Insects are a pre-requisite for many ecosystem functions (Yang and Gratton 2014) and related ecosystem services that are essential for human welfare (Losey and Vaughan 2006; Macadam and Stockan 2015; Rader et al. 2016). Somewhat surprisingly, recent reports on insect decline (e.g., Hallmann et al. 2017; Sánchez-Bayo and Wyckhuys 2019; van Klink et al. 2020) have attracted broader public attention to the global biodiversity crisis and its consequences (Cardoso et al. 2020). A remarkable decline in insect abundance and diversity is observed in many different regions and terrestrial habitats (Lister and Garcia 2018; Wepprich et al. 2019; Wagner 2020), including protected areas (Rada et al. 2019). This is of special concern as it threatens the functioning of ecosystems and the supply of important ecosystem services, such as pollination, food supply, biological control, and soil fertility regulation, and of diverse cultural ecosystem services (Noriega et al. 2018). The documented decline is severe and relevant enough to demand deliberate and consequential political action (Donkersley et al. 2022). Nevertheless, data on most insect populations and their development are still inadequate (Eggleton 2020; Goulson 2019).

Long-term data about population trends are missing for most insect species (IPBES 2019). To date, only two out of 26 European biodiversity indicators are directly based on the abundance and distribution of select taxa: the common bird index and the grassland butterfly index (EEA - European Environment Agency 2012). Butterfly monitoring, with divers methodological approaches, has a long history, with the first schemes starting in Europe and North America in the 1970s (Brereton et al. 2010; Taron and Ries 2015). Due to extensive and constantly growing monitoring networks such as the European butterfly monitoring scheme (eBMS) (Sevilleja et al. 2020), butterflies are one of the best-studied insect groups. Similar schemes for other insects are lacking even though the importance of such programs is widely acknowledged by science (e.g., Magurran et al. 2010; Dicks et al. 2016; Breeze et al. 2021). Recent developments such as the EU pollinator initiative, which proposed a monitoring mechanism as part of the EU Biodiversity Strategy (Potts et al. 2021) and the development of the European soil monitoring (Montanarella and Panagos 2021) show the growing political interest in the monitoring of insect biodiversity.

For the international mapping and monitoring of species populations, the integration of regional and national observation data is needed (Jetz et al. 2019). The combination of data from different monitoring programs can provide benefits such as a broader database with extended spatial and temporal coverage and the potential to draw more robust and general conclusions (Henry et al. 2008). However, data integration can also be challenging due to variations in methods, spatiotemporal and taxonomic scope, collection purposes, and terminologies (Guralnick et al. 2018; König et al. 2019). Due to the high value of long-term monitoring data (Lindenmayer et al. 2012), keeping methods constant over time is often considered more important than uniform methods in the monitoring network when integrating existing schemes. Even if monitoring schemes are newly established, other objectives of the regional programs or differences in local conditions can oppose the use of standardised methods in a monitoring network (Parr et al. 2002). When integrating monitoring schemes, the applied methods should be evaluated regarding their compatibility to ensure that reliable conclusions can be drawn from the data. This is also of importance when studies on different species or taxonomic groups combine different methods depending on the appropriate survey design for the targeted taxa (Montgomery et al. 2021).

Butterflies are one of the few taxa for which a standardized quantitative sampling method is commonly applied in many different countries – Pollard walks, a transect count method (Pollard and Yates 1993). Pollard walks have long been established and used in the eBMS (Schmucki et al. 2015), North American butterfly monitoring networks (Taron and Ries 2015) and many other studies on butterflies with different objectives (e.g., Bruppacher et al. 2016; Dániel-Ferreira et al. 2020; Kolkman et al. 2021). The observation protocols in general follow the original recommendation of Pollard and Yates (1993) regarding weather conditions and the distance from the observer at which butterflies are counted. The length of transects differs substantially and ranges from short transects of 50 m (Loos et al. 2014) to a few kilometres (Herrnando et al. 2016).

Apart from transect counts, area and point counts are widely applied visual survey methods in the monitoring of butterflies and other insects that can be easily identified in the field (Montgomery et al. 2021). In contrast to the mostly standardised transect counts, there are varying applications of the area and point counts. For point counts, observers count all butterflies around them during a defined period of time. Point counts are now commonly applied in citizen science projects such as the Big Butterfly Count (Dennis et al. 2017) or can be applied when sites are difficult to cross, e.g., due to dense vegetation or sensitive habitats (Montgomery et al. 2021). Implementations of the method differ substantially in the time spent surveying, the distance from the surveyor at which butterflies are counted and the shape of the surveyed area (e.g., full or semi-circle) (Henry et al. 2015; Dennis et al. 2017; Lang et al. 2019).

