Food induced variation of thermal constants of development and growth of Autographa gamma (Lepidoptera: Noctuidae) larvae

The development stages of a species may have an identical lower development threshold (LDT) and proportionally dif­ ferent durations. This phenomenon called “rate isomorphy” (RI) has been demonstrated for a number of insect species. In contrast, the growing day degrees accumulated over the period of larval development (sum of effective temperatures SET) should be plastic and vary with environment conditions. The prediction from RI is that, with changing conditions, the uniform LDT should be accom­ panied by differences in development time which remain proportional at different temperatures. This was tested by investigating the effect of diet on thermal requirements for development of larvae of the polyphagous species Autographa gamma (L.) (Lepidoptera: Noctuidae). The larvae were kept at 15.0, 20.3 and 26.7°C and fed on leaves of 13 dicotyledoneous herb and tree species. The pro­ portion of total development time spent on a particular diet was plotted against temperature. The existence of RI was inferred from a zero change in development time proportion with changing temperature. This rigorous test supported RI for 3 of 9 diets where devel­ opment was completed in all temperatures. The LDT observed on 11 diets where the larvae completed development in at least 2 tem­ peratures varied between 9.3 and 11.0°C while SET varied between 167 and 353 day degrees (dd). Assuming RI, LDT and SET for those 9 diets were recalculated. The recalculated LDT was 10.0°C and SET varied between 177-257 dd. The SET increased with decreasing water content and decreasing nitrogen content of food. Worsening food quality decreased food consumption, metabolic and food conversion efficiency, and the relative growth rate of the larvae. Increasing metabolic costs of development were thus posi­ tively correlated with SET. The standardized rate of growth (mg.dd-1) was typical for particular diets. Pupal mass decreased with increasing temperature and, within each temperature, with development length.


INTRODUCTION
In insects the development rate (a reciprocal of devel opment duration) increases with temperature.The increase is linear over a range of ecologically relevant temperatures, although at temperatures approaching the lower point where development ceases and at tempera tures approaching the upper limits of thermal tolerance, the relationship may become non-linear.The linear rela tionship enables the calculation of thermal constants, such as the lower development threshold (LDT), which is a temperature at which the development rate is zero, and the sum of effective temperatures (SET), which is equal to the amount of growing day degrees (dd) needed to complete a development stage.
A huge quantity of data (reviewed by Honek & Kocourek 1990, Honek 1996, Kiritani 1997) became available over the past eighty years, during which time the temperature effects on insect development have been intensively studied.The data revealed not only a large variation of thermal constants between species, but also among populations within species.However, these differ ences are not only caused by biological variation, but also due to bias in the experimental data.This bias appears even in carefully performed experiments and becomes enormously important when comparing the data of dif ferent authors.A recently developed method (Jarosik et al., in press) enabled a test to be made of the relevance of differences in LDT.The method consists of testing the hypothesis of "rate isomorphy"."Rate isomorphy" means that, with changing temperature, the development stages of a population of a species take constant proportions of total development time.Then the LDT is identical for all development stages.A comparison of published data for 342 species from 11 insect orders revealed "rate isomor phy" in 57% of populations.In the rest of the data viola tion of the "rate isomorphy" principle was very small.Thus LDT is probably uniform in developmental stages within populations and species, and probably also in taxonomically related groups of species (Dixon et al., 1997).In contrast to LDT, SET is plastic and reflects the varia tion in environment conditions other than temperature, including food quality of the growing stage (larva), humidity, photoperiod etc. (Squire & Trudgill, in litt.).
With testing for "rate isomorphy" we may prove the identity of LDT of developmental stages but not to estab lish its correct value.A method of calculating the true LDT value may consist in looking at variation of develop ment times of larvae of a highly polyphagous species fed with several diets.A common development threshold may be calculated using a modified application of the "rate isomorphy" principle (described in the Material and Methods).Thermal constants for development were there fore established in larvae of Autographa gamma (L.), an extremely polyphagous noctuid species occurring throughout the Europe which periodically becomes a pest of several agriculture crops.The species has 1-3 genera tions per year, according to weather conditions.Its devel opment was first described by Ostreykowna (1924) who first supposed overwintering in central Europe.Cayrol (1962, 1965) and Novák (1971, 1972, 1988) revealed that the central European populations consist of two constitu ents, a part overwintering in the 4th larval instar and a part immigrating each spring from the southern Europe.The rate of larval development is influenced by tempera ture and humidity (Kozhantshikov, 1939;Hill & Gate house, 1992), population density (Cayrol, 1957;Long, 1953Long, , 1955Long, , 1959) ) and food (Novák, 1960, 1974, Cayrol, 1962;).The widely polyphagous larvae may eat the leaves of more than 200 host plant species (Schwitulla, 1963).The preferred food plants change with the course of the season as a consequence of host plant senescence (Novák, 1960(Novák, , 1974;;Steudel, 1963).In addition to the wide range of natural host plants, under laboratory conditions the larvae eat many plant species not accepted in the open.The food induced plasticity of development rate provides an opportunity for studying the variation of thermal con stants.By looking at parallels in variation of growth rate and food assimilation under different temperatures, we may determine the causes of variation in SET.
In our experiments we tested the hypothesis of "rate isomorphy" in A. gamma larvae provided with leaves of several host plant species.Development time, pupal mass, and food assimilation were studied at 3 experimental tem peratures.The data enabled (1) the determination of the thermal constants of larval development, (2) testing for "rate isomorphy", (3) the establishment of variation in body growth under different trophic conditions and (4) calculations of the relationship between food assimilation and larval development.

