Recent changes in insect distribution are consistent with the expected interacting effects of climate and habitat change. There are many examples of high-latitude margins of insect distributions expanding polewards [1
] and of contractions or local extinction at low-latitude or low-elevation range limits [4
]. Global meta-analysis shows that insect distributions have, on average, shifted polewards and to higher elevations following geographic shifts in isotherms, and such poleward shifts are more rapid in the Northern than in the Southern Hemisphere (18.54 km per year northward shift in centroids) [6
Distributional area changes and the current spreading of animals are usually detected based on the findings of newly established populations, because the spreading of individuals is usually below the limits of observability, especially in the case of insects and other terrestrial invertebrates. Most invertebrates are difficult to find because they are small, are difficult to identify, and are confined to specific ecological niches [7
]. In the case of insects, changes in the areas of distribution are usually documented by random findings of individuals or by standardized monitoring schemes or citizen science projects [8
]. Regardless, the history of spread remains mostly unknown in a given area.
Because of their sensitivity to both land-use and climate change, Orthoptera are generally useful for investigating the effects of recent environmental changes and their effects on range shifts [3
]. In the current research, we selected the bushcricket Ruspolia nitidula
as a model for studying the process of range expansion. R. nitidula
has expanded its area of distribution in Western and Central Europe in recent decades (e.g., [11
]). It is strong flier and therefore has the potential to spread over long distances [14
]. There are currently many spreading species of Orthoptera (e.g., Conocephalus discolor
, and Phaneroptera
spp., and Oecanthus pellucens
) (e.g., [3
]), but R. nitidula
has characteristics that facilitate the detailed mapping of its spread. R. nitidula
males emit a characteristic sound, which is readily detectable at distances of tens of meters [15
]. For this reason, it is possible to detect spreading individuals to map the routes in relation to the landscape topography but also to document the spread rates in a given environment.
is a good model for studying climate-induced range shifts due to good flying ability and free movement through the open non-forested landscape [11
]. Many insect species are habitat specialists and need stepping stones, therefore the spreading could be influenced by the availability of optimal habitat or its fragments. There is limited knowledge about R. nitidula
dispersal abilities, the authors only generally consider R. nitidula
as mobile (e.g., [20
]), capable of long-range dispersal several kilometers distant (e.g., [22
]). No exact study has yet been devoted to flight capabilities, only Monnerat [11
] supposed high mobility due to the recorded colonization of sites that were 10 to 15 km apart.
Expansion rates have traditionally been measured based on the occupancy of square grids and their geographical distances [24
]. However, the geographical (i.e., Euclidean) distance represents the shortest possible path and does not account for the landscape matrix, which is one reason least-cost methods were introduced in the field of landscape ecology [26
]. The purpose of least-cost analyses is to identify optimal routes regarding a raster matrix (i.e., a cost surface), which represents the ease/difficulty of dispersal for a species across particular parts of the landscape [26
]. One option for creating a cost surface is to use habitat suitability models (but see, [28
]). Least-cost analyses are widely used in ecology for identifying potential movement paths and corridors [29
], and have application in the maintaining of habitat connectivity [32
] and in the planning of greenways [33
]. Although Mineur et al. [34
] used least-cost path (LCP) analysis to measure distances to estimate range expansions of marine macrophytes, the LCP analysis was only used to avoid measuring the distance over the mainland, i.e., the coastline was recognized as a barrier that prevented movement of marine macrophytes.
The goal of our study was to estimate the rate of recent range expansion of the actively spreading bushcricket R. nitidula
. The estimates are based on calculated distances from two sources of occurrence data that were obtained at different scales and with different sampling efforts. For one source (a landscape-scale source with uneven sampling), we used a national databases of documented records. For the second source (a regional-scale source with detailed, systematic sampling), we monitored the spread of R. nitidula
for 5 years at the Moravian Gate. The Moravian Gate is a natural pass between the Sudetes and the Carpathians and is located on the areal margin of the R. nitidula
distribution. Although the results of some studies indicated that the Moravian Gate currently plays no part in the south–north spread of some animals and plants [35
], other studies have found that the Moravian Gate is an important ecological corridor [38
We used three methods to estimate the distance of R. nitidula spread: geographic distance, length of the LCP, and length of the randomized-based LCP. We compare the estimates resulting from the analysis of the landscape-scale data that were obtained with an uneven sampling effort with the estimates resulting from the analysis of the regional-scale data that were obtained by systematic monitoring.
