Ground cover vegetation promotes ecological intensification of pear production --Manuscript Draft--Manuscript

: The use of ground cover vegetation is becoming a prominent intervention for promoting biodiversity and associated ecosystem services in Chinese orchards. Despite the large number of studies that have examined the effects of ground cover vegetation on promoting natural enemy populations and related natural pest control, it is less understood whether enhanced natural pest control translates to increase yield and reduced pesticide use. We conducted a 2-year experiment comparing three cover vegetation (ryegrass, clover, and hairy vetch) practices versus a bare ground control in commercial pear orchards in the Yangtze River Delta of East China (YRDEC), China. Natural enemy density (predator and parasitoid abundance), invertebrate herbivore performance (piercing-sucking herbivore abundance and branch-boring and fruit-boring percentage), pesticide input and pear fruit yield were recorded. The results indicated that cover vegetation decreased herbivore abundance and boring percentage by 49.95 and 63.6% respectively, and thus decreased pesticide use by 26.10%. Meanwhile, we found that cover vegetation increased the abundance of natural Abstract The use of ground cover vegetation is becoming a prominent intervention for promoting biodiversity and associated 28 ecosystem services in Chinese orchards. Despite the large number of studies that have examined the effects of 29 ground cover vegetation on promoting natural enemy populations and related natural pest control, it is less 30 understood whether enhanced natural pest control translates to increase yield and reduced pesticide use. We 31 conducted a 2-year experiment comparing three cover vegetation (ryegrass, clover, and hairy vetch) practices versus 32 a bare ground control in commercial pear orchards in the Yangtze River Delta of East China (YRDEC), China. 33 Natural enemy density (predator and parasitoid abundance), invertebrate herbivore performance (piercing-sucking 34 herbivore abundance and branch-boring and fruit-boring percentage), pesticide input and pear fruit yield were 35 recorded. The results indicated that cover vegetation decreased herbivore abundance and boring percentage by 49.95 36 and 63.6% respectively, and thus decreased pesticide use by 26.10%. Meanwhile, we found that cover vegetation 37 increased the abundance of natural enemies by 620.75%, and increased pear fruit yield by 6.82%. Piecewise 38 structural equation modelling indicated that increased natural enemy densities, decreased herbivore performance and 39 pesticide use, while increasing fruit yield. Our results confirm that the use of ground cover vegetations (especially 40 with clover and hairy vetch) can promote ecological intensification and biological pest control in pear orchards. 41

Response: We have added several sentences and tried to make it clearer as far as possible (lines 92-96).
Comment: The information on chemical use and what happened during this experiment is unclear to me. For example, Ll 126-130, how was the application of chemicals decided on? How was this sampling undertaken? Response: Insecticides/miticides were applied according to the observed herbivore densities or herbivory damage in each treatment and control plot, and based on the Economic Injury Level of pest density according to the decisions about spray thresholds from the Fengxian Agricultural Technology Extension and Service Center (FATESC) of Shanghai, China (Wan et al. 2019a(Wan et al. , 2019b(Wan et al. , 2020a, as these commercial pear orchards were under the surveillance of FATESC. In details, how and when farmers were allowed to use pesticides in each plot, was abided by the guidance of FATESC (lines 131-136).
Comment: L122 is this in addition to the pesticide /miticide referred to L126? Response: Herbicides and fungicides were not applied in any blocks during the growth and development stage of pears (lines 130-131).
Comment: It is unclear to me how pests and natural enemies were sampled. Was herbivore data inferred from damage? How about natural enemy data? Analysis would be more useful if data for each group was presented (hoverflies, spiders, ladybirds, lacewings parasitoids, aphids, spidermites, pear midges, boring pests). Response: Thanks for pertinent suggestions. We have added several sentences in the "Sampling methods and fruit yield measurement" to make this clearer as far as possible (lines 144-174).
Comment: L125 'fungicides were not applied in any blocks' Response: Done as suggested (lines 130).
Comment: L135 'spraying with lime sulphur' Is this not inconsistent? Lime sulphur can also act as a miticide Response: Here lime sulphur was used in each block of treatment and control during the dormant period and were considered as a pesticide to kill overwintering pests including insect herbivores, pathogens and mites. However, lime sulphur was not used in the growth and development stage of pears.
Additional Information:

Question Response
Please summarize the main achievement of this work, above and beyond what may have been conveyed in the manuscript title in bullet point style. In brief, the Key Message should state why the work was conducted (knowledge gap(s) as well as key question(s) and/or hypotheses tested) and highlight the main finding(s) and the conclusions derived from this study. The latter should address the wider implications of the work and the relevance for pest control. All text should be generic, seminal and understandable to nonspecialists. This Key Message should also • It was not clear how cover vegetation affects ecological intensification of pears.
• We used Piecewise structural equation modeling to test our hypothesis in the YRD, China.
• Ryegrass, clover and hairy vetch planted in pear orchards increased natural enemy densities and biological control of pest insects.
• These practices suppressed herbivore and pesticide use, while increased fruit yield.
• This study increased our understanding about agrobiodiversity effects on ecological intensification.
be part of your submitted manuscript and will be published in front of the Abstract.
<b><u>Please include 3 to 5 bullet points of maximum 95 characters, including spaces, per bullet point.</b></u> Dear Editor: Please consider our paper, entitled "Ground cover vegetation promotes ecological intensification of pear production" for publication in Journal of Pest Science.
We think our paper represents an important advance in our understanding of the contribution of biodiversity through multi-trophic level production systems in comparing ground cover vegetation and mono-farming production in a two-year experiment.
We think that researchers working in pest management, insect ecology, biodiversity science, ecosystem services, agronomy, population or community ecology, will find our results of interest.

82
In the Yangtze River Delta of East China (YRDEC), the most important invertebrate herbivores in pear orchards 83 are piercing-sucking (aphids, spider mites and pear midges), and branch-and fruit-boring herbivores (oriental fruit 4 moth) (Ma 2019). To reduce the injury and loss of productivity due to herbivores, farmers spray chemical 85 insecticides that kill also beneficial insects (arthropod predators and parasitic wasps) in addition to pest species 86 (Zhang et al. 2014; Ran et al. 2016). In this framework, ground cover vegetation (i.e., ryegrass, white clover, hairy 87 vetch, etc.) has been widely adopted in pear orchards in the YRDEC since late 1990s as a potential way to promote 88 natural enemies of insect herbivores and improve soil fertility (Yi et al. 2010).

89
We hypothesized that the implementation of ground cover vegetation in pear orchards could not only increase the 90 abundance of natural enemies of invertebrate herbivores, decrease both herbivore abundance and herbivory damage, 91 and reduce pesticide use and increase fruit yield. To test this hypothesis, we selected three cover crops among the 92 most common used by farmers in this area. The cover crops selected were ryegrass (a perennial grass), chosen for 93 its ability to scavenge nitrogen and produce biomass; white clover (a perennial legume), chosen for its ability to fix 94 nitrogen and for the lower biomass production ideal in orchards; hairy vetch (an annual legume), chosen for its 95 ability to produce higher biomass and fixing more nitrogen, compared with white clover. We  The experiment was arranged as a randomized complete block design with three replicates (i.e., three blocks). Each 115 block consisted of three treatments (i.e., three plots), of ground cover vegetation of ryegrass (Lolium perenne), white 116 clover (Trifolium repens), or hairy vetch (Vicia villosa) in pear orchards, and one control (i.e., one plot) of pear 117 orchards without ground cover (bare soil). The four plots in each block were randomized (i.e., all three plots of 118 vegetation practices and one control without cover vegetation were randomized 5-to-6-column trees apart in each 119 block). Each treatment and control plot consisted of 6 columns × 30 rows (i.e., 20m × 58m) (Figs. 1 and 2). In each 120 plot, 180 trees (i.e., 6 columns × 30 rows) were planted in a pattern with 16-20m spacing (i.e., 5-6 columns equal to 121 16-20 meters) between neighbouring plots.

122
In 2016, before starting the experiment, the orchard was sown with perennial ryegrass in early November, while 123 perennial white clover, and the annual legume hairy vetch, were sown in early April. As hairy vetch is an annual

175
All analyses were carried out using R4.0.3 (R Core Team 2020). We assessed the independent and interactive 176 effects of ground cover and sampling date on natural enemy abundance, piercing-sucking herbivory abundance and 177 boring percentage using two-way ANOVA with block as a random factor. To improve normality and 178 homoscedasticity of residuals, piercing-sucking herbivory abundance was log-transformed and boring percentage 179 was logit transformed. We then assessed the independent and interactive effects of ground cover and year on amount 180 of commercial pesticides, amount of active ingredient in pesticide, and fruit yield using a two-way ANOVA with 181 block as a random factor. All models were run using the 'lmer' R package (Bates et

204
We found a significant interaction between ground cover and sampling date for all models (Table S1). The eight 205 sampling dates used to monitor the changes over time of the natural enemy abundance during the growing season, 206 showed clearly higher enemy abundance in the treatment with ground cover vegetation than in the control plots 207 (mean percent increase ± SD for ryegrass 582.3 ± 374.4, for clover 761.2 ± 261.8, and for hairy vetch 683.8 ± 208 351.2) (Fig.3a). However, while clover and hairy vetch treatments maintained a higher natural enemy abundance 209 across all the sampling dates, ryegrass treatment showed a decline starting from the fourth sampling date (Fig. 3a).

