INTRODUCTION

Drift is the transport of organisms downstream of a river or stream (Brittain & Eikeland, 1988). To explain this phenomenon, it was originally proposed that the drifting fraction in excess of the carrying capacity of the benthic community (^{Müller, 1954}; Waters, 1961); and that it served to regulate the populations of aquatic organisms (Müller, 1954). )hat there are three types of drift of aquatic macroinvertebrates: beht ) and catastrophic. Waters (1972) also defined the drift of organisms as the active entry into the water column, or as the passive erosive force of the current, considering then that the drift has a behavioral component and an accidental one (Minshall & Winger, 1968; Müller, 1974; Wiley & Kohler, 1984). An increase in drift abundance can be associ nf the substrate, high or low water discharges, increase in temperature and pulses of different pollutants. Therefore, drift dynamics in aquatic macroinvertebrate populations becomes a possible descriptor of lotic ecosystem alteration (Tamaris-Turizo et al., 2013).

Most stream organisms enter drift during at least some period of thei2), with immature insect states being the dominant components (Brittain & Eikeland, 1988). The drifnisms is a factor of great importance in the metabolism of the ergy subsidy that it supplies downstream (Allan, 1995; Rodríguez-Barrios et al., 2007; Tamaris-Turizo et al., 2013). There have been numerous studies that f am to lay eggs, as a way to compensate for larval drift (^{Hershey et al., 1993}). Even mayfly males, feminized by mermithids (Nematoda: Mermithidae), modify their tream, like females (Vance, 1996).

Many biotic and abiotic variables have been cited as influencinity: flow, photoperiod, water chemistry, benthic densities, life cycle stages, and the presence of predators (Brittain & Eikeland, 1988). Some invertebrates drift in order to escape predators; however, drifting to esc makes them susceptible to others (Peckarsky et al., 2002). For certain aquatic organisms, for example fish and larvae of net-building caddisflies (Brit, drift provides an important food source.

In Argentina there a drift assemblage of river and stream invertebrates only in Córdoba (Corigliano et al., 1998; Oberto et al., 2004; ^{Barbero et al., 2013}; Zanotto-Arpellino et al., 2015). In

NW Argentina, only the percentage composition of the drift in two streams in Tucumán was reported (Molineri, 2008).

Changes in the structure and functionin ving ecosystem have been reported in numerous rivers of the world that have suffered the introduction of salmonids, causing significant losses of biodiversity and acting at levels as disparate as genes and the cycle of nutrients (Townsend, 2003). The invaded sites in Argentina show a notable decrease (or disappearance) of native fish (Pa2) and of large and mobile invertebrates (^{Buria et al., 2007}; Molineri, 2008, 2012). In Tucout (Oncorhynchus mykiss), native to Nt ent and widely introduced fish, since the beginning of the 20th century in almost all the Salí sub-basins (Fernández & Fernández, 1998). Its impact on the aquatic community has been studied in one stream (

The objective of this work was to provide a the drift assembly in mountain streams from Yungas of Tucumán and to analyze the differences between sites with and without rainbow trout.

MATERIAL AND METHODS

Study area

The present work was carried out with material collected from ten sites from eight rivers belonging to five different sub-basins in the province of Tucumán: Vipos sub-basin (Chasquivil and Ancajuli streams); Lules sub-basin (Grande and Anfama streams); Medina sub-basin (Las Pavas streams); Los Sosa sub-basin (Azucenas stream); Río Chico sub-basin (Cochuna and Medina streams). All of them belong to the endorheic Salí-Dulce basin. The studied reaches have mountain stream characteristics, with a marked slope, high current velocity, coarse substrate; dissolved oxygen

concentration close to saturation and closed vegetation characterized by Yungas species (^{Sirombra & Mesa, 2010}; Pero & Quiroga, 2019). These courses presented clean, clear, oligotrophic waters ischarges occur in spring- summer (November to April) and the lower in autumn- winter (May to October).

