The zoogeomorphology of case-building caddisfly: Quantifying sediment use

: Caddisfly (Trichoptera) larvae are an abundant and widespread aquatic insect group characterized by the construction of silk structures, including nets and cases. Case-building caddisfly have the potential to modify the sorting and mobility of sand and fine gravel via: (1) case construction, resulting in altered sediment properties; (2) transporting sediment incorporated into cases over the river bed; and (3) changing the structure of river beds via burrowing activity. To investigate these mechanisms, it is necessary to understand the mass, size distribution and spatial variability of sediment use by case-building caddisfly larvae. We quantified the mineral sediment used by individuals and communities of case-building caddisfly in 27 samples, from three sites on a gravel-bed stream. The mass and size distribution of sediment in individual cases varied between taxa (mass = 0.001 – 0.83g, D 50 = 0.17 – 4mm). The mean mass of sediment used by the caddisfly community was 38 gm (cid:1) 2 and varied locally. Sediment use was predominantly coarse sand ( D 50 = 1mm). 64% of sediment use was attributable to Agapetus fuscipes (Glossosomatidae). Due to within-species variability in case mass, the abundance of most taxa, including A. fuscipes , was only weakly associated with the mass of sediment used by this species, at the river scale. Whilst the caddisfly community used a small percentage of the total sediment available (average 2.99% of the 1 – 1.4mm size fraction), A. fuscipes used more fine sediment in their cases at sites where it was more available. Despite variability in local habitat, all sites supported diverse case-building caddisfly communities utilizing mineral sediment. Consequently, geomorphological effects of case-building caddisfly are potentially widespread. The results provide novel insights into the specific grain sizes and quantities of fine sediment used by caddisfly larvae, which represents an important step towards understanding their zoogeomorphic activities. © 2019 The Authors. Earth Surface Processes and Landforms Published by John Wiley & Sons Ltd


Introduction
While sediment dynamics in rivers have traditionally been considered a largely physical phenomenon (Reinhardt et al., 2010), it is increasingly recognized that plants and animals can have substantial effects on hydraulics, bed sediment sorting and sediment transport processes (Rice et al., 2012;Statzner, 2012;Wilkes et al., 2018). A growing body of research has demonstrated the importance of biogeomorphology (Viles, 1988) in rivers, particularly for plants and larger animals, such as fish and mammals (Statzner, 2012). For example, both living plants and dead wood effect instream hydraulics and sediment stability (e.g. Cotton et al., 2006), to the extent that they often have large-scale effects on channel planform and longitudinal profiles (for reviews see Montgomery and Piégay, 2003;Corenblit et al., 2007;Wohl, 2013;Gurnell, 2014). Several reviews have highlighted the geomorphic effects of animals in rivers (Moore, 2006;Rice et al., 2012;Statzner, 2012). For example, the construction of beaver dams alters the hydrology, geomorphology and ecology of smaller rivers and riparian zones (Naimen et al., 1988;Gurnell, 1998;Butler and Malanson, 2005). Salmonid spawning disturbs gravel substrates, reducing their stability (e.g. Gottesfeld et al., 2004;Hassan et al., 2008;Buxton et al., 2015;Fremier et al., 2017). In addition, foraging by benthic-feeding fish can increase sediment mixing and fine sediment suspension (Pledger et al., 2014;Huser et al., 2016), affecting both gravel and sand substrates (Statzner et al., 2003;Pledger et al., 2017).

Zoogeomorphology of riverine invertebrates
Invertebrates have also been shown to have considerable effects on geomorphology, making up for their smaller size (and lower per-capita geomorphic effects) with high abundances and a diverse range of behaviours (Wallace et al., 1993;Albertson and Allen, 2015). In terrestrial ecosystems, some ants are major bioturbators of soils and have been found to excavate >1 tonne of soil per hectare (Seal and Tschinkel, 2006); similar processes exist in lakes (Krantzberg, 1985) and marine environments (Cadée, 2001). In rivers, invertebrates have been shown to both erode and stabilize sediment (Statzner, 2012). Predatory stonefly larvae can transport substantial quantities of sand from gravel interstices (200-400 kg m À2 year À1 ; Statzner et al., 1996), with reduced prey availability increasing foraging activity, and therefore erosion potential (Statzner et al., 1996;Zanetell and Peckarsky, 1996). Similarly, increased erosion of sand-sized sediments is associated with the foraging activity of freshwater shrimps (Pringle et al., 1993;Pringle and Hamazaki, 1998) and mayfly larvae (Soluk and Craig, 1990). Burrowing invertebrates including Asellidae (crustaceans), Tubificidae (worms) and Chironomidae larvae (non-biting midges) can affect the structure and porosity of the bed sediment matrix, via bioturbation, sediment reworking and structure building (Mermillod-Blondin et al., 2003Mermillod-Blondin and Rosenberg, 2006). Crayfish potentially increase the entrainment of fine sediments through a combination of locomotion (foraging and fighting) and burrowing activities (Johnson et al., 2010;Harvey et al., 2014;Rice et al., 2014). Rice et al. (2016) attributed a 32% (474 kg) increase in monthly base flow suspended sediment loads to crayfish activity.

