Ecology

Study sites and examined tree speciesThe study was conducted at three sites in the Andes of southern Ecuador along an elevation gradient at ca. 1000 m (Bombuscaro, Podocarpus NP), ca. 2000 m (San Francisco Reserve) and ca. 3000 m elevation (Cajanuma, Podocarpus NP) in the Provinces of Loja and Zamora-Chinchipe. All sites are located in protected forest areas. At each elevation three permanent 1-ha plots were established in 2018, choosing representative portions of old-growth forest without visible signs of human disturbance (Appendix A1).The forest types at the three sites differ in floristic composition, species richness and structural characteristics49: The premontane rain forest (below 1300 m) at the lowermost site reaches 40 m in height with common tree families being Fabaceae, Moraceae, Myristicaceae, Rubiaceae, and Sapotaceae. It is replaced at 1300–2100 m by smaller-statured lower montane rain forest with Euphorbiaceae, Lauraceae, Melastomataceae, and Rubiaceae as characteristic tree families, and above 2100 m by upper montane rain forest with a canopy height that rarely exceeds 8–10 m. Dominant tree families of the latter forest type are Aquifoliaceae, Clusiaceae, Cunoniaceae, and Melastomataceae. Tree species turnover is complete between premontane and upper montane forest, while a few tree species are shared between lower montane and premontane or upper montane forest types.The climate is tropical humid with a precipitation peak from June to August and a less humid period from September to December. Mean annual temperature decreases with elevation from 20 °C at 1000 m to 9.5 °C at 3000 m, while annual precipitation increases from around 2000 mm at the two lowermost sites to 4500 mm at 3000 m. Typically, there are no arid months with More

In total, 13 culicoid species were found in the present study, with 45.5% of the collected specimens belonging to the Obsoletus Complex while species only occasionally present in previous collections in Germany, accounted for approximately 25% of the sampled individuals. Thus, the species composition is only partly in accordance to earlier studies on the German Culicoides fauna according to which 70 to over 90% of the specimens belonged to the Obsoletus Complex and up to 20% represented members of the Pulicaris Complex, while other culicoid species were present in negligible numbers only12,13. However, previous studies were based on UV-light trap catches12,13,14,15 and targeted active culicoid specimens16. The results obtained in this study are very specific as they represent the species compositions associated with the respective breeding substrates.The gender ratio differed strongly between species, revealing no pattern applicable to all species. The dominance of female Culicoides emerging from breeding sites corresponds to earlier results17,18, even though the sex ratio in the present study showed a much higher proportion of females with 70.7% or a female:male ratio of 2.4:1 than the above studies with 55.6%17 or a female:male ratio of 1.06:118.The evaluation of the diversity of each biotope (excluding the ungrazed meadow where no Culicoides were found) revealed clear differences between the agriculturally used habitats and the more natural biotopes. The Shannon–Weaver index depicted very low diversity for all three studied meadows where biting midges were found. The two meadows (with cattle and sheep) of region 2 reached the lowest possible diversity. This seems plausible as only one species was sampled within each biotope. The meadow with cattle of region 1 revealed at least two species. The Evenness factor of 0.24 depicts the dominance of one of them. The low number of species and unbalanced number of specimens within the biotope result in a low Shannon–Weaver index of 0.24, which describes the poor level of biodiversity.The Simpson index measures the probability that two individuals, randomly selected from a sample, belong to the same species. As only one species was sampled on each meadow from region 2, the probability to choose two specimens which belong to one species is 100% (displayed by the value of D = 1.0). The meadow with cattle of region 2 revealed at least two culicoid species, but the dominance of one species leads to a high Simpson index of 0.92 as well.Opposite to the very low biodiversity of all meadows, the four more natural biotopes of region 3 show an overall high level of biodiversity: according to the Shannon–Weaver index, the level of biodiversity is highest within the AFS (H = 2.96). Compared to the other biotopes of region 3, the AFS revealed by far the highest numbers of culicoid species and specimens. This and the relatively high Evenness factor (E = 0.89) lead to the high H value. The Shannon–Weaver indices for CW and MA are 1.91 and 1.92, respectively. Based on the low numbers of species and specimens in both biotopes, the relatively high H value is mainly caused by its high Evenness values of 0.95 (CW) and 0.96 (MA), respectively. Therefore, the almost equal numbers of all present species leads to the relatively high biodiversity, rather than a high number of species.The Shannon–Weaver index of the DW is the lowest of the four biotopes of region 3 with H = 1.42 and rates this biotope as the one with the lowest diversity of region 3. Though the number of species equal the one of the CW and MA, the higher number of specimens and especially the much lower Evenness factor of 0.71 reduces the H value.Other than the Shannon–Weaver index, the Simpson index rates both, the AFS and the MA, as the two most diverse biotopes. With values of D = 0.13, the probability to randomly select two species of the same species is rather low in both biotopes. As the AFS revealed more than double as many species than the MA, the lower number of caught specimens of the MA must have led to the same biodiversity rate.Study 1—Influence of domestic animals on meadows: up to date, dung-breeding Culicoides have been investigated more thoroughly18,19,20 than most other culicoid species. Most studies have focused on examining selectively either dungheaps or cowpats, rather than conducting a direct comparison between grazed and ungrazed meadows under field conditions. In the present study, we were able to show that the ungrazed meadow seems to be an unsuitable breeding habitat for Culicoides. Therefore, it seems plausible that the suitability of meadows as culicoid breeding sites can be largely, if not completely, attributed to the influence of livestock pasturing.The strong dominance of Obsoletus Complex specimens sampled on grazed meadows is not surprising as this species complex is known to contain typical dung-breeders19,20. The high potential of manure as a breeding substrate has been demonstrated before21,22 and explains the high quantity of Culicoides developing on meadows used by cattle in the present study. While 0.83 midges/sample were found on the meadow with cattle in region 1, only 0.21 midges/sample were collected on the meadow with cattle in region 2. The quantitative differences between these two study sites might be caused by the differing time periods of sampling (April to July for region 1 and August to October for region 2). Previous studies observed population peaks of Obsoletus Complex midges in October, though23, giving reason to expect even higher numbers of midges for region 2 than for region 1, particularly so, as region 2 is an agriculturally dominated area with a higher abundance of potential blood hosts and more suitable breeding habitats than region 1.Compared to the much higher total number of midges emerging from cowpats, sheep dung produced only two specimens. The very low number of midges originating from sheep faeces might be due to the very quick decomposition and desiccation of the rather small droppings, which likely reduces the quality of these remains as culicoid breeding sites. Therefore, it can be assumed that, contrary to pastures with cattle dung, sheep-runs might not play an essential role in promoting the distribution of Culicoides. For modeling approaches, it should be considered, though, that this might only apply to single scattered pieces of faeces as the longer persistence of higher volumes of sheep dung, i.e. on muckheaps, might very likely raise its quality as potential breeding sites as observed by21.All grazed meadows revealed very few culicoid species. Besides members of the Obsoletus Complex, only one individual of C. comosioculatus was found. The present investigation represents a case study though as merely one habitat of each type was sampled. More research to confirm the present results is therefore strongly recommended, even more, as ceratopogonid communities of terrestrial ecosystems have been barely investigated24, with the consequence that breeding sites of Culicoides spp. are still poorly known25.Study 2—Quality of forest-dominated biotopes as culicoid breeding sites: In the present study, the AFS turned out to be very productive as a culicoid breeding site in regards to the number of caught specimens and species diversity. Ten of the 13 collected species were found in the AFS. This is 2.5 times as many species as in the three other biotopes of region 3, which contained four species each in different compositions. Therefore, species-specific requirements for larval development seem to be met for more culicoid species in the AFS than in any of the other study sites.The measured pH values are in accordance to soil analyses conducted in German forests26. As the top layers usually are the most acidic ones, the chosen depth of soil sampling in the present study (upper 0–5 cm) persistently produced low pH values. Additionally, the used solvent (CaCl2) is less sensitive to fast changing weather conditions, but also lowers the measured pH value significantly compared to distilled water26—a solvent often used in earlier studies analyzing the distribution of Ceratopogonidae.The wide variances of the soil factors, especially moisture and organic content, were mainly caused by unequal soil conditions within each biotope rather than changes over time (unpublished data). Nevertheless, the statistical analysis revealed that all four biotopes of region 3 were significantly different from each other regarding the three soil factors. Comparing the means of each soil factor revealed that the AFS contained a higher level of soil moisture, a less acidic pH value and a higher organic content than the other three biotopes of region 3. We could show that significantly more midges (0.4 Culicoides/sample) developed in the AFS compared to the three other biotopes of region 3 with 0.12 (DW), 0.07 (CW) and 0.06 (MA) Culicoides per sample.Previous studies have assumed that the level of moisture be a crucial factor for ceratopogonid development17,20. Also, some studies determined the organic content as pivotal17,27. Our statistical analysis revealed that each soil factor has an impact on the probability of Culicoides to occur. Due to high correlations between the various measured soil factors, it could not be clarified, though, whether they influence the number of specimens, too. But as many culicoid species are known to lay their eggs in batches and previous egg-laying encourages females to oviposit at the same site28, an increase in the probability of biting midge presence should indirectly result in a higher number of specimens, too.The aggregation of larvae in terrestrial habitats29 typically results in a high number of samples completely devoid of midges and an overall low number of specimens sampled by emergence traps30. Thus, the obtained low numbers of collected specimens are not surprising. Nevertheless, emergence traps are still considered to be the best tool for the investigation of breeding site productivity, as it offers a safe assignment of species to their specific developmental sites24,29,31.The Culicoides collected in this study are discussed on species level in regards to existing literature.Culicoides achrayi was found in the AFS. A swamp as a breeding site32 and soil located in stagnant water22 have previously been described for this species. We confirm June as the time of emergence32 and add that C. achrayi co-exists with C. pulicaris.Culicoides albicans was collected in the AFS and DW. Specimens hatched from late April to mid-June, representing one generation per year. We confirm co-habitation with C. pictipennis and C. kibunensis11,33 and the preference for very humid substrates which has been described for the wettest parts of boglands5,34 and for artificially waterlogged soil11. Our results show, that C. albicans larvae can tolerate medium moisture levels, too. The mean organic content of their developmental sites reached from moderate to high, and the pH values lay between strong and ultra-acidic.Culicoides comosioculatus was found on the meadow with cattle dung in mid-June. As only one individual (a gravid female with the presumed intention to oviposit) was collected and no literature regarding breeding sites of this species could be found, our finding only indicates that this species might possibly develop in animal dung although in extremely low numbers.Culicoides grisescens was found within the AFS, the CW and the DW from late May until mid-July. Kremer35 listed soils of swamps and boggy grasslands as developmental sites. We collected C. grisescens in three different biotopes with wide variances of the mean moisture level, mean organic content and mean pH value, which reveals the wide tolerance range of this species towards these three soil factors.Culicoides impunctatus was collected in the AFS and the CW from late May to mid-July, representing one generation per year. This finding differs from earlier observations of two generations per year in Scotland36. Previous studies described breeding sites as acidic, oligotrophic grasslands, swamps, boglands or marshes, often of a peaty consistence5,10,33,34,37 and with soil pH values of 5.0–6.5 (dissolved in distilled water)37. This matches the pH values of the AFS in the present study (lower, but dissolved in CaCl2), but excludes the much lower pH values of the CW. The range considered suitable for C. impunctatus larvae should therefore be extended downwards to as low as pH 2.9–3.9 (CaCl2). We found C. impunctatus in two biotopes comprising a wide variance regarding soil moisture and organic content, which illustrates the wide tolerance range of this species. Individuals of C. impunctatus co-exist with Obsoletus Complex specimens as both were collected within the same sample in the AFS.Culicoides kibunensis was collected in the AFS and MA, which matches earlier observations depicting swamps of eutrophic fresh water bodies17,34, soil of stagnant water bodies22 and acidic grasslands in considerable distances to swamps33 as breeding sites. The AFS and MA revealed pH values between 3.4 and 5.4. Soil moisture and organic content displayed wide variances. All specimens hatched from late May to mid-June. Culicoides kibunensis was found to co-exist with C. albicans as observed by Kettle33. Earlier observations of co-habitations with C. obsoletus s.s. and C. pallidicornis5,34 could not be confirmed.Obsoletus Complex members were present in all study sites except for the ungrazed meadow. In the grazed meadows, Obsoletus Complex midges emerged almost throughout the entire sampling period except for the month of September. Two peaks were observed, one in June/July and a smaller one in October. As in the grazed meadows, the biotopes of region 3 also revealed two generations, but emerging at a slightly earlier time of the year with one peak in May/June and the other one in September/October.Members of the Obsoletus Complex are known to be generalists regarding their choice of breeding sites. Only the identified member species, C. chiopterus and C. obsoletus s.s., are considered here.Culicoides chiopterus was exclusively found on meadows grazed by cattle, which is in accordance to several earlier studies as this species is described as a dung-breeding species developing in cowpats and horse droppings5,34,35,38.Culicoides obsoletus s.s. was mostly sampled in the AFS. Only one individual was collected on a meadow grazed by cattle. Previous descriptions of breeding sites differed widely. Acidic grasslands in considerable distance to bogs/swamps33 and leaf litter compost5,35 could not be confirmed in the present study, although the MA and AFS were of a comparable character. While Uslu and Dik17 could not find any C. obsoletus s.s. in wet organic matter-rich soil, we collected most specimens of this species in the AFS and can therefore confirm previous findings11,29,32,39. The time of C. obsoletus s.s. activity in Germany (April–October) as described by Havelka32 agrees with our observations.Culicoides pallidicornis was found in the MA in late June. This species revealed the smallest variances of all sampled biting midge species regarding the three soil factors, using soil with pH values of 3.6–5.0 (CaCl2) and a relatively low level of moisture. This contradicts earlier observations where C. pallidicornis developed in the mud of eutrophic fresh-water swamps5. While C. pallidicornis larvae are known to co-exist with C. kibunensis5, we can add C. subfagineus to share the same developmental site.Culicoides pictipennis was collected in the DW and, to a minor part, in the AFS. The preferred physicochemical breeding conditions were ultra to extremely acidic with a medium moisture level and a moderate to slightly increased organic content. This differs from previous studies, which have found this species to develop only at the margin of stillwater bodies like pools and ponds, and the littoral of lakes or in artificially waterlogged soil11,32,34. Havelka32 observed C. pictipennis between May and June, while in our investigation the first specimen emerged as early as mid-April. We can confirm the co-existence of C. pictipennis and C. albicans as previously observed by Harrup11.Culicoides pulicaris was sampled in the AFS from late June until September, which agrees with observations denoting May to September as the activity time of this species32. Culicoides pulicaris seems to prefer breeding substrates with a high moisture level and a high organic content, as previously described17,32,34. We can add that C. pulicaris breeds in soil showing pH values at least between 4.0 and 5.4. We collected C. pulicaris together with C. achrayi and found it to simultaneously emerge from one biotope with C. obsoletus s.s. Additionally, we can confirm the co-existence of C. pulicaris with C. punctatus5,40, since both species have similar breeding habitat preferences11.Culicoides punctatus was sampled in the AFS and, to a minor part, in the CW. Time of emergence was from mid-June to late September, which is in accordance with earlier observations listing April-August and October as times of activity32. In the present study, a strong preference for swampy conditions with soil of high moisture, high organic content and a strong to very strong acidity was found. This is in agreement to previous findings11,32,41. The co-existence of C. punctatus with C. pulicaris is well known5,40 and can be confirmed once more. Additionally, we found C. punctatus to co-occur with C. subfasciipennis.Culicoides subfagineus was caught in the MA in late June. The soil was oligotrophic and contained a relatively low moisture level with pH values between 3.6 and 5.0. The first record of this species in Germany was in 2014, when C. subfagineus was observed to attack cattle42.Culicoides subfasciipennis was sampled in mid-June in the AFS. The time and choice of breeding site are in accordance to previous findings17,32. Breeding conditions for the only individual collected revealed a medium soil moisture factor, a pH value of 5.2 and a medium organic content. The species was found to co-develop with C. punctatus. More

