The Effects of Landslide Disturbance on Seasonal Cycles of Allochthonous Inputs to Peruvian Amazonian Headwater Streams

Primary Support from the National Science Foundation and the Andrew W. Mellon Foundation

Return to Naiman Group Homepage

This project is a component of the Andean Amazon Rivers Analysis and Management Research Project (AARAM) cooridinated through Florida International University. AARAM research activities examine the fundamental physical and biological processes controlling the health and dynamics of Andean riverine ecosystems. Emphasis is placed on terrestrial-aquatic and upstream-downstream linkages, with scales ranging from meso- to micro-scale watersheds.

Specific research objectives of AARAM are to

Effect of Disturbance on Allochthonous Input
Efectos de los deslizamientos desde las entradas alóctonas hasta las cabeceras de las quebradas amazónicas (click to view Spanish)

Ann J. Glauber, School of Fisheries, University of Washington

ABSTRACT
We evaluate whether streams adjacent to mature riparian forests experience temporally more stable inputs of litter and nutrients than streams in disturbed forests. Litterfall from three sites (5-years post-landslide, 10-years post-landslide, and mature {"Sites 5Y, 10Y, and M, respectively}) was collected bimonthly to determine mass inputs of total litterfall and its' fractions (plant families, twigs, reproductive structures, unidentified, epiphytes, and frass). Given strong gradients for soil, vegetation and nutrients along landslide scars, data from upper and lower portions of scars were evaluated separately. Nutrient concentrations were determined for each site's largest fractions. We present three months of data suggesting total nutrient inputs, along with component fractions, vary predictably across an ecosystem development gradient. Litterfall input rates were significantly different between upper and lower zones of the youngest landslide. Total litterfall inputs collected between 11/30/98 and 3/14/99 were 71.12 and 124.67 g/m2 at Site 5Y's upper and lower zones, 91.90 and 93.42 at Site 10Y, and 214.24 at Site M. Additionally, nutrient concentrations were found to vary between fractions.

INTRODUCTION
Riparian processes in small Amazonian headwater streams clearly impact nitrogen and organic matter fluxes from land to adjacent aquatic systems, exhibiting disproportionate control of downstream reaches (McClain et al. 1997, 1996, 1994). These findings support substantial research in temperate forests that has underscored the importance of headwater riparian forests and streams in regulating downstream nutrient fluxes (Bormann and Likens 1979, Haycock et al. 1997, Pinay et al. 1990). Headwater streams are small, but numerous, generally comprising the vast majority of stream length within any given river basin. Hence, these streams represent the largest connection between forests soils and the streams that traverse them, removing exported material and nutrients from upper reaches and delivering them to lower aquatic and riparian systems.

While interactions between headwaters and downstream reaches are important ecologically, they also have widespread implications for stream and forest resource management. Nutrient cycling studies provide information on inter- and intra-ecosystem material fluxes, and data describing nutrient cycling can be used to determine the limiting factors in ecosystem production or development, which are important in understanding the potential effects of landuse changes or disturbance within an ecosystem and its surrounding watershed.

Many Amazonian headwater streams are found within the tropical montane forests that surround much of the Amazon Basin (McClain et al. 1996). Tropical montane forests are known to be less productive (Tanner 1980) and have lower nutrient cycling rates than their lowland counterparts (Vitousek and Stanford 1986, Grubb 1977). These differences have been attributed to altitudinal gradients in temperature and radiation, which in turn affect climate, soil composition and pH (Edwards 1977, Marrs et al. 1988, Tanner et al. 1998), species composition, rates of photosynthesis and litterfall, as well as decomposition (Edwards 1977, Vitousek and Stanford 1986). Steep-sloped montane forests are also subject to significant natural disturbances, including landslides (Garwood et al. 1979) which often represent a major material transfer (Swanson et al. 1982). Lowered rates of cycling within tropical montane forests would suggest a delayed post-disturbance recovery from perturbations such as landslides.

Aboveground litterfall often represents the largest nutrient flux between vegetation and the soil surface within a forest ecosystem (Herbohn and Congdon 1998), and fragments of riparian vegetation entering streams are an important link between forest and stream systems. Allochthonous material, comprised of these particulate inputs plus dissolved groundwater inputs, represents a major nutrient input to aquatic ecosystems. The timing and composition of these allochthonous inputs have important consequences for adjacent and downstream stream communities.