MATERIAL AND METHODS
Rearing of larvae.Adult A. gamma moths were collected at Prague Ruzyně (50o06'N, 14o15'E), in August, 1988.The moths flying in an alfalfa stand were captured by use of entomology net when feeding on flowers.Groups of three female and three male moths were placed into cardboard cylinders of 15 cm diameter and 15 cm height covered by a glass lid and provided with a diluted commercial mixed fruit syrup.This source of water and carbohydrates was offered in a petri dish (7 cm diam.)covered with a thin stiff nylon fabric (1.5 mm mesh size) which prevented the moths from drowning.Eggs were laid overnight, on the side-walls of the cylinder.Each day the moths were moved to a new cylinder, the pieces of cardboard with groups of eggs were cut out and until hatching placed into 3.5 x 12 cm glass tubes covered with dense nylon fabric.The freshly hatched larvae were removed at 24 h intervals and used for experiments.Adults and eggs were placed until hatching at room temperature (25 ± 1°C) and natural photoperiod.
(Salicaceae), Urtica dioica L. (Urticaceae).Experiments were made in August (development duration) and September 1988 (food assimilation).Water and total nitrogen content of leaves were measured in early September.For each host plant species c. 200 g of fresh leaves were weighed and dried to a constant mass at 90°C.The dry mass and nitrogen content of each sample was then determined and the water content of the fresh leaves calculated.The determination of N content was made commer cially, by a laboratory using a standard (Kjeldahl) method.
Development time and pupal mass.Cohorts of 0-24 h old larvae hatched at room conditions were transferred to constant temperatures of 15.0, 20.3 and 26.7°C and 18L: 6D photoperiod where the larvae were kept in 3.5 x 12 cm glass tubes covered with dense nylon fabric.First and second instar larvae were reared in groups of 20, and at the end of the 2nd instar the larvae were placed in individual tubes and then kept alone until pupa tion.The larvae were provided with leaves of the 13 host plant species mentioned above.Fresh leaves were supplied and faeces removed every 2 or 3 (weekends) days.Pupation was observed daily, at 07:00 and 19:00.The pupae were sexed and their live body mass was established within 24 h from pupation, with an accuracy of 0.1 mg.Ten randomly selected pupae were killed, dried to constant mass and their dry matter content was deter mined.
Food utilization.The cohorts of 0-24 h old larvae hatched at room conditions were isolated.The larvae were then kept in 500 ml glass cylindric vials covered with dense nylon fabric.Until the 4th instar the larvae were reared in groups of 15-20 per vial, kept at room conditions (25 ± 1 °C, natural photoperiod) and in 2 or 3 d (weekends) intervals supplied with fresh T. officinale leaves.The larvae of early 4th instar were removed from food and starved for 24 h to empty the gut contents.The larvae were then divided into groups of 5 randomly selected individuals and the groups of larvae were weighed.Each group was then placed into a 500 ml vial and supplied with a weighed quantity of fresh leaves.The leaves were collected within 4 h of establishing the experiment and maintained in plastic sacks to prevent desicca tion during handling.Dry matter content of these leaves was determined in 5 samples randomly selected from the leaf supply of each plant species.The vials with larvae and food were then put in constant temperatures of 15.0, 20.3 and 26.7°C and 18L : 6D photoperiod.The larvae were allowed to feed on the leaves for 2 d (at 26.7°C), 3 d (20.3°C) or 5 d (15.0°C).At the end of this feeding period the larvae were removed from food, starved for 24 h and weighed.Dry matter content of the