Biotic Indices and Stream Ecosystem Processes: Results From an Experimental STUDY1

Ecological Applications, 6(1), 1996, pp. 140-151 © 1996 by the Ecological Society of America

BIOTIC INDICES AND STREAM ECOSYSTEM PROCESSES: RESULTS FROM AN EXPERIMENTAL STUDY1 J. BRUCE WALLACE Department of Entomology and Institute of Ecology, University of Georgia, Athens, Georgia 30602 USA JACK W. GRUBAUGH,2 AND MATT R. WHILES3 Institute of Ecology, University of Georgia, Athens, Georgia 30602 USA Abstract. We investigated the ability of the North Carolina Biotic Index (NCBI) and the Ephemeroptera + Plecoptera + Trichoptera (EPT) index to track an experimental manipulation of the invertebrate community and resultant alteration of several ecosystem-level processes in a headwater stream at the Coweeta Hydrologic Laboratory in western North Carolina. Indices were calculated from quantitative monthly or bimonthly benthic samples of moss-covered rockface and mixed substrate habitats, as well as habitat-weighted values based on the proportion of each habitat in the two streams. One stream (C 55) served as a reference stream over the 6-yr period of late 1984 through 1990, whereas the other (C 54) received seasonal treatments with an insecticide for 3 yr (1986-1988). Throughout pretreatment, treatment, and recovery, both the NCBI and EPT indices tracked the disturbance regime of the treatment stream. Indices for the reference stream varied little during the 6-yr period. Both the NCBI and EPT suggested strong changes in the treatment stream during treatment relative to both pretreatment and the reference stream. Following cessation of insecticide treatments, both indices reflected improved biotic conditions during first and second years of recovery in C 54. Compared with fauna of mixed substrates, rockface fauna had lower (better) NCBI values during pretreatment, and exhibited a greater proportional increase in tolerant taxa during treatment than mixed substrates, emphasizing the importance of including rockface communities in environmental monitoring programs. Changes in both the EPT and NCBI indices closely corresponded to changes in ecosystemlevel processes observed in C 54 from pretreatment to treatment, and recovery periods. These processes include: leaf litter processing rates, organic matter storage, fine paniculate organic matter generation and export, and secondary production. With the exception of organic matter storage, all of these processes declined during treatment of C 54, and subsequently increased during recovery. Our results demonstrate the potential of such indices to detect and monitor stream ecosystem changes during and following disturbance. The EPT index was by far the easiest to use from both the standpoint of time required for sample processing and ease of application. Compared with the labor-intensive sample processing, specimen identification and measurement, and data entry required for secondary production calculations, the EPT index was relatively simple and displayed a remarkable ability to track secondary production of invertebrates in the treatment stream. Our data strongly support the inclusion of the EPT and NCBI indices in these southern Appalachian headwater streams as indicators of both degradation and recovery of stream ecosystem processes from chemical-induced disturbance. Key words: biomonitoring; biotic indices; ecosystem processes; EPT index; macroinvertebrates; manipulation; NCBI index; pesticides; recovery; secondary production; streams. INTRODUCTION Biological assessment of aquatic environments has been practiced since the early 1900s (see reviews of Hynes 1960, 1994, Cairns and Pratt 1993). As evidenced by a number of recent books devoted entirely to the subject (e.g., Abel 1989, Plafkin et al. 1989, Rosenberg and Resh 1993, Loeb and Spacie 1994), 1 Manuscript received 15 September 1994; revised 26 January 1995; accepted 16 February 1995. 2 Present address: Department of Biology, Ellington Hall, University of Memphis, Memphis, Tennessee 38152 USA. 3 Present address: Department of Biology, Division of Mathematics and Sciences, Wayne State College, Wayne, Nebraska 68787 USA.

interest in this area has grown tremendously. As demands on water resources increase, ambient biological monitoring is becoming a rapid and accurate means of assessing quality of lotic systems, although specific methods remain limited (Cummins 1994). Many methods have assessed stream quality using invertebrates, ranging from assessing physiological and morphological changes of individuals to various measures of community structure (e.g., Rosenberg and Resh 1993). Biotic indices, which evaluate macroinvertebrate community structure, are widely used. These indices often follow the approach described by Chutter (1972), in which tolerance values (TV) ranging from 0 (very intolerant) to 10 (very tolerant) are based on the ability

