Oecologia DOI 10.1007/s00442-015-3279-5
COMMUNITY ECOLOGY - ORIGINAL RESEARCH
Autumn leaf subsidies influence spring dynamics of freshwater plankton communities Samuel B. Fey · Andrew N. Mertens · Kathryn L. Cottingham
Received: 13 August 2014 / Accepted: 18 February 2015 © Springer-Verlag Berlin Heidelberg 2015
Abstract While ecologists primarily focus on the immediate impact of ecological subsidies, understanding the importance of ecological subsidies requires quantifying the long-term temporal dynamics of subsidies on recipient ecosystems. Deciduous leaf litter transferred from terrestrial to aquatic ecosystems exerts both immediate and lasting effects on stream food webs. Recently, deciduous leaf additions have also been shown to be important subsidies for planktonic food webs in ponds during autumn; however, the inter-seasonal effects of autumn leaf subsidies on planktonic food webs have not been studied. We hypothesized that autumn leaf drop will affect the spring dynamics of freshwater pond food webs by altering the availability of resources, water transparency, and the metabolic state of ponds. We created leaf-added and no-leaf-added field mesocosms in autumn 2012, allowed mesocosms to ice-over for the winter, and began sampling the physical, chemical, and biological properties of mesocosms immediately following ice-off in spring 2013. At ice-off, leaf additions reduced dissolved oxygen, elevated total phosphorus concentrations and dissolved materials, and did not alter temperature or total nitrogen. These initial abiotic effects
contributed to higher bacterial densities and lower chlorophyll concentrations, but by the end of spring, the abiotic environment, chlorophyll and bacterial densities converged. By contrast, zooplankton densities diverged between treatments during the spring, with leaf additions stimulating copepods but inhibiting cladocerans. We hypothesized that these differences between zooplankton orders resulted from resource shifts following leaf additions. These results suggest that leaf subsidies can alter both the short- and longterm dynamics of planktonic food webs, and highlight the importance of fully understanding how ecological subsidies are integrated into recipient food webs. Keywords Terrestrial-aquatic linkages · Food webs · Phenology · Ponds · Zooplankton
“The rain falling on the freshly dried herbs and leaves, and filling the pools and ditches into which they have dropped thus clean and rigid, will soon convert them into tea,— green, black, brown, and yellow teas, of all degrees of strength, enough to set all Nature a-gossiping.” H. D. Thoreau, "Autumnal Tints,” 1862.
Communicated by Robert O. Hall. S. B. Fey and A. N. Mertens have contributed equally to this manuscript. S. B. Fey · A. N. Mertens · K. L. Cottingham Department of Biological Sciences, Dartmouth College, Hanover, NH, USA Present Address: S. B. Fey (*) Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT 06520, USA e-mail: [email protected]
Introduction Ecological subsidies, the energy and materials that move across ecosystem boundaries, influence the structure and composition of food webs (Polis et al. 1997; Takimoto et al. 2002; Gratton et al. 2008). The effect of such subsidies on the recipient community depends on many factors, including the characteristics (Cole et al. 2006) and quantity of materials transferred (Marcarelli et al. 2011), the rate
13
Oecologia
of subsidization (Takimoto et al. 2009), and the behavioral and life history characteristics of the recipient community (Baxter et al. 2005). Tremendous variation exists regarding the time period over which a subsidy can affect the recipient food web. For example, adult insects emerging from temperate aquatic ecosystems into terrestrial ecosystems generally integrate rapidly into the terrestrial food web (Nakano and Murakami 2001; Sabo and Power 2002). By contrast, coarse woody debris that enters aquatic ecosystems from surrounding forests may decay over decades or even centuries (Guyette and Cole 1999). Similarly, the integration of large whale carcasses (which can fall from pelagic habitats following death) into benthic ecosystems first by mobile scavengers, then polychaetes and crustaceans, and finally microbes, can take place over decades (Smith and Baco 2003; Higgs et al. 2014). As such, accurately quantifying the effect of a cross-ecosystem subsidy may require quantifying the recipient ecosystem’s response both immediately and over extended time periods. Leaves from deciduous trees represent an important subsidy for aquatic ecosystems that alters food webs on both short and long time scales (Anderson and Sedell 1979; Earl et al. 2014). Allochthonous inputs to streams from surrounding forests are well appreciated and typically exceed autochthonous inputs in certain ecosystems (Webster and Meyer 1997). By providing both limiting resources and microhabitat (Richardson 1992), leaves aid in the immediate production of bacteria, fungi, and detritivorous insects in streams (Hieber and Gessner 2002). The use of leaf packs containing terrestrial leaf litter also provides strong evidence that leaf subsidies are integrated into recipient ecosystems over long time periods, as shown by how the remaining leaf material continues to decrease in mass following multiple seasons of being utilized by stream biota (Benfield et al. 2001; Swan and Palmer 2004). Moreover, excluding terrestrial leaf litter inputs significantly reduces benthic invertebrate biomass, including shredders, gatherers, and even predators on interseasonal timescales (Wallace et al. 1999). In temperate