Mosquito lines and maintenanceTo introduce Wolbachia into Brazilian Ae. aegypti, an Australian line infected with the wMel strain21 was backcrossed for 8 generations to a natural mosquito population of Rio de Janeiro, Brazil24. Following the genetic background introgression, additional crosses and knockdown resistance (kdr) screening were undertaken to replicate natural insecticide resistance profiling and generate the line wMelRio. To assure a minimal variation in this profiling overtime, and sustain a homogeneous genetic background, wMelRio colony was refreshed with 10% wild males once in every five generations40.To maintain wMelRio, immatures (i.e. larval stages L1 to L4) were reared in dechlorinated water, at 28 °C, and fed Tetramin Flakes (Tetra GmbH, Herrenteich, Germany) until pupal formation. Following adult emersion, groups of 1000 females and 800 males were sorted and kept in BugDorm cages (MegaView Science Co Ltd., Taiwan) at 25 °C, with 10% sucrose solution ad libitum. Every three days, females were fed human blood (from blood donation centers; see details under ‘ethical considerations’), through Hemotek artificial feeders (Hemotek Ltd, UK). Note that, to avoid arboviral contamination of our colony, all blood samples were formerly tested negative for DENV, ZIKV, CHIKV, MAYV and YFV by multiplex qPCR assays36,68. Egg-laying was induced by placing dampened strips of filter paper (i.e. partially immersed in water-containing cups) inside the cages for 2–3 days, after which they were gradually dried at room temperature. Strips loaded with eggs (i.e. ovistrips) were kept at room temperature until further use, either for colony maintenance or field release. Eggs older than 40 days were discarded due to a decay in overall quality60.Egg releasesMass-reared wMel-infected Brazilian Ae. aegypti, wMelRio, were released as eggs in Jurujuba (22°56′ 00″ S, 43°07′ 00″ W), a lower-middle-class community in the city of Niterói (state of Rio de Janeiro, Brazil). Located by the shores of Guanabara bay, this community has grown from a typical fisherman settlement, with informal occupancy, to a total population of 2797 residents in 890 houses. Jurujuba encompasses a total area of 2.53 km2, divided into seven smaller sectors (i.e. sub-areas or localities within the neighborhood): Ponto Final, Várzea, Brasília, Cascarejo, Praia de Adão e Eva, Peixe-Galo and Salinas.wMelRio eggs were released in the field through special deployment devices, referred to as mosquito release containers (MRCs), which consisted of small white plastic buckets (19 cm height × 18 cm top diameter × 15.5 cm base diameter) with four small holes on the side wall, only a few centimeters away from the top lid. Each MRC was loaded with 1 L of water, 0.45 g of Tetramin Tropical Tablets (i.e. one and a half tablet) (Tetra GmbH, Herrenteich, Germany) and an ovistrip containing approximately 150–300 eggs. After six to seven days, about 80% of the immatures were pupae, and after 11 to 12 days, most of the adults had already emerged and left the device from the wall holes. Every 15 days, MRCs were checked and reloaded so that another rearing and release cycle could take place. Release sites were spatially distributed as evenly as possible (Supplementary Fig. S1), so as to maximize the spread of Wolbachia-harboring individuals and promote mating with their wild peers. The release strategy was optimized by splitting the sites into two groups, A and B, with alternate MRC loading schedules. Thus, while MRCs from group A were releasing adults, those from group B were being loaded with new ovistrips, water and food. In the following week, an opposite situation occurred, with MRCs from group B releasing adults. The release schedules, as well as the number of allocated MRCs, varied according to each Jurujuba’s sector (Supplementary Table S1).Ethical considerationsAll methods were carried out in accordance with relevant guidelines and regulations. Study protocol for Wolbachia field release was approved by the National Research Ethics Committee (CONEP, CAAE 02524513.0.1001.0008) and three government agencies: IBAMA (Ministry of Environment), Anvisa (Ministry of Health) and MAPA (Ministry of Agriculture, Livestock and Supply) to obtain the RET (Special Temporary Registry, 25351.392108/2013-96). Prior to mosquito releases, an informed consent was obtained from 70% of Jujuruba households. Also, a written informed consent was obtained from households that hosted BG-sentinel mosquito traps.For the maintenance and mass-rearing of Wolbachia-infected Ae. aegypti, adult females were fed human blood from a donation center (Hospital Antonio Pedro, Rio de Janeiro State University), with supporting regulatory approval (CONEP, CAAE 59175616.2.0000.0008) We only used blood bags which would have been discarded by the donation center, mainly due to insufficient volume to meet their quality assurance policy. Samples had no information on donor’s identity, sex, age and any clinical condition, but were tested negative for several diseases, including Hepatitis B, Hepatitis C, Chagas disease, syphilis, HIV and HTLV, as part of the Brazilian Government routine screening.For vector competence assays, human blood was obtained from Fundação Hemominas as part of a research agreement with Instituto René Rachou (Fiocruz Minas) (OF.GPO/CCO-Nr224/16).
Wolbachia field monitoring and density level assessmentAe. aegypti field population was monitored with BG-Sentinel traps (Biogents AG, Regensburg, Germany), spread across Jurujuba in a semi-homogeneous fashion (Supplementary Fig. S2, Supplementary Table S2, Supplementary Datasheet S1). These monitoring sites were chosen among suitable households who formally agreed with hosting of a trap, and had to be reallocated according to necessity (i.e. household quits hosting the trap). Working traps were checked weekly by removing the catch bags (e.g. small meshed envelopes placed inside the BG-Sentinels to collect trapped insects) and bringing them to the laboratory for species identification and Wolbachia screening. Catch bags were barcoded according to the trap ID and site, so as to create a pipeline for field samples.Screening for Wolbachia in Ae. aegypti samples was undertaken by qPCR. Briefly, individual DNA was extracted by homogenizing head/thorax body parts in Squash Buffer (10 mM Tris–Cl, 1 mM EDTA, 25 mM NaCl, pH 8.2) supplemented with Proteinase K (200 ug/ml) and incubating at 56 °C for 5 min. Extraction ended by enzyme inactivation at 98 °C for 15 min. DNA amplifications were carried out with FastStart Essential DNA Probes Master (Roche), using specific primers and probes to Wolbachia pipientis WD0513 and Ae. aegypti rps17 markers (Supplementary Table S3). Thermocycling conditions were set on a LightCycler 96 Instrument (Roche), as follows: 95 °C for 10 min (initial denaturation), and 40 cycles of 95 °C for 15 s and 60 °C for 30 s. Samples were analyzed using absolute quantification, by comparison to serial dilutions of either gene product, cloned and amplified in the pGEMT-Easy plasmid (Promega). Negative control samples were normalized between plates, and were used as reference to determine a minimum threshold for positive samples.DENV and ZIKV isolation and replication in mosquito cellsZIKV was kindly provided by Instituto Aggeu Magalhães (IAM, Fiocruz) through viral isolation of a symptomatic patient sample from Recife (PE, Brazil) in 2015 (ZikV/H.sapiens/Brazil/BRPE243/2015). DENV was sourced following a viral isolate from a patient sample diagnosed with Dengue type 1 in Contagem (MG, Brazil), also in 2015 (Den1/H.sapiens/Brazil/BRMV09/2015). Both ZIKV and DENV samples were accompanied by patients’ written consent (CONEP, reference number 862.912), being further catalogued into the national database of genetic patrimony and associated knowledge (SISGEN, access number AA1D462).In vitro culture of viral particles were done as previously described36. Briefly, ZIKV and DENV were replicated in Aedes albopictus C6/36 cells, grown at 28 °C in Leibovitz L-15 medium (ThermoFisher) supplemented with 10% fetal bovine serum (FBS) (ThermoFisher). After seven days, supernatants were harvested and virus titers were assessed, first by Reverse Transcription (RT)-qPCR, and later by plaque assay with VERO cells grown under 37 °C, 5% carbon dioxide, in Dulbecco’s Modified Eagle Medium (DMEM) (ThermoFisher) supplemented with 3% Carboxymethylcellulose (Synth) and 2% FBS.Vector competence assaysTo perform vector competence assays with field samples of Ae. aegypti, ovitraps were mounted in both Ponto Final (Jurujuba) and Urca, a Wolbachia-free area in Rio de Janeiro. Ovitraps were collected from the field over 13 weeks, from April to June 2017, which corresponds to the time-frame between 14 and 16 months along the post-release phase in Ponto Final. Once in the insectary, eggs samples were reared to the adult stage in a controlled insectary environment (refer to ‘mosquito lines and maintenance’ for details).For virus challenging assays through oral-feeding, young females (4–6 days old) were starved for 20 to 24 h, and subsequently offered culture supernatant containing ZIKV or DENV mixed with human red blood cells (2:1 ratio), using an artificial membrane feeding system36. It is important to mention that, as for the colony maintenance protocol, blood samples used here were also submitted to quality control prior to its use in the assays, mainly due to putative arbovirus contaminations which could affect the experimental outputs. Likewise, all samples were tested negative for DENV, ZIKV, CHIKV, MAYV and YFV by multiplex qPCR assays36,68. Oral-infections were performed twice for each virus. ZIKV was offered first from fresh (initial virus titer of 4.8 × 108 PFU/mL) and second from frozen culture supernatant (initial virus titer of 7.6 × 106 PFU/mL). In contrast, DENV was offered from fresh supernatants only (virus titers of 2 × 106 PFU/mL and 6.5 × 107 PFU/mL), since frozen versions failed to infect. Specimens were allowed to feed for one hour, after which engorged females were selected and maintained with 10% sucrose solution ad libitum, during the whole extrinsic incubation period. At 14 days post-infection (dpi), viral loads were assessed in heads/thorax extracts by RT-qPCR (refer to ‘Viral diagnosis’ for more details).For saliva-mediated virus challenging assays, ZIKV and DENV pre-exposed females (14 dpi) from Jurujuba (Wolbachia +) and Urca (Wolbachia −) were starved for about 16 h (overnight) before being knocked down and kept at 4 °C for wings and legs removal. Salivation was induced by introducing a 10 µL sterile filter tip, pre-loaded with 5 µl of a solution [30% sucrose (w/v) diluted in 50% fetal bovine serum (FBS) and 50% DMEM medium], into the mosquito proboscis for 30 min. Saliva samples were individually collected, and 276 nL was intrathoracically injected into young naive females (Urca) using a Nanoject II hand held injector (Drummond), as previously described36,68. Each saliva sample was used to inoculate 8–14 naïve Wolbachia-free Ae. aegypti specimens, of which 8 were screened for infective particles. ZIKV and DENV were quantified by RT-qPCR at 5 dpi and 7 dpi, respectively (refer to ‘Viral diagnosis’ for more details). Overall Intrathoracic Saliva Infection index (OISI) was obtained by averaging the percentages (± SD) of infected individuals in each group.Viral diagnosisTo identify ZIKV and DENV particles in individual samples, whole specimens were processed into head/thorax homogenates for RNA/DNA extraction with the High Pure Viral Nucleic Acid Kit (Roche), according to manufacturer’s instructions30. Extracted samples were diluted in nuclease-free water to a concentration of 50 ng/μL. ZIKV, DENV and Wolbachia levels, in vector competence assays, were quantified by RT-qPCR using TaqMan Fast Virus 1-Step Master Mix (ThermoFisher) and specific primers and probes (Supplementary Table S3). Reactions were run on a LightCycler 96 Instrument (Roche), using the following thermocycling conditions: 50 °C for 5 min (initial RT step), 95 °C for 20 s (RT inactivation/DNA initial denaturation), and then 40 cycles of 95 °C for 3 s and 60 °C for 30 s. Each RNA/DNA sample was used in two reactions, one with ZIKV, DENV or Wolbachia primers, and another with Ae. aegypti rps17 endogenous control30. Absolute quantification was achieved by comparing amplification profiles with standard curves generated by serial dilutions of their respective gene products, amplified from a cloned sequence in pGEM-T Easy vector (Promega). Negative control samples (no virus RNA) served as reference to fix a minimum threshold for positive ones. ZIKV and DENV loads were defined as their copy number per sample (head/thorax or saliva), while Wolbachia loads were relative quantifications to the rps17 reference gene. Here, it is worth noting that, while Wolbachia titer is naturally variable and dependent on its whole-body density, the overall expression of rps17 is stable and particularly suitable for internal controls in assays with adult females69, as demonstrated previously by us and others30,62,68.Map creation and source codesThe satellite image map of Jurujuba was created with ArcGIS Desktop 10.7 (Esri Inc., https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview) using Google Earth (Google LLC) source code, under the license and in accordance with the fair use described in ‘https://about.google/brand-resource-center/products-and-services/geo-guidelines/’. Maps with geotagged MRCs and BG-Sentinel traps were created with ArcGIS Desktop 10.7 and OpenStreetMap source code (OpenStreetMap contributors), under the license CC-BY-SA 2.0.Statistical analysesGraphs and statistical analyzes were performed in GraphPad Prism 8 (GraphPad Software Inc., https://www.graphpad.com). Kruskal–Wallis test followed by Dunn’s post-hoc multiple comparisons were used to analyze Wolbachia density data from field-collected and colony samples. ZIKV and DENV loads in head/thorax extracts, from both oral and saliva-challenging samples, were compared using the Mann–Whitney U test. For all statistical inferences, ⍺ was set to 0.05. More