larval body was determined by weighing the fresh and dry mass of 10 randomly selected larvae.For each group, the remaining non-consumed leaves (food) and faeces produced during feeding and post feeding starvation period were collected, dried to constant weight and weighed.Data elaboration.In each temperature T, larval development length D was measured as number of days elapsed from the egg hatching (isolation of the 1st instar larvae cohort) until larvalpupal ecdysis.Development rate R was calculated as D-1.The regression R=a+bxT (where a and b are constants) was calcu lated.From here lower development threshold (°C) was calcu lated as LDT= -bxa-1 and sum of effective temperatures (day degrees dd) as SET=b-1.Standard deviations of LDT were calcu lated according to Janáček et. al (in litt.).Multiple regression of SET on water and N content was calculated using Statistica® (StatSoft, 1994).High SET for poor diets that had been calcu lated from only 2 temperatures were included because they increased the significance of the regression.The mass of neo nate larvae was below the limits of weighing precision and thus negligible compared to pupal mass PM.The growth rate GR Fig. 1.An illustration of the effect of temperature (t) and three different diets on rate of development (R) of larvae within the linear range of the relationship between R and t.The larvae are isomorphic, and at each temperature (t1, t2, t3), 1/6 of the total time spent on the three diets is spent on diet 1, 1/2 on diet 2, and 1/3 on diet 3. Larvae on all three diets have a common lower developmental threshold (LDT).average body mass increment per day) was therefore calculated as GR=PMXD-1.Standardized growth rate SGR (average body mass increment per day degree) was calculated as SGR=PMXSET-1.SGR compensated for differences in growth rate caused by temperature and revealed the variation caused by food.In food utilization experiments, dry mass of ingested food (F), dry mass of excrement (E) and dry body mass increments (I) were calculated.These values were used to calculate the indices of food assimilation, efficiency of food conversion ECI=IXF-1 and metabolic efficiency ECD=IX(F-E)-1.Relative growth rate was calculated as RGR=exp((lnWf-lnWi)xd"1)xWi"1 where Wi is initial mass and Wf final mass of an individual larva in the feeding experiment, and d is duration of the experi ment (days) (Barbehenn et al., 1999).Means are accompanied by ± SE throughout the paper.
Testing for "rate isomorphy".Rate isomorphy implies no change in the proportion of time spent on a particular diet with change in temperature (Fig. 1).Therefore, within the range of the linear relationship between development rate R and tempera ture t, the consequence of rate isomorphy is common LDT for larvae on all diets, in spite of different development rates on each diet.The prediction of rate isomorphy, namely uniform LDT accompanied by proportional variation of SET, was tested by designating the arcsin Jproportion of the time spent on par ticular diet as the response variable, and temperature and diet as factors.The data were analysed by two-way ANOVA with tem perature and diet as fixed effects.The existence of rate isomorphy was inferred from a zero change in the development time proportion at changing temperature, and the existence of variation in SET from differences in the development time pro portions on individual diets.A zero change in the development time proportion at different temperature accompanied by differ ences in the proportions on individual diets indicated rate isomorphy.Different development time proportions on individual diets accompanied by variation in the proportions at different temperatures violated the assumptions of rate isomorphy.

