Weather affects population dynamics in continental grassland over annual and decadal periods

Item Type Article; text

Authors Jonas, J. L.; Wolesensky, W.; Joern, A.

Citation Jonas, J. L., Wolesensky, W., & Joern, A. (2015). Weather affects grasshopper population dynamics in continental grassland over annual and decadal periods. Rangeland Ecology & Management, 68(1), 29–39.

DOI 10.1016/j.rama.2014.12.011

Publisher Society for Range Management

Journal Rangeland Ecology & Management

Rights Copyright © Society for Range Management.

Download date 26/09/2021 13:56:46

Item License http://rightsstatements.org/vocab/InC/1.0/

Version Final published version

Link to Item http://hdl.handle.net/10150/656957 Rangeland Ecology & Management 68 (2015) 29–39

Contents lists available at ScienceDirect

Rangeland Ecology & Management

journal homepage: http://www.elsevier.com/locate/rama

Weather Affects Grasshopper Population Dynamics in Continental Grassland Over Annual and Decadal Periods☆,☆☆

Jayne L. Jonas a, William Wolesensky b,AnthonyJoernc,⁎ a Research Associate III, Department of Forest and Rangeland Stewardship, Colorado State University, Fort Collins, CO 80523, USA b Lecturer, Department of Mathematics, University of Texas, Austin, TX 78712 USA c Professor, Division of Biology, Kansas State University, Manhattan, KS 66506, USA article info abstract

Article history: Understanding the complex dynamics of requires consideration of both exogenous and Received 20 May 2014 endogenous factors at multiple temporal scales. This problem is difficult due to differences in population Accepted 20 October 2014 responses among closely related taxa. Increased understanding of dynamic relationships between exoge- nous and endogenous factors will facilitate forecasting and suggest nodes in the life cycle of economically Keywords: important species susceptible to intervention by managers. This study uses an information-theoretic ap- proach to examine the contributions of weather and density to model population densities and growth atmospheric oscillations (NAO, PDO, SOI) density dependence rates of nine common grasshopper species from continental U.S. grassland over 25 years. In general, exogenous and endogenous feedbacks grass-feeding species and total grass-feeders as a functional group were most closely associated with grasshopper control weather during the year before hatching. Increased variability in prior growing season precipitation was insect herbivores associated with increased densities of bivittata, , Phoetaliotes nebrascensis, and the population growth rate (R) grass-feeding guild. sanguinipes densities tended to be smaller following warm fall seasons, while Amphitoruns coloradus declined during the positive phase of the North Atlantic Oscillation or after warmer than average winters. Population growth rate dynamics of all grouped species combinations were best explained by models including variability in precipitation during the prior year growing season. Large-scale Pacific Decadal Oscillation (PDO) patterns were also associated with growth rate dynamics of the mixed-feeding species group. Density showed a negative relationship with population growth rates of five species. This study indicates the importance of parental and diapause environmental conditions and the utility of incorporating long-term, readily obtained decadal weather indices for forecasting grasshop- per densities and identifying critical years with regard to grasshopper management—at least to the degree that the past will continue to predict the future as global climates change. © 2015 Society for Range Management. Published by Elsevier Inc. All rights reserved.

Introduction conditions (especially temperature) affect insect physiology and per- formance (Casey, 1993; Yang and Joern, 1994), mechanistic links to Even after years of investigation (Andrewartha and Birch, 1954; population dynamics can be hypothesized readily (Begon, 1983; Stige et al., 2007; Uvarov, 1931; White, 2007, 2008), the need remains Kingsolver, 1989; Logan and Powell, 2001). Precipitation may be to better understand the contribution of weather to long- term popu- equally important, most likely acting through host plant quality lation responses of insect herbivores. Because variable weather (Branson, 2008; Joern and Gaines, 1990; Jones and Coleman, 1991; Mattson and Haack, 1987; White, 1993). Abiotic conditions may also influence the strengths and even outcomes of biotic interactions as they affect population dynamics (Gratton and Denno, 2003; Ovadia ☆ Research over this 25-year period was supported in part by NSF/DEB and DOE/ NIGEC. Analyses were supported by funds to AJ from NSF EPS 0553722 (Ecological and Schmitz, 2004; Ritchie, 1996, 2000), or they can have major ef- Forecasting), DEB0456522, and the NSF Konza LTER Program. fects on food quality to herbivores and thus alter biotic interactions in- ☆☆ At the time of research, Jonas was a graduate research assistant, Division of directly (Danner and Joern, 2003; Joern and Mole, 2005; Louda and Biology, Kansas State University, Manhattan, KS 66506, USA. Collinge, 1992; Mattson and Haack, 1987; White, 1993). Finally, ⁎ Correspondence: Anthony Joern, Division of Biology, Kansas State University, larger-scale climate patterns (North Atlantic Oscillation, NAO; El Manhattan, KS 66505, USA. Tel.: +1 785 532 7073. E-mail address: [email protected] (A. Joern). Nino-Southern Oscillation, SOI; Pacific Decadal Oscillation, PDO;

http://dx.doi.org/10.1016/j.rama.2014.12.011 1550-7424/© 2015 Society for Range Management. Published by Elsevier Inc. All rights reserved. 30 J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39