Ethics statementAll procedures were conducted in accordance with the current laws in Thailand on experimental animals and were approved by the safety management committee for experiments of the Laboratory Animal Center, Chiang Mai University (Project Number 2561/FA-0001). The study also followed the recommendations in the ARRIVE guidelines.Species-specific primer designAll the DNA tissue analysed originated from the mucus of the individual giant snakehead. Total DNA was extracted from the mucus sample using the Qiagen DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA). Extracted DNA was used as a template for qPCR assay together with synthetic fragments. DNA samples were quantified using a Qubit fluorometer (Life Technologies) calibrated with the Quant-iT dsDNA HS Assay following the manufacturer’s instructions. For each replicate, 3 µL volumes were measured.Species-specific primers and a minor-groove binding (MGB) probe incorporating a 5′ FAM reporter dye and a 3′ non-fluorescent quencher were designed to amplify an 127 bp targeting within the 16S region for the giant snakehead (C. micropeltes), using Primer Express (V3.0, Life Technologies; Table 3). Probe and primer sequences were matched against the National Centre for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov/) nucleotide database with BLASTn (Basic Local Alignment Search Tool) to confirm the species’ specificity for the giant snakehead in silico assays.Table 3 Details of species-specific primers and the probe designed to amplify a 127 bp fragment of the 16S region of Channa micropeltes (Cuvier, 1831).Full size tableTo ensure that the assay only amplified the giant snakehead, it was deployed on a closely related species commonly found in Thai freshwater environments using conventional PCR amplification and visualization on a 1.5% agarose gel stained with SYBR Safe DNA Gel Stain (Life Technologies).qPCR assayThe qPCR assay was deployed using Environmental Master Mix (Applied Biosystems) on mucus samples from the giant snakehead and related species to ensure the species specificity to the qPCR assay. In addition, eDNA qPCR assay for the giant snakehead, a water sample collected from tank at Phayao Freshwater Aquarium (Phayao Inland Fisheries Research and Development Center) was known to have only the giant snakehead was included as a positive control for the presence of amplifiable eDNA in water samples. The tank contains around 4.5 m3 of water with one individual of giant snakehead resides in the tank (the fish is about 60–70 cm in length).All eDNA qPCR amplifications were performed in three replicates in a final volume of 20 µL, using 10.0 µL of 2 × TaqMan Environmental Master Mix 2.0 (Thermo Fisher Scientific), 2.0 µL of DNA template, 900 nM each of the F/R primers, and 125 nM of the probe. Samples were run under the following conditions: an initial 10 min incubation at 95 °C followed by 50 cycles of denaturation at 95 °C for 15 s and annealing/extension at 60 °C for 1 min. Negative controls with all PCR reagents but no template (three replicates) were run in parallel to assess potential contamination. The quantification cycle (Cq) was converted to quantities per unit volume using the linear regression obtained from the synthesized target gene standard curve (Integrated DNA Technologies Pte. Ltd., Singapore). The giant snakehead eDNA concentrations were then reported as copies/mL. The limit of detection (LOD) and the limit of quantification (LOQ) were also measured using the standard dilution series of synthesized target gene fragment with known copy numbers. A dilution series containing 1.5 × 101 to 1.5 × 104 copies per PCR tube were prepared and used as quantification standards. The calculation of LOD and LOQ was done using published R script by Klymus et al.26.DNA extraction from the filtersDNA trapped on the filters obtained from the aquarium experiments and field collections were extracted using Qiagen DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) using a protocol modified from the manufacturer’s protocol with the following changes: the DNA from all samples were eluted twice with 25 µL AE buffer, in a total volume of 50 µL to obtain a more concentrated eDNA solution. The volume of ATL buffer (360 µL), Proteinase K (40 µL), AL buffer (400 µL) and Ethanol (400 µL) were doubled.Aquarium experimentAn aquarium experiment was used to test the extent to which qPCR of water samples can detect eDNA of giant snakehead at low simulated densities. The juvenile giant snakehead was obtained from the fish store and transported to a laboratory at Chiang Mai University. The giant snakeheads were then held in separate 120 L plastic holding containers in which the water was continuously filtered. The fish were fed frozen shrimp/commercially available flake fish food three times a week, and were held at 23 ± 1 °C.The sensitivity of eDNA detection in the aquaria was evaluated by conducting three aquarium experiments using plastic tanks (30 × 45 × 25 cm) filled with 120 L of aged-tap water. The water in the tanks was continuously aerated through a filter. In each experiment, the giant snakeheads were randomly assigned to the tanks (10 individuals per tank). The average size of the snakeheads was 9.7 cm (body length ranging from 9.1 to 10.6 cm). The average weight was 8.15 g (ranging from 6.7 to 10.6 g). The water in the tanks was maintained at 23 ± 1 °C. A 300 mL water sample from each tank was collected at each time point (0, 3, 6, 12, 24, 48, 72, 96, 120, 144, and 168 after removal of the fishes from the tanks) in triplicate. Collected water was filtered on a GF/F filter (0.7 μm Whatman International Ltd., Maidstone, UK). The eDNA from each sample solution was extracted using a Qiagen DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) in a final volume of 50 µL, detailed in DNA extraction from the filters. To confirm the absence of the giant snakehead eDNA in the water prior to the experiments, three tanks without giant snakehead were prepared and water sample was collected and treated as described above.Real-time PCR was performed with the species-specific primers and probe set using a Rotor-Gene Q system (Qiagen, Hilden, Germany). The reaction conditions were the same as described in qPCR assay. Three replicates were conducted for each sample including the negative PCR control and positive control.eDNA field collectionWater samples were collected at 6 points within Kwan Payao according to the survey locations of the Inland Fisheries Research and Development Center. Additional water samples were collected from 11 and 6 locations in Ing River where water flowed into and out of Kwan Payao, respectively (Fig. 1). To avoid contamination, all field equipment was sterilized using 10% bleach, UV-Crosslinker or autoclaved and sealed prior to transport to the study site, and a separate pair of nitrile disposable gloves were used for each sample. At each site, water samples were collected three replicate in bucket that had been previously decontaminated with a 10% bleach rinse followed by two distilled water rinses.In total, water samples were collected from 6 sites (in Kwan Phayao) and from 17 sites (in the Ing River) from 15th February to 5th March 2019, the middle of the dry season. Each site was sampled in triplicate, 300 mL samples of water were collected and filtered on GF/F filter (0.7 μm Whatman International Ltd., Maidstone, UK). In total, 306 water samples were collected from the surface water of lakes and rivers. For every sampling day, deionised water (300 mL) was filtrated as a negative control. The water samples and real-time PCR was processed as described above in qPCR assay. More

1.Heger, T. & Jeschke, J. M. The enemy release hypothesis as a hierarchy of hypotheses. Oikos 123, 741–750 (2014).Article 

Google Scholar 
2.Elton, C. S. The Ecology of Invasions by Animals and Plants (Springer, 1958).3.Keane, R. M. & Crawley, M. J. Exotic plant invasions and the enemy release hypothesis. Trends Ecol. Evol. 17, 164–170 (2002).Article 

Google Scholar 
4.Mitchell, C. E. & Power, A. G. Release of invasive plants from fungal and viral pathogens. Nature 421, 625–627 (2003).ADS 
CAS 
PubMed 
Article 
PubMed Central 

Google Scholar 
5.Colautti, R. I., Ricciardi, A., Grigorovich, I. A. & MacIsaac, H. J. Is invasion success explained by the enemy release hypothesis? Ecol. Lett. 7, 721–733 (2004).Article 

Google Scholar 
6.Liu, H. & Stiling, P. Testing the enemy release hypothesis: a review and meta-analysis. Biol. Invasions 8, 1535–1545 (2006).Article 

Google Scholar 
7.Meijer, K., Schilthuizen, M., Beukeboom, L. & Smit, C. A review and meta-analysis of the enemy release hypothesis in plant-herbivorous insect systems. PeerJ 4, e2560v1 (2016).Article 

Google Scholar 
8.Jeschke, J. M. & Heger, T. (eds) Invasion Biology: Hypotheses and Evidence (CABI, 2018).9.Levine, J. M., Adler, P. B. & Yelenik, S. G. A meta-analysis of biotic resistance to exotic plant invasions. Ecol. Lett. 7, 975–989 (2004).Article 

Google Scholar 
10.Maron, J. L. & Vilà, M. When do herbivores affect plant invasion? Evidence for the natural enemies and biotic resistance hypotheses. Oikos 95, 361–373 (2001).Article 

Google Scholar 
11.Callaway, R. M. & Ridenour, W. M. Novel weapons: invasive success and the evolution of increased competitive ability. Front. Ecol. Environ. 2, 436–443 (2004).Article 

Google Scholar 
12.Cappuccino, N. & Arnason, J. T. Novel chemistry of invasive exotic plants. Biol. Lett. 2, 189–193 (2006).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
13.Bezemer, T. M., Harvey, J. A. & Cronin, J. T. Response of native insect communities to invasive plants. Ann. Rev. Entomol. 59, 119–141 (2014).CAS 
Article 

Google Scholar 
14.Keeler, M. S. & Chew, F. S. Escaping an evolutionary trap: preference and performance of a native insect on an exotic invasive host. Oecologia 156, 559–568 (2008).ADS 
PubMed 
Article 
PubMed Central 

Google Scholar 
15.Eckberg, J. O., Tenhumberg, B. & Louda, S. M. Insect herbivory and propagule pressure influence Cirsium vulgare invasiveness across the landscape. Ecology 93, 1787–1794 (2012).PubMed 
Article 
PubMed Central 

Google Scholar 
16.Bürki, C. & Nentwig, W. Comparison of herbivore insect communities of Heracleum sphondylium and H. mantegazzianum in Switzerland (Spermatophyta: Apiaceae). Entomol. Gen. 22, 147–155 (1997).Article 

Google Scholar 
17.Cincotta, C. L., Adams, J. M. & Holzapfel, C. Testing the enemy release hypothesis: a comparison of foliar insect herbivory of the exotic Norway maple (Acer platanoides L.) and the native sugar maple (A. saccharum L.). Biol. Invasions 11, 379–388 (2008).Article 

Google Scholar 
18.Cronin, J. T., Bhattarai, G. P., Allen, W. J. & Meyerson, L. A. Biogeography of a plant invasion: plant-herbivore interactions. Ecology 96, 1115–1127 (2015).PubMed 
Article 
PubMed Central 

Google Scholar 
19.Hu, X.-T. & Dong, B.-C. Herbivory and nitrogen availability affect performance of an invader Alternanthera philoxeroides and its native congener A. sessilis. Flora 257, 151412 (2019).Article 

Google Scholar 
20.Agrawal, A. A. & Kotanen, P. M. Herbivores and the success of exotic plants: a phylogenetically controlled experiment. Ecol. Lett. 6, 712–715 (2003).Article 

Google Scholar 
21.Agrawal, A. A. et al. Enemy release? An experiment with congeneric plant pairs and diverse above- and belowground enemies. Ecology 86, 2979–2989 (2005).Article 

Google Scholar 
22.Parker, J. D. & Hay, M. E. Biotic resistance to plant invasions? Native herbivores prefer non-native plants. Ecol. Lett. 8, 959–967 (2005).Article 

Google Scholar 
23.Parker, J. D., Burkepile, D. E. & Hay, M. E. Opposing effects of native and exotic herbivores on plant invasions. Science 311, 1459–1461 (2006).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
24.Parker, I. M. & Gilbert, G. S. When there is no escape: the effects of natural enemies on native, invasive, and non-native plants. Ecology 88, 1210–1224 (2007).PubMed 
Article 
PubMed Central 

Google Scholar 
25.Dostál, P. et al. Enemy damage of exotic plant species is similar to that of natives and increases with productivity. J. Ecol. 101, 388–399 (2013).Article 

Google Scholar 
26.Meijer, K. et al. Phytophagous insects on native and non-native host plants: combining the community approach and the biogeographical approach. PLoS ONE 10, e0125607 (2015).PubMed 
PubMed Central 
Article 
CAS 

Google Scholar 
27.Schultheis, E. H., Berardi, A. E. & Lau, J. A. No release for the wicked: enemy release is dynamic and not associated with invasiveness. Ecology 96, 2446–2457 (2015).PubMed 
Article 
PubMed Central 

Google Scholar 
28.Beckstead, J. & Parker, I. M. Invasiveness of Ammophila arenaria: release from soil-borne pathogens? Ecology 84, 2824–2831 (2003).Article 

Google Scholar 
29.van Kleunen, M., Weber, E. & Fischer, M. A meta-analysis of trait differences between invasive and non-invasive plant species. Ecol. Lett. 13, 235–245 (2010).PubMed 
Article 
PubMed Central 

Google Scholar 
30.Ashton, I. W. & Lerdau, M. T. Tolerance to herbivory, and not resistance, may explain differential success of invasive, naturalized, and native North American temperate vines. Divers. Distrib. 14, 169–178 (2008).Article 

Google Scholar 
31.Seebens, H. et al. No saturation in the accumulation of alien species worldwide. Nat. Commun. 8, 14435 (2017).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
32.Eppinga, M. B., Rietkerk, M., Dekker, S. C., De Ruiter, P. C. & van der Putten, W. H. Accumulation of local pathogens: a new hypothesis to explain exotic plant invasions. Oikos 114, 168–176 (2006).Article 

Google Scholar 
33.Bufford, J. L. et al. Taxonomic similarity, more than contact opportunity, explains novel plant–pathogen associations between native and alien taxa. N. Phytol. 212, 657–667 (2016).CAS 
Article 

Google Scholar 
34.White, E. M., Wilson, J. C. & Clarke, A. R. Biotic indirect effects: a neglected concept in invasion biology. Divers. Distrib. 12, 443–455 (2006).Article 

Google Scholar 
35.Allen, W. J. in Plant Invasions: The Role of Species Interactions (CABI Publishing, 2020).36.Holt, R. D. Predation, apparent competition, and the structure of prey communities. Theor. Popul. Biol. 12, 197–229 (1977).MathSciNet 
CAS 
PubMed 
Article 

Google Scholar 
37.Holt, R. D. & Bonsall, M. B. Apparent competition. Annu. Rev. Ecol. Evol. Syst. 48, 447–471 (2017).Article 

Google Scholar 
38.Sessions, L. & Kelly, D. Predator-mediated apparent competition between an introduced grass, Agrostis capillaris, and a native fern, Botrychium australe (Ophioglossaceae), in New Zealand. Oikos 96, 102–109 (2002).Article 