Our objective is to evaluate potential variations in aboveground litterfall and particulate nutrient inputs to riparian forests and headwaters streams along a successional gradient in the upper Rio Pachitea Basin. This study uses an ecosystem development gradient as a "space-for-time substitution" to evaluate differences in nutrient inputs, and is providing data describing particulate allochthonous inputs across different forest successional stages. Additionally, it extends the range of data on particulate nutrient inputs and litterfall seasonality to the eastern Peruvian Andes. The specific objectives of this study are to determine: (1) the mass balance of aboveground litterfall to riparian soils and headwater streams in an Andean Amazonian lower montane forest and whether it varies seasonally or across forest successional stage; (2) the taxonomic composition (to family) associated with aboveground litterfall; and, (3) the nutrient budget to riparian soils and headwater streams associated with aboveground litterfall, and whether this is affected by season or forest successional stage. At this time we will present data for the first three months of this year-long study.

While this study focuses on alterations in particulate allochthonous inputs following landslides, a natural disturbance, the findings are likely to suggest mechanisms by which ecosystems recover following similar anthropogenic disturbances. As part of the Andean Amazonian Rivers Analyses and Monitoring (AARAM) Pachitea Project, this study investigates spatial and temporal variability in biogeochemical cycles within Peru's Rio Pachitea basin, an Amazonian tributary.

SITE DESCRIPTION:

The study is being conducted within and on the border of Peru's Yanachaga-Chemillen National Park, located north of Oxapampa, Departamento de Pasco, in the eastern Andes. All sites are within headwater streams of the Rio Pachitea, an Amazon tributary (Amazaon Basin Map), upstream of deforested or farmed areas typical of the region. Study sites are located at approximately 2400 m elevation in lower montane tropical forest adjacent to low-order headwater streams, all tributaries to Quebrada El Cedro. No studies have been conducted within the immediate project vicinity to date, but previous studies in the Park have described a system characterized by rich biodiversity of woody plant species with a significant epiphytic component (Foster 1981).

High-gradient forested slopes subject to frequent landslide disturbance characterize the site. Three sites have been selected to represent a natural successional gradient: two landslide-affected sites of differing age (approximately 5- and 10-years post-landslide), and one in mature riparian forest. Each site is along a tributary of a single stream and consists of an area approximately 650 m2 within the riparian zone. For ease of discussion, these sites are hereafter referred to as Site 5Y, 10Y and M, for the 5-year, 10-year, and mature sites, respectively.

This study is one of several conducted through AARAM's Pachitea Project describing controls on biogeochemistry of the Rio Pachitea basin. Ongoing research projects in the vicinity are investigating microcatchment-scale biogeochemistry, through an assessment of forest species assemblages along with a characterization of hydrologic flowpaths and associated biogeochemical fluxes. These data will be compiled into a larger study relating microcatchment data to patterns observed in the Rio Pachitea watershed.

METHODS

Litterfall

Field sampling consists of randomly spaced litter traps placed to quantify the amount and composition of litterfall reaching the riparian forest floor, and falling directly into the stream. At each site there are nine forest floor traps and three stream traps (traps suspended directly above the stream). Both forest floor and stream traps consist of meter-square wooden frames holding 2.0 mm mesh screen. All forest floor traps are terraced (i.e., not parallel to the sloped ground surface) with the uphill edge about 10 cm above the surface. Stream traps are approximately 40 cm above water surface. Thus, at each site 9 m2 of riparian forest floor and 3 m2 of stream surface are sampled.

Since sampling began 11/30/98, each trap has been emptied biweekly to monthly, and contents dried at approximately 55C for 24 hours. Trap contents are sorted into litter fractions (leaf, reproductive parts, twigs, epiphytic, and frass). The leaf fraction is then further sorted into family, genera or species components to the extent possible. Unidentifiable leaves are placed in their own category. Each fraction is identified and weighed.

Nutrient Analyses

Nutrient analyses are conducted on composited samples. Traps from each site are randomly divided into three groups. Given the strong gradients for nutrients, soil, and plant matter existing along landslide scars, samples were further divided into traps from upper and lower zones of affected sites (Sites 1 and 2). Sorted litterfall fractions from each group were composited with other homogenous fractions from traps within each respective group. Traps contents will ultimately be analyzed separately by season (wet vs. dry). All fractions are then ground and homogenized for nutrient analyses. Based on observed litterfall inputs, five families representing the largest total seasonal contributors (by weight) are selected for each site. These five families, along with the reproductive, frass, twig, epiphytic, and unidentified leaf fractions, are then analyzed for nutrient concentrations.

Dried litterfall samples are finely ground (to pass through a 16-mesh-per-cm screen) and analyzed for C, N, P, K, Ca, Mg, S, and Si. Carbon and nitrogen are determined using dry combustion on a C-N analyzer. All cations are determined using an inductively coupled argon plasma analyzer (ICP). For these, samples are acid-digested and centrifuged before being run on the ICP.

RESULTS AND DISCUSSION

The following results are presented representing the first three months of select data collected in this ongoing year-long study.