Palmer Drought Severity Index, PDSI) may act as decadal-scale forcing Methods functions (Jonas and Joern, 2007; Stenseth et al., 2002; Stige et al., 2007). Temporal fluctuations in large-scale atmospheric processes Study Site alter environmental conditions, in turn affecting physiological pro- cesses, biotic interactions, and population processes over multiple- were censused for 25 years at Arapaho Prairie year periods. Although influences of NAO, SOI, or PDO for are (Arthur Co., NE; USA: 41°29'18'N, 101°51'28"W), a protected conti- not well characterized, one can readily envision how large-scale cli- nental grassland research site in south-central Nebraska. The site is mate patterns could synchronize populations at regional scales, affect characterized by upland sandhills grassland composed of large stabi- phenological timing, or coordinate interactions among species at dif- lized sand dunes with steep upper ridges that gradually slope into ferent trophic levels (Post and Forchhammer, 2008; Stenseth and broad flat valleys. Vegetation at Arapaho Prairie is an open-canopy Mysterud, 2005; Stenseth et al., 2002; White, 2008), thus assisting mixed-grass prairie, modified by sandy substrate (Barnes et al., forecasting efforts. 1984). Grasses contribute 80% to total plant biomass, with long- Both biotic (endogenous) and abiotic (exogenous) factors are term ANPP ranging between ~75 and 250 g · m−2 (unpublished known to influence insect population dynamics in complex ways, data). In this region, C3 and C4 grass species typical of eastern tallgrass making it difficult to identify key underlying mechanisms or inter- prairie and western shortgrass steppe grasslands intermix. Dominant actions among multiple factors that limit population abundances plants in this sand dune landscape form recognizable vegetation as- (Andrewartha and Birch, 1954; Barbosa and Schultz, 1987; sociations along the existing topographic gradient (Barnes et al., Belovsky and Joern, 1995; Berryman, 1999; Cappuccino and Price, 1984). The grass canopy is intermingled with extensive bare ground, 1995; Denno and McClure, 1983). Variation in population responses resulting largely from pocket gopher disturbance. Grasshoppers feed seen among even closely related taxa compound the difficulties in on a large proportion of available plant taxa (Joern, 1983). While the addressing this problem (Nothnagle and Schultz, 1987). In addition site was formerly subject to seasonal cattle grazing, management to basic ecological challenges of understanding population time since 1977 included mowing and intermittent light cattle grazing. series (Barbosa and Schultz, 1987; Cappuccino and Price, 1995; Grasshopper population responses were similar to nearby sites that Royama, 1992; Turchin, 2003), applied biologists are keen to un- aregrazedbycattle(Danner and Joern, 2004). derstand dynamic relationships to facilitate forecasting or suggest susceptible nodes in the life cycle of economically important Grasshopper Censuses species amenable to intervention (Branson et al., 2006; Lockwood et al., 2000). Grasshopper populations at Arapaho Prairie were censused in the Some studies link the dynamics of grasshopper populations with same valley area (~0.5 ha) each year for 25 years (1978–2003) dur- variable weather, although regression relationships often exhibit low ing the period of peak seasonal adult density (early–mid August). Al- predictability (Johnson and Worobec, 1988). For example, Capinera though more than 65 grasshopper species have been collected at and Horton (1989) reported significant but somewhat low R2 values Arapaho Prairie, only about 30 species are encountered each year (R2 = 0.14 – 0.32) from multiple regression analyses for several and most are uncommon. Nine species were common enough to be western U.S. grasshopper populations; although weather can be sig- censused reliably each year of the study. All grasshoppers considered nificant, much of the variation in population abundance is not readily here are univoltine with females ovipositing in mid-to-late summer, explained directly by these variables. Alternately, consistently good and overwintering eggs hatch from May to July the next year. fits between weather variables and grasshopper densities have In mixed grasslands with open vegetation structure such as Arap- been observed in other studies (R2 = 0.48 – 0.84) (Fielding and aho Prairie, grasshopper densities are best estimated using the ring Brusven, 1990). This raises an important issue—does variation in transect method (Onsager, 1977). Four transects of 40 0.1-m2 rings weather reliably predict grasshopper population dynamics, or should were placed randomly along each transect and grasshoppers counted we be looking to endogenous biotic causes that might act in concert the next day. Rings were at least 2 m apart so that counting one ring with or independent of variable weather (Belovsky and Joern, did not influence movement into or out of nearby rings. Transects 1995; Danner and Joern, 2004; Joern and Gaines, 1990; Ovadia and were located 20–30 m apart. Density per transect was determined Schmitz, 2004)? Moreover, trends in population dynamics are from rings in that transect (total grasshoppers counted in rings often better framed by examining population growth rate rather along transect divided by area sampled), and an overall average den- than actual population abundance (R rather than N)(Berryman, sity (and SE) for the site calculated by averaging estimates from all 1999; Ellner and Turchin, 1995; Royama, 1992; Turchin, 2003), raising transects. Densities for common grasshopper species were deter- the issue of whether both overall densities and per capita population mined by estimating relative abundances of taxa obtained from growth rates respond to the same factors in the same fashion. 200–400 sweep samples taken in this same area and multiplying Absolute population densities may be more important than growth overall density estimates by relative abundances (Evans et al., 1983). rates to managers concerned with immediate impact, but growth rates provide insight into population dynamics and long-term forecasting. Weather Data We examine the long-term dynamics of generalist feeding grass- hoppers in response to weather and density using a 25-year record of Weather data from 1978–2004 were obtained from the National grasshopper populations from continental North American grassland Oceanic and Atmospheric Administration—National Weather Service (Nebraska, USA). We assess relationships of weather variables (in- station at Arthur, Arthur County, Nebraska (station ID a250369; cluding time-lagged responses) to population dynamics of the nine 101°41′W, 41°34′N; elevation 1067 m). This weather station is locat- most abundant species. Because of the diverse life histories and ed 15 km from Arapaho Prairie. Average annual maximum tempera- food habits of these grasshoppers, each species could respond ture was 16.5 ± 0.1 °C with summer high temperatures of 29.2 ± uniquely to different aspects of weather. We also analyze groups of 0.1 °C and winter highs of 4.1 ± 0.5 °C (Fig. 1A). The amount and sea- species lumped by functional feeding groups because these are sonal timing of precipitation varied greatly among years. From the often used by rangeland managers to predict potential grasshopper NOAA database, 1978–2004 annual mean precipitation was outbreaks. 472.4 ± 18.5 mm, ranging from 244.9 (1989) to 710.4 (1993) mm J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39 31

Table 1 Weather and climatic variables used to build explanatory models in analyses of relationships between weather and grasshopper density and growth rate.

Abbreviation Variable description

CV GSPPT Coefficient of variation of growing season precipitation

CV GSPPTt-1 Coefficient of variation of growing season precipitationt-1

CV GSPPTt+1 Coefficient of variation of growing season precipitationt+1 CV GSTEMP Coefficient of variation of growing season temperature

CV GSTEMPt-1 Coefficient of variation of growing season temperaturet-1

CV GSTEMPt+1 Coefficient of variation of growing season temperaturet+1 DENSITY Density of given grasshopper species or guild

FALLPPT Fall (Septembert-1–Novembert-1) cumulative precipitation FALLTEMP

FALLTEMP Fall (Septembert-1–Novembert-1) mean temperature (C) GSHDD GSHDD Growing season heating degree days

GSHDDt-1 Growing season heating degree dayst-1

GSHDDt+1 Growing season heating degree dayst+1

GSPPT Growing season cumulative precipitation (mm) GSPPTt-1

GSPPTt-1 Growing season cumulative precipitationt-1 (mm) GSPPTt+1

GSPPTt+1 Growing season cumulative precipitationt+1 (mm)

NAO Mean North Atlantic Oscillation index (Januaryt–Marcht)

PDO Mean Pacific Decadal Oscillation index (Septembert-1–Augustt)

PDSI Mean Palmer Drought Severity Index (Septembert-1–Augustt) SOI

SOI Mean Southern Oscillation Index (Septembert-1–Augustt) WINTPPT

WINTPPT Winter (Decembert-1–Marcht) cumulative precipitation (mm)

WINTTEMP Winter (Decembert-1–Marcht) mean temperature (C) See Table S1 (available online at http://dx.doi.org/10.1016/j.rama.2014.12.011) for complete list of models constructed from these variables for use in analysis of grass- hopper densities and growth rates.

