LED streetlight characteristics alter the functional composition of ground-dwelling invertebrates

.

Ecological impacts include the fact that ALAN triggers many negative effects on the development and fitness of highly sensitive nocturnal invertebrates adapted to low light conditions (Falcón et al., 2020;Gaston et al., 2015).These impacts of ALAN on invertebrates include altered behavior (Knop et al., 2017;Lewis et al., 2020;Macgregor et al., 2017;van Geffen et al., 2015), disrupted circadian rhythms (Gaston et al., 2017), changes in reproduction (Honnen et al., 2019;Owens et al., 2020), and modifications in species interactions (Bennie et al., 2015;Deitsch & Kaiser, 2023;Giavi et al., 2021;Grenis & Murphy, 2019;Grubisic & van Grunsven, 2021;Sanders & Gaston, 2018b).As a result, the functional and taxonomic structure of invertebrate communities may change, as previous studies have shown in which predator abundance increased around light sources at night (Brown et al., 2023;Davies et al., 2012).These changes in community composition around sources of ALAN can alter trophic interactions and entire food webs, ultimately leading to a loss or a shift in ecosystem functioning (Owens et al., 2020).While the impacts of ALAN on flight-active invertebrates is rather well-documented (Fabian et al., 2024), ground-dwelling invertebrates, which are likely highly sensitive to ALAN, remain under-researched.Ground-dwelling invertebrates play essential roles in decomposing organic materials, nutrient cycling, and suppression of soilborne diseases and pest control (Cividanes, 2021;Delgado-Baquerizo et al., 2020).Therefore, changes in ground-dwelling invertebrate communities can have cascading effects on higher levels of biological organization.We therefore urgently need a better understanding of the impacts of ALAN on the community composition of ground-dwelling invertebrates.
At the same time, rapid advances in lighting technology have led to low-maintenance, targeted, long-lasting, and energy-efficient Light Emitting Diodes (LED) lighting systems.Their technical properties encompass e.g., the spectral distribution (Donners et al., 2018;Somers-Yeates et al., 2013;Spoelstra et al., 2017), different light levels (dimmability) (Bolliger et al., 2020;Owens & Lewis, 2018;Pauwels et al., 2021;Rowse et al., 2018), or the luminaire shape (Bolliger et al., 2022).Cold-white light-with a smaller wavelength and a higher peak in the blue range-has been shown to have harmful impacts on different species, as demonstrated for bats (Spoelstra et al., 2017) and insects (Brehm et al., 2021;Donners et al., 2018).In terms of illuminance, dimming outdoor lights can be a simple but effective way to reduce ALAN and make nights darker, benefiting nocturnal species (Bolliger et al., 2020;Rowse et al., 2018).In contrast, the shape of the luminaire used for ALAN is an often-overlooked factor, although Bolliger et al. (2022) found that diffusing luminaire shapes attract up to 16% more flight-active insects.The development of LED lighting systems with adjustable parameters-such as color temperature and light intensity-can provide a significant opportunity to mitigate the negative effects of ALAN.To effectively address and mitigate the ecological disruptions caused by ALAN, it is therefore important to quantify how the different characteristics of ALAN, both independently and in combination, affect the environment-especially in the context of LED technology.
We conducted a comprehensive field study over two summers at three different forest sites in Switzerland, to determine how three key attributes of LED lights (color temperature, brightness, and luminaire shape) affect the abundance and community composition of functional arthropod groups (predators, omnivores, and detritivores), as well as the predator-prey ratio, which indicates potential changes in top-down and pest control, respectively.We selected forest sites to test the impact on communities that have not been impacted by ALAN.The following hypotheses were tested: ALAN has an overall attractive effect on all functional arthropod groups, regardless of the attributes of the LED luminaires (color temperature, brightness, and luminaires shape), because light is thought to have a general attraction to arthropods as many processes in insect development and behavior are influenced by light (Owens et al., 2020) (H1).Since predators among arthropod functional groups rely on visual senses to capture prey (Gonzalez-Bellido et al., 2022;Lim & Ben-Yakir, 2020), and ALAN likely facilitates prey location, we expect the attraction effect of ALAN to be strongest for predators (H2).For the interaction of luminaire characteristics, we hypothesize that streetlights with full intensity, cooler light color and a light-diffusing shape are overall more attractive to ground-dwelling arthropods than dimmed, targeted and warmer light as shown in previous studies (Bolliger et al., 2020(Bolliger et al., , 2022;;Donners et al., 2018) (H3).Furthermore, according to our second hypothesis, we expect to catch more predators compared to omnivores and detritivores around street luminaires, indicating increased predation pressure, leading to higher predator-prey (detritivore) ratios (H4).

