- Top of page
- Material and Methods
- Conflict of Interest
In recent decades, changes in weather patterns have been observed, the most pronounced being an increase in ambient temperatures (Solomon et al. 2007). In addition, there is previous research that has recorded changes in species ranges, phenology of life cycles, and interactions (Walther et al. 2002; Parmesan and Yohe 2003; Menzel et al. 2006; Visser and Both 2006; Walther 2010). Species range expansions and, to a lesser extent, retractions have been observed (Parmesan et al. 1999; Parmesan 2006) as well as changes in population dynamics in response to changing winter temperatures and snow cover (Ims and Fuglei 2005; Bale and Hayward 2010). Changes in outbreak patterns (Esper et al. 2007) and in outbreak range (Jepsen et al. 2008) have been attributed to climate warming. One example is the mountain pine beetle (Dendroctonus ponderosae) in North America, which has expanded its range as a result of increased minimum winter temperatures and has achieved higher densities as a result of increased survival and availability of even-aged host plant stands, resulting in unprecedented damage to forests in North America (Robertson et al. 2009; Cudmore et al. 2010). More often patterns of forest pest insect damage have a complex relationship with weather factors or associated changes.
Drought commonly increases herbivore damage (Csóka 1996, 1997; Jactel et al. 2012). Drought stress situations that have a positive effect on host plant quality for insects may arise in situations with high temperature and/or low precipitation (Koricheva et al. 1998). There is evidence that high levels of damage are related to unusual weather patterns (Martinat 1987 and references therein), but the patterns are inconsistent. Insects are ectotherms and could, therefore, be expected to respond strongly to changes in their external environment. Decreased development times increase the potential for multiple generations within the same growing season for multivoltine species as well as for species previously considered to be univoltine (Altermatt 2010; Pöyry et al. 2011). Other studies have found that increased temperatures lead to more fecund adults (Laws and Belovsky 2010). Because temperature often has a direct positive effect on insects in many of their life stages, it is readily concluded that insect populations will exhibit improved performance in a warmer climate. This simple causal linkage fuels the fear of increased damage by pest insects.
Researchers investigating latitudinal gradients showed greater herbivore damage at lower (warmer) latitudes (Adams and Zhang 2009), but other studies have failed to reveal similar herbivore damage patterns (Andrew and Hughes 2004; Sinclair and Hughes 2008). Causal links are unclear and subject of debate, but conclusive findings are lacking (Björkman et al. 2011). Thus, the current evidence seems to suggest no general conclusion concerning levels of herbivore damage in relation to increasing temperatures.
Potential effects of climate change on herbivore damage are not related only to direct effects of weather variables on insect herbivores, indirect effects also need to be considered (Klapwijk et al. 2012). Weather variables can affect insect larvae and the damage they do directly by influencing factors such as winter survival and metabolic rates. However, host plant quality and natural enemy pressure are affected by fluctuations in weather patterns as well, which, in turn, will affect the damage levels as well. In general, the effects of weather may contribute to and strengthen the effects of induced responses of the host tree to herbivore damage (Baltensweiler et al. 2008) and the effects of natural enemies, especially parasitoids, that will be observed in the next or subsequent generations (Liebhold et al. 2000; Klapwijk et al. 2012 and references therein). However, these responses might be asymmetrical (Berggren et al. 2009) resulting in increased predator efficiency as a result of increased temperatures (Kruse et al. 2008). The outcomes of all possible interactions between plants, herbivores, and their enemies in a food web are reflected in the levels of and fluctuations in herbivore damage.
Analysis using the volume of damaged wood (m3) removed from the forest as indicator of bark beetle damage (Ips typographus) showed drought to be one of the factors influencing levels of damage (Marini et al. 2012). They also found endogenous negative feedback with a 2-year lag, suggesting a potentially important role for natural enemies of I. typographus in forests in the Italian Alps. A major obstacle when studying the possible link between weather and dynamics is the lack of long-term data sets as most ecological data collections span 10–20 years. The difficulties are particularly problematic for outbreak species because the time frames of data collection would most likely encompass only one or two outbreaks providing meager data for analyses.
In this study, “area of damage” caused by herbivorous forest pest insects (i.e., the area of forest subjected to more than 20% defoliation) and associated year-to-year fluctuations are examined in relation to observed climate change. We have the unique opportunity to present damage data collected by the Hungarian Forest Research Institute from 1961 to 2009 for six forest insect species (Fig. 1). The damage of these six species could be contributed to specific species in the field whereas other data collected are grouped per family. We analyze the data for trends over time in observed damage area and for relations with temperature and precipitation. Also, we investigate potential changes in the amplitude of the fluctuations over time.
The aim of this study was to investigate whether there are relationships between herbivore damage and temperature or precipitation. The two questions we endeavor to answer are as follows:
- Are there trends in the area of insect damage and/or variability in insect damage during the period 1961–2009?
- How are weather variables in year (t) and year (t − 1) related to damage in year (t)?
