ECOLOGY
AND
BEHAVIOR
Effects of Temperature on Development and Survival of Orthopygia glaucinalis (Lepidoptera: Pyralidae) Reared on Platycarya strobilacea JIAN-FENG LIU,1 MAO-FA YANG,1,2 JI-FENG HU,3 AND CHANG HAN4
J. Econ. Entomol. 108(2): 504–514 (2015); DOI: 10.1093/jee/tov003
ABSTRACT The larvae of Orthopygia glaucinalis (L.) (Lepidoptera: Pyralidae) are used to produce insect tea in Guizhou, China. We investigated the development and survival of O. glaucinalis reared on dried leaves of Platycarya strobilacea under laboratory conditions at 19, 22, 25, 28, 31, 34, and 37 C. The duration of development from egg deposition to adult emergence decreased significantly with increasing temperature from 19 to 31 C, whereas the duration of egg and overall development significantly increased at 34 C. Based on the extreme-value distribution function, the optimal temperature for survival of overall development was 24.89 C, and the larval stage was most susceptible to temperature extremes. The common linear model and the Ikemoto and Takai linear model were used to determine the relationship between temperature and the developmental rate, and estimated the low-temperature threshold (11.44 and 11.62 C, respectively) and the threshold constant (1220.70 and 1203.58 degreedays, respectively) of O. glaucinalis. Nonlinear models were used to assess in fitting the experiment data and to estimate the high temperature thresholds (34.00 to 39.08 C) and optimal temperatures (31.61 to 33.45 C). An intrinsic optimal temperature of 24.18 C was estimated for overall development using the Sharpe–Schoolfield–Ikemoto (SSI) model. Model-averaged parameter estimates and the unconditional standard error were also estimated for the temperature thresholds. Based on the biological parameters and model selection, we concluded that common linear, Lactin-1, and SSI models performed better for predicting the temperature-dependent development of O. glaucinalis. Our findings enable breeders to optimize the developmental rate of O. glaucinalis and improve the yield of insect tea. KEY WORDS Orthopygia glaucinalis, insect tea, developmental rate, critical threshold, model selection Insect tea is a unique specialty tea produced from the frass of certain insect larvae that have fed on particular plant species (You and Zhao 1979, Gao 1996, Xu et al. 2013, Liu et al. 2014). The use of insect tea was first documented in 1578 during the Ming Dynasty in “Compendium of Materia Medica” (Li 1982). The Chengbu Local Records (V) office in Hunan Province also has a record from 1906 related to the use of insect tea during the reign of Emperor Guangxu of the Qing Dynasty. Insect tea is a source of high quality protein and amino acids, and is often used to relieve summer heat, protect the spleen and stomach, and improve digestion (Wu 1997, Xu et al. 2013, Liu et al. 2014). Insect tea is extremely popular in the provinces in
1 Institute of Entomology, College of Agriculture, Guizhou University, Xiahui Rd., Huaxi District, 550025, Guiyang City, Guizhou Province, P. R. China. 2 Corresponding author, e-mail:
[email protected]. 3 Agricultural Bureau of Huaxi District, Mingzhu Rd., Huaxi District, 550025, Guiyang City, Guizhou Province, P. R. China. 4 Department of Forest Protection, College of Forestry, Guizhou University, Xiahui Rd., Huaxi District, 550025, Guiyang City, Guizhou Province, P. R. China.