Total Litterfall Inputs and Composition

Total litterfall inputs between 11/30/98 and 3/14/99 vary predictably amongst the three sites. Total litterfall over this 104-day period ranged between 80.04 g/m2 at Site 5Y, 92.15 at Site 10Y, and 214.24 at Site M. Total litterfall inputs were greater in older sites and in lower portions of landslides as shown in Figure 1. Between 11/30/98 and 3/14/99, total litterfall inputs across the landslide-affected sites were as follows: 71.12 and 124.67 g/m2 at upper and lower zones of Site 5Y, and 91.90 and 93.42 at upper and lower zones of Site 10Y. Litterfall rates at the bottom of Site 5Y, the youngest site, were greater than those at both the upper and lower portion of the medium-aged site (Site 10Y) suggesting soils and organic material deposited by landslides provide nutrient capital to fast-growing pioneer species for at least five years. The composition of total litterfall inputs is presented in Table 1. Plant families contributing the greatest relative percentage to total inputs at each site are as follows: Upper Site 5Y collected primarily Poaceae (22%) and Euphorbiaceae (19.1%), while Lower Site 5Y collected Euphorbiaceae (20.4%) and Solanaceae (17.2%). Site 10Y's upper zone collected Euphorbiaceae (14.6%) and Proteaceae (10.5%), while lower Site 10Y collected 31.1% Solanaceae. Site M's inputs were largely dominated by Juglandaceae (22.3%).

TABLE 1: Total litterfall inputs between 11/30/98 and 3/14/99

Litterfall Rates

Observed litterfall rates ranged between 0.54 and 0.83 g/m2/day at Upper Site 5Y, 1.01 and 1.25 at Lower Site 5Y, 0.67 and 1.13 at Upper Site 10Y, 0.8 and 1.04 and Lower Site 10Y, and 1.3 to 2.57 at Site M. Mean litterfall rates for each collection period are presented in Table 2.

TABLE 2: Mean litterfall rates for each collection period (g/m2/day)

Nutrient Concentrations

Nitrogen, phosphorus and carbon concentration data (in mg/g) are presented in Table 3. These data, like those of Veneklaas (1991), showed significant differences between some analyzed plant fractions. The wood fraction was the lowest in N content (10.9 mg/g + 0.93) while the reproductive parts fraction was the highest (23.3 mg/g + 1.64). Nitrogen concentrations in leaves varied significantly among families (between 12.3 -mg/g + 0.88 and 21.5 mg/g + 0.64 for Poaceae and Solanaceae, respectively).

Phosphorus concentrations also varied among fractions. Reproductive parts contained the highest concentration of P (2.01 mg/g + 0.18), while the lowest values were found in leaves from the Poaceae family (0.56 mg/g + 0.03). Leaf P content varied between Poaceae and Euphorbiaceae (1.53 mg/g + 0.07). Leaves represented highest and lowest values for K concentrations, with unidentified leaves containing 0.52 mg/g + 0.52, and Lauracaceae 10.4 mg/g + 0.3.

Previous studies have shown that forests on moderately fertile soils produce more litter at higher nutrient concentrations than those on other soils (Vitousek and Stanford 1986), suggesting higher rates of litterfall, and higher mass/nutrient ratios, may be indicators of total stand nutrient capital (Vitousek 1984). Varying nutrient concentrations between plant families implies, not surprisingly, that total differences in particulate allochthonous nutrient inputs vary with successional plant community development, as changes in vegetation assemblages would be expected to alter nutrient cycling rates. These may take place in the short-term, as with species-specific differences in seasonal inputs, or in the long-term, with shifting plant communities. The pattern and timing of vegetation recolonization may be likely to exert a controlling influence on nutrient retention within a riparian forest ecosystem, as well as on downstream systems receiving autochthonous inputs.

Early successional forests (upper portions of both 5Y and 10Y) at our sites are dominated largely by bamboo (Poaceae), and ferns, two families characterized by low N and K content. Interestingly, ferns, unlike bamboo, contain relatively high levels of P, a lithophilic and often limiting nutrient, whose levels may be increased after landslides (Zarin and Johnson 1995). N-rich Solanaceae was predominantly found in lower zones of landslides where soil N levels would be expected to be higher. Reproductive parts held the highest N and P loads, and the second highest for K. Contribution of reproductive parts also increased with stand age, with 5Y showing 1.7 + 0.21 g/m2 and 0.47 + 0.18 for upper and lower zones, 10Y supporting 3.33 + 0.75 and 3.07 + 0.99, and P, 15.18 + 11.13.

TABLE 3: Concentrations of nitrogen, phosphorus and carbon in select litterfall fractions

Data are currently being collected to determine whether annual variation of litterfall fractions is occurring at Quebrada El Cedro. Based on these data we will determine whether differences in annual cycles of allochthonous materials exist among different aged riparian forests, and whether older (and less disturbed) riparian forests provide greater amounts of litterfall-derived nutrients to adjacent streams.

