Introduction

The population of flying insects is declining in north western Europe as described by Hallman et al. (Hallmann et al. 2017). However abundance trends in insects are rare, even in a well-studied country like the Netherlands (Kleijn et al. 2018). Multiple studies show a decline in butterflies (Warren et al. 2021) and moths (Fox et al. 2021; Hallmann et al. 2020). Assessing abundance trends in insects is often difficult because they are hard to monitor and identify. Moths however can be sampled in a standardized way with light traps and are nowadays relatively easy to identify in Northwestern Europe with new field guides and image recognition. With approximately 2400 species in the Netherlands, moths are a species rich group. These 2400 species can be divided into approximately 860 macro moths and 1450 species of micro moths (Waring and Townsend 2017). The knowledge about moths is growing, and recent studies underline their importance in the ecosystem as pollinators (Macgregor et al. 2015; Macgregor and Scott-Brown 2020). Furthermore they serve as a food source for insectivorous birds like great tits (Mols and Visser 2007) and as herbivores (Bernays et al. 2004). Moths often have a specific relation with their host plant and, in combination with their short life cycle, they are considered to respond quickly to environmental changes, making them good indicators for the quality of different types of grasslands (Pöyry et al. 2005). The combination of high species diversity and specific habitat preferences, makes that moths occur in almost all natural habitat types in the Netherlands, including agricultural and urban areas. Therefore, moths are likely suitable indicators for habitat quality for many different habitat types. Moth trends can also give insight in other environmental impacts, using trait analyses (Coulthard et al. 2019; van Langevelde et al. 2011).

Moths have been monitored for a long time in the Netherlands, with the oldest records dating back to 1672 (Ellis and van Deijk 2023). In the early days of moth trapping it was most common to collect and rear larvae or use oil lamps to attract them to light (Ingpen 1827) to be able to identify the species. When electricity and generators became commonly available, people more often attracted them with mercury vapor lamps and big white sheets or towers, on which the moths will settle down, which made identification possible. Since this data is collected with different lamp types, difference in effort and not always on exactly the same location, this data cannot be used for population trends.

To assess population trends it is necessary to monitor in a standardized way. This is difficult when moths are manually attracted to light in combination with a sheet or tower, because it is challenging to count all individuals. Also, the trapping effort differs between nights and the monitoring needs to be done on exact coordinates. For these reasons, light traps are used for monitoring schemes, such as the Rothamsted insect survey in the UK since 1968 (Williams 1948), the monitoring scheme in Belgium since 2009 (Veraghtert et al. 2009) and in the Netherlands since 2013 (van Deijk et al. 2019). Within the Rothamsted insect survey, a 200 W incandescent lamp is used (Williams 1948). These lamps have a limited life span, a high proportion of a high wavelength, and little UV, which makes them inefficient in the attraction of moths (Brehm 2017). They also consume a lot of energy. In Belgium the standard lamp type are Mercury Vapor lamps, with more UV and therefore more attraction capability. Although Mercury Vapor lamps were the commonly used lamps in the first years of monitoring in the Netherlands as well, nowadays different lamp types are used, as long as the same type is used for the same location over time. Mercury Vapor lamps can attract many moths, however, besides the disadvantage of them consuming a lot of power, which makes it difficult to use them in remote areas without a generator, they have also been banned in the EU in 2015 (European comission 2021).

To obtain standardized data suitable for various applications such as trend analysis or evaluating habitat restoration measures, as mandated by the European nature restoration law, that aims to restore degraded ecosystems and increase biodiversity, particularly pollinators, monitoring is essential. Therefore, we have developed a lightweight moth trap (called: LedTrap) that can be deployed in large numbers with relatively minimal effort.