RESULTS
Thermal constants of development.The duration of larval development decreased and development rate increased with increasing temperature (Tables 1 and 2).Mortality was generally high, and was not correlated with temperature or kind of food.This might be caused by dis eases introduced to laboratory cultures by food contami nated in the open.Thermal constants were calculated for development on 11 host plant species of which 9 per mitted complete larval development at 3 temperatures while with 2 host plant species development was com pleted at 2 temperatures.Experimental LDT for all spe cies were similar (9.3-11.0°C,average 9.9 ± 0.2°C) (Table 2).As development rate was largely affected by food, larvae fed with leaves of some host plant species extended their development time by up to 3 times com pared that found with optimum food.Consequently, the variation in SET was large and the slow growing larvae  fed with S. babylonica leaves had 2.3 times greater SET (353 dd) than larvae fed with T. officinale (167 dd) (Table 2).
Rate isomorphy.Proportion of the total development time spent on a particular diet significantly interacted with temperature and type of diet (ANOVA: F = 9.236; df = 16, 225; p < 0.001).At particular temperatures, the pro portion significantly varied with diet (Table 3).This con firms that the SET is very plastic and changes with food quality.The proportion did not significantly vary with temperature for larvae reared on Medicago sativa, Pastinaca sativa and Taraxacum officinale (Table 4).It means that on these diets, the variation of SET is propor tional with changing temperature.The larvae fed by these diets have a common LDT and their development is thus isomorphic.The development time proportion varied with temperature on other diets.Larval development thus vio lated rate isomorphy.However, the variation in larval development time on particular diets at different tempera tures (Table 4) was much less than the variation on dif ferent diets at any given temperature (Table 3).The violation of rate isomorphy thus appeared to be generally negligible and the differences in LDT for larvae reared on different diets were small compared to differences in SET  (Table 2).The data for 9 diets where development rate was established in 3 experimental temperatures were recalculated assuming rate isomorphy.The recalculated LDT was 10.0°C (identical with the average LDT calcu lated as the arithmetical mean of the experimental data for these foods).The recalculated SETrec differed by up to 30 dd from SET calculated from experimental data (Table 2).
Effect of food quality.The SET increased with decreasing water content and decreasing nitrogen content of food (Fig. 2).Multiple regression analysis indicated a significant contribution of variation in food dry matter content (t(8) = 3.152, p = 0.014) while the effect of nitrogen content was below the limit of statistical signifi cance (t(8) = -2.011,p = 0.079).The increase of SET was paralleled by decreasing efficiency of food assimilation  which also was largely determined by food quality.ECD and ECI (Fig. 3) decreased with decreasing water content in food.SET established on 4 host plants were conse quently significantly correlated with ECI and ECD (Fig. 4).Food ingestion and RGR were also negatively influ enced by decreasing water content of the food and RGR significantly increased with ECD and ECI (Fig. 5).By contrast, the effect of temperature on larval metabolism was not significant when data were pooled across foods.Food consumption, growth rate, as well as metabolic effi ciency ECD and efficiency of food conversion ECI (Fig. 6) did not significantly change with temperature.
Body growth.Pupal mass increased with temperature and, within each temperature, decreased with the length of development time (Fig. 7).Pupal mass was thus deter mined by growth rate GR which significantly increased with temperature (Fig. 8).Recalculation of data to SGR which removed the effect of temperature revealed that pupal mass increased with SGR and the increase was typical for particular food types (Fig. 8).The increasing SET was inversely proportional to decreasing SGR (Fig. 9).