southwestern China and in certain Southeast Asian countries (Gao 1996). Temperature is a critical abiotic factor that influences the dynamics of insect populations and that of their natural enemies (Huffaker et al. 1999, Nelson et al. 2013). The developmental rate of insects increases almost linearly below the limit of sublethal high temperatures (Wen 1997; Shang et al. 2012, 2013; Liu et al. 2014). The common and Ikemoto and Takai linear models have been used to predict the egg, larval, and pupal development of insects using physiological data (Wen 1997; Shang et al. 2012, 2013; Liu et al. 2014). Linear approximation and reduced major axis methods allow the estimation of low-temperature thresholds and thermal constants within a limited temperature range (Campbell et al. 1974, Ikemoto and Takai 2000). The developmental rate is essentially nil at the low-temperature threshold, and increases with increasing temperature. It reaches at a maximum at the optimal temperature, and decreases rapidly as the high-temperature threshold is approached (Roy et al. 2002, Aghdam et al. 2009). Although the relationship between temperature and the developmental rate is approximately linear at moderate temperatures, it is curvilinear near the high- and
C The Authors 2015. Published by Oxford University Press on behalf of Entomological Society of America. V
All rights reserved. For Permissions, please email:
[email protected]
April 2015
LIU ET AL.: EFFECTS OF TEMPERATURE ON O. glaucinalis DEVELOPMENT
low-temperature thresholds (Wagner et al. 1984). A variety of nonlinear models have been used to describe the relationship between the developmental rate and a wider temperature range in a more realistic manner (Davidson 1944, Logan et al. 1976, Lactin et al. 1995, Brie`re et al. 1999, Roy et al. 2002, Kontodimas et al. 2004, Shi et al. 2011, Ikemoto et al. 2013, Liu et al. 2014). Nonlinear models also provide estimates of high-temperature thresholds and optimal temperatures for development during the various developmental stages. Temperature-driven rate models have been used to predict the activity and seasonal population dynamics of resource insects (Roy et al. 2002, Kontodimas et al. 2004). The larvae of Orthopygia glaucinalis (L.) (Lepidoptera: Pyralidae) are used to produce insect tea in Guizhou Province (Wang 1980, Liu et al. 2013). Previous studies have referred to the frass of O. glaucinalis larvae fed on the dried leaves of Platycarya strobilacea as the O. glaucinalis–P. strobilacea insect tea (You and Zhao 1979, Gao 1996, Liu et al. 2013). Liu et al. (2013) described the morphological characteristics of O. glaucinalis and the tea made from the insect. Recent studies have indicated that temperature has a significant effect on the development and survival of Agloss dimidiata (Haworth, 1809) reared on Malus sieboldii (Regel) Rehd (Wen 1997), A. dimidiata and Pyralis farinalis (Linnaeus, 1758) reared on Litsea coreana Levl. var. sinensis (Allen) Yang et P. H. Huang (Shang et al. 2012, 2013), and Lista haraldusalis (Walker, 1859) reared on P. strobilacea (Liu et al. 2014). Temperature has a greater effect on the daily yield of insect tea (J.F.L. unpublished data. This section has been done by Jian-Feng Liu). Wen (2002) reported that the yield of tea produced using A. dimidiata reared on M. sieboldii was directly proportional to the survival rate of the A. dimidiata larvae. Therefore, to optimize the yield and quality of insect tea, the effect of temperature on the development and survival of the insect must be considered. The optimal temperature for the development and survival of O. glaucinalis remains unclear. We compared the effect of different temperatures on the development and survival of the eggs, larvae, pupae, and all immature stages of O. glaucinalis reared on P. strobilacea. An extreme-value distribution function was used to describe the effect of temperature on the survival of each life stage, and to estimate the maximum survival (k) of O. glaucinalis at the optimal temperature (b). The low-temperature threshold (Tmin) and thermal constant (K) for each immature developmental stage were estimated using linear models. The high-temperature threshold (Tmax) and optimal temperature (Topt) for each developmental stage were estimated using nonlinear models. Based on the biological parameters and model selection, we identified the best model for predicting the temperature-dependent developmental rate of O. glaucinalis. Materials and Methods Insects and Plants. The O. glaucinalis were originally collected in August 2012 from Congjiang