Drift samples

In areas of monsoon climates, such as those of the Yungas region in northwestern Argentina, during the dry season, stability in the flow regime results in high richness, density and biomass of aquatic invertebrates in the benthos (^{Romero et al., 2011}). In contrast to the rainy season where abundance of aquatic inv tent floods (Flecker & Feifarek, 1994; Molineri, 2010). This fluctuation is reflected in the drift assemblage and it has been reported for the region that the biotic effects (predation by trout) are masked by the abiotic (impove nthos by drag) during the flood season (Molineri, 2008). For this reason, drift samples were taken only during the low water season (April to November), between 11 a.m. and 2 p.m. ffects (crepuscular drift increase; ^{Douglas et al., 1994}), with two nets placed simultaneously for a specified period of time. A total of 22 samples corresponding to ten sites in eight streams were studied (Table I), the streams were sampled in different years: Chasquivil (2003), Las Pavas (2008), Ancajuli (2007), Anfama (2007), Grande (2007), Azucenas, Medina (2010) and Cochuna (2010). In Chasquivil and Azucenas, two contiguous sections were studied, and therefore we have a lower section and an upper section for each one. In the lower sections there were trout in high densities but not in the immediately higher sections, due to the presence of waterfalls that the fish cannot overcome. The driftnets used have a square opening (30 x 30 cm) and a mesh opening of 300 |j. At each sampling site, two consecutive samples were taken over time (with two replicates each), leaving the nets one hour at each opportunity (except in Medina, Anfama and Azucena-upper section, with one sample each). The collected material was fixed and preserved in 96° ethanol in plastic bags. Invertebrates were separated from the rest of the organic material using a light stereoscope (Olympus SZ2-ILST) at 10x. The taxonomic identification of the individuals was carried out at the family le taxonomic keys (Domínguez & Fernández, 2009). Biomass was calculated indirectly, using the volume of alcohol displaced by each sample in a small graduated cylinder (5 ml). All the material is deposited in IBN (Tucumán, Argentina).

Trout density at the studied sites was determined qualitatively, classifying the sites into three categories: 1) absence of trout; 2) low density and 3) high density (Table I). Sites with absence of trout were considered to be those that did not present fry plantings in the last ten years, and in which no specimens were observed or captured. Those of category 2 (low density) were the ones that were sown in recent years, or in which no more than two specimens of rainbow trout were observed per 25 m section recorded (ten sections per site were registered). The sections in which trout were not observed, but their presence was indicated by locals and park rangers (e.g, Las Pavas), also fell into this category. Finally, category 3 sites presented three or more trout (observed or captured) per 25 m section, although in the majority more than five large specimens (> 30 cm total length) were observed per section.

Data analysis

The macroinvertebrate abundance data were transformed to the number of individuals per unit volume ("density”, ind./m3), except for Chasquivil, since there was no data on filtered volume for each drift-net. In order to be able to compare this site with the others, a new transformation was performed to percentage data, which was used for an independent analysis. Therefore, two matrices were analyzed: one with drift density data at eight sites (see supplementary data (Appendix 1) available at https://ibn.conicet.gov.ar/recursos/) and the other with percentage abundance data (percentage of each taxon to the total drift per site) at ten sites (the previous eight plus the two sections in Chasquivil, (Appendix 2) also available at https://ibn.conicet.gov.ar/ recursos/).

Drift density data (34 OTUs x 8 sites) were analyzed using NMDS (non-metric multidimensional scaling; ) to study the ordering of sites according to the densities of drifting taxa. The incidence matrix was transformed by the natural logarithm LN (x + 1) to homogenize the variance. Note that there are more OTUs (34) than taxa (31) because larval and pupal Chironomidae and Simuliidae, as well as larval and adult Elmidae, were treated separately.

The following indices were calculated for each sample: Richness, Shannon (H’) and Simpson (D). The similarity between samples was evaluated using the t test (densities) or the Wilcoxon test (percentage data and values of the diversity indices). Comparisons were made by classifying the samples into two categories (0/1). Two classifications were carried out: 1) absence (0) or presence (1) of trout, without taking into account the densities of the salmonid; and 2) sites without trout and those with low densities of this predator (0) were fused in a single category, and on the other hand, sites with high densities of trout (1). Three sites (Chasquivil and Azucenas in their upper reaches, plus the Grande stream) did not present trout; four rivers (Las Pavas, Medina, Cochuna and Anfama) presented trout in low density; and the rest of the sites (Chasquivil and Azucenas in their lower reaches, and Ancajuli) presented high density of trout. All analyses were performed using the Infostat statistical software (Di Renzo et al., 2013).

RESULTS

Drift general description

A total of 31 families of aquatic invertebrates were found in the following taxa: Ephemeroptera, Diptera, Trichoptera, Plecoptera, Coleoptera, Collembola, Hydrachnidia, Nematoda, Oligochaeta, Ostracoda, and Copepoda. Since the larvae and pupae of the families Chironomidae and Simuliidae (Diptera), and the larvae and adults of Elmidae (Coleoptera) were treated independently, the total of OTU's amounted to 34 (Appendices 1 and 2). The most frequent taxa were:

Table I: Diversity indices in ten studied sites. R= family richness, ShaW= Shannon diversity index, Simp= Simpson diversity index. LR= lower reach, HR= higher reach. (1) nindicates number of drift samples analyzed. (2) Trout density categories: 1 (absence), 2 (low density), 3 (high density). 