Due to the fact that most moths are being attracted to near ultraviolet (UV) light (Brehm et al. 2021; van Grunsven et al. 2014a; van Langevelde et al. 2011) the Heath trap (Heath 1965) was designed with a 6 W UV-actinic fluorescent tube, with an emission peak at 365 nm (Donners et al. 2018). The advantage of this, is that this trap can be run on heavy batteries, which makes their usage in remote areas without main power feasible. In combination with an attraction range of 25 m (van Grunsven et al. 2014) they are also used for experiments with moths (Bates et al. 2013; Merckx and Slade 2014; van Grunsven et al. 2020) or within monitoring schemes (Bates et al. 2013). The disadvantage of Heath traps is, that the required batteries are heavy, the actinic does not always reliably switch on and they are not energy efficient. The introduction of energy efficient light-emitting diodes (LED) that emit near UV opens up new possibilities in the monitoring of nocturnal moths because they can be run on a lower capacity power bank, making the whole trap more portable in comparison to the heath trap.

This model of the LedTrap was based on the Heath trap, and, like the Heath trap, intended primarily for use where main power or generators are not available. Therefore its effectiveness was compared to this trap. A downside of LEDs, and many other light sources as well, is that they degenerate over time. This potentially leads to a reduction in the number of moths caught as a result of degeneration of the light sources, that could cause the emitted light to be less bright, and thus reduce its attractiveness to moths. This can result in false negative trends of nocturnal moth species. Therefore we studied the aging effect of the light source comparing new 2835 LEDs with LEDs that had a runtime of 1000 and 2000 h. The 2835 LED-strips are already in use in the Dutch moth monitoring scheme, but the number of moths caught in urban areas with these traps is relatively low. Therefore, we want to have an additional light source, that attracts more moths in urban areas to keep volunteers enthusiastic. In this study we researched the differences in total numbers of individuals, numbers of species and species composition between the Heath trap with an actinic and the Heath trap with the 2835 LED, the effect of run time hours on LED-strips and the effect of a LED with higher output compared to a 2835 LED. We will test the latter question in three different habitat types.

Materials and methods

This study was done with three experiments across the Netherlands; (1) Comparison between the regular Heath trap with the 6 W actinic and the Heath trap with new 2835 LEDs in a mixed forest near Arnhem. (2) A comparison between three different lamp-types, the LedTrap with new 2835 LEDs and LedTraps that were fitted with similar LEDs with 1000 run time hours, and new 5050 LEDs in three habitat types in Onstwedde and Dwingeloo. (3) A comparison between the LedTrap with new 2835 SMD LEDs, the same LEDs with 2000 run time hours, and new 5050 SMD LEDs in the urban area of Almere. Maps of the locations can be found in Online Resource 1.

In the first experiment a Heath trap ((Heath 1965), Fig. 1 left) was used to test the effect of the 2835 LED-strip compared to the standard Heath trap with 6 W actinic tube. In the second and third experiment the novel LedTrap was used with different LED-strips.

Fig. 1
figure 1

Left the original Heath trap and right the LedTrap in combination with a 2835 LED-strip

The LedTrap

The LedTrap is a funnel trap (Fig. 1 right), comparable in design to a Heath trap. The main difference is the LED light instead of an actinic tube and the use of a stackable, round, plastic pail instead of an aluminum box. The light source is a 50 cm LED-strip, which is spiraled around a 21 cm tube, the same length as the 6 W actinic used in the Heath trap, with an outer diameter of 11 mm. The tube is hanging between three plastic interceptors, moths hit these when attracted to the light and fall through the funnel in the trap. Inside the trap two 28 cm*28 cm egg boxes are placed where the moths can hide. The LedTraps were made for the Dutch Butterfly Conservation by the social working place Amfors, Amersfoort. The full construction plan of the LedTrap with all specifications can be found in Online Resource 2.

6 W actinic

The original Heath traps were used with a 6 W actinic fluorescent tube of 21.2 cm, with the highest emissions at 365 nm, connected to a lead-acid battery of 12 volts and 3,400 mAh. The actinic runs about 6 h on this battery. At the end of April, execution of this experiment, a night in the Netherlands lasts about 9 h (KNMI 2017).