An extensive damage area indicates widely distributed populations, and visible herbivore damage indicates high numbers of individuals. Fluctuations in damage area reflect population densities over a certain threshold over time and space. These fluctuations can be used to assess the relationship between the area of herbivore damage and weather conditions (Miller et al. 1989; Marini et al. 2012). The data set includes Lepidoptera species notorious for causing damage in Hungarian forests. These species cause high levels of damage at more or less regular intervals and are often referred to as “outbreak species” (Berryman 1987).
Outbreak species are often characterized by high potential population growth rates (Wallner 1987; Hunter 1991) resulting in the capacity for rapid population increases from 1 year to the next, often leading to high levels of herbivore damage. The population dynamics of species with high potential population growth rates are more vulnerable to fluctuations due to density-dependent regulation than species with low population growth rates (Turchin and Taylor 1992; Turchin 2003). Exogenous variables and endogenous negative feedback together could result in irregular and more or less cyclical patterns in population dynamics (Turchin 2003).
If changes in weather patterns were linked to the overall changes in climate, one would expect changes in insect population fluctuations over time. Theoretically, a possible mechanism behind amplified fluctuations could be increased realized population growth rates (Hassell et al. 1976). Such positive direct effects of climate warming have been found, for example, in the pine processionary moth (Thaumetopoea pityocampa) and the mountain pine beetle (Dendroctonus ponderosae) as a consequence of higher winter temperatures, increasing winter survival, and thus higher realized population growth rates, leading to greater damage (Battisti et al. 2005; Cudmore et al. 2010).
Positive indirect effects might be observed if density-dependent population regulation becomes weaker; this could potentially lead to similar variation around an increasing population mean. On the other hand, if natural enemy pressure becomes stronger or the potential number of individuals sustained in a population is reduced, the population will exhibit greater fluctuations. Higher frequency and severity of outbreaks lead to more damage, but it is still debatable whether or not global warming will result in a higher risk of insect outbreaks (Klapwijk et al. 2012 and reference therein). We found that even though we can observe changes over time, these changes are species specific. Also, there is an indication of a direct relationship between these changes and climatic variables, but the exact mechanisms are hard to identify.
- Top of page
- Material and Methods
- Conflict of Interest
This study is one of only a few to investigate the damage caused by several forest pest insects over a relatively long time period. The general finding is that there is no single trend for all six investigated species. Recent studies suggest that damage by herbivores should increase with increasing temperature (Adams and Zhang 2009), but since most of these studies use temperature gradients involving altitude or latitude as proxies for climate change they are not directly applicable to temporal changes. The analyses of temperature and precipitation data in this study showed that mean yearly temperature has been steadily rising since roughly 1985 (Fig 2A). This steady rise does not directly relate to changes in damage area over the same time span. Precipitation patterns seem not to have changed in that period and also do not show a direct relationship with damage area. This lack of common trends over time indicates the need for more detailed analyses.
The damage data used in the time series (area of damage in hectares) is an indirect measure of population dynamics. The damage recorded is defoliation over 20% (or 10% infestation for R. buoliana) and thus actually reflects the outbreak area of a species at any given time in Hungary. Detailed density sampling would not have provided this type of large-scale information even though it might have given more insight into the actual mechanisms behind the population dynamics observed. The data, as collected and analyzed, give us an insight into the influence of weather variables on the extent of outbreak areas of certain species and reveals several interesting temporal patterns. Changes in damage levels and damage patterns over time could, potentially, be attributed to changes in the area of forest available for these insects. However, over the years examined, the area of forest has increased in Hungary over the study period by planting Robinia spp. stands, the forest area with the preferred tree species for the insects discussed in this study remained unchanged (G. Csóka, pers. comm.). The only observed change is the age structure of the pine forest, the influence of this change in age structure on R. buoliana was investigated for potential effects, but no relationship with the pattern of decreasing damage was found.
Our analyses led us to a number of conclusions. First, no obvious changes in the variability of outbreak area were recorded for any of the species during the data collection period. This means that the outbreaks are not spreading over large areas any more often than they were when data collection began. Second, even though the timing of their larval period is similar for the species examined, different trends in their long-term dynamics were observed. This indicates that changes in population densities do not depend only on direct effects on larval development and performance. Interestingly, both species with increasing trends in damage area over time have low CV's compared with the other species (Table 3). Unfortunately a quantitative comparison was not possible because of limited replication. Third, the observed increasing trends in damage for some species are not related to the overall increasing yearly mean temperatures, implying that the effect of changes in weather is operating during certain sensitive stages that might be species specific. Last, the two species with the highest coefficient of variation, L. dispar and M. neustria (Table 3), do not show an increasing trend. Interestingly, the damage caused by R. buoliana, which also has a high CV, actually reduced over the observation period (Fig. 4).
