UBC Faculty Research and Publications

Predation threat alters community structure and ecosystem function Hammill, Edd; Atwood, Trisha B.; Srivastava, Diane S. Aug 31, 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
52383-Hammill_E _et_al_Predation_threat.pdf [ 860.66kB ]
Metadata
JSON: 52383-1.0132600.json
JSON-LD: 52383-1.0132600-ld.json
RDF/XML (Pretty): 52383-1.0132600-rdf.xml
RDF/JSON: 52383-1.0132600-rdf.json
Turtle: 52383-1.0132600-turtle.txt
N-Triples: 52383-1.0132600-rdf-ntriples.txt
Original Record: 52383-1.0132600-source.json
Full Text
52383-1.0132600-fulltext.txt
Citation
52383-1.0132600.ris

Full Text

   1 Predation threat alters community structure and ecosystem function 1 Edd Hammill1,2, Trisha B. Atwood3,4 and Diane S. Srivastava1 2 1Department of Zoology & Biodiversity Research Centre, University of British Columbia, 3 6270 University Blvd., Vancouver, BC, Canada, V6T 1Z4,  4 2School of the Environment, University of Technology, Sydney, Ultimo, NSW, Australia, 5 2007 6 3Department of Forest and Conservation Sciences, University of British Columbia, 7 Vancouver, BC, V6T 1Z4, Canada 8 4Global Change Institute Building (#20), Staff House Rd, University of Queensland 4072  9  10 Running title: Risk of predation modifies ecosystems 11  12 Correspondence e-mail to Edd Hammill -  edd_hammill@hotmail.com 13 Abstract - 250 14 Main body text – 4700 15 Tables - 1 16 Figures – 4 17 References – 52 18 Appendices – 2, 678 words total 19 Keywords (alphabetical order) 20 bromeliad communities, CO2 dynamics, community ecology, ecosystem function, Odonata, 21 predator-prey, mosquito, non-consumptive effects, trophic cascade 22  23 Author contributions. E.H., T.B.A and D.S.S. designed the study. The experiment was 24 conducted in the field by E.H. and T.B.A. All authors contributed to the writing of the MS.  25 26    2 Abstract 1 Predators can have dramatic effects on food web structure and ecosystem processes. 2 However, the total effect of predators will be a combination of prey removal due to 3 consumption and non-consumptive effects (NCEs) mediated through changes to prey 4 behavioural, morphological, or life history traits induced to reduce predation risk. In this 5 study, we examined how consumptive and non-consumptive effects alter community 6 structure and an ecosystem function using the aquatic ecosystem housed within tropical 7 bromeliads. We exposed emptied bromeliads to no predators, caged predators (NCEs only), 8 or uncaged predators (NCEs and consumptive effects) and recorded densities of all macro-9 invertebrates, microbial densities and in situ CO2 concentrations after 40 days. We found that 10 predators altered community structure and CO2 concentrations largely through NCEs. The 11 magnitude of the effects of NCEs was substantial, contributing > 50% of the total effects of 12 predators on macro-invertebrate communities. The non-consumptive effects of predators 13 were also strong enough to generate a trophic cascade, which significantly increased micro-14 organisms and ecosystem respiration, leading to increased in situ CO2 concentrations. The 15 most likely mechanism behind the NCEs on macro-invertebrate density is detection of 16 predator cues by ovipositing adult females, who actively choose to avoid bromeliads 17 containing predators. Through this mechanism, predator NCEs modified community 18 colonization, the structure of food webs, populations of lower trophic levels, and an 19 ecosystem processes performed by the community. We therefore propose that quantification 20 of the relative strength of predator NCEs in natural ecosystems is critical for predicting the 21 consequences of predator loss from the world’s ecosystems.  22  23 24    3 Introduction 1 The total effect of predators on prey is a combination of direct consumption, and predator 2 non-consumptive effects (NCEs). Predator NCEs may take several forms, such as changes in 3 prey behavioural, morphological or life history traits (Tollrian and Harvell, 1999; Peacor and 4 Werner, 2001). Predator NCEs are prevalent in a wide variety of ecosystems, but are 5 particularly common and strong in ecosystems with sit-and-wait or sit-and-pursue predators 6 (Preisser et al., 2007), as cues from sedentary predators provide a point source indicator of 7 the risk of attack. The magnitiude of the effect of predator NCEs can also be large, and has 8 been shown to be greater than the effects of direct consumption (Trussell et al., 2006a). 9 Previous studies have demonstrated that predators can generate trophic cascades 10 through NCEs alone (Schmitz et al., 1997; Forbes and Hammill, 2013), and affect ecosystem 11 processes (Strickland et al., 2013). However, this past research into predator NCEs has 12 tended to focus on simple, two or three trophic level food chains (Trussell et al., 2006b), 13 often with each trophic level represented by a single species (although see (Peacor et al., 14 2012), and with species densities being determined by the investigators. The effects of 15 predators may be enhanced by artificial manipulation of densities, and reduction of a 16 complex food web to a simple food chain as effects transfer fairly linearly, whereas the 17 reticulate nature of real food webs can dampen trophic cascades (Carpenter, 1996). As 18 predators are currently being lost from the world’s ecosystems at an unprecedented rate 19 (Estes et al., 2011), understanding the strength of predator NCEs is crucial to understanding 20 the consequences of their loss. 21 Predator NCEs may take many forms, and may affect community dynamics through a 22 variety of mechanisms. Prey species that cross ecosystem boundaries through ontogeny, such 23 as insects and amphibians that have terrestrial adult stages, but oviposit in aquatic 24 ecosystems, may avoid locations within a landscape that pose a predation risk (Berendonk 25    4 and Bonsall, 2002; Resetarits and Binckley, 2009; Vonesh and Blaustein, 2010). The decision 1 to avoid ovipositing in locations containing predators can alter colonisation rates, leading to 2 changes in the composition of communities (Kraus and Vonesh, 2010). In addition to altering 3 colonisation rates, the presence of aquatic predators can also increase larval development 4 rates, shortening the length of time individuals are exposed to predation as larvae (Hammill 5 and Beckerman, 2010). Additionally, the threat of predation can reduce foraging rates of 6 competitively dominant prey, influencing community dynamics (Werner and Anholt, 1996). 7 These pre- and post- colonisation processes act to determine the eventual community 8 structure (Vonesh et al., 2009).  9 The structure of ecological communities affects the functions the ecosystem performs 10 (Schulze and Mooney, 1994; Cadotte et al., 2011; Hooper et al., 2012). This relationship 11 between community structure and ecosystem functioning implies predator NCE-mediated 12 changes to communities may alter the functioning of ecosystems, including the production of 13 CO2 through ecosystem respiration. Previous studies have demonstrated that predators are 14 able to alter community respiration to such an extent they change from being sources to sinks 15 of CO2 (Schindler et al., 1997), or vice versa (Atwood et al., 2013). Freshwater ecosystems 16 globally emit similar levels of CO2 (up to 1.65 Pg C yr-1) as land use change (Cole et al., 17 2007).  This relatively high level of CO2 production, coupled with the bi-directional effects of 18 predators, means understanding the mechanisms by which predators alter community 19 respiration has relevance for ecological management actions undertaken to mitigate CO2 20 production (Schmitz et al., 2013). 