140

February 1996

BIOTIC INDICES AND ECOSYSTEM PROCESSES

of a taxon to inhabit stream or river systems differing in water quality (e.g., Hilsenhoff 1977, Plafkin et al. 1989, Lenat 1993). Other organisms such as fish also have been used extensively in biological monitoring programs (e.g., Plafkin et al. 1989, Loeb and Spacie 1994). Construction of biological indices requires considerable effort (i.e., Hilsenhoff 1987, Karr 1991, Lenat 1993, Kerans and Karr 1994). However, once derived, such indices, like most ecological studies, may have problems such as unclear identification of habitats and habitat characteristics sampled, differences in sampling intensity, lack of both pre- and post-impact data, and reliance on only one "unimpaired" reference site (Lenat and Barbour 1994). Years of sampling may be required to collect sufficient data to detect long-term trends, and determining whether an apparent trend is due to anthropogenic causes or natural variation may be impossible (Charles et al. 1994). Alternatively, biological monitoring offers a relatively affordable means of environmental measurement, compared to chemical data, for assessing degradation of aquatic habitats and loss of biological diversity induced by anthropogenic disturbances (Karr 1991, Hynes 1994). Biological evaluation also must take place against a firm knowledge of local fauna and flora (Karr 1991). Karr (1991, 1993) argues that biological indices, which incorporate concepts such as biological diversity and integrity, provide important measures of ecosystem health. Accordingly, the use of such indices, which encompass individual to landscape-level perspectives, should provide an ecologically robust means of assessing ecosystem health (Karr 1993), although other points of view have been expressed (Suter 1993). Further, experimental approaches, which incorporate aspects of biological monitoring into ecosystem-level manipulations, are lacking. A "top-down" ecosystem-level manipulation of a headwater stream at Coweeta Hydrologic Laboratory (North Carolina, USA) was designed to assess the role of macroinvertebrates (primarily insects) in ecosystem processes. Insecticide treatments reduced insect abundance (Wallace et al. 1989, 1991b), biomass, and secondary production (Lugthart and Wallace 1992), as well as leaf litter processing rates (Cuffney et al. 1990, Chung et al. 1993), without altering microbial respiration or abundance (Cuffney et al. 1990, Suberkropp and Wallace 1992). Treatments also caused an accumulation of in-stream leaf litter (Wallace et al. 1995) and reduced fine paniculate organic matter (FPOM) (Cuffney et al. 1990, Wallace et al. 199la) and paniculate inorganic matter (ash) (Wallace et al. 1993) concentrations and export, which subsequently increased during recovery. Thus, this manipulation demonstrated the importance of invertebrates in a series of ecosystem-level processes ranging from organic matter storage to detritus processing and export. We used these

141

ecosystem-level observations to test if biotic indices corresponded to changes in ecosystem processes. Lenat (1993) has presented a biotic index for southeastern U.S. streams, called the North Carolina Biotic Index (NCBI). We applied this index to the 6-yr record of pretreatment, treatment, and recovery data available from the manipulation study described above to examine changes in stream biotic integrity. Additionally, we applied the EPT index (number of taxa belonging to the Ephemeroptera, Plecoptera, and Trichoptera [Crawford and Lenat 1989, Plafkin et al. 1989, Kerans and Karr 1994]) as a second measure of biotic integrity. We compared NCBI and EPT values for the treated stream to those calculated for a nearby reference stream during the same 6-yr period. Specifically, we address the following questions: Where impairments of ecosystem-level processes have been demonstrated, do the NCBI and EPT indices vary significantly from those of a nearby reference stream? Do NCBIs based on abundance and biomass of taxa differ? Do communities inhabiting different substrates differ in their response? Do indices track ecosystem recovery and how do recovery values differ from those of pretreatment and treatment? STUDY SITES Our two study streams were first-order and drain Catchments (C) 54 and 55 at the Coweeta Hydrologic Laboratory (U.S. Forest Service) in western North Carolina. Catchment vegetation was mixed hardwoods. A dense riparian growth of rhododendron resulted in heavy year-round shading of both streams. Elevation, catchment area, thermal regime, discharge, and aspect (southern) were similar for both streams (see Wallace et al. 1991a). Concentrations of most ions were low (50 individuals in subsamples of representative size classes. Abundances and biomass were estimated separately for mixed substrate and rockface substrates. Habitatweighted abundances and biomass for each stream were then calculated according to the proportion of rockface and mixed substrate present in each stream. Information on methods used to estimate secondary production of invertebrates during pre-treatment and the initial

