Contaminants in Linked Aquatic–Terrestrial Ecosystems: Predicting Effects of Aquatic Pollution on Adult Aquatic Insects and Terrestrial Insectivores

BRIDGES*

Contaminants in linked aquatic–terrestrial ecosystems: Predicting effects of aquatic pollution on adult aquatic insects and terrestrial insectivores

Johanna M. Kraus1,2

1US Geological Survey, Columbia Environmental Research Center, 4200 New Haven Road Columbia, Missouri 65201 USA

Abstract: Organisms that move across ecosystem boundaries connect food webs in apparently disparate locations. As part of their life cycle, aquatic insects transition from aquatic larvae to terrestrial adults, thereby linking freshwater ecosystem processes and terrestrial insectivore dynamics. These linkages are strongly affected by contamination of freshwater ecosystems, which can reduce production of adult aquatic insects (i.e., emergence), increase contaminant concentrations in adult insect tissues, and alter contaminant flux to terrestrial ecosystems. Despite the potential impact of contaminants on adult aquatic insects, little is known about predicting these effects. Here, I develop a heuristic model based on contaminant properties and ecotoxicological principles to predict the effects of various classes of aquatic contaminants on adult aquatic insects and discuss implications for terrestrial insectivores living near contaminated freshwaters. The main finding is that contaminant classes vary greatly in how their biologically-mediated effects on aquatic insects affect terrestrial insectivores. Highly bioaccumulative contaminants that are well retained during metamorphosis, like polychlorinated biphenyls (PCBs), are often non-toxic to aquatic insect larvae at concentrations commonly found in the environment. Such contaminants flux from aquatic ecosystems in large quantities in the bodies of emerging adult aquatic insects and expose terrestrial insectivores to toxic levels of pollution. On the other hand, contaminants that are less bioaccumulative, excreted during metamorphosis, and more toxic to insects, like trace metals, tend to affect terrestrial insectivores by reducing production of adult aquatic insects on which they prey. Management applications of this model illustrate type and severity of risk of aquatic contaminants to consumers of adult aquatic insects. Key words: spatial subsidies, aquatic–terrestrial linkages, food web, conceptual model, metamorphosis, insects

The fates of food webs in neighboring ecosystems are often fatty acids, contaminant concentration) of emerging adult spatially linked by movements of nutrients, resources, and aquatic insects can reduce insectivore abundance and bio- predators (Polis et al. 1997, Nakano and Murakami 2001, mass, lower insectivore reproductive success, and increase Baxter et al. 2005, Allen and Wesner 2016). In freshwater predation pressure on alternative prey with consequences ecosystems, aquatic insects that metamorphose from for aquatic and terrestrial food webs (Baxter et al. 2005, Rohr aquatic larvae to terrestrial adults play key roles as prey et al. 2006, Marcarelli et al. 2011). Contaminants, which are subsidies that move nutrients and energy from aquatic to increasingly recognized as a global stressor of aquatic eco- terrestrial food webs (Jackson and Fisher 1986, Nakano systems (Lavoie et al. 2013, Morrissey et al. 2015, Bernhardt and Murakami 2001, Baxter et al. 2005, Allen and Wesner et al. 2017), strongly affect emergence production, assem- 2016). Interruption or modiﬁcation of the magnitude, tim- blage composition, timing of adult aquatic insect emer- ing, or dietary value (e.g., stoichiometry, polyunsaturated gence, concentration of contaminants in adult insect tissues,

E-mail address: [email protected]

*BRIDGES is a recurring feature of FWS intended to provide a forum for the interchange of ideas and information relevant to FWS readers, but beyond the usual scope of a scientiﬁc paper. Articles in this series will bridge from aquatic ecology to other disciplines, e.g., political science, economics, education, chemistry, or other biological sciences. Papers may be complementary or take alternative viewpoints. Authors with ideas for topics should contact BRIDGES Co-Editors, Sally Entrekin ([email protected]) and Allison Roy ([email protected]).

