Ecology

Himalayan marmot habitat analysisThe environmental factors including temperature, vegetation and elevation are the key drivers for the wildlife in alpine ecosystems32. Specific landform attributes such as slope and elevation and vegetation cover affect the population and burrowing of rodents33. For example, rodent burrows in the Western Usambara Mountains in Tanzania were only found at an elevation of above 1600 m33. However, the Himalayan marmot seems to prefer to inhabit areas with low elevation and high land surface temperature34. In this study, the data showed that 76.25% of the Himalayan marmots were found in areas with elevation values of 3400–4600 m. The majority of marmots were found in areas with slopes of 5–20° and vegetation cover higher than 60%. Most marmots were found in alpine meadows, a few were found in temperate grasslands and alpine grasslands, and none were found in other grassland types.Preliminary statistical analysis of vegetation cover, grass type, vegetation type, and Himalayan marmot distribution sample sites obtained using spatial geographic information technology revealed that the meadow grassland areas with lush grass growth, more dominant plants, and abundant food had more marmots. When the vegetation cover reached 0.60–1.00, the number of marmot distribution sample sites was the highest. Dense grass is an ideal habitat and provides concealment for Himalayan marmots, and the abundant plant types provide sufficient food for marmots. In contrast, no marmots were distributed in the alpine scrub, coniferous forest, and alpine snow/ice covered areas where vegetation growth was poor, vegetation cover was low, and food was relatively scarce. Moreover, 70.24% of Himalayan marmots were found in alpine meadows with a wide variety of plant species, including Poaceae, Cyperaceae, and grasses. This finding indicated that alpine meadows are more suitable for Himalayan marmots and have more advantageous habitat conditions compared with other grassland types. The elevation of alpine meadows is 3236–5126 m, and the vegetation is mainly meadows with simple vegetation structure, substantial vegetation cover and dense vegetation growth, and a wide variety of plants, rich food, soft grass, and good palatability. Therefore, alpine meadows provide good natural habitats and foraging sites for marmots.Habitat selection of large rodents is influenced by a combination of vegetation cover availability, food availability, and population density35. Vegetation cover is an important parameter that describes vegetation communities and ecosystems and is closely related to vegetation quantity and productivity. The quality of habitat vegetation is an important factor that affects the spatial distribution of plateau rodents. Both feeding and concealment depend on vegetation, and the height and cover of edible plants and vegetation suitable for concealment determine the choice of vegetation type by marmots. Thus, vegetation cover becomes an important factor for habitat selection by marmots. Different grassland types determine different plant conditions, and selection of different vegetation conditions can increase the chances of survival and improve the reproductive success of marmots; therefore, grassland type is an important ecological factor in habitat selection by marmots. A study showed that the ecological factors affecting habitat selection of Himalayan marmots are mainly topography, anthropogenic disturbance, and vegetation8. Another study concluded that habitat selection by Himalayan marmots is closely related to elements such as topography, landform, temperature, precipitation, and vegetation24.The functions of burrows’ physical parameters is to protect the Himalayan marmots from natural enemies and bad weather36. There is clearly influence of slope on habitat selection by marmots. When the slope is large, wind is strong, and burrows are not well hidden; this makes them difficult to defend against enemies, unsafe for survival, and not conducive to hibernation during winter. In addition, Himalayan marmots prefer to burrow on sunny aspect, because the temperature is suitable and the vegetation is lush, which is suitable for marmots to breed. Therefore, the number of marmot burrows gradually decreases with increasing slope and ubac. Although flat and low-lying areas with small slopes are good for marmots to create dens, rainwater will easily flow into the dens during summer rainfall, which will kill marmots. Therefore, a suitable slope and sunny aspect are also very important for habitat selection by marmots.Application of the predictive spatial distribution map of Himalayan marmots in Qinghai provincePlague surveillance is the main measure used for plague prevention and control in China. Although we have made many improvements in plague surveillance, the traditional method of dragnet surveillance still consumes a lot of human and material resources, is inefficient. The pasture area of Qinghai province is approximately 380,000 km2, and the identified natural plague focus is approximately 180,000 km2; therefore, there is still 200,000 km2 of pasture where the distribution of Himalayan marmots and plague have not been identified. Currently, RS technology is widely used in the fields