21 We quantified predator NCEs on community structure and community respiration 22 using the natural ecosystems contained in the water filled leaf axils of Guzmania bromeliad 23 phytotelmata. Bromeliad leaves are arranged in a rosette structure, and within the wells 24 created by the leaf axils of tank bromeliads exists an aquatic, detrital-based ecosystem 25    5 (Figure 1a). Bromeliad communities experience periodic droughts that can lead to the loss of 1 many members of the community, and are therefore in a seasonal cycle of drought and re-2 colonisation, with community structure being related to colonisation rates (Srivastava et al., 3 2008). Following the onset of rain, bromeliad ecosystems are colonised by macro-4 invertebrate insect larvae, including predatory Odonata, mosquitoes (Culicidae - filter feeders 5 and browsers), as well as detritivorous Chironomidae, Tipulidae and Scirtidae. Throughout 6 the rainy season, adult insects are continuously adding to the macro-invertebrate community 7 though oviposition, and individuals are being lost as they emerge as adults or die. The 8 number of insect larvae present in a bromeliad at any one time is therefore governed by the 9 relative rates of oviposition vs emergence and death. Bromeliad ecosystems also contain a 10 broad microbial community, including ciliates (subphylum Ciliphora), flagellates (subphylum 11 Mastigophora), and rotifers. This microbial community consumes bacteria, fungi, and 12 detritus. The community within the bromeliad therefore contains at least two different 13 compartments, the microbial food web and a detritivore food web, with larvae of the 14 damselfly Mecistogaster modesta (Selys, 1860) acting as top predators in both (Figure 1b).  15 The aquatic ecosystems within bromeliads are net-heterotrophic, fuelled 16 predominantly by the mineralization of organic compounds from allochthonous detritus, 17 resulting in the release of CO2, methane (Martinson et al., 2010; Atwood et al., 2013) and 18 dissolved nutrients (Ngai and Srivastava, 2006). Theoretically, Odonate predators may either 19 increase or decrease community respiration, depending on which trophic pathways are most 20 influenced by predation risk (Figure 1b).  The effect of predators on CO2 concentrations 21 would depend on whether the dominant trophic pathway was even-numbered (detritus-22 microbes-mosquitoes-predator), in which case predators would increase CO2 concentrations, 23 or odd-numbered (detritus-macroinvertebrates-predator), where predators would decrease 24 CO2 concentrations. However, as both mosquitoes and protists exhibit omnivory within 25    6 species and variance in trophic levels between species (Figure 1b), it is difficult a priori to 1 establish the effective number of trophic levels in this pathway. 2  As M. modesta are generalists able to consume a wide variety of macro-invertebrates, 3 risk of consumption and a selection pressure to avoid bromeliads containing M. modesta is 4 likely to be felt by all species.  We therefore made three hypotheses: 1 - the presence of 5 predators would lead to lower densities of all macro-invertebrate species. 2 - The reduction in 6 macro-invertebrates would create a trophic cascade, releasing the microbial community from 7 the pressures of macro-invertebrate predation (especially from mosquitoes), increasing their 8 densities. 3 - Predator NCEs on the macro-invertebrate and microbial communities 9 significantly affect rates of community respiration [as approximated by CO2 concentrations 10 within the water column (Del Giorgio et al., 1999)]. 11  12 Materials and Methods 13 We collected 30 bromeliads from the genus Guzmania from a tropical mid-elevation 14 rainforest (~700 m above sea level) within 5 km of Estación Biológica Pitilla, Área de 15 Conservación Guanacaste, Costa Rica. Prior to being used in the experiment, all bromeliads 16 were thoroughly washed and immersed in water in an inverted position for 24 hrs, then left to 17 dry for 7 days. This method eliminated most if not all invertebrates and residual chemical 18 cues from predators prior to use. The preparation process did not lead to mortality in any of 19 the plants. Our study was conducted during the rainy season (October and November), a time 20 when oviposition by intermediate trophic levels (mosquitoes, Chironomidae, Tipulidae and 21 Scirtidae) is highest and M. modesta are in mid-instars (Srivastava, 2006). Additionally, M. 22 modesta rarely oviposit during the rainy season, which helped ensure that predator free 23 treatments remained free of M. modesta throughout the study (Srivastava, 2006). At the start 24 of the experiment, plants were moved to a 30 m x 30 m patch of secondary forest, and 25    7 suspended from trees that had a diameter at breast height greater than 10 cm. The wells of the 1 plants were filled using commercially available mineral water and leaf litter from Conostegia 2 xalapensis Bonpl. was distributed throughout the plant at a density of 200 mg (dry weight) 3 per 100 ml (total plant volume) to act as food for the community. The leaves of C. xalapensis 4 are highly abundant and easily recognisable, using a single species for leaf litter in all 5 treatments minimised differences in nutrient composition.  6 Prior to the start of the experiment, we recorded the maximum volume of water each 7 plant could hold, and randomly assigned them to three experimental treatments: no predators, 8 caged predators (predator NCEs only) and uncaged predators (NCE’s and consumptive 9 effects of predators, hereafter referred to as total predator effects). Each treatment was 10 represented by 10 replicates, and plants did not significantly differ in size between treatments 11 (volume range 500 – 1500 ml, mean 1013.5ml, F(1,28) = 2.17, p = 0.15). The strength of 12 predator NCEs can be related to predator biomass, with larger predators generating stronger 13 effects (Hill and Weissburg, 2013). To minimise the confounding effect of predator size, all 14 predators used in the experiment were mid-instar M. modesta larvae with body lengths,12 – 15 15 mm body. A single M. modesta larva was added inside the cage for caged (NCE only) 16 replicates, and one outside the cage for uncaged (total predator effect) replicates. Using a 17 single predator per plant mimics natural M. modesta densities for the size of bromeliads used 18 in our study (Srivastava et al., 2005). Predator cages consisted of 50 ml clear centrifuge 19 tubes. Two 15 mm diameter holes were drilled in the sides of the tubes and covered with 80 20 µm mesh. These mesh-covered holes allowed water inside and outside the cages to mix, 21 facilitating diffusion of predator chemical cues. We used 80 µm mesh as it allowed the 22 transfer of chemical cues without clogging with detritus, but did not allow the passage of 23 small macro-invertebrate prey into predator cages. We also added empty cages to no predator 24 and uncaged predator treatments to ensure differences between treatments were not due to the 25    8 presence/absence of cages. All cages were placed inside a well in the second row of leaves 1 from the centre of the plant. Placing the cage relatively close to the centre ensured that 2 chemical cues from a caged predator could diffuse down through the rest of the plant, as 3 water cascades down through the bromeliad during rain showers. Caged predators were fed a 4 single mosquito and chironomid larvae every other day, whereas uncaged predators 5 consumed insects within the bromeliads.  