Caddisfly zoogeomorphology
Caddisfly (Trichoptera: Insecta) are one of the most diverse orders of aquatic macroinvertebrates globally, with over 14 500 species (Zhou et al., 2016). Caddisfly typically spend their larval stages in aquatic environments. Individual taxa are small (usually <30 mm in length) but may have disproportionately large zoogeomorphic effects due to their use of silk (Albertson and Allen, 2015). Caddisfly silk, a tough fibrous protein, has high specific strength and ability to stretch, and can more than double in length before breaking (Brown et al., 2004;Tszydel et al., 2009). Silk allows caddisfly larvae to create a wide variety of structures and Wiggins (1996) classified larvae into five groups . These are larvae which: (i) construct retreats and silk filtration nets; (ii) bind together fine sediments into portable tubular cases; (iii) build domed (saddle-shaped) cases; (iv) construct purse-type cases, primarily composed of silk; and (v) are free living, building a case only for pupation.
Caddisfly larval structures can have measurable zoogeomorphic effects. The nets and retreats of Hydropsychidae ( Figure 1i) can stabilize gravel beds by binding grains together, increasing their effective size and mass, and therefore the force required to move them ( Figure 2N). Hydropsychidae larvae have been found to increase the critical shear stress required to initiate gravel movement by 10-40% (Statzner et al., 1999;Cardinale et al., 2004;Johnson et al., 2009;Albertson et al., 2014a). This consolidation can be increased further by the presence of multiple species (Albertson et al., 2014b(Albertson et al., , 2019. Furthermore, caddisfly can stabilize cobbles (Takao et al., 2006), with a single retreat increasing the force required to move a particle two to threefold (Nunokawa et al., 2008).
However, caddisfly produce a range of structures ( Figure 1) and to date, only the zoogeomorphic effects of nets and retreats have been considered. Most caddisfly species  utilize fine sediment (~0.1 to 5 mm; Hansell, 1968a;Hansell 1968b;Tolkamp, 1980;Statzner et al., 2005;Okano and Kikuchi, 2012;De Gispert, 2018) in the construction of mobile cases, fixed retreats or pupal cases at some point in their life cycle (Wiggins, 1996). We propose three potential mechanisms by which case-building caddisfly larvae may influence sediment transport and hydraulic processes in rivers (Figures 2A-C): (A) Caddisfly may modify the entrainment characteristics of sediments incorporated into their cases ( Figure 2A). Case construction involves binding sediment particles together with silk (Hansell, 1968a;Okano et al., 2016), producing a composite particle of reduced density and relatively large volume. The shear stress required to move caddisfly cases varies with species, case size and whether the case is occupied (Waringer, 1993;Statzner and Holm, 1989;Otto and Johansson, 1995), but there has been no systematic examination of the mobility of cases relative to their constituent grains. (B) Case-building caddisfly taxa may influence fluvial geomorphology by transporting sediment ( Figure 2B). First, caddisfly may perturb sediment during locomotion (i.e. similar to the effect reported for stonefly larvae; Statzner et al., 1996), or during case construction when they sort through and discard many particles (Hansell, 1968a). Second, many caddisfly species crawl large distances across the river bed (~1-4 m day À1 ; Erman, 1986;Jackson et al., 1999;Lancaster et al., 2006), transporting their case and constituent sediment with them. In this manner, caddisfly transport sediment both ii iii iv v i~ 5 mm Figure 1. Diversity of caddisfly larval and pupal architecture (after Wiggins, 1996). (i) Nets filter food from flowing water, often accompanied by a static retreat in which the larvae reside (e.g. Hydropsychidae). (ii) Tubular cases which are usually mobile but may be fixed to stable particles for pupation (e.g. Limnephilidae). (iii) Domed/saddle cases are also usually mobile but fixed for pupation (e.g. Glossosomatidae). (iv) Purse cases (e.g. Hydroptilidae). (v) Pupal cases of free-living or retreat-building taxa (e.g. Rhyacophilidae or Hydropsychidae). The scale bar indicates approximately 5 mm.
R. J. MASON ET AL.

