Orthoptera: Acrididae) at High Latitudes

Home , Melanoplus, Melanoplus sanguinipes

PHYSIOLOGICAL ECOLOGY Developmental Time of Melanoplus sanguinipes (Orthoptera: Acrididae) at High Latitudes

DENNIS J. FIELDING

Subarctic Agricultural Research Unit, USDAÐARS, University of Alaska Fairbanks, P.O. Box 757200, Fairbanks, AK 99775

Environ. Entomol. 33(6): 1513Ð1522 (2004) ABSTRACT Rates of development of Melanoplus sanguinipes F. from Alaska were determined at eight constant temperatures between 21 and 42ЊC. Diurnally alternating temperatures were used to estimate rates of development at temperatures too low for nymphs to complete development under constant temperatures. Two previously published equations were Þt to these data and used to predict developmental rates as a function of temperature. The Þrst equation, nonlinear and only approaching zero rate of development asymptotically, was Þt to the data from constant and alternating temperature treatments. The second equation, which included an intercept, was Þt to the data from constant temperature treatments only. Estimated developmental times based on these equations were tested against observed developmental times in two ßuctuating temperature regimes in growth chambers, and two seasons of Þeld sampling. Elevation of body temperature of grasshoppers above ambient temperatures in the Þeld was modeled as a linear function of solar irradiance, based on Þeld measurements. The effect of behavioral thermoregulation on nymphal developmental times in the Þeld was estimated using standard air temperatures and solar-adjusted temperatures. Rates of development under most of the constant temperature treatments were higher than any previously published for M. sanguinipes. Estimated rates of development in the Þeld using air temperatures only were about one-half those using solar-adjusted temperatures. Observed developmental times in the Þeld were 45 and 42 d in 2000 and 2002, respectively. Because of local adaptation and behavioral thermoregulation, duration of nymphal stages in M. sanguinipes is relatively independent of latitude.

KEY WORDS thermoregulation, subarctic, phenological model

THE GRASSHOPPER, Melanoplus sanguinipes F., has a very (Roff 1992), and temperature is the primary factor wide geographical range (Hebard 1929, Vickery and determining developmental rates in insects. Accurate Kevan 1983), including subarctic regions where it is a and reliable phenological models are critical for many sporadic, but potentially very damaging, pest of small applications: prediction of geographic ranges of inva- grain and vegetable crops. This species of grasshopper sive pests or proposed biocontrol agents (Bradshaw et must overwinter as eggs, and the short summers at al. 2000, Regniere and Nealis 2002); forensic investi- high latitudes allow only a limited amount of time for gations (Smith 1986, Greenberg 2002); prediction of nymphs to complete development, and the adults to the effects of climatic change (Logan and Powell 2001, reproduce. Total mean annual degree-day (DD) ac- Bryant et al. 2002); analysis of life history strategies cumulation for Fairbanks, AK, is 80 and 510, for base (Ayres and Scriber 1994, Bradford and Roff 1997); temperatures of 17.8 and 10ЊC, respectively. For com- development of realistic simulation models (Logan parison, in Montana, Kemp and Onsager (1986) esti- and Bentz 1999); understanding parasitoid/pathogenÐ mated that M. sanguinipes required from 235 to 310 DD host relationships (Carruthers et al. 1992, Bokor and (base 17.8ЊC) to complete nymphal development, and Cagan 1999); and for the optimal timing of pest man- in Saskatchewan, Gage et al. (1976) estimated agement actions (Kemp and Onsager 1986, Berry 1995, nymphal development for three species of grasshop- Berry et al. 1995). A great deal of effort has been per combined, including M. sanguinipes, required 590 expended by scientists to develop reliable models of DD (base 10ЊC). An evaluation of the phenology of insect phenology, but there are numerous challenges subarctic populations of M. sanguinipes may provide in their development (Baker 1980), such as the con- insights into the means by which this insect has founding inßuences of insolation, variable microcli- adapted to northern climates, and it provides an op- mates, and food quality. portunity to evaluate phenological models near the It is well known that, for many insects and partic- limit of this species range. ularly grasshoppers, body temperature is often sub- The duration of immature stages is often an impor- stantially higher than air temperature through absorp- tant determinant of the overall Þtness of an organism tion of solar irradiance (Chappell and Whitman 1990). 1514 ENVIRONMENTAL ENTOMOLOGY Vol. 33, no. 6