We mimicked the feeding procedure in the uncaged 6 predator and no predator treatments to ensure adequate control, in case adult mosquitoes were 7 attracted to the plant by our presence.  8 The experiment was run for 30 days, during which time the insect community 9 accumulated through natural colonisation. After the 30 day colonisation period, we compared 10 differences in macro-invertebrate community structure and density, microbial density, and in 11 situ CO2 concentration. A 30 day study period was used as it allowed for multiple 12 colonization events, while minimizing loss from emergence. Chironomidae, Tipulidae and 13 Scirtidae colonizing the bromeliads have larval stages that are typically greater than one 14 month in duration (Srivastava, 2006). Although 30 days may have exceeded the hydroperiod 15 of mosquito species in this study, oviposition by adult mosquitos is ongoing throughout the 16 rainy season.  17 At the end of the experiment, we randomly selected a well that had not contained the 18 cage, and collected a 1 ml water sample, which was preserved with Lugols media and later 19 used to calculate protist density.  In a different well, 6 ml of water was extracted using a 50 20 ml Pressure-Lok® syringe (VICI Precision Sampling Corp., Baton Rouge, LA), injected in a 21 gas tight vacutainer (Labco Limited High, Wycombe, UK.), chilled and transported within 72 22 h to the Department of Civil Engineering, Environmental Laboratory at the University of 23 British Columbia for analysis of dissolved CO2 gas concentrations. Two CO2 samples were 24 compromised during transit, and thus not included in the analysis. Collections and 25    9 calculations of CO2 concentrations from sample water followed procedures from Hope et al. 1 (Hope et al., 1995).  2 Following CO2 collections we removed all water, insects and detritus from the plant. 3 Insect larvae were sorted and preserved in ethanol within six hours. We identified mosquito 4 larvae to species, while other insects (largely Chironomidae, Tipulidae and Scirtidae) were 5 sorted to family level. For each plant, we calculated macro-invertebrate density (the total 6 number of individuals of each species divided by plant volume) in order to directly compare 7 organisms from plants of different sizes. Data are available on the Knowledge Network for 8 Biocomplexity (http://knb.ecoinformatics.org/knb/metacat/knb.302.1/knb).To estimate 9 densities of micro-organisms, all protists and rotifers were counted in a 50 µm sub-sample of 10 the original Lugols-preserved sample. Five micro-organism samples were compromised 11 during transit to the University of British Columbia, and removed from the analysis. 12 As our macro-invertebrate community composition data required the analysis of 13 multiple response variables (i.e. the density of each species), and a single explanatory 14 variable (predator treatment), we opted to use a multivariate approach.  The “adonis” function 15 from the package “vegan ” (Oksanen et al., 2012), built using the R statistical language (R 16 Development Core Team, 2013) can be used to carry out permutational analysis of variance 17 (PERMANOVA) using distance matrices, and is a generally robust method to investigate 18 differences in multivariate data. We initially ran a PERMANOVA comparing differences in 19 macro-invertebrate community structure among the three experimental treatments (no 20 predator, caged predator, uncaged predator, n = 10 for each treatment). To establish which 21 treatments differed from each other, we carried out post-hoc pair-wise comparisons of each 22 treatment pair, and applied a Bonferroni correction to avoid inflating the chance of finding 23 significant results (Holm, 1979). Differences in macro-invertebrate community structure 24 between the experimental treatments were visualised using multidimensional scaling plots 25    10 (Borg and Groenen, 2005). Multidimensional scaling uses ordination techniques to display 1 the information within a distance matrix. Each replicate is assigned a co-ordinate in each of 2 n-dimensions, by setting n = 2, data can be plotted in 2-dimensional space. Within this space, 3 replicates that are close together are similar to each other, while replicates that are far apart 4 are different. MDS plots therefore represent a method to easily illustrate similarities of 5 difference between replicates in terms of densities of multiple different species (Garpe et al., 6 2006). We subsequently used ANOVAs and post-hoc Tukey’s tests to look at the differences 7 in populations of community members, giving a biological explanation for the community 8 differences expressed in the MDS plots, and to investigate differences in community 9 respiration (dissolved CO2 concentrations in the water). In order to account for non-normality 10 of the data, mosquito, macro-invertebrate detritivore, and micro-organismal densities were 11 log transformed prior to analysis. In order to avoid inflating the chance of finding significant 12 differences due to running multiple tests, we applied a Holm-Bonferroni correction to the P-13 values generated from the ANOVAs, and report the corrected P-values in the results.  14 After we used formal statistical analysis to demonstrate the differences between 15 experimental treatments in consumer densities and dissolved CO2 concentrations, we used 16 randomised bootstrap methods to quantify the proportion of total predator effects accounted 17 for by NCEs. We randomly sampled, with replacement, 10 replicates within each treatment 18 and calculated a mean. We then calculated the difference between the caged predator mean 19 and the no predator mean (NCEs of predators only), and the difference between the uncaged 20 predator mean and the no predator mean (total predator effect). Dividing the NCEs by total 21 predator effects then gave us the relative size of the NCEs (as a percentage of total predator 22 effects). To generate a distribution, this method was repeated 10,000 times for each 23 parameter. Randomised bootstraps are generally more accurate and robust than other methods 24 of analysing the magnitude of differences between treatments (Adams and Anthony, 1996). 25    11 Additionally, as randomised bootstraps generate a distribution of differences between 1 treatment means, confidence limits around the estimate can be reported (Forbes and Hammill, 2 2013). In several instances the difference between no predator treatments and caged predator 3 treatments was greater than the difference between no predator and uncaged predator 4 treatments, resulting in the median reported size of NCEs being greater than 100% of total 5 predator effects. However for all response variables the lower 95% confidence limit was 6 <100%, demonstrating that NCEs were not significantly greater than total predator effects, 7 and values >100% are likely statistical noise.  8  9 Results 10 Macro-invertebrate community structure differed significantly among no predator, caged 11 predator, and uncaged predator treatments (f(2,27) = 14.1, p < 0.001, PERMANOVA, Figure 12 2). Pair-wise comparisons showed that community structure in control treatments was 13 different from caged predator treatments (f(1,18) = 29.4, adjusted p = 0.003, PERMANOVA, 14 Figure 2) and uncaged predator treatments (f(1,18) = 20.6, adjusted p = 0.003, PERMANOVA, 15 Figure 2). Caged and uncaged predator treatments also differed from each other (f(1,18) = 5.6 16 adjusted p = 0.006, PERMANOVA, Figure 2).  