Minimum Winter Temperature as a Limiting Factor of the Potential Spread of Agrilus planipennis, an Alien Pest of Ash Trees, in Europe.

The emerald ash borer, EAB (Agrilus planipennis) is a devastating alien pest of ash trees. It is spreading in European Russia and Ukraine and will appear in other European countries. Our aim was to determine the regions of Europe where the winter temperature drops low enough to prevent A. planipennis establishment. We calculated the minimum daily air temperature from 2003–2019 for each grid square (0.5° × 0.5°) in East Asia, North America and Europe and determined the minimum daily temperature in the grid squares where A. planipennis was recorded. Temperatures of −30 to −33 °C occur in the northern portions of the pest range on all continents. No established population has been recorded in localities where temperatures below −34 °C occur. This temperature is close to the absolute supercooling point of A. planipennis larva (−35.3 °C). It is unlikely that low temperatures could prevent the spread of A. planipennis in northern Western Europe (Sweden, Norway, Finland, etc.), since the temperature in this area did not fall to −34 °C from 2003–2019. However, such temperatures are not rare in eastern European Russia (Kostroma, Vologda, Orenburg regions, etc.), where Fraxinus pennsylvanica and F. excelsior occur. These regions could potentially become refuges for these ash species.


Introduction
The emerald ash borer (Agrilus planipennis Fairmaire) (Coleoptera: Buprestidae) is a devastating alien pest of ash trees in European Russia and North America [1][2][3]. The native range of this wood-boring beetle occupies a restricted territory in East Asia, namely in northeastern China, Japan, the Korean Peninsula and the southern Russian Far East [4]. Since the first record in America in 2002, the pest has spread to 35 states of the USA and five provinces of Canada and killed countless millions of ash trees [5]. In 2003, A. planipennis was first recorded in Europe in Moscow, and a severe outbreak and quick spread of the pest began [1,3]. By 2020, A. planipennis had spread to 16 regions of European Russia and to the east of Ukraine [6,7].
Agrilus planipennis is continuing to spread and will soon appear in other European countries [8,9]. It presents a serious threat to ash trees all over Europe [3,10], so it is included on the list of 20 priority quarantine pests of the EU [11]. It is very important to determine the potential range of A. planipennis in Europe and the factors that could limit the spread of the pest. This information is crucial for quarantine protocols and for the elaboration of measures of pest control. In addition, it is necessary to reveal the regions where ash trees occur but where the climate is not suitable for A. planipennis. It is estimated that A. planipennis will eventually decimate nearly all ash (Fraxinus spp.) in North America [12]. Agrilus planipennis is also dangerous for ash trees all over Europe [10]. The regions where the winter is too cold for A. planipennis could potentially become refuges for ash species. The spread of A. planipennis in Europe could be limited by the native distribution of its host plants: Fraxinus spp. [10] However, it should be taken into account that ash trees are often planted in cities far beyond the native ranges of ash species. In particular, F. pennsylvanica, which was introduced from North America and is highly susceptible to A. planipennis [13], is one of the most popular trees planted in the cities of European Russia and some other European countries [14]. The high-resolution map of Fraxinus spp. distribution in Central Europe is available [15]. However, no such map or city tree inventories for European Russia are available. An additional biotic factor that could theoretically limit the spread of A. plenipennis is the spread of its natural enemies, e.g., the parasitoid Spathius polonicus Niezabitowski [16], but the influence of this factor has not yet been studied.
The only attempt to predict the potential A. planipennis range in Europe by climatic variables was conducted using maximum entropy modelling [17]. This model has a number of serious flaws. First, the survey data from Orlova-Bienkowskaja [18] used as a base for this model were misunderstood: all surveyed locations were regarded as locations of A. planipennis detection, although in fact, the surveys in many locations indicated negative results. Second, the true range of A. planipennis by 2020 was more extensive than the output of this model [7].
It is yet unknown what climatic factors could limit the spread of A. planipennis to the north in Europe. For alien insects, low winter temperatures are one of the most important abiotic factors limiting northern range edges [19]. A study conducted in North America has shown that cold winter temperatures could potentially limit the spread of A. planipennis in the USA and Canada [20,21]. Therefore, it is important to determine the regions of Europe where winter cold could limit the spread of A. planipennis.
The life cycle of A. planipennis is 1 year in warm regions (specifically in Tianjin, China) [22], 2 years in cold regions (specifically in Moscow, Russia) [23] and 1 or 2 years in regions with an intermediate climate (specifically, in New York, NY, USA) [24]. With a 1-year life cycle, A. planipennis overwinters as a J-larva. With a 2-year life cycle, A. planipennis overwinters twice: the first time as a larva in a larval gallery under the bark and the second time as a J-larva in a pupal cell in the sapwood.
The J-larvae survive low temperatures by accumulating high concentrations of glycerol in their body fluids and by synthesizing antifreeze agents [25]. The supercooling point of the overwintering J-larvae of A. planipennis is different in different regions. The absolute minimum value that has been recorded experimentally was −35.3 • C [25]. J-larvae have not been recorded to survive temperatures below this value. These experimental data are supported by observations in nature in North America. In particular, severe mortality (93%) of the overwintering larvae was caused by low temperatures (below −30 • C and up to almost −40 • C) on the 13-15 February 2016 in Syracuse, New York, NY, USA [26].
We determined the minimum mean daily temperatures in localities with A. planipennis in the last 17 years (i.e., after the first record of A. planipennis in European Russia). This allowed us to determine the minimum temperature that did not prevent the survival of A. planipennis populations. Then, we assessed the regions of Europe where winter temperature could potentially prevent the spread of the pest.

