Climate factors associated with the population dynamics of Sitodiplosis mosellana (Diptera: Cecidomyiidae) in central China

Understanding the impacts of climate on insect pest population dynamics is crucial in forecasting pest outbreaks and developing a sustainable pest management strategy. The orange wheat blossom midge, Sitodiplosis mosellana (Géhin), is a chronic winter wheat (Triticum aestivum L.) pest in China, and its population density can strongly fluctuate. We analyzed climate factors (temperature and precipitation) associated with population dynamics of S. mosellana in a large-scale field trial in central China from 1984 to 2013 using Generalized linear mixed effects models. We found total precipitation during January–March was significantly positively correlated with population density of S. mosellana, whereas temperature parameters were not correlated with the population levels. Moreover, S. mosellana population size was significantly negative effected by interaction between temperature and precipitation, which showed that high precipitation with low temperature in spring also reduced the population density. This suggests that annual population size of S. mosellana in Central China is determined by soil moisture in early spring. These results provide basic information that will help in forecasting population levels and in developing a sound pest management strategy for S. mosellana.

and hatch after 4-10 days. Larvae feed on the developing kernels for 3-4 weeks, and mature larvae drop to the soil surface and enter the soil to overwinter 15 .
The populations of S. mosellana fluctuate yearly at the local level 15 but the key factors determining population levels of S. mosellana remain uncertain. Many studies have investigated the key factors driving population dynamics of other insect pests and their natural enemies for the purposes of forecasting future outbreaks 2,15 . However, most studies were not conducted for extended time periods or over large spatial scales. Our research focused on determining the key climate factors (temperature and precipitation) influencing population dynamics of S. mosellana in Henan province, central China. We collected data over a 30-year period .
The annual population densities of S. mosellana varied significantly within each region without a clear trend (southern region, F 29, 148 = 28730.44, P < 0.001; central region, F 29, 148 = 416793.35, P < 0.001; northern region, F 29, 148 = 814126.58, P < 0.001). All regions had a population outbreak in 1989. No significant differences were observed between regions in the population levels of S. mosellana over 30 years (F 2, 498 = 0.801, P = 0.495; Fig. 2). temperature and precipitation variables. Mean temperature were not significantly correlated with the monthly abundance of S. mosellana in all three regions. The population dynamics of S. mosellana was positively related to monthly precipitation in all regions. However, the interaction between temperature and precipitation was negatively related to monthly abundance in all three regions (Table 1).
In Table 2, further analysis shows that total precipitation during January-March was the significant positive predictors of annual variation in population dynamics of S. mosellana (southern and northern, P < 0.01; central regions, P < 0.05). There was a significantly negative role of interaction between temperature and precipitation for S. mosellana (southern and northern, P < 0.01; central regions, P < 0.05).
The population densities of S. mosellana increased linearly with an increase of precipitation during January-March. The linear model for population density in all regions indicated that field population size was generally low if precipitation during January-March was less than about 20 mm (Fig. 3). The interaction between   www.nature.com/scientificreports www.nature.com/scientificreports/ precipitation and temperature during January-March on population density showed high precipitation with low temperature increased population density, while high precipitation with high temperature reduced in all regions ( Fig. 4).