505
Guichushan Biological Tea Development Company (Congjiang, Guizhou, China), and a laboratory colony was established at the Institute of Entomology at Guizhou University. The stock colony was reared on the dried leaves of P. strobilacea in a breeding box at 25 6 1 C. The breeding box consisted of a clear plastic container that was 40 50 cm at the base and 30 cm in height. The container was covered with fine mesh gauze and plastic cap. The P. strobilacea leaves were collected in August 2012 from the same fields from which the O. glaucinalis were obtained. The leaves were washed with water and sun-dried before being heated at 60 C for 24 h in a drying oven to kill any remaining mites or spiders. The stock colony and all of the insects used in the experiments were maintained in an RXZ-380 A incubator (Ningbo Jiangnan Instruments, Ningbo City, China) in an atmosphere of 80 6 5% relative humidity (RH) with a photoperiod of 14:10 (L:D) h. Development and Survival Rates of O. glaucinalis at Different Temperatures. We determined the duration of developmental and survival rate of the eggs, larvae, and pupae of O. glaucinalis at 19, 22, 25, 28, 31, 34, and 37 C. To obtain synchronized eggs of O. glaucinalis, approximately 50 pairs of newly eclosed adults of both sexes were collected from the stock colonies. Each pair was placed in separate 400-ml clear-plastic cups for oviposition at 25 6 1 C. A ball of cotton saturated with 10% honey was placed in the oviposition cup to provide nutrition, and the container was sealed with a muslin cloth. After the 3-d mating period, most of the females had laid eggs on the wall of the container. After 24 h, the adult moths were transferred to a new container. For each experiment, >200 one-day-old eggs were assessed every 4 to 12 h in the rearing container, depending on the temperature, until all of the eggs had either hatched or collapsed. Neonate larvae were collected within 4 h of hatching, and were individually transferred using a fine camel hair brush to a 50-ml centrifuge tube containing 0.3 g of P. strobilacea leaves. The tubes were sealed with fine mesh gauze for ventilation. Additional dried leaves were added to the tubes two times per month. The insects were checked daily, and survival was recorded. The larval exuviae were used to determine molting. Pupae were collected from the centrifuge tube, and placed individually in a new 50-ml centrifuge tube covered with fine mesh gauze. The pupae were observed twice daily, at 0800 and 2000 hours. Adult emergence and the sex of the emerged adults were also recorded. Each treatment was repeated three times in the same incubator. The duration of development and survival rate of O. glaucinalis were calculated for each of the immature stages. For calculation purposes, we assumed that molting or death occurred at the midpoint between two successive observations. The developmental rate of each immature stage was determined by the temperature summation model, and was estimated by using the reciprocal of the average duration of development (days1), as previously described (Campbell et al. 1974). The overall and stage-specific durations of development were compared
506
JOURNAL OF ECONOMIC ENTOMOLOGY
using an analysis of variance (ANOVA). The mean developmental rates were compared using Turkey’s b test at a 5% level of significance. Survival was compared using a chi-squared analysis. The sex ratios are reported as the proportion of females. All statistical procedures were performed using the SPSS 2007, version 17.0, software (IBM, Armonk, NY). Nine models were used to determine the relationship between temperature and the egg, larval, pupal, and overall developmental rate. The common linear model (Campbell et al. 1974) and the Ikemoto and Takai linear model (Ikemoto and Takai 2000) were used to estimate the low-temperature threshold and thermal constant for each stage. The overall and stagespecific developmental rates positively and linearly correlated with temperature up to 31 C. The omission of the developmental rate at 34 C was required to accurately estimate the low-temperature threshold and thermal constant, and ensure the optimal fit of the linear models (De Clercq and Degheele 1992, Sandhu et al. 2010). The various temperature-related variables were calculated from the models using the Excel 2010 software (Microsoft, Redmond, WA). The common linear model is described by the following equation:
1 Tmin 1 ¼ þ T; D K K
(1)
where 1/D is the rate of development (D is the duration of development), T is the temperature ( C), Tmin is the lower temperature threshold, and K is the thermal constant. Ikemoto and Takai (2000) proposed the following equation representing a revised linear model:
DT ¼ K þ Tmin D;
(2)
where DT and D represent y and x coordinates, respectively, on a straight line, K is the thermal constant, and Tmin is the low-temperature threshold. The standard error (SE) of the low-temperature threshold and thermal constant in the common and Ikemoto and Takai linear models were calculated based of the equations described by Ikemoto and Takai (2001). Seven nonlinear models were used to examine the relationship between the developmental rate (1/developmental periods) and temperature from 19 to 34 C. The nonlinear models were used to estimate the lowand high-temperature thresholds and optimal temperature for each immature developmental stage. The following nonlinear models were used in our analysis: one by Davidson (1944), one described by Logan et al. (1976), two described by Lactin et al. (1995), two described by Brie`re et al. (1999), and one described by Ikemoto et al. (2013). The nonlinear models are described in detail in Appendix 1. The analyses using the six of the nonlinear models were performed using the nlinfit function in the MATLAB, version 7.0, program (MathWorks 2005, Natick, MA), whereas the analysis using the Sharpe–Schoolfield–Ikemoto (SSI) model (Ikemoto et al. 2013) was performed using the
Vol. 108, no. 2
R, version 2.11.1, statistical software (R Foundation for Statistical Computing 2010, Vienna, Austria). The results of the regression analyses for the nine models were plotted using the Sigma Plot 2006, version 10, software (Systat, San Jose, CA) and the R software. Temperature-Dependent Survival Model. The survival rate was calculated for each stage by dividing the initial number of individuals at each stage by the number of individuals which developed successfully to the next stage (Son and Lewis 2005). The extremevalue distribution function (Kim and Lee 2003, Son and Lewis 2005, Papanikolaou et al. 2013) was used to describe the effect of temperature on survival during each life stage as follows:
SðTÞ ¼ keð1þðbTÞ=qeðbTÞ=qÞ ;
(3)
where b is the temperature ( C) at which maximum survival occurs, S(T) is the percent survival at the given temperature (b), k is the maximum survival rate, and q is a fitted constant. The estimates of the model parameters for survival were also obtained using the nlinfit function in MATLAB, and the results of the extremevalue distribution function for overall development was plotted using Sigma Plot. Lowand High-Temperature Threshold Estimations. The low-temperature threshold was defined as the temperature at or below which no measurable development was detected (Howell and Neven 2000). The low-temperature threshold was estimated from the linear models as the intercept of the development line on the temperature axis. The lowtemperature threshold was calculated directly from the Briere-1 and Briere-2 model equations (Brie`re et al. 1999). The low-temperature threshold was calculated from the Lactin-2 model, as described (Jandricic et al. 2010). The high-temperature threshold was defined as the temperature at or above which further development did not occur (Kontodimas et al. 2004). With the exception of the logistic model, the high-temperature threshold was calculated directly from the nonlinear model equations. Although the high-temperature threshold could be calculated from the Lactin-2 model, it failed to produce a high-temperature threshold at which the developmental rate was zero. The high-temperature threshold was calculated from the Lactn-2 model, as described previously (Jandricic et al. 2010). Optimal Temperature Estimation. The optimal temperature (Topt) was defined as the temperature at which the maximum developmental rate occurred (Aghdam et al. 2009). In the Lactin-1, Lactin-2, and Logan-6 nonlinear models, the optimal temperature for development was calculated using the following equation, as previously described (Logan et al. 1976):
Topt
lnðe b0 Þ ; ¼ Tm 1 þ e 1 e b0
where e ¼ DT Tm and b0 ¼ q Tm [4b and 4c].
