Aquatic shredder insect diversity declines latitudinally toward the equator, contrary to the general latitudinal biodiversity gradient. Shredder diversity along tropical elevational gradients mimics this shredder latitudinal pattern. One of the hypotheses proposed to explain this pattern of diversity is that dispersal capacity drives variation in shredder assemblages given their low dispersal capacity in tropical streams. Additionally, tropical shredders probably have lower dispersal capacity than the rest of tropical aquatic insects, which have lower dispersal capacities than their counterparts in temperate areas. We tested this hypothesis in an elevational gradient of more than 2000 m in 16 reaches of streams distributed in the Usumacinta, Cahabon, and Polochic river watersheds. We quantitatively sampled aquatic insects and measured 12 environmental variables. We found a regional pool of 118 taxa, with 13 taxa classified as shredders, and 2 taxa of predator-shredders. Contrary to expectations, shredder rarefied richness decreased with increasing elevation, which suggests that dispersal capacity did not change with elevation. Assemblage similarity decreased with increasing distance between reaches due to low capacity to fly long distances. This relationship had a smaller slope when using the shortest spatial distances between pairs of reaches due to potential lateral scatter by flying adults. In sum, the results support the hypothesis that dispersal capacity drove aquatic shredder assemblage structure in these 16 tropical streams.

La diversidad de los insectos acuáticos fragmentadores de hojas declina latitudinalmente hacia el ecuador, contrario al gradiente latitudinal general de biodiversidad. La diversidad de este grupo a lo largo de gradientes de elevación tropicales tiene un patrón similar al patrón latitudinal de diversidad. Una de las hipótesis propuestas para explicar este patrón de diversidad de los insectos fragmentadores es que los insectos acuáticos fragmentadores tienen una capacidad de dispersión menor en zonas tropicales, menor incluso que el resto de insectos acuáticos tropicales que tienen capacidades de dispersión menores que sus contrapartes en áreas templadas. Pusimos a prueba esta hipótesis en un gradiente de elevación de más de 2000 m en 16 secciones de arroyos distribuidas en las cuencas de los ríos Usumacinta, Cahabón y Polochic. Hicimos colectas cuantitativas de insectos acuáticos y mediciones de 12 variables ambientales, simultáneamente. Encontramos 118 taxa entre todos los sitios, de los cuales 13 taxa eran fragmentadores de hojarasca y 2 taxa depredador-fragmentador. Contrario a lo esperado, la riqueza, ponderada por la abundancia, de insectos fragmentadores disminuyó con la elevación, lo que sugiere que la capacidad de dispersión no cambia con la elevación. La similitud de los ensambles de fragmentadores disminuyó con el aumento de la distancia entre sitios debido a la baja capacidad de volar largas distancias. Esta relación tuvo una pendiente menor al usar la distancia más corta entre sitios debido a la potencial dispersión lateral por adultos voladores. Por lo tanto, los resultados dan soporte a la hipótesis de que la capacidad de dispersión controla el proceso de ensamblaje de insectos fragmentadores en estos 16 arroyos tropicales.

Bayesian linear models, Central America, dendritic network structure, freshwater insect assemblages, Guatemala

América Central, Ensambles de insectos acuáticos, Guatemala, Modelos lineares bayesianos, Redes dendríticas

A goal in ecology is to understand the mechanisms driving species assemblage structure in space and time (Hubbell 2001; Soberón and Peterson 2005). Assemblage structure drivers divide into two general groups, niche-based (Losos et al. 1998; Peterson et al. 1999) and dispersal-based mechanisms (MacArthur and Wilson 1967; Hubbell 2001; Thompson and Townsend 2006). Niche-based mechanisms suggest species with similar traits share a like environment, resulting in the resemblance of species assemblages decreasing with increasing environmental dissimilarity between sites (Peterson et al. 1999; Kluge and Kessler 2011; Montaño-Centellas et al. 2021). In dispersal-based mechanisms, species assemblage structure is driven by species dispersal capacity and spatial distance between sites, resulting in the similarity of species assemblages decreasing with increasing distance between sites, accompanied by variation in the strength of the relationship between assemblage similarity and environmental similarity (Brown and Swan 2010; Ge et al. 2023). The inherent hierarchical structure of stream networks makes it possible to contrast the influence of each of these two groups of mechanisms on the structuring of aquatic insect assemblages (Grant et al. 2007; Brown and Swan 2010; Saito et al. 2015).

