The rising temperatures from February to April, driven by climate change, have primarily accelerated the onset of the vegetation season across the analysed territory, shifting on average by nearly two weeks during the investigation period. These changes were statistically insignificant in only a quarter of the study area's grid cells, primarily in regions of highlands (western Lithuania, central Latvia).
The obtained SGS date changes agree with the results of other studies in the research area. In Lithuania, the SGS date shifted earlier by approximately 13 days over the period from 1961 to 2018 (Mačiulytė and Stankūnavičius, 2020). In Latvia, from 1960 through 2008, due to the rising winter temperatures, budburst occurred earlier by 22–25 days, and the temperature rise in April led to an earlier start of flowering by 9–16 days, depending on the plant species (Kampuss et al. 2009). However, such changes in SGS dates and flowering of plants are not solely positive—they not only increase the risk of late frost damage but also may influence the spread of pests and diseases (Meier et al. 2022).
Due to the changing climate and the increasing late winter and early spring temperatures, the last frost date has also become earlier throughout the study area. Nevertheless, the changes in this case were smaller, averaging 6.3 days. Similar changes were observed in other studies, such as in Lithuania from 1961 to 2018 (Mačiulytė and Stankūnavičius 2020) and in the USA from 1900 to 2014 (Kukal and Irmak 2018). The shift of LSF dates from the beginning of the 20th century to the end of the 21st century has also been recorded in Poland and Central Europe (Graczyk and Kundzewicz 2016; Ma et al. 2019; Wypych et al. 2017).
The main factor determining the distribution of LSF dates is the distance from the Baltic Sea, while latitude also influences it. Due to the Baltic Sea's warming effect, in the lower-lying coastal areas, such as western Lithuania and the southwestern part of the study area, the median LSF date is, on average, nearly a month earlier than in central Estonia. The distance from the Baltic Sea also influenced the spatial distribution of FS events. In coastal areas, FS events were rarely recorded during the study period (up to 2 occurrences per decade). Moving away from the Baltic Sea, the number of FS events rapidly increased, with more than half of the study area experiencing 7–8 events per decade. Nonetheless, comparing the first 20 years of the study period with the last two decades (2003–2022), it was found that the greatest increase in FS events occurred in areas characterised by a more rapid rise in spring temperatures.
Although the number of FS events increased over the 73-year study period, the observed changes were small and statistically insignificant. Similar trends have been observed in FS events in other countries or regions. It was found that during the period from 1971 to 2020 in Poland the number of frosts occurring after the SGS date did not change significantly, and the spatial distribution of these phenomena is also primarily influenced by the distance from the sea (Koźmiński et al. 2021; Koźmiński et al. 2023; Nidzgorska-Lencewicz et al. 2024). The risk of frosts causing damage to plants has also remained unchanged since 1864 in the lowlands of Switzerland and Germany (Vitasse and Rebetez 2018). However, frosts remain dangerous for crops and other plants (Liu et al. 2018; Nidzgorska-Lencewicz et al. 2024). Although the number of FS events did not significantly increase during the study period, events capable of causing strong ecological and economic losses still occurred in the eastern Baltic Sea region in the 21st century. For example, in the spring of 2007, 11.8% of the study area's grid cells, primarily in the southwestern and northeastern parts, experienced FS events when a GDD sum of 100 to 200°C was accumulated and the tmin ranged from − 2 to -5°C. Such events occurred during budburst and leaf unfolding when plant frost resistance is minimal (Chamberlain et al. 2019).
In the future, as temperatures continue to rise, the vegetation season is expected to start even earlier, and the dates of leaf unfolding and flowering will also change (Juknys et al. 2017; Ruosteenoja et al. 2016; Szyga–Pluta et al. 2022; Zhu et al. 2019). Therefore, the risk of FS events may persist or even increase (Chmielewski et al. 2018; Ma et al. 2018; Unterberger et al. 2018; Vitasse et al. 2019), especially in areas where the shift of SGS dates exceeds that of LSF dates, and frosts are recorded closer to the date of budburst (Jönsson and Bärring 2011). However, when assessing the damage caused by FS events to vegetation both in the past and the foreseeable future, it is essential to consider that plant frost sensitivity depends on the species (Jönsson and Bärring 2011; Liu et al. 2018; Morin and Chuine 2014) and growth stage (Chamberlain et al. 2019).
The occurrence of FS events in the eastern Baltic Sea region is closely related to atmospheric circulation patterns, and future changes in these patterns may influence both the frequency and intensity of FS events, as well as the extent of the resulting damage. Typical conditions leading to the formation of severe or extreme spring frosts are associated with blocking processes and positive sea level pressure anomalies. Such a situation leads to the advection of cold Arctic air masses from the northwest, north, and northeast (Bielec-Bakowska and Twardosz 2023; Tomczyk et al. 2019). This advection is most often recorded when blocking patterns form in the North Atlantic and northern Europe (Pfahl 2014). It has been found that in Estonia, the frequency of air movement from the north-northwest increased during the 1966–2015 period, with the most significant and statistically significant changes occurring in spring (Lakson et al. 2019). Such changes in the eastern part of the Baltic Sea region may increase the likelihood of spring frosts and, consequently, FS events.
Several factors must be considered when conducting future research on FS events in the eastern part of the Baltic Sea region. In this study, the SGS date was defined solely based on the mean daily air temperature without accounting for other variables influencing the SGS date, such as precipitation, soil moisture and nutrient content, evapotranspiration, and other relevant parameters (Chen et al. 2023; Linderholm 2006; Qu et al. 2015; Rimkus et al. 2017). However, it has been established that for determining the SGS date in spring, the air temperature is crucial (Juknys et al. 2017; Ladwig et al. 2019; Romanovskaja and Bakšienė 2009). Additionally, this study used only two variables to identify FS events: the dates of SGS and LSF. This approach assumes uniform damage from FS events across different species, their life stages and habitats (Chamberlain et al. 2019). Therefore, future research should focus more on specific species to better understand the negative impact and severity of FS events on plants (Ladwig et al. 2019). A deeper understanding of FS events would also facilitate the development of more reliable predictions of such events in the future, thereby helping to mitigate the ecological and economic impacts of FS events, develop climate change adaptation strategies, make more effective decisions in agriculture, forestry, and insurance policies, and cultivate species suitable for growth under altered climate conditions (Chamberlain et al. 2019; Lamichhane 2021).