pond ecosystems, autumn leaf drop can contribute substantially to annual carbon and nutrient budgets (France and Peters 1995; Hongve 1999; Pope et al. 1999), particularly in ecosystems unable to produce large amounts of autochthonous carbon (Rubbo et al. 2006). Furthermore, while there is a growing appreciation for the importance of leaves as a subsidy that immediately impacts producer and consumer components of pond food webs (Oertli 1993; Rubbo et al. 2008; Rubbo and Kiesecker 2004; Cottingham and Narayan 2013; Earl et al. 2014), little is known about the long-term effects of leaf subsidies for such food webs. Our current understanding of this effect comes primarily from two whole-ecosystem experiments that investigated the effect of leaf litter on small ponds by removing or restricting the fall of leaf litter in autumn, and measuring the subsequent
13
Fig. 1 Leaves and leaf leachates impact pond food webs by resourcemediated effects (solid black arrows), transparency-mediated effects (dashed grey arrows), and oxygen-mediated effects (solid grey arrows)
response in the spring. Here, Rubbo et al. (2006) showed that leaf removals decreased ecosystem respiration and reduced inter-seasonal secondary production. However, high between-pond variability produced contrasting results for a similar study investigating the effect of leaf litter on benthic invertebrates (Batzer and Palik 2007). To our knowledge, no study has addressed the long-term effect of leaves on organisms other than benthic invertebrates, and especially not on both the autotroph- and detritus-based food webs. There is reason to believe that autumn leaf subsidies could exert effects on planktonic food webs that extend beyond the autumn months, particularly in temperate ecosystems where reduced light availability and cold temperatures in winter are likely less conducive for planktonic food webs to assimilate the carbon and nutrients provided by autumn leaf subsidies (Twiss et al. 2012; Bertilsson et al. 2013). We identify three pathways through which leaves may affect pond food webs (Fig. 1). First, resource-mediated effects result from the consumption of leaf material, the uptake of nutrients and carbon from leaf leachates, or the consumption of organisms that uptake these materials. Second, transparency-mediated effects result from changes in water transparency that regulates the penetration of incoming solar radiation. Third, metabolism-mediated effects result from changes in the balance of respiration versus photosynthesis in response to leaf additions. The combined changes to the biotic and abiotic environment would have cascading but differential effects on zooplankton orders. Assuming these effects persist into the spring, we predict that cladocerans will benefit relative to copepods due to decreased ultraviolet radiation exposure accompanying leaf additions (Leech and Williamson 2000), cladocerans will be negatively impacted to a greater extent than copepods due to the decreases in oxygen concentrations (Vanderploeg et al. 2009) that accompany leaf additions (Cottingham and Narayan 2013), and the faster numeric responses of cladocerans would allow them to benefit from any increases in phytoplankton stimulated by leaf additions.
Oecologia
Therefore, we hypothesized that autumn leaf drop affects the spring dynamics of pond food webs by altering the availability of resources, the water transparency, and the metabolic state of ponds. We conducted a 10-month mesocosm experiment using an experimental array previously used to understand how variation in leaf chemistry driven by increased soil temperature can alter the quality of leaf subsidies for planktonic food webs (Fey et al. 2015). Here, we test two specific predictions: autumn leaf subsidies affect the initial state of planktonic food webs immediately following ice-off, and have a lasting effect on planktonic food webs beyond ice-off.
Materials and methods Our field experiment involved several stages: 1. Establishing mesocosms and imposing the control (5 g leaves added to mesocosms) and leaf-addition (50 g leaves added to mesocosms) experimental treatments. 2. Following responses to treatments prior to ice-on (Fey et al. 2015). 3. Allowing mesocosms to freeze over during the winter months. 4. Observing the treatment responses immediately following ice-out and throughout the spring. There were five replicates of each treatment, for a total of ten mesocosms. The primary response variables were phytoplankton chlorophyll-a, total nitrogen and total phosphorus concentrations, bacterial and zooplankton densities, and zooplankton community composition. We also characterized the abiotic environment by measuring dissolved oxygen, temperature, and the absorbance due to dissolved substances at 254 nm—a proxy for water clarity and dissolved organic carbon (Brandstetter et al. 1996). The leaf loading rate used (~150 g leaves/m2 pond surface area) falls within the reported rates of areal loading rates observed in the near-shore areas of lakes (Hongve 1999) and previous leaf-addition mesocosm experiments (Rubbo and Kiesecker 2004; Cottingham and Narayan 2013). These leaf additions correspond to a loading rate of ~23 g leaves per meter “shoreline” (as measured by mesocosm circumference), which falls below the 32–720 g/m reported annual shoreline rates (France and Peters 1995). Establishing mesocosm communities We constructed mesocosms from 167-L polyethylene cylindrical refuse bins (0.61-m diameter by 0.80-m depth, 1.95 surface to volume ratio) buried in the ground at the Dartmouth Organic Farm, Hanover, New Hampshire.
