Investigating the Impact of Climate Warming on Phenology of Aphid Pests in China Using Long-Term Historical Data

Global climate warming has significant influence on individual development, population dynamics, and geographical distribution of many organisms, which has drawn much attention in recent years. As a large group of poikilotherms, insects whose life activities are closely linked with ambient temperature are supposed to be influenced by global warming. In order to test the consistency or difference of the effects of long-term climate warming on phytophagous insect pests in different geographical environments, this study collected historical data on the occurrence and population dynamics of three aphid pests (Myzus persicae, Aphis gossypii, and Sitobion avenae) in China, and systematically explored their phenological responses. We found that, during a period of about 60 years, in general, the first occurrence dates and the first migration dates of the three aphids almost moved earlier, while the end of the occurrence and the last migration dates were slightly delayed. However, these responses also represented geographical variation at a local scale. Basically, our results showed that the occurrence and migration seasons of these three aphid pests have been prolonged along with climate warming. This study based on historical literature data provides empirical evidence and valuable implications for understanding the impact of climate warming on insect pests and future management strategies.


Introduction
As one of the main characteristics of global climate change, global warming has been widely concerned. The Intergovernmental Panel on Climate Change (IPCC) points out that the mean global temperature has increased by 0.72 • C during the period of 1880 to 2012 with an average rise rate of 0.12 • C/10 years [1]. Global climate change has a significant impact on geographical distribution, population dynamics, and phenology of many organisms [2]. In recent years, the impact of climate warming on insects has become a research hotspot. As poikilotherms, the life activities of insects such as growth and development, survival, reproduction and migration depend closely on ambient temperature, that is, they may inevitably be affected by the global warming [3][4][5][6]. Therefore, how to cope with climate warming becomes a key issue for individual survival and population development of insects [7].
Previous studies have shown that climate warming can accelerate the growth and development of insects, leading to earlier occurrence and longer life cycle [8][9][10]. For example, the breeding time of spruce beetles Dendroctonus rufipennis in the northwestern North America has been cut in half with China, with vast territory, complex terrain, and diverse climate types, is also significantly affected by global warming. However, the long-term effects of climate warming on aphid pests are largely unknown due to the lack of long-term data on population monitoring. To conquer this obstacle, historical data from the literature may provide valuable information for understanding this issue [35][36][37]. Therefore, our study collected historical data on the occurrence and population dynamics of three important agricultural pests of aphids, Myzus persicae, Aphis gossypii, and Sitobion avenae. All three species can harm lots of economic plants and crops [38,39], among them M. persicae and A. gossypii are rated as TOP 10 Arthropod Pests by the Centre for Agriculture and Biosciences International (CABI) in the State of the World's Plants 2017 report [40]. Based on the long-term data we collected, we investigated the influence of long-term climate change on these aphids by analyzing changes of several life cycle parameters.

Phenological Data of Aphids
The phenological data of aphids in this study were extracted and compiled from historical literature, most of which were retrieved from the CNKI database (http://www.cnki.net), the most extensive and comprehensive database of Chinese periodicals and magazines [41]. First, the species names of the three aphids were used as subject words and respectively searched by using the method of subject word retrieval. Then, the relevant literature reporting the occurrences and geographical distributions of the three aphids in various regions of China were consulted from January 1951 to December 2017 (see Supplementary Table S1). A total of 201 related articles were collected, of which 84 were related to M. persicae, 80 were related to A. gossypii, and 37 were related to S. avenae. The specific time and geographic information on life cycle parameters were extracted and a database was established.
The collected data were organized based on four life cycle parameters, including the first occurrence date, the first migration date, the end of occurrence, and the end of migration. We defined the hatching time of overwintering eggs as the first occurrence date, the time of mating and spawning on the winter host as the end of occurrence, the first time of starting to migrate as the first migration date, and the time of migration back to winter host as the end of migration. There were some vague time descriptions of these parameters in some literature, such as "the beginning of the month", "the end of the month", "the first (or last) ten days". Therefore, we specifically quantified such time information without specific dates. For example, the description about the beginning of a month was set as the first day of that month while the end of a month was set as the last day of that month, and the first, middle, and last ten days of a month were set as 5th, 15th, and 25th of that month, respectively.
The collected data of A. gossypii was mainly concentrated in Xinjiang of northwestern China, as it is the main cotton producing region with suitable conditions for cotton production [42], and therefore, the most historical literature of A. gossypii have been reported data from this region. This gave us an opportunity to test spatial heterogeneity of the impact of climate warming at a local geographical scale. All data collection sites reported in the collected literature of the three aphid pests (M. persicae, A. gossypii, and S. avenae) were georeferenced into geographical maps by using the ArcGIS 10.2 (ESRI, Inc., Redlands, CA, USA), which were presented in the Supplementary Figure S1.