Materials and Methods
We compiled a table of occurrences of A. planipennis in Asia, North America and Europe (Supplementary S1) using information from current databases and articles [4,5,7,27].
The beneath-bark temperature changes slower than the air temperature [20]. Therefore, we used the mean daily temperature to level the short-term air temperature fluctuations. The initial data on air temperature for each day from 1 January 2003 through 1 October 2019 were obtained from the ERA5-Land Global Atmospheric Reanalysis dataset for each 0.5 • × 0.5 • grid in Europe (from 30 to 75 • N and from 20 • W to 70 • E), Asia (from 20 to 60 • N and from 100 to 150 • E) and North America (from 10 to 65 • N and from 50 to 140 • W) [28]. These squares covered the whole current range of the pest in the world and all of Europe. We did not take into account the record from the Xinjiang Uygur Autonomous Region, China. It was unclear if A. planipennis had established in this region, since the number of found specimens and (or) damaged trees was unknown [4].
The daily mean temperature was calculated for each day in each grid square as an average of the temperatures at 00:00, 02:00, 04:00, 06:00, 08:00, 10:00, 12:00, 14:00, 16:00, 18:00, 20:00 and 22:00 UTC. A detailed description of the calculation method and the computer code used is provided in the supplementary material (Supplementary S2). Then, the minimum mean daily temperature recorded from 2003-2019 was determined for each grid (Supplementary S3). Visualization of the minimum temperature and localities of A. planipennis occurrence in the same map allowed us to compare these data. The visualization was created using DIVA-GIS 7.5 [29].
Then, we identified the days where the daily average temperature was below −30 • C in each grid square where A. planipennis was detected (Supplementary S4). This allowed us to determine the daily air temperatures in which the pest populations can survive.
The temperature experienced by the overwintering J-larva could slightly differ from the air temperature in the region. First, the grids in the ERA5 Global Atmospheric Reanalysis dataset were 0.5 • × 0.5 • [28]. Therefore, the temperature in a particular locality could slightly differ from the temperature calculated for the grid. In particular, in the center of large cities, the temperature could be 1-2 • C higher than that in the surrounding territories [30]. Second, when the nightly low temperatures occurred, temperatures beneath the bark were typically about 1 • C warmer than nearby air temperatures [20]. Besides that, the location of a tree (e.g., proximity to buildings, southern exposure and density of neighboring trees) could affect low temperatures that a tree might experience [20]. We did not measure the temperature experienced by individual larvae but used the mean temperature of the coldest day as an index of the severity of winter cold in the region. Our model did not take into account global climate change.