(4a)
April 2015
LIU ET AL.: EFFECTS OF TEMPERATURE ON O. glaucinalis DEVELOPMENT
In the Briere-1 and Briere-2 models, the optimal temperature was calculated using the following equation, as previously described (Brie`re et al. 1999):
0
Topt ¼
1 2mTL þ ðm þ 1ÞT0 @ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi A þ 4m2 TL 2 þ ðm þ 1Þ2 T0 2 4m2 T0 TL 4m þ 2
;
(5) where TL is the lethal temperature (high-temperature threshold), T0 is the low-temperature threshold, and m is an empirical constant. Model Selection. An information theoretic approach was used for the model selection to determine which model provided the best estimates of the temperature thresholds (Burnham and Anderson 2002). The models were ranked from the best to worst was based on the values of Di and wi. We also used the model-averaged parameter estimates and unconditional standard error (SE) for the optimal temperature and the low- and high-temperature thresholds (Burnham and Anderson 2002). The Akaike information criterion (AIC) and the corrected Akaike information criterion (AICc) were defined by the following equations, as previously described (Burnham and Anderson 2002):
AIC ¼ 2ðlog likelihoodÞ þ 2K;
(6)
and 2KðK þ 1Þ ; AICc ¼ 2ðloglikelihoodÞ þ 2K þ ðN K 1Þ
(7) where N is the number of observations, K is the number of model parameters, and the log likelihood of the model reflects the overall goodness of fit of the model based on the data in the statistical output. The change in the Akaike information criterion (Di) was calculated to assess the goodness of fit of each model, relative to the other models evaluated. A value of Di < 2 suggests substantial evidence for the model, and values between 3 and 7 suggest considerably less support, whereas a Di > 10 indicates that the model fits the data poorly (Burnham and Anderson, 2002, Mazerolle 2006). The change in the Akaike information criterion was calculated as follows (Mazerolle 2006):
DAICi ¼ Di ¼ AICi minAIC:
(8)
The Akaike weight (wi) was used to provide another measure of the goodness of fit for each model. Therefore, the model that best fit the data set had the lowest Di and the highest wi. The Akaike weight of the model after i iterations was defined as follows:
expðDi =2Þ : wi ¼ PR r¼1 expðDr =2Þ
(9)
If the parameter of interest did not appear in all of the models, we recalculated the Akaike weights for the models before performing the model averaging. The
507
Table 1. The developmental time (mean 6 SE) of the different stages of O. glaucinalis at six temperatures Temperatures Egg ( C) 19 22 25 28 31 34
Larval
Pupal
Total
12.15 6 0.117a 124.15 6 1.830a 28.42 6 0.755a 164.72 6 1.714a 8.32 6 0.065b 87.14 6 0.621b 17.54 6 0.442b 113.00 6 0.889b 6.61 6 0.133c 66.83 6 0.703c 15.65 6 0.148c 89.09 6 0.761c 5.15 6 0.040d 59.03 6 0.600d 11.25 6 0.196d 75.43 6 0.585d 4.67 6 0.077e 46.98 6 0.761f 10.25 6 0.205d 61.90 6 0.829f 4.83 6 0.099e 51.02 6 0.725e 11.13 6 0.352d 66.98 6 0.827e
Values followed by the same letter in the same column are significantly different (P < 0.05; Tukey’s-b test).
common linear model, which had the lower log likelihood and the least number of parameters, was used to estimate the low-temperature threshold only. Therefore, the model-averaged Tmin 6 SE estimates were independent of the model-averaged Topt and Tmax 6 SE estimates. If one of the models was clearly the Kullback–Leibler best model (K-L best model), as indicated by a wi value 0.90, the inference was supported conditionally. However, in cases where no single model was clearly superior, if the predicted value (hi ) differed markedly across the models (modelR vs. modelR-1), the inference was not supported (Burnham and Anderson, 2002). The estimate of temperature threshold for each model was then weighted by the Akaike weights as follows (Burnham and Anderson 2002, Mazerolle 2006):
P h^ ¼ Ri¼1 wi ^h i ;
(10)
where h^ denotes a model-averaged estimate of h. The unconditional SE was calculated for the temperature thresholds to evaluate the effect of the model selection uncertainty, as previously described, based on the following equation (Burnham and Anderson 2002, Mazerolle 2006).