Context of this study
The design of caddisfly structures has fascinated scientists and incited a large number of detailed studies (e.g. Hansell, 1968a;Tolkamp, 1980;Becker, 2001;Wiggins, 2004). However, research has focused largely on individual species and the functional importance of structures to the larvae (Otto and Johansson, 1995;Limm and Power, 2011;De Gispert et al., 2018) rather than the effects of caddisfly case building on sediment stability and transport. The sedimentary implications are worthy of investigation because fine sediment availability is an important determinant of suspended and bed material transport. Sand fractions constitute the majority of the sediment flux in many rivers, and the sand content of gravel beds is a control on the movement of gravels as bedload (Jackson and Beschta, 1984;Ikeda and Iseya, 1988;Wilcock et al., 2001;Wilcock and Crowe, 2003). Furthermore, while the geomorphic impact of individual caddisfly may be inconsequential, they are typically abundant (De Moor and Ivanov, 2008) and dominate invertebrate biomass in some rivers. Consequently, caddisfly communities can use substantial quantities of sand-sized sediment. For example, preliminary studies indicate that the saddle-cased species Agapetus fuscipes (Glossosomatidae, Figure 1iii) used approximately 1.4 tonnes km À1 of sand to build their cases on the River Dove, UK ( Figure 3). Statzner et al. (2005) estimated that the pupal cases of Rhyacophilidae and Hydropsychidae (groups v and i, respectively in Figure 1) use between 1 and 3 kg m À2 (Statzner et al., 2005;Statzner, 2011). However, the quantity and size distribution of sediment used by the case-building caddisfly community has never been measured. Sediment use by caddisfly larvae is likely to vary with both the composition and abundance of the caddisfly community and within-species variability in case design. The distribution of caddisfly is a result of environmental controls (including flow velocity, oxygen availability and bed sediment characteristics), biological interactions (e.g. competition and predation) and historical factors (Southwood, 1977;Murphy and Davy-Bowker, 2005). The abundance and case designs of caddisfly larvae vary seasonally (e.g. some species only use mineral sediment grains when building pupal cases, Figure 1v; Wiggins, 2004). Furthermore, any impacts of case-building caddisfly on sediment stability will also depend on the geomorphic and hydrological context of the river environment. Consequently, In this study, we quantified mineral sediment use by casebuilding caddisfly in a UK stream. Sampling was restricted to riffle habitats because these typically support a greater abundance and diversity of caddisfly taxa (Brown and Brussock, 1991) and the lowest quantity of fine bed sediments, due to bed armouring (Statzner, 2011). Specifically, we measured the grain size distribution and quantity of sediment used by each of the casebuilding taxa present and compared sediment use between taxa. We also investigated how sediment use varies in space, in relation to the abundance of caddisfly taxa and the availability of sediment in the bed; both likely to be important controls on case design and sediment use. These results provide important information necessary to design ex-situ experiments and field studies to investigate the potential zoogeomorphic role of case-building caddisfly ( Figure 2). Furthermore, knowledge of the controls on sediment use by caddisfly larvae will begin to address the potential extent and importance of case-building caddisfly for fluvial geomorphology. For example, if caddisfly adapt their case design to local conditions, their presence would be expected to be less constrained by the availability of specific grain size fractions and consequently they may have a more widespread zoogeomorphic impact. This paper addresses three specific questions: 1. What is the mass and grain size of sediment used in cases by different caddisfly taxa and by the whole caddisfly community? 2. How does the mass and grain size of sediment used by the caddisfly community and individual taxa vary spatially? 3. To what extent do variations in the mass and grain size of sediment used in cases reflect differences in taxon abundance and sediment availability?

Field sampling
Caddisfly and bed sediment samples were collected from riffle habitats in Wood Brook, Leicestershire, UK ( Figure 4). Wood Brook is a first to second-order gravel-bed stream with an average width of 1.5 m and depth less than 0.1 m in the riffles studied. Wood Brook rises at an elevation of 200 m above sea level and flows for approximately 10 km to its confluence with the River Soar. The geology is largely mudstones and boulder clay underlain by Precambrian volcanic and intrusive igneous rocks (Greenwood et al., 2001). The median discharge of Wood Brook for the five years prior to sampling was 0.03 m s À1 . This was calculated from 15-minute resolution stage data and a rating curve. The slope of Wood Brook over the study area was 0.02 and the mean D 50 of surface sediment was 38 mm.
Caddisfly and sediment samples were collected from three sites along Wood Brook (referred to as upper, mid and lower; Figure 4) in order to capture variability in bed sediment availability and caddisfly communities, associated with longitudinal variations in grain size, channel morphology and local habitat characteristics. Nested within each site, three riffles were selected (each >6 m in length, Figure 4) and three longitudinally distributed samples collected from each riffle, making a total of 27 units (i.e. 3 samples × 3 riffles × 3 sites). Caddisfly communities and case characteristics are known to vary seasonally, therefore all samples were collected during a one-month period between 28 March and 26 April 2017, recognized as an optimum time for instream macroinvertebrate diversity in the UK and therefore likely to be the peak time for sediment use within caddisfly cases.
Caddisfly larvae samples were collected using a Surber sampler (Surber, 1937) with a basal area of 0.1 m 2 (330 × 310 mm frame with a 1 mm diameter mesh net), providing a fully quantitative measure of macroinvertebrate density (Everall et al., 2017). The surface sediment layer was agitated by hand so that macroinvertebrates were carried into the net by the current and larger particles were individually inspected for larvae and cases. A similar Surber sampler frame, with a 0.1 mm net to retain fine sediments, was used to collect a benthic sediment scoop from a location immediately adjacent to the caddisfly sample ( Figure 4C). The bed sediment sample was used to estimate the sediment available to caddisfly for case building, although this will vary depending on the mobility of the individual species (Statzner, 2011). Both caddisfly larvae and sediment were sampled to a depth of 0.05 m (approximately 1.4 × bed D 50 ).