Meteorological Data
We obtained the annual average temperatures of China from 1959 to 2017. Temperature records were retrieved from Chinese meteorological websites (http://data.cma.cn/), the China Environmental Bulletins, and the China Climate Bulletins. Related to the A. gossypii data, we also calculated the annual average temperature of representative areas in Xinjiang from 1951 to 2013. Given that the east-west Tianshan Mountains crosses the central part of Xinjiang, the climate difference between the southern and northern Xinjiang is obvious, which would have a certain impact on the life history of aphids. Therefore, based on our collected geographical distributions of A. gossypii in Xinjiang, we selected six meteorological stations in southern Xinjiang (including Turpan, Kumul, Korla, Aksu, Kashgar, and Shache) and two sites in northern Xinjiang (including Usu and Caijiahu), and calculated the annual average temperatures of the above eight sites, separately. Then, the average of the southern six stations and the average of the two northern sites was used to represent the annual average temperatures of the southern and the northern Xinjiang, respectively. In addition, considering that aphid phenology parameters (first and last occurrence or migration) are more affected by spring and winter temperatures, we also calculated the mean, maximum, and minimum temperatures of spring (March-May) and winter (December-February) in China, as well as representative areas of Xinjiang mentioned above from 1951 to 2014. A simple linear regression method was used to calculate the overall change rates of temperature in China, as well as the southern and the northern Xinjiang in about the past 60 years and analyze the trend and process of temperature change.

Statistical Analysis
We analyzed the phenological responses of the aphids by plotting changes of the life cycle parameters mentioned above (the beginning of occurrence, the end of occurrence, the first migration, and the last migration). The occurrence year was taken as the X-axis, and the change of days for each life cycle parameter was taken as the Y-axis. Linear regression analysis was implemented to establish regression equations, and the trends of the four life cycle parameters in the time series were analyzed, respectively.
To conveniently quantify the "change of days", for each aphid species and each life cycle parameter, we first calculated the differences ("number of days") between the dates of occurrence or migration records in our dataset and 1 January, and then the averages of the numbers of days were calculated. By adopting the average as a reference with a value of zero on the Y-axis, the change of days of occurrence or migration were plotted and used to determine whether the trends of aphid emergence or migration were earlier or later in a long term.
We first carried out a normal test for the quantified data sets, which showed that most data sets were under normal distributions, except for two data sets of the beginning and the end of the occurrence of M. persicae. For the two data sets not following normal distribution, we implemented model fitting to select the most suitable model, which indicated that the fitting degree and significance of the linear model were optimal. In order to improve the accuracy of further analyses, these two data sets were first normalized. The linear regression equation was obtained from regression analysis which was used to investigate the relationship between the year series and the change of days with climate warming. The Pearson coefficient was used to test the correlation between time and the date of occurrence or migration. All analyses were carried out using SPSS for Windows version 24.0 (SPSS Inc., Chicago, IL, USA).  (Figure 1c), within which region the AAT increased from 9.9 • C in 1951 to 13.2 • C in 2013. Moreover, a temperature change rate of about 0.0292 ± SE 0.003 • C year −1 (p < 0.01) was revealed by the linear regression calculation. It could also be seen that while the temperatures in both northern and southern Xinjiang were on the rise in the case of climate warming, the initial AAT, as well as the increasing rate in southern Xinjiang were higher than that in northern Xinjiang. In the past 60 years, the average temperatures of the spring (March-May) and winter (December-February) in China as a whole, as well as in Xinjiang have also been increasing in general, both in terms of average temperature and the maximum or minimum temperature in each season ( Figure 2). Through linear regression calculation, the average temperature of spring (referred as SAT hereafter) and winter (referred as WAT hereafter) in China increased about 0.0176 ± SE 0.004 • C (p < 0.01) and 0.0195 ± SE 0.006 • C (p < 0.01) each year, respectively (Figure 2a,g). The highest (SHT) and lowest (SLT) temperatures of spring rose by 0.0164 ± SE 0.005 • C (p < 0.01) and 0.0195 ± SE 0.004 • C (p < 0.01) per year, respectively ( Figure 2b). Additionally, the change rate of the highest (WHT) and lowest (WLT) temperatures of winter were about 0.0112 ± SE 0.007 • C (p = 0.1004) and 0.0261 ± SE 0.005 • C (p < 0.01) yearly (Figure 2h). Similar rising trends of temperature parameters could also be seen in both northern and southern Xinjiang.

