Effect of microclimatic temperatures on the development period of 3 rice planthopper species (Hemiptera: Delphacidae): a phenology model based on field observations

Abstract Most pest phenology models are temperature dependent. Generally, the air temperature at reference height is used to predict pest development, but the air temperature varies between inside and outside the crop canopy, where pests reside. Here, we sampled 3 rice planthopper species—Nilaparvata lugens (Stål), Sogatella furcifera (Horváth), and Laodelphax striatellus (Fallén)—and micrometeorological observations in paddy fields to analyze how thermal environments inside the canopy affect pest development. Seasonal variations in the population density of these species were surveyed in 3 experimental fields with 2 water temperature plots (normal and low-water temperature plots). The development periods of the 3 species were predicted individually based on pest phenology models using temperatures recorded at 6 heights (0.0–2.0 m). We calculated the root mean square error (RMSE) values from the predicted and observed development periods for each rice planthopper. The development prediction using the temperature inside the canopy was more accurate than that utilizing the temperature at the reference height (2.0 m). In the low-water temperature plot, the RMSE value for N. lugens, S. furcifera, and L. striatellus was 6.4, 5.6, and 4.1 when using the temperature at the reference height (2.0 m), respectively, and 2.8, 3.8, and 2.9 when employing the temperature inside the canopy at 0.25 m, respectively. The development prediction utilizing the air temperature at the bottom (0.25 m) of canopy, where N. lugens resides, was most effective for N. lugens among the 3 species. These findings suggest the importance of utilizing microhabitat-based temperatures to predict pest development.


Introduction
Pest phenology models have been developed to predict the development stages, thereby serving as a tool for determining the optimal timing of pesticide application (Utida 1957, Damos and Savopoulou-Soultani 2012, Tonnang et al. 2017).These models, aimed at forecasting the seasonal prevalence of pest occurrences, have been extensively researched for various insect species, including lepidopteran and hemipteran insects (Davidson 1944, Hilbert and Logan 1983, Honek and Kocourek 1990, Kiritani 2012).Generally, phenology models rely solely on temperature as the environmental environmental conditions, including temperature, vary between inside crop canopies and typical reference heights (Takai et al. 2010, Yoshimoto et al. 2011, 2022, Fukuoka et al. 2012).Previous research has emphasized the importance of using microclimate data to predict diseases and pest damage (Hatfield 1982, Kuwagata et al. 2019).Specifically, several studies have assessed the influence of microclimate on disease occurrence within crop canopies through field surveys (Pangga et al. 2013, Cheng et al. 2015, Haider et al. 2017).However, research focusing on pest development within the canopy is limited, and existing studies have heavily relied on simulation-based validation approaches rather than field surveys (Davidson et al. 1990, Saudreau et al. 2013).Some studies have found that using air temperature inside the canopy improved the accuracy of predicting potato moth development (Rebaudo et al. 2016, Faye et al. 2017).Studies using microclimatic temperature as an environmental variable for predicting pest development have been restricted to dry-field crops and associated pests.In paddy rice, the air temperature inside the canopy is susceptible to flooding, drained conditions, and water temperature (Miyasaka et al. 2011, Tsujimoto et al. 2021, Sun et al. 2022, Maruyama et al. 2023).Consequently, the temperature difference between inside and outside the canopy may be more pronounced in paddy fields than in dry-field crops.To date, no studies have examined temperature inside the canopy and pest development in paddy field crops.Therefore, in this study, we investigated the effect of temperature inside the canopy on pest development in paddy rice.
Nilaparvata lugens (Stål), Sogatella furcifera (Horváth), and Laodelphax striatellus (Fallén), rice planthopper species, are major paddy rice pests in Asia.They considerably decrease paddy rice yield and quality through direct phloem-sap feeding and pathogenic virus transmission (Bottrell and Schoenly 2012, Lee et al. 2013, Horgan et al. 2015, Li et al. 2015, Matsumura 2017, Jiang et al. 2018, Yang et al. 2021).In temperate regions of East Asia, including Japan, N. lugens and S. furcifera are unable to overwinter and establish themselves after long-distance migration from overwintering areas, such as southern China and northern Vietnam (Matsumura 2017).In contrast, L. striatellus is an indigenous insect that can overwinter in Japan and can invade rice paddies through migration from nearby weeds rather than from overseas flights (Shiba et al. 2018).Existing methods to predict the development of the 3 species have utilized temperatures observed at reference heights, as with other insects (Okuda et al. 2019, Tanaka et al. 2019).However, as N. lugens, S. furcifera, and L. striatellus reside within the paddy rice canopy, they experience temperatures different from those at reference heights.Notably, the 3 rice planthopper species exhibit variations in microhabitat preferences within the canopy; N. lugens tends to inhabit the lower parts of rice plants, whereas S. furcifera and L. striatellus do not exhibit the same behavior (Takai and Inou 1970, Noda 1987, Isichaikul and Ichikawa 1993, Matsumura 2017).
In this study, we examined the population density of the 3 rice planthopper species (N.lugens, S. furcifera, and L. striatellus) and conducted meteorological observations inside and outside the canopy.We recorded temperatures at various heights in 3 experimental fields with different water temperature plots to gather a comprehensive dataset of canopy temperatures.Based on the population dynamics of these 3 species within the experimental plots, the relationships between the development prediction in rice planthoppers and the microclimate within the canopy were discussed.
In each paddy field, 2 plots (10 × 25 m) with different water temperatures (normal and low) were established to obtain broadrange temperature data inside the canopy and to ensure easier examination of the effects of the temperature inside the canopy on the development of the 3 planthopper species.Upstream plots created by continuous irrigation and paddy field partitions were defined as "low-water temperature" plots and downstream plots as "normal" plots (Miyasaka et al. 2011).The low-water temperature plots in this study were replicates of paddy fields in mountainous areas and those grown using groundwater.Water levels were adjusted to an average of 80 mm, and no mid-drainage was conducted.

