BIOTROPICA 35(1): 84±93 2003 Food Web Structure and Basal Resource Utilization along a Tropical Island Stream Continuum, Puerto Rico1 James G. March 2 and Catherine M. Pringle Institute of Ecology, University of Georgia, Athens, Georgia 30602, U.S.A. ABSTRACT Tropical stream food webs are thought to be based primarily on terrestrial resources (leaf litter) in small forested headwater streams and algal resources in larger, wider streams. In tropical island streams, the dominant consumers are often omnivorous freshwater shrimps that consume algae, leaf litter, insects, and other shrimps. We used stable isotope analysis to examine (1) the relative importance of terrestrial and algal-based food resources to shrimps and other consumers and determine (2) if the relative importance of these food resources changed along the stream continuum. We examined d15N and d13C signatures of leaves, algae, macrophytes, bio®lm, insects, snails, ®shes, and shrimps at three sites (300, 90, and 10 m elev.) along the RõÂo EspõÂritu Santo, which drains the Caribbean National Forest, Puerto Rico. Isotope signatures of basal resources were distinct at all sites. Results of two-source d13C mixing models suggest that shrimps relied more on algal-based carbon resources than terrestrially derived resources at all three sites along the continuum. This study supports other recent ®ndings in tropical streams, demonstrating that algal-based resources are very important to stream consumers, even in small forested headwater streams. This study also demonstrates the importance of doing assimilation-based analysis (i.e., stable isotope or trophic basis of production) when studying food webs. Key words: Caribbean; Decapoda; Greater Antilles; lotic; prawn; Puerto Rico; rain forest; RCC; river; species interactions. THE RIVER CONTINUUM CONCEPT suggests that as the tinua (Hunte 1978, Marquet 1991, FieÁvet et al. physical structure of a river changes from head- 2001, March et al. 2001) and play a major role in waters to mouth, the relative importance of allocht- determining benthic community composition and honous versus autochthonous resources to consum- the rate of organic matter processing (Pringle et al. ers will also change (Vannote et al. 1980). For ex- 1999, Crowl et al. 2001, March et al. 2001). Initial ample, in forested headwater stream sites, alloch- studies in small headwater streams suggest that thonous inputs such as leaves may form the base shrimp assemblages are highly dependent on ter- of the food web, while in wider low-elevation sites, restrial detritus as a food resource (Covich 1988 a, autochthonous algal resources may be more im- b; March et al. 2001). Research has also shown that portant to consumers. Studies examining this pat- shrimps can have strong negative effects on algal tern in temperate streams, characterized by ®shes abundance (Pringle 1996, Pringle et al. 1999, and insects, generally support this hypothesis March et al. 2002). While we know that shrimps (Winterbourn et al. 1984, Doucett et al. 1996); are important in regulating standing stocks of both however, no published information exists about re- algae and leaves, we know little about the relative source-use patterns along tropical island stream importance of these basal resources to shrimps, or continua. Recent research in the tropics, in both if the relative importance varies along river contin- small streams (Salas & Dudgeon 2001) and large ua. river ¯oodplains (Lewis et al. 2001), suggest a Stable isotopes are a useful tool to identify basal greater reliance of consumers on algal-based re- resources utilized by consumers. Terrestrial and sources. aquatic plants often differ in their 13C signatures Freshwater shrimps are an important compo- (Fry & Sherr 1989). Because 13C signatures of con- nent of tropical streams worldwide. They are es- sumers re¯ect those of the plants they eat, analysis pecially important in tropical island streams, where of 13C signatures have been used successfully to they are often very abundant along entire river con- estimate the relative importance of different basal food resources in stream food webs (Winterbourn 1 Received 19 March 2002; revision accepted 20 Decem- et al. 1984, Rounick & Hicks 1985, Hamilton et ber 2002. 2 Current address: Biology Department, Washington and al. 1992, Doucett et al. 1996). Furthermore, anal- 13 Jefferson College, 60 South Lincoln St., Washington, ysis of C signatures provides an advantage over Pennsylvania 15301, U.S.A. e-mail: [email protected] gut content analysis because it measures the 84 Tropical Island Stream Food Web 85 TABLE 1. Physical parameters at each site. High elevation Mid elevation Low elevation Elevation (m) 300 90 10 Stream order 2 3 4 Channel width (m) 13 15 19 Mean daily dischargea (m3/sec) 0.19 1.11 2.56 % Canopy cover 69 33 21 Drainage area (km2) 3 14 33 a Calculated from USGS data (1 October 1995±30 September 1998). amount of carbon assimilated from each food ested. Stream width widens and canopy cover de- source as opposed to that ingested. This is impor- creases