Counts in a defined area also show varying sampling designs. The main difference in area counts is that while transect counts have a fixed path that is walked by the observers, area counts have no such fixed path. In some studies, large areas are surveyed while recording the survey effort (time) but are not covered systematically (e.g., the 4th of July butterfly count in the US on survey circles with a 7.5-mile radius; Swengel 1990). Area counts as defined by Hardersen and Corezzola (2014) as plot-based surveys are a more standardized approach, as a defined amount of time is spent on each survey site. From here on, we use the term area-time count for such standardized approaches where a patch of a defined size is surveyed for a defined amount of time. Area-time counts were used for butterfly counts in studies that considered different taxonomic groups on the same survey sites (Su et al. 2004; Grill et al. 2005; Marini et al. 2009) or assessed the impact of environmental factors on butterfly communities (Debinski et al. 2001; Fiedler et al. 2017). Area-time counts provide a thorough assessment of species richness at a specific site and are therefore advantageous in heterogeneous landscapes with small habitat patches. Furthermore, they provide better coverage of rare species, facilitating the analysis of drivers of spatial and temporal patterns in butterfly abundance and richness. However, transect counts can be implemented more easily in a citizen science context, as has been done for many decades in different countries (van Swaay et al. 2020), as transects can very often follow existing trails, and hence, disturbance of grassland sites can be minimized.

International biodiversity monitoring approaches such as the envisaged EU Pollinator Monitoring Scheme (Potts et al.2021) call for a standardization of methods as well as an integration of different methods. Ensuring data comparability is crucial when methodologies differ (Parr et al. 2002). We therefore compared two commonly applied butterfly survey methods: the traditional Pollard and Yates (1993) transect counts and area-time counts. For this purpose, we analysed data from 144 sites from a newly established butterfly monitoring scheme in western Austria where both methods were applied simultaneously.

To evaluate whether transect counts and area-time counts could be applied alternatively for the calculation of trend indicators in the context of long-term butterfly abundance monitoring, we analysed whether and how the two methods influence the estimation of different biodiversity variables. At the site level, we used variables such as species richness, species abundance and butterfly habitat quality. At the regional level, we used variables that describe the regional population state (and trend) of individual species, such as relative abundance (summed over all sites) and site occupancy (the number of sites on which a species was detected).


Study area and survey sites

The study area comprises the Austrian federal states of Vorarlberg and Tyrol in the Eastern Alps, with an area of 15.241 km2. In each state, 100 survey sites were designated in grassland habitats using a stratified random and spatially balanced sampling design. The sampling design aims to representatively survey all grassland habitats of the study region and to select homogenous habitats at individual sites. This approach allows for the implementation of probability-based designs – in our case, a stratified random sampling – while maintaining a spatial balance among the selected locations (Theobald et al. 2007). With a spatially balanced sampling design, the sampling locations are evenly spread over the entire survey area (or strata). A spatially balanced sampling hence supports efficient surveys when response variables have a spatial trend, even if the spatial pattern of this trend is not known before the sample is drawn (Kermorvant et al. 2019). The additional use of strata enables the application of variable sampling density if the area covered by the different strata is of substantially different size, shape or habitat quality. The applied strata should ideally cover more or less homogenous habitat types or homogenous response units delimited by basic landscape (e.g., relief) and land use or land cover (LULC) characteristics (Havlík et al. 2011; Schirpke et al. 2020). This enables us to better account for LULC changes when calculating butterfly trend indicators, supports a space-for-time substitution to model past or future trajectories of butterfly communities where historic baseline data are missing (Blois et al. 2013; Montgomery et al. 2020) and facilitates future habitat analysis and modelling.

Therefore, we divided all grassland habitats of the federal states of Tyrol und Vorarlberg into four strata: (1) flat valley meadows with a slope gradient of less than 15%, (2) hillside meadows with a slope gradient of 15% or more, (3) high mountain to alpine meadows - basically all meadows above the current timberline, and (4) all grasslands in protected Natura 2000 sites (European Commission 2020). Slope gradient (to differentiate (1) and (2)) was used as a proxy for management intensity, as the intensity of management practices generally decreases with increasing slope gradient (Rüdisser et al. 2015). In each federal state, 25 sites (covering all strata) are surveyed per year. Therefore, four years are needed to survey all sites, and repetition starts in the fifth year. In Tyrol, data are already available from all 100 sites surveyed in the years 2018–2021. In Vorarlberg, monitoring started in 2020 and hence covered 50 sites from 2020 to 2021.