DOI: 10.1086/705997. Received 15 October 2018; Accepted 30 June 2019; Published online 23 September 2019. Freshwater Science. 2019. 38(4):919–927. © 2019 by The Society for Freshwater Science. 919 920 | Contaminant impacts on aquatic–terrestrial linkages J. M. Kraus and contaminant flux from aquatic to terrestrial food webs about physicochemical characteristics and ecotoxicolog- (Fig. 1; Menzie 1980, Clements et al. 2000, Kraus et al. ical effects of various contaminant classes to develop a heu- 2014a, Chumchal and Drenner 2015). Despite the impor- ristic model that predicts contaminant concentrations in tance of these impacts for terrestrial and aquatic insec- aquatic insects, impacts on emergence production, and flux tivores (e.g., birds, bats, lizards, spiders, rodents, and fish; of contaminants to terrestrial ecosystems via adult aquatic Walters et al. 2008, Sullivan and Rodewald 2012, Kraus insects. This approach complements efforts that conversely et al. 2014a, Chumchal and Drenner 2015, Schiesari et al. use basic ecological principles to predict ecotoxicological 2018), no model currently exists for predicting effects of effects of contaminants on aquatic communities and eco- aquatic contaminants on adult aquatic insects. systems (Rohr et al. 2006, Halstead et al. 2014). I discuss po- Contaminant impacts on adult aquatic insects vary some- tential impacts of model outcomes on terrestrial insectivores, what predictably both among and within contaminant as well as limitations of the model and need for future studies. classes as a function of their toxicity, propensity to bioac- Finally, I provide examples of how this model is currently be- cumulate, and retention in tissues across metamorphosis. ing applied to inform research approaches and management For example, some contaminants strongly reduce metamor- strategies of freshwaters in historically mined, industrially phosis and emergence production (including changes in as- polluted, and agricultural landscapes. Understanding how semblage composition and timing), whereas other contam- the effects of stressors like aquatic contaminants cascade inants bioaccumulate in insect larvae and are retained in across ecosystem boundaries is an important challenge for their tissues across metamorphosis to adulthood (Kraus applied ecology and ecosystem management (Muehlbauer et al. 2014b, Wesner et al. 2019). Trace metal contamination et al. 2019). and neonicotinoid insecticides strongly reduce adult aquatic insect emergence and are implicated in the global decline in HEURISTIC MODEL insect biomass (Hallmann et al. 2014, Kraus et al. 2014a, To predict the potential impacts of aquatic contami- Morrissey et al. 2015, Cavallaro et al. 2017), thus eliminat- nants on emergence production, adult concentrations, and ing an important prey source for many terrestrial insecti- contaminant flux, I modeled the relationship between tox- vores including spiders, young waterfowl, aerial insectivo- icity and contaminant retention during metamorphosis for rous birds, and bats (Hallmann et al. 2014, Morrissey et al. various contaminant classes (Figs 2, 3). It is not possible to 2015). In contrast, other contaminants, such as persistent quantitatively model multiple contaminant classes along organic pollutants and organometals (and possibly algal tox- the same toxicity axis because different classes use differ- ins, personal care products, and pharmaceuticals), do not ent metrics of toxicity. Thus, to examine differences among necessarily reduce emergence production at a site even classes, I qualitatively mapped the relative toxicity and con- though they also bioaccumulate in insect larvae and strongly taminant retention across metamorphosis by contaminant persist in their tissues across metamorphosis (Walters et al. class (Table 1, Fig. 2). To derive predictions about relative 2008, Tweedy et al. 2013, Moy et al. 2016, Kraus et al. 2017, effects of individual contaminants within classes, I used least Richmond et al. 2018). When these highly-contaminated squares linear regressions to model the relationships be- adult aquatic insects fall prey to terrestrial insectivores, they tween toxicity and retention across insect metamorphosis become vectors of contaminant exposure and reduce in- (i.e., the natural log ratio of adult to larval concentrations) sectivore reproduction, health, and juvenile success (Custer within 3 major contaminant classes (i.e., metals, polycyclic et al. 2003, Walters et al. 2010, Chumchal and Drenner aromatic hydrocarbons [PAHs], and polychlorinated bi- 2015). phenyls [PCBs]) based on published toxicity values for con- To increase understanding of how aquatic contamina- taminants with known retention across metamorphosis tion alters aquatic–terrestrial linkages, I use information (Fig. 3).

Figure 1. Flux of aquatic contaminants to the terrestrial ecosystem by adult aquatic insects as a function of emergence production and contaminant concentration within adult insect bodies. White circles represent contaminant ions or molecules within insect bodies. Insect images are courtesy of Jeremy Monroe at Freshwaters Illustrated. Volume 38 December 2019 | 921

Wu et al. 2012, Kraus et al. 2014b). Thus, softer metals may more readily lower emergence production and increase adult aquatic insect tissue concentrations (McCloskey et al. 1996, Kinraide 2009, Li et al. 2012, Wu et al. 2012) than harder metals. Organic compounds vary markedly in terms of their predicted impacts on emergence production and adult insect tissue concentrations. For non-polar organic contaminants with intermediate lipophilicity (i.e., fat solu- bility; log of the octenol-water partitioning coefﬁcient [log Kow] 5 5 – 7), retention across metamorphosis initially increases with increased toxicity, then levels out for compounds with log Kow of ~6 (Fig. 3B; Schultz et al. 1990, Li et al. 2012, Kraus et al. 2014b, Fu et al. 2015). For more hydrophilic non-polar organic compounds (log Kow < 5), retention across metamorphosis decreases with increased toxicity (Fig. 3B; Schultz et al. 1990, Kraus et al. 2014b, Figure 2. Heuristic Model Part I: Predicting effects on emergence of adult aquatic insects and contaminant concentrations Fu et al. 2015). For highly bioaccumulative organic contam- among various classes of aquatic contaminants. As toxicity of a inants (log Kow ~7) like PCBs and tetrachlorodibenzo- given compound/element increases, adult aquatic insect emer- dioxins, relative toxicity within these classes is not necessar- gence is predicted to decrease, leading to loss of prey resources ily correlated with retention (Kraus et al. 2014b) and has for insectivorous consumers. As persistence of the contaminant more to do with congener structure (Fig. 3C; Van den Berg across insect metamorphosis increases, the likelihood of effects et al. 1998, 2006). Thus, to (grossly) generalize for organic of exposure on consumers increases. Gray text shows approxi- contaminants, hydrophilic compounds are more likely to re- mate position within this framework of different classes of con- duce emergence and have lower contaminant concentra- ‘ taminants at environmentally relevant concentrations. many tions in adult insect tissues, contaminants with intermediate ’ ‘ ’ adults and few adults illustrate the relative biomass or abun- lipophilicity tend to reduce emergence and have higher tis- dance of emerging adult aquatic insects given the combination sue concentrations, and lipophilic compounds do not affect of toxicity and retention during metamorphosis in each quadrant of the graph. Similarly, [low] and [high] represent low concentra- emergence but tend to have higher concentrations in adult tions and high concentrations, respectively, of contaminants in aquatic insect tissues. emerging adult aquatic insects in each quadrant of the graph.

