Aquatic Crustaceans in the Driest Desert on Earth: Reports from the Loa River, Atacama Desert, Antofagasta Region, Chile

Global Advances in Biogeography brings together the work of more than 30 scientific authorities on biogeography from around the world. The book focuses on spatial and temporal variation of biological assemblages in relation to landscape complexity and environmental change. Global Advances embraces four themes: biogeographic theory and tests of concepts, the regional biogeography of individual taxa, the biogeography of complex landscapes, and the deep-time evolutionary biogeography of macrotaxa. In addition, the book provides a trove of new information about unusual landscapes, the natural history of a wide array of poorly known plant and animal species, and global conservation issues. This book is well illustrated with numerous maps, graphics, and photographs, and contains much new basic biogeographical information that is not available elsewhere. It will serve as an invaluable reference for professionals and members of the public interested in global biogeography, evolution, taxonomy, and conservation.


Material and methods
The Loa River originates in the Andes Mountains close to the boundary with Bolivia, and flows first from north to south, proceeds in a westerly direction, changes back to a northerly direction, and finally flows west to its confluence with the Pacific Ocean (Fig. 1). We compiled information from two two data sets, one collected during field studies in January 2008 that included nine sites along the Loa River (De los Ríos et al., 2010;Table I), and the other from field studies in 1980 from lower reaches of the Loa River between the outlet to the middle zone. The latter dataset included C. caementarius habitat (Alfaro et al., 1980; Table  II). Altitude, in meters above sea level, was recorded for all sites. We tested for relationships between species richness and two physical variables, salinity (using a YSI-30 sensor) and altitude, using correlation coefficients (Rho-Spearman), calculated in the software SPSS v.12.
Crustacean community structure was explored using a co-occurrencenull model analysis, which tests whether species co-occur less frequently than expected by chance (Gotelli, 2000). A checkerboard score ("C-score") was calculated based upon an absence/ presence matrix, representing a quantitative index of co-occurrence. We used this method for analysis and evalutated results in relation to those of other studies (Tondoh, 2007;De los Ríos et al., 2008;De los Ríos-Escalante, 2011). A community may be structured by competition when the Cscore is significantly larger than that expected by chance (Gotelli, 2000(Gotelli, , 2001. In order to determine whether a particular score is statistically significant, a set of randomizations of the species occurrence data are performed and a null distribution for the coexistence index is created. Gotelli & Entsminser (1997) Tiho &Johens (2006), andTondoh (2007), suggested the following three statistical models for creating randomized communities, with the species placed in rows and the sites in columns: 1. Fixed-Fixed Model. In this model, the row and column sums of the matrix are preserved. Thus, each random community contains the same number of species as the original community (fixed column) and each species occurs with the same frequency as in the original community (fixed row). 2. Fixed-Equiprobable Model. In this algorithm only the row sums are fixed and the columns are treated as equiprobable. This null model considers all the samples (column) as equally available for all species. 3. Fixed-Proportion Model. In this algorithm species occurrence totals are maintained as in the original community, and the probability that a species occurs at a site (column) is proportional to the column total for that sample.
All three of these models exhibit fairly reasonable combinations of Type I and Type II error rates, although model 3 has a high Type I error rate (false positives) using the C-score index, with differences in underlying assumptions and behaviour (Gotelli, 2000). The fixed-fixed model is suggested to be most appropriate for island species lists, in which species-area effects are expected, while the fixed-equiprobable model would be most appropriate for standardized samples in a homogeneous environment (Gotelli, 2000). The fixedproportional algorithm represents an intermediate model, which might be most appropriate in our system due to habitat connectivity and heterogeneity, as well as differences in depth and width along the river. Differing results among models can provide insights into community structure. The null model analysis was using the software ECOSIM, version 7.0 (Gotelli & Etsminger, 1997).

