The push–pull intercrop Desmodium does not repel, but intercepts and kills pests

Over two decades ago, an intercropping strategy was developed that received critical acclaim for synergizing food security with ecosystem resilience in smallholder farming. The push–pull strategy reportedly suppresses lepidopteran pests in maize through a combination of a repellent intercrop (push), commonly Desmodium spp., and an attractive, border crop (pull). Key in the system is the intercrop’s constitutive release of volatile terpenoids that repel herbivores. However, the earlier described volatile terpenoids were not detectable in the headspace of Desmodium, and only minimally upon herbivory. This was independent of soil type, microbiome composition, and whether collections were made in the laboratory or in the field. Furthermore, in oviposition choice tests in a wind tunnel, maize with or without an odor background of Desmodium was equally attractive for the invasive pest Spodoptera frugiperda. In search of an alternative mechanism, we found that neonate larvae strongly preferred Desmodium over maize. However, their development stagnated and no larva survived. In addition, older larvae were frequently seen impaled and immobilized by the dense network of silica-fortified, non-glandular trichomes. Thus, our data suggest that Desmodium may act through intercepting and decimating dispersing larval offspring rather than adult deterrence. As a hallmark of sustainable pest control, maize–Desmodium push–pull intercropping has inspired countless efforts to emulate stimulo-deterrent diversion in other cropping systems. However, detailed knowledge of the actual mechanisms is required to rationally improve the strategy, and translate the concept to other cropping systems.

Data, Figure 4, 5 and 6). None of the previously reported terpenes 12 were constitutively released, 85 nor any terpene or other volatiles that are typically released upon herbivory. Similar results were 86 obtained with D. uncinatum (Extended Data, Figure 7). In contrast, we did confirm that Melinis 87 minutiflora, a Poaceae used previously as a push intercrop, constitutively releases a diverse blend 88 of terpenes in large quantities (Extended Data, Figure 2, 3 and 8). Clearly, independent of soil 89 interactions, Desmodium does not constitutively release volatiles. 90 Although the constitutive release of volatiles is an important precondition for push-pull,  Figure 8), and possibly induced by herbivory that was visible on most sampled plants. 104 Thus, regardless of whether constitutive or induced, Desmodium does not release terpene 105 volatiles, or any other volatiles, in large quantities in the field. Although it cannot be excluded 106 that other conditions or herbivores may induce higher release of reported volatiles, our data with 107 numerous samples under different growth conditions, and from different geographic regions 108 show that this must be very rare, and can therefore not be at the core of a generic strategy. In populated with uncinate trichomes. First instar larvae were somewhat freely moving and grazing 143 between trichomes (Extended Data, Figure 10b,c), but older larvae were seen impaled and 144 immobilized by these trichomes (Figure 4c Figure 10g). Whereas trichomes were flexible at the base, they were fortified with silica toward 147 the tip (Figure 4f), equipping the plant with an effective mechanism to obstruct, damage and 148 immobilize herbivores. Also beneficial insects (Extended Data Figure 10i) and even vertebrates 149 can be trapped by Desmodium 33 . Similar structures are also used by many other plant species 34-36 , 150 and may serve multiple purposes including seed dispersal 37,38 . 151 We thus infer that in the field Desmodium affect fitness of lepidopteran larvae, both directly and 152 indirectly. First, Desmodium entices larval feeding, but truncates larval development. Second, 153 trichomes on Desmodium hinder movement, damage the cuticle and even entirely immobilize 154 larvae on the plant, increasing developmental time, exposure to natural enemies and overall 155 mortality 39,40 . Third, the ingestion of trichomes will damage the intestinal lining and affect 156 digestion, development and survival 40,41 . Indeed, while first instar larvae easily fed around the 157 trichomes, larger larvae did ingest trichomes as evidenced by trichomes found in larval frass. 158 Effectively, rather than functioning as a repellent intercrop, Desmodium appears to be a 159 developmental deathtrap for larvae. 160 Clearly 'push' does not describe the mode-of-action of Desmodium. Instead, the plant exhibits 161 properties reminiscent of a 'pull' crop, a 'dead-end host'. Although superficially similar in mode 162 of action to the 'pull' border crop Napier grass, Desmodium is distinctly different, as it is 163 preferred by larvae, not by adults 8,10 . In addition, Desmodium forms a mechanical barrier to 164 dispersing larvae. Further field studies need to detail how oviposition preference, larval dispersal, 165 development and survival on Desmodium, mechanical obstruction by Desmodium, and additional 166 mechanisms such as parasitization and predation, interplays with crop phenology in suppressing 167 various lepidopteran species across the cropping season. Knowing the exact interaction of 168 mechanisms is critical if we for instance wish to substitute the fodder crop Desmodium with a 169 food crop to enhance food security, or if we are to translate the concept of interceptive 170 intercropping to other cropping systems.

171
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

301
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Switzerland, and were raised on a soybean based semi artificial diet supplemented maize whorls. 408 The third instar larvae were separated into groups of ten individuals in plastic boxes. 409 Pupae were sexed and separated in rearing cages. Adults were provided with a 5 % sucrose 410 solution and 6 days old adults were mated for 6 hrs and used in oviposition experiments. week after the start of the experiment. The volatile headspace was closed for 24 hrs before each 431 sampling and the SPME sampling procedure was the same as described above. sampling procedure was the same as described above.
program was as follows: 40°C/2 min, 8°C/min to 230°C. The solvent delay and mass 450 spectrometry settings were the same as described above. 451 GC-MS results were analyzed using Agilent Mass Hunter B.08.00, the peaks were auto 452 integrated with agile integrator and manual integration. Compounds were tentatively identified 453 by matching their mass spectra with those found in MS Libraries (NIST11 and Wiley12). The  Table S1).

457
Oviposition choice experiments 458 We conducted two experiments to study the short-range/multimodal oviposition repellency and 459 long-range/olfactory oviposition repellency of D. intortum for S. frugiperda females. Larval choice experiments 484 We conducted two-choice feeding bioassays to determine the feeding preference of the first 485 larval instar of S. frugiperda. We cut 8 mm diameter leaf discs from young leaves of 6-7 weeks 486 old maize plants and leaves of 10-12 weeks old D. intortum plants. We put the leaf discs on wet 487 filter paper discs 60 mm apart from each other in 100 mm x 20 mm plastic Petri-dishes. Ten one-488 day old S. frugiperda larvae were placed in each arena and the position of larvae was recorded 489 after 1 h, 2 h and 20 h periods. After 20 h feeding each leaf disk was photographed and the 490 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  In case of each volatile sample the absolute peak areas were divided by the area of the internal 531 standard peak to account for differences in volatile sampling efficiency. The volatile 532 components were categorized into four compound groups: monoterpenoids, sesquiterpenoids, 533 green leaf volatiles and other volatiles. We calculated the total sum of peak areas for these 534 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 24, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 24, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

724
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 24, 2022. ; https://doi.org/10.1101/2022.03.08.482778 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 24, 2022. ; https://doi.org/10.1101/2022.03.08.482778 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The three dimensional model of wind tunnel assays (Extended Data Fig.1) was prepared in 772 SketchUp (version 20.0).

774
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 24, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022