Among contaminant classes, metals and current-use IMPLICATIONS FOR TERRESTRIAL INSECTIVORES pesticides are generally more toxic to aquatic insects, less Placing contaminants within the heuristic model pro- bioaccumulative, and less retained across metamorphosis posed here allows us to generate predictions about where than PCBs and lipophilic organics (Table 1, Fig. 2; Kenaga and how they may affect terrestrial insectivores (e.g., spi- 1982, Hare 1992, Kraus et al. 2014b). Trace metals, PAHs, ders, birds, bats, and lizards) via their effects on aquatic in- and some current-use pesticides increase mortality and re- sect emergence, contaminant concentrations within aquatic duce adult emergence of aquatic insects but are not highly insect tissues, and contaminant ﬂuxes conveyed by adult bioaccumulative nor in most cases well retained during aquatic insects. Contaminants that are lost during insect insect metamorphosis (Table 1, Fig. 2, lower right panel). metamorphosis but lead to high mortality of aquatic in- Organometals and metalloids (e.g., methylmercury and se- sects at concentrations commonly found in the environ- lenomethionine) and lipophilic organic contaminants like ment (either at the adult stage or earlier; Wesner et al. PCBs are not highly toxic to aquatic insects at concentra- 2019) have the potential to cause a loss of food or prey re- tions commonly found in the environment but accumulate source problem for insectivores of adult aquatic insects in relatively high concentrations in larval insects and are (Fig. 2, bottom right panel). In these cases, loss of prey re- well transferred across metamorphosis (Table 1, Fig. 2, up- sources is more likely to drive effects of aquatic contami- per left panel). Finally, bioaccumulative legacy pesticides like nants on insectivores than eating contaminated prey (e.g., dichlorodiphenyltrichloroethane (DDT) and some contami- Kraus et al. 2014a). For example, secondary production nant mixtures may reduce emergence of adult aquatic in- and emergence severely decrease in trace metal-polluted sects while increasing contaminant concentrations in surviv- streams (Clements et al. 2000, Kraus et al. 2014a), whereas ing adults (Table 1, Fig. 2, upper right panel). trace metals are heavily lost during metamorphosis (e.g., Within contaminant classes, metal retention across 39–94% of body burdens; Kraus et al. 2014b). As a result, metamorphosis is positively related to both toxicity and riparian spider biomass has been reported to be related to metal softness (i.e., greater tendency to form covalent bonds; changes in aquatic emergence production but not tissue Fig. 3A; McCloskey et al. 1996, Kinraide 2009, Li et al. 2012, metal concentrations or ﬂux (Paetzold et al. 2011, Kraus 922 | Contaminant impacts on aquatic–terrestrial linkages J. M. Kraus