Statistical Analyses

Analyses were conducted on natural log densities [ln(Nt/m2)] and

population growth rates [R =ln(Nt+1/Nt)](Berryman, 1999; Turchin, 2003) of the nine most common species in addition to three combined categories (all acridids combined, grass-feeding acridids, and forb- plus mixed- feeding acridids). On the basis of in- spection of normal probability plots, the density and population Fig. 1. Growing season A, average maximum temperature (°C) and B, total precipita- growth rate data were judged to have normal distributions for all – tion (mm) and C, atmospheric oscillation indices from 1978 2003 at Arthur, Arthur species and combined groupings. County, NE, USA. SOI = Southern Oscillation Index, NAO = North Atlantic Oscillation Index, PDO = Pacific Decadal Oscillation Index, and PDSI = Palmer Drought Severity Correlation analyses were conducted to check for colinearity Index. Over this period mean growing season maximum temperature and precipita- among weather variables; the Dunn-Sidak alpha correction (Gotelli tion were 29.2 ± 0.1 °C and 321.10 ± 16.37 mm, respectively. and Ellison, 2004) was applied to account for multiple comparisons. None of the 21 weather variables (Table 1) were significantly corre- lated (α'=0.0003atα = 0.05, k = 210). We also checked for corre- (Fig. 1B). During this period, mean growing season and winter pre- lations among species densities and population growth rates, cipitation was 321.1 ± 16.4 mm and 26.5 ± 2.9 mm, respectively. applying the Dunn-Sidak correction for multiple comparisons (α'= Growing season (April–August) heating degree days (HDD) (sum 0.001 at α =0.05,k = 36). of temperatures on days with maximum temperature N20 °C) and We identified 29 and 37 models a priori for analyses of density coefficients of variation (CV) of both temperature and precipitation and growth rate data, respectively (Table S1, available online at were calculated. Data for the Tahiti-Darwin Southern Oscillation http://dx.doi.org/10.1016/j.rama.2014.12.011). These models Index (SOI) and Northern Atlantic Oscillation were downloaded encompassed seasonal weather patterns from prior (t-1) through from the National Center for Atmospheric Research Climate Analysis current year (t) growing seasons that may have direct or indirect ef- Section website (http://www.cgd.ucar.edu/cas/). Pacific Decadal Os- fects on grasshopper life cycle stages (egg production and laying, egg cillation (PDO) values were downloaded from the University of diapause, nymph, and adult survival) (Table S1). Because R is a mea- Washington Joint Institute for the Study of the Atmosphere and sure of population change from t to t+1, analysis of growth rate Oceans website (www.jisao.washington.edu/pdo/). Palmer Drought data included seasonal weather variables for periods t-1, t,and Severity Index (PDSI) values for Nebraska were downloaded from t+1(Table S1). the National Climate Data Center climate monitoring website The information-theoretic approach (Burnham and Anderson, (http://www.ncdc.noaa.gov/oa/climate/research/monitoring.html). 2002) based on Akaike’s Information Criterion corrected for small

Monthly values for SOI, NAO, PDO, and PDSI were averaged to pro- sample size (AICC) was used to assess the contribution of weather duce mean Septembert-1 through Augustt values for each year from to variation in densities, and of both weather and density to popula- 1978 to 2003 (Fig. 1C). This interval corresponds roughly to the life tion growth rates of grasshoppers from 1979–2003. Because these cycle of most grasshoppers from egg laying the fall before sampling datasets are time series, candidate models were fit to the data with (t-1) through maturation to adults in the year of sampling (t). lag 1 autoregression using PROC AUTOREG in SAS v. 9.3 (SAS 32 J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39

Fig. 2. Population densities [natural log (individuals · m−2)] over 25 years of A, all species grouped, B, grass-feeding guild, C, mixed-feeding guild, D, deorum (Ad), E, Phoetaliotes nebrascensis (Pn), F, Melanoplus sanguinipes (Ms), G, coloradus (Ac), H, Mermeria bivittata (Mb). I, Spharagemon collare (Sc), J, Melanoplus angustipennis (Ma), K, Melanoplus foedus (Mf), and L, Opeia obscura (Oo) from Arapaho Prairie (Arthur County, NE, USA). Different scales on the ordinate were used for different species.

Systems, Cary, NC, USA). The ΔAICCi was calculated as the difference specific population densities when the Dunn-Sidak multiple compar- between AICC of model i and that of the model with the lowest AICC ison α-adjustment was applied and only four comparisons had from the candidate set. Models with ΔAICC b 2 were considered to r N 0.5. No significant correlations were observed for species- be the best and equally plausible models, while models with specificpopulationgrowthrates(R) of the nine most common

ΔAICC N 2 were excluded from further analysis. A null model was in- species. Population density was important for understanding cluded in each candidate set; if the null model had ΔAICC b 2 for a species-specific R where population growth rates of all species given analysis, all models were considered ecologically irrelevant decreased significantly as density increased (P b 0.05 except for that response variable. Model weight (wi) based on AICC was , P =0.055)(Fig. 3). used to assess the performance of model i relative to all other models in the set. Model wi values range between 0 and 1; higher values in- Information Theoretic Analyses—Density dicate better performance of that model relative to other models test- ed. Since each explanatory variable appeared in three different The performance of 29 competing models of grasshopper density models in each candidate set, we also calculated relative importance was compared using AICC analysis (Tables 2, S2 available online at values (RIV) for each variable as the sum of wi from all models in [URL]). Plausible models (Table 2) and importance of weather vari- which the given variable occurred (Burnham and Anderson, 2002). ables (Table 3) differed among grasshopper functional groups and

species. The null model was among those with ΔAICC b 2 for the Results total and mixed-feeding groups, as well as for , Melanoplus angustipennis, Mel. foedus, and Spharagemon collare Population Trends density (S2). Variability of prior growing season precipitation

(CV GSPPTt-1) was the only plausible model for densities of the Patterns of population density and growth rates from 1979–2003 grass-feeding functional group, Mer. bivittata, and Phoetaliotes varied among the nine most common acridid species at Arapaho Prai- nebrascensis (Table 2) accounting for approximately 21%, 25%, and rie (Fig. 2). A range of population abundance patterns was observed. 42% of interannual variability (Fig. 4A). Climatic indices and There were no significant pairwise correlations between species- winter temperature accounted for the three plausible models J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39 33

Fig. 3. Regressions of population growth rate (R) as a function of density [natural log (individuals · m−2)] for grasshopper groups (A, all species grouped B, grass-feeding guild, C, mixed- feeding guild) and individual species (D–L). Species abbreviations as for Fig. 2.

Table 2

AICC statistics for plausible models for density of grasshopper A, functional groups and B, species.