Aquatic insects disperse among streams in three major potential ways: a) flying along stream corridors as adults, b) flying overland as adults, and c) drifting within streams as adults and larvae (Brown and Swan 2010; Lancaster and Downes 2013). Under a niche theory framework, flying along stream corridors is the major way to colonize headwaters from main channels (Thompson and Townsend 2006; Grant et al. 2007; Brown and Swan 2010). Therefore, an expectation is that assemblage similarity decreases with lower similarity in ecological conditions, independently of the spatial distance between headwaters (Brown and Swan 2010). One assumption for this expectation to be true is that organisms do not move among headwaters by flight overland because riparian forest is a barrier for dispersal activities. However, riparian forest might not be a complete barrier for overland flying between headwaters (Macneale et al. 2005). Therefore, if flying overland occurs between headwaters, assemblage similarity decreases with increased spatial distance among headwaters. Dispersal capacities drove aquatic insect assemblage structures across large distances (i.e., >10 km) between headwaters streams in both Finland and New Zealand (Thompson and Townsend 2006; Mykrä et al. 2007). Because dispersal capacities vary among species, assemblage similarity decays with spatial distance between streams for species with low or medium dispersal capacities, with a steeper decay for the former. Strong dispersal overrides any relationship between assemblage similarity and spatial distance.

Aquatic insects are classified into six functional feeding groups (FFG) according to the predominant feeding strategy used to obtain food (i.e. shredder, collector-gatherer, collector-filter, piercer, scrapper, and predator; Cummins 1973). It is important to point out that insects from different FFGs can feed on similar food resources, such as diatoms, fungi, or animal tissue (e.g. García et al. 2016; Rosas et al. 2020), but they differ in the strategy to get that food (Cummins and Klug 1979). Some aquatic insect orders are classified almost entirely within an FFG, such as Odonata which are predominantly predators (Cummins et al. 2005; Ramírez and Gutiérrez-Fonseca 2014). While other orders of aquatic insects present two or more FFG states, such as Diptera, Coleoptera, and Trichoptera, where shredder, collector-filterer, predators, and scrapers are FFG that can be found (Cummins et al. 2005; Merritt et al. 2008). Feeding activities of aquatic shredder insects (i.e. breakdown leaf litter into smaller pieces) drive leaf litter decomposition to fine organic matter, and therefore they enable availability food for other aquatic species. Aquatic shredders consume up to 64% of stream leaf litter (Graça et al. 2015), and therefore can greatly reduce leaf litter standings stocks and downstream organic matter transport (Wallace and Webster 1996). Aquatic shredder insects occupy a central link in the food web between leaf litter (which is often the dominant source of organic matter to a forested stream) and predatory animals such as fishes and amphibians. Thus any processes that control aquatic shredder insect diversity in streams indirectly affect ecosystem functions of organic matter cycling and predator abundance.

Aquatic shredder insect assemblages have predominantly low α-diversity in tropical forest headwaters relative to temperate ones, despite γ-diversity of aquatic shredder insects in corresponding tropical regions is as high as in temperate regions (Boyero et al. 2015). This pattern is recurrent in aquatic shredder insects across tropical regions while non-shredder insects have higher α-diversity (Yule et al. 2009; Boyero et al. 2012). A hypothesis is that tropical shredders have lower dispersal capabilities than shredders at temperate streams (Boyero et al. 2015). Also, dispersal capabilities of tropical shredders should be lower than non-shredder insects given that tropical aquatic insects generally have lower dispersal capabilities relative to temperate ones (Polato et al. 2018). Variation in dispersal capacity was related to the strength of the relationship between assemblage similarity and the spatial distance between sites in aquatic insect assemblages of Brazil (Saito et al. 2015). Despite the evidence that dispersal capacity influences aquatic insect assemblage structure, we do not know how much dispersal capacity drives variation in tropical aquatic shredder assemblages, and thus determines patterns of low diversity in tropical streams.