New and aged 2835 LED-strips

The 2835 LED-strip is a 50 cm strip with 30 5 V 2835 SMD LEDs with an emissions peak at 389 nm and a total wattage of 2.3 W, hereafter called 2835 LED. The radiant flux per nanometer can be found in Online Resource 3. For the first experiment with the Heath trap with actinics and the Heath trap with the 2835 LED-strip, the same lead-acid batteries for both lamp types were used. The LED lasted around 16 h on this 12 V 3,400 mAh battery instead of the 6 h of the actinic. In the other two experiments the 5 V strip is connected to a 12 V power bank of 15,000 mAh using a transformer. This gives enough power for about 22 h of light, while being lighter and more compact than a lead-acid battery. During our experiments we initially used waterproof 2835 LED-strips, however, the UV affected the waterproof cover over the LEDs and caused browning, so less UV could get through and fewer individuals were caught. Therefore, we only analyzed the data from traps without waterproofing. For the second experiment, a set of the 2835 LED-strips were daily connected to the main power indoors, for 24 h a day, until they reached a run time of 1,000 h. During the second experiment in Dwingeloo and Onstwedde, they ran for 170 h our in the field, resulting in a total run time of 1170 h. Performing the third experiment in Almere, the same strips were connected to the main power indoors again for another 830 h, resulting in a total run time of 2,000 h.

5050 LED-strip

The 50 centimeter 5050 LED-strip, called 5050 LED from this point, is similar to the 2835 LED-strip but has 30 more powerful 5 V 5050 SMD LEDs, with an emission peak around 410 nm and a total wattage of 4.5 W. The 5 V strip is connected to a 12 V 20,000 mAh power bank using a transformer, giving the strip enough power to run about 15 h.

Experimental design

  1. 1.

    Heath trap with actinic VS Heath trap with 2835 LED-strip.

To test the difference between the Heath trap with the actinic and the Heath trap with the 2835 LED-strip, a randomized block design was used with five blocks consisting of a homogeneous vegetation. In every block three fixed locations were selected at least 50 m apart, between which the lamp types were randomly switched each night. The third location was used for traps with three power LEDs but these suffered from technical difficulties and thus, data from these traps is not used. In total, the traps were placed for six nights in a mixed forest near Arnhem, the Netherlands, between April the 19th and May the 5th, 2018.

  1. 2.

    2835 LED, 1000 runtime hours and 5050 LED experiment.

The second experiment compared the 2835 LED-strip, 2835 LEDs with 1,000 h runtime, and the more powerful 5050 LEDs. This experiment was conducted in Onstwedde and Dwingeloo. A fully randomized block design with twelve blocks with three locations per block was used. Around both villages two blocks were in urban area, two in a forest and two in an open area, to test for differences between illuminated and darker areas. The open area in Onstwedde was a wet grassland and in Dwingeloo it was a heathland. The traps in the open area were at least 100 m apart. In urban areas this distance was sometimes reduced to about ten meters. However, in these instances, the traps were positioned in such a way that they were not visible from each other, thereby avoiding direct competition. Between February the 22th and March 23th, 2021 the 2835 LED and 5050 LED were switched randomly between the six different locations within the block during eleven nights. Between March the 24th and April the 27th the 2835 LEDs with 1000 runtime hours were added to the design and the traps were placed sixteen additional nights.

  1. 3.

    2835 LED, 2000 runtime hours and 5050 LED experiment.

The third experiment, comparing the 2835 LED, the 2835 LED-strips with 2000 h runtime and the 5050 LEDs, was conducted in the city of Almere. The gardens were approximately 40 square meters in size and were separated by fences or hedges for privacy. Some traps were positioned about 10 m apart from each other. However, since only one trap was placed per garden, the lights did not directly compete with each other. A fully randomized block design with five blocks and three locations was used and applies in gardens of inhabitants who opened up their gardens for this experiment. The traps were placed 16 nights in the period between June the 21th and August the 11th, 2021.

During each experiment, the traps were emptied as soon as the sun rose on the morning following their placement. This process was carried out block by block, often taking several hours depending on the number of moths captured in the traps.