Mesocosms were filled with 120 L of groundwater on 5 September 2012 and covered with 1-mm fiberglass mesh to prevent colonization by non-target organisms. The following day, we added homogenized packets containing 5 g of the collected leaf subsidies to all mesocosms to provide a source of organic carbon and nutrients for developing biological communities. On 6–7 September 2012, we added 5 L of water from each of six small lakes and ponds within a 20-km radius of Hanover (30 L total, final mesocosm volume ~150 L) to seed mesocosms with natural phytoplankton and bacterial communities (Table S1). From 9 to 11 September, we added zooplankton from both the littoral (near shore) and pelagic (open water) zones of five of the six ponds. We collected zooplankton using a 12-L Schindler-Patalas trap; each mesocosm received zooplankton filtered from 144 L of pond water. Prior to imposing leaf-addition treatments, there were no differences between mesocosm treatments in any of our response variables (Fey et al. 2015). We did not add benthic invertebrates to our mesocosms in order to focus on the response of the pelagic food web. Generating experimental treatments We collected leaves from the Harvard Forest in central Massachusetts, USA (42°28′N, 72°10′W), rather than from the shoreline of the ponds used to establish mesocosms, to facilitate a prior experiment (Fey et al. 2015). On 28 October 2010 and 20 October 2011, we collected freshly fallen leaves from red maple trees (Acer rubrum), which are common around ponds across eastern North America (Rubbo and Kiesecker 2004). All leaves were collected after abscission but before a rain event. Leaves were dried at room temperature, and then stored in closed paper bags until the experiment so as not to alter leaf chemistry (Cotrufo and Ineson 1996). Leaf subsidies were added to mesocosms on 2 October 2012. We added 45 g of dried leaf litter (10 g from 2010 and 35 g from 2011) to each mesocosm in the leaf-addition treatment. Field sampling We monitored mesocosms through the fall season and noted that mesocosms were fully iced-over the week of 13 November (Fey et al. 2015). Ice-off occurred in all mesocosms during the last week of April 2013. While complete surface ice coverage of the mesocosms persisted during the winter, we did not measure the thickness of the ice to avoid disrupting under-ice dynamics. As the ice and snow melted, we removed additional water to ensure that the mesocosm volume remained at 150 L and that water did not make contact with the fiberglass screens enclosing mesocosms.
13
Oecologia
Beginning on 1 May 2013, we sampled mesocosms weekly between 0800 and 1200 hours. We recorded water temperature, dissolved oxygen, using a YSI model 556 MPS meter (YSI, Yellow Springs, OH). We then homogenized the mesocosms through stirring, removed a 5-L vertical cross-section of the water column using a vertical water sampler, and filled 3 1-L bottles for nutrient, chlorophyll, and bacterial analysis, respectively. All samples were stored in coolers containing ice packs until returning to the laboratory. We collected an additional 5 L of water to sample for zooplankton. We filtered the water through 53-μm mesh, washed the zooplankton off the mesh into 50-mL centrifuge tubes, preserved the zooplankton with 70 % ethanol, and returned the filtered water to the mesocosm. Mesocosm sampling ended at the end of spring, 19 June, when treatment and control mesocosms had converged in most physical and chemical properties. Laboratory analyses Nutrients and chlorophyll-a were analyzed followed previously described procedures (Fey and Cottingham 2012). Total nutrient samples were frozen until digestion and spectrophotometric analysis on a Perkin-Elmer Lambda EZ201 spectrophotometer (PerkinElmer, Waltham, MA). For total nitrogen, we used a basic persulfate digestion and measured nitrate using the second-derivative method (Crumpton et al. 1992). For total phosphorus, we used a persulfate digestion and