Lactin and Johnson (1996) demonstrated that grass- photoperiods, if they have any effect, often act as a hoppers actively regulate internal body temperatures signal that the end of the growing season is approach- (behavioral thermoregulation). A related problem is ing and the insect may respond with accelerated that the temperature where an insect is located may be growth rates (Masaki 1978), it is conceivable that the considerably different than air temperatures recorded extremely long days at high latitudes may provide an by standard meterological devices. During daylight opportunity for more continuous feeding, thus accel- hours while surfaces are heated by solar irradiance, erating development. Growth rates may also differ temperatures are higher closer to the ground. among locally adapted populations. Dingle et al. These considerations complicate the development (1990) showed that developmental times of M. san- of phenological models beyond an accounting of the guinipes differed, probably due to genetic differences, physical processes of, e.g., heat transfer, convection, among populations arrayed along an altitudinal gra- and absorption of solar energy, because the insectÕs dient, with higher altitude populations requiring less behavior also has to be considered, e.g., how much time to complete nymphal development at a constant time it spends in a given microhabitat, its orientation temperature of 33ЊC. to the sunÕs rays, and its preferred temperature (Lactin The objective of this study was to assess the relative and Johnson 1998). Nevertheless, any robust pheno- importance of factors, including thermoregulation, logical model needs to incorporate the effects of be- model selection, local adaptation, and photoperiod, havioral thermoregulation and microclimate. Empir- that inßuence accurate and robust phenological pre- ical models Þt to Þeld observations do this implicitly dictions of nymphal developmental times of M. san- (Gage et al. 1976, Kemp and Onsager 1986), but such guinipes under subarctic conditions. This information models will not be robust with regard to variations in will lead to more robust phenological models for eval- temperature and cloud cover from place to place or uation of pest management tactics and for life history year to year. analyses and also may contribute to a better under- Another challenge involves the selection of appro- standing of the effects of climate change on grasshop- priate equations and algorithms (Wagner et al. 1984). per phenology. Typically, the developmentÐtemperature relationship is determined by measuring developmental rates in the laboratory at a range of constant temperatures. The minimum temperature at which individuals are Materials and Methods able to complete development under constant temperature regimes is often assumed to be a threshold Growth Chamber Experiments. Models of nymphal below which no growth or development occurs. This developmental times were generated from ob- assumption may not be valid when diurnal tempera- servations of growth rates in the laboratory. Devel- ture ßuctuations drop below the putative threshold for opmental rates at temperatures too low for the a few hours daily. It seems likely that some growth and grasshoppers to complete development under con- development takes place at these lower temperatures stant temperatures were estimated from diurnally even though the insects may not be able to complete alternating temperature regimes. The accuracy of development or survive indeÞnitely when maintained the temperatureÐdevelopmental rate models were at constant low temperatures. A phenological model evaluated using two ßuctuating temperature re- that assumes such a threshold will overestimate de- gimes in the growth chambers and 2 yr of Þeld velopmental time when insects spend a substantial observations. portion of the day below “threshold” temperatures Eggs (F1 or F2 generation from adults collected (Worner 1992). Usually the temperatureÐdevelop- near Delta Junction, AK), stored in moist vermiculite, ment rate relationship is nonlinear, especially at lower were allowed to hatch at 25ЊC. Within 24 h of hatching, temperatures (Worner 1992), whereas degree-day grasshoppers were transferred to acetate tubes (10 cm calculations assume a linear relationship. A nonlinear in diameter by 30 cm in length) capped on both ends development versus temperature curve will result in with wire screens, and the tubes were placed in con- fewer degree-days required under cooler temperature trolled temperature chambers. Trials at each temper- regimes than warm (Worner 1992) because develop- ature consisted of three tubes of 20 individuals each. ment is proceeding faster than predicted at low tem- Grasshoppers were fed ad libitum with organically peratures. This seems to be the case in Kemp and grown Romaine lettuce and wheat bran treated with Onsager (1986), where M. sanguinipes required Ϸ25% a 6% solution of sulfamethazine sodium (Sulmet) at a fewer degree-days to complete nymphal development rate of 250 ml/kg of bran. Grasshoppers were exam- in a cool year versus a warmer year. ined daily, and the numbers in each instar was re- Other factors also may affect developmental time corded. Any newly molted adults were removed from of insects. It is a common observation that develop- the tubes, weighed, and placed in separate cages. Dif- mental rates of grasshoppers vary with respect to ferences in adult weight among the temperature re- food quality (Pfadt 1949, Traxler and Joern 1999). gimes was tested with analysis of variance (analysis of Photoperiod has also been shown to affect develop- variance) followed by TukeyÕs multiple comparison mental times of M. sanguinipes (Dean 1982, Dingle test (PROC GLM, SAS Institute 1996). MalesÕ and et al. 1990), with development proceeding more femalesÕ weights were analyzed separately. Differ- quickly under shorter photoperiods. Although short ences in survival among temperature regimes was December 2004 FIELDING:GRASSHOPPER PHENOLOGY AT HIGH LATITUDES 1515

Fig. 1. Temperature regimes within controlled environment chambers. Solid line, 24-h sine wave cycle; dashed line, 48-h ßuctuating cycle. Open and black boxes on the x-axis indicate light and dark photophases, respectively.

