Local thermal extremes shape the nature of herbivore plasticity that controls plant communities

Description

Notes

Methods

Field experiment design

For each grasshopper population, we factorially crossed spider predation risk (present, absent) and experimental warming (warming, ambient) using experimental mesocosm cages in the field. Each cylindrical cage (0.25 m² × 1.0 m H) was constructed using vinyl-coated garden wire fencing; the sides and tops of each cage were covered with insect screening. We deployed 20 cages (each separated by at least 1 m) within each field over naturally growing vegetation. We selected cage locations to ensure they all contained a similar initial composition of grass and Solidago, which was verified by measuring initial percent cover of grass and Solidago.

We created experimental warming treatments by wrapping the sides of assigned cages with translucent plastic sheeting (4 mm Film-Gard, Covalence Plastics, Minneapolis, MN, USA); the bottom 10 cm of the cages was left unwrapped to permit airflow. Experimental warming does not increase temperatures consistently over 24 hours: warming occurred during the day and dissipated at night, resulting in the simultaneous change of mean diel mean temperature (+0.67 °C), mean diel maximum temperature (+2.79 °C), mean diel minimum temperature (-0.27 °C), and mean diel coefficient of variation in temperature (+15%; P < 0.05 in all cases). While wrapping could influence plant growth and development by altering wind and humidity, prior work indicates that end-of-season plant biomass was unaffected directly by such experimental warming treatments.

From late June to early July, we cleared cages of all arthropods before randomly assigning spider predators and grasshoppers to the various treatment cages. We then stocked predator treatment cages with local spiders at natural field densities (1 spider per cage). In mid-July, we stocked all cages with local third-instar grasshoppers at natural field densities (5 grasshoppers per cage). One week after stocking, we replaced any grasshoppers that experienced mortality due to handling stress. Thereafter, we allowed grasshoppers to grow and develop under their respective treatment combinations until late September to early October.

Geographic variation in climatic conditions among local populations

Populations were selected to encompass a range of the thermal regimes (mean temperature, variability, and extremes) present in New England. Remotely sensed data may not capture fine spatial-scale thermal regimes experienced by grasshoppers within a local field site, therefore we recorded temperature hourly within the vegetation canopy at 60 cm above the soil with temperature loggers placed within multiple* *experimental field cages at every site (MX2201, Onset Corporation, Bourne, MA, USA). We included data from temperature loggers deployed in cages for the present study and those deployed in a parallel field experiment that utilized the same 8 populations in identical cages to better resolve and quantify variation in thermal regimes within local population sites. We calculated mean diel mean temperature, mean diel coefficient of variation in temperature, and mean diel temperature extremes for each logger at a population's field site. We then used GLMs to compare metrics with Site type as a fixed effect and population as a random effect nested within Site type (all conducted in R v.4.2.2 with lme4 v.1.1.31 , lmerTest v.3.1.3, DHARMa v.0.4.6, and MuMIn v.1.48.4 packages).

We tested the effect of experimental warming treatments on cage temperatures for each local population by analyzing temperature data from loggers deployed in 'warmed' and 'ambient' treatments with GLMs that compared the mean diel maximum, mean and minimum temperature, and mean diel coefficient of variation in temperature among populations and between Site types. Site type and experimental warming were considered fixed effects, and population was considered a random effect nested within Site type.

Behavioral Plasticity

During sunny days in mid-August, we assayed how grasshoppers from different populations responded behaviorally to predation risk and experimental warming by recording their habitat use within benchtop terraria in their native fields. Terraria (30 cm W × 40 cm L × 90 cm H) had a rectangular plywood base with sides and tops enclosed with fiberglass insect screening. Terraria were planted with 1-2 Solidago (60 cm H) and grasses grown in a greenhouse. We superimposed a 4 × 4 cm measurement grid on one face (84 cm × 40 cm) of each terrarium to allow measurement of the vertical position of grasshoppers in the canopy at each sampling time. We wrapped terraria assigned to experimental warming treatments with clear, thin plastic sheeting to generate warming. We recorded the temperature in a subset of the terraria every 5 minutes (MX2202 loggers from Onset Corporation, Bourne, MA, USA) at a canopy height of 60 cm.

At each field site, we transferred 2-3 grasshoppers from our previously established field experiment cages to a benchtop terrarium with the same experimental warming and predation risk treatments of sourced grasshoppers. For the predation risk treatment, we caught spiders from each respective field; however, due to a spider population crash in one of our VT populations, we sourced spiders from our second VT population. After stocking, we allowed the grasshoppers and spiders to acclimate to the terraria overnight and resume normal behaviors. Every 20 minutes between 09:00-17:40, we recorded the vertical position of each grasshopper within each terrarium. After completing habitat use measurements, grasshoppers were returned to their respective field experiment cages and spiders were confirmed to be alive and mobile in every predation risk terrarium.

