Mixing the Message: Do Dung Beetles (Coleoptera: Scarabaeidae) Affect Dung-Generated Greenhouse Gas Emissions?

By mixing and potentially aerating dung, dung beetles may affect the microbes producing the greenhouse gases (GHGs): carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O). Here, their sum-total global warming effect is described as the carbon dioxide equivalent (CO2e). Our literature analysis of reported GHG emissions and statistics suggests that most dung beetles do not, however, reduce CO2e even if they do affect individual GHGs. Here, we compare the GHG signature of homogenized (“premixed”) and unhomogenized (“unmixed”) dung with and without dung beetles to test whether mixing and burial influence GHGs. Mixing by hand or by dung beetles did not reduce any GHG – in fact, tunneling dung beetles increased N2O medians by ≥1.8x compared with dung-only. This suggests that either: 1) dung beetles do not meaningfully mitigate GHGs as a whole; 2) dung beetle burial activity affects GHGs more than mixing alone; or 3) greater dung beetle abundance and activity is required to produce an effect.


. A summary of the exact GHG fluxes between dung beetles (DB) and dung-only (DO) from data and analyses of past authors.
For Evan et al. 2019's, we assumed DB ~ 0.044 (rounded down to 0) and DO ~ 0.151 (rounded up to 2) for a conservative CH 4 analysis for Table 4.
Less than half of current studies reported total CO 2 e (Penttilä et  cattle, with and without antibiotics. They also measured dung beetles' impact on GHG 116 emissions. They saw dung beetles decreased CH 4 , increased N 2 O, and had no effect on CO 2 117 emissions relative to the non-antibiotic dung-only (Table 1). Interested in the overall warming 118 effects, we also analyzed the supplemental data supplied online (Hammer et  focus only on the more extensive unmixed design, which also forms a part of the combined 167 design (see Table 2's footnotes). pats within cattle pastures using mobile GHG chambers alongside netted cages (Fig 1), which: a) 197 avoided chamber-induced microclimates, b) reduced destructive chamber burial on pastures, and 198 c) prevented dung arthropod entry. Both sites used the same sampling equipment (Fig 2). Fowler repertoire. Following an additive design, we measured each dung-use group together and apart, 218 while also including mixed dung-only, unmixed dung-only, and grass/pasture-only ('vegetative') 219 controls (Table 2). Labor and material constraints restricted additional replicates solely for the 220 pasture-only control (n=4), but this unbalanced design was accounted for statistically by using  224 We measured 3 grams of dung beetles 3 per treatment (Table 2)  respiration studies pre-0d to examine if dung beetles produced ample greenhouse gases, but dung 229 beetles were found only to respire elevated CO 2 and so will only be briefly discussed (see Supp. 230 Section B for more detailed methods and analyses).  between sites (Supp. Fig G6). We will examine two types of statistics here to present a more  Fig G3). Similarly, unmixed dung-282 only produced 2.83x and 3.71x more CH 4 (Supp. 2) different chamber and dung volumes between sites: this could theoretically bias certain 302 treatments toward greater fluxes, but given that unmixed dung-only sizes were random (often 303 larger) and measured using field-based chamber methods (larger volume) -the total dung-to-304 chamber volume ratios between sites were similar (see Supp. Section A for a more in-depth 305 discussion of calculations and analyses). After all, any biases would show up on 0d if the 306 ratios were different, but this did not occur (Supp. Fig G6); and 3) between-site variances: mixed (homogenized) dung in the semi-permanent (buried) chambers 309 could have reduced variation compared with unmixed dung (non-homogenized) of the non-310 standardized (unburied) pastured treatments, thus resulting in reduced power to detect 311 differences. Yet no variation differences were found between the unmixed and mixed dung-312 only (p>0.05) across a week (Supp. Fig G8).
If mixing itself was a factor in affecting GHGs we would expect to see differences between the 315 mixed and unmixed dung-only treatments on 0d. However, there were no differences and the 316 mixed dung-only treatment produced slightly more CH 4 despite being presumably more aerated. 317 Likely it is because fresh cattle dung is liquid-like, and so mixed dung easily reforms and inhibits 318 aeration. However, dung beetles physically affected the dung (Fig 2)  Treatment had no effect on CO 2 fluxes in either the combined (Supp. Fig G1) or unmixed design 325 (Fig 3). Comparatively, time had a strong reductive effect (E=0.70-0.80) compared with 326 treatments (E=0.07-0.10), as CO 2 fluxes declined by >1.9x from 0d to 7d (Supp. Fig G2) or 14d 327 (Fig 4). Oddly, we expected dung beetles to increase CO 2 fluxes given their own respiration and 328 aerating/aerobic-based activities, but our (Supp. Fig B) and Piccini Fig G3), the unmixed dung-only produced less CO 2 fluxes than the mixed 332 dung groups on 0d (Fig 5) and showed slower week-long declines (Fig 6). This was, 333 respectively, due to the pasture-control influencing the unmixed dung-only's reported fluxes, for  2) by producing 1.07x more CO 2 (t=2.41, p=0.056) over time (Fig 6) (Fig 4).
Nitrous Oxide. Unlike the other gases, treatment showed a small-to-medium effect on median 376 N 2 O fluxes for both the unmixed design (E=0.21, Fig 3) and when combining years (E=0.30, 377 Supp. Fig G1), while time had a small-to-strong effect (E=0.22, Supp. Fig G2; E=0.54, Fig 4). 378 As with all main effect analyses, data aggregated across time or treatment obscures differences 379 between high-performance treatments and strong time effects, therefore day-by-day analysis was 380 required for differentiation. For the treatments: the unmixed design showed dung beetle groups 381 producing an average of 2.6x more N 2 O than the mixed dung-only on 3d alone (Fig 5, E=0.62 Fig G8), the dung beetle 386 groups produced a greater frequency of larger fluxes (p=0) and a smaller frequency of minimums 387 (p=0) (Supp. Fig G7) reflecting the significant omnibus and effect size analyses (Fig 3). This 388 suggests tunneler-activity, but not mixing nor dwelling-activity, specifically generated more N 2 O 389 (1.57x) than unmixed dung-only despite its weak effects (Fig 3). Curiously, our vegetative  Table D2) (Fig 6), often generating 0.10-0.50x the amount of the dung-397 containing treatments on any given day. Across time, the dung-containing groups produced 2.59-7.26x more N 2 O than their respective vegetative controls, thus showing dung's propensity for 399 N 2 O generation (Fig 6) until complete decay (see the desiccated dung in Fig 2). Dung (Supp. Fig   400   G9) and soil (Supp. Fig G10)  CO 2 , and CO 2 e were positively correlated with dung moisture loss (Supp. Fig G9, p≤0.01), but 404 not soil moisture loss (Supp. Fig G10, p≥0.05 GHGs collected was solely CO 2 , but since CH 4 and N 2 O enjoy a larger greenhouse effect, they 427 respectively contributed to 23.07 and 7.28% of the total effect (Supp. Fig G5). Even so, CO 2 428 commands 69.65% of the sum-total CO 2 e which is why CO 2 e graphs (Figs 3-6, Supp. Figs G1-429 G4) predominately follow CO 2 trends. Treatment had no effect (~1x) on CO 2 e (E<0.12, Fig 3), 430 and the small reduction of CO 2 e on 0d was attributed to the vegetative differences as described in 431 the Carbon dioxide section. Comparatively, time steadily reduced CO 2 e by 3.24x (Fig 6) over the 432 course of a week (E=0.85, Fig 4)