Google Scholar 
39.Dangremond, E. M., Pardini, E. A. & Knight, T. M. Apparent competition with an invasive plant hastens the extinction of an endangered lupine. Ecology 91, 2261–2271 (2010).PubMed 
PubMed Central 
Article 

Google Scholar 
40.Bhattarai, G. P., Meyerson, L. A. & Cronin, J. T. Geographic variation in apparent competition between native and invasive Phragmites australis. Ecology 98, 349–358 (2017).PubMed 
Article 
PubMed Central 

Google Scholar 
41.Simberloff, D. & Von Holle, B. Positive interactions of nonindigenous species: invasional meltdown? Biol. Invasions 1, 21–32 (1999).Article 

Google Scholar 
42.Allen, W. J. et al. Community-level direct and indirect impacts of an invasive plant favour exotic over native species. J. Ecol. 108, 2499–2510 (2020).Article 

Google Scholar 
43.Morris, R. J., Lewis, O. T. & Godfray, C. J. Experimental evidence for apparent competition in a tropical forest food web. Nature 428, 310–313 (2004).ADS 
CAS 
PubMed 
Article 
PubMed Central 

Google Scholar 
44.Tack, A. J. M., Gripenberg, S. & Roslin, T. Can we predict indirect interactions from quantitative food webs? – an experimental approach. J. Anim. Ecol. 80, 108–118 (2011).PubMed 
Article 

Google Scholar 
45.Frost, C. M. et al. Apparent competition drives community-wide parasitism rates and changes in host abundance across ecosystem boundaries. Nat. Commun. 7, 12644 (2016).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
46.Bardgett, R. D. & Wardle, D. A. Aboveground–Belowground Linkages: Biotic Interactions, Ecosystem Processes and Global Change (Oxford University Press, 2010).47.Heinen, R., Biere, A., Harvey, J. A. & Bezemer, T. M. Effects of soil organisms on aboveground plant-insect interactions in the field: patterns, mechanisms and the role of methodology. Front. Ecol. Evol. 6, 106 (2018).Article 

Google Scholar 
48.Bever, J. D., Westover, K. M. & Antonovics, J. Incorporating the soil community into plant population dynamics: the utility of the feedback approach. J. Ecol. 85, 561–573 (1997).Article 

Google Scholar 
49.Kulmatiski, A., Beard, K. H., Stevens, J. R. & Cobbold, S. M. Plant–soil feedbacks: a meta‐analytical review. Ecol. Lett. 11, 980–992 (2008).PubMed 
Article 

Google Scholar 
50.Levine, J. M., Pachepsky, E., Kendall, B. E., Yelenik, S. G. & Lambers, J. H. Plant-soil feedbacks and invasive spread. Ecol. Lett. 9, 1005–1014 (2006).PubMed 
Article 
PubMed Central 

Google Scholar 
51.Suding, K. N., Harpole, W. S., Fukami, T., Kulmatiski, A., MacDougall, A. S., Stein, C. & van der Putten, W. H. Consequences of plant–soil feedbacks in invasion. J. Ecol. 101, 298–308 (2013).Article 

Google Scholar 
52.Crawford, K. M. et al. When and where plant-soil feedback may promote plant coexistence: a meta-analysis. Ecol. Lett. 22, 1274–1284 (2019).PubMed 
PubMed Central 

Google Scholar 
53.Cornelissen, T., Fernandes, G. W. & Vasconcellos-Neto, J. Size does matter: variation in herbivory between and within plants and the plant vigor hypothesis. Oikos 117, 1121–1130 (2008).Article 

Google Scholar 
54.Price, P. W. The plant vigor hypothesis and herbivore attack. Oikos 62, 244–251 (1991).Article 

Google Scholar 
55.Waller, L. P. et al. Biotic interactions drive ecosystem responses to plant invaders. Science 368, 967–972 (2020).CAS 
PubMed 
Article 
PubMed Central 

Google Scholar 
56.Kozlov, M. V., Lanta, V., Zverev, V. & Zvereva, E. L. Global patterns in background losses of woody plant foliage to insects. Glob. Ecol. Biogeogr. 24, 1126–1135 (2015).Article 

Google Scholar 
57.Zas, R., Moreira, X. & Sampedro, L. Tolerance and induced resistance in a native and an exotic pine species: relevant traits for invasion ecology. J. Ecol. 99, 1316–1326 (2011).Article 

Google Scholar 
58.Croy, J. R., Meyerson, L. A., Allen, W. J., Bhattarai, G. P. & Cronin, J. T. Lineage and latitudinal variation in Phragmites australis tolerance to herbivory: implications for invasion success. Oikos 129, 1341–1357 (2020).Article 

Google Scholar 
59.Liu, G., Huang, Q.-Q., Lin, Z.-G., Huang, F.-F., Liao, H.-X. & Peng, S.-L. High tolerance to salinity and herbivory stresses may explain the expansion of Ipomoea cairica to salt marshes. PLoS ONE 7, e48829 (2012).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
60.Paynter, Q. et al. Why did specificity testing fail to predict the field host-range of the gorse pod moth in New Zealand. Biol. Control 46, 453–462 (2008).Article 

Google Scholar 
61.Groenteman, R., Fowler, S. V. & Sullivan, J. J. St. John’s wort beetles would not have been introduced to New Zealand now: a retrospective host range test of New Zealand’s most successful weed biocontrol agents. Biol. Control 57, 50–58 (2011).Article 

Google Scholar 
62.Blossey, B. & Nötzold, R. Evolution of increased competitive ability in invasive nonindigenous plants: a hypothesis. J. Ecol. 83, 887–889 (1995).Article 

Google Scholar 
63.Felker-Quinn, E., Schweitzer, J. A. & Bailey, J. K. Meta-analysis reveals evolution in invasive plant species but little support for evolution of increased competitive ability (EICA). Ecol. Evol. 3, 739–751 (2013).PubMed 
PubMed Central 
Article 

Google Scholar 
64.Rotter, M. C. & Holeski, L. M. A meta-analysis of the evolution of increased competitive ability hypothesis: genetic-based trait variation and herbivory resistance trade-offs. Biol. Invasions 20, 2647–2660 (2018).Article 

Google Scholar 
65.Shelby, N. et al. No difference in the competitive ability of introduced and native Trifolium provenances when grown with soil biota from their introduced and native ranges. AoB Plants 8, plw016 (2016).PubMed 
PubMed Central 
Article 

Google Scholar 
66.Harvey, J. A., Bukovinszky, T. & van der Putten, W. H. Interactions between invasive plants and insect herbivores: a plea for a multitrophic perspective. Biol. Conserv. 143, 2251–2259 (2010).Article 

Google Scholar 
67.Allen, W. J. et al. Multitrophic enemy release of invasive Phragmites australis and its introduced herbivores in North America. Biol. Invasions 17, 3419–3432 (2015).Article 

Google Scholar 
68.Kim, T. N. & Underwood, N. Plant neighborhood effects on herbivory: damage is both density and frequency dependent. Ecology 96, 1431–1437 (2015).PubMed 
Article 

Google Scholar 
69.Bartomeus, I., Vilà, M. & Santamaría, L. Contrasting effects of invasive plants in plant-pollinator networks. Oecologia 155, 761–770 (2008).ADS 
PubMed 
Article 
PubMed Central 

Google Scholar 
70.Lekberg, Y., Gibbons, S. M., Rosendahl, S. & Ramsey, P. W. Severe plant invasions can increase mycorrhizal fungal abundance and diversity. ISME J. 7, 1424–1433 (2013).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
71.Fernandez-Conradi, P., Jactel, H., Robin, C., Tack, A. J. M. & Castagneyrol, B. Fungi reduce preference and performance of insect herbivores on challenged plants. Ecology 99, 300–311 (2018).PubMed 
Article 
PubMed Central 

Google Scholar 
72.Reinhart, K. O. & Callaway, R. M. Soil biota and invasive plants. N. Phytol. 170, 445–457 (2006).Article 

Google Scholar 
73.Gioria, M. & Osborne, B. A. Resource competition in plant invasions: emerging patterns and research needs. Front. Plant Sci. 5, 501 (2014).PubMed 
PubMed Central 
Article 

Google Scholar 
74.Frost, C. M. et al. Using ecological network theory to predict biological invasions. Trends Ecol. Evol. 34, 831–843 (2019).PubMed 
Article 
PubMed Central 

Google Scholar 
75.Sauve, A. M. C., Thébault, E., Pocock, M. J. O. & Fontaine, C. How plants connect pollination and herbivory networks and their contribution to community stability. Ecology 97, 908–917 (2016).PubMed 
PubMed Central 

Google Scholar 
76.Pilosof, S., Porter, M. A., Pascual, M. & Kéfi, S. The multilayer nature of ecological networks. Nat. Ecol. Evol. 1, 0101 (2017).Article 

Google Scholar 
77.Weir, B. S., Turner, S. J., Silvester, W. B., Park, D. C. & Young, J. M. Unexpectedly diverse Mesorhizobium strains and Rhizobium leguminosarum nodulate native legume genera of New Zealand, while introduced legume weeds are nodulated by Bradyrhizobium species. Appl. Environ. Microbiol. 70, 5980–5987 (2004).CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
78.Cappuccino, N. & Carpenter, D. Invasive exotic plants suffer less herbivory than non-invasive exotic plants. Biol. Lett. 1, 435–438 (2005).PubMed 
PubMed Central 
Article 

Google Scholar 
79.Liu, H., Stiling, P. & Pemberton, R. W. Does enemy release matter for invasive plants? evidence from a comparison of insect herbivore damage among invasive, non-invasive and native congeners. Biol. Invasions 9, 773–781 (2007).Article 

Google Scholar 
80.Howell, C. Consolidated List of Environmental Weeds in New Zealand. DOC Research & Development Series 292 (Department of Conservation, 2008).81.Ghanizadeh, H. & Harrington, K. C. Weed management in New Zealand pastures. Agronomy 9, 448 (2019).CAS 
Article 

Google Scholar 
82.Kos, M., Tuijl, M. A. B., de Roo, J., Mulder, P. P. J. & Bezemer, T. M. Species-specific plant–soil feedback effects on aboveground plant-insect interactions. J. Ecol. 103, 904–914 (2015).CAS 
Article 

Google Scholar 
83.Heinen, R., Biere, A. & Bezemer, T. M. Plant traits shape soil legacy effects on individual plant–insect interactions. Oikos 129, 261–273 (2020).CAS 
Article 

Google Scholar 
84.Bezemer, T. M et al. Above‐and below‐ground herbivory effects on below‐ground plant–fungus interactions and plant–soil feedback responses. J. Ecol. 101, 325–333 (2013).Article 
CAS 

Google Scholar 
85.Heinze, J., Wacker, A. & Kulmatiski, A. Plant–soil feedback effects altered by aboveground herbivory explain plant species abundance in the landscape. Ecology 101, e03023 (2020).PubMed 
Article 
PubMed Central 

Google Scholar 
86.Müller, C. B., Adriaanse, I. C. T., Belshaw, R. & Godfray, H. C. J. The structure of an aphid-parasitoid community. J. Anim. Ecol. 68, 346–370 (1999).Article 

Google Scholar 
87.R Core Team. R: a language and environment for statistical computing. Version 3.6.1. R Foundation for Statistical Computing http://www.R-project.org (2019).88.Bates, D. et al. lme4: linear mixed-effects models using ‘Eigen’ and S4. R package version 1.1-21 http://CRAN.R-project.org/package=lme4 (2019).89.Lenth, R., Singmann, H., Love, J., Buerkner, P. & Herve, M. emmeans: estimated marginal means, aka least-squares means. R package version 1.3.5.1 http://CRAN.R-project.org/package=emmeans (2019).90.Dormann, C. F., Fruend, J. & Gruber, B. bipartite: visualising bipartite networks and calculating some (ecological) indices. R package version 2.13 http://CRAN.R-project.org/package=bipartite (2019). More

1.Gómez, C., White, J. C. & Wulder, M. A. Optical remotely sensed time series data for land cover classification: A review. ISPRS J. Photogramm. Remote. Sens. 116, 55–72 (2016).ADS 
Article 

Google Scholar 
2.Dale, V. H. The relationship between land-use change and climate change. Ecol. Appl. 7, 753–769 (1997).Article 

Google Scholar 
3.Dai, A. Increasing drought under global warming in observations and models. Nat. Clim. Chang. 3, 52–58 (2013).ADS 
Article 

Google Scholar 
4.Houghton, R. A. et al. Carbon emissions from land use and land-cover change. Biogeosciences 9, 5125–5142 (2012).ADS 
CAS 
Article 

Google Scholar 
5.Lambin, E. F. et al. The causes of land-use and land-cover change: Moving beyond the myths. Glob. Environ. Chang. 11, 261–269 (2001).Article 

Google Scholar 
6.Song, X. P. et al. Global land change from 1982 to 2016. Nature 560, 639–643 (2018).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
7.Grimm, N. B. et al. Global change and the ecology of cities. Science 319, 756–760 (2008).ADS 
CAS 
PubMed 
Article 
PubMed Central 

Google Scholar 
8.Utkin, L. V. et al. A weighted random survival forest. Knowl. Based Syst. 177, 136–144 (2019).Article 

Google Scholar 
9.Gašparović, M., Zrinjski, M. & Gudelj, M. Automatic cost-effective method for land cover classification (ALCC). Comput. Environ. Urban Syst. 76, 1–10 (2019).Article 

Google Scholar 
10.Hu, Y., Dong, Y. & Batunacun. ,. An automatic approach for land-change detection and land updates based on integrated NDVI timing analysis and the CVAPS method with GEE support. ISPRS J. Photogramm. Remote Sens. 146, 347–359 (2018).ADS 
Article 

Google Scholar 
11.ESA. Sentinel-2 Mission Requirements Document. Earth (2007).12.Main-Knorn, M. et al. Sen2Cor for Sentinel-2. In 3 (2017). https://doi.org/10.1117/12.2278218.13.Gong, P. et al. Stable classification with limited sample: Transferring a 30-m resolution sample set collected in 2015 to mapping 10-m resolution global land cover in 2017. Sci. Bull. 64, 370–373 (2019).Article 

Google Scholar 
14.Homer, C., Huang, C., Yang, L., Wylie, B. & Coan, M. Development of a 2001 National Land-Cover Database for the United States. Photogramm. Eng. Remote. Sens. 70, 829–840 (2004).Article 