Our results indicate that there are direct as well as indirect effects of temperature on insect performance, potentially mediated by the host plant or natural enemies. Of all species, only T. processionea larvae seem to be affected by temperature during the larval growth period (temperatures in March and April). The negative relationship between March temperatures and subsequent outbreaks could be because high March temperatures cause hatching to be asynchronous with leaf flush; neonate larvae can survive some starvation, but if the period is too long it will have a negative effect on survival (Meurisse et al. 2012). However, if the high temperatures are a month later in April, they have a positive effect on damage area, indicating a positive effect on survival and feeding rates (Wagenhoff and Veit 2011). Lymantria dispar appears to be affected by March temperatures but with a delayed effect on damage in the subsequent year. Several species seem to be affected by July temperatures; E. chrysorrhoea and T. processionea in the same generation and T. viridana in the next generation. In July, these species are either in their pupal or adult stage and the direct effect on damage area could be the result of positive effects of temperature on dispersal, extending the damage area. The effect of dispersal can be observed in the same generation when the number of individuals dispersing is high or it can be seen in the following generation if the numbers are limited or the dispersed females have low fecundity, but their offspring are very successful in the newly colonized, previously undefoliated areas. Successful progeny of dispersed females, that is, a delayed effect of dispersal, could occur when dispersing individuals have low fecundity, a common feature in Lepidoptera (Saastamoinen et al. 2010). For both R. buoliana and L. dispar, temperatures in October have a negative relationship with damage area. This could indicate that warm Octobers lead to larvae (neonate larvae in the case of L. dispar) that are poorly prepared for overwintering, potentially because they use too much resource (Han and Bauce 1998 cited from Bale and Hayward 2010) or (perhaps more likely for L. dispar but also possible for R. buoliana) the extra heat provides parasitoids with a longer window of opportunity for attack and is associated with higher attack rates (Dhillon and Sharma 2009), reducing damage area in the next generation. For R. buoliana we identified a negative relationship between January temperatures and damage area; perhaps higher January temperatures keep metabolism up through winter, leading to individuals in poor condition when temperatures increase in spring (Hahn and Denlinger 2007). The lagged negative effect of September temperatures on T. viridana damage area is not straightforward to interpret. Effects on egg/larval survival should be observed in the damage caused by the generation in the following year. Potentially there is an effect of temperature on natural enemy survival, for example parasitoids; such increased parasitism success would only be observed in the herbivores 2 years later (Liebhold et al. 2000).
Using minimum and maximum temperatures might have led to clearer patterns in the data. However, mean monthly temperatures do reflect the minima and maxima to a certain extent; therefore, using these data in the analyses should have revealed the most important patterns of the relationship between temperature and area of damage for these forest insect Lepidoptera.
Certain outbreak species, like L. dispar, are known to have strong cyclical patterns of outbreaks (Turchin and Taylor 1992). Our analysis confirms this for the L. dispar damage area in Hungary. This cyclicity is often attributed to regulation by natural enemies based on theoretical models and some empirical data. The population cycles of L. dispar fit a second-order autoregressive model with a moving average. Including weather variables in this model does not add explanatory value. This could indicate that the population processes affecting L. dispar dynamics in Hungary are relatively insensitive to weather conditions. Alternatively, the processes may be sensitive to weather, as shown by the first-order regressive model with the weather variables included; this generates the type of pattern described by an ARMA model. However, it is unlikely that the population dynamics of L. dispar are independent of weather influences, even though previous studies have also suggested that the drivers of the cycles are endogenous (Liebhold et al. 2000). Euproctis chrysorrhoea and T. viridana show some indication of periodicity, but the noise in the data prevents the detection of true cycles over time. Potentially, the dynamics of the latter two species are not solely based on interactions with natural enemies or bottom-up control; the high percentages explained by the model (Table 4) support the relative importance of exogenous variation (e.g., weather fluctuations).
For most species, potential and detected cycles span more than about 10 years. This means that potential changes in periodicity are difficult to detect since a 49-year time series only covers four complete cycles at most. The potential cycles of M. neustria would span a period of more than 20 years, so this data set only covers, perhaps, one cycle. In summary, the actual frequency of peaks in damage area has not changed for any of the species. The potential and detected presence of 10-year cycles in the damage was also the reason to divide the damage in 10-year periods for the analyses of changes in the magnitude of variation over time. Approaching the variability in this manner the risk of unevenly divided peaks in the damage is brought to a minimum and the actual effect of higher amplitude of changes will be analyzed.
As discussed in the introduction, weather can have a great influence on many insect traits, resulting in observed changes in insect emergence and development times recorded as being a result of climatic change (Walther et al. 2002; Parmesan and Yohe 2003; Parmesan 2006). According to our analyses, temperature changes affect the damage caused by insect herbivores more strongly than precipitation. Our study does show that not all species are affected by climate change and that the species with a relatively high CV (indicating high rates of population change) will not respond in a similar manner to species with a low CV (indicating low rates of population change). Potentially, species with high population growth rates have population dynamics that are ruled more by endogenous feedback than by exogenous feedback; our results for L. dispar and M. neustria indicate that the feedback does not change. However, for E. chrysorrhoea, T. viridana, and T. processionea, it seems that the changes in exogenous feedback allow them to cause more than 20% damage to a greater area of forest. The observed reduction in R. buoliana damage area strengthens the finding of earlier research that autumn temperatures can be important for overwintering success, and that there is an additive effect of January temperatures (West 1936; Bogenschütz 1976). However, the observed relationships between damage and weather are not always intuitive and therefore highlight the need for a better understanding of the interactions between temperature, precipitation, insect physiology, and ultimately their effects on population dynamics.