Unconditional SE ¼
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi c ^h i jgi Þþð^h i h^ Þ2 ; varð i¼1 wi
PR
(11) c ^h i jgi Þ represents the variance of the estiwhere varð c ^h i jgi Þ the mated ^h i given the gi of the model and varð SE of the ^h i squared. Based on the model-averaged estimates and unconditional SE, confidence intervals were also used to assess the magnitude of the effect (Burnham and Anderson 2002, Mazerolle 2006). Upper 95% confidence limit ¼ estimate þ ð1:96Þ SE ½12 and; Lower 95% confidence limit ¼ estimate ð1:96Þ SE ½13;
Narrow intervals indicate precise estimates. Results Developmental Time. The durations of development for the immature developmental stages of O. glaucinalis at the six temperatures evaluated are listed in Table 1. The one-way ANOVA showed that
508
JOURNAL OF ECONOMIC ENTOMOLOGY
Vol. 108, no. 2
Table 2. The survival (mean) rate of O. glaucinalis at six temperatures Temperatures ( C)
Survival rate (%) Egg
Larval
Pupal
Total development
19 22 25 28 31 34
73.60% (n ¼ 131) 83.22% (n ¼ 124) 87.26% (n ¼ 137) 83.47% (n ¼ 207) 82.32% (n ¼ 163) 81.55% (n ¼ 137)
77.10% (n ¼ 101) 78.23% (n ¼ 97) 81.75% (n ¼ 112) 81.64% (n ¼ 169) 76.69% (n ¼ 125) 74.45% (n ¼ 102)
88.12% (n ¼ 89) 96.91% (n ¼ 94) 97.32% (n ¼ 109) 89.94% (n ¼ 152) 82.40% (n ¼ 103) 80.39% (n ¼ 82)
50.00% (n ¼ 89) 63.09% (n ¼ 94) 69.43% (n ¼ 109) 61.29% (n ¼ 152) 52.02% (n ¼ 103) 48.81% (n ¼ 82)
Female ratio
53.93% 52.13% 55.96% 50.66% 47.97% 42.68%
n, number of individuals. Table 3. Parameter estimates (mean 6 SE) of temperaturedependent survival model for O. glaucinalis Parameter Egg k b q R2
Larval
Pupal
Total development
85.99 6 1.91 81.11 6 1.18 95.79 6 0.55 66.71 62.71 27.1055 6 0.9875 25.2713 6 0.7490 23.818 6 0.9738 24.8939 6 0.5228 16.7506 6 3.2268 19.2332 6 3.2037 13.9435 6 2.5107 8.9412 6 1.0739 0.8142 0.837 0.8719 0.9077
The values of mean 6 SE are estimated by temperature-dependent survival model.