of mapping and ecological surveillance18,19,21,22,37.Applications of RS technology in areas such as malaria, dengue, schistosomiasis and plague have been previously reported27,37. Using GIS combined with remotely sensed data, Proches Hieronimo et al. found that the presence of small mammals was positively influenced by elevation, whereas the presence of fleas was clearly influenced by land management features, and thus these observations have positive implications for plague surveillance27. In this study, RS technology combined with field validations were used to determine the distribution and areas of different types of grasslands in Qinghai province, and the average density of Himalayan marmot distribution in different types of grasslands. The high-, low-, and very low-density areas of Himalayan marmot distribution were identified. The soil map, vegetation map, administrative map, and marmot density statistics were merged to form the spatial data and attribute data basis for the information system to map the distribution of Himalayan marmot and determine the area of Himalayan marmot distribution. Generally speaking, the occurrence of human plague epidemic is closely related to the local animal plague epidemic2. However, a large part of the high-density distribution of Himalayan marmots is located in uninhabited areas and the areas are generally sparsely populated, which also indicates that we should reasonably allocate plague prevention and control resources to areas where human plague is most likely to occur to prevent the occurrence of human plague epidemics.Field validation for verificationThrough field validation and information from local farmers and herdsmen, we confirmed that Himalayan marmots inhabited 68 sample sites in Tongde, Zeku, Guinan, Xunhua, Haiyan, Ulan, Qilian, Hualong, and Huzhu counties. Among them, Tongde, Zeku, Guinan, Xunhua, Haiyan, Ulan, and Qilian counties have all historically experienced marmot plague outbreaks and can be considered as reliable natural plague foci38. The data from this field validation are consistent with the previous survey data and the epidemic history of the counties in Qinghai province39.MAE can better reflect the actual number of errors in prediction values; the smaller the MAE value, the higher the prediction accuracy. The MAE derived from the field validation data was 0.1331 and the prediction accuracy was 0.8669. The accuracy of the predicted Himalayan marmot spatial distribution reached 87%, which indicated that the predicted probability map of the Himalayan marmot spatial distribution can better predict the potential marmot distribution.The predicted spatial distribution map of Himalayan marmot in Qinghai province was then compared with environmental information such as elevation, vegetation, grass type, slope, and aspect of 352 field survey sites. The obtained RS data showed that the prediction results were excellent, and the predicted spatial distribution map of Himalayan marmot in Qinghai province was drawn with high accuracy. The prediction map visually reflects the different density distribution of Himalayan marmots; this allows us to optimize the settings and reasonable spatial layout of animal plague surveillance sites and improve surveillance efficiency.Application of marmot information collection system V3.0Marmot information collection system V3.0 was developed based on the “3S” technology standardizing the collection of surveillance data, and makes the management and analysis of information more convenient and faster. This study revolutionized the traditional method of considering plague-stricken counties as the plague foci, and effectively reduces the work intensity of operators and improves the data collection efficiency. In 2016 and 2017, we applied this system to the animal plague surveillance tasks in the plague-stricken counties of Haidong, Hainan, and Haibei in Qinghai province, and standardized the collection of provincial geographic location data of animal plague surveillance (data not shown). In 2018, we also applied this system in Wulan County, which frequently experiences plague, and achieved a good application effect (data not shown).In the next step, we will expand the pilot areas (mainly national and provincial plague surveillance sites), collect surveillance data from each surveillance site, continuously optimize and update the system, improve the efficiency of data analysis and utilization, detect the plague epidemic in marmot in a timely and accurate manner, correctly determine the epidemic trend of plague in marmots, and attempt to strictly prevent the plague from spreading to humans. We plan to use a new model of drone surveillance to create a multidimensional, three-dimensional, real-time big data plague surveillance information reporting system to enhance early plague warnings and prediction in Qinghai province and even in the country, which will be of positive practical significance to serve and guarantee the Belt and Road Initiative. These approaches are expected to provide new technical means for plague investigation and research, and to provide references for setting up plague surveillance programs and prediction for the natural Himalayan marmot plague focus in Qinghai province and the QTP. More