Study sites
The study was conducted for two consecutive summers during a total of 18 weeks-from June 09 to August 05, 2021 (8 weeks), and May 25 to August 17, 2022 (10 weeks)-at three long-term forest observation sites in Switzerland (https://lwf.wsl.ch/en/):(1) Birmensdorf (lowland site 518m asl with temperate mixed deciduous forest), (2) Lägern (eastern fringe of the Jura mountains at 866m asl with a temperate mixed deciduous forest), and (3) Alpthal (pre-alpine site at 996m asl with a temperate mixed conifer forest; Appendix A, Fig. 1).The sites were supplied with electricity, equipped with meteorological stations, and were not under the direct influence of ALAN prior to this study.

Light treatments
The experimental setting of each of the three study sites consisted of 12 commercial state-of-the-art LED streetlights (type Izylum 1, Schréder Swiss AG) mounted to aluminum poles at 2.5 m height plus two dark control plots (without artificial light, but otherwise same installation) at each site, giving us a total of 48 streetlights and 6 control plots.To avoid light interference between the treatments while taking into account the technical constraints of the infrastructure, a buffer distance of 30 m was maintained between all the light treatments and the controls.The LEDs were professionally installed, surveyed, and maintained by the employees of the electricity works of the canton of Zurich EKZ.
Three LED characteristics were tested individually and as interactions: three light colors (neutral white, 3700 K; warm white, K; and amber, 2200 K; Fig. 1 Aa), two light levels (100%, 65 ± 6.5 lux, and 50%, 32.5 ± 3.25 lux, of the full light intensity; Fig. 1 Ab), and two luminaire shapes (luminaire emitting light vertically (standard) and an enhanced horizontal light emission triggered by smooth-white plexiglass tubes attached to the streetlights (diffused);Fig. 1 Ac).The Plexiglass tubes used had a diameter of 150 mm, a length of 245 mm, and a light transmittance of 44% (Appendix A; Fig. 2).We employed a full factorial design for the three LED-light characteristics, assigning two light levels for each LED-light color temperature and accounting for two luminaire shapes.This resulted in 12 different LED-light treatments per site, supplemented by two dark controls, yielding a total of 14 plots per site (Appendix A; Fig. 1).

Luminaire description and technical adjustments
We used state-of-the-art LED streetlights of the type Izylum (Schréder Swiss AG).The luminous flux of all luminaires was standardized to 1500 lm (Appendix B; Table 1).The standardization was performed at the optical laboratory at the Swiss Metrological Institute METAS in Bern.
The spectral ranges of all luminaires showed a small (2200 K), intermediate (2900 K), or strong (3700 K) peak in the blue part of the spectrum (Appendix C, Fig. 1).As no differences in impact on invertebrates was observed between the 2200 K and the 3700 K luminaire in the first year (2021), we hypothesized that the blue peak in the 2200 K luminaire might be responsible for the lack of effects.Thus, in the second year (2022), we omitted the blue peak in the 2200 K streetlights by adding a filter.This resulted in a light source with a color temperature of 1900 K (the spectral range is shown in Appendix C; Fig. 1).Installing these filters allowed to assess potential attraction effects due to the spectrum's blue peak.However, the elimination of the blue peak in the 2200 K luminaire did not change the arthropod's responses relative to the 2200 K luminaire with the blue peak (Appendix D Fig. 1; Tables and 2).Therefore, we combined the data of the filtered 2200 K (1900 K) and unfiltered 2200 K luminaire, which will subsequently be referred to as 2200 K.
N. van Koppenhagen et al.