in a downstream direction (Table 1). The tant because food sources can differ greatly in qual- EspõÂritu Santo enters the ocean ca 6kmdown- ity. Assimilation ef®ciencies of leaves are typically stream of the low-elevation site. low compared to algal resources (Benke & Wallace Ten species of omnivorous freshwater shrimps 1980, Whitledge & Rabeni 1997). Stable isotopes inhabit rivers of the LEF, representing three fami- are also useful in determining trophic level because lies: Palaemonidae, Atyidae, and Xiphocarididae. the 15N signature is generally enriched by 2.5 to The palaemonid shrimps Macrobrachium carcinus 3.5½ with every trophic transfer (Peterson & Fry (L), M. faustinum (De Saussure), M. crenulatum 1987, Hershey & Peterson 1996). Holthuis, M. acanthurus (Wiegmann), and M. he- In this study, we used stable isotope analysis to terochirus (Wiegmann) are thought to consume de- describe the basic structure of the food web along composing leaf litter, ®ne particulate organic mat- a tropical island stream continuum in Puerto Rico. ter, algae, macrophytes, insects, mollusks, small We addressed the following questions: (1) What is ®sh, and other shrimps (Covich & McDowell the relative importance of different basal food re- 1996). The atyid shrimps, represented by Atya la- sources (e.g., algae and leaves) to shrimps and other nipes Holthuis, A. scabra (Leach), A. innocous consumers? and (2) Does the relative importance (Herbst), and Micratya poeyi (GueÂrin-MeÂneville), of algal and leaf resources to consumers vary along consume ®ne particulate organic matter, leaf ma- the stream continuum? terial, algae, and small sessile insects by brushing with their cheliped fans (Pringle et al. 1993). They METHODS use these same cheliped fans to ®lter feed when ¯ow conditions are suitable (Covich 1988b). Xip- STUDY SITES AND NATURAL HISTORY. We conducted hocaris elongata (GueÂrin-MeÂneville) consumes leaf this study along the RõÂo EspõÂritu Santo in north- matter, periphyton, ®ne particulate organic matter, eastern Puerto Rico. Three 100 m study sites were insects, small ¯owers, and fruit using tiny pincers selected. The high-elevation site (300 m elev.) is in (Covich & McDowell 1996, Pringle 1996). Gut the second-order Quebrada Sonadora in the Carib- content information for these shrimp taxa is lim- bean National Forest, which is synonymous with ited. Pringle et al. (1993) found mostly silt/detritus the Luquillo Experimental Forest (LEF). Land use and ``unidenti®ed green material'' in both A. lanipes consists of tabonuco forest characterized by Dacr- and X. elongata. No gut content studies exist for yodes excelsa Vahl. Allochthonous inputs enter the Macrobrachium spp. from this river; however, in river system throughout the year with two peaks, ponds in Jamaica, M. carcinus consumed mostly from March to May and September to October detritus but also algae, leaves, plant seeds, aquatic (Covich & McDowell 1996). The mid-elevation insects, ®sh, mollusks, and other crustacea (Lewis site (90 m elev.) is located ca 3.1 km downstream et al. 1966). In southern Australian streams, atyid from the high-elevation site in a third-order por- and palaemonid shrimps were omnivorous; how- tion of the RõÂo EspõÂritu Santo. The low-elevation ever, bio®lm was a signi®cant portion of their diet site (10 m elev.) is in a fourth-order segment of the (Burns & Walker 2000). No information is avail- RõÂo EspõÂritu Santo, ca 4.2 river km downstream able on food preference. from the mid-elevation site. Land use is mostly res- All of these shrimp species undergo extended idential and pasture at the mid- and low-elevation larval development and are amphidromous (Covich sites, but the riparian zone is predominantly for- & McDowell 1996, Johnson et al. 1998); adult 86 March and Pringle females release planktonic larvae that drift down- with an electroshocker and by angling. A single stream to the estuary (March et al. 1998). Larvae crab was collected by hand at the high-elevation spend 50±110 days in the estuary before migrating site. Samples were stored frozen or dried until pre- back upstream as metamorphosed postlarvae (Cha- pared for stable isotope analysis. ce & Hobbs 1969, Hunte 1978, Benstead et al. 2000). Adult shrimps differ in their distribution SAMPLE PREPARATION AND ANALYSIS. All samples along the elevation gradient; high-elevation sites are were re-dried at 608C for at least 48 hours. Rep- dominated by atyid shrimps (mostly A. lanipes) and licates of leaves and macrophytes were ground into X. elongata, while Macrobrachium spp. and X. elon- powder with a ball mill. Insect samples were com- gata comprise the majority of the shrimp assem- posites of multiple individuals. At both the mid- blage at lower-elevation sites (March et al. 2001, and low-elevation sites, leptophlebiid and baetid 2002). may¯ies were combined to obtain suf®cient mass The insect assemblage at all three sites is de- for isotope analysis. Samples of muscle tissue were pauperate relative to continental tropical streams taken from snails, shrimps, and ®shes.