After 50 years of absence, R. nitidula reappeared in the Czech Republic in 2006. In that year, the first individuals were detected in southern Moravia near the Czech–Austrian border. The species then began to continually spread to the northwest and northeast. Using three methods to calculate the spread distance, including two methods that accounted for habitat suitability (LCP and random LCP), we estimated the expansion rates and possible spreading routes.
According to the analyses of the maximum annual shift of all occurrence data at the landscape scale, the estimated maximum expansion rates was high (13.8–16.2 km/year). We are aware that this estimation of the overall expansion is partly based on spatially and temporally uneven occurrence data, and could therefore be biased. Moreover, the great disadvantage of the landscape-scale data is the effect of possibly insufficient sampling effort. We tried to assess this bias by estimating expansion rate on gradually randomly reduced data. Although reducing the dataset by 90% substantially reduced the estimated expansion rate (by 47–48% for all three distance methods), the prediction of the best-fit Weibull model for a 10-fold increase of sampling effort had virtually no effect on the expansion rate estimate (Figure 6
); this suggested that the sampling effort was sufficient for these estimations. The estimates derived from the two longest paths (11.5–13.5 km/year) correspond better to linear model estimates and annual distances from our systematic field monitoring on the areal margin than to linear model estimates based on the landscape-scale data. During monitoring of dispersed males in the Odra River basin (2016–2020), we recorded the highest annual expansion between years 2017 and 2018. The distance estimated based on geographical distance was 13.7 km, and we consider this to be the lowest estimate of how far R. nitidula
moves, i.e., we assume that the real distance is greater because the organisms are unlikely to move in a straight line. In contrast, the estimated distance based on LCP analysis was 14.8 km. The estimate using passage LCP analysis, which we consider more realistic than the estimate using LCP analysis, was only slightly lower than 14.6 km between these years. In other years, however, the annual distances moved were shorter (see Figure 7
). Linear regression of maximum distances in our systematic survey provided estimates of expansion rates ranging from 11.1 to 11.7 km/year, depending on which method was used to calculate distance moved. Even with the mentioned unlikely LCP path between years 2017 and 2018, all three methods of distance calculation resulted in similar estimates. Moreover, the difference in calculated distances between the three methods did not change (Figure 8
), unlike the distances at the landscape scale (Figure 4
). We assume that this is caused by the openness of the landscape in the Odra River basin, which might be favorable for R. nitidula
As noted in the Section 1
, R. nitidula
has expanded its area of distribution in Western and Central Europe in recent decades and the supposed trigger is a warming trend (e.g., [11
]). Simmons and Thomas [69
] predicted range expansion to be characterized by two distinct phases. First, populations at the expanding margin should consist of large numbers of spreading individuals due to their selective advantage during the founding of new populations. Second, as additional populations in the landscape are established and the selective advantage of dispersal is reduced, the costs of expansion should select for lower dispersal rates, such that these populations should have characteristics of long-established rather than spreading population. Travis and Dytham [70
] showed that these two phases of evolution should occur during expansion of species. The R. nitidula
individuals monitored in the current study are apparently in the first phase of range expansion, i.e., the individuals are spreading and establishing temporary (or possibly permanent) populations that are the sources of next season’s migrants. We estimated possible spreading routes from two origin occurrences (2006) to two occurrences on the areal margin of the distribution (2020). In the northwest direction, the path goes from Morava and the Thaya conflux area through parts of the Dyje–Svratka Valley and Svitava Valley to the eastern part of Polabí; in the northeast direction, the path goes through the Lower Morava Valley, the Moravian-Silesian foothills, and the Moravian Gate to the Ostrava Basin. These modelled paths are unsurprising regarding the ecological requirements of the species. Although passive dispersal has been reported for orthopterans [71
] and other insect species [72
], and the most west-north occurrence in this study is isolated (Figure 1
), we assume that the similar spread rate of the longest path on the contrary direction supports the plausibility that this occurrence record is the result of natural expansion. We also recognize that records west of the origin occurrences from 2006 may be the result of direct expansion from northern Austria (see [73
]). Unfortunately, we could not examine this in detail, because precise data from Austria are not available. However, the first occurrence of Ruspolia in this area of the Czech Republic was recorded in 2011, and according to our regional survey estimates, the distance would be achievable by gradual expansion from origin occurrences reported in 2006. In any case, because we calculated our estimates only from a cumulative maximum distance from all data described in the Section 2.2
, we assume that this will not have a significant effect on our estimates.