Laboratory analysis
Caddisfly samples were preserved in 4% formaldehyde solution, washed through a 0.25 mm mesh sieve and manually processed for caddisfly larvae, pupae and cases, which were stored in 70% industrial methylated spirit. Most larvae were identified to species level (Edington and Hildrew, 1995;Wallace et al., 1990), although early instar taxa (e.g. Athripsodes sp.) were not identified beyond genus. Larvae and case length were measured to the nearest millimetre (Tolkamp, 1980). When more than 10 individuals of a species were present, length was estimated for a random subsample (average n = 12). Cases in each sample were combined within each taxon for grain size analysis (regardless of size or instar). For most taxa, only cases that were complete and retained their original length and width were analysed, although Rhyacophilidae and Hydropsychidae pupae were included even if incomplete due to low total abundance. To break down the silk structure and disaggregate the mineral particles, caddisfly cases were immersed in 30% hydrogen peroxide (H 2 O 2 ) at 80°C for at least 5 h and stirred periodically. For cases with high organic content, some larger organic particles were removed manually. H 2 O 2 was removed by washing the samples with deionized water through a 38 mm diameter, 0.063 mm aperture sieve and then dried overnight at 80°C. The disaggregated sediment was then mechanically sieved through 5.6-0.063 mm (38 mm diameter) sieves at half-phi intervals and weighed. Individual fractions were weighed on a four decimal place scale. To reduce sieving error, case samples with mass less than 0.01 g, or samples where the mass lost during the sieving process exceeded 5%, were removed from the analysis (9 and 1 removed, respectively, leaving 151 samples). Bed sediment samples were dried at 105°C for 12 h and then sieved into half-phi intervals down to 0.063 mm using a mechanical sieve shaker.

Sediment use by individual caddisfly taxa
The grain size distribution of an average individual case of each taxon in each sample, was calculated by dividing the mass of each grain size fraction by the number of cases in that sieved sample. Percentiles were determined for each of these grain size distributions. The mass of an average individual case was estimated as the sum of all size fractions divided by the number of cases sieved. To compare the mass and size distribution of sediment used in individual cases of different taxa at the river scale (question 1), the grain size distributions, percentiles and mass were averaged across all samples in which that taxon used sediment. The mean and standard deviation of larvae length, case length, case mass, D 50 , D 5 and D 95 are reported for each taxon, where the bar indicates the mean of all values across the river. D 5 and D 95 are used to give a measure of the range of grain sizes used by each taxon, excluding the most extreme sizes.

Sediment use by the caddisfly community
To calculate the total mass of sediment used by each taxon in each Surber sample, the sediment use by an individual of that taxon was multiplied by the number of cases recorded in the Surber sample (note: not all cases in a sample were subject to sieve analysis). The fragility of Glossosomatidae cases meant that the number of larvae recorded was typically greater than the number of surviving cases. Therefore, the total sediment used by this taxon was determined by multiplying the average sediment use per case by the number of larvae present. The total sediment mass used by the case-building caddisfly community in each Surber sample was then calculated as the sum of the totals for each taxon. The grain size distribution was similarly calculated, multiplying the mass of each grain size fraction in an individual case by the number of cases (or larvae for Glossosomatidae) present. The community grain size distribution was calculated as the sum of each grain size fraction used by each taxon. The mean and standard deviation, percentiles, skew and kurtosis were calculated for community sediment use. The mass of sediment used by caddisfly larvae in each sample was scaled from 0.1 m 2 Surber samples to 1 m 2 for ease of interpretation.

Spatial variability in sediment use by the caddisfly community
In order to address question 2, the mean mass and grain size distribution of sediment used by the caddisfly community in each sample was combined by site (upper, mid and lower), THE ZOOGEOMORPHOLOGY OF CASE-BUILDING CADDISFLY with nine samples in each (i.e. three replications in three riffles within each site). A Shapiro-Wilk test revealed that these predominantly displayed non-normal distributions and therefore a non-parametric Kruskal-Wallis test was used to compare the variability in the mass and grain size percentiles (D 50 , D 16 and D 84 ) of the community-level data between sites. Subsequently, Dunn's tests were used to determine which sites were significantly different. Shapiro-Wilk, Kruskal-Wallis and Dunn's tests were conducted in R studio (version 3.3.3, R Core Team, 2017).

Relation between abundance and sediment use
To address this question, we assumed that differences in the total sediment mass used in different locations (i.e. variability between samples) reflect: (1) differences in the abundance and caddisfly taxa present and (2) differences in the mass of cases built by individuals of the same species (potentially reflecting unknown biotic and abiotic drivers, including sediment availability). To explore the relative importance of these factors, we conducted linear regression between abundance (independent variable) and the total mass of sediment used (dependent variable) for each taxon and the whole community. A high value (R 2 ) indicates that spatial variability in sediment use is associated with the abundance of each taxon. A low value means that within-species variability explains differences in sediment use between samples. Samples without sediment use by a particular taxa were excluded from the regression analysis while samples with sediment use (cases) but no larvae (larvae may have been lost in sampling or emerged as adults) were included. To further investigate within-species variability in case building we concentrated on A. fuscipes (Glossosomatidae) because they were the most abundant taxon and accounted for most sediment use at the river scale. To determine if variability in A. fuscipes case mass corresponded to variability in the size of sediment used, we plotted the mean and standard deviation of grain size distributions for each site. Furthermore, to investigate potential explanations of variability in case mass we examined the correlations between mean case mass, mean case D 50 and mean larvae length across all sites for this species.

Relation between sediment availability and sediment use
To investigate how sediment use varied with sediment availability, the grain size distributions of mean sediment use by the caddisfly community were compared to the bed sediment availability, both at the river scale and for each site. The mass of sediment used by the caddisfly community was divided by the mass of available sediment and multiplied by 100 to give the percentage of sediment used. This was done for each grain size, for each sample, and averaged for each site and the whole river.