tested using a multiple comparisons test for propor- (per day) were estimated for the low temperatures in ϭ tions, analogous to TukeyÕs test, as described by Zar the alternating temperature regimes as 1/d 0.5 r(33) ϩ ϭ (1999). Photoperiod in all trials was 20:4 (L:D) h, 0.5 r(low), which was rearranged to give r(low) Ϫ unless otherwise noted, to simulate summertime day- 2/d r(33), where r(33) and r(low) are developmental light hours at 65Њ N latitude. Light was provided by rates at 33ЊC and the low temperature, respectively, ßuorescent tubes (Sylvania cool-white). Relative hu- and d is the observed median number of days from egg midity varied from 35 to 50%. hatch to adult molt under the alternating tempera- Developmental rates were determined 1) at eight tures. constant temperatures from 21 to 42ЊCat3Њ intervals, Two equations were used to model developmental 2) at three alternating temperature regimes, and 3) at rates as a function of temperature. The Þrst was a two ßuctuating temperature regimes. The alternating nonlinear equation derived by Logan et al. (1976), temperature regimes consisted of diurnal, rectangular which only approaches zero asymptotically at low thermoperiods (12 h high:12 h low), alternating be- temperatures. Њ tween the high temperature of 33 C in each case, and ϭ pT Ϫ ͓pTmax Ϫ͑Tmax Ϫ T͒/⌬͔ low temperatures of 21, 18, or 15ЊC. The ßuctuating r͑T͒ e e [1] temperature regimes consisted of 1) a sine-wave pat- where r(T) is the rate of development at temperature tern with a 24-h period with a high of 33ЊC and a low T, and Tmax, p, and ⌬ are parameters controlling the of 15ЊC, and 2) an irregular pattern that repeated over shape and height of the curve. The second equation a 2-d cycle (Fig. 1), with a high temperature of 38ЊC was a modiÞcation of the Þrst equation as described by the Þrst day and 30ЊC the second day, and a low Lactin et al. (1995), which provides an intercept, ␭, temperature of 14ЊC each day. In addition, grasshop- representing a low-temperature developmental pers were reared at 27ЊC constant temperature under threshold: short day conditions of 10:14 (L:D) h to assess the ϭ pT Ϫ ͓pTmax Ϫ͑Tmax Ϫ T͒/⌬͔ ϩ ␭ effect of photoperiod. r͑T͒ e e [2] Median date of the Þnal molt was determined with Parameters were Þt to equation 1 by using observed probit analysis (SAS Institute 1996) for each temper- developmental rates from the constant temperature ature regime. Each tube of 20 individuals was treated regimes (excepting the 21ЊC treatment) plus the es- as a separate replicate. There were three replicates per timated rates of development at 15, 18, and 21ЊC from group. Rates of development were calculated for each the alternating temperature regimes. Parameters were constant temperature regime as the reciprocal of the Þt to equation 2 by using data from the constant tem- median number of days from egg hatch to adult molt. perature regimes only. The observed times from egg hatch to adult under To compare developmental rates of the Alaskan alternating temperature regimes were used, in con- population of M. sanguinipes to those from lower lat- juction with observed developmental rate at 33ЊC con- itudes, the Þrst equation was used to model data from stant, to estimate developmental rates at temperatures previously published reports of nymphal developmen- too low for grasshoppers to complete development tal times for M. sanguinipes (Parker 1930, Shotwell under constant temperatures. Developmental rates 1941, Brett 1947, Pfadt 1949, Smith 1959, Putnam 1963, 1516 ENVIRONMENTAL ENTOMOLOGY Vol. 33, no. 6

Anderson et al. 1979, Kemp and Dennis 1989, Dingle of maximum numbers of Þrst instars in the pop- et al. 1990.) All equations were Þt using the GaussÐ ulation. Median adult molt was estimated using Newton least-squares method (PROC NLIN, SAS In- probit analysis as described above for the growth stitute 1996). chamber observations. Observations were weighted I evaluated the three models mentioned above, in by the number of individuals in each sample. All the absence of thermoregulation, by comparing pre- sampling locations were within a 10-km radius of a dicted with observed time from egg hatch to adult weather station at the University of Alaska experi- under the ßuctuating temperature regimes in con- ment farm (64Њ N, 114.5Њ W), southeast of Delta trolled environment chambers. Predictions were gen- Junction. All sites were chosen to represent a variety erated by the rate summation method (Stinner et al. of vegetation types typical of the area. The weather 1974, Worner 1992). Incremental developmental units station measured air temperature at 1.25 m above were calculated on an hourly basis by dividing by 24 ground level, and solar irradiance (Watts per square the modeled daily rates of development at each tem- meter) with a class I pyranometer [model CM-6, perature of the daily cycle. Predicted time to adult Kipp & Zonen (U.S.A.) Inc., Bohemia, New York] molt was taken to be when summed cumulative de- The weather station recorded hourly means of read- velopmental units equalled 1. The equations described ings made at 1-min intervals. above were evaluated by comparing cumulative de- Developmental times under Þeld conditions were velopmental units at the actual median time of adult predicted using four methods. First, developmental molt. rates from the growth chamber observations were Insolation and Grasshopper Internal Tempera- used to Þt equation 1, which was then used to es- tures. Internal body temperatures were measured in timate developmental times in the Þeld based on freshly killed (by freezing) grasshoppers of various 1.25-m air temperatures. Second, predictions were ages and sizes by inserting copper-constantin ther- made with the same equation based on solar ad- mocouples into the thorax through the sternum. justed temperatures. Third, data from the constant Grasshoppers were weighed before temperatures temperature treatments in the growth chamber were measured. Temperatures were measured with were used to Þt equation 2 (with a developmental the grasshoppers at ground level and perpendicular threshold), which was then used to predict devel- to the sunÕs rays. This location and position allowed opmental times based on solar adjusted tempera- estimation of the maximum possible temperature tures. Finally, published developmental rates for increase by behavioral thermoregulation (Lactin M. sanguinipes from lower latitudes were used to Þt and Johnson 1998). Measurements were taken with equation 1, and predictions of developmental times individuals on three different substrates: bare ground, green leaves, and dry leaves. Simulta- in the Þeld were generated using solar adjusted neously, solar irradiance was measured with a sili- temperatures. Predicted developmental times were con pyranometer (LI-200X, LiCor, Lincoln, NE), generated by hourly summation of growth incre- and shaded air temperature was measured at 1.25 m. ments, as described above, for each of the four All measurements were recorded within1hofre- methods. Solar-adjusted temperatures were gener- moving the grasshoppers from the freezer, to avoid ated using the previously determined linear rela- effects of drying. tionship between solar irradiance and elevation of Elevation of body temperatures was expressed as body temperatures. An upper limit of estimated Њ degrees above air temperature. The relationship grasshopper temperatures was set at 39 C, which is between elevation of body temperature and solar near the observed preferred maximum temperature irradiance was modeled using linear regression for several species of grasshopper {Kemp 1986, (PROC REG, SAS Institute 1996). To assess the Carruthers et al. 1992, Lactin and Johnson 1996). importance of substrate on elevation of body tem- When solar irradiance was too low to signiÞcantly peratures, an analysis of covariance (ANCOVA) elevate body temperature, as determined by the was conducted with the difference between air and intercept of the linear solar-body temperature equa- body temperature as the dependent variable, sub- tion, body temperatures were assumed to be equal strate as the main factor, and solar irradiance as the to recorded air temperature. covariable (PROC GLM, SAS Institute 1996, Zar 1999). Field Observations. Seven sampling locations Results were selected in spring 2000 before any grasshoppers had hatched. Samples were taken in 2000 and Nymphal Development at Constant and Alternat- 2002. Grasshopper populations were too low in 2001 ing Temperatures. Survival rates and fresh weights of for meaningful data. At each location, two or three adults were lower at the high and low extremes of the times per week, beginning before egg hatch and constant temperature treatments (Table 1). There continuing throughout the nymphal stages, grass- was little difference in survival or weight among any hoppers were collected with a sweep net. Sweep of the alternating temperature regimes and the middle samples were frozen and grasshoppers were later range of constant temperatures. Developmental times sorted according to species and instar. Median date decreased with increasing constant temperatures up of egg hatch was estimated to be 2 d before the date to 39ЊC. There was no difference in developmental December 2004 FIELDING:GRASSHOPPER PHENOLOGY AT HIGH LATITUDES 1517