Sampling and grouping invertebrates
Ground-dwelling invertebrates were collected using standard pitfall traps (funnel traps with a diameter of 150 mm, see Lange et al., 2011).Two traps were deployed at each streetlight: the center trap (C) at approximately 1 m distance from the street-light pole (measured to yield 100% of the full light intensity; 65 ± 6.5 lux at ground level), and the peripheral trap (P) was set approximately 4.3 m from the street-light pole (measured to yield 10% of the full light intensity; 6.5 ± 0.65 lux; Fig. 1 B).
The pitfall traps consisted of a plastic pipe (150 mm × 280 mm) dug into the ground and a 750 ml bottle with a standard PVC funnel (150 mm diameter) attached (Lange et al., 2014), containing a 0.5% biocide solution (Rocima GT™).The funnel was installed level with the ground.The traps were emptied weekly over the two sampling periods.We identified and sorted the collected samples into 41 arthropod groups using a stereo microscope (Appendix E, Table 1 & Appendix F, Fig. 1).We grouped the sorted invertebrates into functional groups: predators (spiders, harvestmen, ground beetles, rove beetles, centipedes), omnivores (ants, slugs/snails, grasshoppers/crickets), and detritivores (springtails, mites, woodlice, millipedes) to measure changes in the functional group composition of the community (Fig. 2).Additionally, we built a predator-detritivore ratio to test whether predator-prey relationships change in response to LED treatments, which would indicate shifts in trophic interactions and ecosystem dynamics.Omnivores were excluded from this ratio, because they are predators and prey at the same time, although the relative importance may vary depending on the species.

Statistical analysis
All statistical analyses were performed using R Statistical Software (v4.2.2, R Core Team, 2021).We used linear mixed-effect models (LMM, package lme4; v1.1.30,Bates et al., 2015) to assess (1) the effects of color temperature, light intensity, and luminaire shape on the abundance of different ground-dwelling arthropod functional groups.We followed a two-step approach.The first step was to assess whether the invertebrates are attracted to light (regardless of the LED light characteristics combination) compared to the control dark plots and to assess the differences between the two pitfall trap positions (central position C & peripheral position P), each functional group (predators, omnivores, or detritivores) was fitted independently as a function of a control term (2 levels; dark vs. light (yes or no control)) and the pitfall trap position (2 levels; position C or P).Light color (3 levels), light intensity (2 levels), light shape (2 levels) were included as covariates.Two random effects were considered (week nested in year, plot nested in site).In a second step, we excluded the control term (as the control treatment is not full factorial) to focus on the effects of the light treatments and the  N. van Koppenhagen et al. interactions between these.Therefore, each functional group (predators, omnivores, or detritivores) and each pitfall trap position (central C or peripheral P) was fitted independently as a function of light color (three levels), light intensity (2 levels), light shape (2 levels), all pairwise interactions between light variables (i.e., color*intensity, color*shape, intensity*shape) as fixed effects with two random effects (week nested in year, plot nested in site).
To analyze the effects on the predator-detritivore ratio (2), we used a general linear mixed-effect model (GLMM, package lme4;v1.1.30, Bates et al., 2015) to assess the individual and interacting effects of color temperature, light intensity, and luminaire shape.The effects on the ratio of predators to detritivores (using the cbind command) were fitted as a function of light color (3 levels), light intensity (2 levels), light shape (2 levels, all pairwise interactions between light variables (i.e., color*intensity, color*shape, intensity*shape) as fixed effects.Three random effects were considered (week nested in year, plot nested in site, and an observation-level random term to account for overdispersion).
The explanatory variables entering the models were checked for overall model performance using the DHARMa package (v 0.4.6;Hartig, 2020) and multicollinearity using the performance package (v0.11.0,Lüdecke et al., 2021).Model performance was further assessed calculating the R 2 using the MuMIn package (v1.47.5;Barton, 2023) and the Akaike information criterion (AIC; Bonakdari and Zeynoddin, 2022) for the full model.

Arthropod abundance
A total of 349,779 invertebrates representing 12 taxonomic groups were captured at 48 streetlights during the 18 weeks over two summers.The most frequent taxa were springtails, mites, ants, spiders, and ground beetles, making up around 95% of the total number of invertebrates (Fig. 2).Among functional groups, detritivores were the most abundant with a total of 256,997 individuals, followed by omnivores with 53,088 and the predators with 39,694 individuals (Fig. 2).Detritivores consisted of 60% springtails, 36% mites, 2% woodlice, and 2% millipedes.

Effects of light versus dark controls on functional arthropod groups
Our study showed significantly higher abundances of predators and omnivores at the light treatments compared to the dark control plots (light vs. dark; averaged over both pitfall trap positions).The effect was most pronounced in omnivores with on average 275% more individuals at light treatments, similarly strong in predators with an increase of 70% and not statistically significant but still an increase of 15% in detritivores (Fig. 3 A, B & C).For detailed regression results see Appendix G (Tables 1 and 2).