temperature significantly affected the duration of development from 19 to 31 C in each of the developmental stages evaluated, with the exception of pupal development at 28 and 31 C. Significant increases in the larval and overall duration of development were recorded at 34 C. The mean proportion and 95% confidence interval (CI) of the durations of development for the egg and pupal stages were 7.29% (95% CI: 7.03–7.55%) and 16.41% (95% CI: 15.34–17.47%) of the duration of overall development, respectively, whereas the duration of development for the larval stage was 76.30% (95% CI: 75.05–77.56%) of the duration of overall development, regardless of the temperature. As shown in Table 2, the overall survival and survival of pupal stage were significantly affected by temperature (pupal stage: v2 ¼ 26.558, df ¼ 5, P ¼ 0.00007; overall development: v2 ¼ 13.63, df ¼ 5, P ¼ 0.018). Eggs hatched at all of the temperatures evaluated. The survival rate for the larval stage was slightly lower than that of the other developmental stages, and none of the larvae completed development at 37 C. The overall survival for the immature stages was highest at 25 C, and was lowest at 34 C. Model Evaluation. The parameter estimates (mean 6 SE) of the temperature-dependent survival model for eggs, larvae, pupae, and all immature stages of O. glaucinalis are shown in Table 3. The extremevalue function adequately described the relationship between temperature and survival rate for all stages. Because the O. glaucinalis were more susceptible to higher temperatures, the pattern of the bell-shaped curves for overall survival was skewed (Fig. 1). The values of R2 were > 0.81 for each developmental stage and for overall development. The parameters estimated in the regression analysis or calculated by solving the equations are listed in Tables 4–6 and Supp Table 1 [online only]. The lowtemperature threshold and thermal constant were
estimated using the linear models. The rate of development increased linearly with increasing temperature from 19 to 31 C, as evidenced by the high R2 values (R2 > 0.944, P < 0.001), which indicated superior goodness of fit for the linear models. The lowest lowtemperature threshold (11.33 C) and the highest thermal constant (939.66 degree-days) were observed for the larval stage using the common linear model. According to the common linear model and the Ikemoto and Takai linear model, the low-temperature threshold and thermal constant for overall development estimated using the common linear model and Ikemoto and Takai linear model were similar. However, the Ikemoto and Takai linear model had a higher R2adj value than the common linear model. In addition, the Ikemoto and Takai linear model provided a slightly higher value for the low-temperature threshold and a slightly lower thermal constant, compared with the common linear model. The Ikemoto and Takai linear model consistently produced more narrow estimates of the SE of the low-temperature threshold and thermal constant for each of the developmental stages, compared with the estimates based on the common linear model. The high-temperature threshold could not be estimated by either linear model because the curve did not intersect the x-axis at higher temperatures. The values of the measurable parameters, AICC, D i, and wi, are presented in Tables 5 and 6. The modelaveraged temperature threshold estimate and the unconditional SE for the various immature developmental stages are showed in Table 7. The influence of temperature on the overall developmental rate for the nine models is presented in Figure 2. With the exception of the SSI model, the values of AICC, D i, and wi were used to determine the goodness of fit for the nonlinear models. The Lactin-1 model seems to closely fit the data for larval, pupal, and overall development based on the lower AICC and D i, and higher wi. The Briere-1 model fit the data for the egg stage better, with the lower AICC and D i, and higher wi. The model-averaged lowtemperature threshold and unconditional SE estimates for egg, larval, pupal, and overall development obtained using the common linear model were similar. A similar trend was observed in the model-averaged optimal temperature and unconditional SE for larval, pupal, and overall development. The model-averaged high-temperature threshold and unconditional SE for larval, pupal, and overall development were similar to those obtained using the Lactin-1 model, whereas those for egg development were similar to those of the Briere-1 model.
April 2015
Fig. 1.
LIU ET AL.: EFFECTS OF TEMPERATURE ON O. glaucinalis DEVELOPMENT
509
Temperature-dependent model of survival rate of O. glaucinalis for total immature development.
Table 4. Low temperature threshold (mean 6 SE) and thermal constant (mean 6 SE) of immature stages of O. glaucinalis estimated by two linear temperature-driven rate models at 19–31 C temperature ranges Stage
Method
Egg Larval Pupal Total Immature
Linear regression
Common Ikemoto and Takai Common Ikemoto and Takai Common Ikemoto and Takai Common Ikemoto and Takai
Equation
R2adj
P
1/D ¼ 0.129 þ 0.011T DT ¼ 86.169 þ 11.850D 1/D ¼ 0.012 þ 0.001T DT ¼ 928.890 þ 11.484D 1/D ¼ 0.062 þ 0.005T DT ¼ 180.698 þ 12.590D 1/D ¼ 0.009 þ 0.001T DT ¼ 1,203.584 þ 11.616D
0.981 0.991 0.981 0.987 0.944 0.967 0.994 0.995