Ecological Applications Vol. 6, No. 1

treatment year (1986) can be found in Lugthart and Wallace (1992). Biotic indices Two separate indices were used for assessment of stream biotic integrity in C 54 and C 55: the North Carolina Biotic Index (NCBI) and taxonomic richness of intolerant taxa (Ephemeroptera + Plecoptera + Trichoptera, EPT). EPT is a sensitive indicator of stream perturbations (Crawford and Lenat 1989, Eaton and Lenat 1991) and is widely used by various agencies as part of their environmental monitoring programs (Lenat 1988, Plafkin et al. 1989). The NCBI is based on an extensive data set of benthic stream samples collected throughout North Carolina and is designed to be specific to mountain, Piedmont, or coastal ecoregions of the southeastern United States (Lenat 1993). The NCBI is calculated as:

where S is the number of taxa, TV, is the tolerance value of the i*1 taxon, N, is density of the /'lh taxon as either abundance (numbers per square metre) or biomass (milligrams ash-free dry mass per square metre), and N, is total abundance (or biomass) of macroinvertebrates in the sample. Tolerance values range from 0 (highly intolerant taxa) to 10 (highly tolerant taxa). Taxa from our streams were assigned TVs provided by Lenat (1993) with several modifications. Taxa classified to genus (i.e., Serratella spp. [Ephemeroptera] and Rhyacophila spp. [Trichoptera]) were assigned mean tolerance values of species known to occur or likely to occur in the study streams. Chironomids (Diptera) were assigned a tolerance value of 5.7, based on a mean value provided by Lenat (1993), which may be conservative for our streams. Based on data from Wallace et al. (1991 ft), mean tolerance values of drifting chironomid taxa during insecticide applications ranged from 4.2 to 5.6. For crayfish, the tolerance value of 8.1 provided for collective Cambarus spp. is probably too high for our insecticide treatment, because C. bartonii was quickly eradicated from C 54 during treatment and did not return during recovery (Lugthart and Wallace 1992, Whiles and Wallace 1992). Thus, crayfish were excluded from NCBI calculations. Sampling and sample processing equipment was fitted with meshes of 250 u,m. Larger mesh sizes (i.e., 1 mm) are generally used in bioassessment and biomonitoring protocols (see Plafkin et al. 1989), and collections made with smaller meshes will bias results toward abundant smaller taxa such as Nematoda and Chironomidae. Further, any TVs assigned to nematodes are at best tenuous because they are rarely, if ever, included in freshwater biotic indices (D. Lenat, personal communication). Lugthart et al. (1990) found =55% of non-Tanypodinae chironomids collected in C 54 and C 55 were in the 1-mm size class (). The reference stream received no insecticide throughout the study. Data analyses Several within- and among-stream comparisons did not meet the assumption of equal variance; therefore, nonparametric methods were used for most analyses (Elliott 1977). To test for differences among multiple comparisons for specific periods (i.e., pretreatment, treatment, recovery year 1, and recovery year 2) and between streams, we used a Kruskal-Wallis ANOVA on ranks, and all pairwise comparisons were tested US-

143

TABLE 1. Average NCBIs (North Carolina Biotic Index) for invertebrate abundances and biomass on rockface and mixed substrates in treatment and reference streams with and without crayfish, nematodes, and chironomids 0.11-0.96, Mann-Whitney rank sum test, Table 4). Randomized intervention analysis (RIA) rejected the null hypothesis that no significant changes in NCBIs occurred in the treatment stream relative to the reference for pretreatment and treatment periods, and treatment and 1st- or 2nd-yr recovery for either habitatweighted abundances or biomass (Table 5). In contrast, between-stream RIA analysis did not reject the null hypothesis that no significant differences existed between streams during the pretreatment, or among re-