Table 1. Development times, survival, and weights (؎ SD) of M. sanguinipes reared at eight constant temperatures and three alternating temperatures

Median days to Adult weight (g) Temperature regime Survival adult molt n Male n Female (oC) (%) (95% CL) Mean Ϯ SD Mean Ϯ SD 21 constant 65.2 (63.6Ð66.7) 38c 12 0.20 Ϯ 0.020cd2 11 0.20 Ϯ 0.030d2 24 constant 41.6 (40.1Ð42.9) 72ab 13 0.22 Ϯ 0.039c 31 0.24 Ϯ 0.030c 27 constant (short days)1 27.5 (26.1Ð29.0) 83ab 22 0.29 Ϯ 0.036b 29 0.31 Ϯ 0.054b 27 constant (long days)1 28.0 (26.7Ð29.4) 85ab 25 0.30 Ϯ 0.040ab 27 0.32 Ϯ 0.052b 30 constant 23.4 (21.7Ð25.0) 90ab 30 0.30 Ϯ 0.038ab 25 0.36 Ϯ 0.052ab 33 constant 17.8 (16.1Ð19.2) 92a 27 0.31 Ϯ 0.041ab 26 0.40 Ϯ 0.063a 36 constant 16.7 (15.6Ð18.2) 69b 16 0.32 Ϯ 0.054ab 25 0.37 Ϯ 0.056ab 39 constant 15.8 (13.7Ð19.7) 69b 19 0.29 Ϯ 0.038b 22 0.29 Ϯ 0.038b 42 constant 21.7 (20.1Ð23.0) 28c 5 0.18 Ϯ 0.034d 11 0.20 Ϯ 0.040d 33/21 alternating 25.8 (24.2Ð27.1) 81ab 28 0.33 Ϯ 0.040a 25 0.36 Ϯ 0.064ab 33/18 alternating 29.5 (27.7Ð31.0) 86ab 20 0.31 Ϯ 0.033ab 28 0.35 Ϯ 0.061ab 33/15 alternating 27.6 (25.5Ð28.8) 76ab 18 0.31 Ϯ 0.054ab 26 0.35 Ϯ 0.073ab

1. Short days, 10:14 (L:D); long days, 20:4 (L:D). 2. Means within a column followed by the same letter are not signiÞcantly different (P Ͻ 0.05, LSD test) times between short days and long days at 27ЊC developmental rates were nearly all greater than any (Table 1). previously published times at the same constant tem- Nonzero developmental rates at 15 and 18ЊC, esti- peratures (Fig. 2). mated from the alternating temperature regimes, in- Both equations provided a good Þt to the data: dicate that some growth and development was taking equation 2 to the constant temperature regimes and place at these temperatures (Fig. 2). Developmental equation 1 to the constant temperatures plus low tem- rate at 21ЊC estimated from the alternating tempera- perature developmental rates from the alternating re- tures seemed to be greater than that observed for the gimes (Table 2; Fig. 2). Equation 2 predicted a devel- 21ЊC constant temperature (Fig. 2). Survival rate and opmental threshold at 16.4ЊC. Equation 1 Þt to the weights at constant temperatures of 21, 24, and 42ЊC Alaskan data provided the best estimates of propor- were lower than other treatments (Table 1), indicat- tional development under the ßuctuating regimes ing that grasshoppers were stressed when maintained (Table 3). Equation 2 underestimated cumulative de- at these temperatures constantly. Observed nymphal velopmental units under ßuctuating temperatures by 13Ð22%, and equation 1, Þt to lower latitude data, underestimated cumulative developmental units by 27Ð31% (Table 3). Field Observations. M. sanguinipes consisted of Ϸ25% of all grasshoppers collected in 2000 and Ϸ15% in 2002. Median hatch dates were estimated to be somewhat earlier in 2002 than in 2000 (29 May versus 2 June). In both years, there was a clearly deÞned pulse of Þrst instars of M. sanguinipes over a short period of time (Fig. 3). The median date of adult molt was 16 July and 9 July, in 2000 and 2002, respectively, resulting in somewhat shorter nymphal developmental times in 2002 than in 2000 (Table 3). Fresh weight and femur length of Þeld-collected females was within the range of published data from lower latitude populations (Fig. 4). Elevation of grasshopper body temperatures over Fig. 2. Developmental rates (reciprocal of number of ambient was primarily inßuenced by solar irradiation days from egg hatch to adult eclosion) of M. sanguinipes from (ANCOVA, F ϭ 241.0, P Ͻ 0.0001). Substrate type may previously published reports and from results of experiments have had some effect on thermoregulation (AN- with a population from Alaska (see Table 1). Solid line, COVA, F ϭ 2.75, P ϭ 0.071). Because the effect was not equation 1 (Logan) Þt to Alaskan data only, including rates strong, and because of the difÞculties in attempting to at 15, 18, and 21ЊC estimated from alternating temperature estimate the amount of time a grasshopper spent on regimes; dotted line, equation 1 (Logan) Þt to previously different substrates, all data were combined in a single published data from lower latitudes, including rates at 12 and 2 Њ linear equation that provided a reasonable estimate (r 22 C estimated from alternating temperature regimes ϭ (Parker 1930); dashed line, equation 2 (Lactin) with inter- 0.78) of the effect of solar irradiance on internal cept Þt to data from Alaskan populations at constant tem- temperatures of grasshoppers (Fig. 5). Below 98 peratures only. See text and Table 2 for details on equations. W/m2, elevation of body temperature was negligible. 1518 ENVIRONMENTAL ENTOMOLOGY Vol. 33, no. 6