17 The effects of predator treatment on community composition appeared to be due to 18 significant density changes among treatments in both mosquitoes and macro-invertebrate 19 detritivores (Table 1, Figure 3a). Compared to no-predator controls, densities of all mosquito 20 genera were significantly reduced in the presence of caged predators (Table 1, Figure 3a), 21 and uncaged predators (Table 1, Figure 3a). Densities of Culex and Wyeomyia were 22 significantly lower in uncaged compared to caged predator treatments, while Anopheles 23 densities were not different between caged and uncaged treatments (Table 1, Figure 3a). The 24 majority of the predator-associated reductions in mosquito density for all species were due to 25    12 NCEs, as caged predators (NCEs only) caused a reduction all mosquito species > 50% that of 1 uncaged predators (Figure 4). Compared to no predator controls, all macro-invertebrate 2 detritivore densities were significantly reduced by both caged predators and uncaged 3 predators (Table 1, Figure 3b-d); although we found no evidence of significant differences 4 between caged and uncaged predator treatments for any of the benthic detritivore families 5 (Table 1, Figure 3b-d). Therefore, NCEs appeared to account for the vast majority (~100%) 6 of the total predator effects on detritivore densities (Figure 4).  7 Effects of predators on mosquito and detritivore densities appeared to differentially 8 alter densities of the microbial community (Table 1, Figure 3e). Compared to bromeliads 9 without predators, Ciliophora densities were increased by the presence of uncaged, but not 10 caged predators, while Mastigophora densities were significantly higher in the presence of 11 caged predators, and higher still when predators were uncaged (Table 1, Figure 3e). Rotifera 12 densities were unaffected by any experimental treatment (Table 1, Figure 3e). Predator NCEs 13 accounted for the minority of total predator effects (< 50%) for all micro-organism Phyla 14 (Figure 4).  15 Community respiration also significantly differed among predator treatments (Table 16 1, Figure 3f). Compared to bromeliads without predators, dissolved CO2 concentrations were 17 significantly higher in caged and uncaged predator treatments (Table 1, Figure 18 3f).Furthermore, dissolved CO2 concentrations in uncaged predator treatments were 19 significantly higher than caged predator treatments (Table 1, Figure 3f). The size of predator 20 NCEs on CO2 concentrations was a relatively large (61.4%) percentage of total predator 21 effects (95% CI = 34.8% - 96.2%, Figure 4).  22  23 Discussion 24    13 We have demonstrated that predator NCEs are strong enough to generate changes in 1 community composition, leading to altered food web structure and ecosystem processes in a 2 natural bromeliad ecosystem. Within our experiment, the threat of predation alone 3 substantially reduced macro-invertebrate densities, generating a trophic cascade that 4 increased microbial densities. We believe that higher microbial densities led to the increased 5 dissolved CO2 concentrations in the water through greater community respiration. Our study 6 suggests that predator NCEs may play a crucial role in determining community composition, 7 and differences in community composition alter ecosystem respiration.  8 Our results showed that all macro-invertebrate species decreased in the presence of 9 caged predators (NCEs only). This result provides clues as to the mechanisms by which 10 predator NCEs affected prey densities. NCEs can alter communities through changes in 11 completive interactions between prey. Prey may induce defences to predation that affect their 12 ability to compete with other species, reducing some prey species densities and increasing 13 others (Mowles, Rundle & Cotton, 2011). However, our data do not support this as all macro-14 invertebrate species decreased in the presence of caged predators, suggesting no species 15 gained an advantage over its competitors following predation risk. Although we are unable to 16 determine the exact mechanism by which caged predator treatments reduced macro-17 invertebrate densities, predator NCEs could have been mediated through changes in; i) 18 oviposition behaviour, affecting how the community was assembled, and/or ii) changes in 19 larval development rate, increasing the rate at which individuals left the community through 20 emergence. Many insect species have the ability to detect the presence of predators in an 21 ecosystem and choose to oviposit elsewhere (Brodin et al., 2006; Vonesh and Blaustein, 22 2010), reducing the number present in the community. Diptera species have also been shown 23 to increase larval development rates in response to the threat of predation (Hammill and 24 Beckerman, 2010), which would reduce the density within the community through faster 25    14 emergence rates. Although the most parsimonious explanation for predator NCEs on larval 1 macro-invertebrate densities is reduced oviposition rates and/or increased development rates, 2 we cannot discount the possibility that predator NCEs also operate via indirect means. In 3 pitcher plants, larval mosquito growth is facilitated by detrital breakdown by detritivore 4 larvae (Heard 1994), but it is unknown if similar effects occur in bromeliads.  As a result, 5 negative effects of NCEs on detritivore oviposition rates could have reduced facilitative 6 effects of detritivores on larval mosquitoes.  7 The threat of predation alone was sufficient to cause an increase in the density of 8 protists, presumably related to the decrease in mosquito densities, but not rotifers, suggesting 9 they are unaffected by changes in other trophic levels. It is unlikely that protist densities were 10 affected by predator NCEs on detritivores, as previous experiments have shown no effect of 11 free roaming odonates on protozoan (ciliates, flagellates) or rotifer densities in the absence of 12 mosquitoes (Srivastava and Bell, 2009). By contrast, increased protist densities following a 13 decrease in mosquito abundance (a trophic cascade) are well-documented in aquatic systems 14 (Eisenberg et al., 2000; Kneitel and Miller, 2002).  15 For macro-invertebrates, the proportion of total predator effects accounted for by 16 NCEs was large, >50% for all mosquito species and ~100% for Anopheles and benthic 17 detritivores. Predatory M. modesta are voracious, generalist predators, able to consume all 18 other bromeliad-dwelling insects (Srivastava et al., 2005), meaning all species will be under 19 selection pressure to avoid them.  However, we postulate that differences in the contribution 20 of NCEs between prey of M. modesta are related to the life-histories of the species involved, 21 and the mechanisms used to avoid predator encounters. Prey may avoid M. modesta in two 22 ways, adults may avoid ovipositing in bromeliads containing M. modesta, or larvae avoid 23 encounters with M. modesta within the bromeliad wells. As M. modesta are found 24 predominantly in the leaf litter at the base of bromeliad tanks (Srivastava et al., 2005), they 25    15 inhabit the same micro-habitat as benthic detritivores. This micro-habitat sharing will mean 1 benthic detritivores have a high chance of encountering predators. Anopheles larvae lack a 2 breathing siphon and are therefore constrained to the water surface, meaning when viewed 3 from underneath they are silhouetted and easily detected by M. modesta. Conversely to 4 detritivores and Anopheles, Culex and Wyeomyia larvae may move through the water column, 5 allowing them to efficiently avoid contact with predators. As it would appear benthic 6 detritivores and Anopheles have a lower ability to reduce predator encounters in the water, 7 they may experience a relatively higher pressure to avoid ovipositing in predator locations. 