328

Research Needs and Applications to Reduce Erosion and Sedimentation in Tropical Steeplands (Proceedings of the Fiji Symposium, June 1990): IAHS-AISH Publ. No.192,1990. Selection of riparian buffer zones in humid tropical steeplands F.N. SCATENA Institute of Tropical , Southern Experiment Station, U.S.D.A. Forest Service, Rio Piedras, Puerto Rico ABSTRACT This paper discusses various aspects of the design and costs of riparian protection zones in the humid steeplands of the Luquillo Mountains of Puerto Rico. This riparian community is dominated by successional species and can be protected without large losses in commercial timber. However, the cost of buffers increases geometrically as smaller order channels are buffered. Generally the costs associated with protecting intermittent streams is prohibitive in this highly dissected terrain. Depending on the design criteria used, protection zones along perennial streams can include between 5 and 20% of the commercial basal area while protecting 9 to 25% of the watersheds area.

INTRODUCTION

The use of riparian protection zones in forestry operations is based on the premise that the structure of the riparian zone has a controlling influence on the environmental conditions of the aquatic habitat. This paper discusses the influence of several ecosystem characteristics that affect the selection of riparian buffers in the Luquillo Mountains of Puerto Rico. In addition, some economic consequences of different buffer designs are considered. Buffer zones are defined here as strips of undisturbed terrain adjacent to the stream channel. Buffer width is the ground surface distance from the bankfull channel margin to the edge of permitted utilization along one side of the channel. The study was conducted on two watersheds in the Bisley sector of the Luquillo Experimental Forest of Puerto Rico (Lat 18 18'N, Long 65 50'W). The physiography, land-use history, and ecology of the area are described in elsewhere (Scatena 1989, Brown et al. 1983). In general, these 6.7 and 6.3 ha watersheds drain a mature secondary forest in a subtropical wet forest life zone (sensu Holdridge 1967). Both watersheds are underlaid by volcanoclastic sandstones and tuffs that weather into a clayey pedon. Over 50% of the study area has slopes greater than 45% and shallow landslides are common.

GENERAL DESIGN CONSIDERATIONS

In the Caribbean region, riparian protection zones have been used for over 260 years. In 1721 laws prohibiting the of "around" a series of reservoirs were established in Antigua (Beard 1949). In 1839, a protected zone was recommended along the 329

"margins and headwaters of streams" in Puerto Rico (Wadsworth 1948). Regardless of this long history, most buffer design in the region is still based on imprecise adjectives rather than concise or ecologically sound specifications. In a recent review of riparian buffers, Clinnick (1985) concluded that the most recommended buffer width for temperate is 30 meters. While it was recommended that buffers extend to the spring head or runoff confluence it was acknowledged that this is rarely practiced. Moreover, buffers are usually only required along perennial streams. Although buffers are commonly recommended, and even legislated for tropical streams, both implementation and published design criteria are rare. In some Australian tropical forests, a minimum strip of 20 m of undisturbed vegetation is required on either side of channels draining 100 ha or more (Phillis 1980). Furthermore, no is permitted within 100 m of the top waterline of any "major" water storage site. In other Australian tropical forests, a minimum width of 10 m is required on streams draining 60 ha or more (Ward et al. 1980). Because of variations in site conditions, at their best, design recommendations only provide a guide to catchment managers (Trimble et al. 1957, Steinblum et al. 1982, Clinnick 1985, Borg et al. 1988). Nevertheless, the ideal buffer would provide the maximum environmental protection at the minimum economic cost. Maximizing environmental benefits can presumably be met by: 1) minimizing changes in light, temperature, sediment and channel morphology within the riparian zone; and 2) reducing offsite exports of sediment and nutrients by providing storage sites adjacent to the active stream channel. The environmental costs of buffers can be reduced by: 1) limiting the amount of valuable timber within the buffer, and 2) reducing associated extraction costs. The determination of effective buffer width and extent has typically been established in two ways (Clinnick 1980): 1)'establishing and protecting the minimum area contributing the runoff, and 2) by determining the sediment trapping efficiency of a vegetative strip. In this paper, buffer delineation is also been based on the hypothesis that the riparian corridor is a mosaic of dynamic patches whose interactions are fundamental to the ecological properties of the zone (Naiman et al. 1988). Therefore, buffer boundaries based on existing patches will reduce changes to the ecological properties of the riparian mosaic. This is particularly relevant to humid tropical streams where the stream community may be structured to reduce or recycle nutrients draining from the surrounding terrestrial environment (Dudgeon 1984, Covich 1988).