Constraints on LDT.
The prediction of uniform LDT derived from the assumption of rate isomorphy (RI) was confirmed only on 3 from 9 diets where development was completed in all temperatures.It is not surprising because we had available only 3 experimental temperatures.Even development rate (Campbell et al., 1974).The low preci sion of LDTs is obvious from their large standard errors (Table 2).
The largest departure from expected proportions assuming the existence of rate isomorphy is at the lowest  or the highest temperatures (Jarosik et al., in prep.).There are three reasons why these temperatures may violate rate isomorphy.At low temperature there may be differential mortality.The individuals with the fastest development complete their development early while the rest succumb to adverse conditions, the more so if their development is prolonged.The second reason is an imprecise measuring of developmental time at high temperatures.As develop mental time decreases with temperature, the number of observations per stage also decreases if monitoring is made at constant intervals at low and high temperatures.The third reason is crucial from a statistical point of view.An important determinant of the slopes of the linear regressions, from which the LDTs are inferred, are the highest and the lowest values (see Crawley 1993, p. 78-82).Therefore, a relatively small bias in the develop mental rates measured at the highest or the lowest tem peratures will cause a large shift in the LDT.
The common LDT for A. gamma larvae calculated using data for 11 host plant species was 9.9 ± 0.2°C, by 2.3°C higher than the LDT calculated for a UK popula-Fig.9. Standardized growth rate (mg/dd) in relation to SET for different host plants (data pooled over temperatures).Regression for average value for each host plant: growth rate = 2.9 -0.0073 SET.F = 16.80;df = 1, 9; p < 0.003; R2= 65.1%.Acronyms as in Fig. 8, meaning of points as in Fig. 7. tion of this species (Hill & Gatehouse, 1992) which was 7.7°C (male) and 7.6°C (female) (Table 5).The differ ence, tested using SD calculated for the UK data (Janáček et al., in litt.), was not significant (t2,4 = 1.434, p > 0.05).In fact it is smaller than variation observed between popu lations of other species studied by more than one author, e.g.Heliothis armigera (Hübner) (LDT between 8.6-12.9°C),Heliothis virescens (F.) (7.6-11.8°C),or Trichoplusia ni (Hübner) (9.0-14.7°C)(Table 5).The differ ence was apparently caused by variation of methods used in the two studies.Data on variation of LDT between populations were available for 12 noctuid species (Table 5).If we exclude the outlier data for populations of A. ipsilon (Hufnagel), Pseudaletia unipuncta (Haworth) and Spodoptera frugiperda (Smith) because of the large extent of variation in the differences between local popu lations, which were in the order of 1.7-5.7°C(mean = 3.9 ± 0.4°C).In general differences between the studies of different authors were greater than the differences "within" the studies.Dixon et al. (1997) extended the idea of rate isomorphy (existence of a LDT constraint) to higher taxonomic  groups of similar biology.Noctuid moths may be consid ered in this respect because they have a similar larval biology.To compare the available data we compiled a review of 48 studies on larvae of noctuid species of the temperate zone (Table 5).The included species are all folivorous facultatively polyvoltine phytophagans.In fact, the average LDT for 108 populations of 34 species, 10.2 ± 0.3°C, is surprisingly close to the average LDT of A. gamma established in this study.Thus, despite the enor mous variation of the literature data the assumption of rate isomorphy for the family of Noctuidae could not be rejected.