Figure 3. Heuristic Model Part II: Predicting effects of contaminants on adult aquatic insect emergence and contaminant concentration within contaminant classes. Empirical relationships between retention across insect metamorphosis (dependent variable) and different measures of toxicity (independent variable) for compounds/elements within 3 contaminant classes: metals (A), PAHs (B), and PCBs (C). Retention is measured as the natural log of the ratio of the concentrations observed in adult and larval aquatic insects (extracted from Kraus et al. 2014b). For metals, the computed toxicity scale is the predicted log10 of the inverse concentration of different metal ions required to reduce a variety of metrics (survival, growth, etc.) for a variety of organisms by 50% (extracted from Kinraide 2009). For PCBs, toxic equivalency factors were extracted from Van den Berg et al. (2006) for mammals, which is similar to fish and birds (Van den Berg et al. 1998). For PAHs, log10(1/EC50) is the log10 of the inverse of the baseline (non-polar) predicted concentration at which the growth rate or yield was reduced by 50% for multiple species of green algae (extracted from Fu et al. 2015). Statistically significant least squares regressions were (A) y 5 1.5x 2 2.1, R2 5 0.41, (B) y 521.4x 1 6.7, R2 5 0.89 when log Kow < 5; (black circles and dashed line) and y 522.8x2 1 39.8x – 139.5, R2 5 0.86 when log Kow 5 5 to 7 (gray circles and dashed line). et al. 2014a). On the other hand, contaminants that are consumers such as song birds and bats might be less likely highly retained but not very toxic at concentrations com- to be attracted to the ecological trap of mercury-laden prey, monly found in the environment have the potential to ex- thus severing the linkage between aquatic and terrestrial pose insectivores to detrimental concentrations of dietary exposure. Many freshwater ecosystems are simultaneously contaminants (i.e., an exposure problem; Fig. 2, top left of exposed to multiple contaminants, and a strength of this panel). For example, PCBs exhibit low aquatic toxicity but framework is its ability to generate hypotheses regarding high persistence in tissues. Insects lose at maximum only what effects different contaminant mixtures will have on 17% of their PCB body burdens across metamorphosis. PCB aquatic–terrestrial linkages. concentrations are 1.3 higher in adult insects compared A final strength of this heuristic model is that it can help with larvae (i.e., bioamplification; Kraus et al. 2014b) be- predict how contaminant properties alter the relevance cause insects also lose 20 to 80% of their body mass during of contaminant concentration vs contaminant flux con- metamorphosis. As a result, patterns of PCB bioaccumula- veyed by adult aquatic insects in driving diet-mediated ef- tion in riparian spiders and modeled exposure of aerial in- fects of aquatic contaminants on terrestrial insectivores. sectivorous birds are similar to aquatic insect and sediment Changes in emergence production and aquatic contami- concentrations (e.g., Walters et al. 2010, Kraus et al. 2017, nant concentrations in adult aquatic insects have well- Walters et al. 2018). studied implications for insectivores. However, the product Contaminants and contaminant mixtures that are both of these 2 factors, contaminant flux, which is a direct mea- highly toxic and retentive in adult aquatic insects at con- sure of how much contaminant is moving from water to centrations commonly found in the environment could land in prey (Table 1, Fig. 1; Menzie 1980, Chumchal and lead to a variety of outcomes depending on the relative Drenner 2015), varies in its role as a driver of consumer ex- magnitude of these processes. For example, at concentra- posure and risk depending on contaminant class. For ex- tions found in most ecosystems, the effects of methyl mer- ample, mercury flux in small-bodied adult aquatic insects cury bio accumulation are more important than its lethal emerging from ponds is positively correlated with concen- toxicity to adult aquatic insects, leading to high exposure trations of mercury in riparian spiders (Tweedy et al. 2013). andbioaccumulationintheirterrestrialconsumers(Tweedy However, this straightforward relationship between con- et al. 2013, Chumchal and Drenner 2015, Gann et al. 2015, taminant flux and consumer risk appears to occur only M. M. Chumchal, Texas Christian University, personal when contaminants both biomagnify through the food web communication). On the other hand, dietary exposure of and are well retained across metamorphosis at concentra- terrestrial insectivores to mercury could be very low near tions that do not significantly reduce adult aquatic insect aquatic ecosystems affected by neonicotinoid insecticides emergence. For contaminants that do not biomagnify or because there would be few adult aquatic insects left to eat. are lost across metamorphosis, or are at concentrations If there are fewer adult insect prey available, mobile aerial that strongly reduce adult emergence, contaminant flux Table 1. Data to inform Part I of the Heuristic Model: Relative toxicity, bioaccumulation, metamorphic retention, and contaminant flux of various contaminant classes to adult aquatic insects. Contaminant flux is predicted based on relative toxicity and mean retention of that class of contaminants across insect metamorphosis. Contaminant y Relative toxicity Relative bioaccumulation Metamorphic retention flux Contaminant Compound/element classes examples Level Cite Level Cite Mean (SE) Cite Predicted

Trace metals Ba, Cd, Cr, Cu, Hg, High Kenaga 1982, Mc- Low/Med Hare 1992 21.81 (0.37) Kraus et al. Low Mn, Mo, Ni, Pb, Sr, Closkey et al. 1996, 2014b Ti, Zn Kinraide 2009 Organometals MeHg, Se-met Low Chumchal and High Chumchal and 20.28 (0.24) Kraus et al. High Drenner 2015 Drenner 2015 2014b PCBs Various congeners Low Kenaga 1982, Van den High Walters et al. 2016 0.13 (0.10) Kraus et al. High Burg 2006 2014b PAHs Fluorene, pyrene, High Fu et al. 2015 Low Fu et al. 2015, Walters 20.88 (0.19) Kraus et al. Low hexachlorobenzene, et al. 2016 2014b indeno(1, 2, 3, c, d)- pyrene z Current-use Neonicotinoids, High Morrissey et al. 2015 Low Hladik et al. 2015 0.28 (0.56) Table S1 Low/Med pesticides bifenthrin, atrazine Legacy DDT degradates, High Kenaga 1982 Med/High Grier 1982, Walters 20.47 (0.35) Kraus et al. Med pesticides hexachlorobenzene et al. 2016 2014b