Temperatures below −30 • C in the Native Range of A. planipennis in Asia
Winter temperatures below −30 • C are not rare in the northern part of the native range in the Russian Far East and in northeastern China ( Figure 1). From 2003-2019, days with mean daily temperatures below −30 • C were recorded in at least 14 localities of Asia where A. planipennis was recorded during this period ( Table 1). The coldest days with mean temperatures below −33 • C were recorded in Troitskoe and Khabarovsk city (Russia) and in Yichun (China).   No direct data on the percentage of larvae that survived the low temperature in Asian A. planipennis populations were available. However, a comparison of the pest distribution with the climatic data showed that the populations lived in localities where a very low temperature was usual. For example, a population of A. planipennis in Khabarovsk was first recorded in 2006 and still existed in 2016 despite temperatures below −30 • C occurring there every year from 2010 to 2014, and the minimum mean daily temperature being below −33 • C [31]. The mean daily temperature did not fall to −34 • C in any A. planipennis locality in Asia from 2003-2019.

Temperatures below −30 • C in the Current Range of A. planipennis in North America
A similar analysis of climatic data made for the current range of A. planipennis in North America ( Figure 2) has shown that very cold days with mean temperatures below −30 • C were recorded in several A. planipennis localities from 2003-2019 ( Table 2) Table 2.   Table 2.

Temperatures below −30 • C in the Current Range of A. planipennis in European Russia
The days with mean temperatures below −30 • C occurred in most localities of A. planipennis in European Russia from 2003-2019 ( Figure 3, Table 3). However, no cases of extinction or decline of populations after severe cold have been recorded. In particular, in January 2006, temperatures below −30 • C were recorded in the Moscow region. However, despite this, a severe outbreak and quick spread of A. planipennis was recorded in the Moscow region from 2006-2009 [1,[44][45][46].
The northernmost European population of A. planipennis was discovered in 2013 in the city of Yaroslavl [18]. This population survived despite frosts in January 2017: the mean daily temperature was −33.76 • C on 7 January and −33.69 • C on 8 January. Moreover, the range of A. planipennis in Yaroslavl expanded: in summer 2019, A. planipennis was recorded in 11 new localities and killed hundreds of trees in the city [47].
No populations of A. planipennis have been recorded in European Russia in regions where mean daily temperatures below −34 • C occurred from 2003-2019.
Yaroslavl [18]. This population survived despite frosts in January 2017: the mean daily temperature was −33.76 °C on 7 January and −33.69 °C on 8 January. Moreover, the range of A. planipennis in Yaroslavl expanded: in summer 2019, A. planipennis was recorded in 11 new localities and killed hundreds of trees in the city [47].
No populations of A. planipennis have been recorded in European Russia in regions where mean daily temperatures below −34 °C occurred from 2003-2019.  Table 3.  . Some localities where the daily temperature fell below −30 • C are numbered. Detailed information for each of these localities is provided in Table 3. Table 3. Some Agrilus planipennis localities in European Russia where the mean temperature of the coldest day from 2003-2019 was below −30 • C *. Figure 3)

Discussion
The data from Asia, North America and European Russia indicated that populations of A. planipennis could survive in regions where mean daily temperatures below −30 • C occurred. Even mean daily temperatures of −33 • C did not destroy A. planipennis populations. The winters in northern Western Europe were warmer than those in the current northern range of A. planipennis in European Russia and in the northern native range in Asia ( Figure 4). Therefore, it is unlikely that winter cold could prevent the spread of A. planipennis in Norway, Sweden, Finland and other European countries.
Established populations of A. planipennis have not been recorded in regions with severe winters where days with mean temperatures below −34 • C occur. According to the experimental data, such temperatures cause high mortality in the overwintering J-larvae [25,48]. The absolute supercooling point, i.e., the minimum temperature at which A. planipennis J-larvae could survive, is −35.3 • C [21].
Temperatures below −34 • C are not rare in eastern European Russia (see Figure 4 and Supplementary S2). Therefore, winter cold could probably become the limiting factor of A. planipennis spread to this region. In particular, it is unlikely that A. planipennis will establish in the Arkhangelsk region, Republic of Komi, most of the Vologda region, Kostroma region, Kirov region, Udmurt Republic, Republic of Bashkortostan or most of the Orenburg region. Temporary populations of A. planipennis could theoretically appear in these regions in the case of several subsequent warm winters, but these populations would probably be destroyed by the next cold winter.