8 This increased pressure to avoid ovipositing in predator locations potentially explains why 9 NCEs accounted for such a large portion of total predator effects on Anopheles and benthic 10 detritivores, but not Culex and Wyeomyia.  11 The risk of odonate predation also affected community respiration, as measured by 12 concentrations of dissolved CO2 in the water. Community respiration increased in the 13 presence of both caged and uncaged predatory odonates, even though these bromeliads 14 contained fewer macro-invertebrates. A probable explanation for higher CO2 concentrations 15 in predator treatments was the increase in microbial densities. Bacterivorous protists have 16 important consequences for rates of detrital decomposition (Ribblett et al., 2005), and these 17 rates of decomposition are correlated with rates of community respiration (Young et al., 18 2008). It would appear that the higher CO2 concentrations generated through respiration and 19 decomposition by the microbial community were greater than the amount lost through 20 reductions in the macro-invertebrate community, resulting in an overall net increase in 21 community respiration. We believe it is unlikely that the presence of predators themselves 22 was enough to increase CO2 concentrations. Based on the densities of micro-organisms we 23 found and respiration estimates from previous investigations (Lawton, 1971; Glazier, 2009) 24 we estimate predatory damesflies contributed <0.1% of the CO2 generated by the micro-25    16 organisms (full calculation in Appendix 1). The increase in CO2 following addition of a 1 predator is also contrary to previous studies in the same ecosystem that showed predators 2 decrease CO2 (Atwood et al., 2013; Atwood et al., 2014). The previous investigations by 3 Atwood et al (2013, 2014) used a 3-tier food chain (Damselfly-macroinvertebrates-detritus), 4 whereas the present investigation contained a 4-tier component (Damselfly-mosquito-protists 5 and rotifers-detritus). Our results therefore agree with earlier investigations showing top 6 predators increase CO2 production in 4-teir food chains (Schindler et al., 1997). The results 7 we present, in conjunction with the earlier work of Atwood (2013, 2014) and Schindler 8 (1997) demonstrate the relationship between predator effects on CO2 production and food 9 chain length. Encouraging top predators has been proposed as a management initiative to 10 reduce atmospheric CO2 (Schmitz et al., 2013), we propose that without knowing food chain 11 length, encouraging predators may increase rather than decrease CO2 concentrations.  12 The use of cages to isolate predators from the remainder of the community allows 13 chemical cues to disperse through the ecosystem (Hettyey et al., 2010), but means that prey 14 are unable to use visual or tactile cues to detect predators. There is also growing evidence that 15 prey species choose locations within a landscape based not only on the quality of the location 16 itself, but also the quality of nearby locations (Resetarits and Binckley, 2009). Previous 17 studies have shown prey avoid predator-free ecosystems that are located close to ecosystems 18 containing predators (Resetarits and Binckley, 2009). Within our experiment, all plants were 19 ~ 1.5 m apart, and arranged randomly (to avoid “clumping” of treatments). However, we 20 cannot rule out that the proximity of bromeliads containing predators made our predator-free 21 control plants more or less desirable to ovipositing females than “isolated” predator-free 22 plants would have been.  23  The addition of mosquitoes and chironomids to our caged treatment as food items 24 would have introduced nutrients into the ecosystem, potentially providing extra resources for 25    17 the microbial community. While we cannot rule out that these extra nutrients may have had 1 an effect, the number of prey items introduced as food was relatively low compared to the 2 densities of individuals that naturally colonised (< 8% in any one treatment). In terms of 3 carbon additions, we estimate the total carbon content added as food in the caged predator 4 treatments to be <5% of what was originally added as leaf litter (full calculation given in 5 Appendix 2). This estimation is based on length-weight regressions of the food items (Sabo et 6 al., 2002), and estimations of the carbon content of aquatic insects (Kraus and Vonesh, 2012) 7 and leaf litter (Martin and Thomas, 2011). The estimation of percentage carbon introduced as 8 food does not account for leaf litter falling into bromeliads naturally through the experiment, 9 suggest the estimated 5% contribution may be high. Our results also suggest that nutrients 10 being introduced as food items did not substantially affect the ecological community as 11 micro-organism densities were lower in our caged predator treatments than in our uncaged 12 predator treatments (increased nutrient additions should generate the opposite effect).  The 13 low amount of nutrients contributed by food additions, and the relatively low density of 14 micro-organisms (compared to uncaged predators treatments) suggests the addition of prey 15 into our caged predators treatments did not bias our results. 16 In many complex natural systems, predator NCEs may be overlooked due to 17 difficulties associated with the quantification of their relative contribution. However, we 18 show that failing to quantify and account for NCEs may lead to misunderstandings of the 19 mechanisms by which predators affect community assembly, food web structure and 20 ecosystem function. Here we show that predator NCEs altered prey densities, and produced 21 trophic cascades that affected ecosystem processes. Global predator densities are in serious 22 decline (Estes et al., 2011), and management strategies designed to replace predators 23 ecologically must account for both consumptive and non-consumptive predator effects. In a 24 changing world, failing to understand the consequences of predator NCEs may have serious 25    18 implications for the structure of natural communities, and the ecological functions they 1 perform.  2  3 Acknowledgments Permission to work in the Área de Conservación Guanacaste (ACG) was 4 obtained from MINAE, the Costa Rican ministry for environment and energy. Particular 5 thanks to Róger Blanco for assistance with administration and logistics. This work was 6 funded through an NSERC E.W.R. Steacie Memorial Fellowship awarded to D.S.S. We 7 thank Calixto Moraga and Petrona Rios Castro for invaluable help in the field. We would like 8 to also thank Oswald J. Schmitz for his insightful comments on earlier versions of this 9 manuscript. 10 11    19 References 1 Adams, D.C., Anthony, C.D., 1996. Using randomization techniques to analyse behavioural 2 data. Animal Behaviour 51, 733-738. 3 Atwood, T., Hammill, E., Srivastava, D., Richardson, J., 2014. Competitive displacement 4 alters top-down effects on carbon dioxide concentrations in a freshwater ecosystem. 5 Oecologia 75, 353-361. 6 Atwood, T.B., Hammill, E., Grieg, H., Kratina, P., Shurin, J.B., Srivastava, D.S., Richardson, 7 J.S., 2013. Predator-induced reduction of freshwater carbon dioxide emissions. Nature 8 Geoscience 6, 191-194. 9 Berendonk, T.U., Bonsall, M.B., 2002. The phantom midge and a comparison of 10 metapopulation structures. Ecology 83, 116-128. 11 Borg, I., Groenen, P., 2005. Modern multidimensional scaling: Theory and applications. 