Thresholds of acidification impacts on macroinvertebrates adapted to seasonally acidified tropical streams: potential responses to extreme drought-driven pH declines

Introduction

Anthropogenic impacts on aquatic ecosystems include acidification caused by acid rain (Schindler, 1988) and acid mine drainage (MacCausland & McTammany, 2007). Acidification of freshwater systems is also caused by other mechanisms, including CO₂ inputs (e.g., Hasler et al., 2018) and dilution of alkalinity by seasonal snowmelt (Lepori, Barbieri & Ormerod, 2003). Acidification can alter many ecosystem characteristics, including food web structure (Gray et al., 2016), decomposition rates (Dangles et al., 2004a), nutrient cycling (Ely, Von Schiller & Valett, 2010), algal biomass (Smucker & Vis, 2011), and biodiversity of fish Baker et al. (1996) and macroinvertebrates (Rosemond et al., 1992; Lepori, Barbieri & Ormerod, 2003; Baldigo et al., 2009). Macroinvertebrates in particular are widely known as indicators of water quality and ecosystem health, and the effects of acidification on benthic macroinvertebrates are well documented. Low pH can lead to shifts in community composition (Rosemond et al., 1992; Kullberg, 2002; Woodward, Jones & Hildrew, 2002), increased drift (Hall et al., 1980; Courtney & Clements, 1998), and reduced survival and emergence (Bell, 1971; Courtney & Clements, 1998; Thomsen & Friberg, 2002). Macroinvertebrates are frequently negatively affected in environments that have been exposed to anthropogenic acidification via acid deposition or acid mine drainage (Gray, 1998; Kowalik & Ormerod, 2006), and CO₂-induced acidification can also cause physiological stress and increased mortality in molluscs (reviewed in Hasler et al., 2018) and decreased feeding responses in crayfish (Allison, Dunham & Harvey, 1992).

However, macroinvertebrate assemblages are often tolerant to acidification in habitats that exhibit historically low or variable pH, such as temporary ponds (Batzer, Palik & Buech, 2004), streams with high concentrations of organic acids (Winterbourn & Collier, 1987; Collier et al., 1990; Dangles, Malmqvist & Laudon, 2004b), and mountain streams acidified by seasonal snowmelt (Petrin, Laudon & Malmqvist, 2007). In these cases, macroinvertebrates may have physiologically adapted to tolerate increased acidity; additionally, high concentrations of organic material in some systems may help to bind toxic aluminum ions that are released under acidic conditions (Dangles, Malmqvist & Laudon, 2004b). Thresholds of acidification effects vary greatly, both between and within taxa: larval midges (Chironomidae: Diptera) have been found to be tolerant to conditions with pH <3 (e.g., Orendt, 1999), while other taxa exhibit striking differences in local adaptation to acidic conditions (Petrin, Laudon & Malmqvist, 2007). The vast majority of pH-related studies have been conducted in the temperate zone, and little is known about acidification tolerance in tropical stream communities (but see Cranston et al., 1997; Roque et al., 2010).

Drought-driven stream pH declines are geographically widespread, occurring in Europe, North America, and Central America, and are predicted to increase with shifting precipitation regimes due to climate change (Laudon et al., 2004; Nordstrom, 2009; Small et al., 2012). These pH declines may be driven by dilution, organic acidity, oxidation of sulfides in soil and rock (Laudon et al., 2004; Nordstrom, 2009), pyrite oxidation (Sammut et al., 1995) and, particularly in the tropics, by biogenic CO₂ inputs from soil respiration following dry periods (Small et al., 2012) and potentially iron oxidation from lateritic soils (Ganong, 2015). Effects of episodic acidification on stream macroinvertebrates have been described mainly in the context of acidification events driven by anthropogenic acid deposition (e.g., Kowalik et al., 2007) or mining (e.g., MacCausland & McTammany, 2007). Stream macroinvertebrate tolerance to natural drought-driven pH declines remains virtually unstudied save for a CO₂-induced experimental stream acidification in Puerto Rico, which showed decreased macroinvertebrate abundance and richness at a pH of 5.4 (Klem & Gutiérrez-Fonseca, 2017).