Discussion
The data from Asia, North America and European Russia indicated that populations of A. planipennis could survive in regions where mean daily temperatures below −30 °C occurred. Even mean daily temperatures of −33 °C did not destroy A. planipennis populations. The winters in northern Western Europe were warmer than those in the current northern range of A. planipennis in European Russia and in the northern native range in Asia ( Figure 4). Therefore, it is unlikely that winter cold could prevent the spread of A. planipennis in Norway, Sweden, Finland and other European countries. Established populations of A. planipennis have not been recorded in regions with severe winters where days with mean temperatures below −34 °C occur. According to the experimental data, such temperatures cause high mortality in the overwintering J-larvae [25,48]. The absolute supercooling point, i.e., the minimum temperature at which A. planipennis J-larvae could survive, is −35.3 °C [21].
Temperatures below −34 °C are not rare in eastern European Russia (see Figure 4 and Appendix S2). Therefore, winter cold could probably become the limiting factor of A. planipennis spread to this region. In particular, it is unlikely that A. planipennis will establish in the Arkhangelsk region, Republic of Komi, most of the Vologda region, Kostroma region, Kirov region, Udmurt Republic, Republic of Bashkortostan or most of the Orenburg region. Temporary populations of A. planipennis could theoretically appear in these regions in the case of several subsequent warm winters, but these populations would probably be destroyed by the next cold winter.
Fraxinus pennsylvanica is very common in urban and roadside plantings in eastern European Russia. It has become an established nonnative plant in the Orenburg region and Republic of Bashkortostan and occurs in natural habitats there [49]. This ash species is very cold-tolerant. In particular, plantings of F. pennsylvanica in Ufa (Bashkortostan) survived extremely cold winters when the temperatures fell below −40 °C [50].
Fraxinus excelsior is also sometimes planted in the cities of eastern European Russia. In addition, it is native in the Kostroma, Vologda and Orenburg regions and occurs in the forests there [51][52][53]. These regions, where the winters are too cold for A. planipennis but the climate is suitable for F. excelsior and F. pennsylvanica, could potentially become a refuge for these ash species. Fraxinus pennsylvanica is very common in urban and roadside plantings in eastern European Russia. It has become an established nonnative plant in the Orenburg region and Republic of Bashkortostan and occurs in natural habitats there [49]. This ash species is very cold-tolerant. In particular, plantings of F. pennsylvanica in Ufa (Bashkortostan) survived extremely cold winters when the temperatures fell below −40 • C [50].
Fraxinus excelsior is also sometimes planted in the cities of eastern European Russia. In addition, it is native in the Kostroma, Vologda and Orenburg regions and occurs in the forests there [51][52][53]. These regions, where the winters are too cold for A. planipennis but the climate is suitable for F. excelsior and F. pennsylvanica, could potentially become a refuge for these ash species.
It is interesting that the range of A. planipennis in European Russia has not expanded to the north since 2013, although the range has significantly expanded to the south [7]. This additional indirect evidence suggests that the current northernmost A. planipennis locality (Yaroslavl) is close to the northern border of the potential range.
We used the data since 2003, because A. planipennis was first recorded in Europe in 2003. Little is known about the distribution of A. planipennis in Asia before records in North America and European Russia. There is no evidence that A. planipennis survived the temperatures below −34 • C in its native range before 2003.
The minimum winter temperature is important but probably not the only factor determining the potential A. planipennis range. The possible impact of accumulated degree days during the growing season and other factors should be studied in the future. It would also be interesting to make direct year-round measurements of the temperature under the bark of ash trees in Moscow, Yaroslavl and other localities of A. planipennis in European Russia. This would allow us to determine what temperatures are experienced by the larvae and J-larvae in nature.