2 Model variables AICC LL R Δ AICC wi of Amphitoruns coloradus density (Table 2B). The North Atlantic Oscil- lation index was the most important single variable (RIV = 0.29, A Feeding guilds fi Grass-feeding species Table S3) and, when combined with the Paci c Decadal Oscillation

CV GSPPTt-1 (+) 43.67 −18.2 0.21 0.00 0.30 (PDO) and Southern Oscillation Index (SOI), accounted for 32% of B Species A. coloradus density (Table 2B, Fig. 4B). Melanoplus sanguinipes densi- ty was negatively associated with fall temperature (Table 2B, Fig. 4C). NAO (−) 80.04 −36.4 0.18 0.00 0.18 The models with prior year density (density ) and density plus CV WINTTEMP (−) 81.34 −37.1 0.13 1.30 0.10 t-1 t-1 NAO (−) PDO (+) SOI (−) 81.68 −34.2 0.32 1.64 0.08 GSPPTt-1 and prior intraseasonal variability in temperature (CV Melanoplus sanguinipes GSTEMPt-1) provided the best fittoOpeia obscura density (Table 2B, FALLTEMP (−) 60.72 −26.8 0.30 0.00 0.58 Fig. 4D); densityt-1 was at least 50% more important than any weath- Mermiria bivittata er variable (Table S3 available online at [URL]) and alone had R2 = CV GSPPTt-1 (+) 77.73 −35.3 0.25 0.00 0.32 Opeia obscura 0.92 (Table 2B).

CVGSPPTt-1 (−) CVGSTEMPt-1 (+)

DENSITYt-1 (+) 64.75 −25.7 0.95 0.00 0.61

DENSITYt-1 (+) 65.98 −29.4 0.92 1.23 0.33 Information Theoretic Analyses—Population Growth Rates (R) Phoetaliotes nebrascensis − CV GSPPTt-1 (+) 67.16 30.0 0.42 0.00 0.80 We assessed the relative likelihood of 37 competing models using 2 fi AICC = AIC corrected for small sample size, LL = log likelihood, R =regressioncoef - AICC criteria to explain R (Tables 4, S4 available online at [URL]). Gen- cient, ΔAICC = difference in AICC between given model and best model, wi = model erally, densityt alone, CV GSPPTt-1 alone, or the densityt, CV GSPPTt-1, weight. All models with ΔAICC b 2 are shown for each grouping. If the null model and CV GSTEMPt-1 model best explained trends in R for 9 of 12 spe- had ΔAICC b 2, then none of the candidate models were ecologically meaningful. See Table S2 (available online at http://dx.doi.org/10.1016/j.rama.2014.12.011) for com- cies/groups (Table 3). In most cases, R decreased as densityt (Fig. 3), plete results from all density analyses. CV GSPPTt-1 (Fig. 5), and CV GSTEMPt-1 increased. Although CV 34 J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39

Table 3 variables) and endogenous (densities) factors on grasshopper popu- AICC statistics for plausible models for growth rates (R) of grasshopper A, functional lations, as well as obtain ecological insights into the nature of the un- groups and B, species. derlying contributions of these drivers to interannual population 2 Model variables AICC LL R Δ wi dynamics, including differences among common species. Models AIC C with a combination of weather-dependent and density variables fit A Feeding guilds grasshopper population responses from a single site over 25 years. All species This is similar to the finding for locusts in a 1 900-year-long dataset − − CV GSPPTt-1 ( ) 37.62 15.2 0.52 0.00 0.65 from China, where the model containing previous year abundance, CV GSPPT (−) CV GSTEMP (+) 39.19 −12.9 0.60 1.57 0.30 t-1 t-1 temperature, and precipitation accounted for nearly 50% of current DENSITYt (−) Grass-feeding species year locust abundance (Tian et al., 2011). However, species-specific − − CV GSPPTt-1 ( ) 31.59 12.2 0.63 0.00 0.80 responses were quite distinct in our analysis. Some grasshopper Mixed-feeding species and locust species are known to exhibit seemingly regular periods CV GSPPT (−) 59.51 −26.2 0.29 0.00 0.23 t-1 of high and low densities over variable periods of about 10–15 CV GSPPTt-1 (−) CV GSTEMPt-1 (−) 59.80 −23.2 0.45 0.29 0.20 years in length (Branson et al., 2006; Capinera, 1987; Criddle, 1932; DENSITYt (−) PDO (−) 60.23 −26.5 0.32 0.72 0.16 Joern and Gaines, 1990; Jonas and Joern, 2007; Smith, 1954), while − − − CV GSPPTt-1 ( ) CV GSTEMPt-1 ( ) 60.47 21.8 0.54 0.96 0.14 many other grasshopper species show a range of different responses, NAO including minimal fluctuations among years, regular short-term cy- (+) PDO (−) B Species cling, or long-term increasing or decreasing trends. We document Ageneotettix deorum the same range of trends among species. Moreover, our observation − − DENSITYt ( ) 62.05 27.4 0.29 0.00 0.22 that species-specific dynamics differed greatly from one another is − − CV GSPPTt-1 ( ) 62.15 27.5 0.23 0.10 0.21 different from previous North American studies that documented CV GSPPT (−) CV GSTEMP (−) 62.38 −24.5 0.43 0.32 0.19 t-1 t-1 significant correlations among population abundances of grasshop- DENSITYt (−) Amphitornus coloradus per species (Fielding and Brusven, 1990; Gage and Mukerji, 1977).

DENSITYt (−) 83.30 −38.1 0.28 0.00 0.28 Analyses used estimates of actual densities (i.e., number of Melanoplus angustipennis individuals · m−2). Long-term records of actual grasshopper densi- − − − CV GSPPTt-1 ( ) CV GSTEMPt-1 ( ) 65.77 26.2 0.57 0.00 0.70 ties are rare as surrogate estimates of abundance (e.g., annual esti- DENSITYt (−) Melanoplus foedus mates of area requiring chemical-control) are more often used in

DENSITYt (−) 99.10 −46.0 0.37 0.00 0.61 time series analyses of rangeland grasshoppers. Although surrogate Melanoplus sanguinipes measures are useful to detect trends when there are no other op- − CVGSTEMPt+1 (+) 67.27 30.0 0.29 0.00 0.24 tions, their use may be misleading and possibly lead to incorrect or − CVGSTEMP (+) 69.21 31.0 0.20 1.93 0.09 incomplete inferences as they weaken or even obscure the signal be- Mermiria bivittata WINTPPT (−) 76.17 −34.5 0.29 0.00 0.31 tween weather and abundance with respect to mechanisms and Opeia obscura density-dependent effects. Because density is a fundamental ecolog- − GSHDDt-1 (+) 63.86 28.3 0.37 0.00 0.33 ical unit that underpins most problems in population ecology, fore- Phoetaliotes nebrascensis casting efforts should use it whenever possible. We also used an CV GSPPTt-1 (−) 67.09 −30.0 0.43 0.00 0.72 Spharagemon collare information-theoretic (IT) approach rather than traditional multiple