covery years, and indicated a strong treatment effect (Table 5). EPT index Before treatment, average number of EPT taxa did not differ among streams (Table 6). Throughout the study, EPT indices remained relatively consistent in the reference stream, ranging from 15 to 24 taxa on each collection date. In contrast, dramatic changes occurred in the treated stream (Fig. 2). Number of EPT taxa ranged from 13 to 21 taxa in the treatment stream prior to treatment and declined dramatically to 2-8 taxa following initial treatment (=day 430, Fig. 2). EPT taxa remained low throughout the treatment period and initial 6 mo of recovery in the treatment stream (summer 1989, =day 1750, Fig. 2). In midsummer 1989, EPT taxa increased sharply as recolonization occurred. Following the sharp decrease with initial treatment, the average EPT index for the treatment stream was only 5.4 during the treatment period, 10.8 during Ist-yr recovery, and increased to 16.5 during 2nd-yr recovery (Table 6). In contrast, EPT of the reference stream averaged 19 during treatment, 20.5 during Ist-yr recovery, and 18.3 during 2nd-yr recovery of the treatment stream. Treatment stream EPTs during treatment and Ist-yr recovery were significantly lower than those of pretreatment (Table 6). Second-yr recovery EPTs in the treatment stream did not differ significantly from pretreatment. Between-stream comparisons show that EPTs did not differ between treatment and reference streams during pretreatment and 2nd-yr recovery pe-

TABLE 4. Between-stream (treatment vs. reference) probability values for comparisons of NCBIs (North Carolina Biotic Index) based on abundances and biomass during different time periods of the study. Periods refer to insecticide treatment of the treatment stream and are as follows: pretreatment = October 1984 to November 1985; treatment = December 1985 to December 1988; recovery year 1 = January to December 1989; and recovery year 2 = January to December 1990. Pretreatment Treatment Recovery year 1 Comparison! 0.57 NS 0.0006* 0.12 NS Habitat-weighted abundances 0.0006* 0.61 NS 0.51 NS Habitat-weighted biomass 0.02* 0.02* 0.0006* Mixed substrate abundances 0.24 NS 0.0012* Mixed substrate biomass 0.18 NS 0.044* 0.0006* 0.33 NS Rockface abundances 0.64 NS 0.0006* 0.006* Rockface biomass t Comparisons made by Mann-Whitney rank sum test, *P < 0.05, NS = P > 0.05.

Recovery year 2 0.82 NS 0.61 NS 0.11 NS 0.96 NS 0.96 NS 0.13 NS

146

J. BRUCE WALLACE ET AL.

Ecological Applications Vol. 6, No. 1

TABLE 5. Results of Randomized Intervention Analysis (RIA) performed on NCBIs (North Carolina Biotic Index) based on habitat-weighted abundances and biomass between the treatment stream (C 54) and reference stream (C 55), testing the null hypothesis that no change occurred in the treatment stream relative to the reference stream. Comparison periods Pretreatment vs. treatment Pretreatment vs. recovery year 1 Pretreatment vs. recovery year 2 Treatment vs. recovery year 1 Treatment vs. recovery year 2 Recovery year 1 vs. recovery year 2 * = strong effects of treatment. riods. In contrast, treatment and Ist-yr recovery EPTs differed significantly from those of the reference stream (Table 6). Biotic indices and ecosystem processes Previous studies on these streams have documented several ecosystem-level consequences resulting from this top-down manipulation of the invertebrate community. For example, these heavily shaded, heterotrophic streams primarily depend on allochthonous inputs of leaf litter, which serve as the energy base of the system. Following invertebrate reduction, leaf litter decomposition within the treated stream was greatly reduced compared with pretreatment and untreated reference streams. The time required for 95% processing of leaf litter was about twice that of untreated streams (Cuffney et al. 1990, Chung et al. 1993). Decomposition rates observed for leaf litter corresponded with the EPT index patterns observed during pretreatment, treatment, and recovery in C 54. During treatment, when the EPT index was lowest, decomposition rates (days to 5% of original litter mass remaining in litter bags) decreased. Decomposition rates subsequently increased, along with EPT values during recovery (Fig. 3). Previous work in these streams has also demonstrated strong linkages between invertebrate shredders, leaf

Habitat-weighted abundances, P