y Natural log of adult divided by larval concentrations z Data provided in the supplemental materials, Table S1. Represents transfer from larvae with low tissue concentrations. 924 | Contaminant impacts on aquatic–terrestrial linkages J. M. Kraus can be uncorrelated with consumer risk. For example, in to aquatic contaminants (Baxter et al. 2005, Sullivan and a study of insect emergence from metal-contaminated Rodewald 2012, Sullivan and Manning 2019). For example, streams in the central Rocky Mountains, flux of the trace assemblages in metal-contaminated streams have relatively metals Zn, Cu, and Cd via adult aquatic insects was highest few metal-sensitive mayflies but more true flies (dipterans) from low-metal streams (Kraus et al. 2014a). However, ri- and metal-bioaccumulating stoneflies (Clements et al. 2000, parian insectivores of adult aquatic insects were unaffected Kraus et al. 2014a). Riparian spiders like long-jawed orb by metal flux from low metal streams. Instead, consumers weavers (Tetragnathidae) can eat mayflies and dipterans, were more affected by loss of prey near high-metal streams. but stoneflies tend to be too large-bodied to capture. Thus, Spider densities across a gradient of stream metal concen- even if stoneflies contribute to the emergence production trations were unrelated to contaminant concentration or and metal flux from a stream, they are unlikely to affect flux but were positively correlated with dipteran emergence riparian tetragnathid biomass. Consumer mobility can also (Kraus et al. 2014a). strongly affect insectivore responses to changes in adult aquatic insect emergence. For example, diminished re- MODEL ASSUMPTIONS AND DATA GAPS sources, regardless of prey contaminant concentrations, The purpose of a heuristic model is to provide a baseline are likely to have larger effects on specialist consumer set of predictions as a practical, albeit simplistic, starting populations with small foraging ranges than contaminant point for solving a problem. The current model assumes exposure (e.g., Power et al. 2004). However, for generalist, that the physicochemical properties of contaminants and higher-order and highly-mobile consumers, like birds and their known ecotoxicological effects on aquatic insects at bats, loss of aquatic insect resources from 1 local source concentrations commonly found in the environment are would not be expected to affect populations as much as ex- the main drivers of 1) adult aquatic insect production, con- posure from eating contaminated prey (e.g., Custer et al. fl taminant concentration, and flux from contaminated fresh- 1998). More research is needed to esh out the role of these water ecosystems and 2) effects on insectivorous consum- factors in driving the effects of aquatic contaminants on ers of these insects. Deviations from model predictions adult aquatic insects and insectivorous consumers. point to situations where these simplifying assumptions do not hold and other factors may be equally or more im- RESEARCH AND MANAGEMENT APPLICATIONS portant in driving the effects of aquatic contaminants on Regionally, fate and transport models, as well as empir- adult aquatic insects and their insectivores. For example, ical trend and assessment measurements, can provide in- in circumstances where the contaminants themselves are formation at multiple scales about contaminant distribu- not the major driver of insect production, other factors be- tion across the landscape (MacLeod et al. 2001, Kolpin et al. sides exposure drive contaminant flux from aquatic to ter- 2002). Building on these frameworks, the model presented restrial food webs. Tweedy et al. (2013) observed that in here can be used to predict whether aquatic-dependent in- mercury contaminated ponds, nutrient enrichment (a sectivores in a particular region are likely to suffer from prey bottom-up ecological process) and presence of fish (a top- resource depletion or contaminant exposure. These pre- down ecological process), and not mercury concentrations, dictions can be used to shape experimental design and sup- largely drive aquatic insect emergence and, thus, patterns port efficient allocation of resources for research designed of mercury flux and uptake by riparian spiders. In other cir- to inform management decisions. For example, the Min- cumstances, such as comparisons across geologically and eral Belt of the United States Rocky Mountains is a highly- topographically complexlandscapes,variationinwaterchem- mineralized and historically-mined region. We predicted istry and species characteristics (e.g., contaminant concen- based on the information used in this model that the metals tration, organic matter, pH, water hardness, timing of expo- mobilized into the streams of this region would accumulate sure, species identity and genetic adaptation; Clements et al. in larvae but not be retained across metamorphosis to ter- 2016), rather than contaminant concentrations alone, drive restrial adults (Hare 1992, Kraus et al. 2014b, Wesner et al. toxicity and bioaccumulation of contaminants. In these cases, 2019). However, aqueous concentrations were great enough measurements of local conditions are key to site-specific in some streams to cause high mortality in adult aquatic in- predictions. sects (Kraus et al. 2014b), potentially making reduced food Landscape, food-web, consumer, and prey characteris- availability a major problem for consumers (Custer et al. tics can also be affected by contaminants and will influence 2003, Paetzold et al. 2011, Kraus et al. 2014a). Because of how contaminant impacts on aquatic insects play out for the predictions driven by the model presented here, we fea- their consumers. Factors such as habitat loss and alter- tured ecological measures of emergence production heavily ation, aquatic and terrestrial diversity, trophic structure, in the experimental design of these studies, which allowed prey preferences and ability to capture insects, consumer accurate assessment of the ecotoxicological effects of trace mobility, prey size and behavior, and timing of emergence metals on emergence and riparian spiders in these regions are all common potential modifiers of insectivore responses (Paetzold et al. 2011, Kraus et al. 2014a). In contrast, Volume 38 December 2019 | 925 methylmercury uptake by riparian spiders are driven by LITERATURE CITED both concentrations of mercury in adult aquatic insect prey Allen, D. C., and J. S. Wesner. 