the molybdate colorimetric reaction. Samples for dissolved materials were pre-filtered using PURADISC 25 GF/F disposable filters (Whatman, Maidstone, Kent, UK), and were measured spectrophotometrically by absorbance in a 1-cm quartz cuvette at 254 nm (Brandstetter et al. 1996). Finally, to measure chlorophyll-a concentration, we vacuum-filtered 100 mL of water onto 47-mm Whatman GF/C filters; froze the filters, extracted the chlorophyll in methanol, and measured concentrations using the Welschmeyer method (Welschmeyer 1994). Zooplankton were enumerated at 10× magnification on a Leica MZ-12 dissecting microscope (Leica Microsystems, Bannockburn, IL). Cladocerans were identified to genus; adult copepods were identified to order (Calanoida versus Cyclopoida), while copepodites were enumerated as a bulk category. We did not detect any rotifers in mesocosms, similar to previous mesocosm experiments (Fey and Cottingham 2012; Cottingham and Narayan 2013). Bacteria were enumerated using a modified 4′,6-diamidino-2-phenylindole (DAPI) staining method (Porter and Feig 1980). Upon returning from the field, whole water samples were fixed immediately with glutaraldehyde (3 % final concentration; Acros Organics, Geel, Belgium). Prior to enumeration, samples were filtered at low pressure ( 0.6). The effect of leaf additions on total phosphorus persisted over winter, such that total phosphorus in the leaf-addition treatment concentrations remained five times higher immediately following ice-off (Fig. 3b; ice-out t4.06 = 5.57, P = 0.005). Leaf additions initially altered chlorophyll-a concentrations and bacterial densities following ice-off. Leaf-addition treatments initially had reduced chlorophyll-a concentrations by 50 % (Fig. 4a; t5.55 = −3.73, P = 0.011). By contrast, bacterial densities were two-fold higher in
Fig. 3a, b Total phosphorus and nitrogen concentrations through the spring for leaf-addition and control mesocosms. Black squares are data from leaf-addition mesocosms and white circles are data from control mesocosms. Left panels are data from 13 November 2012, the last sample date before ice-over. Data are treatment mean ± SE (n = 5 replicates)
mesocosms receiving leaves immediately following ice-off (Fig. 4b; t7.57 = 3.20, P = 0.014). Leaf additions initially decreased cladoceran densities to near-zero densities relative to control treatments, although cladoceran densities were initially low in both treatments (Fig. 5a, c, e, g). Consequently, total cladoceran, Chydorus, and Daphnia densities were higher in control treatments immediately following ice-off (t3.15 = −4.97, P = 0.014; t3.83 = −4.55, P = 0.012; and t6.97 = −1.09, P > 0.3, respectively), while Bosmina densities did not differ between treatments (t6.96 = −10.05, P 0.3). Autumn leaf subsidies impact throughout the spring By late spring, most abiotic variables had converged or were converging between treatments. The absorbance of dissolved substances measured at 254 nm (1/m) decreased throughout the spring in the leaf-addition treatment,
13
Oecologia
Fig. 4a, b Basal resource response to leaf addition, as measured by chlorophyll-a concentration and bacteria densities. Black squares are data from leaf-addition mesocosms and white circles are data from control mesocosms. Left panels are data from 13 November 2012, the last sample date before ice-over (Fey et al. 2015). Data are treatment mean ± SE (n = 5 replicates)
while the control treatment did not change (Fig. 2a; RMANOVA time × treatment F7,59 = 10.89, P 0.6; end of spring, t6.61 = −1.24, P > 0.2). Despite the initial differences in phosphorus to leaf additions at ice-off, the chemical environments between treatments converged by the end of spring. Nitrogen concentrations increased similarly in both treatments (Fig. 3a; RM-ANOVA time F7,64 = 9.55, P 0.1). Phosphorus concentrations decreased in leafaddition treatments, while increasing in control treatments (Fig. 2b; RM-ANOVA time × treatment F7,64 = 11.32, P 0.7).