DENSITYt (−) 96.57 −44.7 0.36 0.00 0.37 regression methods to assess the relative contributions of a range of WINTPPT (−) 97.59 −45.2 0.24 1.02 0.22 variables identified a priori to population abundances. This approach 2 allowed us to step back and develop consensus models on the basis AICC = AIC corrected for small sample size, LL = log likelihood, R =regressioncoeffi- cient, ΔAICC = difference in AICC between given model and best model, wi = model of available weather variables, at least in terms of which factors are Δ b weight. All models with AICC 2 are shown for each grouping. If the null model repeatedly included in the top models (not just the single best Δ b had AICC 2, then none of the candidate models were ecologically meaningful. See model) and the relative importance of factors compared with one Table S3 (available online at http://dx.doi.org/10.1016/j.rama.2014.12.011) for com- plete results from all growth rate analyses. another. One rule of thumb for assessing grasshoppers in northern North GSTEMPt-1 tended to have the highest RIV after CV GSPPTt-1 American grasslands is that hot, dry weather conditions persisting 2 (Table S4), its R was b 0.001 to 0.01 alone (Table S4) and it showed over several years favor high densities or outbreaks (Edwards, the opposite (positive) relationship to R for the all species group 1960; Gage and Mukerji, 1977). However, this conclusion is not gen- (Table 4A). For the mixed-feeding group, a negative relationship to eral (Branson et al., 2006; Capinera, 1987; Capinera and Horton, PDO was also indicated among plausible models (Table 4A, Fig. 5B). 1989; Dempster, 1963; Joern and Gaines, 1990). Often lost in such Decadal climate indices were not included in the top models for discussions is the critical frame of reference as changes in actual den- explaining R of any other group or species. Mel. sanguinipes increased sities, and population growth rates are used interchangeably. The with CV GSTEMPt and CV GSTEMPt+1 (Table 4B, Fig. 5D), while need for this distinction has been emphasized in times series analy- O. obscura was positively related to prior growing season heating de- ses (Berryman, 1999; Royama, 1992; Turchin, 2003). Using the IT ap- gree days (GS HDD t-1)(Table 4B, Fig. 5E). There was a negative rela- proach, we examined multiple models for fitting grasshopper tionship between winter precipitation and Mer. bivittata (Table 4B, population dynamics in continental North American grassland Fig. 5F). Winter precipitation was also a plausible model for experiencing variable weather. S. collare, but RIV of densityt was nearly two times higher (Table S3). Variation in Density Discussion Overall, weather variables contributed significantly to patterns of Weather has been historically considered a major driver of insect variation in densities of several groups or species, and large-scale de- population dynamics, including grasshoppers (Andrewartha and cadal relationships were prominent for one of them (Tables 2,S2).An 2 Birch, 1954; Dempster, 1963; Uvarov, 1931). Our aim was to under- exception is O. obscura, where densityt-1 alone had R = 0.92. How- stand the importance of combinations of exogenous (weather ever, data for this species are the least reliable because of generally J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39 35

Fig. 4. Relationships between densities of grasshoppers or functional groups [natural log (individuals · m−2)] in relation to weather variables or climatic indices: A, Variability in prior growing season precipitation, CV GSPPTt-1, B, North Atlantic Oscillation index, NAO, C, fall temperature, and D, prior year density, densityt-1. Species abbreviations as for Fig. 2. low population densities at the edge of detection with the sampling White (2007, 2008) revisits the contentious issue about the role methods and effort used. The null model was among those identified of weather in limiting and regulating consumer abundances. He con- as plausible for all species and mixed-feeding groups and four species cludes that the availability of food above a significant quality thresh- (A. deorum, Mel. angustipennis, Mel. foedus, S. collare)indicatingthat old determines final densities of populations, and that the amount of none of our candidate models were particularly meaningful with re- food available is determined by weather conditions during the grow- gard to densities of these taxa. ing season. Within-season variability of precipitation in the previous

Table 4 Relative importance values (RIV) of variables compiled across all competing models in the A, density and B, growth rate analyses.

A Density B Growth rate

Explanatory variable All species Grass- feeding species Mixed- feeding species All species Grass- feeding species Mixed- feeding species CV GSPPT 0.029 0.031 0.030 11 0.006 33 3 11 CV GSPPTt+1 0.008

CV GSPPTt-1 0.029 0.317 0.059 0.990 0.999 0.574 CV GSTEMP 0.093 0.029 0.031 11 0.006 33 3 11 CV GSTEMPt+1 0.008

CV GSTEMPt-1 0.045 0.040 0.036 0.344 0.202 0.346 DENSITY2 0.041 0.059 0.123 0.299 0.185 0.283 FALLPPT 0.073 0.028 0.027 11 0.004 FALLTEMP 0.033 0.024 0.061 11 0.014 GSHDD 0.053 0.024 0.028 11 0.004 33 3 11 GSHDDt+1 0.005 11 GSHDDt-1 0.044 0.039 0.048 0.007 33 3 11 GSPPTt+1 0.012 GSPPT 0.028 0.050 0.051 11 0.006 1 GSPPTt-1 0.035 0.029 0.043 0.001 0.007 NAO 0.029 0.053 0.080 0.049 0.018 0.174 PDO 0.084 0.041 0.069 0.050 0.018 0.329 PDSI 0.042 0.046 0.051 0.001 1 0.037 SOI 0.028 0.064 0.137 0.001 1 0.039 WINTPPT 0.188 0.027 0.027 11 0.004 WINTTEMP 0.030 0.034 0.030 11 0.005

Values calculated for a given variable as the sum of Akiake’s weights (wi) for all models in which that variable appeared (Burnham and Anderson, 2002). The highest RIV for each group is in bold. See Table S4 (available online at http://dx.doi.org/10.1016/j.rama.2014.12.011) for RIV results of individual species. 1 b0.001. 2 Density explanatory variable at t-1 for Density response variables and t for Growth Rate response variables. 3 t + 1 explanatory variables not relevant to Density response variables. 36 J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39

Fig. 5. Population growth rates (R) of grasshopper functional groups in response to A, variability of prior growing season precipitation, CV GSPPTt-1 and B, Pacific Decadal Oscillation,

PDO. Responses of individual species population growth rates to C, CV GSPPTt-1, D, variability of prior growing season temperature, CV GSTEMPt-1, E, winter precipitation, and F, prior growing season heating degree days, GS HDDt-1. Species abbreviations as for Fig. 2.