Sediment use by individual caddisfly taxa
There was considerable variability in the mass and size distribution of cases of individual taxa (Table I, Figure 5). The tubular cases of Leptoceridae and Lepidostomatidae (Figure 1ii) used fine sand (D 50 = 0.17 mm) and the mean mass for both taxa was only 0.001 g, less than 1% of the mean Limnephilidae case mass (0.146 g), which used coarser sediment (D 50 = 1.12 mm). Pupal cases of Rhyacophilidae (free living) and Hydropsychidae (net and retreat builder) were substantially heavier (mean mass = 0.83 g) and coarser (D 50 = 4 mm) than all other taxa ( Figure 5). Goeridae cases are unusual because they secure coarser particles to the case edges. This resulted in a bimodal sediment distribution ( Figure 5). The size distribution of Glossosomatidae cases (saddle cases, D 50 = 0.94 mm) was similar to Limnephilidae (tube cases), but the mean mass of each individual case was less (0.02 g, Table I, Figure 5). Therefore, case-building caddisfly exhibited considerable variability in larvae and case size, particle size distribution and case mass, not just between structure groups (Figure 1) but also between families and species (Table I, Figure 5). Sediment use by the whole caddisfly community A diverse caddisfly community was recorded with 24 taxa from 10 families (Table II) and a mean taxonomic richness of 7.4 taxa per sample (minimum 3, maximum 12). Mean abundance was 2250 larvae m À2 , but this varied from 180 to 7460 larvae m À2 . Caddisfly using mineral sediments for case building accounted for approximately 50% of the taxa present (Table II), but 94% of individual caddisfly larvae recorded. All taxa combined, the average total mineral mass used by case-building caddisfly was 37.57 g m À2 , but this varied across the 27 samples, ranging from 3.74 to 138.83 g m À2 ( Figure 6A). Due to the differences in case grain size distributions between taxa, the size range of mineral sediment used by the community was broad, from 0.063 to 11 mm ( Figure 5). Therefore, caddisfly utilize the entire sand size range as well as fine gravel particles for case construction (0.063 mm < sand < 2 mm gravel; Wentworth, 1922). On average, sediment use was 84% sand and 16% fine gravel. The average particle size distribution of sediment used ( Figure 6B) was bimodal with a primary mode at 1-1.4 mm and another mode at 4-5.6 mm. D 50 was 1.06 mm and the distribution is fairly symmetrical (skew Sk: À0.008) and leptokurtic (kurtosis K: 1.238). The total mass of sediment used by case-building caddisfly (all samples combined) was dominated by Glossosomatidae (64%, Figure 6B), followed by pupae of Rhyacophilidae and Hydropsychidae (15%) and Limnephilidae (11%), with other taxa accounting for less than 5%.

Spatial variability in community sediment use
Kruskal-Wallis tests confirmed that there were no significant differences in the total mass of sediment used between the three sites (Table III, Figure 7A, mean at the upper site = 29.59, mid = 38.02, lower = 34.11 g m À2 ). However, the size of sediment used was significantly finer at the mid site than at the upper and lower sites for D 50 , D 84 and D 16 (Table III, Figure 7B, mean at the upper site = 1.61, mid = 0.79, lower = 1.11 mm). The upper and lower sites were not significantly different for any grain size parameter (Table III). The mass of sediment used by each taxon varied between sites (Figures 7C-E).
At the upper site, Glossosomatidae and Limnephilidae accounted for a similar proportion of the 0.5-4 mm size fraction while at the mid and lower sites, sediment use was dominated by Glossosomatidae. Both the upper and lower sites have community grain size distributions skewed towards coarser particles due to (1) pupal cases of Rhyacophilidae and Hydropsychidae which dominated sediment use at larger grain sizes (2.8-11 mm), and (2) differences in the size distribution of sediment used by Glossosomatidae.

Relation between abundance and sediment use
The abundance of the caddisfly community (all taxa) was a poor predictor of the mass of sediment used, explaining only 17% of variation (Table IV), meaning that differences in case mass between individuals are important at a community level. When considering each taxon separately, for some, abundance displayed a strong association with sediment mass used (Leptoceridae, R 2 = 0.79 and Goeridae, R 2 = 0.75), but for others the association was weak (Glossosomatidae, R 2 = 0.34). When grouped by site, the association between abundance and sediment use was stronger for some taxa, including Glossosomatidae, which displayed a strong linear association between sediment mass and abundance at the upper and lower sites but not at the mid site (Table IV, Figure 8A). Within each of the three sites, Glossosomatidae larvae built cases of similar mass and consequently Glossosomatidae abundance had a strong association with total Glossosomatidae sediment use.

THE ZOOGEOMORPHOLOGY OF CASE-BUILDING CADDISFLY
However, at the river scale there was considerable inter-site variability in case mass.
In addition to having lower mass, Glossosomatidae cases at the mid site contained finer sediment particles ( Figure 8B). Correlations between mean case mass and larvae abundance across all samples revealed that there was a negative association (R = À0.55, Figure 8C). However, in samples with low abundance, case mass varied substantially, while at high abundance, mass was always lower. A very strong correlation exists between mean case mass and D 50 ( Figure 8D) and between case mass and mean larvae length ( Figure 8E). Overall, the abundance of most taxa, and particularly Glossosomatidae, is a poor predictor of sediment mass used by this taxa due to within-taxa variations in the mass of cases built and the grain sizes used.

