It is well known that temperature is the most influencing factor in the development of insects known to be typical poikilothermal animals (Wagner et al., 1984a). Unsurprisingly, temperature significantly affected the development time of A. ipsilon at all stages as seen in Table 1. Our interest was not only in the effect of temperature itself, but also in variations of developmental time among previous reports at comparable temperatures. Previous pupal periods were within the 95% confidence intervals of our estimated line in most cases. The small variation was consistent with previous reports showing that the pupal development period was less affected by environmental factors such as humidity and host plants at the same temperature (El-Kifl et al., 1972; Zaazou et al., 1973; Amin & Abdin, 1997; Nikolov, 1977; Santos & Shields, 1998), although pupal duration for larvae fed lettuce was shorter than that for larvae fed other host plants (Muştu et al., 2021). However, egg and larval periods were somewhat largely various among previous reports as indicated by scattered points outside of 95% confidence limits (see Figs. 2A, 2B).
Especially, egg incubation time showed a very large difference at a temperature of 30 ℃ or higher (Fig. 2A). It was 2.0 d at 30 ℃ and 1.1 d at 35 ℃ in the present study. However, almost all previous studies reported longer incubation time than our results: 3.0 d at 30 ℃ by Hasegawa & Chiba (1969), 3.1 d at 30 ℃ and 2.7 d at 35 ℃ by Fahmy et al. (1973), 2.9 d at 33 ℃ by Blair (1976), 4.0 d at 30 ℃ and 3.1 d at 35 ℃ by Chaudhary & Malik (1980), and 2.81 d at 30 ℃ by Dahi et al. (2009). These differences might have little impact on the phenology prediction of A. ipsilon in the field because the maximum difference was only 2 days that only appeared at high temperatures, which did not occur frequently in the field condition.
The larval period also showed a large variation among previous studies (Fig. 2B). The larval development of insects can be affected by various environmental factors such as temperature, photoperiod, food quality and quantity, rearing density, and humidity (Esperk et al., 2007). Among them, the most important environmental factor was host plant (namely larval diet) when temperature was constant in A. ipsilon (El-Kifl et al., 1972; Zaazouet al., 1973; Busching& Turpin, 1977; Kim, 1991; Amin & Abdin, 1997; Nikolov, 1977; Santos & Shield, 1998; Muştu et al., 2021). Differently from that of eggs, the variation of larval period can significantly affect the accuracy of a phenology model. As reported by Muştu et al. (2021), for example, the larvae A. ipsilon fed on lettuce completed approximately 20 days earlier than those fed on maize. This difference might be large enough to cause control timing to fail since the prediction of cutting larval instars is critical for successful control of A. ipsilon. In comparison with previous studies, partial or all data points from four study cases were placed outside the lower confidence limit (namely longer development time) of our estimated curve (Fahmy et al., 1973; Blair, 1976; Archer et al., 1980; Santos & Shields, 1998). Whereas studies of Rivnay (1964), Olufade (1972), Chaudhary & Malik (1980), Kaster (1983), and Dahi et al. (2009) showed a similar trend as ours. Therefore, it seems necessary to consider host plants in the field application of a phenology model.