While macroinvertebrates might be expected to be more tolerant to natural pH declines that recur over time than to acute, anthropogenically-driven pH declines, natural drought-driven pH declines are likely to intensify in many areas due to climate change. The Neotropics have been labeled a “hot spot” for climate change (Giorgi, 2006), and drought-driven pH declines are predicted to become increasingly frequent and severe due to increased precipitation variability and dry conditions in coming decades (Rauscher et al., 2008; Min et al., 2011). In lowland Costa Rica, wet-season pH declines can last for months and are negatively correlated with dry-season rainfall in small (first- to third-order) rainforest streams (Small et al., 2012). Extreme acidification events (pH < 4.5) have been observed after unusually dry seasons (Small et al., 2012).

Our objectives were to: (1) determine thresholds of lethal and sub-lethal effects of acidification on benthic macroinvertebrates from lowland streams at La Selva Biological Station (La Selva) in Costa Rica; and (2) examine these thresholds in the context of pH variability from a 16-year stream chemistry dataset, which includes an extreme pH drop resulting from a prolonged dry period associated with a strong ENSO event (1998). We used laboratory mesocosm experiments to test thresholds of macroinvertebrate response to pH declines in two streams: a poorly-buffered stream that is prone to seasonal pH fluctuation (pH 4.3–6.9; typical of many tropical streams) and a stream that is naturally well-buffered (pH 5.1–6.9) due to inputs of bicarbonate-rich regional groundwater. We tested: (1) thresholds of acidification effects on survival of three focal taxa that included the most abundant benthic macroinvertebrates (mayfly naiads, adult shrimp, and midge larvae); and (2) whether chironomid (midge) larvae are uniformly adapted to historic acidification regimes or whether there is local adaptation (i.e., differential tolerance in terms of growth and survival) to acidification between chironomid larvae from the poorly-buffered vs. the well-buffered stream. We hypothesized that macroinvertebrates are adapted to acidification within their natal stream’s characteristic pH range (based on the 16-year pH dataset); previous work has shown that La Selva stream insects are adapted to their natal streams’ phosphorus concentration, another key physicochemical parameter that varies across La Selva streams (Small, Wares & Pringle, 2011), suggesting that stream macroinvertebrates may be adapted to the physicochemistry of their natal stream. We predicted that: (1) macroinvertebrate survival and growth rates over would decrease at, or slightly below, the lowest level in this pH range; and (2) the threshold of growth/survival effects would occur at a higher pH level for macroinvertebrates from the well-buffered stream than for macroinvertebrates from the poorly-buffered stream.

Materials & Methods

Study site

Our study site was La Selva Biological Station, a 1,536-ha reserve located in lowland rainforest on the Caribbean slope of Costa Rica (Fig. 1A). La Selva is drained by numerous small streams, some of which receive natural interbasin flows of geothermally-modified regional groundwater that emerges in seeps at the base of Pleistocene lava flows (Pringle & Triska, 1991; Pringle et al., 1993). Due to incorporation of ions derived from weathering bedrock, this regional groundwater has high concentrations of many solutes, including bicarbonate, and is naturally well-buffered. In contrast, local groundwater—derived from precipitation within the watershed—is solute-poor and poorly-buffered. Consequently, streams draining La Selva can be solute-poor or solute-rich, depending on whether they receive regional bicarbonate-rich groundwater (Fig. 1B), resulting in very different pH regimes due to the natural buffering capacity of the solute-rich groundwater: well-buffered streams have a relatively high pH (∼6), while poorly-buffered streams have a lower pH that is more prone to fluctuation. Monthly pH measurements at 14 sites have been taken by the same technician since 1997 (methodological details below). Stream pH varies on multiple time scales and is driven by rainfall, with declines due in part to inputs of CO₂ derived from soil respiration (Small et al., 2012). Transient pH declines of 0.3 to 0.5 units occur immediately after individual precipitation events (Small et al., 2012; Ardón et al., 2013). Longer pH declines occur during the wet season, and low dry-season precipitation is correlated with strong subsequent wet-season stream pH declines that may last for months, with more pronounced pH declines in poorly-buffered streams (Small et al., 2012). Extreme climatic events cause large inter-annual variation: after a severe ENSO-driven drought in 1998, pH in some poorly-buffered streams dropped below 4 (Ramírez, Pringle & Douglas, 2006; Small et al., 2012). Using a one-year monthly dataset from six streams, Ramírez, Pringle & Douglas (2006) showed that insect biomass and density in these streams increased with pH, declining during the major pH drop in 1998. An examination of a 15-year dataset from two poorly-buffered streams indicated that macroinvertebrate richness and abundance were most strongly influenced by stream discharge (Gutiérrez-Fonseca, Ramírez & Pringle, 2018).