2016. Synthesis: Comparing effects and their emergence production (i.e., mercury flux; Tweedy of resource and consumer fluxes into recipient food webs us- et al. 2013, Chumchal and Drenner 2015), which required ing meta-analysis. Ecology 97:594–604. measurement of both emergence and tissue concentrations Ballingall, A. 2018. Ottawa to phase out pesticides linked to bee for these studies. deaths. The Star, Toronto, Canada. (Available from: https:// Understanding the effects of aquatic pollution on www.thestar.com/news/canada/2018/08/14/ottawa-to-phase aquatic–terrestrial linkages is of interest to regulatory agen- -out-pesticides-linked-to-bee-deaths.html) Baxter, C. V., K. D. Fausch, and W. C. Saunders. 2005. Tangled cies and local stakeholders. For example, adult aquatic in- fl fl webs: Reciprocal ows of invertebrate prey link streams and sect mediated-PCB ux to terrestrial insectivores near the riparian zones. Freshwater Biology 50:201–220. Great Lakes and other industrially polluted water bodies is Bernhardt, E. S., E. J. Rosi, and M. O. Gessner. 2017. Synthetic being used as a measure of remedy effectiveness (e.g., sed- chemicalsasagentsofglobalchange.FrontiersinEcologyand iment dredging; Custer et al. 1998, Walters et al. 2010, the Environment 15:84–90. Kraus et al. 2017, Walters et al. 2018, Muehlbauer et al. Carson, R. 1962. Silent spring. Houghton Mifflin, Boston, 2019). Effects of current-use pesticides on the emergence Massachusetts. and contamination of adult aquatic insects in wetlands crit- Cavallaro, M. C., C. A. Morrissey, J. V. Headley, K. M. Peru, and ical to waterfowl production are being used in decision- L. Liber. 2017. Comparative chronic toxicity of imidacloprid, making by the Canadian government and are of interest clothianidin, and thiamethoxam to Chironomus dilutus and to the United States Fish and Wildlife Service (Table S1; estimation of toxic equivalency factors. Environmental Toxi- cology and Chemistry 36:372–382. Ballingall 2018, Cavallaro et al. 2017). Thus, by predicting Chumchal, M. M., and R. W. Drenner. 2015. An environmental and identifying problems due to the disruption and con- problem hidden in plain sight? Small human-made ponds, emer- – tamination of aquatic terrestrial linkages, conceptual and gent insects, and mercury contamination of biota in the Great quantitative models, such as the one presented here, can Plains. Environmental Toxicology and Chemistry 34:1197–1205. help prioritize research directions and decisions regarding Clements, W. H., D. M. Carlisle, J. M. Lazorchak, and P. C. John- natural resource management. son. 2000. Heavy metals structure benthic communities in Col- The genesis of the field of ecotoxicology (Truhaut 1977) orado mountain streams. Ecological Applications 10:626– and the publication of Silent Spring (Carson 1962) alerted 638. the public to how exposure to contaminants can cascade Clements, W. H., D. R. Kashian, P. M. Kiffney, and R. E. Zuellig. through food webs and ecosystems, such as occurred be- 2016. Perspectives on the context-dependency of stream com- tween DDT and a massive reproductive crash in bald ea- munity responses to contaminants. Freshwater Biology 61: 2162–2170. gles (Grier 1982). Also highlighted in this early work, al- Custer, C. M., T. W. Custer, P. D. Allen, K. L. Stromborg, and though not as appreciated, are the cascading problems of M. J. Melancon. 1998. Reproduction and environmental con- aquatic contamination on ecological disruptions, extinc- tamination in tree swallows nesting in the Fox River drainage tions, and population/community declines across ecosys- and Green Bay, Wisconsin, USA. Environmental Toxicology tem boundaries caused by prey resource losses (Carson and Chemistry 17:1786–1798. 1962, Rohr et al. 2006, Bernhardt et al. 2017, Ballingall Custer, C. M., T. W. Custer, P. M. Dummer, and K. L. Munney. 2018, Schiesari et al. 2018). Predicting under what condi- 2003. Exposure and effects of chemical contaminants on tree tions each (or both) of these effects occur informs the range swallows nesting along the Housatonic River, Berkshire County, – of possible outcomes for insectivorous consumers and the Massachusetts, USA, 1998 2000. Environmental Toxicology – approaches needed to measure those outcomes in food and Chemistry 22:1605 1621. webs (e.g., measure concentrations for exposure problems Fu, L., J. J. Li, Y. Wang, X. H. Wang, Y. Wen, W. C. Qin, L. M. Su, and Y. H. Zhao. 2015. Evaluation of toxicity data to green and ecological metrics for food loss). Ultimately, this heu- algae and relationship with hydrophobicity. Chemosphere ristic model for predicting how aquatic contaminants af- 120:16–22. fect adult aquatic insects will help streamline decision Gann, G. L., C. H. Powell, M. M. Chumchal, and R. W. Drenner. making about what are often cryptic and non-intuitive ef- 2015. Hg-contaminated terrestrial spiders pose a potential fects of contaminant stressors at the land-water interface. risk to songbirds at Caddo Lake (Texas/Louisiana, USA). En- vironmental Toxicology and Chemistry 34:303–306. ACKNOWLEDGEMENTS Grier, J. W. 1982. Ban of DDT and subsequent recovery of repro- This work was funded by the United States Geological Survey duction in bald eagles. Science 218:1232–1235. (USGS) Contaminant Biology Program and the USGS Menden- Hallmann, C. A., R. P. Foppen, C. A. van Turnhout, H. de Kroon, hall Program. Jeremy Monroe of Freshwaters Illustrated pro- and E. Jongejans. 2014. Declines in insectivorous birds are as- duced the insect images used in Fig. 1. This research was subjected sociated with high neonicotinoid concentrations. Nature 511: to USGS review and approved for publication. Any use of trade, 341. firm, or product names is for descriptive purposes only and does Halstead, N. T., T. A. McMahon, S. A. Johnson, T. R. Raffel, J. M. not imply endorsement by the United States Government. Romansic, P. W. Crumrine, and J. R. Rohr. 2014. Community 926 | Contaminant impacts on aquatic–terrestrial linkages J. M. Kraus