During the behavioral assays in field terraria, experimental warming raised mean temperatures by 5.99 °C, maximum temperature by 7.25 °C, and minimum temperatures by 3.17 °C (P < 0.05 in all cases). Experimental warming also increased the coefficient of variation in temperature by 13%, but this trend was not significant (P = 0.0831). The influence of experimental warming on temperatures did not differ significantly among Site types (P > 0.05 in all cases). The effects of experimental warming were larger in the terraria than those in our field cages because we only assessed behavior on clear, sunny days, whereas field loggers within experimental cages integrated weather conditions over the whole season. We tested for behavioral plasticity by analyzing the percent change in mean daily vertical position within benchtop terraria from ambient baseline conditions with GLMs that considered predation risk and Site type as fixed effects with population as a random effect nested within Site type.

Physiological Plasticity

From late September to early October, we randomly selected one grasshopper from every treatment cage that had surviving grasshoppers and grouped them according to their treatment of origin (i.e., each factorial combination of experimental warming and predation risk). Grasshoppers were immediately transported to the laboratory at Yale University where their thermal performance was measured as mass specific respiration rate at 25°C, 30°C, and 35°C.

We evaluated the thermal performance of grasshoppers from different populations and rearing treatments, by measuring  the mass specific CO₂ respiration of grasshoppers across a 10°C gradient in temperature. We measured the respiration rates of grasshoppers at 25°C, 30°C, and 35°C as a metric of physiological stress because respiration integrates stress induced by both temperature and predation risk. After being collected in the field, grasshoppers were kept in an environmental control chamber (Percival Scientific, Inc., Perry, IA) at ~ 22°C with constant light levels and 50% relative humidity. Grasshoppers were given an ad libitum supply of water and no food for at least 24 hours after collection to put them in a post absorptive state. Just before beginning respiration rate measurements, we weighed (g) each grasshopper to allow calculation of their mass-specific respiration rate. We measured respiration as the volume of CO₂ exhaled by grasshoppers with a Qubit S-151 infrared CO2 analyzer (Q-S151 model, 1 ppm resolution: Qubit Biology Inc., Kingston, Ontario, Canada) after scrubbing the air of water vapor. We used an incurrent flow rate of ~ 110mL/min. We measured respiration at 25°C and increased temperature sequentially to 30°C and then 35°C with the Qubit gas switching system in the 8 tube stop flow arrangement. In this arrangement, we used the following parameters: Stop Time = 14 minutes, Dwell Time = 2 minutes, Delay Short = 0.2 minutes, Delay Long = 1.5 minutes, Event Time = 0.5 minutes. We decided on these intervals to avoid saturating the CO₂ analyzer, to produce repeatable results for an individual grasshopper, and to manage flow constraints of the system. In order to avoid saturating the system when assessing respiration, we used Airgas Ultra Zero Grade Air. Working with air of a known CO₂ concentration improved the accuracy of our measurements and allowed us to extend dwell times without the risk of saturating the CO₂ analyzer. We allowed grasshoppers to acclimate to each temperature for 32 minutes prior to measurement.

We tested for physiological plasticity by analyzing the percent change in the mass specific respiration rates of grasshoppers relative to baseline ambient conditions using GLMs that considered predation risk, Site type (i.e., Cool vs Warm), and assay temperature (25°C, 30°C, or 35°C) as fixed effects and population as a random effect nested within Site type. To account for repeated measures on individual grasshopper across all three temperature treatments, we also included 'individual' as a random effect nested within each unique combination of population, predation risk, and Site type.

Survival

At the end of the season, we tested for survival differences among grasshopper populations by analyzing the percent change in grasshopper counts from baseline ambient conditions with GLMs that considered predation risk and Site type as fixed effects and population as a random effect nested within Site type.

Geographic variation in the link between grasshopper trait plasticity and community-level impacts

At each field site, we removed all above-ground plant biomass within each experimental cage. Because grasshopper herbivory can reduce the biomass of grass and Solidago, we sorted all plant biomass from each cage by Solidago and grass. Solidago and grass samples were weighed after drying for 48 hours at 60°C. Cool sites and Warm sites did not differ with respect to initial Solidago cover, initial grass cover, or precipitation during the experiment (P>0.05 in all cases; 800 m² resolution precipitation from the PRISM Climate Group; http://www.prism.oregonstate.edu).

We evaluated the impacts of grasshoppers on their local plant community by testing for treatment effects on end-of-season Solidago biomass and grass biomass using GLMs that considered experimental warming, predation risk, and Site type as fixed effects and population as a random effect nested within Site type. To meet the assumptions of GLMs, Solidago biomass data was log₍₁₀₎ transformed before analysis.