Results and discussion
Our results revealed the presence of a small number of crustacean species in the Loa River (Table 1). . In the Quillagua River, no crustacean species were found, but the introduced fish species H. kochi (González & Watling, 2001) Table 2. Geographical location, altitude, conductivity, salinity, and species reported for the studied sites in the Loa River during the 1980 sampling period.
Among data collected in 2008, Spearman rho correlation values indicated no relation between species number and salinity (r = -0.39; P = 0.149) however, a significant and positive relationship was detected between species richness and altitude (r = 0.61; P = 0.041) ( Table 2). In contrast, among data collected in 1980, the Spearman rho correlation values indicated no relation between species number and salinity (r = -0.391; P = 0.149) or between diversity and altitude (r = 0,189; P > 0,05) (  Table 3. Results of correlation and null-model analyses. Correlation coefficients are provided between species richness and conductivity, salinity, and altitude, respectively, along the Loa River (sampling period 2008). The null-model analysis (see text) suggests that crustacean community structure is random.
The results of the null model analysis among all simulations within 2008 data revealed that crustacean community composition appears to be random (Table 4). However, small sample size can mask other underlying patterns (De los Ríos-Escalante, 2011). A different situation was observed within the 1980 data, where the fixed-proportional denoted the presence of regulatory factors, and the fixed-equiprobable analysis indicated a weak presence of regulator factor presence (Table 4).
Unfortunately chemical and other physical parameters were not measured in situ; however, human influences on the river may be potentially regulatory factors. Such influences have been documented on central Chilean rivers (Figueroa et al., 2003). The upper reaches of the Loa River are subject to lower levels of anthropogenic impact, given that human population size is small in relation to that in the lower reaches, which are subject to urbanization and and mining activities (Alvarez, 1999;Melcher, 2004).
The present literature on Chilean rivers only contains descriptions of invertebrate species associations in south-central rivers, and most studies attempt to use such information for bio-indication of water quality (Figueroa et al., 2003(Figueroa et al., , 2007. These studies generally report differences in macroinvertebrate assemblages (mainly insects and crustaceans) in relation to water pollution along the river courses (Figueroa et al., 2003(Figueroa et al., , 2006. These studies indicate that the fluvial aquatic fauna of Chilean is regulated by deterministic factors, namely water quality as influenced by the level of human alteration (Figueroa et al., 2003(Figueroa et al., , 2006. However, Figueroa et al., (2003Figueroa et al., ( , 2006 only described the riverine biota to the family level, and this lack of species-level data precludes more precise statistical treatments of community structure (Gotelli & Graves, 1996;Gotelli & Ellison, 2000;Jaksic, 2001). Our study included only crustaceans, but emphasized species-level taxonomy, whereas Figueroa et al., (2003Figueroa et al., ( , 2006 studied all benthic invertebrates including aquatic insects, which are diverse compared to crustaceans. Future studies of riverine community structure should expand upon our studies in order to include aquatic insects at the species level. Aquatic insects are likely to ecologically interact with crustaceans in fluvial benthic communities, in turn affecting abundances and patterns of co-occurrence among both taxonomic groups (Parra et al, 2001).
Biogeographically, the presence of the amphipods H. fossamanchini and H. kochi in the Loa River has been previously described by Gonzalez (2003) and Jara et al. (2006). However, those descriptions did not specify details about the localities at which the species were found. Thus, our study presentsnew information to the knowledge of crustacean distribution. The absence of specimens of the northern river shrimp, C. caementarius, indicates that this species is threatened by excessive fishing as human food (Jara et al., 2006), and probably also by the presence of exotic fishes, such as Gambusia affinis¸ Oncorhynchus mykiss and Salmo trutta. All of those fish species are active predators on native aquatic invertebrates (Leyse et al., 2005).
Regarding the presence of ostracods in Chile, there are currently only records of H. panningi and Cucacandona spp., which also have been reported for other South American inland waters (Martens & Behen, 1994). Spatial differences in community composition were detected, with the presence of ostracods related to the absence of amphipods in the lower reaches of the Loa River, reaches that exhibit relatively high salinity and conductivity (Table  3). In contrast, the high-altitude zones of the Loa, as well as the Salvador River, have relatively lower salinity, with species associations between the amphipods H. fossamanchini and H. kochi (Table 1). The presence of microcrustaceans, specifically copepods and cladocerans in Conchi Reservoir, corroborates their distribution as reported in the literature (Araya & Zúñiga, 1985;Reid, 1985, Ruiz & Bahamonde, 1989. Although distributional data denote a segregation of species into low and high altitude reaches (Table 1), with significantly higher species richness at higher altitudes, the null model results suggest that crustacean species associations are random (Table 3). These results are seemingly in disagreement; however, the low total species richness and small number of study sites may preclude the detection of a true underlying pattern (De los Ríos-Escalante, 2011).
Nevertheless, the negative relationship between species richness with salinity and altitude (Table 4), is explained by predation by C. caementarius on microcrustaceans (Alfaro et al., 1980;López et al., 1986). This may represent a top-down trophic cascade that affects community structure. This scenario would be similar to descriptions of Chilean (Soto et al., 1994;De los Ríos, 2003) and Argentinean (Reissig et al., 2004) Patagonian lakes. Considering the present results, two main forces likely regulate aquatic community structure: salinity variation and predator presence. Such a combination of factors have been reported in shallow Andean mountain saline and sub-saline lakes (Hurlbert et al., 1986; De los Ríos-Escalante, 2011).
Our results contribute to understanding of the community ecology of inland water invertebrates in northern Chilean streams. Unfortunately, these ecosystems have been understudied to date due primarily to problems of accessibility. Further work on the crustacean and other aquatic faunae of northern Chilean rivers is clearly needed.  Table 4. Results of correlation and null-model analyses. Correlation coefficients are provided between species richness and conductivity, salinity, and altitude, respectively, along the Loa River (sampling period 1980). The null-model analysis (see text) suggests that crustacean community structure is random.

Acknowledgements
The present study was supported by a grant to PR from the Research Direction of the Catholic University of Temuco (Funding for Development of Limnology). Also, we gratefully express our appreciation to Eliana Ibáñez, Elizabeth Escalante, Luis Escalante, and Elisa Pistán for their helpful logistical assistance during the field work.