Table 2. Parameter estimates (؎ SE of estimate) for equations modeling developmental rates (1/d) of Melanoplus sanguinipes

Parameters ⌬ ␭ Equations PTmax 1. Logan (Eq. 1) Þtted to Alaskan data 0.1400036 44.92730234 7.14040716 n.a. Ϯ0.0067114 Ϯ0.43901831 Ϯ0.3417696 2. Lactin (Eq. 2) Þtted to Alaskan data 0.00306514 50.94143178 2.64423622 Ϫ1.05145087 Ϯ0.000217713 Ϯ1.93310510 Ϯ0.6250147 Ϯ0.00612136 3. Logan (Eq. 1) Þtted to lower-latitude data 0.14482974 44.83717143 6.90320288 n.a. Ϯ0.01059514 Ϯ1.28134427 Ϯ0.504557377

Because the grasshoppers were in contact with the ers et al. (1992) found that grasshopper Camnula pel- substrate, this relationship includes the effects of sub- lucida (Scudder) maintained body temperatures near strate heating and ground-level microclimate. Use of 39ЊC, and Lactin and Johnson (1996) found that 40ЊC solar-adjusted temperatures with equation 1 Þtted to was the modal body temperatures for nymphs of Alaskan data provided predictions nearest to observed M. sanguinipes, when allowed to freely position them- developmental times (Table 3), overestimating cumu- selves in a temperature gradient. Observed develop- lative developmental units by 5% in 2000 and under- mental rates in the Alaskan population peaked near estimating by 11% in 2002. Using air temperatures only 39ЊC, making it a logical choice, although survival and with the same equation underestimated cumulative adult weights were not at the optimum at 39ЊC con- development by 39 and 53% in 2000 and 2002, respec- stant temperature. In reality, estimated body temper- tively. Equation 2, which predicted no development atures did not reach 39ЊC frequently and at most for below 16.4ЊC, underestimated cumulative develop- only a few hours per day. Another assumption was that ment by 16 and 29%, in 2000 and 2002, respectively, by grasshoppers maintained orientation and location that using solar-adjusted temperatures. Equation 1 Þt to maximized aborption of solar energy, i.e., that grass- lower latitude data, by using solar-adjusted tempera- hoppers attempted to elevate body temperatures as tures, underestimated cumulative development by 19 near to the optimum (Ϸ39ЊC) as possible. Thus, mea- and 33% in 2000 and 2002, respectively (Table 3). All surements of grasshopper internal body temperatures methods underestimated cumulative developmental were made with grasshoppers in contact with a sub- units to a greater degree in 2002 than in 2000 (Table 3). strate that was also solar heated, and with body ori- ented perpendicular to sunÕs rays. To incorporate the effects of behavioral thermoregulation with as few Discussion assumptions as possible, I used simple linear regression M. sanguinipes in subarctic Alaska completed de- to relate temperature elevation to solar irradiance and velopment in Ͻ45 d, similar to that reported in warmer disregarded effects of wind. Although wind is un- climates (Kemp and Onsager 1986), even though am- doubtedly an important factor in thermoregulation bient temperatures were much lower. By modeling (Lactin and Johnson 1998), the difÞculties in estimat- developmental rates by using various methods, I was ing wind speed at ground level preclude the use of this able to assess the relative contributions of thermoreg- factor in operational situations. ulation, “below threshold” development, and local ad- I assumed that grasshoppers were not able to ther- aptation to rapid nymphal development in high-lati- moregulate in the growth chambers, thereby provid- tude populations of M. sanguinipes. ing a means to assess the contribution of behavioral Some assumptions were necessary to estimate ther- thermoregulation to the rapid growth. Endothermy, or moregulation under Þeld conditions. One is that grass- metabolic thermoregulation, has only been demon- hoppers did not allow body temperatures to exceed strated in grasshoppers while in ßight (Chappell and 39ЊC. Observations of Þeld populations (Parker 1930, Whitman 1990). Equation 1, Þt to Alaskan data and Kemp 1986) suggest grasshoppers avoid temperatures that only approaches developmental zero asymptoti- higher than this. In laboratory experiments, Carruth- cally, gave a reasonable prediction of developmental at the (1.0 ؍ Table 3. Observed median no. days from egg hatch to adult eclosion and cumulative developmental units (adult molt time of observed median adult molt using four methods applied to two ﬂuctuating temp regimes in a controlled environment and to two years of ﬁeld observations

a Observed median Cumulative Developmental Units Temperature days to adult Eq. 1 - AK, Eq. 1 -lower- regime Eq.1-AK Eq.2-AK molt (95% C.L.) without solar latitude Sine waveb 34.7 (33.1Ð36.1) n.a. 1.04 0.88 0.81 Fluctuatingb 37.1 (35.7Ð38.6) n.a. 1.01 0.82 0.79 Field, 2000 45 (40Ð50) 1.05 0.61 0.82 0.81 Field, 2002 42 (40Ð44) 0.89 0.47 0.71 0.70

a Solar-adjusted temperatures were used in the calculation of developmental units, unless otherwise noted. b See Fig. 1 for temp cycles. December 2004 FIELDING:GRASSHOPPER PHENOLOGY AT HIGH LATITUDES 1519