METHODOLOGY

In order to delineate characteristic traits of the riparian zone and thereby define potential buffer boundaries, systematic surveys and site classifications of the study area were made. Vegetation and site data were collected on systematically spaced, 10 m in diameter plots located on a 40 x 40 meter grid that covers both 330

watersheds. In addition, 17 transects situated across stream valleys were also established to characterize riparian vegetation (Heaton and Letourneau 1989). These transects were divided into subplots of areas equal to the plots that grid the watersheds. At each plot, the following criteria was collected: 1) topographic location (ridge slope, valley, stream channel), slope, and distance from bankfull margin; 2) vegetation: species, height class, diameter at breast height for each stem equal to or greater than 2.5 cm; 3) structure: presence or absence of emergences, dominants, codominants, midstory, understory, and herbaceous strata. Analysis of topography and hydraulic geometry are based on 1:500 scale topographic maps of the watersheds and were supplemented with direct field measurements. Permanently saturated areas were located in the field and were defined by the presence of gley saturated soils.

RESULTS

Geomorphic conditions

The study watersheds are characterized by steep slopes, knife-like divides and a relatively high drainage density. Over 65% of the plane watershed area is, classified as slopes. Ridges and valley bottoms each comprise about 17% of the landscape. The channel network in the watersheds consists of steep gradient streams supported by a dense network of ephemeral channels (Table 1). The drainage net is dendritic and follows the geometric laws of stream order and number (sensu Straler 1952). Furthermore, stream channels can be divided into three morphological types: main channels, intermittent channels, and leaf-lined swales (Scatena 1989).

Table 1. Morphological characteristics of Bisley stream channels. First order segments are intermittent channels definable on a 1:500 scale map and do not include leaf filled swales.

Stream Number Mean Total length order per ha length (m) (m/ha)

1 1.9 30 57 2 0.61 61 37 3 0.31 61 19 4 0.16 141 23

The main channels of the watersheds are third and fourth order segments with concave profiles, high flow channels, small isolated floodplains and steep side walls. Floodplains are rare and when present are usually formed from toes of landslips. First and second order channels are intermittent and typically lined by plastic clay and moss covered boulders. Although these channels only have water flowing in them several times a month, 331

they are sufficiently active to prevent the establishment of woody vegetation or the persistent accumulation of organic debris. Draining into these intermittent channels are 1 to 2 m wide swales that are armored with mats of fine roots and are filled with moist leaves and occasional saplings. These swales are definable up to the watershed divides and are so numerous that in places they form a corrugated surface on the upper slopes of the basins. Landslips and fall gaps are common features along the steep margins of the 3rd and 4th order channels. For example, over a 1.6 years period, gap induced canopy turnover was twice as fast for riparian canopies compared to the more stable ridgetops. Furthermore, nearly 60% of the annual bedload export in a 1987/88 water year was due to a shallow landslide adjacent to one of the channels (Scatena unpublished data). Saturated, gley soils are also common features of the riparian zone. Gley soils underlie 8.5% of the watershed area and primarily occur along channel margins and at the bottom of oil catenas. Approximately 75% of these soils are within 10 m of 3rd and 4th order channels.

Vegetation

Vegetation in the study area consists of a mature secondary Dacryodes-Sloanea type forest. Characteristic traits of this forest are the distribution of species by topography (Briscoe and Wadsworth 1970) and a well stratified canopy (Beard 1930, Richards 1966). Typically, stable ridge sites are covered by the dominate hardwoods while valleys floors are dominated by palms (Prestoea montana), herbs, and fast growing light demanding species. The canopy of perennial channels is generally two-layered, and consists of an open upper strata of sprawling lateral branches, and a nearly continuous herbaceous layer (Table 2). In contrast, upland canopies are usually closed and structured by dominant, codominant and understory strata.