Our two study streams were the poorly-buffered Sura-60 and the well-buffered Arboleda, which are similar in terms of size and discharge (Sura-60 0.19 m³/s, Arboleda 0.17 m³/s; Small & Pringle, 2010). Sura-60 substrate consists mainly of boulders and gravel, while the Arboleda’s substrate is principally sediment with some boulders. These streams have served as focal sites for previous comparative work on aquatic macroinvertebrates (e.g., Small & Pringle, 2010; Small, Wares & Pringle, 2011).

Results

Monthly pH readings from 1997–2013 show a mean pH of 5.63 (range 4.32–6.94) in the poorly-buffered focal stream (Sura-60) and a mean pH of 6.20 (range 5.11–6.92) in the well-buffered focal stream (Arboleda). A yearlong monthly time series (July 2005–July 2006) demonstrates the seasonal variability in pH in both streams: pH in Sura-60 had a range of 1.62 units (pH 4.72 to 6.34) and fluctuated up to 1.02 units in consecutive months, while pH in the Arboleda had a range of only 0.43 unit (pH 5.89–6.32), varying at most 0.30 unit between consecutive months (Fig. 2A).

Fluctuations in pH that we observed at monthly intervals are also evident at seasonal and interannual timescales. In the 16-year time series (1998-2013), the mean annual pH of the well-buffered stream remained relatively constant (range 6.00 to 6.41), with mean monthly pH shifting by 0.01 to 0.35 units between consecutive dry and wet seasons (Fig. 2B). In contrast, mean annual pH in the poorly-buffered stream varied by 1.21 units (range 5.15 to 6.36), and mean monthly pH shifted by 0.06 to 0.68 unit between consecutive dry and wet seasons. The exception was during the 1998 ENSO-driven drought, when mean monthly pH of the poorly-buffered stream during the dry season (January–May) was 6.29, but dropped sharply to 4.80—a decline of 1.49 units—in the ensuing wet season (August–December) (Fig. 2C).

Mayfly survival time decreased significantly as pH declined (P < 0.0001, power = 0.99, effect size = 1.06), with survival significantly longer at and above pH 4 than at pH 3.5 and 3.0 (Fig. 3). Shrimp survival time also decreased significantly with declining pH (P = 0.001, power = 0.87, effect size = 1.06), with survival in the pH 4 treatment intermediate between survival at pH 5 and 6 vs. pH 3 and 3.5 (Fig. 3).

Survival of larval chironomids from both Sura-60 and Arboleda ranged from 56–70% from pH 3.5 to 6.0; survival decreased sharply between pH 3.5 and 3.0, with total mortality at pH 2.5 (Fig. 4A). There was no significant difference in survival, or in initial size, between Sura-60 and Arboleda chironomids in any pH treatment (P > 0.05, Sura: power = 0.94, effect size = 0.81; Arboleda: power = 0.96, effect size = 0.90) (Fig. 4A).