The Response of the First Migration
Based on scatter plots of the first migration times of the three aphid species (Figure 3), except the A. gossypii in southern Xinjiang, the scatter points of change of first migration of M. persicae, S. avenae, and A. gossypii in northern Xinjiang showed a tendency to lower and more negative values in the Y-axis with the year series, indicating that the date of the first migration of these aphids appeared earlier. Among them, the first migration date of M. persicae and S. avenae was advanced by 0.770 ± SE 0.202 days year −1 (t = −3.682, p< 0.01) and 0.216 ± SE 0.103 days year −1 (t = −2.092, p < 0.05), respectively (Figure 3a,b). The first migration date of A. gossypii in northern Xinjiang was with a change rate of 0.632 ± SE 0.250 days (t = −2.527, p < 0.05) earlier per year (Figure 3c). However, this life cycle parameter of A. gossypii in southern Xinjiang was almost not changed (−0.0024 ± SE 0.186 days per year; t = −0.013, p = 0.990) (Figure 3d).   (Figure 4c,d). However, this contrast may be still uncertain due to few data points.

The Response of the End of Occurrence
In general, scatter plots of the last occurrence dates of M. persicae, as well as A. gossypii in Xinjiang showed an upward shift, which indicated that the times for mating and spawning on winter hosts were slightly delayed. The oviposition time of the green peach aphid advanced by 0.018 ± SE 0.017 (t = 1.053, p = 0.313) (Figure 6a), while this life cycle parameter of A. gossypii in northern and southern Xinjiang advanced by 1.949 ± SE 3.263 days each year (t = 0.597, p = 0.582) and 0.281 ± SE 0.670 days each year (t = 0.419, p = 0.689), respectively (Figure 6b,c). As too little information about the end of occurrence of S. avenae had been reported in historical literature, we discarded analysis of it for this species.