Butterfly surveys

The butterfly survey took place from 2018 to 2021. To optimize the sampling effort in relation to the additional information gain, we opted to survey each site four times in one year (as described above) between mid-May and early September (Barkmann 2020; Hardersen and Corezzola 2014; Lang et al. 2016; Roy et al. 2007). Observers selected survey dates flexibly to optimize weather conditions (see next paragraph) and to achieve on average three to four weeks and a minimum of at least one week between two consecutive surveys. This approach aims to broadly capture the butterfly species composition of the entire season (Stewart et al. 2020; Hardersen and Corezzola 2014).

At six of the study sites, only three surveys took place due to unfavourable weather conditions at high elevations. These six sites were excluded from the analysis, leaving 144 sites (Fig. 1).

Fig. 1
figure 1

Map showing the location of the 144 survey sites in the Austrian states of Tyrol and Vorarlberg

The butterfly surveys were carried out between 10:00 and 17:00 in warm (13 C or more), dry and calm windy conditions (Beaufort scale: 0–3) following Pollard and Yates (1993). At temperatures below 17 °C, surveys were only conducted in full sunshine. At temperatures over 17 °C, cloud cover of up to 40% was permitted. To allow a comparison of the two methods, all surveys were carried out in two phases:

First, a 50-m transect was sampled as described by Pollard and Yates (1993) by walking slowly at a speed of approximately 10 m min− 1, sampling all sitting and passing butterflies in an imaginary rectangle of 2.5 m on the sides and 5 m in front. In exceptional cases, the transect was unilateral, i.e., 5 m on one side. This variant was chosen, for example, if there was no suitable path through the area and the transect was only possible at the edge of the area. The resulting walking time per 50-m transect was approx. 5 min.

Second, an area-time count was carried out for an additional 25 min. For the area-time count, the area around the original transect was extended to 1000 m², i.e., the area 10 m to the left and right of the transect. During the area-time count, the area was walked slowly in winding lines, avoiding multiple counting of the same individuals.

Double-counting during the survey cannot be completely ruled out but was avoided as best as possible. Individuals that were already detected during the transect count were therefore not recorded again during the following 25-min area-time count. Although the complete area-time count of the survey sites comprises both phases of the survey, the two phases were separated for the data analysis presented herein to obtain two independent datasets.

All butterflies were identified either in-flight or caught and identified in-hand before release. Animals were caught only if necessary to ensure species identification. If this was the case, the observer interrupted the survey time to either identify the butterfly, take a picture or collect for later identification and then resumed the survey where they had interrupted it. Only in exceptional cases were individuals collected for later identification. The cryptic species pairs Aricia agestis/artaxerxes, Colias hyale/alfacariensis, Leptides sinapis/realis and Pyrgus malvae/malvoides are difficult to differentiate in the field and were therefore treated as species complexes.

Indicators and data analysis

Data analysis aimed to understand and describe the influence of the two different survey methods on butterfly abundance and diversity estimates. For this purpose, we compared the results of the 50 m transect counts, which took 5 min, with the results of the 25-min area-time counts for species richness, butterfly abundance, and butterfly habitat quality (BHQ) at the site level and regarding individual species abundance and site occupancy at the regional level. BHQ is an indicator that combines min-max normalized abundance and species richness and is a suitable metric to compare different sites regarding their general habitat quality for butterflies (Rüdisser et al. 2017). BHQ was calculated as

(Eq. 1)

where \({S}_{i}\) is the species richness at site i and \({A}_{i}\) is the abundance at the corresponding site. \({S}_{Min}\) and \({S}_{Max}\) are the minimal and maximal species richness of all sites, and \({A}_{Min}\) and \({A}_{Max}\)are the minimal and maximal species abundance of all sites.

For all analyses, data from the four visits per site were summed. For individual species abundance at the regional level, all counts were summed, and site occupancy was determined by identifying the number of sites where the specific species occurred at least once.