We tested how dispersal capacity drove variation in aquatic shredder and non-shredder insect assemblages across 16 reaches of Neotropical headwaters streams in northern Central America. First, we tested the assumption that shredder richness increased with increasing elevation, and the assumption that environmental dissimilarity increased with increasing spatial distances between pairs of reaches because there is a wide environmental gradient among 16 reaches. Second, there are two expected predictions that follow if dispersal limitation is driving assemblage structures (Brown and Swan 2010), 1) assemblage similarity decreases with increasing distance between pairs of reaches (i.e., Distance-decay relationship, DDR), 2) no relationship should exist between assemblage similarity and environmental dissimilarity between pairs of reaches. If shredders have a lower dispersal capability than non-shredders, then we expected that non-shredder assemblage similarity should decrease with increasing spatial distance at a slower rate than shredder assemblage. Third, given that dispersal can occur by adults flying overland (Brown and Swan 2010; Lancaster and Downes 2013), we expected a weaker relationship between shredder assemblage similarity and spatial distance between pairs of sites than the relationship between shredder assemblage similarity and shortest watercourse distance between pairs of sites (i.e., in-network distance). An alternative hypothesis to dispersal limitation as a driver of assemblage structure is that environmental factors drive variation in aquatic insect assemblages (Brown and Swan 2010). If that was the case, we expected 1) no relationship between assemblage similarity and spatial distance between pairs of reaches, and 2) a strong relationship between assemblage similarity and environmental dissimilarity between pairs of reaches.

Most taxa were non-shredder (118) with 13 taxa classified as shredder functional feeding group (Suppl. material 1: table S2), and 2 taxa were predator-shredder. Shredder richness varied between 2 to 10 taxa by reach, while non-shredder richness varied between 8 to 50 taxa by reach. Predator groups had the largest sampled richness with 40 taxa, and the other 4 taxa were identified as collector-filter predators. Four taxa were classified as algae-piercers.

Contrary to expectation based on previous findings of positive relationship between richness and elevation (Camacho et al. 2009; Yule et al. 2009), rarefied richness, both shredder and non-shredder, did not vary with increasing elevation after weighting richness by the number of individuals collected by reach (Fig. 2). Low shredder richness could reflect low abundances e.g., Totonicapan reach with 8 individuals of shredders within 2 species. However, shredder abundance was not related to general low abundance in the Totonicapan reach given that non-shredders had a high sampled abundance (1546 individuals) relative to other reaches in El Amay or Biotopo sites (10 and 96 individuals, respectively).

Figure 2. 

Rarefied richness, both shredder and non-shredder, did not vary with increasing elevation. Dots are rarefied richness for non-shredder and shredder of 15 reaches in an elevational gradient from 156 to 2720 m asl in northern Central America. Black solid lines are the mean fit Bayesian linear model between variables with a Poisson distribution. Blue faded lines are a 1000 possible lines from the Bayesian posterior distribution of the fitted model. Third panel shows a density plot of estimated values for slopes, dashed red line indicates the zero value.