During each experiment, the traps were emptied starting at sunrise, the morning after they were placed. This was done block by block, often taking several hours depending on the number of moths captured in the traps. All moths were directly identified according to Waring and Townsend (2017) for the macro moths and Sterling and Parsons (2012) for the micro moths. Species that cannot be distinguished on the basis of external characteristics were recorded as aggregated (morpho-)species. All moths were counted per trap and released, at least one hundred meters away from the nearest trap. For experiment 1, the comparison with the Heath trap, only macro moths were identified. For the other two experiments both the macro and micro moths were identified. All moths in or on the trap were recorded and used in the analysis.

Statistical analysis

We compared the number of species and specimens in:

  • The multivariate prediction model demonstrated satisfactory diagnostic performance with an AUC of 0.946 (0.896–0.996).

  • the Heath trap with actinic vs. the Heath trap with a 2835 LED-strip.

  • the 2835 LED-strip vs. 2835 LED-strips with a of run time of 1000 and 2000 h.

  • the 2835 LED-strip vs. the 5050 LED-strip.

Similar models and tests were conducted for the separated analyses.

Heath trap with actinic VS 2835 LED-strip

We used generalized linear mixed models with a negative binomial distribution from the package lme4 (function ‘glmer’) (Lenth et al. 2022) in R version 4.3.1 (R Core Team 2023). The number of moths was used as response variable, and lamp type as explanatory variable of interest. Date was included as random factor and the block as fixed factor (glmer.nb(number ~ (1|date) + block + lamptype, data = dataset). Data with many zero observation records were all tested for zero inflation with the Rpackage DHARMa (Hartig 2022), no dataset was found to be zero inflated. The vegan package (Oksanen et al. 2022) was used to test for differences in species community. The raw data was transformed with the decostand function, whereafter the dissimilarity was calculated using the vegdist function (method Bray for Bray-Curtis dissimilarity). The multivariate dispersion was calculated with the betadisper function and the homogeneity of the variance was tested using the permutest function with 999 permutations. When the outcome of the permutest showed no significant inhomogeneity of variance, the data was tested with the adonis2 function with 99,999 permutations, where the overall significance over all terms was calculated. Families with more than fifteen observed individuals are analyzed on family level with the same method as above.

Runtime hours

The same model was used in the comparison between the Heath trap with actinics and the 2835 LED-strip. Data of the two different experiments could be combined, because the block, which includes the location, and date were included in the model. The same method as above was used to test for differences in species community. The input in the model was the data of the experiment with 1000 or 2000 h runtime.

5050 LED

The same analysis and models are used to compare the 2835 and 5050 LED-strip. For this experiment two datasets of different locations were also combined, because the block which includes the locationand date were included in the model, and because of the fully randomized block design.

Results

Heath trap with actinic VS heath trap with 2835 LED-strip

A total of 1788 macro moths of seven different families and 63 different species were caught. After the waterproof cover was removed, the Heath trap with the 2835 LED-strip attracted 2.3 times (lcl 1.8, upl 3.3, negative binominal glm, p < 0.01) more moths compared to the Heath traps with the actinic (Table 1, negative binominal glm, p < 0.01). The number of species trapped in the Heath trap with the 2835 LED-strip was also higher compared to the Heath trap with the actinic (Fig. 2, negative binominal glm, p < 0.001).

Fig. 2
figure 2

Significant more moth species were attracted to the Heath trap with the 2835 LED-strip compared to the Heath trap with actinic (glm negative binominal p < 0.001 with a 95% confidence interval). The y-axis shows the expected number of species of the model

Table 1 Average number of attracted individual moths per trap per night for all families together and separated per family. The second column gives the total number of observed (morpho-)species within the subgroup. The third column gives the average number in the Heath trap with 2835 LED-strip with the 95% confidence levels between brackets. The same in the fourth column for the Heath trap with the actinic. The last column gives the ratio of the Heath trap with 2835 LED over the Heath trap with actinic, with the 95% confidence levels between brackets

The higher number of individuals in the Heath trap with the 2835 LED-strip compared to the Heath trap with actinic was observed within four different families (negative binominal glm, Table 1): Erebidae (p = < 0.01), Geometridae (p < 0.01), Noctuidae (p < 0.01) and Notodontidae (p < 0.01). We did not find a difference in total number for the Drepanidae and Saturniidae (negative binominal, glm p > 0.05). The species composition of the moths caught with the two light sources also differed (permanova, p = 0.002).