year (CV GSPPTt-1) was positively related to density of the grass- Variation in Growth Rate feeding group and two late-hatching grass-feeding species (Mer. bivittata, P. nebrascensis)(Pfadt, 1994)(Table 2). It is also wor- Understanding population dynamics and regulation from an eco- thy to note that a similar positive relationship was found between logical standpoint is best approached by characterizing density- abundance of late-season grasshoppers and CV GSPPTt-1 in Kansas dependent changes in population growth rate, R (Berryman, 1999; tallgrass prairie (Jonas and Joern, 2007). Although the mechanism Royama, 1992; Turchin, 2003). Negative relationships are expected could not be tested here, such a lagged effect likely reflects for density-dependent regulation. All species examined here showed weather-related effects on host plant quantity and quality available negative relationships between R and N, although much variation to the parental generation. For example, Branson (2008) found was not explained in all cases. In the two species where Densityt by that a late-season precipitation event had no effect on grass bio- itself and a single weather variable were among plausible models, mass but increased grass quality, which was associated with Densityt explained 29–36% of the variability in R while the weather decreased apparent resource limitation and dampened density- variable accounted for 23–24%. dependent effects on survival and fecundity in P. nebrascensis It is noteworthy that (1) the null model was never among plausi- and Mel. sanguinipes. Consistent responses to such variables ble models for R and (2) the best models for explaining variation in R support White’s conclusions that weather-dependent variability among years differed from models that best explained changes in in food quality is critical for understanding final densities, a density for most species. Despite different explanatory models, the result of great interest to grazing-land managers, and warrants amount of variability explained by weather was similar for density further investigation. and R in most species and plausible results were seen under both J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39 37 response variables. This suggests that weather is equally important between grasshopper population responses and climate indices for understanding both metrics of grasshopper populations. Again, cannot be drawn here, an increasing number of population studies O. obscura was the exception to this pattern. can now be linked to large-scale climate variability (Stenseth and Population growth rates were most closely related to aspects of Mysterud, 2005; White, 2008). For example, Stige et al. (2007) precipitation and/or temperature during the growing season of the found that decadal-scale locust abundance in China was associat- previous year for 5 of the 9 species examined. As was discussed for ed with frequency of floods and droughts using a 1 000-yr long density, these lagged effects of temperature and precipitation most dataset. They hypothesize that the pattern was likely driven by a likely act on population growth rates by altering host plant nutrition- longer-term large-scale climatic oscillation influencing mon- al quality available to the parental generation of grasshoppers. In ad- soons. These findings suggest the existence of unrecognized eco- dition, Jonas et al. (in review) found that prior growing season logical processes and opportunities for better understanding weather was associated with changes in native plant diversity in a long-term population dynamics. Colorado shortgrass prairie and Kansas tall- and mixed-grass prairies. This further suggests the importance of weather acting on grasshop- Implications per populations indirectly through effects of altered species composi- tion on host plant availability or quality. In the two cases where the Grasshoppers are usually among the most important group of same weather variable was important (i.e., among top models and grassland insect herbivores prone to outbreak in western US grass- highest RIV) for both density and growth rates (CV GSPPTt-1 for lands, in some cases causing severe defoliation of neighboring agri- both the grass-feeding guild and P. nebrascensis), the relationship cultural fields and pastures (Hewitt and Onsager, 1983; Joern and was positive for density and negative for growth rate. These seeming- Gaines, 1990; Cunningham and Sampson, 2000; Onsager, 2000; ly disparate results are likely related to the combined influences of Branson et al., 2006;). Because of this, rangeland managers need to density-dependence (biotic regulation) and weather (abiotic regula- know about trends in actual grasshopper densities so that grasshop- tion) acting in concert on growth rates. It is also interesting that per management activities can be planned in advance. Plant defolia- although slopes differed among grasshopper functional groups and tion due to grasshoppers can be economically significant (Hewitt and species, the point at which R went from increasing to decreasing Onsager, 1983; Meyer et al., 2002; Mitchell and Pfadt, 1974), with es- populations was at approximately the same value of CV GSPPTt-1 timates of forage loss of N$1 billion US or more each year (Branson (Fig. 5A,C). The ecological significance of this value is unclear, though et al., 2006), leading to significant competition with dominant ungu- it may signal a plant quality threshold. Winter precipitation was asso- late grazers (especially cattle) (Onsager, 2000). Grazing-land man- ciated with decreasing populations of Mer. bivittata and S. collare, agers historically employed large-scale chemical control during which suggests that soil moisture may exert important controls on outbreak periods to reduce densities (Cunningham and Sampson, population dynamics of these species through a variety of mecha- 2000), an approach that also incurs a significant economic and envi- nisms acting at the egg stage (e.g., rate of development, survival, sus- ronmental burden. Recent approaches seek to significantly reduce ceptibility to pathogens). However, due to limited understanding of (or eliminate) pesticide application rates and indiscriminate use of environmental controls over demographic processes in these partic- exotic biological control agents over broad areas (Cunningham and ular species, it is difficult to speculate on mechanisms driving such Sampson, 2000; Lockwood, 1993; Lockwood et al., 2000), but better species-specific responses to different weather variables. forecasting information is needed to succeed in these efforts. Under- standing the relative influences of density-dependent and weather- Decadal Climate Oscillations influenced factors on grasshopper species most prone to outbreak is an important first step in moving grasshopper control programs to-

On the basis of AICC criteria, the three decadal climate models ward a more preventative rather than reactive focus when it comes (NAO, PDO, SOI) examined were successful predictors of to managing outbreaks. On the basis of the results presented here, A. coloradus density, although NAO was strongest. Density of similar investigations in other regions and under different manage- A. coloradus tended to be highest during the negative phase of NAO, ment regimens are warranted to assess the generality of our findings. which is primarily associated with colder and to a lesser extent Such insights are needed for all types of intervention, whether direct- drier winters (Hurrell et al., 2003; Ottersen et al., 2001). Given that ly through chemical intervention or indirectly through vegetation winter temperature was also identified as a plausible model and management activities (Branson et al., 2006), in order to reinforce that A. coloradus density was negatively related to it, it is likely that those that act in concert with natural mechanisms that limit grass- this species is particularly responsive to the thermal environment hopper population abundances. during the egg or hatchling stage. Past attempts at predicting grasshopper responses for developing The PDO index was among plausible models for mixed-feeding management activities were primarily based on counts either in the grasshopper population growth rates. The relationship indicates current or previous year, approaches that provide little lead time that during the peak of the positive phase when winter temperatures for planning. Some managers advocated the need to incorporate tend to be higher than average and precipitation lower than average changes in density over multiple years, although data were seldom in northern North America (Mantua, 2002), mixed-feeding grasshop- available over large areas that historically experienced grasshopper pers as a group decline. The effect of PDO on North American weather outbreaks. Ultimately, predictions from loose rules of thumb such patterns is similar to that of the NAO, but on a 40- to 60-yr time scale as the existence of 10-yr cycles may be useful for planning if underly- instead of 10- to 20-yr cycle (Hurrell, 1995; Mantua, 2002). There- ing mechanisms can be identified to explain the periodicity. We fore, this relationship between mixed-feeders and PDO is likely sim- identify significant weather patterns that underlie grasshopper re- ilar to that between mixed-feeders and NAO in Kansas tallgrass sponses, making them useful for planning long-term forecasting prairie shown by Jonas and Joern (2007). and management goals. Decadal climate indices influence broad geographic areas over variable lengths of time (Stenseth and Mysterud, 2005), and dif- Acknowledgments ferent oscillations may act in phase with one another (Latif and Barnett, 1996; Mantua et al., 1997; Stenseth and Mysterud, Arapaho Prairie is owned by the Nature Conservancy and man- 2005; Zhang et al., 1997). While the exact mechanistic linkages aged by the School of Biological Sciences at the University of 38 J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39