13
While chlorophyll-a concentrations differed immediately following ice-off, concentrations increased faster in leaf-addition treatments (Fig. 4a; RM-ANOVA time × treatment F7,64 = 4.85, P 0.2). Bacterial densities, however, followed the opposite trajectory. Bacteria densities were generally higher in leaf-addition mesocosms (Fig. 4b; RM-ANOVA treatment F1,64 = 10.283, P = 0.003) and increased more rapidly in the control treatment throughout the spring (Fig. 4b; RM-ANOVA time × treatment F4,64 = 5.26, P = 0.002), reaching densities 50 % higher in control than leaf-addition mesocosms by the end of spring (Fig. 4b; t6.98 = −2.44, P = 0.045). In contrast to the physical and chemical environment, zooplankton populations generally continued to diverge between treatments as the spring progressed. Leaf additions continued to suppress cladoceran densities compared to control treatments. Total cladoceran and Chydorus population densities were higher in control mesocosms throughout the spring (Fig. 5a, e; RM-ANOVA treatment F1,64 = 26.81, P
COMMUNITY ECOLOGY - ORIGINAL RESEARCH
Autumn leaf subsidies influence spring dynamics of freshwater plankton communities Samuel B. Fey · Andrew N. Mertens · Kathryn L. Cottingham
Received: 13 August 2014 / Accepted: 18 February 2015 © Springer-Verlag Berlin Heidelberg 2015
Abstract While ecologists primarily focus on the immediate impact of ecological subsidies, understanding the importance of ecological subsidies requires quantifying the long-term temporal dynamics of subsidies on recipient ecosystems. Deciduous leaf litter transferred from terrestrial to aquatic ecosystems exerts both immediate and lasting effects on stream food webs. Recently, deciduous leaf additions have also been shown to be important subsidies for planktonic food webs in ponds during autumn; however, the inter-seasonal effects of autumn leaf subsidies on planktonic food webs have not been studied. We hypothesized that autumn leaf drop will affect the spring dynamics of freshwater pond food webs by altering the availability of resources, water transparency, and the metabolic state of ponds. We created leaf-added and no-leaf-added field mesocosms in autumn 2012, allowed mesocosms to ice-over for the winter, and began sampling the physical, chemical, and biological properties of mesocosms immediately following ice-off in spring 2013. At ice-off, leaf additions reduced dissolved oxygen, elevated total phosphorus concentrations and dissolved materials, and did not alter temperature or total nitrogen. These initial abiotic effects
contributed to higher bacterial densities and lower chlorophyll concentrations, but by the end of spring, the abiotic environment, chlorophyll and bacterial densities converged. By contrast, zooplankton densities diverged between treatments during the spring, with leaf additions stimulating copepods but inhibiting cladocerans. We hypothesized that these differences between zooplankton orders resulted from resource shifts following leaf additions. These results suggest that leaf subsidies can alter both the short- and longterm dynamics of planktonic food webs, and highlight the importance of fully understanding how ecological subsidies are integrated into recipient food webs. Keywords Terrestrial-aquatic linkages · Food webs · Phenology · Ponds · Zooplankton
“The rain falling on the freshly dried herbs and leaves, and filling the pools and ditches into which they have dropped thus clean and rigid, will soon convert them into tea,— green, black, brown, and yellow teas, of all degrees of strength, enough to set all Nature a-gossiping.” H. D. Thoreau, "Autumnal Tints,” 1862.
Communicated by Robert O. Hall. S. B. Fey and A. N. Mertens have contributed equally to this manuscript. S. B. Fey · A. N. Mertens · K. L. Cottingham Department of Biological Sciences, Dartmouth College, Hanover, NH, USA Present Address: S. B. Fey (*) Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT 06520, USA e-mail: [email protected]
Introduction Ecological subsidies, the energy and materials that move across ecosystem boundaries, influence the structure and composition of food webs (Polis et al. 1997; Takimoto et al. 2002; Gratton et al. 2008). The effect of such subsidies on the recipient community depends on many factors, including the characteristics (Cole et al. 2006) and quantity of materials transferred (Marcarelli et al. 2011), the rate
13
Oecologia
of subsidization (Takimoto et al. 2009), and the behavioral and life history characteristics of the recipient community (Baxter et al. 2005). Tremendous variation exists regarding the time period over which a subsidy can affect the recipient food web. For example, adult insects emerging from temperate aquatic ecosystems into terrestrial ecosystems generally integrate rapidly into the terrestrial food web (Nakano and Murakami 2001; Sabo and Power 2002). By contrast, coarse woody debris that enters aquatic ecosystems from surrounding forests may decay over decades or even centuries (Guyette and Cole 1999). Similarly, the integration of large whale carcasses (which can fall from pelagic habitats following death) into benthic ecosystems first by mobile scavengers, then polychaetes and crustaceans, and finally microbes, can take place over decades (Smith and Baco 2003; Higgs et al. 2014). As such, accurately quantifying the effect of a cross-ecosystem subsidy may require quantifying the recipient ecosystem’s response both immediately and over extended time periods. Leaves from deciduous trees represent an important subsidy for aquatic ecosystems that alters food webs on both short and long time scales (Anderson and Sedell 1979; Earl et al. 2014). Allochthonous inputs to streams from surrounding forests are well appreciated and typically exceed autochthonous inputs in certain ecosystems (Webster and Meyer 1997). By providing both limiting resources and microhabitat (Richardson 1992), leaves aid in the immediate production of bacteria, fungi, and detritivorous insects in streams (Hieber and Gessner 2002). The use of leaf packs containing terrestrial leaf litter also provides