New LEDs VS old LEDs

In total 2879 moths were trapped; 2059 macro moths of 109 species and 820 micro moths of 85 (aggregated) species. No difference in attraction of individuals was found between new 2835 LED-strips, and ones with 1000 h (negative binominal glm, p = 0.87) or 2000 h (negative binominal glm, p = 0.38, Fig. 3) runtime. We found no difference in species composition after 1000 h runtime (permanova, p = 0.63), but we did after 2000 h runtime (PERMONOVA, p = 0.039). When testing for the different habitat types separately, we also found no effect of runtime hours on the number of attracted individuals in urban areas (negative binominal glm, p = 0.41) nor the forest (negative binominal glm,p = 0.74), nor the open field (negative binominal glm, p = 0.18).

Fig. 3
figure 3

No significant difference was found between the number of individuals per trap per night between a new 2835 LED-strip compared to the ones with 1000 h (negative binominal glm, p = 0.87) or 2000 h (negative binominal glm, p = 0.38) runtime. The error bars show 95% confidence intervals

2835 LED VS 5050 LED

A total of 9209 moths were trapped: 7624 macro moths of 115 species and 1585 micro moths of 87 species. The 5050 LED trapped 1.99 (1.73–2.35) times as many moths compared to the 2835 LED (negative binominal glm p < 0.001, Table 2). Even though the numbers in the forest were 5.66 (4.71–7.11) times higher than in the urban area (negative binominal glm, p < 0.01) and 14.66 (12.09–17.23) times as high as in the open area (negative binominal glm, p < 0.01), we found similar ratio’s between both LED types in the traps in the forest, urban areas and the open area. The number of macro moths was 2.19 (1.88–2.63) times higher (negative binominal glm, p < 0.01) and the number of micro moths was 1.35 (1.15–1.64) times higher (negative binominal glm, p < 0.01, Table 2) in the trap with a 5050 LED compared to the 2835 LED. When the data was split and tested on family level, we found no significant difference between the number of species in the 2835 LED compared to the 5050 LED. Also, the species compositions of the catches with the two lamp types did not differ (permanova, p = 0.48).

Table 2 Average number of attracted number of individual moths per trap per night for both micro and macro moths together and split up. The second column gives the total number of observed species within the group. The third and fourth columns gives the average number in the traps with the 95% confidence levels between brackets. The last column gives the ratio of the 5050 LED over the 2835 LED, with the 95% confidence levels between brackets

Discussion and conclusion

The Heath trap in combination with a 6 W actinic is a commonly used trap for scientific experiments (Bates et al. 2013; Merckx and Slade 2014; van Grunsven et al. 2020). However, the heavy and unsafe lead acid batteries, the unreliable actinic lights and the limited running hours during a night makes this trap often impractical. Therefore we developed the LedTrap. The results show that the 2835 LEDs attracts on average 2.3 times as many moths as the Heath trap with the actinic. This is considered to be partly due to the fact that the Heath trap with actinics was running 2/3 of the night compared to the ones with 2835 LED. The latter ran the whole night, while they were connected to the same 12 V, 3,400 mAh battery, typically utilized in conjunction with this trap. This emphasizes that energy efficiency is one of the advantages of using LED over actinic. We saw a higher number of moths in the trap with 2835 LED within four different families.