Nebraska–Lincoln. We are grateful for extensive, long-term logistical Hurrell, J.W., 1995. Decadal trends in the North Atlantic Oscillation: regional temper- – – atures and precipitation. Science 269, 676 679. support by Cedar Point Biological Station (University of Nebraska Hurrell, J.W., Kushnir, Y., Ottersen, G., Visbeck, M. (Eds.), 2003. The North Atlantic Lincoln). Many field assistants provided invaluable help in this pro- Oscillation: climatic significance and environmental impact. Geophysical Mono- ject over the years. A. Laws and S. Parsons and anonymous reviewers graph Series Volume 134. American Geophysical Union (279 pp.). Joern, A., 1983. Host plant utilization by grasshoppers (: Acrididae) from a provided constructive comments on the manuscript. sandhills prairie. Journal of Range Management 36, 793–797. Joern, A., Gaines, S.B., 1990. Population dynamics and regulation in grasshoppers. In: Chapman, R.F., Joern, A. (Eds.), Biology of grasshoppers. John Wiley & Sons, New Appendix A. Supplementary data York, pp. 415–482. Joern, A., Mole, S., 2005. The plant stress hypothesis and variable responses by blue Supplementary data to this article can be found online at http:// grama grass (Bouteloua gracilis) to water, mineral nitrogen and insect herbivory. Journal of Chemical Ecology 31, 2069–2090. dx.doi.org/10.1016/j.rama.2014.12.011. Johnson, D.L., Worobec, A., 1988. Spatial and temporal computer analysis of insects and weather: grasshoppers and rainfall in Alberta. Memoirs Entomological Socie- ty Canada 146, 33–46. References Jonas, J.L., Joern, A., 2007. Grasshopper (Orthoptera: Acrididae) communities respond to fire, bison grazing and weather in North American tallgrass prairie: a long-term Andrewartha, H.G., Birch, L.C., 1954. The distribution and abundance of . study. Oecologia 153, 699–711. University of Chicago Press, Chicago, IL, USA. Jonas, J. L., A. Symstad, and D. Buhl. in review. Weather, management, and location im- Barbosa, P., Schultz, J.C. (Eds.), 1987. Insect outbreaks. Academic Press, New York, NY, pact long-term patterns of plant richness and diversity in Great Plains grasslands. USA. Ecology. Barnes, P.W., Harrison, A.T., Heinisch, S.P., 1984. Vegetation patterns in relation to to- Jones, C.G., Coleman, J.S., 1991. Plant stress and insect herbivory: toward an integrated pography and edaphic variation in Nebraska Sandhills Prairie. The Prairie Natural- perspective. In: Mooney, H.A., Winner, W.E., Pell, E.J. (Eds.), Response of plants to ist 16, 145–158. multiple stresses. Academic Press, San Diego, CA, USA, pp. 249–280. Begon, M., 1983. Grasshopper populations and weather: the effects of insolation on Kingsolver, J.G., 1989. Weather and population dynamics if insects: integrating phys- brunneus. Ecological Entomology 8, 361–370. iological and population ecology. Physiological Zoology 62, 314–334. Belovsky, G.E., Joern, A., 1995. Regulation of grassland grasshoppers: differing domi- Latif, M., Barnett, T.P., 1996. Decadal climate variability over the North Pacific and nant mechanisms in time and space. In: Cappucino, N., Price, P.W. (Eds.), Novel North America: dynamics and predictability. Journal of Climate 9, 2401–2432. approaches for the study of population dynamics: examples from insect herbi- Lockwood, J.A., 1993. Environmental issues involved in biological control of rangeland vores. Academic Press, New York, NY, pp. 359–386. grasshoppers (Orthoptera, Acrididae) with exotic agents. Environmental Ento- Berryman, A.A., 1999. Principles of population dynamics. Stanley Thornes Publishers, mology 22, 503–518. Ltd., Cheltenham, UK. Lockwood, J.A., Latchininsky, A.V., Sergeev, M. (Eds.), 2000. Grasshoppers and grass- Branson, D.H., 2008. Effects of a large late summer precipitation event on food limita- land health. Kluwer Academic Publishers, Dordrecht, The Netherlands. tion and grasshopper population dynamics in a northern Great Plains grassland. Logan, J.A., Powell, J.A., 2001. Ghost forests, global warming, and the mountain pine Environmental Entomology 37, 686–695. beetle. American Entomologist 47, 160–172. Branson, D.H., Joern, A., Sword, G.A., 2006. Sustainable management of insect herbi- Louda, S.M., Collinge, S.K., 1992. Plant resistance to insect herbivores: a field test of the vores in grassland ecosystems: new perspectives in grasshopper control. Biosci- environmental stress hypothesis. Ecology 73, 153–169. ence 56, 743–755. Mantua, N.J., 2002. Pacific-Decadal Oscillation (PDO). In: MacCracken, M.C., Perry, J.S. Burnham, K.P., Anderson, D.R., 2002. Model selection and multimodel inference: a (Eds.), Encyclopedia of global environmental change. Wiley & Sons, Chichester, UK, practical information-theoretic approach. Springer, New York, NY, USA. pp. 592–594. Capinera, J.C., 1987. Population ecology of rangeland grasshoppers. In: Capinera, J.C. Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M., Francis, R.C., 1997. A Pacific (Ed.), Integrated pest management on rangeland: a shortgrass perspective. interdecadal climate oscillation with impacts on salmon production. Bulletin of Westview Press, Boulder, CO, pp. 162–182. the American Meteorological Society 78, 1069–1079. Capinera, J.C., Horton, D.R., 1989. Geographic variation in effects of weather on grass- Mattson, W.J., Haack, R.A., 1987. The role of drought in outbreaks of plant-eating in- hopper infestation. Environmental Entomology 18, 8–14. sects. Bioscience 37, 110–118. Cappuccino, N., Price, P.W. (Eds.), 1995. Population dynamics: new approaches and Meyer, C.K., Whiles, M.R., Charleton, R.E., 2002. Life history, secondary production, and syntheses. Academic Press, San Diego, CA, USA. ecosystem significance of acridid grasshoppers in annually burned and unburned Casey, T.M., 1993. Effects of temperature on foraging of caterpillars. In: Stamp, N.E., tallgrass prairie. American Entomologist 