strong evidence that leaf subsidies are integrated into recipient ecosystems over long time periods, as shown by how the remaining leaf material continues to decrease in mass following multiple seasons of being utilized by stream biota (Benfield et al. 2001; Swan and Palmer 2004). Moreover, excluding terrestrial leaf litter inputs significantly reduces benthic invertebrate biomass, including shredders, gatherers, and even predators on interseasonal timescales (Wallace et al. 1999). In temperate pond ecosystems, autumn leaf drop can contribute substantially to annual carbon and nutrient budgets (France and Peters 1995; Hongve 1999; Pope et al. 1999), particularly in ecosystems unable to produce large amounts of autochthonous carbon (Rubbo et al. 2006). Furthermore, while there is a growing appreciation for the importance of leaves as a subsidy that immediately impacts producer and consumer components of pond food webs (Oertli 1993; Rubbo et al. 2008; Rubbo and Kiesecker 2004; Cottingham and Narayan 2013; Earl et al. 2014), little is known about the long-term effects of leaf subsidies for such food webs. Our current understanding of this effect comes primarily from two whole-ecosystem experiments that investigated the effect of leaf litter on small ponds by removing or restricting the fall of leaf litter in autumn, and measuring the subsequent
13
Fig. 1 Leaves and leaf leachates impact pond food webs by resourcemediated effects (solid black arrows), transparency-mediated effects (dashed grey arrows), and oxygen-mediated effects (solid grey arrows)
response in the spring. Here, Rubbo et al. (2006) showed that leaf removals decreased ecosystem respiration and reduced inter-seasonal secondary production. However, high between-pond variability produced contrasting results for a similar study investigating the effect of leaf litter on benthic invertebrates (Batzer and Palik 2007). To our knowledge, no study has addressed the long-term effect of leaves on organisms other than benthic invertebrates, and especially not on both the autotroph- and detritus-based food webs. There is reason to believe that autumn leaf subsidies could exert effects on planktonic food webs that extend beyond the autumn months, particularly in temperate ecosystems where reduced light availability and cold temperatures in winter are likely less conducive for planktonic food webs to assimilate the carbon and nutrients provided by autumn leaf subsidies (Twiss et al. 2012; Bertilsson et al. 2013). We identify three pathways through which leaves may affect pond food webs (Fig. 1). First, resource-mediated effects result from the consumption of leaf material, the uptake of nutrients and carbon from leaf leachates, or the consumption of organisms that uptake these materials. Second, transparency-mediated effects result from changes in water transparency that regulates the penetration of incoming solar radiation. Third, metabolism-mediated effects result from changes in the balance of respiration versus photosynthesis in response to leaf additions. The combined changes to the biotic and abiotic environment would have cascading but differential effects on zooplankton orders. Assuming these effects persist into the spring, we predict that cladocerans will benefit relative to copepods due to decreased ultraviolet radiation exposure accompanying leaf additions (Leech and Williamson 2000), cladocerans will be negatively impacted to a greater extent than copepods due to the decreases in oxygen concentrations (Vanderploeg et al. 2009) that accompany leaf additions (Cottingham and Narayan 2013), and the faster numeric responses of cladocerans would allow them to benefit from any increases in phytoplankton stimulated by leaf additions.
Oecologia
Therefore, we hypothesized that autumn leaf drop affects the spring dynamics of pond food webs by altering the availability of resources, the water transparency, and the metabolic state of ponds. We conducted a 10-month mesocosm experiment using an experimental array previously used to understand how variation in leaf chemistry driven by increased soil temperature can alter the quality of leaf subsidies for planktonic food webs (Fey et al. 2015). Here, we test two specific predictions: autumn leaf subsidies affect the initial state of planktonic food webs immediately following ice-off, and have a lasting effect on planktonic food webs beyond ice-off.
Materials and methods Our field experiment involved several stages: 1. Establishing mesocosms and imposing the control (5 g leaves added to mesocosms) and leaf-addition (50 g leaves added to mesocosms) experimental treatments. 2. Following responses to treatments prior to ice-on (Fey et al. 2015). 3. Allowing mesocosms to freeze over during the winter months. 4. Observing the treatment responses immediately following ice-out and throughout the spring. There were five replicates of each treatment, for a total of ten mesocosms. The primary response variables were phytoplankton chlorophyll-a, total nitrogen and total phosphorus concentrations, bacterial and zooplankton densities, and zooplankton community composition. We also characterized the abiotic environment by measuring dissolved oxygen, temperature, and the absorbance due to dissolved substances at 254 nm—a proxy for water clarity and dissolved organic carbon (Brandstetter et al. 1996). The leaf loading rate used (~150 g leaves/m2 pond surface area) falls within the reported rates of areal loading rates observed in the near-shore areas of lakes (Hongve 1999) and previous leaf-addition mesocosm experiments (Rubbo and Kiesecker 2004; Cottingham and Narayan 2013). These leaf additions correspond to a loading rate of ~23 g leaves per meter “shoreline” (as measured by mesocosm circumference), which falls below the 32–720 g/m reported annual shoreline rates (France and Peters 1995). Establishing mesocosm communities We constructed mesocosms from 167-L polyethylene cylindrical refuse bins (0.61-m diameter by 0.80-m depth, 1.95 surface to volume ratio) buried in the ground at the Dartmouth Organic Farm, Hanover, New Hampshire.