Relation between sediment availability and sediment use
The mean D 50 of bed material samples was lowest at the mid site (32 mm) and similar between the upper and lower sites (38 and 40 mm, respectively). Considering only the size range Table II. Caddisfly taxa found in this study. Taxa are sorted according to their behaviour (after Wiggins, 1996) and use in this study: (1) taxa that used mineral sediments in their cases which were included in sediment analysis and (2)  Cases for which a larva was never found and therefore could not be identified were classified as unknown. Unknown cases were all similar in architecture and are believed to be discarded pupal cases of Leptoceridae species. b Pupal cases of Hydropsychidae and Rhyacophilidae were identified only to family level but are likely to be the same species found elsewhere. used by case-building caddisfly in this study, the average mass of bed sediment availability increased with grain size ( Figure 9A). Some samples displayed a bimodal distribution with a second peak in the 0.25-0.5 mm size range, particularly at the mid site. Sediment use was also skewed towards finer sediment at the mid site ( Figure 9B). Sediment availability for all size fractions greatly exceeded the mass of sediment used by caddisfly larvae, resulting in a low percentage used overall ( Figure 9C). As an average across all samples, the percentage of sediment use peaked at the 1-1.4 mm size range with a mean of 2.99% (matching the mode of sediment use, Figure 9 B). In this size fraction, the percentage used ranged considerably from 0.25% up to 24.86% of available sediment. Therefore, sediment of all grain sizes was abundant and available  to taxa at all sites. Despite this, the use of sediment by both the case-building caddisfly community and Glossosomatidae was skewed towards finer sediment at the mid site where this sediment was more abundant.