Our model of predicted larval chironomid survival yielded similar results: we excluded pH 2.5 due to complete mortality, and 95% credible intervals for pH 2.75 were lower than those for pH 3.0, which in turn were lower than those for pH 3.5 to 6.0; the 95% CI for pH 3.5–6.0 all overlapped (Fig. 4B). The 95% credible interval for stream effect (−0.618 to 0.455) overlapped with 0, indicating no significant stream effect (Table 1). The size effect was positive (95% CI [0.161–0.703]), showing a positive effect of initial chironomid size on survival, and the random effect ranged from [0.275–1.478], indicating a fairly strong “bucket” effect (i.e., unexplained variation among replicates) (Table 1). The effect on survival (exponentiation of the regression coefficient) was less than 1 for pH 2.75 and 3.0, indicating reduced survival, and greater than 1 for all other pH treatments (Table 1). R-hat values for all effect sizes ranged from 1.000 to 1.003, indicating convergence.

Initial size and pH treatment together significantly influenced percent biomass increase at the Arboleda (P = 0.04, R² = 0.19) and Sura-60 (P < 0.001, R² = 0.54). In the Arboleda, this trend was driven by the strongly significant negative relationship between growth rate and initial size of larval chironomids (P = 0.02, β = −0.398), with a nonsignificant effect of pH treatment (P = 0.24, β = 0.192). In Sura-60, growth rate was also negatively impacted by initial size (P < 0.001, β = −0.705) but was also positively related to pH (P = 0.003, β = 0.342) (Fig. 5). Initial size and pH treatment were not significantly correlated at either site.

Discussion

Our results indicate that stream macroinvertebrates are tolerant of typical seasonal and interannual pH shifts in poorly-buffered streams, but our data also suggest that stream macroinvertebrates are vulnerable to extreme pH declines caused by severe drought. Monthly pH readings taken from poorly-buffered streams during/after a severe ENSO event in 1998 demonstrate the lowest pH values ever recorded in several streams at La Selva, and these pH values include four instances in which survival thresholds of focal taxa overlap with pH measurements from three streams (Fig. 6), including Sura-100 (pH 3.90) in July, Saltito-100 (pH 3.80) and Sura-100 (pH 3.98) in November, and Carapa (pH 3.62) in December.

Ours is the first study (of which we are aware) to determine thresholds of pH tolerance for macroinvertebrates in streams experiencing extreme natural drought-driven acidification. Severe dry seasons in lowland Costa Rica lead to sharp stream pH declines in subsequent wet seasons (Small et al., 2012): in 1998, intra-annual pH variation (i.e., the difference between the lowest and highest monthly pH) ranged from 1.73 (pH 5.93 to 4.20, at the Taconazo) to 3.01 units (pH 6.91 to 3.90 at Sura-100). Such pH declines are not necessarily restricted to our study system and could be widespread in poorly-buffered streams draining weathered, nutrient-poor tropical soils in areas with seasonal precipitation regimes. These characteristics apply to large areas of the tropics that lack longterm pH records; however, sharp pH declines (from 7.4 to 4) have been recorded in small streams in the Ecuadorean Amazon during floods (Capps et al., 2011).

A comparison of pH declines in La Selva streams and other naturally- and anthropogenically-acidified streams indicates that the magnitude of pH declines (up to 3.01 units) during the extreme acidification event in 1998 is similar to natural pH declines recorded in two other systems, the Ecuadorean Amazon and western US (Table 2). However, the duration of the pH declines in unbuffered streams of La Selva was much longer, with pH values <5 recorded for up to seven months, in one case including pH values <4 for two consecutive months. Notably, the magnitude of natural pH drops driven by precipitation regime (droughts, rainfall, floods, snowmelt) is similar to—and sometimes greater than—the magnitude of declines driven by episodic anthropogenic acidification (Table 2).

Natural drought-driven stream acidification is not only a tropical phenomenon: it also occurs in North American and European streams due to oxidation of sulfide soils in desiccating wetlands (Laudon & Bishop, 2002; Eimers et al., 2007) or acid rock (Verplanck, McCleskey & Nordstrom, 2006; Nordstrom, 2009) (Table 2), as well as in Australian lakes and wetlands (Sammut et al., 1995; Sommer & Horwitz, 2001). These systems are also vulnerable to greater drought-driven pH declines due to climate change (Nordstrom, 2009; Erlandsson, Laudon & Fölster, 2010), so knowledge of the thresholds of macroinvertebrate pH tolerance could have important management implications for temperate as well as tropical streams.