The number of moths captured in the 2835 and 5050 experiments was found to be lower in urban (i.e. illuminated) areas in comparison to forests, with the lowest numbers observed in open areas. This pattern may be attributed to the moment the experiment was conducted, in early spring. Most of the larvae of the species that fly at this moment of the year feed on trees and shrubs, leading to a higher concentration of moths in forested environments. The urban area where the experiment was conducted was also relatively green with many trees and shrubs. This might have resulted in relatively high number of moths. Still we found lower number in our urban areas than in forested areas, which can partly be a result of light competition of the trap with streetlights, similar to the negative effect of brighter moon phases on catches with light traps (Yela and Holyoak 1997). To enhance moth capture rates, particularly in urban areas, we investigated the brighter 5050 LED. This LED attracted on average twice as many individuals compared to the 2835 LED-strip in both urban areas and nature reserves. Since we want to use both LED-strips, 2835 and 5050, in our Dutch monitoring scheme, we also tested the difference in species composition between the LEDs, and did not find a difference. This is in accordance with what Infusio et al. (2017) found, comparing species composition between incandescent lamps and UV LEDs, even though these two light sources differ markedly in spectral composition. Although we only sampled in short periods, we expect that the temporal limitations in our experiments are unlikely to influence the results, as we expect the relative differences to be similar throughout the year. We did find a higher proportion of macro moths compared to micro moths which could potentially be caused by a lower mobility of micro moths. Micro moths are generally smaller than macro moths, and it has been shown that for Noctuids, smaller species, measured as forewing length, indicates a lower relative dispersal capability compared to larger species (Jones 2014). Also, a significant majority of moth species in our study were associated with forests. Slade et al. (2013) showed that for macro moths that tend to inhabit forest environments, larger species have a better dispersal ability. Therefore it is possible that the smaller micro moths also are less mobile that the larger macro moths.

LEDs are often promoted as being durable and lasting for 20,000 h. However, shifts in spectral composition are common (Davis et al. 2016). Whether environmental conditions potentially speed up the degradation process remains to be studied but the brown discoloration of the waterproof coating could be an indication. Since we are already using the 2835 LedTrap within our Dutch monitoring scheme, we want to know that results do not reflect the decrease of attractiveness of the LED, but do truly reflect changes in moth composition. Therefore, we tested the attractiveness of the 2835 LEDs after 1000 and 2000 running hours. We found no difference between the LEDs with 1000 running hours compared to new ones, but we did find a small effect on the species composition after 2000 h runtime. This is only a small proportion of the 20.000 h, but reflect a significant number of trapping nights. It appears, considering our own experience and the experience of several hundred volunteers who are using the traps, LED strips develop mold or rust when wet, especially during autumn or spring. This might have happened for the strips after 2000 h runtime as well. The 30 LEDs in a strip are controlled by 10 drivers, each driving 3 LEDs. These drivers sometimes break down, which causes the 3 LEDs in a row to stop working. Therefore, we advise to check the LED-strips regularly, and replace strips when a driver has failed or when the strip has 1000 h runtime.

Possible usage

The newly designed LedTrap comes at a low cost and can be seen as a very portable, versatile trap which can both be used within a monitoring scheme, but also for other experiments that require standardized data collection of nocturnal moths. However, if a long term monitoring scheme already uses a different trap (i.e. Heath trap) we would recommend not to switch at once but use both traps for some time and gradually replace them, for example with the LedTrap.

The LED traps are already frequently used within the Dutch monitoring scheme. The 5050 LED will be a solid replacement for the 2835 LED in regions with a low moth density like the Netherlands, but in species rich countries like Spain the 2835 LED already attracts many moths (preliminary results SPRING project), and identifying all individuals might become cumbersome. In case when volunteers need to be kept enthusiastic without taking too much time to empty a trap, the 2835 LED would be recommended. The benefits of using LEDs for moth monitoring are also underlined by Infusino et al. (2017) by their energy efficiency and thereby their usage in remote areas without main power or the option to leave a generator behind.

The lights of this research can also be used for different experiments, besides a monitoring scheme. The weaker lights compared to, for example, a MV bulb results in attracting insects from a shorter distance. The attraction span of the 6 W actinic used in the Heath trap is around ten meters (van Grunsven et al. 2014), which is very likely to be more or less the same for the 2835 LED. This indicates that the traps can serve as tools to assess moth populations on a very local scale, comparing different habitats or evaluating effects of management measures. In the Netherlands, the traps are also used in agricultural areas where farmers contribute to data collection to monitor the effect of management actions (van Deijk et al. 2022).