For each of the abovementioned variables, the correlation between the data from the transect count and from the area-time count was calculated using the Pearson correlation coefficient. For analysis of BHQ, the square root of the indicator was used for both methods for a less skewed distribution of the data. There were some outliers with a high number of individuals for the abundance data for both the individual species and survey sites. Thus, the data from both methods were square-root transformed, and the correlation was calculated for both the original and the transformed data.

Additionally, we compared the individual-based species accumulation curves of both survey methods for all four strata. The species abundance data from all sites within a stratum were summed, and species richness was interpolated and extrapolated as described in Chao et al. (2014) with 50 bootstrap replications. The data analysis was conducted for all sites and additionally for the different strata individually.

Data analysis was conducted in R (Core Team 2021). The iNEXT package (Hsieh et al. 2016) was used to calculate rarefaction and extrapolation curves, and the ggplot2 package (Wickham 2019) was used for data visualisation.


During the 576 surveys, we made 9817 butterfly detections and observed 119 species. A total of 7499 butterflies were counted during the area-time counts and 2318 during the transect counts. The time between two consecutive surveys at the same site ranged from 7 to 63 days, with a mean of 23.9 days (SD: 8.9 days). For 95% of two consecutive surveys, the time between the visits ranged from 11 to 42 days.

The butterfly abundances detected per site with the area-time counts ranged from two to 211 (mean: 54.2, SD: 41.9) and with the transect counts from one to 103 (mean: 16.6, SD: 15.7). With the area-time counts one to 30 (mean: 12.8, SD: 6.2) and with the transect counts one to 18 (mean: 6.4, SD: 3.8) species were detected per site. Individual species were detected on average at 2.3 times more sites with the area-time counts than with the transect counts. On average, 2.4 times more species per site were detected with the area-time counts than with the transect counts (mean of the proportion calculated for individual sites).

Individual species abundances at the regional level, and hence over all survey sites, ranged from zero to 905 (mean: 63.0, SD: 119.4) individuals detected with the area-time counts and from zero to 292 (mean: 19.5, SD: 37.1) individuals detected with the transect counts. With the area-time counts, individual species were detected at zero to 106 (mean: 15.5, SD: 19.7) sites, and with the transect counts, they were detected at zero to 64 (mean: 7.8, SD: 11.6) sites (Supplementary information Table S1). Three species were detected during transect counts only, and 20 species were detected during area-time counts only.

The correlation analysis for the data at the regional level showed a strong linear relationship between the area-time counts and the transect counts, with a Pearson correlation coefficient of r = 0.96 for the species site occupancy, r = 0.96 for the untransformed species abundance and r = 0.96 for the square-root transformed species abundance (Fig. 2). The correlation coefficients for the individual strata ranged from r = 0.77 to r = 0.94 for site occupancy, from r = 0.70 to r = 0.98 for untransformed abundance and from r = 0.84 to r = 0.96 for the square-root transformed abundance. All correlations were significant with p < 0.001. The abundances of two species showed a noticeable deviation: Pieris rapae was counted more often and Cupido minimus less often on the area-time counts compared to the transect counts than would be expected from the linear correlation between the two variables.

Fig. 2
figure 2

Scatterplots of the regional indicators for 119 species obtained with data from the transect counts and with data from the area-time counts; r = Pearson’s correlation coefficient; dashed line: 1:1 reference line; solid line: linear regression line; p value < 0.001 for both. a: square root of the abundance per species over all survey sites; b: site occupancy (number of sites a species was detected on)

The rarefied species accumulation curves based on the data of the two survey methods consistently overlapped for all strata except ‘high mountain to alpine meadows’. Species accumulation curves based on the area-time count data were largely saturated across all strata, while those based on the transect data were not saturated (Fig. 3).

Fig. 3
figure 3

Comparison of the species accumulation curves of the two survey methods (area- time count and transect count) based on the number of individuals recorded. a: flat valley meadows with a slope gradient of less than 15%; b: hillside meadows with a slope gradient of 15% or more; c: high mountain to alpine meadows - basically all meadows above the current timberline; d: grasslands in Natura 2000 sites, e: all sites

At the site level, the correlation analysis also showed a clear linear relationship between the data from the transect counts and the area time counts, with Pearson correlation coefficients of r = 0.80 for butterfly abundance per site, r = 0.85 for square-root-transformed abundance, r = 0.81 for species richness, and r = 0.86 for the square root of BHQ (Fig. 4a-c). The correlations at site level within the individual strata ranged from r = 0.72 to r = 0.87 for untransformed abundance, from r = 0.79 to r = 0.85 for the square root transformed abundance, from r = 0.73 to r = 0.81 for species richness and from r = 0.80 to r = 0.88 for the square root of the BHQ. All correlations were significant at p < 0.001.

