A novel method for monitoring ground-dwelling arthropods on hard substrates: characterizing arthropod biodiversity among survey methods

Sampling approaches are commonly adapted to reflect the study objectives in biodiversity monitoring projects. This approach optimizes findings to be locally relevant but comes at the cost of generalizability of findings. Here, we detail a comparison study directly examining how researcher choice of arthropod trap and level of specimen identification affects observations made in small-scale arthropod biodiversity studies. Sampling efficiency of four traps: pitfall traps, yellow ramp traps, yellow sticky cards, and a novel jar ramp trap were compared with respect to an array of biodiversity metrics associated with the arthropods they captured at three levels of identification. We also outline how to construct, deploy, and collect jar ramp traps. Trapping efficiency and functional groups of arthropods (flying, crawling, and intermediate mobility) varied by trap type. Pitfalls and jar ramp traps performed similarly for most biodiversity metrics measured, suggesting that jar ramp traps provide a more comparable measurement of ground-dwelling arthropod communities to pitfall sampling than the yellow ramp traps. The jar ramp trap is a simple, inexpensive alternative when the physical aspects of an environment do not allow the use of pitfalls. This study illustrates the implications for biodiversity sampling of arthropods in environments with physical constraints on trapping, and the importance of directly comparing adapted methods to established sampling protocol. Future biodiversity monitoring schemes should conduct comparison experiments to provide important information on performance and potential limitations of sampling methodology.


Introduction 24
There are many ways to observe populations and communities of insects.

2
Arthropod sampling methodology may be particularly prone to introducing contextual biases to data, which 40 makes biodiversity monitoring difficult to approach in a comprehensive, standardized way (Montgomery et al., 41 2021). Each collection method has variable trapping efficiency that depends on arthropod biology and behavior as 42 well as trap design (Montgomery et al., 2021). These biases do not eliminate the utility of the collected data, but 43 additional information about the goals, constraints, and methods of a given experiment or monitoring strategy must 44 be used to contextualize and understand the limitations and further use of these data. This contextual information 45 also aids effective synthesis of data across biodiversity studies (Elphick, 2008). Within insect ecology, there is a 46 strong cultural precedent of 'do-it-yourself' approaches for developing novel trapping methods, customized to a

64
Use of common approaches may aid synthesis of monitoring data for arthropod populations, but may be 65 constrained by the environments that these techniques are deployed in. For instance, pitfall traps are a commonly 66 used method to sample ground-dwelling arthropods (Greenslade, 1964; Hohbein & Conway, 2018) and consist of a 67 container filled with a killing fluid dug into the soil so that the rim is flush with the ground's surface (Figure 1a

82
The objective of this study was to investigate how the design of arthropod traps affect the observations of 83 arthropod communities, particularly for trap designs that had been adapted to contend with physical constraints of 84 their deployment environment. Specifically, we compared the performance of two traps designed to minimize 85 3 disruption to soil substrates to two classical trapping methods. We compared arthropod communities among 86 traditional pitfall traps, commercially available ramp traps, sticky cards, and a novel, alternative design to the 87 commercial ramp trap, the jar ramp trap. Herein, we outline how to construct and deploy jar ramp traps. We 88 predicted that the arthropod community captured by each trap would vary based on the structure of the trap and the 89 functional biology of the arthropods. In addition to comparing these four sampling methods, we compared how 90 multiple approaches to insect identification may impact the findings. We predicted that different identification levels 91 will produce variable statistical results, each revealing and obscuring different parts of the community, and 92 suggesting tradeoffs between both trapping and sample processing approaches. We discuss recommendations for 93 comparison studies which will improve the interoperability of data produced by specialized insect sampling 94 methodology.

Materials and Methods 96
Study sites

97
We selected study sites with similar abiotic attributes to the thin-soil environments our adaptive traps were 98 designed for: exposed to solar radiation, precipitation, and wind, but with deeper soils to accommodate the use of

132
Trap contents were collected after 48 hours. Samples from the yellow ramp traps were strained with 133 noseeum mesh in the field and preserved in 70% ethanol in gallon plastic zipper-top bags. Yellow sticky cards were 134 placed directly into gallon plastic zipper-top bags. Jar ramp traps and pitfall traps had plastic lids secured on the 135 glass jar or plastic container, respectively, directly in the field. Samples were processed in the laboratory and 136 specimens were identified with the aid of a dissecting microscope. For the duration of the study, yellow sticky cards 137 were stored in the freezer and all other samples in vials with 70% ethanol.