Meteorological Observation
Meteorological observations were conducted at the center of each plot throughout the rice growing season to minimize edge effects.Air temperatures inside and outside the canopy were observed using ultra-fine T-type thermocouples (φ 0.08 mm; OMEGA Engineering, Norwalk, CT, USA) to minimize the influence of radiation.The effect of solar radiation increases with sensor size (Bergman et al. 2020).Maruyama et al. (2020) reported a daily mean error of 0.24 °C when using sensors with 0.25-mm fine thermocouples.The sensor used in this study was 0.08 mm, indicating that the effect of solar radiation was even smaller than that reported in the previous studies.In addition, the effect of solar radiation was further diminished as the temperature was observed inside the canopy.Therefore, using the 0.08-mm sensor rendered the effect of solar radiation on temperature negligible.At 2.0-m height, the air temperature was measured above the ground using a naturally ventilated shelter (41303-5A, Young, MI, USA), whereas air temperatures inside the canopy were measured at 1.0, 0.75, 0.5, and 0.25 m above the ground without a shelter to eliminate the effect of shelter heating by radiation in the canopy where adequate ventilation was difficult.Water temperature was observed mid-depth at 3 points with T-type thermocouples (φ 0.65 mm, OMEGA Engineering); the water temperature averages were used during analysis.Meteorological observations were made every 10 s with data loggers (CR1000 and CR1000X, Campbell Scientific, Logan, UT, USA) and recorded as 10-min averages.

Surveys of Rice Planthopper Species
In each experimental plot, N. lugens, S. furcifera, and L. striatellus, 3 major Asian pest species, were surveyed.The arrival time of N. lugens and S. furcifera in rice paddies was determined based on the date of capture of adults via net traps set at 10-m height (Kisimoto 1976, Otuka 2013).Net traps placed next to paddy fields were checked daily during the growing season of paddy rice.A total of 320 adult N. lugens male and female individuals (male-female ratio of 1:1) were released into each experimental plot on 30 July because there was no sufficient invasion to establish themselves in the experimental plots.After the first 2 wk of transplantation, the population density of rice planthoppers in paddy fields was surveyed twice a week using the standard sticky board method.In this method, a sticky board was placed horizontally on one side of a rice hill at 10 cm above the ground, and the rice hill was manually struck on the other side to dislodge the insects (Nagata and Masuda 1978).The survey was conducted in 3 replicates of 20 rice hills per experimental plot, and the survey data averages were used to calculate the population density of each species per plot.The development period of rice planthoppers, defined according to previous reports, is an indicator used to verify the accuracy of development prediction using temperature (Kuno 1968, Tanaka et al. 2019).Specifically, the development period was defined as the period between the date of first arrival of a species (including the date of release of N. lugens) and the peak of population density (or the percentage of adult population) after the predicted development period (occurring after 20-30 days of arrival in the case of rice planthoppers) or as the period between peak of population density (or the percentage of adult) and peak of population density (occurring after 20-30 days of population density peak).The percentage of adult population was calculated using adult and nymph population densities.In this study, only life stages after the middlenymph stage were counted because early nymphs are difficult to distinguish among the 3 species.