Negative DDR slopes indicated potential dispersal dynamics driving assemblage structure of shredder and non-shredder assemblages (Fig. 3). Assemblage similarity, both shredder and non-shredder, decayed with increasing Euclidean distances at similar rate. Shredder assemblage similarity decayed with increasing in-network distances, while non-shredder assemblage similarity did not vary with increasing in-network distances. The DDR relationship was 1.6 times weaker for the shredder assemblage similarity and Euclidean distance relationship (i.e., Slope = –6.4 × 10^-3, C.I. = –9.6 × 10^-3) than for the shredder assemblage similarity and in-network distance relationship (i.e., Slope = –3.9 × 10^-3; C.I. = –5.4 × 10^-3), which suggests that adult shredder dispersed overland. Assemblage similarity, both shredder and non-shredder, did not vary with increasing environmental dissimilarity. There was a positive relationship between environmental dissimilarity and increasing Euclidean distances between pairs of reaches (Fig. 4), which supports dispersal dynamics as a driver of assemblage structure across study sites.

Figure 3. 

Relationships between assemblage similarity and distance show stronger strength of dispersal role on structuring aquatic shredder insect assemblages compared to aquatic non-shredder insect assemblages. Dots are assemblage similarity and distance between pairs of 16 reaches in an elevational gradient from 156 to 2720 m asl in northern Central America. Black solid lines are the mean fit Bayesian linear model between variables with a Zero-inflated Beta distribution. Blue faded lines are a 1000 possible lines from the Bayesian posterior distribution of the fitted model. Third column shows density plots of estimated values for slopes, dashed red line indicates the zero value.

Figure 4. 

Positive relationship between environmental dissimilarity and spatial distance agrees with the predictions for an assemblage structure controlled by dispersal dynamics (Brown and Swan 2010). Dots are environmental dissimilarity and spatial distance between pairs of 16 reaches in an elevational gradient from 156 to 2720 m asl in northern Central America. Black solid lines are the mean fit Bayesian linear model between variables with Normal distribution. Blue faded lines are a 1000 possible lines from the Bayesian posterior distribution of the fitted model.

Distance-decay relationships (DDR) documented here indicate lower dispersal capacity for shredders than non-shredders along the river network. Aquatic insect richness did not vary with increasing elevation. Shredder state of the functional feeding group trait had no effect on DDR when using Euclidean distance, while it turned a DDR with a slope of zero into a DDR with a negative slope when in-network distance was used. Shredder DDR decayed 1.6 times faster with Euclidean distance between pairs of sites than with in-network distance between pairs of sites. Assemblage similarity, both shredder and non-shredder, did not change with environmental variation between pairs of reaches.

Patterns of shredder richness across elevation did not match expectation based in previous studied tropical elevational gradients; richness did not vary with increasing elevation. Previous findings have shown an increasing shredder richness with increasing elevation in the humid tropics of Australia and Malaysia (Camacho et al. 2009; Yule et al. 2009). One potential source of our contrasting result may be the use of rarefied richness because previous work did not weigh richness by abundance, despite the known influence of sample size on richness (Colwell et al. 2012). Lack of variation in shredder richness along elevational gradients has been observed in Kenya and Panama, even when richness was not weighed by abundance (Dobson et al. 2002; Camacho et al. 2009). A plausible explanation for the general low richness of shredders could be misassignment of taxa as shredders, in combination with low leaf litter availability in quantity and quality (Dobson et al. 2002). Here, misassignment was unlikely because the assignment to functional feeding groups was based on information specific for Neotropical taxa (Cummins et al. 2005; Ramírez and Gutiérrez-Fonseca 2014).

Species richness at a site is influenced by sample size (Colwell et al. 2012). Therefore, we cannot find more species than the number of individuals sampled at any site. Weighting richness by abundance (i.e. rarefied richness) can change where the peak of richness is in an elevational gradient, as it is observed for richness for vegetation with a peak of richness at low elevations instead of mid-elevations (Ibanez et al. 2016). Given this, to ensure comparability, it is recommended that richness between localities be done by comparing rarefied richness (e.g. Griffiths et al. 2021).