Mesocosms were filled with 120 L of groundwater on 5 September 2012 and covered with 1-mm fiberglass mesh to prevent colonization by non-target organisms. The following day, we added homogenized packets containing 5 g of the collected leaf subsidies to all mesocosms to provide a source of organic carbon and nutrients for developing biological communities. On 6–7 September 2012, we added 5 L of water from each of six small lakes and ponds within a 20-km radius of Hanover (30 L total, final mesocosm volume ~150 L) to seed mesocosms with natural phytoplankton and bacterial communities (Table S1). From 9 to 11 September, we added zooplankton from both the littoral (near shore) and pelagic (open water) zones of five of the six ponds. We collected zooplankton using a 12-L Schindler-Patalas trap; each mesocosm received zooplankton filtered from 144 L of pond water. Prior to imposing leaf-addition treatments, there were no differences between mesocosm treatments in any of our response variables (Fey et al. 2015). We did not add benthic invertebrates to our mesocosms in order to focus on the response of the pelagic food web. Generating experimental treatments We collected leaves from the Harvard Forest in central Massachusetts, USA (42°28′N, 72°10′W), rather than from the shoreline of the ponds used to establish mesocosms, to facilitate a prior experiment (Fey et al. 2015). On 28 October 2010 and 20 October 2011, we collected freshly fallen leaves from red maple trees (Acer rubrum), which are common around ponds across eastern North America (Rubbo and Kiesecker 2004). All leaves were collected after abscission but before a rain event. Leaves were dried at room temperature, and then stored in closed paper bags until the experiment so as not to alter leaf chemistry (Cotrufo and Ineson 1996). Leaf subsidies were added to mesocosms on 2 October 2012. We added 45 g of dried leaf litter (10 g from 2010 and 35 g from 2011) to each mesocosm in the leaf-addition treatment. Field sampling We monitored mesocosms through the fall season and noted that mesocosms were fully iced-over the week of 13 November (Fey et al. 2015). Ice-off occurred in all mesocosms during the last week of April 2013. While complete surface ice coverage of the mesocosms persisted during the winter, we did not measure the thickness of the ice to avoid disrupting under-ice dynamics. As the ice and snow melted, we removed additional water to ensure that the mesocosm volume remained at 150 L and that water did not make contact with the fiberglass screens enclosing mesocosms.
13
Oecologia
Beginning on 1 May 2013, we sampled mesocosms weekly between 0800 and 1200 hours. We recorded water temperature, dissolved oxygen, using a YSI model 556 MPS meter (YSI, Yellow Springs, OH). We then homogenized the mesocosms through stirring, removed a 5-L vertical cross-section of the water column using a vertical water sampler, and filled 3 1-L bottles for nutrient, chlorophyll, and bacterial analysis, respectively. All samples were stored in coolers containing ice packs until returning to the laboratory. We collected an additional 5 L of water to sample for zooplankton. We filtered the water through 53-μm mesh, washed the zooplankton off the mesh into 50-mL centrifuge tubes, preserved the zooplankton with 70 % ethanol, and returned the filtered water to the mesocosm. Mesocosm sampling ended at the end of spring, 19 June, when treatment and control mesocosms had converged in most physical and chemical properties. Laboratory analyses Nutrients and chlorophyll-a were analyzed followed previously described procedures (Fey and Cottingham 2012). Total nutrient samples were frozen until digestion and spectrophotometric analysis on a Perkin-Elmer Lambda EZ201 spectrophotometer (PerkinElmer, Waltham, MA). For total nitrogen, we used a basic persulfate digestion and measured nitrate using the second-derivative method (Crumpton et al. 1992). For total phosphorus, we used a persulfate digestion and the molybdate colorimetric reaction. Samples for dissolved materials were pre-filtered using PURADISC 25 GF/F disposable filters (Whatman, Maidstone, Kent, UK), and were measured spectrophotometrically by absorbance in a 1-cm quartz cuvette at 254 nm (Brandstetter et al. 1996). Finally, to measure chlorophyll-a concentration, we vacuum-filtered 100 mL of water onto 47-mm Whatman GF/C filters; froze the filters, extracted the chlorophyll in methanol, and measured concentrations using the Welschmeyer method (Welschmeyer 