Plasticity of SET.
The time (or heat) requirements for completing development of a stage as reflected by varia tion of the SET are plastic.At the species level this varia tion parallels taxonomic differences between species, but also trophic specialization of the taxa and the differences in body size (Honek, 1999).Within populations of some species there exist adaptive differences between the sexes in development time (Nylin et al., 1993).However, in many other species the importance of sex linked differ ences is small (Honek, 1997).Probably the most impor tant differences in SET are caused by variation of food quality (Slansky & Scriber, 1985;Slansky, 1993).In this study the differences in food quality caused a 2.6 fold variation in the SET.The leaves of 13 host plant species differed in their water and nitrogen content, which both together explained 62.2% of the variance in SET.This indicates a small effect of secondary factors, leaf surface quality and allelochemicals, on food acceptability for A. gamma larvae.This is apparently consistent with the broad polyphagy of this species which should overcome the defence barriers of a wide range of plant species.
The water content of food is an important factor in food utilization by caterpillars (Slansky, 1993).Its importance has been demonstrated with natural foods (Scriber, 1977(Scriber, , 1979a)), synthetic diets (Schmidt & Reese, 1988) and with stored products of different humidity (Nawrot, 1979;Hagstrum & Milliken, 1988).The effects of water content on food digestibility and larval growth was also demon strated in other insect orders (e.g.Merkel, 1977).The importance of water content of food (47.2% of variance in SET explained) for larval performance of A. gamma thus parallels other polyphagous insect species.The nitrogen content of food explained only 19.3% of total variance in SET of A. gamma.Although N content is an important factor of food quality for Lepidoptera larvae (Scriber 1979b;Lindroth et al., 1991;Bauce et al., 1994;Soontiens & Bink, 1997), its relative importance is appar ently smaller than in sucking insects where nitrogen con tent of phloem or xylem sap may become a limiting factor of larval growth and adult reproductive performance (e.g.Honek et al., 1998;Ponder et al., 2000).
Growth.As with other insects, temperature and devel opment time influenced final size (pupal mass) of the larva.The effect of temperature on body size is unimodal, with a monotonic decrease of body size below and above the "optimum" temperature (e.g.David et al., 1994).The temperature where insects grow to largest body size is always below the optimum for development time and reproduction efficiency.In A. gamma this temperature is less than or equal to 15.0°C, since the average pupal mass decreased over the range of experimental temperatures (15.0-26.7°C).Within each temperature pupal size decreased with increasing development time.There was no trade-off between body size and development time which might be expected under optimum trophic condi tions (Begon et al., 1990).Extended larval development accompanied by lower final mass was apparently associ ated with the inability to compensate for reduced food quality (Slansky, 1993).In fact, the indices of efficiency of food assimilation decreased in parallel with decreasing growth rate and pupal mass.The variation of pupal size thus may appear as a non-adaptive result elicited by an environmental constraint.However, if we consider a wide spectrum of situations that the organism may face in the open, decreasing body size might still be adaptive (Wik lund et al., 1991;Nylin & Gotthard, 1998).Pupating at small size may be advantageous in areas with constrained thermal unit availabilty (Ayres & Scriber, 1994;Scriber, 1996).The study of fitness consequences of body size variation in A. gamma remains to be studied.

Table 1 .
Number of pupated individuals (N), per cent mortality (M), larval development time (D) and fresh pupal mass (W) of larvae kept at 3 constant temperatures and fed with leaves of 13 host plant species.

Table 2 .
Food quality as indicated by per cent water (WC) and total N content (NC) in leaves, regression constants a, b (develop ment rate = a x temperature + b), thermal constants SET (dd) and LDT (°C) calculated from experimental data, and SET recalcu lated under assumption of "rate isomorphy" principle (SETrec).

Table 3 .
ANOVA of the proportion (angular transformation) of the total development time spent on different diets at 15, 20.3 and 26.7°C.

Table 4 .
ANOVA of the proportion (angular transformation) of the total development time spent at different temperatures on indi vidual diets.

Table 5 .
Thermal constants for larval development in noctuid moths of temperate regions.Number of populations investigated in a study (Pn), number of experimental temperatures where development length was determined (N), range of experimental tempera tures (Range), lower development threshold (LDT) and sum of effective temperatures (SET).