Gillooly, J. F., Brown, J. H., West, G. B., Savage, V. M. & Charnov, E. L. Effects of size and temperature on metabolic rate. Science 293, 2248–2251 (2001).Article 
CAS 

Google Scholar 
Davidson, E. A. & Janssens, I. A. Temperature sensitivity of soil carbon decomposition and feedbacks to climate change. Nature 440, 165–173 (2006).Article 
CAS 

Google Scholar 
Lopez-Urrutia, A., San Martin, E., Harris, R. P. & Irigoien, X. Scaling the metabolic balance of the oceans. Proc. Natl Acad. Sci. USA 103, 8739–8744 (2006).Article 
CAS 

Google Scholar 
Yvon-Durocher, G. et al. Reconciling the temperature dependence of respiration across timescales and ecosystem types. Nature 487, 472–476 (2012).Article 
CAS 

Google Scholar 
Crowther, T. W. et al. Quantifying global soil carbon losses in response to warming. Nature 540, 104–108 (2016).Article 
CAS 

Google Scholar 
Bar-On, Y. M., Phillips, R. & Milo, R. The biomass distribution on Earth. Proc. Natl Acad. Sci. USA 115, 6506–6511 (2018).Article 
CAS 

Google Scholar 
Rivkin, R. B. & Legendre, L. Biogenic carbon cycling in the upper ocean: effects of microbial respiration. Science 291, 2398–2400 (2001).Article 
CAS 

Google Scholar 
Friedlingstein, P. et al. Uncertainties in CMIP5 climate projections due to carbon cycle feedbacks. J. Clim. 27, 511–526 (2014).Article 

Google Scholar 
Smith, T. P. et al. Community-level respiration of prokaryotic microbes may rise with global warming. Nat. Commun. 10, 5124 (2019).Article 

Google Scholar 
Antwis, R. E. et al. Fifty important research questions in microbial ecology. FEMS Microbiol. Ecol. 93, fix044 (2017).Bardgett, R. D., Freeman, C. & Ostle, N. J. Microbial contributions to climate change through carbon cycle feedbacks. ISME J. 2, 805–814 (2008).Article 
CAS 

Google Scholar 
Enquist, B. J. et al. Scaling from traits to ecosystems: developing a general trait driver theory via integrating trait-based and metabolic scaling theories. Adv. Ecol. Res. 52, 249–318 (2015).Article 

Google Scholar 
Allen, A. P., Gillooly, J. F. & Brown, J. H. Linking the global carbon cycle to individual metabolism. Funct. Ecol. 19, 202–213 (2005).Article 

Google Scholar 
Schramski, J. R., Dell, A. I., Grady, J. M., Sibly, R. M. & Brown, J. H. Metabolic theory predicts whole-ecosystem properties. Proc. Natl Acad. Sci. USA 112, 2617–2622 (2015).Article 
CAS 

Google Scholar 
Alster, C. J., Koyama, A., Johnson, N. G., Wallenstein, M. D. & von Fischer, J. C. Temperature sensitivity of soil microbial communities: an application of macromolecular rate theory to microbial respiration. J. Geophys. Res. Biogeosci. 121, 1420–1433 (2016).Article 

Google Scholar 
Yvon-Durocher, G. et al. Five years of experimental warming increases the biodiversity and productivity of phytoplankton. PLoS Biol. 13, e1002324 (2015).Article 

Google Scholar 
Garzke, J., Connor, S. J., Sommer, U. & O’Connor, M. I. Trophic interactions modify the temperature dependence of community biomass and ecosystem function. PLoS Biol. 17, e2006806 (2019).Foster, K. R. & Bell, T. Competition, not cooperation, dominates interactions among culturable microbial species. Curr. Biol. 22, 1845–1850 (2012).Article 
CAS 

Google Scholar 
Coyte, K. Z., Schluter, J. & Foster, K. R. The ecology of the microbiome: networks, competition, and stability. Science 350, 663–666 (2015).Article 
CAS 

Google Scholar 
Machado, D. et al. Polarization of microbial communities between competitive and cooperative metabolism. Nat. Ecol. Evol. 5, 195–203 (2021).Article 

Google Scholar 
Bradford, M. A. et al. Cross-biome patterns in soil microbial respiration predictable from evolutionary theory on thermal adaptation. Nat. Ecol. Evol. 3, 223–231 (2019).Article 

Google Scholar 
Garcia-Martin, E. E., McNeill, S., Serret, P. & Leakey, R. J. G. Plankton metabolism and bacterial growth efficiency in offshore waters along a latitudinal transect between the UK and Svalbard. Deep Sea Res. I 92, 141–151 (2014).Article 
CAS 