Fig. 3. Phenology of M. sanguinipes near Delta Junction, AK, in 2000 (a) and 2002 (b). Observed proportion of the population as Þrst instars (triangles) and adults (Þlled circles). Solid line, results of probit analysis used estimate median adult molt. Dashed line, cumulative developmental units using equation 1 (see text and Table 1) with solar adjusted temperatures; dotted line, same as dashed line except with air temperatures without solar adjustment. Fig. 4. Size of Þeld-collected adult female M. sanguinipes from near Delta Junction, AK, compared with populations times under ßuctuating temperatures in the growth from lower latitudes. Mean femur length: AK, from Alaska; chambers, lending some conÞdence that it accurately ID-low, southern Idaho below 1,500-m elevation (unpub- describes developmental rates within the temperature lished data); ID-high, southern Idaho above 1,500-m eleva- range examined (14Ð39ЊC). In the Þeld, low temper- tion (unpublished data); WY1Ð4, population from Wyoming atures were often lower than low temperatures in the reared in laboratory with different food plants (Pfadt 1949). Mean weight of mature, Þeld-collected female adult M. san- growth chambers. If the models do not accurately guinipes: AK, Alaskan population; MT81Ð86, population from describe development at these lower temperatures, it western Montana in years from 1981 through 1986 (Belovsky would alter the predicted number of days required to and Slade 1995). Error bars indicate 1 SD (not reported for complete development in the Þeld, but it would have WY femur length; Pfadt 1949). little effect on the relative ranking of the different methods in terms of accuracy of predicted development. Comparison of developmental times predicted to elevate body temperatures above ambient for Ϸ15 by using ambient temperatures with predictions by h per day at latitude 64Њ N compared with 13 h per day using solar-adjusted temperatures indicates that grass- for Miles City, MT (latitude 46.5Њ N, NOAA 2002). The hoppers cut the time spent in nymphal stages by ap- relative beneÞts, in terms of developmental rates, de- proximately one-half through behavioral thermoreg- rived from thermoregulation at higher latitudes would ulation. be much greater because lower ambient temperatures Days are long at high latitudes, but the low angle of would allow grasshoppers (via behavioral thermoreg- the sun for much of the time probably does not allow ulation) to maximize absorption of solar energy with- thermoregulation for many more hours per day than out exceeding their optimum temperature. At lower at lower latitudes. Examining the solar irradiance read- latitudes, daytime summer temperatures may regu- ings, assuming that 100 W/m2 is required to raise body larly exceed 30ЊC, so grasshoppers would not need to temperatures signiÞcantly (Fig. 5), on clear days near elevate body temperatures as much to reach their the summer solstice, grasshoppers could only expect optimum temperature and the relative increase in 1520 ENVIRONMENTAL ENTOMOLOGY Vol. 33, no. 6

of subarctic populations. Without rearing the insects from different populations simultaneously under iden- tical conditions, I cannot unequivocally state that genetic differences are responsible, because rearing conditions (e.g., food and humidity) may account for some of the discrepancy between developmental rates of M. sanguinipes in this study versus previously published reports. Nevertheless, populations of M. sanguinipes from differing thermal environments have been shown to develop at different rates under iden- tical conditions (Dingle et al. 1990), and because the developmental rates in this study are the highest ever reported for M. sanguinipes at several temperatures, it seems that genetically based population differences in developmental rates are likely. Fig. 5. Internal body temperatures, as degrees above Another assumption placed median egg hatch at 2 d shaded air temperatures at 1.25 m, of M. sanguinipes nymphs previous to the observed peak percentage of Þrst in- as a function of solar irradiance. Grasshoppers were at ground stars. Unlike lower latitudes, where hatch of M. san- level in contact with bare ground (Þlled circles), green leaves guinipes is often spread out over several weeks (On- (open circles), or leaf litter (triangles). Relationship tem- sager 1987), hatching in Alaska occurs within a very perature elevation with solar irradiance: green leaf substrate, limited time frame, making estimation of median hatch dotted line; bare ground, short dashed line; leaf litter, long less challenging. All model predictions of cumulative dashed line; all substrates combined, heavy solid line. development were lower for 2002 than 2000 or the ßuctuating regimes in the controlled environments (Table 3). If the actual median hatching date was, for body temperature and developmental rates would be instance, 2 d later than estimated, all models would less. have two extra days of accumulated developmental Estimation of low-temperature developmental rates units, and so predictions of the date of median adult from alternating temperature regimes assumed no molt would be relatively early. Nevertheless, starting physiological acceleration due to variable tempera- from different hatching dates would not greatly tures, i.e., development at the two alternating tem- change the performance of the different equations peratures was not inßuenced by the alternating tem- relative to one another. perature, or the span between the two temperatures, I also assumed food quality did not affect nymphal or the thermoperiod. Liu et al. (1995) analyzed the developmental rates in the Þeld. There was a great literature on insect developmental rates and found variety of plants at the locations where grasshoppers little evidence for such effects. Equation 1, Þt to Alas- were collected, and it seems likely that M. sanguinipes kan data and that only approaches developmental zero could select an adequate diet from among them (Be- asymptotically, gave a reasonable description of de- hmer and Joern 1993). Also, rainfall was such that velopmental times under ßuctuating temperatures in plants remained green and actively growing through- the growth chambers, whereas equation 2, based on out the nymphal stage. constant temperature data only and that predicted a The data used to estimate developmental rates from developmental rate of zero below 16.4ЊC, underesti- lower latitudes were from a variety of published ex- mated development. In the Þeld also, equation 2, using periments. The studies used in the analysis did not all solar-adjusted temperatures, underestimated devel- use grasshoppers reared under optimal conditions, opmental times. These results, along with the results intentionally, thus introducing some bias. Neverthe- of the alternating temperature regimes, indicate that less, it seems that local adaptation may be an important signiÞcant growth and development occurred at tem- source of error in phenological models. This means peratures below that which is sufÞcient for M. sangui- that a single, regionally independent phenological nipes to complete development under constant tem- model for M. sanguinipes may not be possible. Pheno- peratures. It seems that, at least for M. sanguinipes, it logical models must be calibrated to local populations is not very useful to estimate developmental thresh- and the geographical boundaries where each model is olds based only on the results from constant temper- applicable need to be determined. ature regimes. Incorrect thresholds will weaken the Through behavioral thermoregulation, and proba- robustness of the model, even if calibrated to local bly genetic adaptation, M. sanguinipes in Alaska was conditions, to seasonally changing temperature re- able to complete nymphal development in about the gimes or interannual variation in weather (Worner same amount of time as at lower latitudes. This ob- 1992). servation suggests that M. sanguinipes is well adapted The same equation, but Þt to data from lower lati- to moderate climatic perturbations and that climate tudes, underestimated development in the Þeld, indi- change may not greatly affect nymphal developmental cating the inßuence of local adaptation on phenology times. December 2004 FIELDING:GRASSHOPPER PHENOLOGY AT HIGH LATITUDES 1521