1994). Zooplankton were enumerated at 10× magnification on a Leica MZ-12 dissecting microscope (Leica Microsystems, Bannockburn, IL). Cladocerans were identified to genus; adult copepods were identified to order (Calanoida versus Cyclopoida), while copepodites were enumerated as a bulk category. We did not detect any rotifers in mesocosms, similar to previous mesocosm experiments (Fey and Cottingham 2012; Cottingham and Narayan 2013). Bacteria were enumerated using a modified 4′,6-diamidino-2-phenylindole (DAPI) staining method (Porter and Feig 1980). Upon returning from the field, whole water samples were fixed immediately with glutaraldehyde (3 % final concentration; Acros Organics, Geel, Belgium). Prior to enumeration, samples were filtered at low pressure ( 0.6). The effect of leaf additions on total phosphorus persisted over winter, such that total phosphorus in the leaf-addition treatment concentrations remained five times higher immediately following ice-off (Fig. 3b; ice-out t4.06 = 5.57, P = 0.005). Leaf additions initially altered chlorophyll-a concentrations and bacterial densities following ice-off. Leaf-addition treatments initially had reduced chlorophyll-a concentrations by 50 % (Fig. 4a; t5.55 = −3.73, P = 0.011). By contrast, bacterial densities were two-fold higher in
Fig. 3a, b Total phosphorus and nitrogen concentrations through the spring for leaf-addition and control mesocosms. Black squares are data from leaf-addition mesocosms and white circles are data from control mesocosms. Left panels are data from 13 November 2012, the last sample date before ice-over. Data are treatment mean ± SE (n = 5 replicates)
mesocosms receiving leaves immediately following ice-off (Fig. 4b; t7.57 = 3.20, P = 0.014). Leaf additions initially decreased cladoceran densities to near-zero densities relative to control treatments, although cladoceran densities were initially low in both treatments (Fig. 5a, c, e, g). Consequently, total cladoceran, Chydorus, and Daphnia densities were higher in control treatments immediately following ice-off (t3.15 = −4.97, P = 0.014; t3.83 = −4.55, P = 0.012; and t6.97 = −1.09, P > 0.3, respectively), while Bosmina densities did not differ between treatments (t6.96 = −10.05, P 0.3). Autumn leaf subsidies impact throughout the spring By late spring, most abiotic variables had converged or were converging between treatments. The absorbance of dissolved substances measured at 254 nm (1/m) decreased throughout the spring in the leaf-addition treatment,
13
Oecologia
Fig. 4a, b Basal resource response to leaf addition, as measured by chlorophyll-a concentration and bacteria densities. Black squares are data from leaf-addition mesocosms and white circles are data from control mesocosms. Left panels are data from 13 November 2012, the last sample date before ice-over (Fey et al. 2015). Data are treatment mean ± SE (n = 5 replicates)
while the control treatment did not change (Fig. 2a; RMANOVA time × treatment F7,59 = 10.89, P 0.6; end of spring, t6.61 = −1.24, P > 0.2). Despite the initial differences in phosphorus to leaf additions at ice-off, the chemical environments between treatments converged by the end of spring. Nitrogen concentrations increased similarly in both treatments (Fig. 3a; RM-ANOVA time F7,64 = 9.55, P 0.1). Phosphorus concentrations decreased in leafaddition treatments, while increasing in control treatments (Fig. 2b; RM-ANOVA time × treatment F7,64 = 11.32, P 0.7).
13
While chlorophyll-a concentrations differed immediately following ice-off, concentrations increased faster in leaf-addition treatments (Fig. 4a; RM-ANOVA time × treatment F7,64 = 4.85, P 0.2). Bacterial densities, however, followed the opposite trajectory. Bacteria densities were generally higher in leaf-addition mesocosms (Fig. 4b; RM-ANOVA treatment F1,64 = 10.283, P = 0.003) and increased more rapidly in the control treatment throughout the spring (Fig. 4b; RM-ANOVA time × treatment F4,64 = 5.26, P = 0.002), reaching densities 50 % higher in control than leaf-addition mesocosms by the end of spring (Fig. 4b; t6.98 = −2.44, P = 0.045). In contrast to the physical and chemical environment, zooplankton populations generally continued to diverge between treatments as the spring progressed. Leaf additions continued to suppress cladoceran densities compared to control treatments. Total cladoceran and Chydorus population densities were higher in control mesocosms throughout the spring (Fig. 5a, e; RM-ANOVA treatment F1,64 = 26.81, P