ecology theory predicts the effects of agrochemical mixtures contamination of global surface waters and associated risk on aquatic biodiversity and ecosystem properties. Ecology Let- to aquatic invertebrates: A review. Environment International ters 17:932–941. 74:291–303. Hare, L. 1992. Aquatic insects and trace metals: Bioavailability, Moy, N. J., J. Dodson, S. J. Tassone, P. A. Bukaveckas, and L. P. bioaccumulation, and toxicity. Critical Reviews in Toxicology Bulluck. 2016. Biotransport of algal toxins to riparian food 22:327–369. webs. Environmental Science and Technology 50:10007–10014. Hladik, M. L., M. Vandever, and K. L. Smalling. 2015. Exposure of Muehlbauer, J. D., C. A. Lupoli, and J. M. Kraus. 2019. Aquatic- native bees foraging in an agricultural landscape to current- terrestrial linkages provide novel opportunities for freshwater use pesticides. Science of the Total Environment 542:469–477. ecologists to engage stakeholders and inform riparian man- Jackson, J. K., and S. G. Fisher. 1986. Secondary production, agement. Freshwater Science 38:946–952. emergence and export of aquatic insects from a Sonoran Des- Nakano, S., and M. Murakami. 2001. Reciprocal subsidies: Dy- ert stream. Ecology 67:629–638. namic interdependence between terrestrial and aquatic food Kenaga, E. E. 1982. Predictability of chronic toxicity from acute webs. Proceedings of the National Academy of Sciences 98: toxicity of chemicals in fish and aquatic invertebrates. Envi- 166–170. ronmental Toxicology and Chemistry 1:347–358. Paetzold, A., M. Smith, P. H. Warren, and L. Maltby. 2011. En- Kinraide, T. B. 2009. Improved scales for metal ion softness and vironmental impact propagated by cross-system subsidy: toxicity. Environmental Toxicology and Chemistry 28:525–533. Chronic stream pollution controls riparian spider popula- Kolpin, D. W., E. T. Furlong, M. T. Meyer, E. M. Thurman, S. D. tions. Ecology 92:1711–1716. Zaugg, L. B. Barber, and H. T. Buxton. 2002. Pharmaceuticals, Polis, G. A., W. B. Anderson, and R. D. Holt. 1997. Toward and hormones, and other organic wastewater contaminants in US integration of landscape and food web ecology: The dynamics streams, 1999–2000: A national reconnaissance. Environmen- of spatially subsidized food webs. Annual Review of Ecology tal Science & Technology 36:1202–1211. and Systematics 28:289–316. Kraus, J. M., P. P. Gibson, D. M. Walters, and M. A. Mills. 2017. Power, M. E., W. E. Rainey, M. S. Parker, J. L. Sabo, A. Smyth, S. Riparian spiders as sentinels of polychlorinated biphenyl con- Khandwala, J. C. Finlay, F. C. McNeely, K. Marsee, and C. An- tamination across heterogeneous aquatic ecosystems. Envi- derson. 2004. River-to-watershed subsidies in an old-growth ronmental Toxicology and Chemistry 36:1278–1286. conifer forest. Pages 217–240 in G. A. Polis, M. E. Power, Kraus, J. M., T. S. Schmidt, D. M. Walters, R. B. Wanty, R. E. and G. R. Huxel (editors). Food webs at the landscape level. Zuellig, and R. E. Wolf. 2014a. Cross-ecosystem impacts of The University of Chicago Press, Chicago, Illinois. stream pollution reduce resource and contaminant flux to ri- Richmond, E. K., E. J. Rosi-Marshall, D. M. Walters, J. Fick, T. parian food webs. Ecological Applications 24:235–243. Brodin, A. Sundelin, and M. R. Grace. 2018. A diverse suite Kraus, J. M., D. M. Walters, J. S. Wesner, C. A. Stricker, T. S. of pharmaceuticals contaminates stream and riparian food Schmidt, and R. E. Zuellig. 2014b. Metamorphosis alters con- webs. Nature Communications 9:1–9. taminants and chemical tracers in insects: Implications for Rohr, J. R., J. L. Kerby, and A. Sih. 2006. Community ecology as a food webs. Environmental Science and Technology 48:10957– framework for predicting contaminant effects. Trends in Ecol- 10965. ogy and Evolution 21:606–613. Lavoie, R. A., T. D. Jardine, M. M. Chumchal, K. A. Kidd, and Schiesari, L., M. A. Leibold, and G. A. Burton Jr. 2018. Meta- L. M. Campbell. 2013. Biomagnification of mercury in aquatic communities, metaecosystems and the environmental fate of food webs: A worldwide meta-analysis. Environmental Sci- chemical contaminants. Journal of Applied Ecology 55:1553– ence & Technology 47:13385–13394. 1563. Li, Y., X. L. Li, X. Y. Du, M. Wang, J. Xin, Y. Hu, and Y. Wang. Schultz, T. W., N. L. Wyatt, and D. T. Lin. 1990. Structure-toxicity 2012. Using the QICAR model to correlate metal ion charac- relationships for nonpolar narcotics: A comparison of data from teristics with toxicity order numbers. Human and Ecological the Tetrahymena, Photobacterium and Pimephales systems. Bul- Risk Assessment: An International Journal 18:1255–1270. letin of Environmental Contamination and Toxicology 44:67– MacLeod, M., D. G. Woodfine, D. Mackay, T. McKone, D. Ben- 72. nett, and R. Maddalena. 2001. BETR North America: A region- Sullivan, S. M. P., and D. W. P. Manning. 2019. Aquatic–terrestrial ally segmented multimedia contaminant fate model for North linkages as complex systems: Insights and advances using net- America. Environmental Science and Pollution Research 8:156. work models. Freshwater Science. 38:936–945. Marcarelli, A. M., C. V. Baxter, M. M. Mineau, and R. O. Hall. Sullivan, S. M. P., and A. D. Rodewald. 2012. The energetic path- 2011. Quantity and quality: Unifying food web and ecosystem ways that move contaminants from aquatic to terrestrial envi- perspectives on the role of resource subsidies in freshwaters. ronments. Environmental Toxicology and Chemistry 31:1175– Ecology 92:1215–1225. 1183. McCloskey, J. T., M. C. Newman, and S. B. Clark. 1996. Predict- Truhaut, R. 1977. Ecotoxicology: Objectives, principles and per- ing the relative toxicity of metal ions using ion characteristics: spectives. Ecotoxicology and Environmental Safety 1:151–173. Microtox bioluminescence assay. Environmental Toxicology Tweedy, B. N., R. W. Drenner, M. M. Chumchal, and J. H. Ken- and Chemistry:® An International Journal 15:1730–1737. nedy. 2013. Effects of fish on emergent insect-mediated flux of Menzie, C. A. 1980. Potential significance of insects in the removal methyl mercury across a gradient of contamination. Environ- of contaminants from aquatic systems. Water, Air, and Soil mental Science & Technology 47:1614–1619. Pollution 13:473–479. Van den Berg, M., L. Birnbaum, A. T. Bosveld, B. Brunström, Morrissey, C. A., P. Mineau, J. H. Devries, F. Sanchez-Bayo, M. P. Cook, M. Feeley, J. P. Giesy, A. Hanberg, R. Hasegawa, S. W. Liess, M. C. Cavallaro, and K. Liber. 2015. Neonicotinoid Kennedy, and T. Kubiak. 1998. Toxic equivalency factors Volume 38 December 2019 | 927

(TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Walters, D. M., M. A. Mills, K. M. Fritz, and D. F. Raikow. 2010. Environmental Health Perspectives 106:775. Spider-mediated ﬂux of PCBs from contaminated sediments Van den Berg, M., L. S. Birnbaum, M. Denison, M. De Vito, to terrestrial ecosystems and potential risks to arachnivorous W. Farland, M. Feeley, H. Fiedler, H. Hakansson, A. Hanberg, birds. Environmental Science & Technology 44:2849–2856. L. Haws, and M. Rose. 2006. The 2005 World Health Organi- Walters, D. M., R. R. Otter, J. M. Kraus, and M. A. Mills. 2018. zation reevaluation of human and mammalian toxic equiva- Riparian spiders indicate the magnitude and sources of poly- lency factors for dioxins and dioxin-like compounds. Toxico- chlorinated biphenyl (PCB) contamination at a large contam- logical Sciences 93:223–241. inated sediment site. Environmental Toxicology and Chemis- Walters, D. M., K. M. Fritz, and R. R. Otter. 2008. The dark side try 37:2467–2474. of subsidies: Adult stream insects export organic contami- Wesner, J. 2019. Using stage structured food webs to assess the nants to riparian predators. Ecological Applications 18:1835– effects of contaminants and predators on aquatic–terrestrial 1841. linkages. Freshwater Science. 38:928–935. Walters, D. M., T. D. Jardine, B. S. Cade, K. A. Kidd, D. C. G. Wu, F., Y. Mu, H. Chang, X. Zhao, J. P. Giesy, and K. B. Wu. 2012. Muir, and P. Leipzig-Scott. 2016. Trophic magniﬁcation of or- Predicting water quality criteria for protecting aquatic life ganic chemicals: A global synthesis. Environmental Science & from physicochemical properties of metals or metalloids. Technology 50:465024658. Environmental Science & Technology 47:446–453.