Baetidae, Chironomidae and Simuliidae larvae (in all sites), Hydrachnidia (in all but one site), followed by Chironomidae mpupae, Elmidae, Collembola and Leptohyphidae larvae. Regarding abundance, the Chironomidae larvae reached the highest value with 39% of the total samples followed by Hydrachnidia with 30%, Baetidae with 11%, Chironomidae pupae with 6% and Simuliidae larvae with 4%.

Drift density and diversity

The ordering of the samples according to the density of the drifting taxa resulted in three groups of sites, coinciding with the presence and density of trout (Fig. 1). Sites without trout were ordered to the negative side of axis 2 along with sites that had trout at low density. Sites with high trout densities are located on the positive side, axis 2.

At high trout densities (more than three trout per 25 m stretch), a decrease in abundance of some groups with large individuals (Baetidae, Gripopterygidae and Leptoceridae) was observed, but by the the contrary, those with smaller individuals (Diptera, Hydrachnidia, Nematoda and Oligochaeta) increased (Fig. 2).

The sites were not significantly different in their diversity indices (Table I). The richness, diversity and dominance did not vary in agreement with the density of trout, except when comparing lower and upper sections of the same river (Azucenas and Chasquivil). In both rivers the indices were lower for the lower sections, which presented high trout densities.

DISCUSSION

The taxa found in the drift were similar to those reported in the Yungas river benthos ; ^{Romero et al., 2011};) and are commonly found drifting in other mountainous regions of South America (e.g., in Colombia). Also in Colombia,) reported a high frequency and abundance of Hydracarina, which is in agreement with our findings. In any case, insects clearly dominated, as it occurs in other latite al., 2013).

Our results show that in Yungas streams notable differences in macroinvertebrate drift exist, depending on the density of salmonids. These differences (decrease in density of large and conspicuous taxa) are not surprising since rainbow trout is a visual predator with a poorly specialized diet ; completely disappear even if there are trout in very low ^{Albariño & Buria, 2011).} However, we did not expect the densities (;, absence of impact in cases of low trout density, since it). This could be due to the advantage of insects has been shown that even chemical traces of salmonids for presenting life stages in different habitats, being in the water can significantly decrease insect drift terrestrial as adults, generating a constant influx of (Peckhis lack of effect on individuals as a result of oviposition and drift from invertebrates does not coincide with reports for tributaries small enough not to harbor trout. Furthermore, native aquatic vertebrates (fish and amphibians), which the availability of in situ shelters and the size of

invertebrate populations can be considered much larger than for aquatic vertebrates.

Fig. 1: NMDS plot of 8 sites with density data. Site icons indicate: absence of trout (open triangles), low density of trout (gray triangles) and high density of trout (black ircles). Minimum convex polygons are showed for better visualization. Codes: LP1A, LP1B, LP2A & LP2B (Las Pavas); AZa1A, AZa1B, AZa2A & AZa2B (Azucenas lower reach); AZup6A & AZup6B (Azucenas higher reach); ME1 (Medina); COCH1 & COCH2 (Cochuna); ANC6A & ANC6B (Ancajuli); ANF1 (Anfama); RG1 & RG2 (Rio Grande). See supplementary data available at https://ibn.conicet.gov.ar/recursos/. 

Fig. 2: Rank-abundance curves for the macroinvertebrates in drift from streams without trout (first panel), with low density of trout (second panel), and with trout in high densities (third panel). Abundance is expressed as percentage. 

Regarding the diversity indices, we believe that the absence of significant differences reflects the high beta diversity of the area (; ), since when comparing the sections with and without trout of the same river (lower and upper sections in Azucenas and Chasquivil) changes in the values of richness, dominance and diversity were observed. Some works pointed to a nocturnal shift in the drift of larger insects in the presence of trout ( ; ^{McIntosh et al., 2002}). Future studies regarding the diel periodicity of the drift may give additional clues to understand the effect of invasive trout in Yungas.

ACKNOWLEDGMENTS

To the Faculty of Natural Sciences and Institute M. Lillo, to the Miguel Lillo Foundation and the Neotropical Biodiversity Institute for the workplace and equipment. We also appreciate the comments of Carolina Nieto, Hugo Fernández, Eduardo Domínguez and Ueso Montero in a preliminary version of the manuscript. Comments of two anonymous reviewers are greatly acknowledged. The following projects partially funded this study: CIUNT Project 26/G416 (H.R. Fernández), PIP0330 and PICT1067 (E. Domínguez) and PIP0845 (CM). CONICET is thanked for the constant support.