Discussion
There were large fluctuations in both the population size of S. mosellana and climate factors during 1984-2013 in central China. We explored key climate factors associated with the field population density of S. mosellana. At the month scale, abundance is related to precipitation and the interaction between precipitation and temperature. At the annual scale, precipitation from January to March is a particularly important predictor.
S. mosellana diapause in the soil as larvae from summer to the following spring. Diapause is broken in response to rising soil temperatures, after which larvae enter a moisture-sensitive period. Most larvae developed when the soil moisture reached 17.5%, whereas most larvae went into an extended diapause for another year when the soil moisture was <12% 14 . Shanower (2005) highlighted the key role of soil moisture in enabling larvae to leave their cocoon 16 . Soil moisture also is the main factor limiting pupation and successful adult emergence 13,17,18 . Therefore, soil moisture levels during moisture-sensitive period is predicted to be a key factor affecting the yearly population abundance of S. mosellana. In central China, the soil moisture of wheat fields before April is determined by natural rainfall. The total precipitation during January-March in central China was greater in 1989 than in other years, which resulted in S. mosellana population outbreaks. On the contrary, the population size was very small if total precipitation during January-March was <20 mm. This result is consistent with the previous report, when precipitation exceeded 20 mm in May or early June in western Canada, the larvae terminated diapause, left their cocoons and pupated in several days, or most larvae remained dormant until the following year 18 .
Temperature may impact insects in many ways, directly via physiological or behavioral changes, or indirectly by influencing plant-insect interactions 19 . Within a certain range, increased temperature tends to have positive effects on insects 20 , especially multivoltine insects in temperate climate zones. However, our results indicate that the field population dynamics of S. mosellana was not directly determined by temperature. S. mosellana larvae pupate and emerge within the soil, and soil could therefore buffer these stages from the direct impacts of extreme temperatures. Similar results were also reported in other species such as parasitoids 21 , indicating that parasitoid development can be independent of the macroenvironment, and changes in temperatures are less likely to alter the dynamics of the host-parasitoid system 22 . The change of soil temperature lags behind that of air temperature, which may influence the population dynamics of S. mosellana. This requires further research as no data on soil temperature were available in the current study.
Many univoltine insect herbivores evolved their life history to synchronize their larval stages with the appearance of target host organs, and this synchronization determines the quality and quantity of available food resources and the population size of herbivores 23 . The life span of S. mosellana adults is usually a few days, and the damage occurs only if adult emergence coincides with the sensitive stage of wheat (ear emergence through to flowering). Therefore, the synchronization influences S. mosellana more severely than other insects. Climate changes may disturb the synchronization of herbivores with their host 23 . In this study, S. mosellana population size was significantly negative effected by interaction between temperature and precipitation, which showed that high precipitation with low temperature in spring also reduced the population density. The temperature requirement for S. mosellana adult emergence is different from that for wheat ear emergence; a cold spring may the later emergence of S. mosellana but may speed the ear emergence of wheat. In addition, the synchronization is affected by wheat variety, because the growth period will vary among wheat varieties 24 .

conclusions
Many studies and reviews report the effects of climate factors on multivoltine insects 2,5 , but there are fewer studies of climate impacts on univoltine insects. S. mosellana has one generation per year and spends more than 9 months in the soil as a diapause larva in a cocoon. The feeding-stage larva lives in a relatively confined space (wheat glume). Our results indicate that the population dynamics of S. mosellana were significantly related to precipitation during January-March in central China. Soil water evaporation is low during January-March in central China due to the low ambient temperatures 25 , so the soil moisture in late March and April was highly correlated to precipitation level during January-March. The soil moisture in late March and April could be a key  km 2 ), located in the central part of China. It is the main agricultural province and the largest winter wheat producer in China, producing 25% of the national wheat output. The climate in Henan province is continental monsoon and varies from the subtropical zone in the southern area to the temperate zone in the northern area. The mean annual temperature in Henan province ranges from 15.7 °C in the southern areas to 9.5 °C in the northern areas. The annual mean precipitation ranges from 533 to 1,380 mm, with the majority of rain fall occurring in summer. In this study, Henan province was divided into southern research regions (31°23′-33°50′N, 63,000 km 2 ), which are located in the Huaihe river valley and belong to a subtropical humid monsoon climate with abundant rainfall and sunshine; northern research regions (35°00′-36°22′N, 26,000 km 2 ), which are located north of the yellow river and belong to a warm temperate semi-humid and semi-arid continental monsoon climate; and central research regions (33°50′-35°00′N, 78,000 km 2 ), which belong to a transition zone (Fig. 5). The mean annual temperature is 15.1, 14.6, and 13.6 °C in the southern, central, and northern regions, respectively. The mean annual precipitation levels are 806, 634, and 528 mm in the southern, central, and northern regions, respectively. Climate factor data (temperature and precipitation) were obtained from China Meteorological Data Sharing Service System, China Meteorological Administration (http://cdc.nmic.cn/home.do/). The monthly and annual climate data, mean temperature and total precipitation during January-March of each year were used to analyze the association with the population dynamics of S. mosellana in the three regions during 1984-2013. Data analysis. Data were evaluated for normality using the Shapiro-Wilk test prior to performing the analyses. One-way analysis of variance was used to compare the differences of S. mosellana population density, temperature, and precipitation among years using SPSS statistics software, version 22 (IBM Corp., Armonk, NY, US). The contribution of monthly precipitation and mean temperature on the population density of S. mosellana in the three regions of Henan province during 1984-2013 was evaluated using Generalized linear mixed effects models (GLMM) in package glmmADMB in R Statistical Software 2.13.1, using 'temperature' , 'precipitation' as independent variables, 'year' as random variables, and Poisson error distributions. Post-diapause larval development and adult emergence period are moisture-and temperature-sensitive during late March and April in central China, so the temperature and precipitation during January to March were assumed the key factors determining the population dynamics of S. mosellana. To test this hypothesis, We used the generalized linear model (GLM) software package, relating the density of S. mosellana to temperature and precipitation during January to March.