Google Scholar 
15.Fry, J. A. et al. Completion of the 2006 national land cover database for the conterminous united states. Photogramm. Eng. Remote Sens. 77, 858–864 (2011).
Google Scholar 
16.Homer, C. et al. Completion of the 2011 national land cover database for the conterminous United States—Representing a decade of land cover change information. Photogramm. Eng. Remote Sens. 81, 345–354 (2015).
Google Scholar 
17.Yang, L. et al. A new generation of the United States National Land Cover Database: Requirements, research priorities, design, and implementation strategies. ISPRS J. Photogramm. Remote Sens. 146, 108–123 (2018).ADS 
Article 

Google Scholar 
18.Chen, J. et al. Global land cover mapping at 30 m resolution: A POK-based operational approach. ISPRS J. Photogramm. Remote Sens. 103, 7–27 (2015).ADS 
Article 

Google Scholar 
19.Hoang, T. T., Truong, V. T., Hayashi, M., Tadono, T. & Nasahara, K. N. New JAXA high-resolution land use/land cover map for Vietnam aiming for natural forest and plantation forest monitoring. Remote Sens. 12, 2707 (2020).20.Phan, D. C., Trung, T. H., Nasahara, K. N. & Tadono, T. JAXA high-resolution land use/land cover map for Central Vietnam in 2007 and 2017. Remote Sens. 10, 1406 (2018).ADS 
Article 

Google Scholar 
21.Nemani, R. Nasa earth exchange: Next generation earth science collaborative. ISPRS Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. XXXVIII-8/, 17–17 (2012).22.Mugiraneza, T., Nascetti, A. & Ban, Y. Continuous monitoring of urban land cover change trajectories with landsat time series and landtrendr-google earth engine cloud computing. Remote Sens. 12, 2883 (2020).23.Jin, Y., Liu, X., Yao, J., Zhang, X. & Zhang, H. Mapping the annual dynamics of cultivated land in typical area of the Middle-lower Yangtze plain using long time-series of Landsat images based on Google Earth Engine. Int. J. Remote Sens. 41, 1625–1644 (2020).ADS 
Article 

Google Scholar 
24.Hu, Y. & Hu, Y. Detecting forest disturbance and recovery in Primorsky Krai, Russia, using annual landsat time series and multi-source land cover products. Remote Sens. 12, 129 (2020).Article 

Google Scholar 
25.Miettinen, J., Shi, C. & Liew, S. C. 2015 Land cover map of Southeast Asia at 250 m spatial resolution. Remote Sens. Lett. 7, 701–710 (2016).Article 

Google Scholar 
26.Ghorbanian, A. et al. Improved land cover map of Iran using Sentinel imagery within Google Earth Engine and a novel automatic workflow for land cover classification using migrated training samples. ISPRS J. Photogramm. Remote Sens. 167, 276–288 (2020).ADS 
Article 

Google Scholar 
27.Huang, H. et al. The migration of training samples towards dynamic global land cover mapping. ISPRS J. Photogramm. Remote Sens. 161, 27–36 (2020).ADS 
Article 

Google Scholar 
28.Bagan, H. & Yamagata, Y. Land-cover change analysis in 50 global cities by using a combination of Landsat data and analysis of grid cells. Environ. Res. Lett. 9, 064015 (2014).29.Radoux, J. et al. Automated training sample extraction for global land cover mapping. Remote Sens. 6, 3965–3987 (2014).ADS 
Article 

Google Scholar 
30.Tran, H., Tran, T. & Kervyn, M. Dynamics of land cover/land use changes in the Mekong Delta, 1973–2011: A Remote sensing analysis of the Tran Van Thoi District, Ca Mau Province, Vietnam. Remote Sens. 7, 2899–2925 (2015).ADS 
Article 

Google Scholar 
31.Chi, V. K. et al. Land transitions in Northwest Vietnam: An integrated analysis of biophysical and socio-cultural factors. Hum. Ecol. 41, 37–50 (2013).Article 

Google Scholar 
32.Villamor, G. B., Catacutan, D. C., Truong, V. A. T. & Thi, L. D. Tree-cover transition in Northern Vietnam from a gender-specific land-use preferences perspective. Land Use Policy 61, 53–62 (2017).Article 

Google Scholar 
33.Truong, V. T. et al. JAXA annual forest cover maps for Vietnam during 2015–2018 Using ALOS-2/PALSAR-2 and auxiliary data. Remote Sens. 11, 2412 (2019).ADS 
Article 

Google Scholar 
34.And, R. D. of A. Vietnam’S Modified Submission on Refreence Levels for Redd+ Results Based Payments Under Unfccc. https://redd.unfccc.int/files/vietnam_frl_modified__submission_final_for_posting.pdf (2016).35.Xu, X., Jain, A. K. & Calvin, K. V. Quantifying the biophysical and socioeconomic drivers of changes in forest and agricultural land in South and Southeast Asia. Glob. Chang. Biol. 25, 2137–2151 (2019).ADS 
PubMed 
Article 
PubMed Central 

Google Scholar 
36.Hansen, M. C. et al. High-resolution global maps of 21st-century forest cover change. Science (80- ) 342, 850–853 (2013).ADS 
CAS 
Article 

Google Scholar 
37.Di Gregorio, A., and Jansen, L. J. M. Land Cover Classification System (LCCS): Classification Concepts and User Manual. Fao http://www.fao.org/3/x0596e/x0596e00.htm (2000).38.Van Thinh, T., Phan, D. C., Nasahara, K. N. & Tadono, T. How does land use/land cover map’s accuracy depend on number of classification classes? Sci. Online Lett. Atmos. 15, 28–31 (2019).
Google Scholar 
39.Talukdar, S. et al. Land-use land-cover classification by machine learning classifiers for satellite observations—a review. Remote Sens. 12, 1135 (2020).ADS 
Article 

Google Scholar 
40.Song, S., Gong, W., Zhu, B. & Huang, X. Wavelength selection and spectral discrimination for paddy rice, with laboratory measurements of hyperspectral leaf reflectance. ISPRS J. Photogramm. Remote Sens. 66, 672–682 (2011).ADS 
Article 

Google Scholar 
41.Szantoi, Z., Smith, S. E., Strona, G., Koh, L. P. & Wich, S. A. Mapping orangutan habitat and agricultural areas using Landsat OLI imagery augmented with unmanned aircraft system aerial photography. Int. J. Remote Sens. 38, 2231–2245 (2017).ADS 
Article 

Google Scholar 
42.Hill, R. A. & Thomson, A. G. Mapping woodland species composition and structure using airborne spectral and LiDAR data. Int. J. Remote Sens. 26, 3763–3779 (2005).ADS 
Article 

Google Scholar 
43.Kontgis, C., Schneider, A. & Ozdogan, M. Mapping rice paddy extent and intensification in the Vietnamese Mekong River Delta with dense time stacks of Landsat data. Remote Sens. Environ. 169, 255–269 (2015).ADS 
Article 

Google Scholar 
44.Kontgis, C. et al. Monitoring peri-urbanization in the greater Ho Chi Minh City metropolitan area. Appl. Geogr. 53, 377–388 (2014).Article 

Google Scholar 
45.D’Amour, C. B. et al. Future urban land expansion and implications for global croplands. Proc. Natl. Acad. Sci. USA 114, 8939–8944 (2017).Article 
CAS 

Google Scholar 
46.Ha, T. V., Tuohy, M., Irwin, M. & Tuan, P. V. Monitoring and mapping rural urbanization and land use changes using Landsat data in the northeast subtropical region of Vietnam. Egypt. J. Remote Sens. Sp. Sci. 23, 11–19 (2020).
Google Scholar 
47.Nhan, T. Q., Van Ly, L. T. & Tan, L. V. How much do rice farmers earn from their crops? Evidence from a rice-exporting country. J. Agric. Stud. 8, 302 (2020).
Google Scholar 
48.Truong, T. D. & Do, L. H. Mangrove forests and aquaculture in the Mekong river delta. Land Use Policy 73, 20–28 (2018).Article 

Google Scholar 
49.Lam-Dao, N., Pham-Bach, V., Nguyen-Thanh, M., Pham-Thi, M.-T. & Hoang-Phi, P. Change detection of land use and riverbank in Mekong Delta, Vietnam using time series remotely sensed data. J. Resour. Ecol. 2, 370–374 (2011).
Google Scholar 
50.Ha, T. T. P., van Dijk, H. & Visser, L. Impacts of changes in mangrove forest management practices on forest accessibility and livelihood: A case study in mangrove-shrimp farming system in Ca Mau Province, Mekong Delta, Vietnam. Land Use Policy 36, 89–101 (2014).Article 

Google Scholar 
51.Le, T. N., Bregt, A. K., van Halsema, G. E., Hellegers, P. J. G. J. & Nguyen, L. D. Interplay between land-use dynamics and changes in hydrological regime in the Vietnamese Mekong Delta. Land Use Policy 73, 269–280 (2018).Article 

Google Scholar 
52.Khuc, Q. V., Tran, B. Q., Meyfroidt, P. & Paschke, M. W. Drivers of deforestation and forest degradation in Vietnam: An exploratory analysis at the national level. For. Policy Econ. 90, 128–141 (2018).Article 

Google Scholar 
53.Meyfroidt, P., Vu, T. P. & Hoang, V. A. Trajectories of deforestation, coffee expansion and displacement of shifting cultivation in the Central Highlands of Vietnam. Glob. Environ. Chang. 23, 1187–1198 (2013).Article 

Google Scholar 
54.Ngo-Duc, T., Kieu, C., Thatcher, M., Nguyen-Le, D. & Phan-Van, T. Climate projections for Vietnam based on regional climate models. Clim. Res. 60, 199–213 (2014).Article 

Google Scholar 
55.Lindesay, J. A. et al. International geosphere-biosphere programme/international global atmospheric chemistry SAFARI-92 field experiment: Background and overview. J. Geophys. Res. Atmos. 101, 23521–23530 (1996).ADS 
CAS 
Article 

Google Scholar 
56.Klemas, V. V., Dobson, J. E., Ferguson, R. L. & Haddad, K. D. A coastal land cover classification system for the NOAA Coastwatch Change Analysis Project. J. Coast. Res. 9, 862–872 (1993).
Google Scholar 
57.Saah, D. et al. Primitives as building blocks for constructing land cover maps. Int. J. Appl. Earth Obs. Geoinf. 85, 101979 (2020).Article 

Google Scholar 
58.Keys, R. G. Cubic convolution interpolation for digital image processing. IEEE Trans. Acoust. 29, 1153–1160 (1981).MathSciNet 
MATH 
Article 

Google Scholar 
59.Farr, T. G. et al. The shuttle radar topography mission. Rev. Geophys. 45, RG2004 (2007).60.Filipponi, F. Sentinel-1 GRD preprocessing workflow. Proceedings 18, 11 (2019).Article 

Google Scholar 
61.Soenen, S. A., Peddle, D. R. & Coburn, C. A. SCS+C: A modified sun-canopy-sensor topographic correction in forested terrain. IEEE Trans. Geosci. Remote Sens. 43, 2148–2159 (2005).ADS 
Article 

Google Scholar 
62.Saleous, N. & Kutler, J. LEDAPS Calibration, Reflectance, Atmospheric Correction Preprocessing Code, Version 2 Summary: Data Citation: Model Product Description: Oak Ridge National Laboratory Distributed Active Archive Center http://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=1146 (2012) 10.3334/ORNLDAAC/1146.63.Vermote, E., Justice, C., Claverie, M. & Franch, B. Preliminary analysis of the performance of the Landsat 8/OLI land surface reflectance product. Remote Sens. Environ. 185, 46–56 (2016).ADS 
PubMed 
Article 
PubMed Central 

Google Scholar 
64.Foga, S. et al. Cloud detection algorithm comparison and validation for operational Landsat data products. Remote Sens. Environ. 194, 379–390 (2017).ADS 
Article 

Google Scholar 
65.Louis, J. et al. Sentinel-2 Global Surface Reflectance Level-2a Product Generated with Sen2Cor. In International Geoscience and Remote Sensing Symposium (IGARSS) 8522–8525 (2019). https://doi.org/10.1109/IGARSS.2019.8898540.66.Roy, D. P. et al. A general method to normalize Landsat reflectance data to nadir BRDF adjusted reflectance. Remote Sens. Environ. 176, 255–271 (2016).ADS 
Article 

Google Scholar 
67.Roy, D. P. et al. Examination of Sentinel-2A multi-spectral instrument (MSI) reflectance anisotropy and the suitability of a general method to normalize MSI reflectance to nadir BRDF adjusted reflectance. Remote Sens. Environ. 199, 25–38 (2017).ADS 
Article 

Google Scholar 
68.Lee, J. S., Ainsworth, T. L., Wang, Y. & Chen, K. S. Polarimetric SAR speckle filtering and the extended sigma filter. IEEE Trans. Geosci. Remote Sens. 53, 1150–1160 (2015).ADS 
Article 

Google Scholar 
69.Roy, D. P. et al. Characterization of Landsat-7 to Landsat-8 reflective wavelength and normalized difference vegetation index continuity. Remote Sens. Environ. 185, 57–70 (2016).ADS 
PubMed 
Article 
PubMed Central 

Google Scholar 
70.Claverie, M. et al. The Harmonized Landsat and Sentinel-2 surface reflectance data set. Remote Sens. Environ. 219, 145–161 (2018).ADS 
Article 

Google Scholar 
71.Flood, N. Seasonal composite landsat TM/ETM+ Images using the medoid (a multi-dimensional median). Remote Sens. 5, 6481–6500 (2013).ADS 
Article 

Google Scholar 
72.Kaufman, Y. J. & Tanré, D. Atmospherically Resistant Vegetation Index (ARVI) for EOS-MODIS. IEEE Trans. Geosci. Remote Sens. 30, 261–270 (1992).ADS 
Article 

Google Scholar 
73.Saah, D. et al. Land cover mapping in data scarce environments: Challenges and opportunities. Front. Environ. Sci. 7, 150 (2019).Article 

Google Scholar 
74.Li, C., Wang, J., Wang, L., Hu, L. & Gong, P. Comparison of classification algorithms and training sample sizes in urban land classification with landsat thematic mapper imagery. Remote Sens. 6, 964–983 (2014).ADS 
Article 

Google Scholar 
75.Richards, J. A. Remote sensing digital image analysis: An introduction. Remote Sensing Digital Image Analysis: An Introduction vol. 9783642300 (2013).76.Kruse, F. A. et al. The spectral image processing system (SIPS)-interactive visualization and analysis of imaging spectrometer data. Remote Sens. Environ. 44, 145–163 (1993).ADS 
Article 