Discussion
These arthropods affected the decomposition rates and microbial pathways driving carbon (C) 445 and nitrogen (N) storage/release throughout the environment. For example, by reducing C and N 446 lost to the atmosphere, it is instead used and stored terrestrially (Sylvia et al. 2005). 447 Theoretically, dung beetles are capable of similarly affecting GHGs, but lacked supportive 448 research (see Introduction). By improving the power of our study (combined-years design) and 449 testing potential methodological problems (unmixed design), we suggest that dung beetles are 450 ultimately carbon neutral and that the physical 'mixing' of dung may not be a significant 451 mechanism in reducing GHGs. 452 453 GHG trends and their potential causes 454 As the pat ages, the constant decay physically alters the dung by leaching/evaporating water (by 455 1.89x from 0-14d, Supp. Table D6) and loosening the dung structure -a process aided by the 456 disturbance and disassembly of dung pats by dung beetles (Fig 2). The disintegrating and 457 desiccating dung allows for deeper oxygen (O 2 ) penetration and permeation such as seen in soil 458 (Sylvia et al. 2005). If true, we would predict decreased CH 4 , increased N 2 O, and increased CO 2 459 over time. However, we saw both dung-based CH 4 and CO 2 (and so CO 2 e) decline permanently 460 over time until they mirrored the vegetative-control fluxes (Fig 6). Though expected for CH 4 , 461 CO 2 's decline was a surprise. Presumably transitioning from an anaerobic to an aerobic dung pat 462 by mixing or aging should predictably increase CO 2 emissions via environmental respiration or 463 enhanced gas transport, though not dung beetle respiration (Iwasa et al. 2015). After all, we 464 (Supp. Section B) and Piccini et al. (2017) showed that dung beetle respiration was less <1.5% of 465 the total CO 2 , and that dung beetles did not release more CH 4 and N 2 O than the control (Supp. 466 Fig B), despite consuming methanogen and denitrifier-rich dung (Yokoyama et al. 1991a).
Ultimately, no CO 2 differences existed between any treatments (Fig 3) even after two weeks of 468 decomposition and desiccation (Fig 2). Meanwhile, N 2 O followed our predictions but offered a 469 surprise. Collectively this suggests that cow dung is an obvious moisture and fertilizer source, and that 488 dung beetles may increase the incomplete denitrification rate. 489 The increased N 2 O fluxes from 1-7d and the sharp 14d decline in dung-based treatments suggest 491 that our treatments supported large enough N-pools and a mostly anaerobic state sufficient for 492 N 2 O production until 14d. However, we wondered if dung or soil was the main source of 493 denitrifier activity. Consider that: higher soil moisture, but not dung moisture (Supp. Fig G9), 494 predicted higher N 2 O emissions (Supp. Fig G10) (Table 3).  (Fig 2). However, by comparing dung beetle treatments to time, we can more easily deduce if 562 dung beetle activity, despite all other pressures, is a more powerful GHG predictor -it wasn't. 563 We also hypothesized that premixing dung, an activity reflecting dung beetle activity, might 564 obscure the dung beetle effect -it didn't. suggestion that dung tunnels may increase CH 4 's release from the pat applies only to dung wet 570 enough for anaerobic maintenance, but chunky/dry enough to support sturdy microtunnels. When 571 mixing fresh dung, the dung reforms and reconnects when wet, so likely there was no aeration in 572 the mixed dung-only without O 2 for confirmation. At the outset we hypothesized that mixing 573 multiple dung pats and relocating them alters GHGs, but this was not borne out -however, it 574 does question whether mixing itself (Table 4) is an influential factor, especially compared to 575 time-based decay (Table 3).