Effects of central versus peripheral pitfall trap position on functional arthropod groups
All functional groups showed a significantly higher abundance close to the streetlights with higher light intensities at the central trap position (C; 100% full light intensity) than lower light intensities at the peripheral position (P; 10% of the full light intensity) (Fig. 3 D, E, F).Predators showed the highest difference between the two pitfall trap positions, with on average 100% higher numbers at the C position, followed by omnivores with an increase of 50%, and detritivores with 28%.For detailed statistical results see Appendix G (Tables 1 and 3).

Effects of the individual and interacting luminaire characteristics on functional arthropod groups
In the analysis of the individual LED characteristics on functional arthropod abundance, we observed no significant effects of the three LED-light characteristics (color temperature, light levels, and luminaire shapes) on the abundance of predators and detritivores (Fig. 4 A, D, G & C, F, J).Similarly, omnivores exhibited no significant response to LED color temperature and luminaire shape at 100% light levels (Fig. 4 B & E).However, omnivores showed significantly reduced numbers (− 57%)  when dimming the light from 100% intensity 50% intensity at position C (Fig. 4 H; t(24) = − 2.13, p 0.04).For detailed statistical results, see Appendix H (Tables 1, 2 & 3).
Testing interacting LED characteristics on arthropod functional groups, we rather consistently observe reduced abundances at light levels of 50% compared to 100%.However, it was only statistically significant in three cases, for predators at 3700K (with a decrease of 53%; Fig. 5 A), omnivores at luminaires with diffused light emittance (decrease of 74%; Fig. 5 E), and detritivores at standard luminaires (decrease of 27%, Fig. 5 H).Analyzing the interacting effect of the three light color temperatures compared to the two luminaire shapes, no significant attraction to either combination was detected in all three functional arthropod groups (Fig. 5 Thus, among all three individual and interacting LED parameters tested, we summarize that dimming from 100% to 50% is the most important measure to reduce the number of arthropod functional groups around artificial lights.While the analysis of the individual LED characteristics only revealed a significant result for omnivores (Fig. 4 E), and the interaction is only rarely statistically significant (Fig. 5 A, E, H) a clear trend is visible indicating mainly reduced numbers around streetlights dimmed to 50%.For detailed statistical results, see Appendix H (Tables 1 and 4).
To show the results of the interactions between the LED characteristics, we focused on the pitfall trap position C, as position P did not reveal significant results (for the results of the P position see Appendix H; Fig. 2; Tables 2 and 5).

Effects of luminaire characteristics and their interactions on predator-detritivore ratio
The analysis of the three LED characteristics on the predatordetritivore ratio resulted in no significant differences across all individual characteristics (color temperature, light level, and luminaire shape; Fig. 6 A, B & C).No effects were observed for the interactions between light level and luminaire shapes (averaged over the three LED color temperatures) and color temperature and luminaire shape (averaged over the two luminaire shapes; Fig. 6 E, F).However, a significant color temperaturelight level interaction effect on the predatordetritivore ratio was observed.At the 3700 K luminaires, the ratio was higher at 100% compared to the 50% light level (Fig. 6 D).The significantly higher ratio indicates an increased presence of predators relative to detritivores at luminaires with 3700 K and 100% intensity compared to those with 3700 K that are dimmed to 50%, potentially resulting in elevated top-down pressure on detritivores.For detailed statistical results, see Appendix I (Tables 1, 2 & 3).
To show the results of the individual LED characteristic and the interactions on the predator-detritivore ratio, we only included the results of the pitfall trap position C, as position P did not reveal significant results (for the results of the P position see Appendix I; Fig. 1; Table 4, 5 & 6).