Google Scholar 
77.Breiman, L. Random forests. Mach. Learn. 45, 5–32 (2001).MATH 
Article 

Google Scholar 
78.He, Y., Lee, E. & Warner, T. A. A time series of annual land use and land cover maps of China from 1982 to 2013 generated using AVHRR GIMMS NDVI3g data. Remote Sens. Environ. 199, 201–217 (2017).ADS 
Article 

Google Scholar 
79.Zhao, H. & Chen, X. Use of normalized difference bareness index in quickly mapping bare areas from TM/ETM+. In International Geoscience and Remote Sensing Symposium (IGARSS) vol. 3 1666–1668 (IEEE, 2005).80.Baloloy, A. B., Blanco, A. C., Raymund Rhommel, R. R. C. & Nadaoka, K. Development and application of a new mangrove vegetation index (MVI) for rapid and accurate mangrove mapping. ISPRS J. Photogramm. Remote Sens. 166, 95–117 (2020).ADS 
Article 

Google Scholar 
81.García, M. J. L. & Caselles, V. Mapping burns and natural reforestation using thematic mapper data. Geocarto Int. 6, 31–37 (1991).Article 

Google Scholar 
82.Wright, C. & Gallant, A. Improved wetland remote sensing in Yellowstone National Park using classification trees to combine TM imagery and ancillary environmental data. Remote Sens. Environ. 107, 582–605 (2007).ADS 
Article 

Google Scholar 
83.Hutchinson, C. F. Techniques for combining Landsat and ancillary data for digital classification improvement. Photogramm. Eng. Remote Sens. 48, 123–130 (1982).
Google Scholar 
84.Tadono, T. et al. Precise global DEM generation by ALOS PRISM. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. II–4, 71–76 (2014).Article 

Google Scholar 
85.Jokar Arsanjani, J., Zipf, A., Mooney, P. & Helbich, M. OpenStreetMap in GIScience. OpenStreetMap in GIScience: Experiences, Research, Applications (2015). https://doi.org/10.1007/978-3-319-14280-7.86.Open Development Mekong. OpenDevelopmentMekong. East-West Management Institute https://data.opendevelopmentmekong.net/organization/6f37a27d-2790-4b9a-8570-a36cb1d8108f?res_format=KML (2015).87.Truong, V. T. et al. JAXA annual forest cover maps for Vietnam during 2015–2018 Using ALOS-2/PALSAR-2 and auxiliary data. Remote Sens. 11 (2019).88.Olofsson, P. et al. Good practices for estimating area and assessing accuracy of land change. Remote Sens. Environ. 148, 42–57 (2014).ADS 
Article 

Google Scholar 
89.Congalton, R. G. A review of assessing the accuracy of classifications of remotely sensed data. Remote Sens. Environ. 37, 35–46 (1991).ADS 
Article 

Google Scholar 
90.Schmidt, M. The Sankey diagram in energy and material flow management—part II: Methodology and current applications. J. Ind. Ecol. 12, 173–185 (2008).Article 

Google Scholar 
91.Zha, Y., Gao, J. & Ni, S. Use of normalized difference built-up index in automatically mapping urban areas from TM imagery. Int. J. Remote Sens. 24, 583–594 (2003).ADS 
Article 

Google Scholar 
92.Lacaux, J. P., Tourre, Y. M., Vignolles, C., Ndione, J. A. & Lafaye, M. Classification of ponds from high-spatial resolution remote sensing: Application to Rift Valley Fever epidemics in Senegal. Remote Sens. Environ. 106, 66–74 (2007).ADS 
Article 

Google Scholar 
93.Tucker, C. J. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sens. Environ. 8, 127–150 (1979).ADS 
Article 

Google Scholar 
94.McFeeters, S. K. The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features. Int. J. Remote Sens. 17, 1425–1432 (1996).ADS 
Article 

Google Scholar 
95.Huete, A. R. A soil-adjusted vegetation index (SAVI). Remote Sens. Environ. 25, 295–309 (1988).ADS 
Article 

Google Scholar 
96.Penuelas, J., Baret, F. & Filella, I. Semi-empirical indices to assess carotenoids/chlorophyll a ratio from leaf spectral reflectance. Photosynthetica 31, 221–230 (1995).CAS 

Google Scholar 
97.Jayamanna, S., Kawamura, M. & Tsujiko, Y. Relation between social and environmental conditions in colombo, sri lanka and the urban index estimated by satellite remote sensing data. Int. Arch. Photogram. Remote Sens. 31, 321–326 (1996).
Google Scholar 
98.Shen, L. & Li, C. Water body extraction from Landsat ETM+ imagery using adaboost algorithm. In 2010 18th International Conference on Geoinformatics, Geoinformatics 2010 (2010). https://doi.org/10.1109/GEOINFORMATICS.2010.5567762.99.Perry, C. R. & Lautenschlager, L. F. Functional equivalence of spectral vegetation indices. Remote Sens. Environ. 14, 169–182 (1984).ADS 
Article 

Google Scholar 
100.As-syakur, A. R., Adnyana, I. W. S., Arthana, I. W. & Nuarsa, I. W. Enhanced built-UP and bareness index (EBBI) for mapping built-UP and bare land in an urban area. Remote Sens. 4, 2957–2970 (2012).ADS 
Article 

Google Scholar 
101.Liu, H. Q. & Huete, A. Feedback based modification of the NDVI to minimize canopy background and atmospheric noise. IEEE Trans. Geosci. Remote Sens. 33, 457–465 (1995).ADS 
Article 

Google Scholar 
102.Gitelson, A. A., Viña, A., Ciganda, V., Rundquist, D. C. & Arkebauer, T. J. Remote estimation of canopy chlorophyll content in crops. Geophys. Res. Lett. 32, 1–4 (2005).Article 
CAS 

Google Scholar  More

Our monetary compensation programme for poor rural communities in Mexico’s Sierra de Manantlán Biosphere Reserve encouraged them to forgo exploitation of their natural resources to provide ecosystem services for the city of Colima (see Nature 591, 178; 2021). But focusing solely on the monetary value of ecosystems isn’t enough.The National Forestry Commission of Mexico and the Fund for Natural Protected Areas have made compensatory payments of almost US$1 million to communities in the Cerro Grande region of the reserve since 2003, backed since 2013 by a local fiduciary fund of voluntary contributions from Colima’s citizens. However, the voluntary payments are minimal because most people don’t acknowledge the importance of the mountain forest that provides 90% of their water.As well as monetary schemes, the cutural traditions and the land-tenure rights of communal organizations must be recognized. They should be empowered to draw up contracts between owners of natural resources and urban beneficiaries that will promote their common social, economic and livelihood interests. To increase productivity and family income for impoverished small landowners, payment for ecosystem services could be implemented by using diverse marketing approaches that include sustainable agroforestry and livestock production. More