Phenology Model
A linear model was used to predict pest development (Uvarov 1931, Utida 1957, Damos and Savopoulou-Soultani 2012).Details of each parameter based on previous laboratory experiments are shown in Table 1.The models predicted the development of the 3 rice planthopper species at each of the 3 different stages (egg, nymph, and preoviposition [PO]).Rice planthoppers progressed to the next stage when the hourly integrated value of the developmental rate reached R = 1, and the predicted development period was defined as the period when all development stages were completed.The temperatures used in the phenology model were hourly averages of the observed values.
The linear model with effective accumulated temperature is the most commonly applied model; it describes the relationship between development rate and temperature as shown in Equation (1) (Noda 1989, Vattikuti et al. 2019): where R is the developmental rate, P is the developmental period, T is the temperature (the explanatory variable), T min is the lower threshold temperature for development, and K is the thermal constant.In this study, the model was expressed with Eq. ( 2) when considering the effects of high temperatures (Okuda et al. 2019): where T upper is the upper-temperature threshold and T stop is the temperature at which development stops and the survival rate is zero (i.e., the critical thermal maximum).The threshold temperatures have been studied in various regions and found to be similar (Horgan et al. 2020).Specifically, Kiritani (2012) reported that numerous studies have examined T min .In this study, we used the values reported by Noda (1989) who examined all stages, including the PO stage.For each life stage of N. lugens, T upper was the temperature reported by Horgan et al. (2020), whereas T stop was derived from the temperatures reported by Krishnaiah et al. (2005) and Piyaphongkul et al. (2012).Similar previously reported values were referenced for S. furcifera and L. striatellus (Table 1).

Data Analysis
The effects of water temperature differences on air temperature inside the canopy were confirmed using the Tukey honest significant difference (HSD) test (P ≤ 0.05).The prediction method using the air temperature inside the canopy and that using the air temperature at the reference height (2.0 m) was compared by calculating the root mean square error (RMSE) and mean absolute error (MAE) values between the predicted and observed development periods.Furthermore, the prediction methods were verified with skill scores based on RMSE.The RMSE skill score (RMSE ss ) showed the proportion of RMSE improvement over a reference method.RMSE ss was defined as RMSE ss = 1 − (RMSE h /RMSE ref ), where RMSE h is the RMSE calculated using the temperature inside the canopy at a specific height and RMSE ref is the RMSE calculated using the temperature at the reference height (2.0 m).All statistical analyses were performed using the "stats" package in R (v.4.1.2) (R Core Team 2021).PO represents preoviposition.T min is the lower threshold temperature for development, K is the thermal constant, T upper is the upper threshold temperature limit for development, and T stop is the temperature at which development stops or the survival rate is zero (i.e., the critical thermal maximum).and paddy field partitions resulted in 3-4 °C differences in water temperature during the day on the heading date.During the day, the air temperatures inside the canopy gradually increased from 0.25 to 2.0 m, with the lowest temperature observed at the water level (0.0 m).At night, the air temperatures inside the canopy were lower than the air temperature at the reference height (2.0 m) and the water temperature (0.0 m).