12 Springer-Verlag, New York. 13 Brodin, T., Johansson, F., Bergsten, J., 2006. Predator related oviposition site selection of 14 aquatic beetles (Hydroporus spp.) and effects on offspring life-history. Freshwater Biology 15 51, 1277-1285. 16 Cadotte, M.W., Carscadden, K., Mirotchnick, N., 2011. Beyond species: functional diversity 17 and the maintenance of ecological processes and services. Journal of Applied Ecology 48, 18 1079-1087. 19 Carpenter, S.R., 1996. Microcosm experiments have limited relevance for community and 20 ecosystem ecology. Ecology 77, 677-680. 21 Cole, J.J., Prairie, Y.T., Caraco, N.F., McDowell, W.H., Tranvik, L.J., Striegl, R.G., Duarte, 22 C.M., Kortelainen, P., Downing, J.A., Middelburg, J.J., Melack, J., 2007. Plumbing the 23 global carbon cycle: Integrating inland waters into the terrestrial carbon budget. Ecosystems 24 10, 171-184. 25    20 Del Giorgio, P.A., Cole, J.J., Caraco, N.F., Peters, R.H., 1999. Linking planktonic biomass 1 and metabolism to net gas fluxes in northern temperate lakes. Ecology 80, 1422-1431. 2 Eisenberg, J.N.S., Washburn, J.O., Schreiber, S.J., 2000. Generalist feeding behaviors of 3 Aedes sierrensis larvae and their effects on protozoan populations. Ecology 81, 921-935. 4 Estes, J.A., Terborgh, J., Brashares, J.S., Power, M.E., Berger, J., Bond, W.J., Carpenter, 5 S.R., Essington, T.E., Holt, R.D., Jackson, J.B.C., Marquis, R.J., Oksanen, L., Oksanen, T., 6 Paine, R.T., Pikitch, E.K., Ripple, W.J., Sandin, S.A., Scheffer, M., Schoener, T.W., Shurin, 7 J.B., Sinclair, A.R.E., Soule, M.E., Virtanen, R., Wardle, D.A., 2011. Trophic downgrading 8 of planet earth. Science 333, 301-306. 9 Forbes, C., Hammill, E., 2013. Fear in the dark? Community-level effects of non-lethal 10 predators change with light regime. Oikos 122, 1662-1668. 11 Garpe, K.C., Yahya, S.A.S., Lindahl, U., Ohman, M.C., 2006. Long-term effects of the 1998 12 coral bleaching event on reef fish assemblages. Marine Ecology Progress Series 315, 237-13 247. 14 Glazier, D.S., 2009. Metabolic level and size scaling of rates of respiration and growth in 15 unicellular organisms. Functional Ecology 23, 963-968. 16 Hammill, E., Beckerman, A.P., 2010. Reciprocity in predator-prey interactions: exposure to 17 defended prey and predation risk affects intermediate predator life history and morphology. 18 Oecologia 163, 193-202. 19 Hettyey, A., Zsarnoczai, S., Vincze, K., Hoi, H., Laurila, A., 2010. Interactions between the 20 information content of different chemical cues affect induced defences in tadpoles. Oikos 21 119, 1814-1822. 22 Hill, J.M., Weissburg, M.J., 2013. Predator biomass determines the magnitude of non-23 consumptive effects (NCEs) in both laboratory and field environments. Oecologia 172, 79-24 91. 25    21 Holm, S., 1979. A simple sequentially rejective multiple test procedure. Scandinavian Journal 1 of Statistics 6, 65-70. 2 Hooper, D.U., Adair, E.C., Cardinale, B.J., Byrnes, J.E.K., Hungate, B.A., Matulich, K.L., 3 Gonzalez, A., Duffy, J.E., Gamfeldt, L., O'Connor, M.I., 2012. A global synthesis reveals 4 biodiversity loss as a major driver of ecosystem change. Nature 486, 105-U129. 5 Hope, D., Dawson, J.J.C., Cresser, M.S., Billett, M.F., 1995. A method for measuring free 6 CO2 in upland streamwater using headspace analysis. Journal of Hydrology 166, 1-14. 7 Kneitel, J.M., Miller, T.E., 2002. Resource and top-predator regulation in the pitcher plant 8 (Sarracenia purpurea) inquiline community. Ecology 83, 680-688. 9 Kraus, J.M., Vonesh, J.R., 2010. Feedbacks between community assembly and habitat 10 selection shape variation in local colonization. Journal of Animal Ecology 79, 795-802. 11 Kraus, J.M., Vonesh, J.R., 2012. Fluxes of terrestrial and aquatic carbon by emergent 12 mosquitoes: a test of controls and implications for cross-ecosystem linkages. Oecologia 170, 13 1111-1122. 14 Lawton, J.H., 1971. Ecological energetics studies on larvae of damselfly Pyrrhosoma 15 nymphula (sulzer) (Odonata-Zygoptera). Journal of Animal Ecology 40, 385-423. 16 Martin, A.R., Thomas, S.C., 2011. A reassessment of carbon content in tropical trees. Plos 17 One 6, e23533. 18 Martinson, G.O., Werner, F.A., Scherber, C., Conrad, R., Corre, M.D., Flessa, H., Wolf, K., 19 Klose, M., Gradstein, S.R., Veldkamp, E., 2010. Methane emissions from tank bromeliads in 20 neotropical forests. Nature Geoscience 3, 766-769. 21 Ngai, J.T., Srivastava, D.S., 2006. Predators accelerate nutrient cycling in a bromeliad 22 ecosystem. Science 314, 963-963. 23    22 Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O'Hara, R.B., Simpson, 1 G.L., Solymos, P., Henry, M., Stevens, H., Wagner, H., 2012. vegan: Community Ecology 2 Package. R package version 2.0-3. 3 Peacor, S.D., Pangle, K.L., Schiesari, L., Werner, E.E., 2012. Scaling-up anti-predator 4 phenotypic responses of prey: impacts over multiple generations in a complex aquatic 5 community. Proceedings of the Royal Society B-Biological Sciences 279, 122-128. 6 Peacor, S.D., Werner, E.E., 2001. The contribution of trait-mediated indirect effects to the net 7 effects of a predator. Proceedings of the National Academy of Sciences of the United States 8 of America 98, 3904-3908. 9 Preisser, E.L., Orrock, J.L., Schmitz, O.J., 2007. Predator hunting mode and habitat domain 10 alter nonconsumptive effects in predator-prey interactions. Ecology 88, 2744-2751. 11 R Development Core Team, 2013. R: A language and environment for statistical computing. 12 In: Computing, R.F.f.S. (Ed.), Vienna. 13 Resetarits, W.J., Jr., Binckley, C.A., 2009. Spatial contagion of predation risk affects 14 colonization dynamics in experimental aquatic landscapes. Ecology 90, 869-876. 15 Ribblett, S.G., Palmer, M.A., Coats, D.W., 2005. The importance of bacterivorous protists in 16 the decomposition of stream leaf litter. Freshwater Biology 50, 516-526. 17 Sabo, J.L., Bastow, J.L., Power, M.E., 2002. Length-mass relationships for adult aquatic and 18 terrestrial invertebrates in a California watershed. Journal of the North American 19 Benthological Society 21, 336-343. 20 Schindler, D.E., Carpenter, S.R., Cole, J.J., Kitchell, J.F., Pace, M.L., 1997. Influence of food 21 web structure on carbon exchange between lakes and the atmosphere. Science 277, 248-251. 22 Schmitz, O., Raymond, P., Estes, J., Kurz, W., Holtgrieve, G., Ritchie, M., Schindler, D., 23 Spivak, A., Wilson, R., Bradford, M., Christensen, V., Deegan, L., Smetacek, V., Vanni, M., 24 Wilmers, C., 2013. Animating the carbon cycle. Ecosystems 17, 344-359. 25    23 Schmitz, O.J., Beckerman, A.P., Obrien, K.M., 1997. Behaviorally mediated trophic 1 cascades: Effects of predation risk on food web interactions. Ecology 78, 1388-1399. 2 Schulze, E.-D., Mooney, H.A., 1994. Biodiversity and ecosystem function: with 22 tables. 3 Springer. 4 Srivastava, D.S., 2006. Habitat structure, trophic structure and ecosystem function: 5 interactive effects in a bromeliad-insect community. Oecologia 149, 493-504. 6 Srivastava, D.S., Bell, T., 2009. Reducing horizontal and vertical diversity in a foodweb 7 triggers extinctions and impacts functions. Ecology Letters 12, 1016-1028. 8 Srivastava, D.S., Melnychuk, M.C., Ngai, J.T., 2005. Landscape variation in the larval 9 density of a bromeliad-dwelling zygopteran, Mecistogaster modesta (Odonata: 10 Pseudostigmatidae). International Journal of Odonatology 8, 67-79. 11 Srivastava, D.S., Trzcinski, M.K., Richardson, B.A., Gilbert, B., 2008. Why are predators 12 more sensitive to habitat size than their prey? insights from bromeliad insect food webs. 13 American Naturalist 172, 761-771. 