Climate-induced range shifts have been detected in many animals, and a significant portion of these shifts have been found in records collected over a long period of time [24
]. There are various ways to calculate species expansions or shifts, but because detailed ecological knowledge for many species is missing, less complex methods based on species presence data are often used to assess distributional changes [25
]. The majority of range shifts are detected by the simple comparison of recent and historical distributions. Hickling et al. [75
] analyzed a 25-year dataset of 329 animal species in Britain and found that 275 of the species shifted northwards and 52 shifted southwards at their range margin; the latter study also reported that 22 species of Orthoptera had a northward shift at the range margin of 34 km during the 25-year period. Preuss et al. [25
] evaluated the range expansion of the non-native bushcricket Roeseliana roeselii
in Sweden using range margin statistics and the area-based method of grid occupancy. Both approaches resulted in estimates of spreading rates ranging from 1.5 to 3.0 km/year. However, Hochkirch and Damerau [76
] reported that macropterous individuals of R. roeselii
can disperse for distances up to 19.1 km/year. Simmons and Thomas [69
] tested predictions for the evolution of dispersal during range expansion for four species of wing-dimorphic bushcrickets. The observed range expansion of macropterous Conocephalus discolor
increased over time, from a rate of 1.22 km/year in the period 1976–1989 to 7.44 km/year in the period 1990–2002). Walker and Nickle [77
] estimated the maximum spread rate of the mole crickets Scapteriscus vicinus
and S. acletus
at 20 km/year. Our study included the analysis of the maximum annual shift of occurrence data from long-term, landscape-scale observation and a 5-year annual systematic monitoring of dispersed males in the Odra River basin. Our use of direct monitoring of spreading males to detect annual range shifts is unique among Orthoptera, because we monitored spreading individuals before they established populations that are usually used for range shift estimation. Spreading individuals can provide clearer insight on used routes than reconstructions based on the established populations, those existences depend on the quality and position of a suitable breeding habitat.
One of the main pitfalls of using LCP analysis is a deterministic path calculation that leads to a single optimal path [46
]. The known limitations and shortcomings of the use of LCP analysis in connectivity modelling include the assumption that moving individuals have knowledge of the entire landscape and the destination locality and will therefore follow only the one optimal path; on the other hand, LCP analysis introduced path stochasticity in the estimates of connectivity [27
]. We suggest that these assumptions may not be so limiting in the context of calculating the rate of expansion, because we did not aim to describe the ability of an individual to travel a certain path (as in connectivity modelling), but instead aimed to assess the overall range expansion rate of the species. However, one optimal path generated with LCP analysis is problematic also in this case. Because LCP analysis relies on a cost surface, different input surfaces result in different routes. Researchers previously reported that distances of LCP analysis based on habitat suitability depend on the chosen HSM [79
]. We therefore used the randomized shortest path (i.e., passage LCP analysis) as an alternative to geographic distance and LCP analysis. On the two longest paths (Figure 5
), the lines representing geographic distances ignore unsuitable regions for R. nitidula
(e.g., mountainous areas) and therefore estimated the shortest distance travelled per year. Paths estimated by LCP analysis, in contrast, strictly followed the most suitable places (i.e., pixels on the provided cost surface), and could overestimate the expansion rate. In the case of our regional-scale monitoring, the longest path 2016–2017 generated with the LCP analysis (Figure 7
), leads to different point than other two methods, resulting in higher calculated distances. Additionally, in our preliminary analysis (unpublished data), we include altitude in the HSM, (which showed majority relative variable importance among other variables) and further least-cost path between the most distant northwestern occurrence and origin points followed the lowest possible altitudes and greatly circumvents the highlands ignoring its valleys, which results in measured distance of ca. 250 km.
Based on our results, we recommend caution in the use of LCP analysis to estimate an insect’s rate of expansion. On the other hand, use of geographic distance will always underestimate the rate to some extent. The difference in the estimated distances depend on the overall distance of expansion (as indicated by the diverging trend lines in Figure 4
) and presumably on the impermeability of the landscape for the specific species. We suggest that adding some degree of randomization to the analysis, as is done with passage LCP, can provide more ecologically reasonable results than those obtained with the use of geographic distance or LCP analysis. Additional research is needed to determine which level of randomization is appropriate and the degree to which the suitability of the methods depend on the specific organism and landscape.
Recent climate-induced range shifts have been detected in many insect taxa in Central Europe, and R. nitidula seems to be a good model for the detailed evaluation of spreading paths and expansion rates due to good flying ability and free movement through the non-forested landscape. LCP analysis has indicated potential corridors for the spread of native or alien thermophilic species in association with climate warming. The expansion rate of R. nitidula documented in the current study indicates a natural and rapid shift northward of the areal margin across the entire width of the Czech Republic. Similar expansion rates can be expected for other insects that are strong fliers and that can freely move through open non-forested landscapes.