Fig. 4
figure 4

Scatterplots of the biodiversity indicators over 144 survey sites obtained with data from the transect counts compared with data from the area-time counts; p value < 0.001 for all correlations. r = Pearson’s correlation coefficient; dashed line: 1:1 reference line; solid line: linear regression line; a: butterfly abundance at the individual survey sites; b: species richness at the individual survey sites; c: square root of the butterfly habitat quality indicator (BHQ) at the individual survey sites


In the face of the observed insect decline, the importance of sound insect monitoring is being increasingly emphasized. The envisaged establishment of EU-wide pollinator monitoring (Potts et al. 2021) could be a major and important step towards better insect conservation (Samways et al. 2020; Warren et al. 2021) and the establishment of comprehensive European biodiversity monitoring. Although there already exists broad experience in butterfly monitoring, the development and establishment of transcontinental pollinator monitoring is still in its infancy. Standardization of methods as well as the development and integration of new innovative approaches are essential for successful and sustainable implementation. For this purpose, butterfly survey methods must be compared to assess their efficiency and mutual compatibility. We added to this evaluation with an extensive comparison between transect and area-time counts.

Not surprisingly, we observed more species per site with the more extensive area-time counts than with the transect counts. The 2.4-times higher species richness per site for the area-time counts also resulted in 2.3-times the number of sites where a species was detected. This led to a more complete coverage of the butterfly community composition in the different strata. The species accumulation curves were saturated across all strata when based on area-time count data. The species accumulation curves for transect count data were not yet saturated due to the lower sampling effort, but were very similar to the curves for the area-time counts for all but one stratum. For the alpine meadows the curves diverge slightly. This might be caused by the high habitat diversity combined with naturally lower butterfly abundance in this stratum. However, we assume that species accumulation curves for transect counts would saturate too with increased sampling effort (more sites or higher sampling frequency).

Both methods – transect counts and area-time counts – led to surprisingly consistent proportional results regarding relative butterfly abundance (r = 0.96) and species occupancy (r = 0.96) on the regional scale. The observed strong linear relationship between the local diversity estimates of abundance (r = 0.80), species richness (r = 0.81), and BHQ (r = 0.86) based on the transect counts and on the more comprehensive area-time counts is very good news. This confirms both the robustness of the transect counts and their compatibility with the more comprehensive area-time counts.

While the transect counts with 50-m-long transects and four surveys resulted in a cumulative survey time of approximately 20 min per transect and year – covering an area of 250 m² – the area-time counts resulted in a cumulative sampling time of 100 min per site and year – covering an area of 1000 m². However, the strong linear correlation between the data from the transect counts and from the far more comprehensive area-time counts indicate that even a few surveys with relatively low effort can be highly informative. This is true for both population monitoring at the regional scale and for local site comparison regarding diversity and overall habitat quality.

While documented differences in sampling effort can be accounted for when combining data from multiple monitoring schemes, differences in the detectability of distinct species can lead to severe biases (Kellner and Swihart 2014). It is well known that visual survey methods can underestimate the presence of cryptic or sedentary species (Dennis et al. 2006). Distance sampling and mark-recapture studies revealed that butterfly detectability during transect counts varies substantially among species and is influenced by colour, wingspan, and behaviour (Isaac et al. 2011; Pellet et al. 2012). Therefore, the data obtained from active visual counting methods do not reflect true butterfly abundance but rather species-specific relative abundances (van Swaay et al. 2008). In contrast to the transect counts, the area-time counts do not follow a fixed path and hence could lead to a changed detectability of specific species via disturbance or better capturing of the spatiotemporal variation of resources (e.g., nectar availability, microclimate, water) (Dennis et al. 2006). We showed that this was not the case. This is in line with Kadlec et al. (2012), who compared area-time counts and transect counts at ten sites in the Czech Republic and found no difference between the detectability of mobile and imperceptible species between the two methods.