Table 2. Canopy characteristics for perennial stream channels of the Bisley watersheds.

Distance from channel (m) Canopy trait 0-10 10-20 20-30 <30

Number of plots 30 17 11 20

.lean number of 1.3 2.1 2.4 2.8 non-shrub (.45) (.56) (.50) (.64) strata (SD)

Percent of plots 100 82 46 45 with herbaceous strata 332

Over 85% of the upland forests have three or more definable strata whereas only 30% of the riparian canopy has three or more layers. Moreover, a continuous herbaceous layer occurred in all the plots within 10 m of perennial channels but in only 46% of the upland locations. Generally, herbaceous undergrowth is dense along the margins of perennial channels but has a patchy distribution along intermittent channels and at distances greater than 25 m from perennial channels. An open riparian canopy with dense understory and large individuals standing apart from each other has been noted elsewhere in the Caribbean (Shreve 1914, Beard 1949) and is considered a common trait of rain forest streams (Richards 1966). In addition, basal areas (Table 3), leaf area index (Odum et al. 1970) and interception losses (Scatena 1989) are known to increase away from the channel toward the more stable ridge tops in the Luquillo Forest.

Table 3. Vegetation characteristics by topographic zone for perennial stream channels in the Bisley drainages.

Topographic Distance Number Basal area m2/ha) zone from of Mean (standard deviation) channel plots Total Commercial

Active channel 0-10 16 16.7 (22.3) 8.1 (20.0) Riparian 10-20 19 22.9 (14.9) 11.5 (13.4) Transition 20-30 9 32.8 (18.1) 10.5 (14.0) Upland <30 57 35.3 (24.6) 13.7 (9.5) Ridges Variable 11 50.4 (26.6) 18.7 (26.0) a) Basal areas are average values for topographic zone and are based on unpublished data and information in Heaton and Letourneau (1989). Estimates of commercial basal area exclude all but the five most valuable timber species: Guarea guidonia, Dacryodes excelsa, Manilkara bidentata, Swietenia macrophylla, Khaya nyasica.

Dacryodes excelsa is both the dominant species and most abundant commercial species in these forests. However, Dacryodes rarely grows long the riparian corridor and only attains large sizes on the more stable ridgetops. The only commercial valuable species that is common along permanent stream channels is Guarea guidonia. This native of the family commonly occurs in the small floodplains of the drainages.

DISCUSSION

Hydrologie and Geomorphic Considerations

Saturated overland flow is considered to be the dominant runoff mechanism in the clayey soils of the humid Caribbean (Walsh 1980). Tn the Bisley watersheds, quickflow from channel precipitation and 333

areas of saturated soils account for approximately one-third of the annual discharge (Scatena unpublished data). Therefore, while these areas comprise less than 9% of the watershed area, they have a disproportionate effect on stream runoff and chemistry. Likewise, protecting these areas will produce disproportionate benefits in reducing changes in the hydrology of the system. In the study area, 75% of the saturated areas are within 10 m of the perennial channels. This suggests that a 10 m buffer along perennial channels is the minimum buffer area needed to reduce, but not necessarily eliminate, changes to stream hydrology. The effectiveness of buffers as traps of sediment eroded from the upland sites is questionable in these particular watersheds. First, the steeply sloping channel walls provide few deposition sites for sediment. Secondly, personal observations and the abundance of intermittent channels suggest that during large storms most transport to perennial channels is by channelized flow in the intermittent streams. Under these conditions the trapping and filtering capacity of the buffers is greatly diminished. The establishment of riparian buffers will not eliminate existing erosion along the channel margins. However, it may reduce accelerated input from these sources. Regardless of their location, these unstable areas should be included in buffer ones to avoid any accelerated erosion or instability. Unstable areas are occupied by successlonal species, are generally devoid of valuable timber, and can be buffered without significant losses in commercial volume.