Sediment use by individual taxa and the case-building caddisfly community
Caddisfly are a widespread, abundant and diverse group of invertebrates. Caddisfly larvae build complex structures from fine sediment and silk (Wiggins, 2004). Whilst case building by individual caddisfly taxa has received attention from an ecological perspective, this is the first study to focus on the granulometry of cases from a geomorphic perspective and to measure sediment use (mass and grain sizes) by a range of taxa and for entire communities. The caddisfly community of Wood Brook was diverse, both in terms of species richness and the design of silk structures, with all five categories recorded (Table II, Figure 1). About half of taxa were case-building caddisfly utilizing mineral sediments, but due to their high abundance they accounted for over 90% of individuals. This suggests that if the mechanisms identified in Figure 2 are significant, case-building caddisfly could be important zoogeomorphic engineers in this stream. The considerable variability in the size of sediment used by different taxa in this study ( Figure 5) probably reflects differences in the function of the larval cases (Otto and Svensson, 1980;Wiggins, 2004;Boyero et al., 2006). For example, the fine grains used in Sericostomatidae cases (D 50 = 0.27, Table I) result in smooth, tapered cases, which may facilitate burrowing. Most taxa in this study used predominantly mineral sediments (Table II), which require a much greater investment of energy than organic cases (Otto and Svensson, 1980). Consequently, using mineral sediment must be advantageous to larvae in gravel-bed rivers, perhaps providing ballast, protection from crushing by moving sediment and reduced risk of predation (Webster and Webster, 1943;Statzner and Holm, 1989;Otto and Johansson, 1995). The results of this study demonstrate that both the size and magnitude of mineral sediment used are dependent on the caddisfly taxa present, and consequently any geomorphic effect is likely to vary between taxa. As a result of the diversity in grain size used by different taxa, the caddisfly community used sediment from 0.063 to 11 mm in diameter ( Figure 6B). Consequently, caddisfly have the potential to directly affect the distribution and mobility of a broad sediment size range. However, sediment use was predominantly coarse sand and fine gravel ( Figure 6B), which are important size fractions geomorphologically, as they are often the most mobile bed material size fractions (Jackson and Beschta, 1984;Ashworth and Ferguson, 1989;Kuhnle, 1993). (independent) and the mass of sediment used (dependent) for each taxon across all samples (whole river) and for each site (upper, mid and lower) independently. The regression was also conducted for the whole caddisfly community All Upper Mid Lower Glossosomatidae 0.34(25)** 0.71(7)* 0.15(9) 0.99(9)*** Goeridae 0.75(16)*** 0.80(6)* 0.75(9)** (1) Sericostomatidae 0.38(24)** 0.05(6) 0.78(9)** 0.25(9) Lepidostomatidae 0.52(6) (0) (0) 0.52(6) Leptoceridae 0.79(13)*** (2) 0.84(9)*** (2) Limnephilidae 0.55(18)*** 0.25(9) 0.93(8)*** (1) Community 0.17(27)* 0.09(9) 0.14(9) 0.97(9)*** The number of samples included in each regression (n) is shown in brackets (regression not conducted when n < 5). Significance is indicated by * 0.05 > p > 0.01; ** 0.01 > p > 0.001; *** p < 0.001. Sediment use by key taxa and potential zoogeomorphological importance Despite the abundance and diversity of case-building caddisfly larvae in Wood Brook, several taxa are of particular interest. Glossosomatidae accounted for over 60% of sediment use and were by far the most abundant taxa (up to 6710 larvae m À2 ). Glossosomatidae cases are built from sediment particles accessible to them on the bed surface (Marchant, 1988;Becker, 2001;Statzner, 2011;Hansell, 1968b) and they may therefore have a disproportionate effect on sand availability at the bed surface. In particular, Glossosomatidae usually reside on the exposed upper surfaces of larger gravel particles where the flow of water facilitates respiration (Figure 3; Nijboer, 2004;Becker, 2005;Morris and Hondzo, 2013;Morris et al., 2015). Glossosomatidae therefore redistribute surface particles vertically, potentially increasing the exposure of constituent sand grains by transporting them from interstitial spaces and sheltered patches onto the upper surfaces of larger cobbles. This study showed that Glossosomatidae used on average 25.95 but up to 88.92 g m À2 of sand. On average, this represents a vertical redistribution of 1 kg of sand for every 40 m 2 of occupied river bed. Despite their low abundance in this study, Rhyacophilidae and Hydropsychidae pupal cases were constructed from large particles and, by mass, therefore accounted for 15% of the overall sediment used by case-building caddisfly ( Figure 6B) but nearly 50% at the upper site ( Figure 7C). Statzner et al. (2005) and Statzner (2011) measured sediment use by these taxa in the River Furan, France and found them to use substantial quantities of sediment (up to~3 kg m À2 ). They found considerably higher abundances of pupal cases (250 m À2 for Hydropsyche siltalai; Statzner, 2011) compared to the average of 3 m À2 in this study. This is probably due to both higher abundances of larvae in the River Furan than Wood Brook, and the time of year sampled. Statzner et al. (2005) sampled in June and August, while our study was conducted in April, prior to pupation for most species (Wallace et al., 1990). In situations where pupal cases of free-living and net-building taxa are abundant (>1000 larvae m À2 ; Cardinale et al., 2004), sediment use is expected to be skewed towards coarser particles. During pupation, these taxa may have geomorphic effects by modifying the hydraulic properties of coarse sand and gravel (Figure 2A). Pupal cases are usually attached to more stable particles (unlike mobile cases) and consequently the mobility of constituent sediment is likely to be reduced. Cases may also reduce the mobility of the larger particles they are attached to (Figure 1i; Nunokawa et al., 2008).
Spatial variability in community sediment use and the distribution of case-building caddisfly taxa Both the size and mass of sediment used were spatially variable at the river scale (Figure 7), even within this relatively narrow range of environmental conditions (riffle habitats in~3.6 km of stream; Figure 1). There was considerable variability in the mass of sediment used (4-139 g m À2 ) but this was not structured by site ( Figure 7A). This is potentially related to the patchy distribution of benthic invertebrates in rivers (Pringle et al., 1988); a result of habitat variability and biological controls (Cummins and Lauff, 1969;Southwood, 1977). The size of sediment used by the caddisfly community was statistically lower at the mid site than at the upper and lower sites (Table III, Figure 7B). This is interesting because the greatest environmental difference would be expected between the upper and lower  (Vannote et al., 1980;Rice and Church, 1998). Nevertheless, the mid site had the lowest slope, bed sediment D 50 and water depth, suggesting that it was distinct from the other sites in terms of habitat characteristics.
The greater use of fine sediments by the caddisfly community at the mid site is the result of both the absence of Rhyacophilidae and Hydropsychidae pupal cases and the use of finer sediments by Glossosomatidae larvae ( Figure 7D). The upper site had the coarsest D 50 where Rhyacophilidae and Hydropsychidae pupae used a greater proportion of coarsesediment grains, resulting in a bimodal sediment distribution with a primary peak in the coarse sand range and a second peak of fine gravel for the site (4-5.6 mm, Figure 7C). The upper site had, on average, the lowest abundance of Glossosomatidae and the mass of each case was very variable ( Figure 8C). This could be due to shading by riparian vegetation reducing the availability of their algae food source (Mcneely and Power, 2007) and increasing inputs of allochthonous organic matter, favouring shredders (Vannote et al., 1980), such as the Limnephilidae larvae present.

