded (Neill et al., 1999(Neill et al., , 1997a, b;, b;Koutika et al., 1999;Garcia-Montiel, 2000;Almeida et al., 2005).The decline in N levels with flooding was also seen in previous study in the Amazon (Koutika et al., 1999).In non-flooded forests there may be a similar amount of ammonium as nitrate in the soil (Neill et al., 1995), however, there is more ammonium in flooded forests (Koschorreck and Darwich, 2002) as lack of oxygen reduces bacterial growth and the N mineralization (Myster and Schaefer, 2003;Li et al., 2005) which produces the nitrate sampled here.Phosphorus and K levels were similar to other sampling (Garcia-Montiel et al., 2000) and increased P and K concentrations with flooding have been widely reported (Vitousek and Sanford, 1986).
This was a sampling study.It cannot demonstrate causality.Results, however, suggest the effects of a flooding gradient, determined by both the length of inundation and the maximum water level attained, analogous to the elevation gradient of decreasing organic matter and nitrogen extending from the Andes to the Amazon (Townsend-Small and McClain, 2005).The nonflooded forests, however, benefit from improved physical and chemical soil properties and more efficient biological cycling of C and N.This may be important because these soils are acidic and unfavorable for nitrification (Koschorreck and Darwich, 2003).In the flooded forests, N is also an issue and can be the most limiting nutrient (Forsberg, 1984) because nitrification (converting ammonium to more mobile nitrate) depends on oxygen (Carmo and Cerri, 2007).While P is the primary limiting nutrient to plant production in the humid tropics (Vitousek and Sanford, 1986); results suggest that N may also be important in forests that are extensively flooded.Results also suggested, in some cases, a correspondence between the vegetation of terra firme forest and soil characteristics.For example, Tuomisto et al. (2002) found a negative correlation between extractable soil cations and both plant density and species richness and Ruokolainen et al. (2007) reported that soil differences explained 50% of floristic differences among terra firme plots while geographic distances explained only 16%.

Conclusion
There was substantial variation (1) in the non-flooded forests for all parameters except soil pH and there was some evidence that floristically similar forests were also similar in soil parameters; (2) among the flooded forests for all parameters except soil pH, nutrients correlated with the flooding gradient, that is all nutrients increased with increased flooding except N which decreased, and (3) flooded forest were more basic than the non-flooded forests; SOM, P and K were significantly greater in the flooded forests, and N decreased significantly when the forest was flooded.In general flooding seems to have a greater impact on Amazon soils compared to the variation found among non-flooded forests.

Table 1 .
Mean ± standard error (n = 10 for each forest type) of soil parameters for each of the five common non-flooded forest types.

Table 2 .
Mean ± standard error (n = 10 for each forest type) of soil parameters for each of the three common flooded forest types.

Table 3 .
Mean ± standard error of soil parameters for the five-pooled non-flooded forests vs. the three-pooled flooded forests.