Google Scholar 
Davidson, E. A., Richardson, A. D., Savage, K. E. & Hollinger, D. Y. A distinct seasonal pattern of the ratio of soil respiration to total ecosystem respiration in a spruce-dominated forest. Glob. Change Biol. 12, 230–239 (2006).Article 

Google Scholar 
Dutkiewicz, S., Follows, M. J. & Bragg, J. G. Modeling the coupling of ocean ecology and biogeochemistry. Glob. Biogeochem. Cycles 23, GB4017 (2009).Article 

Google Scholar 
Follows, M. J., Dutkiewicz, S., Ward, B. & Follett, C. in Microbial Ecology of the Oceans 3rd edn (eds Gasol, J. & Kirchman, D.) Ch. 12 (John Wiley, 2018).Letten, A. D. & Stouffer, D. B. The mechanistic basis for higher-order interactions and non-additivity in competitive communities. Ecol. Lett. 22, 423–436 (2019).Article 

Google Scholar 
Grilli, J., Barabás, G., Michalska-Smith, M. J. & Allesina, S. Higher-order interactions stabilize dynamics in competitive network models. Nature 548, 210–213 (2017).Article 
CAS 

Google Scholar 
Maynard, D. S., Crowther, T. W. & Bradford, M. A. Competitive network determines the direction of the diversity–function relationship. Proc. Natl Acad. Sci. USA 114, 11464–11469 (2017).Article 
CAS 

Google Scholar 
Fiegna, F., Moreno-Letelier, A., Bell, T. & Barraclough, T. G. Evolution of species interactions determines microbial community productivity in new environments. ISME J. 9, 1235–1245 (2015).Article 

Google Scholar 
Lawrence, D. et al. Species interactions alter evolutionary responses to a novel environment. PLoS Biol. 10, e1001330 (2012).Article 
CAS 

Google Scholar 
Harcombe, W. R., Chacón, J. M., Adamowicz, E. M., Chubiz, L. M. & Marx, C. J. Evolution of bidirectional costly mutualism from byproduct consumption. Proc. Natl Acad. Sci. USA 115, 12000–12004 (2018).Article 
CAS 

Google Scholar 
Goldford, J. E. et al. Emergent simplicity in microbial community assembly. Science 361, 469–474 (2018).Article 
CAS 

Google Scholar 
Yvon-Durocher, G. et al. Methane fluxes show consistent temperature dependence across microbial to ecosystem scales. Nature 507, 488–491 (2014).Article 
CAS 

Google Scholar 
Fox, J. W. & Harpole, W. S. Revealing how species loss affects ecosystem function: the trait-based price equation partition. Ecology 89, 269–279 (2008).Article 

Google Scholar 
Kontopoulos, D., Smith, T. P., Barraclough, T. G. & Pawar, S. Adaptive evolution shapes the present-day distribution of the thermal sensitivity of population growth rate. PLoS Biol. 18, e3000894 (2020).Article 
CAS 

Google Scholar 
Wilson, W. G. & Lundberg, P. Biodiversity and the Lotka–Volterra theory of species interactions: open systems and the distribution of logarithmic densities. Proc. R. Soc. Lond. B 271, 1977–1984 (2004).Article 

Google Scholar 
Rossberg, A. G. in Food Webs and Biodiversity 181–191 (John Wiley & Sons, 2013).Schloss, P. D. et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl. Environ. Microbiol. 75, 7537–7541 (2009).Article 
CAS 

Google Scholar 
Garcia, F. C., Bestion, E., Warfield, R. & Yvon-Durocher, G. Changes in temperature alter the relationship between biodiversity and ecosystem functioning. Proc. Natl Acad. Sci. USA 115, 10989–10999 (2018).Article 
CAS 

Google Scholar 
Padfield, D., O’Sullivan, H. & Pawar, S. rTPC and nls.multstart: a new pipeline to fit thermal performance curves in R. Methods Ecol. Evol. 12, 1138–1143 (2021).Article 