There was a strong linear relationship between the species abundance counts from the two methods. Only two species showed a considerable deviation from the linear correlation (Fig. 2a). For Pieris rapae, more individuals were detected with the area-time counts than would have been expected from the abundance counted on the transects. In contrast, Cupido minimus was counted in proportionally higher abundance on the transect counts. This deviation is likely due to mud-puddling, a behaviour that is often observed in adult butterflies and can lead to aggregations of individuals (Downes 1973; Boggs and Dau 2004). Such aggregations can cause high local densities of adult butterflies on open spots such as paths. This was observed on one transect in early July, during which 70 individuals of Cupido minimus were counted on a transect. On the subsequent area-time count, only two individuals were detected. Consequently, the respective site also deviates from the linear correlation for on-site butterfly abundance (Fig. 4a). In this case, the transect count likely overestimates abundance, while the area-time count provides a more realistic assessment. In this context, note that we separated the data from the transect counts and the consecutive area-time counts to obtain independent variables for the presented methodological analysis. This might have contributed to an underestimation of the butterfly abundance and species richness estimates based on area-time counts in relation to the estimates based on transect counts.

The similar detectability of butterfly species with both methods supports the integration of data from area-time counts and transect counts for the analysis of trends in butterfly abundance. Differences in the time spent surveying and the area of survey sites do not hinder a comparison and combination of abundance data because, unlike species richness (Gotelli and Colwell 2001), species abundance is scale-independent. Due to this and other favourable mathematical characteristics, such as monotonicity and proportionality to species’ declines and increases (van Strien et al. 2012), the geometric mean of species abundance is often used for the calculation of indicators. These indicators, such as the living planet index (Loh et al. 2005) and the UK wild bird indicator (Gregory et al. 2008), generally assess and combine population trends of multiple species. The EU butterfly indicator for grassland species (van Swaay et al. 2019) and additional indicators developed in the eBMS framework (Schmucki et al. 2015; van Swaay et al. 2020) are also based on changes in the geometric mean of species abundance.

For the calculation of international butterfly indicators, density estimates of the contributing national (or regional) butterfly monitoring schemes are combined (van Swaay et al. 2020). A comparison of methods – as was done here – allows us to assess the factor by which the detected species densities differ between methods and hence enables the conversion of density estimates and facilitates the calculation of collated indicators.

Depending on the aims and requirements of a monitoring scheme, area-time counts can be a reasonable alternative to the commonly applied transect counts. Transect counts can be implemented much easier in a citizen science context, as has been done for many decades in different countries (van Swaay et al. 2020), because transects can follow existing trails and hence disturbance of grassland sites can be minimized. Shorter transects that stay within the same LULC class can give a good impression of butterfly habitat quality, as shown in this study, but more detailed information on butterfly fauna, such as data on rare species, requires more extensive surveys. Area-time counts provide a more complete assessment of species richness and community composition at a specific site and are therefore especially useful in small-structured anthropogenic-influenced landscapes. Next to the information gain, area-time counts are also more efficient regarding the time spent travelling and surveying. Much effort in systematic monitoring does not result solely from the mere survey time but rather from time travelling to sites and documenting site and weather conditions. Area-time counts allow more thorough assessments of a site when focusing surveys on a given location. They enable a closer connection to specific habitat types and better comparability with other plot-based monitoring schemes (Levanoni et al. 2011; Hardersen and Corezzola 2014). Especially in heterogeneous landscapes, this approach facilitates the selection of sites that cover a homogenous LULC class and are influenced by the same local environmental factors. In contrast, transects with a comparable survey effort often lead through diverse habitats (Caritg et al. 2011; Taron and Ries 2015), which makes it more challenging to relate monitoring data to environmental variables. In some monitoring schemes, transects are divided into shorter sections that lead through homogenous habitats to overcome this obstacle (e.g., Luppi et al. 2018). However, sections of uniform length do not often match changes in habitat, while sections with variable length impede the comparison of data from different habitat types due to differences in survey effort. Area-time counts are therefore better suited than transect counts when aiming to analyse the drivers of temporal and spatial variations in butterfly species richness, abundance and community composition, particularly on small scales (Guariento et al. 2022). Going beyond the detection of population trends and investigating the causes of declines in butterfly abundance is crucial for taking adequate conservation measures.


Area-time counts and transect counts resulted in similar assessments of relative abundance and site occupancy on regional level. Both methods can therefore be used alternatively or synergistically for the calculation of abundance-based trends. Area-time counts have many benefits for the analysis of drivers of spatial and temporal patterns in butterfly abundance and richness because they provide a better coverage of small habitat patches. However, transect counts can be a valuable complementation to area-time counts in order to enhance spatial or temporal coverage e.g., in a citizen science context.