Discussion
Our study on the influence of artificial light at night (ALAN) on the abundance of three functional groups of ground-dwelling invertebrates in forests revealed a strong nocturnal attraction to ALAN in general, particularly of higher intensity light, with the most pronounced effect on predators and omnivores.This suggests that the functional composition of ground-dwelling arthropod communities that have not yet been exposed to artificial light is significantly altered by ALAN, likely affecting trophic interactions and food webs, and ultimately ecosystem functioning.We found no significant effects of the LED color temperatures and luminaire shapes for all three functional arthropod groups, suggesting that color temperature and luminaire shapes may affect these functional groups less.However, we observed significant interactions between light intensities and certain LED color temperatures (3700 K) In line with our first hypothesis (H1), we observed an increased abundance in omnivores (+275%), predators (+70%) and detritivores (+15%) in light treatment plots compared to control plots.The significant and much stronger attraction of predators and omnivores to ALAN, stands partially in line with our second hypothesis (H2) and with the studies of Brown et al. (2023) and Davies et al. (2012).Both studies found a strong increase in predatory invertebrates under artificial light.In contrast, detritivores did not respond significantly and much weaker to ALAN, which might suggest a potential shift in community composition favoring higher numbers of predators and omnivores around ALAN sources.Increased abundances of predators and omnivores could lead to increased top-down pressure, influence predator-prey dynamics (Ditmer et al., 2021;Moyse et al., 2023;Nelson et al., 2022), and potentially compromise ecosystem services.A previous study evaluated the top-down and bottom-up effects of ALAN on invertebrates in grassland systems and found that under certain lights, the presence of predators can lead to a decrease in prey species of around 55% (Bennie et al., 2018).
Our study found no effects of the different commercially used LED color temperatures in the three functional ground-dwelling arthropod groups.This contrasts with our third hypothesis (H3) and previous studies emphasizing the impactful role of the cold-white light spectrum on invertebrates (Brehm et al., 2021;Briolat et al., 2021;Donners et al., 2018;Somers-Yeates et al., 2013) and bats (Spoelstra et al., 2017).Furthermore, our study showed no general effect of the luminaire shape on ground-dwelling functional groups.These results again stand in contrast to our third hypothesis (H3) and the study of Bolliger et al. (2022), who found that diffuse luminaire shapes significantly attract flight-active insects.A possible explanation for this could be, that ground-dwelling invertebrates may not be affected by luminaire shapes, as at a vertical distance of >2.5m the effect of a diffusor could be attenuated, in contrast to the exposure of aerial insects.When comparing the two light levels of 100% and dimmed to 50%, no significant effects on predators and detritivores were found.This contrasts with our third hypothesis (H3) and previous studies on flight-active insects (Bolliger et al., 2020;Owens & Lewis, 2018) and ground-dwelling arthropods (Davies et al., 2017) where they found reduced ecological impacts when dimming the light.However, in our study, omnivores showed higher abundance at the full light intensity.This suggests that omnivorous invertebrates respond more sensitively to the variations in the tested light intensities compared to the other two groups.Although, the pronounced differences observed between the two pitfall trap positions and between control (dark) sites and treatment (light) sites across all three arthropod groups suggest the presence of potential thresholds of light intensity for predators and detritivores as well.Therefore, exploring a wider range of light levels, including levels lower than 50% could help to identify specific thresholds at which the different arthropod groups exhibit significant behavioral changes.These Fig. 5.Estimated marginal means for the interactions of three luminaire characteristics at the pitfall position C (100% light intensity) (color temperature, light level, luminaire shape) on arthropod abundance for each functional group, predators (A, B, C), omnivores (D, E, F), and detritivores (G, H, J).The interactions between the light levels and temperatures are averaged across both luminaire shapes (A, D, G); the interactions between the light levels and luminaire shapes are averaged across the three color temperatures (B, E, H) and the interactions between the luminaire shapes and the color temperatures are averaged across the two light levels (C, F, J).The error bars show the standard error (SE).Significant interactions (p < 0.05) are indicated by an asterisk (*).findings show that other general factors (such as light or no light) of ALAN, than individual LED characteristics like color temperature, light levels, and luminaire shapes have stronger effects on ground-dwelling invertebrate communities in previously not light-exposed forests.
Therefore, even more important than investigating individual LEDlight characteristics is to understand how different light characteristics (LED color temperature, light levels, and luminaire shapes) interact to influence arthropod responses.While predators did not show significant responses to variations in light intensity, a significant interaction with LED color temperature was observed, with predatory invertebrates being less attracted to 3700 K luminaires dimmed down to 50% intensity (-53%) compared to luminaires at full intensity.Additionally, interactions of LED characteristics were found in omnivores and detritivores, with a significantly lower attraction to lower light intensity in diffused luminaires for omnivores (-74%) or respectively standard luminaires in detritivores (-27%).This suggests that different functional arthropod groups exhibit varied responses to different LED light characteristic combinations, resulting in reduced numbers at luminaires dimmed from 100% to 50% light intensity.
Furthermore, individual treatments did not yield significant effects on the predator-detritivore ratio, similar to a multi-year study assessing the effects of ALAN on trophic arthropod structure where they found no evidence of ALAN affecting the numbers of herbivores and predators (Firebaugh & Haynes, 2020).However, again when analyzing the interacting effects, we observed that combining higher light intensity with 3700 K luminaires resulted in a higher ratio.This indicates a significantly higher abundance of predators compared to detritivores around streetlights with our LED-light combination, in contrast to those dimmed down to 50%.This higher predator-detritivore ratio stands in line with our fourth hypothesis (H4) and can cause a decrease in detritivores either by predation or by avoidance and lead to a loss of ecosystem functioning, as shown by Hines and Gessner (2012), where they found a decrease of decomposition by detritivores of over 50% in the presence of predatory spiders.
We used three forest sites differing in tree composition and climatic factors and focused on different ground-dwelling functional groups to test for the generality of effects related to ALAN.However, our study was limited by the number of sites and taxonomic resolution.Future research should explore the generalizability of our findings and investigate the broader impacts of ALAN on terrestrial invertebrates in different ecosystems.In particular, this should include urban and suburban areas where ALAN is prevalent (e.g.cities, newly developed areas, forest boarders, e.g.Lockett et al., 2021).This could be used to assess whether and how invertebrate communities have already changed in response to ALAN (e.g.Kaunath & Eccard, 2022).To further enhance the depth of understanding, future studies should use a trait-based approach using finer taxonomic resolution (Hölker et al., 2021).Additionally, the cascading effects of altered arthropod community composition on ecosystem functioning and the long-term effects of ALAN on arthropod communities and ecosystem dynamics should be investigated in more detail with attention at a high taxonomic and spatial resolution.