Sediment use in relation to the abundance of taxa and availability of sediment
The weak association between the abundance of the whole case-building caddisfly community and the total mass of sediment used (Table IV) is unsurprising because the mass of cases varied by several orders of magnitude between taxa (Table I). Therefore, to achieve a 0.85 g increase in sediment use would require just one Rhyacophilidae or Hydropsychidae pupal case but 850 Leptoceridae cases (0.001 g each, Table I). Interestingly, the abundance of many individual taxa was also only weakly related to sediment mass used at the river scale (Table IV). This indicates that within-taxa differences in case mass between samples explained more variability in total sediment use than the abundance of that taxa. In this study, differences in case mass between individuals of the same taxa may be explained to some extent by the presence of multiple instars of larvae, which were combined for grain size analysis. Furthermore, within the Limnephilidae family, the presence of multiple species (with different case designs) and differing use of organic material (Table I) also explains this weak association.
A. fuscipes (Glossosomatidae) cases had lower mass and used finer sediment at the mid site than the upper and lower sites ( Figures 8B and C). This difference was associated with smaller A. fuscipes larvae ( Figure 8E) and less developed (early instar) larvae are likely to build smaller cases with lower mass. The greater abundance of A. fuscipes at this site may increase competition for resources and limit larvae development. Alternatively, the greater availability of fine sediment at the mid site ( Figure 9A) may cause larvae to build cases which contain more fine sediment and consequently weigh less ( Figure 8D). At the river scale only a small percentage of available sediment was used (greatest at the 1-1.4 mm fraction where mean percentage used = 2.99%; Figure 9C) and therefore, it is unlikely that taxa are limited in their case design by access to their preferred grain sizes. Glossosomatidae larvae, however, typically use easily accessible sediment (Becker, 2001) and therefore, while specific size fractions are not limiting at any site, the greater availability of finer fractions at the mid site may explain the lower D 50 of A. fuscipes cases here.
Overall, the percentage of available sediment used in this study is considerably lower than that of Statzner (2011); 2.99% compared to 15-25% for the dominant sediment size. This is likely to be at least partly due to a difference in sampling technique. Statzner (2011) sampled surface sediments and the first subsurface layer and, whilst they do not specify a precise depth, this is unlikely to be as deep as the 5 cm sampled in this study. In addition, Wood Brook drains agricultural lands and consequently contains a high quantity of fine sediment. There is, however, considerable variability in the proportion of sediment used, up to 25% of the 1-1.4 mm size fraction was used at one lower sample. This is explained by a low quantity of fine sediment in this sample, rather than high caddisfly use of sediment. Therefore, in rivers with lower quantities of fine sediment, such as mountain streams (Church, 2010) or downstream of reservoirs (Brandt, 2000), case-building caddisfly may use substantial proportions of the available fine sediment.

Limitations
Quantifying sediment use by a caddisfly community presented multiple methodological challenges. Determining the size distribution of very small masses of sediment (as low as 0.01 g) was difficult, but the use of small-diameter sieves (38 mm) reduced sediment loss and allowed comparison with sieved bed sediment samples (unlike photographic methods; Statzner et al., 2009). Quantifying sediment use for enough individuals and sites necessitated combining taxa of different instars within each sample. However, sediment distributions are known to change with larvae development, which means that our grain size estimates are broader than many in the literature (e.g. Tolkamp, 1980). Last, the size distribution of Rhyacophilidae and Hydropsychidae pupal cases is likely to be less accurate than those for the other taxa, due to the low number analysed and the combination of two families, known to differ in pupal case design (Statzner et al., 2005). Nevertheless, within these constraints, the results present an important step towards understanding the magnitude of sediment use by case-building caddisfly and the effect of cases on fine-sediment sorting and transport in rivers.

Conclusions
This study quantifies sediment use by the case-building caddisfly community from riffle habitats in a small stream. In doing so it extends research on both the ecology of case building and the zoogeomorphic role of caddisfly in rivers (Cardinale et al., 2004;Johnson et al., 2009;Albertson et al., 2014b). Considerable variability was found in the mass and size distribution of cases between caddisfly taxa at every taxonomic level, between (1) silk structure design groups (Figure 1), (2) families ( Figure 5), (3) species (Table I) and (4) individuals of the same species (Table IV, Figure 8). Case mass ranged from 0.001 to 0.85 g and D 50 from 0.17 to 4 mm. Therefore, the mechanisms and extent of caddisfly zoogeomorphic effects are likely to vary between taxa. Community sediment use was on average 37.57 g m À2 , and was 84% sand and 16% fine gravel. A. fuscipes (Glossosomatidae) accounted for over 60% of sediment use, Rhyacophilidae and Hydropsychidae (15%) and Limnephilidae (11%). These taxa should therefore be the focus of future research concerning the mechanisms of zoogeomorphic impact outlined in Figure 2.
Due to considerable variability in case mass between sites, abundance was weakly related to sediment use for the caddisfly community and for most taxa at the river scale. In this stream, caddisfly used a small percentage of the total sediment available to them for all size fractions (on average <3%), suggesting that access to preferred sediment was not a limiting R. J. MASON ET AL. factor in case design. Nevertheless, A. fuscipes cases varied in case design between sites, with a lower D 50 at the mid site where finer sediment was more readily available. Intraspecific variability is rarely considered in zoogeomorphology (Albertson and Allen, 2015). This raises important questions about the replicability of localized biogeomorphic studies which consider only a few individuals, or individuals from a singular site.
Despite variability in community composition, abundance and case architecture, the caddisfly community used sediment in all 27 samples. Consequently, within the riffle habitats studied, sediment use by the caddisfly community is not constrained in space by the presence of a particular species, abundance or sediment availability. It follows that if cases do have geomorphic effects (mechanisms in Figure 2), then these are likely to be widespread across a range of habitats in which case-building taxa are known to be common. Furthermore, individual taxa appear to be adaptable in their use of sediment. The distribution of A. fuscipes, in particular, is not limited by the availability of a specific range of grain sizes and consequently may have widespread zoogeomorphic effects.
Small streams, such as Wood Brook, make up 70-80% of the total channel length of river networks and are an important but often neglected waterbody in catchment geomorphology research (Downing, 2012;Wohl, 2017). Caddisfly are abundant in most small streams and rivers and consequently, their zoogeomorphic role could be widespread. Future work is required to characterize sediment use by caddisfly larvae at larger spatial and temporal scales, as well as to quantify the effects of case building on the mobility of sediment particles ( Figure 2). This study provides insights into the size and quantity of sediment use by case-building caddisfly, essential to guide further research on caddisfly zoogeomorphology.