Vegetation

Minimal to light, litter fall, and other environmental conditions adjacent to the channel is necessary to maintain th environmental integrity of the riparian zone. This should occur if the canopy affeeing those processes is not disturbed. Presence of well defined herb layer appears to be an adequate indicator of riparian conditions and can be used to delineate riparian zones. Furthermore, since continuous herbaceous undergrowth also occupies saturated and unstable areas, defining buffers on the basis of herbaceous undergrowth will also include these areas. The use of buffers as micro-refugia to maintain and wildlife habitat is a potential objective, or at least added benefit of buffers. This can be particularly useful in highly disturbed areas where buffers act not only as a refuge but also as corridors for migration to less disturbed areas. However, given the relatively low densities and species compositions found along streams, the protection of riparian areas cannot be substituted for the protection of suitable upland habitat.

Economie Considerations

Although riparian protection zones have potential environmental benefits, they have an associated economic cost. The relative costs of various buffer designs can be compared if watershed area 334

is considered a surrogate for environmental benefit and the commercial basal area included in the buffers a measure of economic cost. Opportunity cost are implicitly considered in ths model. Additional costs associated with buffer establishment or added extraction costs are not included. Nevertheless, costs resulting from complex extraction paths needed to avoid riparian buffers may be substantial. The total buffered area within a watershed is the sum of the area buffered along each stream segment: likewise, the total basal area excluded from commercial use is the sum of that included in the buffer. Using actual channel lengths (Table 1), and basal areas (Table 3) both the watershed area and commercial basal area included in different buffers designs were calculated for the study area (Fig. 1 and Table 4).

Figure 1. Percent watershed area and commercial basal area included in riparian buffers of various widths in the Bisley watersheds. % Watershed area 100

% total oommarteal Basal area 100 / 90 / Intermittent 80 / Channels 70 h / 60 50 - X 40 yS ^^ s^ ^^^^ Perennial 30 ^^ ^^^ Channels 20 10 ^^\^^^^^ •**«z--r'~~~~\. i i i i i i i i i 10 15 20 25 30 35 40 45 50 55 60 Buffer Width (m) 335

Table 4. Comparison of various riparian buffer designs for the Bisley watersheds.

Buffer Buffer % total % Commercial Buffer design width extent area of basal area (m) buffer in buffer

Riparian 10 Perennial Q9 saturated channels areas only

Marginal cost 22 Perennial 18 13 equal benefit

Distribution 25 Perennial 17 16 of herbaceous vegetation

Kiparian zone 25+ Perennial 25 20 saturated and springs, unstable unstable areas areas

While actual solutions will vary with local conditions and refinements could be added, this model indicates that both cost and benefits of buffers increases geometrically as buffers are included on smaller order channels. This is a direct consequence of the law of stream numbers: channel number and total length increase geometrically with decreasing channel order. If all perennial and intermittent channels in the watersheds were protected by a 35 m wide zone on each side of the channel, the entire watershed would be buffered. Generally, buffering intermittent channels will not only remove prohibitive amount of area from use, it will also create a labyrinth of usable areas that will greatly increase extraction costs. Buffers based on the presence of dense herbaceous vegetation along perennial channels will exclude 16% of the commercial basal area from exploitation. When all saturated and unstable areas are included, approximately 25% of the watershed can be protected at a cost of 20% of the commercial basal area. Actual losses are less since most riparian trees have relatively short boles, wide crowns and poor commercial form. Since the perennial channel margins in these forests are dominated by successional type species while commercial hardwoods are generally restricted to upland areas, both the total and average cost of buffer increase with buffer widths. However, once buffers extend into areas occupied by upland hardwoods, total cost increases rapidly with increases in buffer width. Therefore, an optimal width from an economic perspective may be considered the point where the marginal gain in buffer area equals the marginal increase in commercial basal area included in the buffer: i.e., marginal benefits equal marginal costs. Although this design may not be an ecological optimum, buffers larger than this will include commercial basal area at a rate 336

greater than the increase in watershed area. Based on values from the Bisley watersheds, this optimum averages 22 in for perennial channels and less than 10 m for intermittent channels. While the actual point where marginal benefits equal marginal cost will vary with site conditions, it will generally be similar to the extent of herbaceous undergrowth because the lack of herbaceous undergrowth corresponds to a dense upland canopy and greater commercial basal areas.