14 Strickland, M.S., Hawlena, D., Reese, A., Bradford, M.A., Schmitz, O.J., 2013. Trophic 15 cascade alters ecosystem carbon exchange. Proceedings of the National Academy of Sciences 16 of the United States of America 110, 11035-11038. 17 Tollrian, R., Harvell, C.D., 1999. The Ecology and Evolution of Inducible Defenses. 18 Princeton University Press, Princeton. 19 Trussell, G.C., Ewanchuk, P.J., Matassa, C.M., 2006a. Habitat effects on the relative 20 importance of trait- and density-mediated indirect interactions. Ecology Letters 9, 1245-1252. 21 Trussell, G.C., Ewanchuk, P.J., Matassa, C.M., 2006b. The fear of being eaten reduces 22 energy transfer in a simple food chain. Ecology 87, 2979-2984. 23    24 Vonesh, J.R., Blaustein, L., 2010. Predator-induced shifts in mosquito oviposition site 1 selection: a meta-analysis and implications for vector control. Israel Journal of Ecology & 2 Evolution 56, 263-279. 3 Vonesh, J.R., Kraus, J.M., Rosenberg, J.S., Chase, J.M., 2009. Predator effects on aquatic 4 community assembly: disentangling the roles of habitat selection and post-colonization 5 processes. Oikos 118, 1219-1229. 6 Werner, E.E., Anholt, B.R., 1996. Predator-induced behavioral indirect effects: 7 Consequences to competitive interactions in anuran larvae. Ecology 77, 157-169. 8 Young, R.G., Matthaei, C.D., Townsend, C.R., 2008. Organic matter breakdown and 9 ecosystem metabolism: functional indicators for assessing river ecosystem health. Journal of 10 the North American Benthological Society 27, 605-625. 11  12  13 14    25 Appendix 1. Contribution of damselfly larva to community respiration 1 As the addition of a damselfly in the predator treatments may increase community 2 respiration, we estimated total microbial microbial respiration and compare it to the 3 respiration rates of damselfly to help ascertain the major source of CO2. The micro-organisms 4 we measured ranged in size, with a volume of ~3500µm3 being average. According to Galizer 5 (Glazier, 2009) a single celled heterotroph of this size will have a metabolic rate (nL O2 cell-1 6 h-1) of 0.01. In the caged predator treatment containing the lowest micro-organismal 7 densities, the total bromeliad contained ~450000000 protists, meaning the total respiration 8 per hour attributable to the micro-organisms was 450000000*0.01 = 4500000 nL O2 cell-1 h-1, 9 or 450µL per hour. Damselfly larvae respire at a rate of between 0.2 and 3.0µL per hour 10 (Lawton, 1971), suggesting that even at their highest respiration rates the single caged 11 damselfly we added to the predator treatments would contribute less than 0.1% of the CO2 12 that is attributable to micro-organisms. 13 In addition to the above calculation, our conclusion that community respiration is 14 attributable mainly to the microbial increase is justified on the basis of two previous studies.  15 These previous investigations used the exact same system, but without the mosquito-microbe 16 component of the food web (Atwood et al., 2013; Atwood et al., 2014). In both of these 17 studies, predator presence reduced CO2 of three-tier bromeliad food webs. The difference in 18 the results of the current investigation and these previous papers is attributed to differences in 19 food chain length. In the previous papers the authors study a simplified odd-numbered food 20 chain, while the current investigation looks at a more reticulate even-numbered food chain. If 21 the increase in community respiration we observe in the present study were due mainly to the 22 addition of predators, we would have expected to see an increase in community respiration 23 (and therefore CO2) in these earlier studies, when in fact we see the opposite.  24  25    26 Appendix 2. Calculation of the potential for macroinvertebrates offered as food to act as 1 a nutrient subsidy.  2 The possibility exists that increased CO2 concentrations in the caged predator treatments 3 (NCEs alone) may have been due to nutrient subsidies offered to the predator as food. To 4 estimate the importance of the carbon added as food, we calculated the total mass of carbon 5 added as food, and compared this value to the amount of carbon added to all bromeliads as 6 leaf litter.  7  Leaf litter from tropical forests is composed of ~47% carbon (Martin and Thomas, 8 2011). In each bromeliad, we added 200mg (dry weight) leaf litter per 100ml plant volume. 9 The mean volume across plants was 1013.5ml, meaning we added 2027 mg dry weight leaf 10 litter, which equates to 952.7mg C. We added a total of 30 food items (mosquito and 11 chironomid larvae) to each of the caged treatments over the course of the experiment. These 12 food items were all ~5mm total length, we therefore used the length of 5mm to calculate dry 13 mass using length-weight regression values for aquatic insects (Sabo et al., 2002). Using 14 these values, we calculate we added a maximum of 98.6mg of food, which equates to 15 44.37mg C according to estimates of the carbon composition of aquatic insects (Kraus and 16 Vonesh, 2012). This therefore means that the amount of carbon we introduced as food in the 17 caged treatments was < 5% of what was initially introduced as leaf litter. This value of 5% is 18 around 1/10th of the size of the error bars around the CO2 estimate in figure 4, suggesting that 19 the contribution to CO2 concentrations made by the addition of food items has not 20 substantially affected the results. This value of 5% would also be an over-estimate as it does 21 not include carbon introduced to all replicates as leaf litter falling naturally from the trees 22 during the experiment. For these reasons we believe that although adding food items to the 23 caged predator treatments may have slightly increased CO2 concentrations, the observed 24 difference in CO2 concentrations between our no predator and caged predator treatments is 25    27 primarily due to changes in community composition, rather than the introduction of food 1 items.   2 3    28 Table legends 1 Table 1. Results from ANOVAs performed on all species and CO2 concentrations, with 2 Tukey tests where applicable.  “Corrected P” denotes the P-value following Holm-Bonferroni 3 correction (Holm, 1979). Bolded values indicate significant differences at 0.05 level.  4 5    29 Figure legends 1 Figure 1. (a) Illustration of the structural nature of Guzmania bromeliads, and the technique 2 used to generate experimental treatments, damselfly larvae and tubes enlarged 4x relative to 3 bromeliads to improve clarity.  (i) No predator treatments, (ii) Caged predator treatments, 4 non-consumptive effects (NCEs) only, (iii) Uncaged predator treatments, both consumptive 5 effects and NCEs. (b) Simplified bromeliad food web demonstrating proposed energy flow 6 between trophic groups.  7  8 Figure 2. Multidimensional scaling plot (MDS) illustrating how predators alter the macro-9 invertebrate community composition in bromeliads. Within the plot, each point represents a 10 single bromeliad community. Treatments containing predators are represented by circles, 11 either caged (NCEs) , or uncaged (total predator effects ). No predator controls (no 12 predator effects) are represented by crosses (). The distance between points is proportional 13 to the similarity in community composition, meaning nearby points represent similar 14 communities. 15  16 Figure 3. Predator effects on densities of (a), mosquitoes (filter-feeders) (b), Chironomidae 17 (c), Tipulidae, (d), Scirtidae, (e), Micro-organisms and (f) in situ CO2 concentrations of 18 bromeliad ecosystems. Caged predator treatments were only exposed to the non-consumptive 19 effects of predators, while uncaged predator treatments were exposed to both the non-20 consumptive and consumptive effects of predators.  