48, 40–49. Casey, T.M. (Eds.), Caterpillars: ecological and evolutionary constraints on forag- Mitchell, J.E., Pfadt, R.E., 1974. A role of grasshoppers in shortgrass prairie ecosystem. ing. Chapman & Hall, New York, NY, USA. Environmental Entomology 3, 358–360. Criddle, N., 1932. The correlation of sunspot periodicity with grasshopper fluctuation Nothnagle, P.J., Schultz, J.C., 1987. What is a forest pest? In: Barbosa, P., Schultz, J.C. in Manitoba. Canadian Field Naturalist 46, 195–199. (Eds.), Insect outbreaks. Academic Press, San Diego, CA, USA, pp. 59–80 Cunningham, G.L., Sampson, M.W., 2000. Grasshopper integrated pest management Onsager, J.A., 1977. Comparison of five methods for estimating density of rangeland user handbook. USDA/APHIS Technical Bulletin 1809. grasshoppers. Journal of Economic Entomology 70, 187–190. Danner, B.J., Joern, A., 2003. Food-resource mediated impact of predation risk on per- Onsager, J.A., 2000. Suppression of grasshoppers in the Great Plains through grazing formance in the grasshopper Ageneotettix deorum (Orthoptera). Oecologia 137, management. Journal of Range Management 53, 592–602. 352–359. Ottersen, G., Planque, B., Belgrano, A., Post, E., Reid, P.C., Stenseth, N.C., 2001. Ecological Danner, B.J., Joern, A., 2004. Development, growth and egg production of the common effects of the North Atlantic Oscillation. Oecologia 128, 1–14. grasshopper, Ageneotettix deorum (Orthoptera: Acrididae) in response to indirect Ovadia, O., Schmitz, O.J., 2004. Weather variation and trophic interaction strength: risk of spider predation. Ecological Entomology 29, 1–11. sorting the signal from the noise. Oecologia 140, 398–406. Dempster, J.P., 1963. The population dynamics of grasshoppers and locusts. Biological Pfadt, R.E., 1994. Grasshopper species fact sheets. Wyoming Agricultural Experiment Reviews 38, 490–529. Station, Bulletin 912. Denno, R.F., McClure, M.S. (Eds.), 1983. Variable plants and herbivores in natural and Post, E., Forchhammer, M.C., 2008. Climate change reduces reproductive success of an managed systems. Academic Press, New York, NY, USA. Arctic through trophic mismatch. Philosophical Transactions of the Edwards, R.L., 1960. Relationship between grasshopper abundance and weather con- Royal Society London (B) 363, 2369–2375. ditions in Saskatchewan, 1930–1958. Canadian Entomologist 92, 619–624. Ritchie, M.E., 1996. Interaction of temperature and resources in population dynamics Ellner, S., Turchin, P., 1995. Chaos in a noisy world: new methods and evidence from and experimental test of theory. In: Floyd, R.B., Shepard, A.W., DeBarro, P.J. time- series analysis. American Naturalist 145, 343–375. (Eds.), Frontiers of Population Ecology. CSIRO Publishing, Melbourne, Australia, Evans, E.W., Rogers, R.A., Opfermann, D.J., 1983. Sampling grasshoppers (Orthoptera: pp. 79–91. Acridide) on burned and unburned tallgrass prairie: night trapping vs. sweeping. Ritchie, M.E., 2000. Nitrogen limitation and trophic vs. abiotic influences on insect her- Environmental Entomology 12, 1449–1454. bivores in a temperate grassland. Ecology 81, 1601–1612. Fielding, D.J., Brusven, M.A., 1990. Historical analysis of grasshopper (Orthoptera: Royama, T., 1992. Analytical population dynamics. Chapman & Hall, London, UK. Acrididae) population responses to climate in southern Idaho, 1950–1980. Envi- Smith, R.C., 1954. An analysis of 100 years of grasshopper populations in Kansas ronmental Entomology 19, 1786–1791. (1854–1954). Transactions of the Kansas Academy of Science 57, 397–433. Gage, S.H., Mukerji, M.K., 1977. A perspective of grasshopper population distribution in Stenseth, N.C., Mysterud, A., 2005. Weather packages: finding the right scale and com- Saskatchewan and interrelationship with weather. Environmental Entomology 6, position of climate in ecology. Journal of Ecology 74, 1195–1198. 469–479. Stenseth, N.C., Mysterud, A., Ottersen, G., Hurrell, J.W., Chan, K.S., Lima, M., 2002. Eco- Gotelli, N.J., Ellison, A.M., 2004. A primer of ecological statistics. Sinauer, MA, USA, Sun- logical effects of climate fluctuations. Science 297, 1292–1296. derland (510 pp.). Stige, L.C., Chan, K.-S., Zhang, Z., Frank, D.A., Stenseth, N.C., 2007. Thousand-year-long Gratton, C., Denno, R.F., 2003. Seasonal shift from bottom-up to top-down impact in Chinese time series reveals climatic forcing of decadal locust dynamics. Proceed- phytophagous insect populations. Oecologia 134, 487–495. ings of the National Academy of Science (USA) 104, 1618–16193. Hewitt, G.B., Onsager, J.A., 1983. Control of grasshoppers on rangeland in the United Tian, H., Stige, L.C., Cazelles, B., Kausrud, K.L., Svarverud, R., Stenseth, N.C., Zhang, Z., States—a perspective. Journal of Range Management 36, 202–207. 2011. Reconstruction of a 1 910-y-long locust series reveals consistent J.L. Jonas et al. / Rangeland Ecology & Management 68 (2015) 29–39 39

associations with climate fluctuations in China. Proceedings of the National Acad- White, T.C.R., 2007. Resolving the limitation-regulation debate. Ecological Research 22, emy of Science (USA) 108, 14521–14526. 354–357. Turchin, P., 2003. Complex population dynamics: a theoretical/empirical synthesis. White, T.C.R., 2008. The role of food, weather and climate in limiting the abundance of Princeton University Press, Princeton, NJ. animals. Biological Reviews 83, 227–248. Uvarov, B.P., 1931. Insects and climate. Transactions of the Entomological Society of Yang, Y., Joern, A., 1994. Influence of diet, developmental stage and temperature on London 79, 1–247. food residence time. Physiological Zoology 67, 598–616. White, T.C.R., 1993. The inadequate environment: nitrogen and the abundance of an- Zhang, Y., Wallace, J.M., Battisti, D.S., 1997. ENSO-like interdecadal variability: 1900– imals. Springer-Verlag, Berlin, Germany. 93. Journal of Climate 10, 1004–1020.