Acknowledgments Greenberg, B. 2002. Entomology and the law: ßies as forensic indicators. Cambridge University Press, Cambridge, Linda DeFoliart provided technical assistance. The manu- NY. script was improved by suggestions from John Fox, Peter Hebard, M. 1929. The Orthoptera of Colorado, pp. 303Ð425. Bechtel, and anonymous reviewers. In Proceedings of the Academy of Natural Sciences Phil- adelphia 80. Kemp, W. P. 1986. Thermoregulation in three rangeland References Cited grasshopper species. Can. Entomol. 118: 335Ð343. Kemp, W. P., and B. Dennis. 1989. Development of two Anderson, R. V., C. R. Tracy, and Z. Abramsky. 1979. Hab- rangeland grasshoppers at constant temperatures: devel- itat selection in two species of short-horned grasshoppers. opment thresholds revisited. Can. Entomol. 121: 363Ð371. Oecologia (Berl.). 38: 359Ð374. Kemp, W. P., and J. A. Onsager. 1986. Rangeland grasshop- Ayres, M. P., and J. M. Scriber. 1994. Local adaptation to pers (Orthoptera: Acrididae): modeling phenology of regional climates in Papilio canadensis (Lepidoptera: Pap- natural populations of six species. Environ. Entomol. 15: ilionidae). Ecol. Monogr. 64: 465Ð482. 924Ð930. Baker, C.R.B. 1980. Some problems in using meteorological Lactin, D. J., and D. L. Johnson. 1996. Behavioral optimi- data to forcast the timing of insect life cycles. EPPO Bull. zation of body temperature by nymphal grasshoppers 10: 83Ð91. (Melanoplus sanguinipes, Orthoptera: Acrididae) in tem- Behmer, S. T., and A. Joern. 1993. Diet choice by a grass- perature gradients established using incandescent bulbs. feeding grasshopper based on the need for a limiting J. Thermal Biol. 21: 231Ð238. nutrient. Funct. Ecol. 7: 522Ð527. Lactin, D. J., and D. L. Johnson. 1998. Environmental, phys- Belovsky, G. E., and J. B. Slade. 1995. Dynamics of two ical, and behavioural determinants of body temperature Montana grasshopper populations: relationships among in grasshopper nymphs (Orthoptera: Acrididae). Can. weather, food abundance and intraspeciÞc competition. Entomol. 130: 551Ð577. Oecologia (Berl.). 101: 383Ð396. Lactin, D. J., N. L. Holliday, D. L. Johnson, and R. Craigen. Berry, J. S. 1995. Computer models in integrated pest man- 1995. Improved rate model of temperature-dependent agement: a case study of the grasshopper integrated pest development by arthropods. Environ. Entomol. 24: 68Ð75. management project. J. Agric. Entomol. 12: 229Ð240. Liu, S. S., G. M. Zhang, and J. Zhu. 1995. Inßuence of tem- Berry, J. S., W. P. Kemp, and J. A. Onsager. 1995. Within- perature-variations on rate of development in insects - year population dynamics and forage destruction model analysis of case-studies from entomological literature. for rangeland grasshoppers (Orthoptera: Acrididae). En- Ann. Entomol. Soc. Am. 88: 107Ð119. viron. Entomol. 24: 212Ð225. Logan, J. A., and B. J. Bentz. 1999. Model analysis of moun- Bokor, P., and L. Cagan. 1999. Phenology, basic biology and tain pine beetle (Coleoptera: Scolytidae) seasonality. parasitism of Microgaster tibialis (Hymenoptera, Bra- Environ. Entomol. 28: 924Ð934. conidae), a parasitoid of the European corn borer, Os- Logan, J. A., and J. A. Powell. 2001. Ghost forests, global trinia nubilalis, in central Europe. Biologia 54: 567Ð572. warming, and the mountain pine beetle (Coleoptera: Bradford, M. J., and D. A. Roff. 1997. An empirical model of Scolytidae). Am. Entomol. 47: 160Ð173. diapause strategies of the cricket, Allonemobius socius. Logan, J. A., D. J. Wollkind, S. C. Hoyt, and L. K. Tanigoshi. Ecology 78: 442Ð451. 1976. An analytic model for description of temperature Bradshaw, W. E., S. Fujiyama, and C. M. Holzapfel. 2000. dependent rate phenomena in arthropods. Environ. En- Adaptation to the thermal climate of North America by tomol. 5: 1133Ð1140. the pitcher-plant mosquito, Wyeomyia smithii. Ecology Masaki, S. 1978. Seasonal and latitudinal adaptations in the 81: 1262Ð1272. life cycles of crickets, pp. 72Ð100. In H. Dingle [ed.], Brett, C. H. 1947. Interrelated effects of food, temperature, Evolution of insect migration and diapause. Springer, and humidity on the development of the lesser migratory New York. grasshopper, Melanoplus mexicanus mexicanus (Saussure) [NOAA] National Oceanic and Atmospheric Administra- (Orthoptera). Department of Entomology. Oklahoma tion. 2002. National Climatic Data Center, www.ncdc. Agricultural Experiment Station, Stillwater, OK. noaa.gov/oa/ncdc.html. National Oceanic and Atmo- Bryant, S. R., C. D. Thomas, and J. S. Bale. 2002. The inßu- spheric Administration. ence of thermal ecology on the distribution of three Onsager, J. A. 1987. Integrated management of rangeland nymphalid butterßies. J. Appl. Ecol. 39: 43Ð55. grasshoppers, pp. 196Ð204. In J. L. Capinera [ed.], Inte- Carruthers, R. L., T. S. Larkin, H. Firstencel, and Z. Feng. grated pest management on rangeland: a shortgrass prai- 1992. Inßuence of thermal ecology on the mycosis of a rie perspective. Westview Press, Boulder, CO. rangeland grasshopper. Ecology 73: 190Ð204. Parker, J. R. 1930. Some effects of temperature and mois- Chappell, M. A., and D. W. Whitman. 1990. Grasshopper ture upon Melanoplus mexicanus mexicanus Saussure and thermoregulation, pp. 143Ð172. In R. F. Chapman and A. Camnula pellucida Scudder (Orthoptera), pp. 132. Mon- Joern [eds.], Biology of grasshoppers. Wiley, New York. tana Agricultural Experiment Station Bulletin 223. Dean, J. M. 1982. Control of diapause induction by a change Pfadt, R. E. 1949. Food plants as factors in the ecology of the in photoperiod in Melanoplus sanguinipes. J. Insect Lesser Migratory grasshopper Melanoplus mexicanus Physiol. 28: 1035Ð1040. (Sauss.). University of Wyoming Agricultural Experi- Dingle, H., T. A. Mousseau, and S. M. Scott. 1990. Altitudinal ment Station Bulletin 290. variation in life cycle syndromes of California populations Putnam, L. G. 1963. The progress of nymphal development of the grasshopper, Melanoplus sanguinipes (F.). Oeco- in pest grasshoppers (Acrididae) of western Canada. Can. logia (Berl.). 84: 199Ð206. Entomol. 95: 1210Ð1216. Gage, S. H., M. K. Mukerji, and R. L. Randell. 1976. A pre- Regniere, J., and V. Nealis. 2002. Modelling seasonality of dictive model for seasonal occurrence of three grasshop- gypsy moth, Lymantria dispar (Lepidoptera: Lymantri- per species in Saskatchewan (Orthoptera: Acrididae). idae), to evaluate probability of its persistence in novel Can. Entomol. 108: 245Ð253. environments. Can. Entomol. 134: 805Ð824. 1522 ENVIRONMENTAL ENTOMOLOGY Vol. 33, no. 6