Different letters denote treatments that 21 differ significantly from each other according to post-hoc Tukeys testing within a genus (a), 22 family (b-d), or phyla (f). Bars represent means ± standard errors.  23  24    30 Figure 4. Relative magnitude of non-consumptive effects (NCEs), compared to total predator 1 effects, on species and families of organisms, as well as CO2 concentrations, within 2 bromeliad communities. For parameters where the bar height is greater than 100% the 3 difference between caged predator and no predator treatments was greater than the difference 4 between uncaged predator and no predator treatments. Data are means ± 95% confidence 5 limits.   6 7    31 Table 1 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17 18  f-statistic corrected P Tukey results (predator treatment)    none vs.  uncaged none vs.  caged caged vs.  uncaged Mosquitoes     Culex f(2,27) = 102.6 <0.001 <0.001 <0.001 0.002 Wyoemyia f(2,27) = 37.4 <0.001 <0.001 <0.001 0.007 Anopheles f(2,27) = 13.1 <0.001 0.003 0.003 0.991 Benthic detritivores     Chironomidae f(2,27) = 6.94 0.012 0.007 0.010 0.990 Tipulidae f(2,27) = 6.17 0.012 0.009 0.024 0.910 Scirtidae f(2,27) = 4.19 0.026 0.032 0.073 0.920 Micro-organisms     Ciliphora f(2,22) = 9.755 <0.001 0.61 0.001 0.008 Mastigophora f(2,22) = 23.8 <0.001 0.002 <0.001 0.010 Rotifera f(2,22) = 1.75 0.20 NA NA NA Respiration     CO2 f(2,21) = 11.42 0.004 0.003 <0.001 0.010    32 Figure 1 1  2  3  4 5    33 Figure 2 1  2  3 4    34 Figure 3 1  2  3 4    35 Figure 4 1  2  3 4    36 Supplementary material 1 Appendix 1. Contribution of damselfly larva to community respiration 2 As the addition of a damselfly in the predator treatments may increase community 3 respiration, we estimated total microbial microbial respiration and compare it to the 4 respiration rates of damselfly to help ascertain the major source of CO2. The micro-5 organisms we measured ranged in size, with a volume of ~3500µm3 being average. 6 According to Galizer (Glazier 2009) a single celled heterotroph of this size will have a 7 metabolic rate (nL O2 cell-1 h-1) of 0.01. In the caged predator treatment containing the 8 lowest micro-organismal densities, the total bromeliad contained ~450000000 protists, 9 meaning the total respiration per hour attributable to the micro-organisms was 10 450000000*0.01 = 4500000 nL O2 cell-1 h-1, or 450µL per hour. Damselfly larvae respire 11 at a rate of between 0.2 and 3.0µL per hour (Lawton 1971), suggesting that even at their 12 highest respiration rates the single caged damselfly we added to the predator 13 treatments would contribute less than 0.1% of the CO2 that is attributable to micro-14 organisms. 15 In addition to the above calculation, our conclusion that community respiration 16 is attributable mainly to the microbial increase is justified on the basis of two previous 17 studies.  These previous investigations used the exact same system, but without the 18 mosquito-microbe component of the food web (Atwood et al. 2013, Atwood et al. 2014). 19 In both of these studies, predator presence reduced CO2 of three-tier bromeliad food 20 webs. The difference in the results of the current investigation and these previous 21 papers is attributed to differences in food chain length. In the previous papers the 22 authors study a simplified odd-numbered food chain, while the current investigation 23 looks at a more reticulate even-numbered food chain. If the increase in community 24 respiration we observe in the present study were due mainly to the addition of 25    37 predators, we would have expected to see an increase in community respiration (and 1 therefore CO2) in these earlier studies, when in fact we see the opposite.  2  3 Appendix 2. Calculation of the potential for macroinvertebrates offered as food to 4 act as a nutrient subsidy.  5 The possibility exists that increased CO2 concentrations in the caged predator 6 treatments (NCEs alone) may have been due to nutrient subsidies offered to the 7 predator as food. To estimate the importance of the carbon added as food, we calculated 8 the total mass of carbon added as food, and compared this value to the amount of 9 carbon added to all bromeliads as leaf litter.  10  Leaf litter from tropical forests is composed of ~47% carbon (Martin and 11 Thomas 2011). In each bromeliad, we added 200mg (dry weight) leaf litter per 100ml 12 plant volume. The mean volume across plants was 1013.5ml, meaning we added 2027 13 mg dry weight leaf litter, which equates to 952.7mg C. We added a total of 30 food items 14 (mosquito and chironomid larvae) to each of the caged treatments over the course of 15 the experiment. These food items were all ~5mm total length, we therefore used the 16 length of 5mm to calculate dry mass using length-weight regression values for aquatic 17 insects (Sabo et al. 2002). Using these values, we calculate we added a maximum of 18 98.6mg of food, which equates to 44.37mg C according to estimates of the carbon 19 composition of aquatic insects (Kraus and Vonesh 2012). This therefore means that the 20 amount of carbon we introduced as food in the caged treatments was < 5% of what was 21 initially introduced as leaf litter. This value of 5% is around 1/10th of the size of the 22 error bars around the CO2 estimate in figure 4, suggesting that the contribution to CO2 23 concentrations made by the addition of food items has not substantially affected the 24 results. This value of 5% would also be an over-estimate as it does not include carbon 25    38 introduced to all replicates as leaf litter falling naturally from the trees during the 1 experiment. For these reasons we believe that although adding food items to the caged 2 predator treatments may have slightly increased CO2 concentrations, the observed 3 difference in CO2 concentrations between our no predator and caged predator 4 treatments is primarily due to changes in community composition, rather than the 5 introduction of food items.   6  7 References for Supplementary material 8 Atwood, T., E. Hammill, D. Srivastava, and J. Richardson. 2014. Competitive 9 displacement alters top-down effects on carbon dioxide concentrations in a 10 freshwater ecosystem. Oecologia 75:353-361. 11 Atwood, T. B., E. Hammill, H. Grieg, P. Kratina, J. B. Shurin, D. S. Srivastava, and J. S. 12 Richardson. 2013. Predator-induced reduction of freshwater carbon dioxide 13 emissions. Nature Geoscience 6:191-194. 14 Glazier, D. S. 2009. Metabolic level and size scaling of rates of respiration and growth in 15 unicellular organisms. Functional Ecology 23:963-968. 16 Kraus, J. M., and J. R. Vonesh. 2012. Fluxes of terrestrial and aquatic carbon by emergent 17 mosquitoes: a test of controls and implications for cross-ecosystem linkages. 18 Oecologia 170:1111-1122. 19 Lawton, J. H. 1971. Ecological energetics studies on larvae of damselfly Pyrrhosoma 20 nymphula (sulzer) (Odonata-Zygoptera). Journal of Animal Ecology 40:385-423. 21 Martin, A. R., and S. C. Thomas. 2011. A reassessment of carbon content in tropical trees. 22 Plos One 6:e23533. 23    39 Sabo, J. L., J. L. Bastow, and M. E. Power. 2002. Length-mass relationships for adult 1 aquatic and terrestrial invertebrates in a California watershed. Journal of the 2 North American Benthological Society 21:336-343. 3  4  5  6 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.52383.1-0132600/manifest

Comment

Related Items