138
In contrast to the commercial yellow ramp trap, the ramps on the jar ramp traps were at a lower angle and 139 the noseeum mesh provided a substrate that was easier to grip than the smooth plastic. Additionally, the design of

Results 206
Seven sampling periods at our three sites yielded 165 samples (accounting for three pitfalls lost to 207 disturbance by mammal excavation), which contained a total of 13,634 arthropod specimens. Overall, yellow ramp 208 traps caught the greatest number of individuals (7,758); followed by sticky cards (4,199); then jar ramp traps

209
(1,099); with pitfall traps catching the least (578) (see abundances by identification level: Table S1). Trap types had The total number of orders captured varied from 9 (pitfall traps) to 12 (yellow ramp traps and jar ramp 215 traps) (Figure 3a). When compared with first order jackknife richness estimates, pitfall trap efficiency was 90%; 216 yellow ramp trap efficiency was 100%; yellow sticky card efficiency was 100%; and jar ramp trap efficiency was 217 86%.

218
Overall differences in richness, abundance, Shannon diversity, and evenness were observed between the 219 trap types in generalized linear mixed effect models (Table S2; Figure 4a). Higher arthropod richness, abundance, 220 and diversity were observed in yellow ramp traps than other trap types. For richness and abundance, yellow ramp    (Table S1). When compared with first order jackknife richness estimates, 233 pitfall trap efficiency was 84%; yellow ramp trap efficiency was 84%; yellow sticky card efficiency was 84%; and 234 jar ramp trap efficiency was 79%.

235
Overall differences in richness, abundance, Shannon diversity, and evenness were observed between the 236 trap types in generalized linear mixed effect models (Table S2; Figure 4b). Similar to the order level analyses, 237 yellow ramp traps collected the highest richness, abundance, and diversity. Jar ramp traps and pitfall traps collected 238 the lowest richness. For abundance, yellow ramp traps were followed by yellow sticky cards, jar ramp traps, and

261
The total number of focal taxa captured varied from 3 (pitfalls and jar ramp traps) to 9 (yellow sticky cards) 262 (Figure 3c). When compared with first order jackknife richness estimates, pitfall trap efficiency was 78%; yellow 263 ramp trap efficiency was 69%; yellow sticky card efficiency was 77%; and jar ramp trap efficiency was 78%.

264
There were no differences in richness, abundance, Shannon diversity, or evenness detected between trap 265 types in generalized linear mixed effects models (Table S2;   arthropod communities. However, our study also suggests an important caveat: the ability to detect differences 276 between sampling types is also affected by sample size. Studies that focus at a high taxonomic resolution will 277 require many more individual samples to be able to detect differences. Our analyses of focal taxa were not able to 278 detect statistical trends due to the relatively small number of specimens from each group.

279
In our study, the jar ramp trap and pitfall trap communities were very similar, suggesting they had similar

335
Though the jar ramp trap has many advantages over the yellow ramp trap, and performed similarly to pitfall

338
Larger jars could easily be adapted to this design, however, at the compromise of ramp steepness. Although the 339 rocks provide a natural means of securing the noseeum mesh to the ground, the presence of rocks could affect 9 movement of some ground-dwelling arthropods. In spaces where some substrate exists, the traps may be lightly 341 covered by soil to secure the mesh.

342
Our study demonstrated that the level to which taxa are identified impacts the study results and researcher 343 interpretation of the biodiversity data. In this study, arthropods were identified using three approaches (i.e.

369
Although the focal beetle taxa required larger sample sizes to detect differences that may exist among trap 370 types, we still observed some members of the important predatory beetle families Carabidae, Staphylinidae, and 371 Coccinellidae as well as which trap types captured them (Table S1). This allowed us to dive deeper into two ground-

385
Seven species of Coccinellidae were captured during the study, with over 70% of the lady beetle specimens 386 10 collected by yellow sticky cards, which are considered reliable traps for measuring activity density in this family 387 (Bahlai et al., 2013). Yellow ramp traps captured some lady beetles as well, but neither the pitfall nor the jar ramp 388 trap captured this taxon reliably.

389
It is not uncommon for biodiversity monitoring to occur in sensitive habitats with unique constraints, 390 requiring customized approaches to monitoring. However, these modifications to standardized trapping methods 391 limit the comparability of study findings. This study illustrates the implications for biodiversity sampling of 392 arthropods in environments with physical constraints on trapping, and the importance of directly comparing adapted 393 methods to established sampling protocol. We have shown that conducting a comparison of those methods can 394 provide important contextual information on how that method performs, and its potential limitations in monitoring 395 protocol.

396
Comparison studies should ideally be conducted in the environment where monitoring will occur.

397
However, in our case the thin-soil environments that jar ramp traps and yellow ramp traps are meant for did not 398 allow for the use of pitfall traps. By conducting the comparison in an environment with similar abiotic attributes as 399 thin-soil sites, we were able to comprehensively examine the efficacy of these trap types to inform future arthropod 400 monitoring study designs in these sensitive habitats. This study leverages sites that were accessible and relatively 401 uniform in environmental conditions to demonstrate that such comparisons of methodology can be relatively small 402 scale and accomplished with limited labor. Indeed, the experimental work for this study was completed on a 403 university campus when travel and support labor was highly limited by the COVID-19 lockdown.

404
Despite its importance to environmental management, developing standards for biodiversity monitoring 405 comes with many challenges. Between idiosyncratic biology of target taxa and habitat effects on trapping efficiency,