The damage of A. ipsilon to maize is a major concern perhaps because maize is an important food crop. Thus, maize was one of the most preferred host plants by many researchers (Archer et al., 1980; Kaster, 1983; Santos & Shields, 1998; Mushtaq et al., 2021). However, maize is not an optimal host plant of A. ipsilon for biological performance. In comparative many studies, maize was the least suitable host in terms of larval period (i.e., the longer the period, the less suitable the host plant) as reviewed by Kaster (1983). The shortest larval period was 20.9 d on alfalfa (Medicago sativa L.). It was longer at 22.3 d on maize at 22.5 ℃ (El-Kifl et al., 1972). The most extended larval duration was on maize (29.4 d) among nine host plants (Nikolov, 1977). The most rapid larval development (approximately 24.0 d) was on wheat (Triticum aestivum L.) or oats (Avena sativa L.) compared to maize (27.0 d) (Busching & Turpin, 1977). Furthermore, Amin & Abdin (1997) have reported the shorter larval period of 23.9 d on Egyptian clover (Ttifolium alexandrinum L.) but longer at 28.2 d on maize at 27.0 ℃. Data of Archer et al. (1980) and Santos & Shields (1998) obtained from feeding on maize apparently showed longer larval development time as seen in Fig. 2B (host plant not available in Fahmy et al., 1973; Blair, 1976). However, not all A. ipsilon populations had a long development period when fed on maize. For instance, larval developments from the study of Kaster (1983) were well matched with our results despite rearing on maize seedling leaves. Also, Kim (1991) reported larval development on various crops at 25 ℃ without delayed development time on maize: 23.64 d on Kimchi cabbage (Brassica campestris L.), 24.06 d on maize (Zea mays L.), 25.16 d on radish (Raphanus sativus L.), 25.59 d on welsh onion (Allium fistulosum L.), 27.56 d on red pepper (Capscium annuum L.), 29.26 d on sweet potato (Ipomoea batatas (L.) Lam.), and 33.67 d on soybeans (Glycine max (L.) Merrill). The biological response of this moth to its host plant seems to vary by regional populations. In Korea, our development models including degree-day details will be able to be used at least in the fields of Kimchi cabbage, maize, radish, and welsh onion since its development time does not appear to have much variation among those crops as shown in the report of Kim (1991).
As described in Materials and Methods, the larvae of A. ipsilon have six basic instars. However, sometimes they have seven or eight instars. In rare cases, they have five or nine instars. Therefore, the sum of the average development time for each instar may overestimate the actual average larval development time. That is, when the average development time of larvae is calculated by the sum of the average development time for each instar, an error can occur when adding the development time of the seventh instar obtained from a separate individual even though the development is completed at the sixth instar. Therefore, when individuals that complete development at the sixth or seventh instar are mixed, the actual average development time of larvae should be calculated as the development time based on each individual. Because individual-based data sets are unavailable, we combined the sixth instar and remaining instars using the development time up to completion of the fifth instar. A common lower threshold temperature was applied to calculate degree days for each stage. Applying the common lower threshold temperature, it can be very convenient to track the phenology of each stage of A. ipsilon because degree days can be calculated on a single scale. Total degree days (eggs to adults) of the present study were exactly the same as those of Archer et al. (1980), although there were considerable differences in egg, sixth, and remaining instars and pupal stage. Also, total degree days were much higher in Luckmann et al. (1976) than in ours. Average total degree days for the completion of the period of eggs to adults were 588.3 DD (average of Luckmann et al. (1976), Archer et al. (1980) and present study), including 50.8 DD for eggs, 39.9 DD for the first instar, 33.8 DD for the second instar, 33.7 DD for the third instar, 38.8 DD for the fourth instar, 51.9 DD for the fifth instar, 149.9 DD for the sixth and remaining instars, and 189.3 DD for pupae. This information for the degree days as well as the lower threshold temperature and thermal constant for each stage (Table 2) will be useful for understanding the phenological biology of A. ipsilon.
The best equation for the development models was selected on the basis of various criterion as described in Materials and Methods. Finally, the Lactin model with three parameters (Lactin et al. 1995) was adopted as the best model for eggs, larvae, and pupae. This Lactin model was also applied to each instar by omitting the evaluation process of the model selection criteria to prevent proliferating the number of equations because it was already examined for total larvae. Although the Lactin model had a small number of parameters, its ability to fit asymmetric insect development data was excellent as in other cases (Choi and Kim 2014; Hyun et al. 2017; Kim and Kim 2018).
Unlike the degree day models shown above, the stage transition models could project stochastically the proportion of individuals that completed the development to the next stage from a specific cohort, in which the model of development rate was incorporated into the distribution model of development time using environmental temperature. That is, the development rate was used to calculate the mean (or median) rate of development per day at a given temperature and daily development rates were accumulated to produce the physiological age. Finally, the function of the distribution model determines the cumulative proportion of cohort development at a given physiological age (Curry et al. 1978a, b; Wagner et al. 1985; Kim et al. 2001). For a stochastical projection of A. ipsilon populations in variable temperature conditions (i.e., in the field), development rate can be accumulated by daily air temperatures. It can be used as input value to the distribution function, which can arrange transition proportions through time. We provided all parameters for the development rate models and distribution model of development time in the immature stages of A. ipsilon. These stage transition models could be used to construct a population model for A. ipsilon that could simulate its phenology pattern and population dynamics in the field.