Seizing resilience windows to foster passive recovery in the forest-water interface in Amazonian lands.

Forest regeneration has increased in many tropical abandoned lands and current restoration commitments in this region aim to restore over 1,400,000 km2 of degraded land by 2030. Although regenerating forests recover biomass, biodiversity, and processes with time, the recovery trajectories may be uncertain due to past disturbances. Currently, there is a lack of knowledge to sustain the effectiveness of passive regeneration for the recovery of riparian forests and the adjacent waterbodies in the tropics, which may compromise the outcomes of ongoing and future tropical riparian restoration programs. We evaluated the drivers of riparian forest structural recovery and how this relates to stream conditions in 12 abandoned pasturelands in eastern Brazilian Amazonia. These pasturelands range across regeneration age (pasture (PA) - 0 to 4 years; young regeneration (YR) - 8 to 12 years; old regeneration (OR) - 18 to 22 years) and years of past land-use (PA - 23.25 average years of past land-use, YR - 18.25, OR - 7). We compared the conditions of these sites to 4 reference sites with conserved forests (REF, >100 years), where there was no recorded pasture use in the past. Short-term responses of forests and streams to passive regeneration indicated high ecosystem resilience after low to intermediate past land-use intensity, reflected in the improvement of stream ecosystems. Such high resilience is possibly attributable to low- to intermediate-intensity pasture-related disturbances, remaining forest matrix, and residual structures (e.g. roots, sprouts, and in-stream wood) observed in the area. Our results suggest a recovery by 12 to 20 years for riparian forests of this region. However, areas degraded by intensive land-use apparently showed delayed recovery. We conclude that seizing resilience windows (defined here as the period when ecosystems retain high potential resilience) is essential to foster passive recovery of riparian forests and streams more cost-effectively in the tropics.

Evaluating ecosystem effects of climate change on tropical island streams using high spatial and temporal resolution sampling regimes.

Climate change is expected to alter precipitation patterns worldwide, which will affect streamflow in riverine ecosystems. It is vital to understand the impacts of projected flow variations, especially in tropical regions where the effects of climate change are expected to be one of the earliest to emerge. Space-for-time substitutions have been successful at predicting effects of climate change in terrestrial systems by using a spatial gradient to mimic the projected temporal change. However, concerns have been raised that the spatial variability in these models might not reflect the temporal variability. We utilized a well-constrained rainfall gradient on Hawaii Island to determine (a) how predicted decreases in flow and increases in flow variability affect stream food resources and consumers and (b) if using a high temporal (monthly, four streams) or a high spatial (annual, eight streams) resolution sampling scheme would alter the results of a space-for-time substitution. Declines in benthic and suspended resource quantity (10- to 40-fold) and quality (shift from macrophyte to leaf litter dominated) contributed to 35-fold decreases in macroinvertebrate biomass with predicted changes in the magnitude and variability in the flow. Invertebrate composition switched from caddisflies and damselflies to taxa with faster turnover rates (mosquitoes, copepods). Changes in resource and consumer composition patterns were stronger with high temporal resolution sampling. However, trends and ranges of results did not differ between the two sampling regimes, indicating that a suitable, well-constrained spatial gradient is an appropriate tool for examining temporal change. Our study is the first to investigate resource to community wide effects of climate change on tropical streams on a spatial and temporal scale. We determined that predicted flow alterations would decrease stream resource and consumer quantity and quality, which can alter stream function, as well as biomass and habitat for freshwater, marine, and terrestrial consumers dependent on these resources.

Implications of guppy (Poecilia reticulata) life-history phenotype for mosquito control.

Guppies (Poecilia reticulata) are frequently introduced to both natural and artificial water bodies as a mosquito control. Laboratory studies have demonstrated that guppies can consume large numbers of larval mosquitoes. Our study investigates how intraspecific variability in guppy phenotype affects their importance as a mosquito biocontrol and how habitat conditions (natural ponds vs. water storage containers) may influence insect biomass and guppy feeding. Using a blocked experimental design, we established stream-side mesocosm ponds with half receiving gravel substrate to simulate pond-bottom habitat. To provide realistic diet choices and insect abundances, we allowed the mesocosms to colonize naturally with aquatic insect larvae for 1 month before introducing guppies. We tested two distinct guppy phenotypes (from high- and low-predation streams) alongside fish-free controls. After 1 month, we measured insect biomass in the mesocosms and examined guppy gut contents to document direct predation. While overall insect biomass was not significantly different across the three fish treatments, we observed a significant reduction in mosquito biomass in fish treatments compared to fish-free controls, as well as intraspecific differences in feeding. Overall insect biomass was significantly higher in mesocosms without gravel, while habitat condition had no effect on mosquito biomass. As guppy phenotype responds to changes in their environments, it is an important consideration for biocontrol policy to anticipate potential ecosystem effects. We close by relating our findings to other studies and by discussing the implications and potential risks of using guppies to control mosquitoes.

Occurrence, leaching, and degradation of Cry1Ab protein from transgenic maize detritus in agricultural streams.