Conclusions
Our study highlights the complex relationship between the different characteristics of LED streetlights on functional ground-dwelling arthropod groups in forests that have not been strongly exposed to ALAN.While we observed a strong attraction of predators and omnivores to ALAN in general, detritivores exhibited minimal response.Increased abundance of predators and omnivores in illuminated areas could lead to increased top-down effects on lower trophic levels, potentially altering the abundance and distribution of prey.This disturbance of trophic interactions and food webs may impact processes such as nutrient cycling and decomposition rates and lead to a loss or shift in ecosystem functioning.Individual light properties such as color temperature, light intensity, and luminaire shapes showed no significant effects on predators and detritivores.The exception were omnivores with reduced numbers (− 57%) in traps exposed to luminaires dimmed  down to 50%.Instead, interactions between light characteristics emerged as influential factors, with specific luminaire shapes and color temperature interacting with light intensity showed significantly reduced numbers of invertebrates at lower intensities.Our findings thus underscore the importance of considering the combined effects of different LED light characteristics on arthropod responses.To ultimately achieve a more comprehensive understanding of the broader ecological consequences of LED light technology on terrestrial invertebrates and ecosystem dynamics, future research should be extended to a gradient from urban to different, more natural ecosystems.The use of trait-based approaches with a finer taxonomic resolution could lead to a better mechanistic understanding of the consequences of light pollution.This is essential for informing effective mitigation strategies and preserving biodiversity in the face of increasing urbanization and artificial lighting.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.A) LED characteristics tested in the field experiments: (a) three LED color temperatures, (b) two light levels (no dimming (100%), dimmed to 50%), and (c) two luminaire shapes (standard, diffused); B) Distribution of the pitfall traps-the center trap (C; 100%, full light intensity) and the peripheral trap (P; 10% of the full light intensity).

Fig. 2 .
Fig. 2. The total number of captured individuals of each invertebrate taxonomic group classified into three functional groups (predators, omnivores, detritivores) on a logarithmic scale (total number of caught individuals = white numbers inside the bar).

Fig. 3 .
Fig. 3.Estimated marginal means of the abundance of functional arthropod groups (predators, omnivores, and detritivores) for two light treatments (light, dark) (A, B, C), and two pitfall trap positions (central C, peripheral P) (D, E, F).The error bars show the standard error (SE).Significant interactions (p < 0.05) are indicated by an asterisk (*).

Fig. 4 .
Fig. 4. Estimated marginal means of the abundance of functional arthropod groups (predators, omnivores, and detritivores) for three LED color temperatures (A, B, C), two luminaire shapes (D, E, F), and two light levels (G, H, J).The error bars show the standard error (SE).(For the graphic showing the raw means including the control, see Appendix H; Fig. 1).

Fig. 6 .
Fig. 6.Estimated marginal means for the response of the predator-detritivore ratio (at pitfall position C) to (A) the three LED color temperatures, (B) two light levels, and (C) two luminaire shapes.(D) The interaction effects between the light levels and temperatures averaged across both luminaire shapes; (E) the effects between the light levels and luminaire shapes averaged across the three LED color temperatures; (F) effects between the luminaire shapes and the LED color temperatures averaged across the two light levels.The error bars show the standard errors (SE).Significant interactions (p < 0.05) are indicated by an asterisk (*).