Google Scholar  More

Study areaThe Qinghai–Tibet Plateau (26°00′-39°47′N, 73°19′-104°47′E), one of the most important pastoral areas in the world, straddles the southwest regions of China, and it includes 244 counties, which belong to six provinces: Tibet, Qinghai, Xinjiang, Gansu, Sichuan, and Yunnan. It is characterized by rich natural grassland resources, including desert steppes, alpine steppes, and alpine meadows (Fig. 1a). The grassland areas account for over 56% of this region34. The grassland plays a vital role in providing regional and national animal husbandry products and fodder35, which enables the local herders to obtain almost all of the resources required for survival36. The grazing density distribution is extremely unbalanced (Fig. 1a) owing to the high spatial heterogeneity of economic development (Fig. 1b-1) and grassland production (Fig. 1b-2), resulting from the differences in resources and environmental factors37. Over the past few decades, there has been a significant change in the number of livestock animals, and the number of sheep exceeded 160 million by 2020. Therefore, it is urgent to obtain a high-resolution gridded grazing dataset for its evaluating spatiotemporal changes and coordinating the relationship between human beings and the grassland ecosystem.Fig. 1Location of the Qinghai–Tibet Plateau: (a) grassland type and distribution, and grazing density (GD) in 244 counties; (b) spatial heterogeneity of economic development (ED) and grassland production (GP) in 244 counties. GD, ED, and GP are represented by sheep unit per grassland area per county (SU/hm2), human footprint index per pixel (HF/pixel) per county, and net primary production per grassland area per county (gC/m2), respectively.Full size imageFig. 2Methodological framework for grazing spatialization.Full size imageMethodological frameworkWe developed a methodological framework for high-resolution gridded grazing dataset mapping. The framework mainly includes four parts: (i) identifying features affecting grazing, (ii) extracting theoretical suitable grazing areas, (iii) building grazing spatialization model, and (iv) correcting the grazing spatialization dataset. Each step is explained in more detail below (Fig. 2).Step 1: Identifying features affecting grazingGrazing activities are affected by the spatial heterogeneity of resources and environmental factors, regulated by the grazing behavior of herders and the foraging behavior of herds, and restricted by ecological protection policies. Therefore, the specific implications of the 14 influencing factors from the above four aspects are presented in Table 1. These factors are necessary for spatializing the county-level grazing data.Table 1 The identified features affecting grazing.Full size tableStep 2: Extracting theoretical suitable grazing areasThe decision tree approach38 was adopted to extract the theoretical suitable grazing areas for further grazing spatialization (step 2 in Fig. 2). First, the potential grazing area was identified according to the boundary of the grassland ecosystem, because grazing behavior only occurs in the grassland. Then, the unsuitable areas for grazing, i.e., extremely-high-altitude areas and areas adjacent to towns, were removed from the potential grazing area stepwise. The areas strictly prohibited for grazing, i.e., the core areas of national nature reserves39 within grassland areas, were also deemed unsuitable for grazing. Finally, the extracted areas were the theoretically suitable grazing areas.Step 3: Building grazing spatialization model(i) Extracting cross-scale feature (CSFs)In the traditional method, the spatial resolution of the training data (i.e., the average value at the administrative level) differs from that of the predicting data (i.e., the value at the pixel level), and the trained model can only capture the characteristics within the training data. However, the extreme value of the predicting data inevitably exceeds the range of the training data, which can result in underestimation in these parts40. To reduce these mismatches, we built an improved method for CSFs extraction (Fig. 2, first part of step 3).First, the census grazing data are simply distributed from county level to pixel level using the weight of the absolute disturbance (AD) index as Eq. (1). The AD index is measured by Mahalanobis distance using Eq. (2), which is calculated according to the deviation between the potential and observed normalized difference vegetation index (NDVI) values22. Second, the distributed grazing data are graded via the hierarchical clustering method, and the optimal number of the group can be determined using the Davies–Bouldin index (DBI)41 as Eq. (3), an index for evaluating the quality of clustering algorithm. The smaller the DBI, the smaller the distance within each group. Therefore, the DBI can be used to select the best similar values to minimize the deviation within each group. Finally, we can obtain the scope of the groups within each county using the above two steps and obtain the average value of all independent variables and the dependent variable accordingly. As expected, we can decompose the average value at the county level (traditional features in Fig. 2) into the average value at the group level (improved features in Fig. 2).$$S{U}_{i}=S{U}_{j}^{C}frac{{w}_{A{D}_{i}}}{{w}_{A{D}_{j}}}$$