The insecticidal Cry1Ab protein expressed by transgenic (Bt) maize can enter adjacent water bodies via multiple pathways, but its fate in stream ecosystems is not as well studied as in terrestrial systems. In this study, we used a combination of field sampling and laboratory experiments to examine the occurrence, leaching, and degradation of soluble Cry1Ab protein derived from Bt maize in agricultural streams. We surveyed 11 agricultural streams in northwestern Indiana, USA, on 6 dates that encompassed the growing season, crop harvest, and snowmelt/spring flooding, and detected Cry1Ab protein in the water column and in flowing subsurface tile drains at concentrations of 3-60ng/L. In a series of laboratory experiments, submerged Bt maize leaves leached Cry1Ab into stream water with 1% of the protein remaining in leaves after 70d. Laboratory experiments suggested that dissolved Cry1Ab protein degraded rapidly in microcosms containing water-column microorganisms, and light did not enhance breakdown by stimulating assimilatory uptake of the protein by autotrophs. The common detection of Cry1Ab protein in streams sampled across an agricultural landscape, combined with laboratory studies showing rapid leaching and degradation, suggests that Cry1Ab may be pseudo-persistent at the watershed scale due to the multiple input pathways from the surrounding terrestrial environment.

Biodiversity and ecosystem risks arising from using guppies to control mosquitoes.

Deploying mosquito predators such as the guppy (Poecilia reticulata) into bodies of water where mosquitoes breed is a common strategy for limiting the spread of disease-carrying mosquitoes. Here, we draw on studies from epidemiology, conservation, ecology and evolution to show that the evidence for the effectiveness of guppies in controlling mosquitoes is weak, that the chances of accidental guppy introduction into local ecosystems are large, and that guppies can easily establish populations and damage these aquatic ecosystems. We highlight several knowledge and implementation gaps, and urge that this approach is either abandoned in favour of more effective strategies or that it is used much more rigorously. Controlling mosquitoes does not need to come at the expense of freshwater biodiversity.

Evidence for the persistence of food web structure after amphibian extirpation in a Neotropical stream.

Species losses are predicted to simplify food web structure, and disease-driven amphibian declines in Central America offer an opportunity to test this prediction. Assessment of insect community composition, combined with gut content analyses, was used to generate periphyton-insect food webs for a Panamanian stream, both pre- and post-amphibian decline. We then used network analysis to assess the effects of amphibian declines on food web structure. Although 48% of consumer taxa, including many insect taxa, were lost between pre- and post-amphibian decline sampling dates, connectance declined by less than 3%. We then quantified the resilience of food web structure by calculating the number of expected cascading extirpations from the loss of tadpoles. This analysis showed the expected effects of species loss on connectance and linkage density to be more than 60% and 40%, respectively, than were actually observed. Instead, new trophic linkages in the post-decline food web reorganized the food web topology, changing the identity of "hub" taxa, and consequently reducing the effects of amphibian declines on many food web attributes. Resilience of food web attributes was driven by a combination of changes in consumer diets, particularly those of insect predators, as well as the appearance of generalist insect consumers, suggesting that food web structure is maintained by factors independent of the original trophic linkages.

Occurrence of maize detritus and a transgenic insecticidal protein (Cry1Ab) within the stream network of an agricultural landscape.

Widespread planting of maize throughout the agricultural Midwest may result in detritus entering adjacent stream ecosystems, and 63% of the 2009 US maize crop was genetically modified to express insecticidal Cry proteins derived from Bacillus thuringiensis. Six months after harvest, we conducted a synoptic survey of 217 stream sites in Indiana to determine the extent of maize detritus and presence of Cry1Ab protein in the stream network. We found that 86% of stream sites contained maize leaves, cobs, husks, and/or stalks in the active stream channel. We also detected Cry1Ab protein in stream-channel maize at 13% of sites and in the water column at 23% of sites. We found that 82% of stream sites were adjacent to maize fields, and Geographical Information Systems analyses indicated that 100% of sites containing Cry1Ab-positive detritus in the active stream channel had maize planted within 500 m during the previous crop year. Maize detritus likely enters streams throughout the Corn Belt; using US Department of Agriculture land cover data, we estimate that 91% of the 256,446 km of streams/rivers in Iowa, Illinois, and Indiana are located within 500 m of a maize field. Maize detritus is common in low-gradient stream channels in northwestern Indiana, and Cry1Ab proteins persist in maize leaves and can be measured in the water column even 6 mo after harvest. Hence, maize detritus, and associated Cry1Ab proteins, are widely distributed and persistent in the headwater streams of a Corn Belt landscape.

Rapid decomposition of maize detritus in agricultural headwater streams.

Headwater streams draining agricultural landscapes receive maize leaves (Zea mays L.) via wind and surface runoff, yet the contribution of maize detritus to organic-matter processing in agricultural streams is largely unknown. We quantified decomposition and microbial respiration rates on conventional (non-Bt) and genetically engineered (Bt) maize in three low-order agricultural streams in northwestern Indiana, USA. We also examined how substrate quality and in-stream nutrient concentrations influenced microbial respiration on maize by comparing respiration on maize and red maple leaves (Acer rubrum) in three nutrient-rich agricultural streams and three low-nutrient forested streams. We found significantly higher rates of microbial respiration on maize vs. red maple leaves and higher rates in agricultural vs. forested streams. Thus both the elevated nutrient status of agricultural streams and the lability of maize detritus (e.g., low carbon-to-nitrogen ratio and low lignin content) result in a rapid incorporation of maize leaves into the aquatic microbial food web. We found that Bt maize had a faster decomposition rate than non-Bt maize, while microbial respiration rates did not differ between Bt and non-Bt maize. Decomposition rates were not negatively affected by genetic engineering, perhaps because the Bt toxin does not adversely affect the aquatic microbial assemblage involved in maize decomposition. Additionally, shredding caddisflies, which are known to have suppressed growth rates when fed Bt maize, were depauperate in these agricultural streams, and likely did not play a major role in maize decomposition. Overall, the conversion of native vegetation to row-crop agriculture appears to have altered the quantity, quality, and predictability of allochthonous carbon inputs to headwater streams, with unexplored effects on stream ecosystem structure and function.