Negative slopes of the relationship between shredder assemblage similarity and spatial distance suggest that shredder assemblage structure may be partially driven by dispersal dynamics (Brown and Swan 2010). Shredder assemblages had negative DDR slopes, which suggest that shredders had low to moderate dispersal capacity given that high dispersal capacity weakens or eliminates this relationship (Thompson and Townsend 2006; Ge et al. 2023), as in the case of non-shredder assemblages. Furthermore, a shallower DDR slope with Euclidean distance than with in-network distance suggests that lateral dispersal may control similarity among shredder assemblages (Brown and Swan 2010). One caveat is that we did not measure directly the distance that shredder species dispersed at the study sites. However, overland dispersal movement is a plausible explanation for the observed assemblage similarity between sites, as Euclidean distances between sites (e.g., 200 and 900 m) were shorter than distances over watercourses in the corresponding sites (e.g., 1.5 to 9 km). We know that aquatic insects can fly from 200 to 900 m overland and over watercourses (Macneale et al. 2005; Finn and Poff 2008). Additionally, estimated DDR slopes for shredders were 25 times smaller than slopes estimated for freshwater benthic macroinvertebrate assemblage similarity across 91 sites in North America (Brown and Swan 2010), suggesting detection of small effect size.

Assemblage similarity, for both shredders and non-shredders, did not change as a function of environmental dissimilarity between pairs of sites. This pattern in combination with the positive relationship between environmental dissimilarity and Euclidean distance between pairs of reaches suggests that local environmental factors did not drive assemblage structure (Brown and Swan 2010). These results contrast to previous findings in temperate zones where environmental factors and dispersal drove assemblage structure (Thompson and Townsend 2006; Mykrä et al. 2007; Brown and Swan 2010; Heino and Tolonen 2017). The positive relationship between environmental dissimilarity and Euclidean distance between sites, as expected, rules out the possibility that the studied environmental gradient was too small. Null relationships between assemblage similarity and environment have been previously reported for tropical aquatic insects with low dispersal capacity (Boyero et al. 2015; Saito et al. 2015). Altogether, these absences of relationships suggest that dispersal limitation, rather than environmental factors, was the main driver of species assembly structure.

This work supported the hypothesis that dispersal capacity drives shredder assemblage structure in Neotropical streams. We found evidence that aquatic shredder insects have lower dispersal capacity than aquatic non-shredder insects, and flying overland could be a pathway to connect assemblages between headwaters streams. To date, studies have focused on assessing environmental variables as drivers of aquatic insect assemblage structure (Heino et al. 2003; Heino and Tolonen 2017). However, there is accumulating evidence that neutral processes such as dispersal control the assemblage structure (Thompson and Townsend 2006; Boyero et al. 2015; Saito et al. 2015). In the case of shredder assemblages, we assessed dispersal capacity as a driver of assemblages of a small number of reaches. Negative slopes of DDRs, together with the lack of relationship between assemblage similarity and environmental dissimilarity, suggest that dispersal limitations contribute to structuring shredder assemblages in these streams, despite the small number of sites studied. Therefore, future work should assess a) how far do aquatic shredder insects fly overland and over watercourses? and b) how much gene flow exists among shredder populations that are spatially close overland, but far in terms of stream network distance?

We thank administrators, owners, and rangers for the permission, logistic facilities, and access to the field sites. We thank Luis Velázquez for helping us to get to some nice streams in Huehuetenango. We also thank the University of San Carlos of Guatemala, the Fulbright-LASPAU Program, the Russell E. Train Education for Nature program of the World Wildlife Fund, the Flathead Lake Biological Station, and the University of Montana for the scholarships and funding. Jim Elser, Rosa Jiménez, Laurel Genzoli, and Maury Valett provided comments that improved this manuscript. We thank the feedback provided by an anonymous reviewer which improved our manuscript. Sampling done for this work followed the rules and regulations of Consejo Nacional de Areas Protegidas -CONAP- from Guatemala. We appreciate the support provided by the Dirección de Valoración y Conservación de la Diversidad Biológica, CONAP, especially from José Luis Echeverría.