(1)
where SUi and (S{U}_{j}^{C}) are the grazing value for pixel i and the census grazing value for county j; ({w}_{A{D}_{i}}) is the weight of the AD index for pixel i and ({w}_{A{D}_{j}}) represents the summed weight of the AD index values for all pixels in county j.$$begin{array}{cll}A{D}_{i} & = & sqrt{{({D}_{i}-u)}^{T}co{v}^{-1}({D}_{i}-u)}\ {D}_{i} & = & NDV{I}_{i}^{T}-NDV{I}_{i}^{P}end{array}$$
(2)
where ADi is the AD index for pixel i; the vector composed of its observed NDVI (left(NDV{I}_{i}^{T}right)) and potential NDVI (left(NDV{I}_{i}^{P}right)) time-series data could be considered as two points in the feature space for pixel i, and Di and u are the difference and the mean value of the vector, respectively; cov is the covariance matrix.$$DB{I}_{k}=frac{1}{k}{sum }_{x=1}^{k}ma{x}_{yne x}left(frac{overline{{a}_{x}}+overline{{a}_{y}}}{left|{delta }_{x}-{delta }_{y}right|}right)$$
(3)
where DBIk is the DBI coefficient when the cluster number is k; (overline{{a}_{x}}) and (overline{{a}_{y}}) are the average distances of the group xth and the group yth, respectively; δx and δy are the center distance of the group xth and the group yth, respectively.Different from the traditional method, our method can decompose features into multiple features using the grading AD index. The differences among counties will not be easily averaged out. Moreover, our method is less affected by scale mismatch and can be transferred to cross-scale modeling26.(ii) Building RF model with partitioningA single model cannot accurately obtain the variation information of the Qinghai–Tibet Plateau with high spatial heterogeneity. The partition model, a widely used method for estimating population distribution and others42,43, can be incorporated into the proposed model to improve its performance. The thresholds (0.43, 0.35 and 0.21 SU/hm2), determined according to the theoretical livestock carrying capacity (equation S1), were calculated and used to separate independent variables and dependent variable for each grassland types: alpine meadow, alpine steppe and alpine desert steppe (see Section 6.1 for details). Then, the RF models were established, and the training and testing samples were randomly divided in the proportion of 3:1. It is notable that transforming the response variable using natural log prior to RF model fitting is necessary to achieve higher prediction accuracies44. Finally, the independent variables at the pixel level were inputted into the two trained RF models, and the corresponding grid grazing dataset was output by combining the two results (Fig. 2, second part of step 3).(iii) Validating the accuracy of the methodsThe performance of the grazing spatialization model was evaluated through a comparison of the predicted value with census value26. Accuracy validation indexes, including coefficients of determination (R2), root mean square error (RMSE), and mean absolute error (MAE), were used to evaluate the performances of the proposed RF-based models (Table 2), as presented in Eq. (4).$$begin{array}{ccc}{R}^{2} & = & 1-frac{{sum }_{j=1}^{N}{left(S{U}_{j}^{C}-S{U}_{j}^{P}right)}^{2}}{{sum }_{j=1}^{N}{left(S{U}_{j}^{C}-overline{S{U}^{C}}right)}^{2}}\ RMSE & = & sqrt{frac{{sum }_{j=1}^{N}{left(S{U}_{j}^{C}-S{U}_{j}^{P}right)}^{2}}{N}}\ MAE & = & frac{{sum }_{j=1}^{N}| S{U}_{j}^{C}-S{U}_{j}^{P}| }{N}end{array}$$
(4)
where (S{U}_{j}^{C}) and (S{U}_{j}^{P}) are the census grazing value and the predicted grazing value for county j, respectively; (overline{S{U}^{C}}) is the average census data for all counties; and N is the number of all counties.Table 2 The proposed methods and their descriptions.Full size tableStep 4: Correcting grazing spatialization dataset(i) Correcting residuals of datasetCorrecting residuals is necessary to obtain datasets with higher accuracy45,46, because propagating the cross-scale relationship in the RF models will inevitably generate errors47. The residuals, calculated by the difference between the average census grazing and predicted grazing values at the administrative level, were used to calibrate the errors related to all pixels within this county. The revised dataset after residual correction is the final product provided in this study. The residual correction method is expressed by Eq. (5), and the process is shown in the fourth step in Fig. 2.$$S{U}_{i}^{RP}=S{U}_{i}^{P}+{R}_{j}$$