Diurnal and Seasonal Variations in Air Temperature Inside the Canopy
Table 2 shows the mean daily air temperatures inside and outside the canopy, water temperatures in each experimental plot, and the result of the Tukey-HSD test after the release of N. lugens (30 July-30 September 2021).The air temperature inside the canopy (0.25 m) in the normal and low-water temperature plots was 1.3 and 2.1 °C lower than the air temperature at the reference height (2.0 m), respectively.The results of the Tukey-HSD test revealed statistically significant differences between air temperatures inside (0.25-1.0 m) and above the canopy (2.0 m) in both plots.Additionally, there were also significant differences (P ≤ 0.05) in the air temperatures inside the canopy at several different heights.

Rice Planthopper Development Period
The results of adult population peaks observed in the 3 experimental fields with 2 different water temperature plots based on the number of 3 rice planthopper species are shown in Table 3.The development periods were determined based on the results of adult peaks (population density and adult percentage).The starting dates of the development periods for N. lugens were 30 July, when their initial release was carried out, and 14 August, when an invasion from overseas was observed.Sogatella furcifera invaded multiple times, with major dates of 9 June, 9 July, and 14 August.Laodelphax striatellus is an indigenous insect that constantly invades paddies from nearby grasslands.Therefore, the dates of the first observation of L. striatellus or those of the adult peaks (population density and adult percentage) were used as the starting dates.
The average development periods observed in the normal and low-water temperature plots were 29.11 and 31.44 days for N. lugens, 27.92 and 27.80 days for S. furcifera, and 28.33 and 28.50 days for L. Striatellus, respectively.There were no statistically significant differences (P > 0.05) in the development periods between the plots for any species.

Prediction of Rice Planthopper Development Period
Figure 2 shows the relationship between the development periods predicted using air temperatures inside and outside the canopy or water temperatures and the observed development periods obtained from the population dynamics of the 3 rice planthopper species.For all 3 species, the development periods predicted using air temperature at the reference height (2.0 m) tended to be shorter than the observed development periods.The RMSE values for the 3 rice planthopper species were smaller when calculated using air temperatures inside the canopy (0.25-0.75 m) than when calculated using air temperatures at the reference height (2.0 m) (Table 4).In the normal plots, the RMSE minimum value for N. lugens, S. furcifera, and L. striatellus was 3.10 at 0.5 m height, 3.88 at 0.25-0.5 m height, and 3.61 at 0.5 m height, respectively.Similar tendencies were observed for the MAE values.In the low-water temperature plots, the RMSE and MAE minimum values for N. lugens, S. furcifera, and L. striatellus were observed at 0.25, 0.0, and 0.25 m heights, respectively.The RMSE ss for N. lugens was 0.31 and 0.56 in the normal and low-water temperature plots when calculated using air temperature inside the canopy at 0.25 m height; the scores for S. furcifera and L. striatellus were 0.24 and 0.31, and 0.20 and 0.29, respectively.

Discussion
Influence of Water Temperature Differences on Canopy Air Temperature Diurnal temperature variation in paddy fields is known to be smaller than that in upland fields due to the large volumetric heat capacity of water.In particular, the diurnal temperature variation is lower inside the canopy than outside the canopy.Diurnal temperature variation is known to affect pest development (Chen et al. 2015).It has been reported that the development rate of herbivorous arthropods decreases when the mean daily temperature is around 30 °C and  Among the observed heights, 0.25-2.00m is the air temperature, and 0.00 m is the water temperature.Different letters indicate a significant difference according to Tukey's honest significant difference (HSD) test (P ≤ 0.05).
the diurnal temperature variation is over 10 °C (Vangansbeke et al. 2015).In this study, diurnal temperature variation inside the canopy was lower than 10 °C even on days when the diurnal temperature variation outside the canopy was 10 °C or higher.Although the development of rice planthoppers depends on the effect of the diurnal temperature variation, it is possible that the environment inside the paddy field canopy functions as a buffer, thereby reducing the effect of the diurnal temperature variation on the accuracy of development prediction.Lowering the water temperature by continuous irrigation has been proven to be effective for improving rice quality and preventing sheath blight disease (Arai and Ito 2001, Nagata et al. 2005, Miyasaka et al. 2011).In this study, we established low-water temperature plots via continuous irrigation to collect broad-range temperature data inside the canopy.Continuous irrigation resulted in an average reduction of 2 °C in water temperature per day, consistent with the results of previous studies (Arai andIto 2001, Miyasaka et al. 2011).However, no relationship was found between the observed development periods of the 3 rice planthopper species and the water treatment plots, nor was there a statistically significant difference.The 2 °C water temperature difference inside the canopy had an average effect of 0.8 °C on the air temperature at the most affected height (0.25 m).For rice planthoppers such as N. lugens, a temperature difference of 0.8 °C may result in a difference of 1-2 days in their development period of approximately 30 days.Given that the survey frequency in this study was 3-4 days, it may not have been possible to detect differences in the developmental period due to temperature differences between inside and outside the canopy.However, our results suggest that extending the development periods of rice planthopper by more than 3 days is difficult, even with continuous irrigation to lower the water temperature.Therefore, further investigation with more frequent observations or drastic changes in water temperature (>2 °C) is required to verify the effects of water temperature control on the development rate of rice planthoppers.