Discussion
Our analyses indicated that, during a period of about 60 years, the annual average temperatures of China as a whole, as well as the Xinjiang region have shown rising trends, and the average temperatures in the spring and winter showed similar trends. Under the condition of climate warming, in general, the first migration times of the three aphids (i.e., M. persicae, S. avenae, and A. gossypii) occurred earlier. This may be due to the warming climate that leads to earlier spring and faster reaching of the minimum threshold of population size aphids needed to migrate. Similarly, the temperature increase could make earlier incubation of overwintering eggs of the three aphids, therefore the first occurrence dates were also advanced. We also found that the last migration dates mostly showed slight delay. The frequent occurrence of warm winter may make the aphids that originally rely on spawning for overwintering continue to grow in parthenogenesis on secondary hosts, instead of flying back to winter hosts [33,43,44]. This may also cause a slight delay of the end of occurrence defined by mating and spawning on winter hosts, as shown by our data. Basically, these responses indicate prolonged occurrence and migration seasons of these aphid pests throughout the time series. However, data accumulated so far have indicated that the effects of climate change are much more complex than the simple linear response to the increasing average temperature, and there may be differences between different regions [45][46][47]. In fact, our data in northern and southern Xinjiang may also indicate spatial heterogeneity of the effect of climate warming at a local scale. For instance, the average temperature in southern Xinjiang was higher than that in northern Xinjiang. These may lead to earlier occurrence and migration times of aphids in southern Xinjiang than in northern Xinjiang. The average first migration time of A. gossypii in southern Xinjiang was around mid-April, while the average first migration time in northern Xinjiang was around mid-late April. The data for the other two species (M. persicae and S. avenae) in Xinjiang region also supported that, the further north the distribution sites were, basically the later the first migration time would be in a same year, as was also found in a previous study in the UK [16]. In addition, the long-term trends of last migration and the first occurrence of the cotton aphid showed different patterns between the northern and southern Xinjiang. There is a possibility that this contrast may be a real reflection of spatial heterogeneity of the impact of climate warming. However, uncertainty still exists due to few data accumulation for specific life cycle parameters.
Considering that aphid phenology is temperature-dependent, the weather anomaly in a specific year may introduce "noise" (i.e., unusual values for the life cycle parameters) in a long-term data set. For example, in 1996, the first migration of A. gossypii in southern Xinjiang was reported on 15 June, later than the average. According to literature records, the temperature was unstable in May of that year, heavy rain and low temperature appeared, which may delay the migration of aphids. However, in general, the first migration of the cotton aphid in southern Xinjiang is slightly ahead of time. Similarly, severe cold damage occurred in Heilongjiang province of the northeastern China in 1980, and the first migration time of M. persicae was postponed to mid-late July. Differences in geography or host plant may also affect aphid phenology. For example, in the Huangzhong county of Qinghai province with an altitude of about 2800 m, the first migration of M. persicae was later than the average. This is understandable considering the temperature decreases along with an increase of the altitude. Moreover, the establishment of greenhouses close to cities may affect the overwintering strategy of aphids and therefore introduce specific unusual records. There was one case of influence of greenhouse on the first migration of aphids in our dataset: M. persicae on peach trees in the greenhouse in Hami region of Xinjiang showed much earlier migration than the average. The extension of occurrence and migration seasons of these aphid pests mean a possible increase of damage time, which will be more harmful for the growth and development of crops. Our results can provide theoretical guidance for the future prediction and prevention of aphid pests. In addition, it should be noted that the time of aphid damage has been advanced caused by temperature increase, which may lead to mismatch of phenological synchronicity between aphids and host plants or natural enemies [48,49]. If the insects hatch early, but the host plants it feeds on do not germinate at the same time, then the young individuals cannot feed normally and die [50]. As an alternative response, insects may turn to new plants for food in order to survive, resulting in the expansion of host plant range. As described in a previous study, the host plants that the willow psyllid Cacopsylla groenlandica feeds on has expanded from just one type of willow to four along a climatic gradient [51]. The advanced occurrence of aphids may also affect the original phenological synchronization between aphids and parasitic wasps, leading to the decrease of parasitic rate [52]. Equally important is that rising temperatures could make aphids more prone to parthenogenesis and spread to higher altitudes and latitudes, potentially increasing the range of crop damage [32,53]. These are challenges to monitoring and controlling aphid pests in the future.
Understanding the long-term impact of climate warming on insect pests is of great significance in both science and application. However, evidence on this issue has still been limited [35][36][37], mainly due to the lack of long-term dataset. The present paper provides another empirical study by collecting nearly 60 years dataset on population dynamics of three notorious aphid pests. It should be noted that due to the ambiguity of time information provided by some literature, certain uncertainty could occur when quantifying the time data. Therefore, to minimize the impact of uncertainty in such historical literature data, careful data collection, standardization, and analyses are indispensable. Cautious analyses based on such historical dataset can still contribute valuable implications for uncovering the effects of climate warming on aphid occurrence and migration in general. This study will provide a reference for future studies to explore related issues in more organism groups and geographical regions. In the future, a more detailed investigation of the long-term impact of climate warming on aphids and other insect pests, as well as standard monitoring network by implementing, for example, suction traps, are very needed in China.

Conclusions
Based on the collection and analyses of historical data of three aphid pests (Myzus persicae, Aphis gossypii, and Sitobion avenae) in China, this study revealed the long-term effects of climate warming on phenology of these aphids across large temporal and spatial scales. Results showed that the annual average temperatures and average temperatures in the spring and winter of China as a whole as well as the Xinjiang region have shown rising trends over the past 60 years. With climate warming, in general, the first occurrence and migration dates of these aphids have almost moved earlier, while the end of occurrence and the last migration dates have been slightly delayed, indicating prolonged occurrence and migration seasons. The data of A. gossypii in northern and southern Xinjiang may also indicate spatial heterogeneity of phonological responses to climate warming at a local scale. The present paper proves the value of historical literature data for uncovering long-term impact of climate warming on insect phenology, and provides important implications for future prediction and control of insect pests.
Supplementary Materials: The following are available online at http://www.mdpi.com/2075-4450/11/3/167/s1. Table S1: A list of source references for retrieving the phenological data of the three aphids used in this study. Figure S1: The data collection sites of the three aphids used in this study.