Results

Ecosystem subsidies and food web interactions Watershed inputs and excretion by gizzard shad both represent important nutrient fluxes to the Acton Lake water column. Both of these nutrient sources can be considered “new” nutrient inputs to phytoplankton because they represent fluxes of nutrients from outside the water column (Vanni 2002). The relative importance of these two nutrient sources (watershed vs. fish) varies greatly over various temporal scales (Vanni et al. 2001, 2005, 2006a, 2011; Kelly et al. 2018; Williamson et al. 2018). In wet years, the watershed provides more soluble reactive P (SRP) than gizzard shad. However, in dry years P excretion by gizzard shad exceeds SRP inputs from the entire watershed averaged over the entire growing season (April-September) (Williamson et al. 2018) and this excretion can support a significant proportion of phytoplankton primary production (Vanni et al. 2006b). The relative importance of the two nutrient sources also varies over shorter time intervals, which are highly relevant to phytoplankton generation times. During storms (i.e. over a few days), watershed P inputs greatly exceed nutrient excretion by gizzard shad, but during intervening low-flow periods P excretion by shad exceeds SRP inputs from the watershed, even in wet years (Williamson et al. 2018). These shorter time scales are important because phytoplankton become P-limited within a week or so after storms (Vanni et al. 2006a). Other sources of new P to the euphotic zone such as direct release from sediments (including microbally-mediated P flux), entrainment from hypolimnetic water, and excretion by benthic invertebrates are usually much less important than watershed inputs or excretion by shad (Devine and Vanni 2002; Nowlin et al. 2005; Domine et al. 2010; Vanni et al. 2011). Subsidies of allochthonous sediment are also important in Acton Lake, and interact with nutrient subsidies at several time scales. Storms deliver large quantities of sediments, inducing a shift from nutrient- to light-limitation (Vanni et al. 2006a). Over longer time scales, sediment and dissolved nutrient subsidies can directly and indirectly increase detrital food resources for gizzard shad, thereby increasing shad abundance and the importance of nutrient translocation. Data from stable isotopes (deuterium) suggest that terrestrial detritus sustains about 30% of gizzard shad production in Acton Lake (and between 20-50% in 10 other reservoirs) (Babler et al. 2011). However, among-lake gizzard shad biomass is negatively correlated with “allochthony” (i.e., reliance on terrestrial detritus), suggesting that phytodetritus is a better food resource (Babler et al. 2011). Suspended sediments can also shade phytoplankton, reducing their productivity. Thus, the input of sediments from the watershed can represent a ‘negative subsidy’ that actually reduces ecosystem productivity, at least in the short-term.

Changes in agricultural practices Agricultural practices in Acton Lake’s watershed have changed markedly since the early 1990s, described in Renwick et al. (2008, 2018). Similar agricultural changes are occurring throughout the Midwest USA (and many other areas), driven by conservation and economics. The Acton Lake watershed exemplifies these changes, and provides an excellent model system in which to study linkages between changing agricultural management at the landscape scale and ecological responses in streams and lakes. The most important change in the watershed is a pronounced increase in conservation tillage, defined as practices that leave > 30% of the soil surface covered with crop residue (this includes no-till and reduced tillage). Conservation tillage in the Acton watershed increased dramatically in the 1990s, from ~15% of cropland in 1990 to >60% in 2000. This increase was facilitated by economic incentives provided to farmers toward the goal of reducing soil erosion and subsequent sedimentation rates in Acton Lake. Since 2000, conservation tillage has been relatively stable at 60-65% of cropland. Many watersheds throughout the Midwest USA have experienced similar patterns in the use of conservation tillage. Another potentially important change is a decline of ~50% in the number of hogs in the watershed from 1990 to 2002, but a moderate increase since that time. Finally, data on the use of fertilizers is inconclusive. On the one hand, fertilizer deliveries declined sharply in Preble County, OH, where about 80% of the Acton watershed lies. As with conservation tillage, this decline occurred in the 1990s with no consistent trend since ~2000. On the other hand, surveys of farmers reveal no temporal trends in fertilizer application rates over our study period (Renwick et al. 2018).

Trends in nutrient and sediment subsides from inflow streams to Acton Lake Trends in the concentrations of sediments and nutrients in Acton Lake’s inflow streams can be divided roughly into two decades (Renwick et al. 2008, 2018). Over approximately the first decade of our research (mid-1990s to mid-2000s), flow-weighted concentrations of suspended sediment (SS) and soluble reactive phosphorus (SRP) declined greatly, whereas nitrate concentration (the dominant nitrogen form) did not change much. However, during roughly the second decade of our research, the dynamics of stream nutrient concentrations changed greatly. SRP is no longer declining and actually increased over this period, and SS is still declining but at a slower rate. In contrast, nitrate declined greatly during this second decade. Over the entirety of our study, all of these constituents show a tendency toward declining concentrations at the watershed scale; however, declines have been greater for nitrate (~5% per year, 1994-2014) and SS (~4% per year) than for SRP (~1% per year) (Renwick et al. 2018). Loads of these constituents (mass per year delivered to the lake) also follow these trends. Thus, loads of nitrate (as well as total N) and SS show declines over the 21-year period (although loads are variable due to interannual variation in precipitation and runoff). In contrast, loads of SRP (and total P) declined over the first decade but show no temporal trend since then.

Trends in Acton Lake: Response to variable nutrient and sediment subsidies The response of Acton Lake phytoplankton to changes in nutrient and sediment inputs has been strong and includes some ‘ecological surprises’ (Twombly et al in review). During the period when stream P declined, lake phytoplankton biomass actually increased substantially (Kelly et al. 2018), even though phytoplankton were usually limited by P during this time (Hayes et al. 2015). The increase in phytoplankton was partly due an alleviation of light-limitation, because of declining suspended sediment concentrations in the lake. However, phytoplankton could not have increased without an alternative nutrient source. Most likely, this source is excretion by sediment-feeding fish (gizzard shad, Dorosoma cepedianum), which provide a nutrient subsidy to phytoplankton by feeding on sediment detritus and excreting nutrients into the water column. During the period of phytoplankton increase, ecosystem-wide excretion rates of gizzard shad increased substantially, fueled by an increase in their biomass (Williamson et al. 2018). Thus, fish (by providing nutrients via excretion) and reduced sediment concentrations (by alleviating light limitation) both provide resilience to the Acton ecosystem against reversal of eutrophication (Kelly et al. 2018). Over roughly the second decade of our study period, phytoplankton biomass has been relatively stable, but we have observed changes in the degree to which phytoplankton are limited by nitrogen vs. phosphorus. Phytoplankton are becoming more N-limited over time, especially in comparison to P limitation. Before ~2010, P was consistently the limiting nutrient for Acton phytoplankton in summer; N was limiting only in severe drought years, presumably because of lower nitrate supply from the watershed as well as increased denitrification efficiency (Hayes et al. 2015). However, in recent years we observe phytoplankton N-limitation even in years of average precipitation/runoff. The trend towards increasing N limitation, relative to P limitation, is consistent with declining nitrate concentrations and stable/increasing SRP concentrations in Acton inflow streams over the past decade. However, it appears that the trend in N vs. P limitation is best explained by a combination of declining watershed load N:P and an increase in the relative contribution of gizzard shad excretion. Gizzard shad excrete nutrients at a much lower N:P than is provided by watershed loads (Williamson et al. 2018); thus, as gizzard shad biomass increases, and as watershed loads decline, gizzard shad excretion becomes relatively more important to phytoplankton. The result is a lower N:P provided to phytoplankton, in aggregate (i.e., considering N and P supply from both fish and the watershed). Edit