Influence of Air Temperatures Inside the Canopy on Predicting Pest Developments
In this study, analyzing the development periods of the 3 rice planthopper species and air temperatures by height revealed that the predictions of the development period using air temperatures inside the canopy were in better agreement with the observed values compared to predictions using only the reference heights.In most instances, the development periods predicted using the air temperature at the reference height (2.0 m) were shorter than the actual observed periods.This discrepancy may be attributable to the air temperature at the reference height (2.0 m) being higher than that within the canopy, where the 3 rice planthopper species reside.The phenology model used in this study indicated that the development rate of N. lugens was the highest at temperatures ranging from 28.5 to 34.9 °C.The air temperature at the reference height (2.0 m) was often within this range compared to the temperature inside the canopy.Additionally, the phenology model did not show a decrease in development rate at higher temperatures, as the maximum temperature during the study was 34.6 °C below T stop (34.9 °C).Consequently, the predicted development periods were likely shorter because the predicted development rates exceeded the actual rates.It is also possible that the observed development periods were extended owing to stunted development under high-temperature conditions in the fields.Indeed, there are reports of stunted development of and reduced egg-laying by rice planthoppers at 32 °C or higher (Park et al. 2011, 2013, Sujithra and Chander 2013, Wang et al. 2013, Vattikuti et al. 2019, Horgan et al. 2020).However, development period extensions (i.e., a decrease in the development rate) were not observed when N. lugens individuals were reared at 34-38 °C for a short time (11:00-16:00) instead of continuously at 32 °C (Yang et al. 2021).This finding suggests that the discrepancy between the predicted and observed development periods using air temperature at the reference height (2.0 m) is unlikely to be attributed to the omission of observing the reduced development rates between T upper and T stop .Nighttime air temperatures inside the canopy dropped below T min (12.7 °C) in late October, which was the water drainage time before paddy rice harvest.In most cases, temperatures at the reference height (2.0 m) also dropped below T min .Differences in air temperatures inside and outside the canopy may influence development rates when daytime air temperatures within the canopy drop below T min .However, temperatures below 15 °C are generally unsuitable for rice cultivation (Sánchez et al. 2014, Kuwagata et al. 2019).While data remain insufficient, considering air temperature differences inside and outside the canopy are rare when temperatures are near T min in Asian paddy rice cultivation.In this study, N. lugens showed the highest RMSE ss , an indicator of improvement in the RMSE, in both normal and low-water temperature plots.Nilaparvata lugens tends to inhabit the bottom of rice plants, whereas S. furcifera and L. striatellus have not been reported to prefer specific parts of rice plants at any developmental stages (e.g., eggs, nymphs, and adults) (Takai and Inou 1970, Noda 1987, Isichaikul and Ichikawa 1993, Matsumura 2017).Considering that there are differences of 0.7 °C in the normal plots and 1.1 °C in the low-water temperature plots between the bottom (0.25 m) and top (1.0 m), even within the same canopy, differences in the microhabitats of the three rice planthopper species may affect the RMSE ss .Thus, these results suggest the importance of using environmental information on the habitats of pests to predict their development periods.For all 3 species, the RMSE ss values based on the prediction of the development period using the air temperature at the reference height (2.0 m) and 0.25 m inside the canopy were higher in the low-water temperature plots than in the normal plots.We observed that the lower the water temperature, the greater the difference between the air temperature at the reference height (2.0 m) and the bottom n is the sample size for the development period of each rice planthopper in each plot; the normal plot sample size is on the left, and the low-water temperature plot sample size is on the right. of the canopy (0.25 m), which is consistent with previous reports (Maruyama et al. 2023).These findings revealed that in paddy fields where the water temperature is low, predicting pest development using the air temperature inside the canopy (where pests live) would be more effective than using air temperature at the reference height.As it is often difficult to observe the weather in mountainous paddy fields, incorporating models that estimate the temperature inside a canopy using micrometeorological models along with regional atmospheric models such as WRF is necessary for accurate predictions of pest development (Maruyama and Kuwagata 2010, Chen et al. 2011, Yoshimoto et al. 2011, Skamarock et al. 2021).