210
Density of piercing-sucking herbivores was lower in three treatment plots than in the control plots (mean percent 211 decrease ± SD for ryegrass 52.1 ± 7.1, for clover 36.6 ± 15.6, and for hairy vetch 53.8 ± 11.3) (Fig. 3b). Compared 212 to the two other vegetation cover treatments, the clover treatment showed generally a greater abundance of piercing-213 sucking herbivores (Fig. 3b and Fig. S2). Likewise, four sampling dates indicated that boring percentage was lower 214 in three treatment plots than in the control plots (mean percent decrease ± SD for ryegrass 57.9 ± 13.7, for clover 215 64.9 ± 11.0, and for hairy vetch 67.2 ± 8.7) (Fig. 3c). In this case, hairy vetch showed generally a lower boring 216 percentage ( Fig. 3c and Fig. S3).

217
Pesticide use and fruit yield

218
We found a significant effect of ground cover on pesticide use and fruit yield, but not on year or interaction between 219 ground cover and year (Table S2) 24.0% in orchards with ryegrass, clover and hairy vetch, respectively, during the whole experiment (Fig. 3d, e). The 222 two-year data collected on fruit yield, showed a significant increase in the treatment plots compared with the control 223 plots (i.e., vegetations of ryegrass, clover and hairy vetch increased pear yield by 5.7, 6.7 and 8.1 t•ha -1 , 224 respectively) ( Fig. 3f and Fig. S6).

225
Piecewise SEM analysis 226 The piecewise SEM was well supported by the data (Fisher's C = 5.45, df = 2, p = 0.066 for model including 227 abundance of piercing-sucking herbivores; Fisher's C = 4.53, df = 2, p = 0.103 for model including percentage of 228 branch-boring and fruit-boring) and none of the independence claims implied by the model were statistically 229 significant (p > 0.05) suggesting that all the important relationships were specified in the model (Fig. 4). We found 230 that natural enemy abundance directly (Fig. 4a) or indirectly (Fig. 4b) affected pear yield. In the first model, natural 231 enemy abundance showed a strong negative effect on abundance of piercing-sucking herbivores, a negative effect 232 on chemical pesticide use, and a strong positive effect on pear yield (Fig. 4a). In the second model, we found that 233 the effect of natural enemy abundance on pear yield was mediated by branch-boring/fruit-boring percentage (Fig.   234 4b). Natural enemy abundance showed a strong negative effect on boring percentage, while boring percentage had a 235 negative effect on pear yield (Fig. 4b).

238
Our study confirmed that the implementation of ground cover vegetation is an effective intervention for promoting 239 ecological intensification in pear orchards. We found that ground cover vegetation not only increased the abundance 240 of natural enemies and decreased herbivore abundance or damage, but also reduced pesticide use and increased fruit

251
Although the three cover vegetation treatments showed a similar positive contribution to support natural enemy 252 populations, we observed some differences among the treatments. For example, natural enemy abundance was, on 253 average, 21.8 and 14.2% higher in pear plots with white clover than in those with ryegrass and hairy vetch, 254 respectively. This positive effect was also maintained across the sampling season. This would suggest a greater 255 ability of white clover to promote natural enemy populations, which might be due to the contribution of more 256 invertebrate herbivores as "prey pool". Indeed, we observed many different herbivores in the clover plots (e.g., 257 noctuids, aphids, whiteflies) that might represent complementary food resources for natural enemies in orchards.

258
Furthermore, both white clover and hairy vetch are nectar plants, which can provide nectars and honeydews to 259 enemies, while ryegrass has no such function. Interestingly, when the temperature increased in July and August, 260 ryegrass started to wilt and die. Ryegrass growing cycle usually reaches its peak in biomass production in May, and 261 starts declining its growth after June, while white clove reaches the highest biomass production from May to 262 September, and start declining afterwards (Hodgson 1990). This might explain why ryegrass showed the lowest 10 natural enemy abundance among the three ground cover vegetation treatments. Despite these differences, all ground 264 cover vegetation treatments showed generally a similar performance in controlling herbivory abundance and 265 damage (Table S1)  villosa] showed with blue "+") and control practice (i.e., pear orchards without cover vegetation). Pear trees were 463 spaced in 4 m (East to West direction; column) × 2 m (North to South direction; row) grids. The three replicated 464 blocks were more than 50 meters apart. In each block, the orchard was arranged with four randomized plots (i.e., all 465 three plots of vegetation practices and one control without cover vegetation were randomized 5-to-6-column trees 466 apart in each block). Each treatment or control plot was 6 columns × 30 rows.