CONCLUSIONS

Due to the high stream channel densities in humid tropical steeplands, riparian buffers can eliminate significant areas of the watershed from commercial use. However, since these riparian zones are typically low in valuable timber species, the actual cost of buffers is less compared to areas where the riparian community contains the dominate commercial species. Both the watershed area and commercial basal area included in buffers increases geometrically as smaller and smaller channels are protected. Due to the relatively high stream channel densities in this, and presumable other, tropical steeplands, protection of intermittent channels is economically prohibitive. Furthermore, since storm runoff is dominated by channelized flow from intermittent swales, and the steep channel margins cannot store large volumes of sediment, the sediment trapping efficiency of these buffers is greatly reduced. Delimiting buffers on the presence of herbaceous vegetation provides both a practical and ecological sound solution to buffer selection. Not only are the buffers relatively easy to distinguish, they also protect areas with few commercial hardwoods. Furthermore, since herbaceous layers are a diagnostic feature of these riparian zones, buffers based on their presence should reduce disruption to the existing vegetation mosaic and therefore reduce changes to ecological characteristics of the system.

REFERENCES

Beard, J.S. (1949). The natural vegetation of the Windward and Leeward Islands. Clarendon Press, Oxford. Borg H., Hordace A., Batini, F. (1988). Effects of logging in stream and river buffers on watercourses and water quality in the Southern Forest of Western Australia. Aust. For., (2):98-105. Briscoe, C.B., Wadsworth, F.H. (1970). Stand structure and yield in the Tabonuco forest of Puerto Rico. Chapter 156 ijl Odum, II.T. and Pigeon, R.F., eds. A tropical rain forest, 13-6. NTIS, Springfield, Va. Brown, S., Lugo, A.E., Silander, S., Liegel, L. (1983). Research history and opportunities in the Luquillo Experimental Forest. U.S. Forest Service General Technical Report, S0-44. Clinnick, P.F. (1985). management in forest operations: a review. Aust. For. 48(l):43-45. 337

Covich, A. (1988). Geographical and historical comparisons of neotropical streams: blotic diversity and detrital processing in highly variable habitats. J. North American Benthological Society 7(4):361-386. Dudgeon, D. (1984). The importance of streams in tropical rain forests systems. Tropical rain-forests: The Leeds Symposium 99, 71-82. Keaton, K., Letourneau, A. (1989). Changes in forest structure and composition along a gradient from streams to ridges in a subtropical moist forest in Puerto Rico. Intership report to the Institute of Tropical Forestry for the School of Forestry and Environmental Studies, Yale University. Phillis, K.J. (1985). Operational protection of forest values: in managing the . Pages 153-164 In K.R. Shepard, ed. Australian National University Press. Naiman, R.J., Decamps, H., Pastor, J. Johnston, C.A. (1988). The potential importance of boundaries to fluvial systems. J. North American Benthological Society 7(4):289-306. Richards, P.W. (1966). The tropical , an ecological study. Cambridge University Press. Scatena, F.N. (1989). An introduction to the physiography and history of the Bisley Experimental Watersheds in the Luquillo Mountains of Puerto Rico. U.S. Forest Service. General Technical Report SO-72. Shreve, F. (1914). A montane rain-forest. A contribution to the physiological plant geography of Jamaica. Publ. Carneg. Inst. 199. Steinblum, I.J., Froehlick, J.K., Lyons, J.K. (1984). Designing stable buffer strips for stream protection. J. For. 82:49-52. Trimble, G.R., Sartz, R.S. (1957). How far from a stream should a logging road be located? J. For. 55:339-341. Wadsworth, F.il. (1949). The development of the forest land resources of Luquillo Mountains of Puerto Rico. PhD Dissertation, University of Michigan, Ann Arbor. Ward, J.P., ixanowski, P.J. (1985). Implementing controls of harvesting operations in North Queenslands rainforests. Pages 1Ô5-186 j_n_ Shepard, K.R., ed. Managing the tropical forests. Australian National University Press. Walsh, R.P.D. (1980). Runoff processes and models in the humid tropics. Z. Geom. N.F. Supp. Bd. 36:176-202.