(5)
where (S{U}_{i}^{RP}) denotes the predicted grazing value revised by the residuals for pixel i, (S{U}_{i}^{P}) denotes the predicted grazing for pixel i, and Rj denotes the residuals calculated from the difference between census grazing and predicted grazing data for county j.(ii) Validating the accuracy of datasetTwo goodness-of-fit indexes were used to validate the consistency of spatial distribution and the temporal trend between predicted grazing data and census grazing data. Generally, the coefficient of determination (R2), defined in Eq. (4), is used to verify the consistency of spatial distribution, and the Nash–Sutcliffe efficiency (NSE, Eq. (6)) is used to verify the consistency of temporal trend. An index value closer to 1 corresponds to a more accurate dataset. Meanwhile, we also collected field grazing data from 56 sites to further validate the spatial accuracy of the dataset, and it measured using the R2 in Eq. (4).$$NSE=1-frac{{sum }_{t=1}^{T}{left(S{U}_{t}^{RP}-S{U}_{t}^{C}right)}^{2}}{{sum }_{t=1}^{T}{left(S{U}_{t}^{C}-overline{S{U}^{{C}^{{prime} }}}right)}^{2}}$$
(6)
where (S{U}_{t}^{RP}) and (S{U}_{t}^{C}) are the predicted grazing value revised by residuals and the census grazing value of all counties in year t, respectively; (overline{S{U}^{{C}^{{prime} }}}) is the average census grazing value of all years; and T is the number of time steps.(iii) Identifying uncertainties associated with datasetThe uncertainties associated with the dataset originate from the following two aspects: First, the unreasonableness of our method, owing to the errors related to cross-scale modeling or the inappropriate selection of influencing factors, is an important source of uncertainties. Second, the incompleteness of auxiliary variables also introduces uncertainties. In this instance, grassland-free areas are not accurately identified in some counties, but livestock animals are raised in these counties. These counties have no effective value for livestock density prediction. Overall, the uncertainties can be identified in terms of the mean relative error (MRE) in Eq. (7).$$MRE=frac{{sum }_{j=1}^{N}left|frac{S{U}_{j}^{C}-S{U}_{j}^{RP}}{S{U}_{j}^{C}}right|}{N}ast 100 % $$
(7)
where (S{U}_{j}^{C}) is the census grazing value for county j, (S{U}_{j}^{RP}) is the predicted grazing value revised by residuals for county j, and N is the number of counties.Data sourceCensus grazing data at county levelEight types of livestock, namely cattle, yaks, horses, donkeys, mules, camels, goats, and sheep, were considered according to the regional characteristics, and livestock stocking quantity at the end of year for each county can be determined from statistical yearbooks. However, the numbers of livestock at the county level for some years between 1982 and 2015 were not recorded. The missing data were indirectly approximated from city- or provincial-level data (e.g., interpolation using their temporal trends). Each type of livestock stocking quantity was converted into standard sheep unit (SU) according to the national standards using Eq. (8)48, namely the calculation of rangeland carrying capacity (NY/T 635-2015). Of the 244 counties of the Qinghai–Tibet Plateau, only 242 counties were considered, as the census grazing data for the other 2 counties were unavailable. The unit of grazing statistics data at the county level is defined as SU per county per year (SU·county−1·year−1).$$begin{array}{l}SU={N}_{sheep}+0.8times {N}_{goats}+5times {N}_{cattle}+5times {N}_{yaks+}+\ 6times {N}_{horses}+3times {N}_{donkeys}+6times {N}_{mules}+7times {N}_{camels}end{array}$$
(8)
where Nsheep, Ngoats, Ncattle, Nyaks, Nhorses, Ndonkeys, Nmules, Ncamels are the number of sheep, goats, cattle, yaks, horses, donkeys, mules, and camels at the year-end, respectively. SU denotes the standard sheep unit (SU·county−1·year−1).Data of grazing influencing factors at pixel levelThe types of features affecting grazing were obtained from the first step described in Methods, and the detailed information, such as original spatiotemporal resolution, format, and source, is shown in Table 3. The format (i.e., GeoTIFF), spatial resolution (i.e., 0.083°), and the number of rows and columns of the gridded features were leveraged to further produce a high-resolution grazing dataset.Table 3 Data source of grazing influence factors.Full size table More