The main limitation of this study was focusing solely on the effects of temperature on development period prediction for the 3 rice planthopper species.Furthermore, competition among species and other stress responses was not investigated.Predicted development periods in this study tended to be shorter than the observed period, even when considering air temperature within the canopy.Additionally, factors other than temperature may influence development periods (Rashid et al. 2016, Ikeuchi and Kubota 2018, Vailla et al. 2019, Liu et al. 2020).It has been established that feeding S. furcifera with paddy rice damaged by the same species slows their development rate (Matsumura and Suzuki 2003).In addition, L. striatellus is known to migrate to or from maize during the cultivation of paddy rice because L. striatellus hosts crops other than paddy rice (Yoshida et al. 2014).We believe that this migration is one factor that decreases the accuracy of on-site development prediction, not only for rice planthoppers but also for other pests.In L. striatellus, a species that has multiple host crops, the development prediction based on the temperature inside the canopy showed slightly improved accuracy; however, this effect was limited compared to that of the development prediction of N. lugens, which prefers to inhabit the lower sections of rice plants.The migration of pests may obscure population peaks, which are important in determining their development period.In addition, the migration of pests leads to changes in not only the canopy environment conditions, such as temperature, experienced by the pest but also in the development rate owing to changes in diet (Wang et al. 2013, HE et al. 2021).Therefore, depending on the type of pest, population density may need to be surveyed on surrounding crops other than the target crop.In the future, developing and modifying the model to consider these factors, such as the migration between crops and feeding conditions, will likely be the key to further improving prediction accuracy and reducing paddy rice damage caused by rice planthoppers.
Accurate pest development predictions are vital for optimizing labor utilization and ensuring a stable food supply in agriculture.In this study, we demonstrated the importance of considering the microclimate within a crop canopy in predicting crop growth and disease outbreaks, as well as developing pests.In particular, our results indicated that N. lugens may be strongly affected by microclimate inside the lower canopy, which represents a microhabitat within the rice canopy.These results highlight the importance of considering the specific temperature point at which data is recorded while acknowledging that other factors might be equally important in predicting pest development.We anticipate the utilization of microclimate-based predictions of pest development to serve as an effective means of bridging the gap between laboratory-developed phenology models and actual pest developments observed in the field.

Figure 1
Figure1shows the comparison results of daily variations in air temperatures inside and outside the canopy and water temperatures on the heading date of paddy rice transplanted in June 2021 between normal and low-water temperature plots.The average height of paddy rice plants during this time was 1.1 m.Continuous irrigation

Fig. 1 .
Fig. 1.Diurnal variation in air and water (0.0 m) temperatures at different heights in normal and low-water temperature plots during the heading stage.

Table 2 .
Daily temperatures (mean ± standard deviation) based on height in normal and low-water temperature plots during the growing season after the release of Nilaparvata lugens (Stål) (30 July-30 September 2021) in 3 fields and a result of 2-way ANOVA testing the effect of height and experimental plot

Table 4 .
RMSE and MAE values were calculated using the observed and predicted development periods for the 3 rice planthopper species: Nilaparvata lugens (Stål), Sogatella furcifera (Horváth), and Laodelphax striatellus (Fallén).The development periods were predicted based on air temperature at each height in normal and low-water temperature plots