Contrary to our hypothesis, chironomid larvae appear to exhibit general, rather than local, adaptation to stream pH in terms of survival and growth: survival did not differ between sites (Fig. 4), and while pH did significantly impact growth rate at Sura-60, pH explained a very low percentage of variability in growth rate (Fig. 5). Our laboratory results indicating general tolerance of chironomids are supported by two previous field studies involving multiple taxa. First, macroinvertebrate drift (and shrimp movement downstream) during an in-stream acidification experiment at La Selva increased when pH decreased to 4.3–4.35 (Ardón et al., 2013), just above our recorded threshold values. The drifting/downstream-movement response suggests that, in cases of extreme (natural) pH declines, well-buffered downstream reaches could potentially serve as refugia at a landscape scale for organisms moving downstream from poorly-buffered sites (Fig. 1). Second, macroinvertebrate community composition does not differ between well-buffered and poorly-buffered streams (Ramírez, Pringle & Douglas, 2006), suggesting general macroinvertebrate tolerance to low pH. This similarity in macroinvertebrate assemblage composition between well- and poorly-buffered streams contrasts with the findings from other studies in which community composition varied greatly between streams in close proximity with different pH regimes (Lepori, Barbieri & Ormerod, 2003; Kowalik et al., 2007).

The magnitude of natural pH declines within La Selva streams is similar to or greater than that of pH drops in other systems (Table 2), but the ecological impacts—or lack thereof—are more likely due to the source and historical duration of the declines, rather than the magnitude. The prevalence of tolerant taxa in La Selva streams—which have presumably undergone natural acidification for millennia—highlights the importance of historical pH regimes in structuring macroinvertebrate communities (Collier et al., 1990; Dangles et al., 2004a; Dangles, Malmqvist & Laudon, 2004b). Comparison of our findings with pH tolerances from other systems (Table 3) shows that systems that have experienced natural acidification for millennia, such as mountain streams with pH declines driven by snowmelt, often include macroinvertebrate communities that are highly resistant to acidification (Lepori, Barbieri & Ormerod, 2003; Petrin, Laudon & Malmqvist, 2007). In contrast, macroinvertebrates in streams that have historically exhibited circumneutral pH levels are often severely impacted by sudden exposure to acidic conditions caused by acid rain or acid mine drainage (Guérold et al., 2000; Kowalik & Ormerod, 2006; MacCausland & McTammany, 2007).

Increased incidence of severe pH declines due to climate change would likely have ecosystem-level effects due to potential loss of key taxa. The lower tolerance of mayflies and shrimp (Fig. 3) compared to chironomids (Fig. 4) suggests that extreme pH declines might simplify macroinvertebrate communities as mayflies and shrimp die, drift, or move downstream to escape low pH levels. Shrimp can play key roles in reducing sediment accrual and driving insect densities and algal community composition (Pringle & Hamazaki, 1998; Rosemond et al., 2001), as well as contributing up to 94% of stream invertebrate biomass and 40% of macroinvertebrate secondary production (Snyder, 2012). Therefore, the loss of Macrobrachium could potentially cause a significant shift in ecosystem structure and function due to changes in algal taxa, higher chironomid density, and greater sediment accrual.

Conclusion

Our work provides evidence that macroinvertebrates in La Selva streams are tolerant to naturally low pH levels, but are vulnerable to the extreme pH declines (pH < 4) observed following severe droughts. Chironomid growth and survival were unaffected by pH regimes above 3.0, indicating tolerance to pH levels experienced in the poorly-buffered stream during natural seasonal acidification, while shrimp and mayfly survival were impacted at pH levels below 4. Given that droughts are predicted to increase in frequency and severity in the future, climate change potentially threatens macroinvertebrate taxa in these streams. These findings also signal the need for further research in streams worldwide that are vulnerable to drought-driven pH declines.

Supplemental Information