Future Work

These data represent the first three months of a yearlong study, the goal of which is to provide an annual budget for particulate OM litterfall and its' associated nutrient load into riparian soils and streams across a natural landslide-driven disturbance gradient within a headwater Amazonian stream system. Results obtained will aid in the understanding of controls on ecosystem resilience and recovery of montane tropical forests within the upper Pachitea River Basin. The field portion of this study will be completed in December 1999. Nutrient analyses will be completed by April 2000, and a full report and results can be expected in August 2000. Given the wide-spread clearing of Amazonian forests by humans, results from this study may help provide insights into natural controls on disturbance which may be important to developing management strategies which minimize long-term affects associated with anthropogenic disturbance.

REFERENCES

Bormann, F.H. & G.E. Likens. (1979) Pattern and Process in a Forested Ecosystem. Springer-Verlag, New York.

Edwards, P.J. (1977). Studies of mineral cycling in a montane rain forest in New Guinea: II. The production and disappearance of litter. Journal of Ecology 65: 971-992.

Foster, R. (1981) Brief inventory of plant communities and plant resources of the Palcazu Valley in Central Selva Natural Resources Management Project. JRB Associates, Lima. USAID Project No. 527-0240, Vol. II, App. F.

Garwood, N.C., D.P. Janos & N. Brokaw. (1979) Earthquake-caused landslides: A major disturbance to tropical forests. Science 205:997-999. September 7.

Grubb, P.J. (1977) Control of forest growth and distribution on wet tropical mountains: with special reference to mineral nutrition. Annual Review of Ecology and Systematics 8:83-107.

Haycock, N.E., T.P. Burt, K.W.T. Goulding & G. Pinay. (1997) Buffer Zones: Their Processes and Potential in Water Protection. Quest Environmental, Hertfordshire, UK.

Marrs, R.J., J. Proctor, A. Heaney & M.D. Mountford. (1988) Changes in soil nitrogen &endash; mineralization and nitrification along an altitudinal transect in tropical rainforest in Costa Rica. Journal of Ecology 76:466-482.

McClain, M.E., J.E. Richey & T.P. Pimentel. (1994) Groundwater nitrogen dynamics at the terrestrial-lotic interface of a small catchment in the Central Amazon basin. Biogeochemistry 34:45-56.

McClain, M.E., J.E. Richey & R.L. Victoria. (1996) Andean contributions to the biogeochemistry of the Amazon River system. Bulletin de l'Institut Fracais d'Etudes Andines 24:425-437.

McClain, M.E., J.E. Richey, J.A. Brandes & T.P. Pimentel. (1997) Dissolved organic matter and terrestrial-lotic linkages in the central Amazon Basin, Brazil. Global Biogeochemical Cycles 11:295-311.

Pinay, G., H. Decamps, E. Chauvet & E. Fustec. (1990) Functions of ecotones in fluvial systems, in The Ecology and Management of Aquatic-Terrestrial Ecosystems (ed. by R.J. Naiman & H. Decamps), Man and Biosphere Series, Vol. 4 (ed. by J.N.R. Jeffers), Unesco Publishing, Paris, France, and Parthenon Publishing Group, Lancs, UK.

Swanson., F.J., R.L. Fredriksen & R.M. McCorison. (1982) Material transfer in a Western Oregon watershed, in Analysis of Coniferous Forest Ecosystems in the Western United States (R. Edmonds, ed.) US/IBP Synthesis Series #14. Hutchinson Ross Publishing Co., Stroudsburg, Pennsylvania, USA.

Tanner, E.V.J. (1980) Studies on the biomass and productivity in a series of montane rain forests in Jamaica. Journal of Ecology. 68:573-588.

Tanner, E.V.J., P.M. Vitousek & E. Cuevas. (1998) Experimental investigation of nutrient limitation of forest growth on wet tropical mountains. Ecology 79(1):10-22.

Veneklaas, E.J. (1991) Litterfall and nutrient fluxes in two montane tropical rain forests, Columbia. Journal of Tropical Ecology 7:319-336.

Vitousek, P.M. (1984) Litterfall, nutrient cycling, and nutrient limitation in tropical forests. Ecology 65(1):285-298.

Vitousek, P.M., and R.L. Stanford, Jr. (1986) Nutrient cycling in moist tropical forest. Annual Review of Ecology and Systematics 17:137-167.

Zarin, D.J. & A.H. Johnson. (1995) Nutrient accumulation during primary succession in a montane tropical forest, Puerto Rico. Soil Science Society of American Journal 59:1444-1452.