Insect culturePhyllotreta armoraciae beetles were reared on potted Brassica juncea cultivar “Bau Sin” plants or on potted Brassica rapa cultivar “Yo Tsai Sum” plants (Known-You Seed Co., Ltd) in mesh cages (Bugdorm, MegaView Science Co., Ltd) in a controlled environment chamber at 24 °C, 60% relative humidity, and a 16-h photoperiod. Food plants were cultivated in a growth chamber at 24 °C, 55% relative humidity, and a 14-h photoperiod. Beetles were provided with 3–4-week-old plants once per week, and plants with eggs were kept separately for larval development. Larvae were allowed to pupate in the soil, and after 3 weeks, the soil with pupae was transferred into plastic boxes (9 L, Lock&Lock) until the new generation of beetles emerged.RNA isolation, RNA sequencing, and de novo transcriptome assemblyTotal RNA was extracted from dissected foregut, midgut, hindgut, and Malpighian tubule tissue of newly emerged P. armoraciae beetles that were reared on B. juncea using the innuPREP RNA Mini Kit (Analytik Jena). Tissues from at least ten beetles were pooled per sample. RNA integrity was verified using an Agilent Technologies 2100 Bioanalyzer with the RNA 6000 Nano Kit (Agilent Technologies). RNA quantity was determined using a Nanodrop ND-1000 spectrophotometer (PEQlab Biotechnologie GmbH). One set of RNA samples was sequenced by GATC Biotech on the HiSeq 2500 System from Illumina in Rapid Run mode, using the paired-end (2 × 125 bp) read technology at a depth of 15–25 million reads for each sample. For a second set of samples consisting of four biological replicates per tissue, we additionally performed an on-column DNA digestion with the innuPREP DNase I Digest Kit (Analytik Jena) according to the manufacturer’s instructions. RNA samples were poly(A)-enriched, fragmented, and sequenced at the Max Planck Genome Centre Cologne on the HiSeq 3000 Sequencing System from Illumina, using the paired-end (2 × 150 bp) read technology at a depth of 22 million reads for each sample. Sequencing reads were filtered to remove bad-quality reads based on fastq file scores and trimmed based on read length using CLC Genomics Workbench software version 10.1. With a randomly sampled set of 420 million reads from the two sets of sequencing data, a transcriptome was assembled de novo with the following parameters: nucleotide mismatch cost = 1; insertion = deletion costs = 2; length fraction = 0.6; similarity = 0.9. Conflicts among the individual bases were resolved by voting for the base with the highest frequency. After removing contigs shorter than 250 bp, the final assembly contained 36,445 contigs with an N50 contig size of 2115 bp.Identification of coleopteran MFS transportersWe predicted a protein dataset for P. armoraciae by translating each contig of the gut and Malpighian tubule-specific transcriptome into all six reading frames. After removing sequences shorter than 50 amino acids, we submitted the protein dataset (267,568 sequences) to the TransAAP hosted at the TransportDB 2.0 web portal35. This initial annotation predicted a total of 1401 putative transporter sequences and revealed the MFS and the ABC transporters to be the largest transporter families (Supplementary Data 1). We focused on MFS transporters, which are classified into >80 families53. We used one protein sequence from each MFS family as a query to search candidate MFS transporters in the protein dataset from P. armoraciae using Blastp (E value threshold of 10−5), and assigned each candidate to an MFS family based on sequence similarity to transporter sequences deposited in TCDB. Additional candidates were identified by repeating the search procedure with an extended dataset including the candidate MFS transporters from P. armoraciae. The number of TMDs for each candidate was predicted using the TMHMM Server v.2.054. Partial sequences encoding less than six predicted TMDs were removed from the dataset. The same strategy was used to identify putative MFS transporters in protein datasets that were predicted from the genomes of Leptinotarsa decemlineata, genome annotations v.0.5.355, Anoplophora glabripennis, assembly Agla_1.056, and Tribolium castaneum, assembly Tcas3.057, respectively. The predicted protein sequences are provided in Supplementary Data 2.Digital gene expression analysisDigital gene expression analysis of putative MFS transporters identified in the P. armoraciae transcriptome was carried out using CLC Genomics Workbench v.10.1 by mapping the Illumina reads from the second set of samples onto the reference transcriptome, and counting the reads to estimate gene expression levels. For the cloned MFS genes, complete open-reading frames (ORFs) were used as reference sequences for mapping. For read alignment, we used the following parameters: nucleotide mismatch cost = 2; insertion = deletion costs = 3; length fraction = 0.6; similarity fraction = 0.9; maximum number of hits for a read = 15. Each pair of reads was counted as two. Biases in the sequence datasets and different transcript sizes were corrected using the TPM (transcripts per kilobase million) normalization method to obtain correct estimates for relative expression levels between samples.Phylogenetic analyses of coleopteran MFS transportersWe inferred the lineage-specific diversification patterns of putative MFS transporters from P. armoraciae, L. decemlineata, A. glabripennis, and T. castaneum in phylogenetic analyses with two different datasets, one containing all identified MFS transporters (867 sequences), the other containing a subset of putative sugar porters (120 sequences) from the above four species and 35 sugar porters from C. populi52. The corresponding protein sequences were aligned using the MUSCLE algorithm58 implemented in MEGA 7 with default parameters. The alignments were trimmed manually and the best substitution models were determined using ProtTest 3.4.259. Maximum-likelihood phylogenetic trees were constructed in IQ-TREE version 1.6.060 using the VT + G + F substitution model with 1000 ultrafast bootstrap replicates for the full dataset, and the LG + G + F substitution model with 1000 bootstrap replicates for the subset of putative sugar porters.Identification and sequencing of candidate transportersBased on our phylogenetic analysis, we selected the largest clade of putative MFS transporters that was specifically expanded in P. armoraciae for further studies. Transcriptome analyses revealed the presence of a pseudogene (PaMFS9-ps) that shares between 43 and 95% nucleotide sequence identity with members of the focal clade. The protein encoded by this pseudogene is predicted to possess only two transmembrane domains due to a premature stop codon caused by frameshift mutations in the coding sequence. To obtain the full-length ORFs of partial transcripts, we synthesized 5′- and 3′-rapid amplification of cDNA (complementary DNA) ends (RACE)–cDNA using the SMARTerRACE cDNA Amplification Kit (Clontech) and performed 5′- and 3′-RACE-PCR according to the manufacturer’s protocols (Clontech). All full-length ORFs were cloned into the pCR™4-TOPO® TA vector (Thermo Fisher Scientific) for sequence verification.Tissue-specific expression of candidate transportersWe used quantitative PCR (qPCR) to analyze the expression of candidate transporter genes in the foregut, midgut, hindgut, Malpighian tubules, and other tissues of 1-day-old adult P. armoraciae beetles (n = 4 biological replicates, each with two technical replicates), respectively. In addition, the expression of candidate transporter genes was analyzed in the proximal, central, and distal Malpighian tubule regions (Supplementary Fig. 2d) dissected from 4-day-old adult P. armoraciae beetles (n = 3 biological replicates, each with two technical replicates). Total RNA was extracted using the InnuPrep RNA Mini Kit (Analytik Jena), treated with TURBO DNase (Thermo Fisher Scientific), and purified using the RNeasy MinElute Cleanup Kit (Qiagen). First-strand cDNA was synthesized using the Verso cDNA Synthesis Kit (Thermo Fisher Scientific) using a 3:1 mixture (v/v) of random hexamer and oligo dT primers according to the manufacturer’s protocol. Quantitative PCR was performed in optical 96-well plates (Bio-Rad) on a Bio-Rad CFX Connect Real-Time System using the Absolute Blue qPCR SYBR Green Kit (Thermo Fisher Scientific). The PCR program was as follows: 95 °C for 15 min, then 40 cycles at 95 °C for 15 s, 57 °C for 30 s, and 72 °C for 30 s, followed by a melt cycle from 55 to 95 °C in 0.5 s increments. Primers (Supplementary Data 4) were designed using Primer3web version 4.1.0. We verified the amplification specificity by sequencing each PCR product after cloning into the pCR™4-TOPO® TA vector (Thermo Fisher Scientific) and by melting-curve analyses. Primer efficiencies were calculated using a cDNA template dilution series. Gene expression was normalized to the expression level of eukaryotic initiation factor 4A (eIF4A), which showed the lowest variability across tissues among four tested reference genes (Supplementary Table 4).Expression of candidate transporters in insect cellsFor protein expression, we cloned each ORF without stop codon into the pIEx-4 expression vector (Novagen) in frame with the vector-encoded carboxy-terminal 6× His-tag and sequenced the resulting constructs. Primer sequences are listed in Supplementary Data 4. One construct of each candidate gene was used for transfection of High Five™ insect cells (Gibco) cultured in Express Five® SFM medium (Gibco) supplemented with 20 mM glutamine (Gibco) and 50 μg/mL gentamicin (Gibco). Confluent insect cells were diluted 1:5, dispensed in 500 μL aliquots into 24-well plates, and incubated at 27 °C. The next day, we transfected the cells using FuGENE HD Transfection Reagent (Promega) according to the manufacturer’s protocol. Cells treated with transfection reagent only were used as a negative control. After 48 h, we harvested the cells for Western blotting and uptake assays.Western blottingTo confirm protein expression, transfected insect cells were washed twice with phosphate-buffered saline (PBS; pH 7.4), collected by centrifugation, and resuspended in hypotonic buffer (20 mM Tris-HCl (pH 7.5), 5 mM EDTA, 1 mM dithiothreitol, 0.1% benzonase nuclease (Merck Millipore) (v/v), and protease inhibitors (cOmplete Mini, ETDA-free, Roche Diagnostics GmbH)). After incubation on ice for 10 min, the samples were frozen in liquid nitrogen, thawed, and centrifuged (16,000 × g for 15 min at 4 °C). The resulting cell pellet was resuspended in hypotonic buffer and used for Western blotting using horseradish peroxidase-conjugated anti-His antibody (1:10,000; Novex, Life technologies).Glucoside uptake assays with transfected insect cellsCells were washed with PBS (pH 5.5) by pipetting and incubated with different glucoside substrates at 200 µM in PBS (pH 5.5) for 1 h at 27 °C. Assays were performed with substrate mixtures containing seven different glucosinolates (2-propenyl glucosinolate (Roth), 4-methylsulfinylbutyl glucosinolate (Phytoplan), 4-methylthiobutyl glucosinolate (Phytoplan), 2-phenylethyl glucosinolate (Phytoplan), benzyl glucosinolate (Phytoplan), 4-hydroxybenzyl glucosinolate (isolated from Sinapis alba seeds61), and I3M glucosinolate (Phytoplan)), or five other plant glucosides (salicin (Sigma-Aldrich), linamarin (BIOZOL), dhurrin (Roth), catalpol (Sigma-Aldrich), and aucubin (Sigma-Aldrich)). After incubation, cells were washed three times with ice-cold PBS (pH 5.5) by pipetting, collected in 300 μL 80% (v/v) methanol, frozen in liquid nitrogen, thawed, and centrifuged at 3220 × g for 10 min at 4 °C. The supernatant was dried by vacuum centrifugation, dissolved in ultrapure water, and analyzed by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). All glucoside uptake assays were performed in triplicates. The amount of each substrate in transporter-expressing cells was compared with that detected in control cells. Transporters were considered active when the average glucoside amounts detected in transporter-expressing cells were at least twofold higher than those detected in control cells.Cloning of PaGTR1 into the pNB1u vector and cRNA synthesisWe amplified the ORF of PaGTR1 without stop codon by PCR using uracil-containing primers (Supplementary Data 4). The 3′ primer was designed to encode a human influenza hemagglutinin tag to enable the detection of recombinant protein by Western blotting if necessary. The pNB1u vector was digested overnight at 37 °C with PacI and Nt.BbvCI (New England Biolabs) to generate 8-nt overhangs. One microliter gel-purified PCR product (100 ng/µL) was combined with 1 µL gel-purified vector (50 ng/µL), 1 U USER enzyme, 2 μL 5× PCR reaction buffer and 5 μL H2O, incubated at 37 °C for 25 min, followed by 25 min at room temperature. After the transformation of chemically competent E. coli cells, colonies containing the appropriate insert were identified by Sanger sequencing. The DNA template for complementary RNA (cRNA) synthesis was amplified by PCR from the X. laevis expression construct using pNB1uf/r primers (Supplementary Data 4) and cRNA was synthesized using the mMESSAGE mMACHINE™ T7 Transcription Kit (Invitrogen) according to the manufacturer’s manual.Biochemical characterization of PaGTR1 in Xenopus oocytesThe cRNA concentration was adjusted to 800 ng/μL with RNase-free water for oocyte injection. Xenopus laevis oocytes (Ecocyte Bioscience) were injected with 50 nL containing 40 ng cRNA or with 50 nL pure water as a control using a Drummond NANOJECT II (Drummond Scientific Company) or a Nanoliter 2010 Injector (World Precision Instruments).Injected oocytes were incubated in Kulori buffer (90 mM NaCl, 1 mM KCl, 1 mM MgCl2, 1 mM CaCl2, and 5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, pH 7.4) supplemented with 50 μg/mL gentamicin at 16 °C for 3 days until assaying. All transport assays were performed at room temperature. Injected oocytes were pre-incubated for 5 min in Kulori buffer (90 mM NaCl, 1 mM KCl, 1 mM MgCl2, 1 mM CaCl2, and 5 mM 2-(N-morpholino)ethanesulfonic acid (MES), pH 6.0) before they were transferred into the same buffer containing the substrate(s). To determine the substrate preference of PaGTR1, we incubated oocytes in Kulori buffer (pH 6.0) containing an equimolar mixture of 2-propenyl glucosinolate, 4-methylsulfinylbutyl glucosinolate, 4-methylthiobutyl glucosinolate, 2-phenylethyl glucosinolate, benzyl glucosinolate, 4-hydroxybenzyl glucosinolate, and I3M glucosinolate, each at 200 µM, for 1 h.The pH dependency of glucosinolate transport was determined by incubating the injected oocytes with 100 μM I3M glucosinolate for 10 min in Kulori buffer adjusted to different pH values. The effects of different ionophores on glucosinolate transport was studied by incubating the PaGTR1-expressing oocytes in Kulori buffer at pH 6.0 containing either 20 μM carbonyl cyanide m-chlorophenyl hydrazone (H+ ionophore, Sigma-Aldrich), 20 μM nigericin (K+/H+ exchanger, Abcam), or 20 μM valinomycin (K+ ionophore, Abcam) for 15 min. Afterwards, we incubated the oocytes in Kulori buffer containing 100 μM I3M glucosinolate and the corresponding ionophore for 10 min. Assays performed with oocytes incubated in Kulori buffer without any ionophore served as a control.The time course of I3M glucosinolate uptake was analyzed by incubating oocytes with 100 μM I3M glucosinolate in Kulori buffer (pH 6.0) for 5, 10, 20, 30, 45, 60, 120, 180, and 240 min, respectively. The apparent Km value of PaGTR1 for I3M glucosinolate was determined by incubating injected oocytes in Kulori buffer (pH 6.0) for 10 min with different substrate concentrations. The Km value was calculated by nonlinear regression analysis in SigmaPlot 14.0 (Systat Software Inc.).Each assay consisted of 14–15 oocytes and was stopped by washing oocytes four times with Kulori buffer. Afterwards, 12 of the washed oocytes were distributed into four Eppendorf tubes, with three oocytes per tube and immediately homogenized in 100 µL of 50% (v/v) methanol. After centrifugation (21,380 × g or 16,000 × g for 15 min), the supernatant was incubated at −20 °C for at least 1 h to precipitate proteins, which were pelleted by centrifugation (21,380 × g or 16,000 × g for 15 min). Finally, 60 μL sample was diluted with 120 μL ultrapure water, filtered through a 0.22 μm PVDF-based filter plate (Merck Millipore), and analyzed by LC-MS/MS. The glucosinolate concentration in oocytes was calculated by assuming an oocyte volume of 1 μL62.LC-MS/MSGlucosinolates were quantified by LC-MS/MS using an Agilent 1200 HPLC system connected to an API3200 tandem mass spectrometer (AB SCIEX). Separation was achieved on an EC 250/4.6 NUCLEODUR Sphinx RP column (250 mm × 4.6 mm, 5 μm; Macherey-Nagel) using a binary solvent system consisting of 0.2% (v/v) formic acid in water (A) and acetonitrile (B), with a flow rate of 1 mL/min at 25 °C. The elution gradient was as follows: 0–1 min, 1.5% B; 1–6 min, 1.5–5% B; 6–8 min, 5–7% B; 8–18 min, 7–21% B; 18–23 min, 21–29% B; 23–23.1 min, 29–100% B; 23.1–24 min, 100% B; 24–24.1 min, 100 to 1.5% B; 24.1–28 min, 1.5% B. Glucosinolates were detected in negative ionization mode. The ion spray voltage was set to −4500 V. Gas temperature was set to 700 °C, curtain gas to 20 psi, collision gas to 10, nebulizing gas to 70 psi, and drying gas to 60 psi. Nonhost glucosides were quantified using an Agilent 1200 HPLC system connected to an API5000 tandem mass spectrometer (AB SCIEX). Separation was achieved on an Agilent XDB-C18 column (5 cm × 4.6 mm, 1.8 μm) using a binary solvent system consisting of 0.05% (v/v) formic acid in water (A) and acetonitrile (B) with a flow rate of 1.1 mL/min at 25 °C. The elution gradient was as follows: 0–0.5 min, 5% B; 0.5–2.5 min, 5–31% B; 2.5–2.52 min, 31–100% B; 2.52–3.5 min, 100% B; 3.5–3.51 min, 100 to 5% B; 3.51–6 min, 5% B. Compounds were detected in negative ionization mode with ion spray voltage set to −4500 V. The gas temperature was set to 700 °C, curtain gas to 30 psi, collision gas to 6, and both nebulizing gas and drying gas to 60 psi. Multiple reaction monitoring (MRM) was used to monitor the transitions from precursor ion to product ion for each compound (Supplementary Table 5). Compounds were quantified using external standard curves. Analyst Software 1.6 Build 3773 (AB Sciex) was used for data acquisition and processing.Samples from the pH dependency experiment were analyzed by LC-MS/MS using an Advance UHPLC system (Bruker) connected to an EVOQ Elite TripleQuad mass spectrometer (Bruker) equipped with an electrospray ion source. Separation was achieved on a Kinetex 1.7u XB-C18 column (100 mm × 2.1 mm, 1.7 µm, 100 Å, Phenomenex) using a binary solvent system consisting of 0.05 % (v/v) formic acid in water (A) and acetonitrile with 0.05% (v/v) formic acid (B), with a flow rate of 0.4 mL/min at 40 °C. The elution gradient was as follows: 0–0.2 min, 2% B; 0.2–1.8 min, 2–30% B; 1.8–2.5 min, 30–100% B; 2.5–2.8 min, 100% B; 2.8–2.9 min, 100 to 2% B; and 2.9–4.0 min, 2% B. Glucosinolates were detected in negative ionization mode. The instrument parameters were optimized by infusion experiments with pure standards. The ion spray voltage was set to −4000 V. Cone temperature was set to 350 °C, cone gas to 20 psi, heated probe temperature to 200 °C, and probe gas flow to 50 psi. Nebulizing gas was set to 60 psi, and collision gas to 1.6 mTorr. MRM parameters are provided in Supplementary Table 5. Bruker MS Workstation software (Version 8.2.1, Bruker) was used for data acquisition and processing of glucosinolates. All other samples from experiments using the oocyte expression system were analyzed using the LC-MS/MS method described above for insect cell-based assays. The concentrations of all glucosinolates in the substrate preference assay were determined using external standard curves. Assays performed to characterize I3M glucosinolate transport were quantified using 2-propenyl glucosinolate as an internal standard.Double-stranded RNA synthesisWe synthesized seven different double-stranded RNAs (dsRNAs) between 120 and 298 bp in length, one specific for each PaGTR1/2/3/5/6/7/8, respectively, and a 223-bp fragment of the inducible metalloproteinase inhibitor (IMPI) from the greater wax moth Galleria mellonella (AY330624.1) (dsIMPI) using the T7 RiboMAX™ Express RNAi System (Promega). In silico off-target prediction was done by searches of all possible 21 mers of both RNA strands against the local P. armoraciae transcriptome database allowing for two mismatches. Except for PaGTR5, the prediction did not find any off-target towards putative transporter genes in the transcriptome (Supplementary Data 5).Function of PaGTR1 in vivoTo analyze the function of PaGTR1, we injected newly emerged adult P. armoraciae beetles (reared on B. juncea) with 100 nL ultrapure water containing 80 ng of dsPaGTR1 or 80 ng of dsIMPI, respectively, using a Nanoliter 2010 Injector (World Precision Instruments). Injected beetles were provided with detached leaves of 3–4-week-old B. juncea plants and kept in plastic containers with moistened tissue in the laboratory under ambient conditions. Four days after dsRNA injection, we collected dsIMPI- and dsPaGTR1-injected beetles for gene expression analysis (n = 5 replicates, three beetles per replicate) and glucosinolate analysis (n = 10 replicates, three beetles per replicate), respectively. The remaining beetles were used for a sequestration experiment with Arabidopsis thaliana Col-0 (Arabidopsis) plants that had been cultivated in a growth chamber at 21 °C, 55% relative humidity, and a 10-h photoperiod. To compare the accumulation and excretion of ingested glucosinolates in dsIMPI- and dsPaGTR1-injected beetles, we fed beetles with detached Arabidopsis leaves in Petri dishes (60 mm diameter) that contained 50 μL of ultrapure water and were sealed with parafilm to prevent leaf wilting (n = 10 replicates, five beetles per replicate). Feeding assays were performed in the laboratory under ambient conditions, and leaves were exchanged every day for five consecutive days. To estimate how much the beetles fed, we determined the weight of each leaf before and after feeding. Feces were collected every day using 100 µL of ultrapure water per replicate, combined with 300 μL of pure methanol in a 1.5 mL Eppendorf tube and dried by vacuum centrifugation. Feces samples were then homogenized in 200 μL of 80% (v/v) methanol using metal beads (2.4 mm diameter, Askubal) in a tissue lyzer (Qiagen) for 1 min at 30 Hz. After feeding, adults were starved for one day, weighed, frozen in liquid nitrogen, and stored at −20 °C until glucosinolate extraction. Beetle samples were homogenized using a plastic pestle in 200 μL 80% (v/v) methanol. All samples were then extracted with 1 mL 80% (v/v) methanol containing 25 μM 4-hydroxybenzyl glucosinolate as an internal standard. After centrifugation (16,000 × g for 10 min), glucosinolates were extracted from the supernatant, converted to desulfo-glucosinolates, and analyzed by high-performance liquid chromatography coupled with diode array detection (HPLC-DAD) as described below. The glucosinolate content in adults or feces was calculated in nanomole per adult, respectively.To confirm the specificity of PaGTR1 knockdown, we analyzed the effect of dsPaGTR1 injection on the expression of PaGTR2, PaGTR3, PaGTR9, and PaGTR10. PaGTR9 and PaGTR10 share the highest nucleotide sequence similarity (69% sequence identity) with PaGTR1. PaGTR2 and PaGTR3 expression was analyzed because the recombinant transporters also used I3M glucosinolate as a substrate. RNA extraction, purification, cDNA synthesis, and qPCR were performed as described above.Function of PaGTR5/6/7/8 in vivoTo analyze the function of PaGTR5/6/7/8, we injected newly emerged adult P. armoraciae beetles that had fed for 2 days on B. juncea leaves with 100 nL ultrapure water containing 100 ng of dsIMPI or each 100 ng of dsPaGTR5/6/7/8 using a Nanoliter 2010 Injector (World Precision Instruments). A subset of the dsRNA-injected beetles was fed with detached leaves of Arabidopsis plants in plastic containers with moistened tissue for gene expression analysis (n = 6 replicates, two beetles per replicate). The remaining dsRNA-injected beetles were used for a sequestration experiment with Arabidopsis plants to compare the accumulation and excretion of previously stored and ingested glucosinolates in dsIMPI- and dsPaGTR5/6/7/8-injected beetles. We fed the injected beetles with detached Arabidopsis leaves in Petri dishes (60 mm diameter) that contained 30 μL of ultrapure water and were sealed with parafilm to prevent leaf wilting (n = 10 replicates, six beetles per replicate). Feeding assays were performed in the laboratory under ambient conditions, and leaves were exchanged every day for six consecutive days. To estimate how much the beetles fed, we determined the weight of each leaf before and after feeding. Starting from the second day, feces were collected as above for 5 days. After drying by vacuum centrifugation, feces were homogenized in 1 mL of 80% (v/v) methanol containing 25 μM 4-hydroxybenzyl glucosinolate as an internal standard using metal beads (2.4 mm diameter, Askubal) in a tissue lyzer (Qiagen) for 1 min at 30 Hz. Fed adults were starved for 1 day, weighed, and frozen in liquid nitrogen until glucosinolate extraction. Beetle samples were homogenized using a plastic pestle in 200 μL of 80% (v/v) methanol and then extracted with 1 mL of 80% (v/v) methanol containing 25 μM 4-hydroxybenzyl glucosinolate as an internal standard. After centrifugation (16,000 × g for 10 min), glucosinolates were extracted from the supernatant, converted to desulfo-glucosinolates, and analyzed by HPLC-DAD as described below.To confirm the specificity of PaGTR5/6/7/8 knockdown, we analyzed the effect of dsPaGTR5/6/7/8 injection on the expression of PaGTR9 and PaGTR10, which share the highest nucleotide sequence similarity (67–69% sequence identity) with PaGTR5/6/7/8. RNA extraction, purification, cDNA synthesis, and qPCR were performed as described above.Function of PaGTR2/3 in vivoTo analyze the functions of PaGTR2 and PaGTR3, we injected third instar larvae of P. armoraciae (reared on B. rapa) with 100 nL ultrapure water containing 80 ng of dsPaGTR2 and 80 ng of dsPaGTR3 (dsPaGTR2/3) or 80 ng of dsIMPI, respectively, using a Nanoliter 2010 Injector (World Precision Instruments). Injected larvae were provided with detached B. rapa petioles and kept in plastic tubes with moistened tissue in the laboratory under ambient conditions until pupation. Newly emerged adults were again injected with dsRNAs and provided with detached B. rapa leaves. Three days after the second dsRNA injection, we collected dsIMPI- and dsPaGTR2/3-injected beetles for gene expression analysis (n = 6 replicates, two beetles per replicate) and glucosinolate analysis (n = 10 replicates, three beetles per replicate), respectively. The remaining beetles were used for a feeding experiment with Arabidopsis. Each replicate consisted of one Arabidopsis leaf and three beetles that were placed in a Petri dish (60 mm diameter) with 50 μL of ultrapure water (n = 12–13 replicates). Each leaf was photographed before and after feeding to determine the consumed leaf area. Fed leaves were frozen in liquid nitrogen, freeze-dried, and homogenized using metal beads (2.4 mm diameter, Askubal) in a tissue lyzer (Qiagen) for 2 min at 30 Hz. Fed beetles were starved for 1 day, weighed, frozen in liquid nitrogen, and stored at −20 °C until glucosinolate extraction. Beetle samples were homogenized using a plastic pestle in 200 μL 80% (v/v) methanol. All samples were extracted with 1 mL 80% (v/v) methanol containing 25 μM 1-methylethyl glucosinolate (extracted from Sisymbrium officinale seeds) as an internal standard. Extracts were applied to DEAE-Sephadex A-25 (Sigma-Aldrich) columns in 96-well filter plates (Nunc) that were preconditioned with 1 mL of ultrapure water and 2 × 500 μL of 80% (v/v) methanol. After loading 900 μL of extract, the columns were washed with 500 μL of 80% (v/v) methanol, followed by 2 × 1 mL of ultrapure water. To adjust the pH condition, 500 μL of 0.02 M MES buffer (pH 5.2) was added to each column. After adding 30 μL of Helix pomatia sulfatase solution to each column and overnight incubation at room temperature, desulfo-glucosinolates were eluted using 300 or 500 μL of ultrapure water into 96-deep well plates (Nunc). Samples were analyzed by desulfo-HPLC-DAD as described below. The glucosinolate content in adults and fed leaves was calculated in nanomole per adult and nanomole per cm2 leaf, respectively. The ingested glucosinolate amount was calculated based on the ingested leaf area and the corresponding leaf glucosinolate content. To elucidate which proportion of the ingested glucosinolates was sequestered, we expressed the glucosinolate amount detected in the beetles relative to the ingested glucosinolate amount from the leaves, which was set to 100%.HPLC-DADSamples were analyzed by HPLC on an Agilent Technologies HP1100 Series instrument equipped with a photodiode array detector. After injection of 100 μL of each sample, separation was achieved on an EC 250/4.6 NUCLEODUR Sphinx RP column (250 mm × 4.6 mm, 5 μm; Macherey-Nagel; samples of the RNA interference (RNAi) experiments of PaGTR1, and PaGTR2/3) or an EC 250/4.6 NUCLEODUR 100-5 C18ec column (250 mm × 4.6 mm, 5 μm; Macherey-Nagel; samples of the RNAi experiments of PaGTR5/6/7/8, and PaGTR2/3) using a binary solvent system consisting of ultrapure water (A) and acetonitrile (B), with a flow rate of 1 mL/min. The elution gradient was as follows: 0–1 min, 1.5% B; 1–6 min, 1.5–5% B; 6–8 min, 5–7% B; 8–18 min, 7–21% B; 18–23 min, 21–29% B; 23–23.5 min, 29–100% B; 23.5–26 min, 100% B; 26–26.1 min, 100 to 1.5% B; and 26.1–31 min, 1.5% B. The eluent was monitored by diode array detection between 190 and 360 nm. Desulfo-glucosinolates were identified based on a comparison of retention times and absorption spectra with those of known standards63. The glucosinolate content in each sample was calculated from the peak areas at 229 nm relative to the peak area of the internal standard using relative response factors64.Glucosinolate concentration in the hemolymph of adult P. armoraciae
Hemolymph was collected from 7-day-old adult P. armoraciae reared on B. juncea by cutting off an abdominal leg and collecting the extruding droplet using glass capillaries (0.5 µL, Hirschmann® minicaps®) (n = 6 replicates, 50 beetles per replicate). The capillaries were marked with 1 mm intervals (corresponding to 15.6 nL) to estimate the volume of collected hemolymph. The hemolymph was diluted in 500 µL of 90% (v/v) methanol, homogenized using metal beads (2.4 mm diameter, Askubal) in a tissue lyzer (Qiagen) for 1 min at 30 Hz, and boiled for 5 min at 95 °C. After two centrifugation steps (13,000 × g for 10 min each), the supernatant was dried by vacuum centrifugation, dissolved in 50 µL 50% (v/v) methanol, diluted in ultrapure water, and analyzed by LC-MS/MS.Morphology of the Malpighian tubule system of P. armoraciae
To investigate the structure of the Malpighian tubule system, we dissected the gut and Malpighian tubules of 4-day-old P. armoraciae adults in PBS (pH 6.0) under a stereomicroscope. The tracheae that attach Malpighian tubules to the gut were removed using fine forceps to release the tubules. Pictures were taken with a Canon EOS 600D camera.pH of hemolymph and excretion fluid of isolated Malpighian tubules of P. armoraciae
The pH of the hemolymph and Malpighian tubule excretion fluid of 5-day-old P. armoraciae adults was assessed using the pH indicator bromothymol blue (Alfa Aesar). Hemolymph was collected by cutting off an abdominal leg and collecting the extruding droplet using a pipette (n = 3 replicates, six to ten beetles per replicate). Excretion fluid was collected from dissected Malpighian tubules that were incubated in saline A as described for the Ramsay assay (n = 4 replicates, one tubule per replicate). Hemolymph and excretion fluid were immediately mixed with the same volume of 0.16% (w/v) bromothymol blue dissolved in 10% (v/v) ethanol, respectively, under water-saturated paraffin oil in a Sylgard-coated petri dish. The resulting color of the droplet was compared with those of citric acid–Na2HPO4 buffer solutions ranging from pH 5.2 to 6.6 in 0.2 increments mixed with 0.16% (w/v) bromothymol blue.Fate of plant glucosides injected in P. armoraciae beetlesTo analyze the fate of plant glucosides in vivo, we injected 100 nL of an equimolar mixture of 2-propenyl glucosinolate, 4-hydroxybenzyl glucosinolate, linamarin, salicin, and catalpol, each at 10 mM, and 0.15% (w/v) amaranth into the hemolymph of 2-day-old adult P. armoraciae (reared on B. rapa). One group of beetles was sampled 30 min after injection by freezing beetles in liquid nitrogen (n = 10 replicates, five beetles per replicate). The remaining beetles were fed with detached leaves of Arabidopsis in Petri dishes (60 mm diameter) in the laboratory under ambient conditions (n = 10 replicates, five beetles per replicate). We added 30 μL of ultrapure water to each Petri dish and sealed them with parafilm to prevent leaf wilting. After 1 day, we sampled the beetles as described above (n = 10 replicates, five beetles per replicate). Feces were collected using 100 µL of ultrapure water per replicate and combined with 300 μL of pure methanol in a 1.5 mL Eppendorf tube. All samples were stored at −20 °C until extraction. Beetle and feces samples were homogenized as described in the RNAi experiment. After centrifugation (16,000 × g for 10 min), the supernatant was dried by vacuum centrifugation, dissolved in 100 µL of ultrapure water, and analyzed by LC-MS/MS. The glucoside content in adults or feces was calculated as nanomole per adult, respectively.Ramsay assayTo analyze the excretion of plant glucosides in situ, we performed Ramsay assays65 with dissected Malpighian tubules from 4- to 5-day-old P. armoraciae adults reared on B. rapa (Supplementary Fig. 6). Malpighian tubules were dissected in saline A (100 mM NaCl, 8.6 mM KCl, 2 mM CaCl2, 8.5 mM MgCl2, 4 mM NaH2PO4, 4 mM NaHCO3, 24 mM glucose, 10 mM proline, 25 mM 3-(N-morpholino)propanesulfonic acid (MOPS), pH 6.0)66. Single tubules were transferred into a 10 μL droplet of saline B (60 mM NaCl, 10.3 mM KCl, 2.4 mM CaCl2, 10.2 mM MgCl2, 4.8 mM NaH2PO4, 4.8 mM NaHCO3, 28.8 mM glucose, 12 mM proline, 30 mM MOPS, 1 mM cyclic AMP, pH 6.0) under water-saturated paraffin oil in a Sylgard-coated petri dish. The proximal end of the tubule was pulled out of the droplet, attached to a metal pin, and cut using a glass capillary to allow the collection of excretion fluid. To start the assay, we added 2 μL of an equimolar glucoside mixture consisting of 2-propenyl glucosinolate, 4-methylsulfinylbutyl glucosinolate, 4-hydroxybenzyl glucosinolate, 2-phenylethyl glucosinolate, I3M glucosinolate, linamarin, salicin, and catalpol, each at 40 mM, and 0.6% (w/v) amaranth) to saline B. After 2–3 h, we collected the excretion fluid and 2 μL of the bathing droplet in 300 μL of 80% (v/v) methanol, respectively. The Malpighian tubule was washed three times in ~15 mL of saline A and afterwards transferred into 300 μL of 80% (v/v) methanol. All samples were stored at −20 °C until extraction. Malpighian tubule samples were homogenized using a plastic pestle. After centrifugation (16,000 × g for 10 min), the supernatant was dried by vacuum centrifugation, dissolved in 70 µL ultrapure water, and analyzed by LC-MS/MS. Out of 36 assays, 25 assays were excluded because no excretion fluid was visible within 2–3 h.Statistical analysisNo statistical methods were used to predetermine sample size. Statistical analyses were conducted in R 3.5.167 or in SigmaPlot 14.0 or 14.5 (Systat Software Inc.). Two groups were compared by two-tailed Student’s t test, Mann–Whitney U test, or the method of generalized least squares68, depending on the variance homogeneity and the normality of residuals. Data of the Ramsay assays were compared by paired two-tailed Student’s t test using FDR (false discovery rate)-corrected P values69. Three or more groups were compared by one-way analysis of variance, followed by post hoc multiple comparisons test, or the method of generalized least squares68. If necessary, data were transformed prior to analysis. For comparisons using the method of generalized least squares, we applied the varIdent variance structure, which allows each group to have a different variance. The P value was obtained by comparing models with and without explanatory variables using a likelihood ratio test70. Significant differences between groups were revealed with factor level reductions71. Information about data transformation, statistical methods, and results of the statistical analyses are summarized in Supplementary Tables 2 and 3 and Supplementary Data 6.Reporting summaryFurther information on research design is available in the Nature Research Reporting Summary linked to this article. More