Roff, D. A. 1992. The evolution of life histories; theory and oligophagous grasshopper Hesperotettix viridis (Acridi- analysis. Chapman & Hall, New York. dae). Oikos 87: 239Ð250. SAS Institute. 1996. The SAS System for Windows, release Vickery, V. R., and D.K.M. Kevan. 1983. A monograph of 6.12. SAS Institute, Cary, NC. the orthopteroid insects of Canada and adjacent regions. Shotwell, R. L. 1941. Life histories and habits of some grass- Pp. 681Ð1462. Lyman Entomol. Mus. Res. Lab. Mem. No. hoppers of economic importance on the great plains. 13: iÐiv. USDA Tech. Bull. 774. Wagner, T. L., H.-I. Wu, P.J.H. Sharpe, R. M. Schoolﬁeld, and Smith, D. S. 1959. Utilization of food plants by the migratory R. N. Coulson. 1984. Modeling insect development grasshopper Melanoplus mexicanus mexicanus (Sauss.) rates: a literature review and application of a biophysical (Orthoptera: Acrididae), with some observations on the model. Ann. Entomol. Soc. Am. 77: 208Ð225. nutritional value of the plants. Ann. Entomol. Soc. Am. 52: Worner, S. P. 1992. Performance of phenological models 674Ð680. under variable temperature regimes: consequences of the Smith, K.G.V. 1986. A manual of forensic entomology. Cor- Kaufmann or rate summation effect. Environ. Entomol. nell University Press, Ithaca, NY. 21: 689Ð699. Stinner, R. E., A. P. Gutierrez, and G. D. Butler. 1974. An Zar, J. H. 1999. Biostatistical analysis. Prentice-Hall, Engle- algorithm for temperature-dependent growth rate simu- wood Cliffs, NJ. lation. Can. Entomol. 106: 519Ð524. Traxler, M. A., and A. Joern. 1999. Performance tradeoffs for two hosts within and between populations of the Received 30 September 2003; accepted 10 June 2004.