Now consider a generalized model in which the species interactions are heterogeneous. A natural way of introducing heterogeneity in the system is by having a species diversify into subpopulations with different interaction strengths12,13,14,15. This way of modeling heterogeneity is useful as it can describe different kinds of heterogeneity. For example, the subpopulations could represent polymorphic traits that are genetically determined or result from plastic response to heterogeneous environments. A population could also be divided into local subpopulations in different spatial patches, which can migrate between patches and may face different local predators. We can also model different behavioral modes as subpopulations that, for instance, spend more time foraging for food or hiding from predators. We study several kinds of heterogeneity after we introduce a common mathematical framework. By studying these different scenarios using variants of the model, we show that our main results are not sensitive to the details of the model.We focus on the simple case where only the prey species splits into two types, (C_1) and (C_2), as illustrated in Fig. 1b. The situation is interesting when predator A consumes (C_1) more readily than predator B and B consumes (C_2) more readily than A (i.e., (a_1 / a_0 > b_1 / b_0) and (b_2 / b_0 > a_2 / a_0), which is equivalent to the condition that the nullclines of A and B cross, see section “Resources competition and nullcline analysis”). The arrows between (C_1) and (C_2) in Fig. 1b represent the exchange of individuals between the two subpopulations, which can happen by various mechanisms considered below. Such exchange as well as intraspecific competition between (C_1) and (C_2) result from the fact that the two prey types remain a single species.The system is now described by an enlarged Lotka-Volterra system with four variables, A, B, (C_1), and (C_2): $$begin{aligned} dot{A}&= varepsilon _A ,alpha _{A1} , A , C_1 + alpha _{A2} , A , C_2 – beta _A , A end{aligned}$$
(3a)
$$begin{aligned} dot{B}&= varepsilon _B , alpha _{B1} , B , C_1 + alpha _{B2} , B , C_2 – beta _B , B end{aligned}$$
(3b)
$$begin{aligned} dot{C_1}&= C_1 , (beta _C – alpha _{CC} , C)-alpha _{A1} , C_1 A-alpha _{B1} , C_1 B – sigma _1 , C_1 + sigma _2 , C_2 end{aligned}$$
(3c)
$$begin{aligned} dot{C_2}&= C_2 , (beta _C – alpha _{CC} , C) -alpha _{A2} , C_2 A -alpha _{B2} , C_2 B + sigma _1 , C_1 – sigma _2 , C_2 end{aligned}$$
(3d)
The parameters in these equations and their meanings are listed in Table 1. Here we assume that the prey types (C_1) and (C_2) have the same birth rate and intraspecific competition strength, but different interaction strengths with A and B. Note that (C_1) and (C_2) are connected by the (sigma _i) terms, which represent the exchange of individuals between these subpopulations through mechanisms studied below; these terms indicate a major difference between our model of a prey with intraspecific heterogeneity and other models of two prey species. For the convenience of analysis, we transform the variables (C_1) and (C_2) to another pair of variables C and (lambda), where (C equiv C_1 + C_2) is the total population of C as before, and (lambda equiv C_2 / (C_1 + C_2)) represents the composition of the population (Fig. 1c). After this transformation and rescaling of variables (described in “Methods”), the new dynamical system can be written as: $$begin{aligned} dot{A}&= A , big ( C , (a_1 (1-lambda ) + a_2 lambda ) – a_0 big ) end{aligned}$$
(4a)
$$begin{aligned} dot{B}&= B , big ( C , (b_1 (1-lambda ) + b_2 lambda ) – b_0 big ) end{aligned}$$
(4b)
$$begin{aligned} dot{C}&= C , big ( 1 – C – A (a_1 (1-lambda ) + a_2 lambda ) – B (b_1 (1-lambda ) + b_2 lambda ) big ) end{aligned}$$
(4c)
$$begin{aligned} dot{lambda }&= lambda (1-lambda ) , big ( A (a_1 – a_2) + B (b_1 – b_2) big ) + eta _1 (1-lambda ) – eta _2 lambda end{aligned}$$
(4d)
Here, (a_i) and (b_i) are the (rescaled) feeding rates of the predators on the prey type (C_i); (a_0) and (b_0) are the death rates of the predators as before; (eta _1) and (eta _2) are the exchange rates of the prey types (Table 1). The latter can be functions of other variables, representing different kinds of heterogeneous interactions that we study below. Notice that Eqs. (4a–4c) are equivalent to the homogeneous Eqs. (2a–2c) but with effective interaction strengths (a_text {eff} = (1-lambda ) , a_1 + lambda , a_2) and (b_text {eff} = (1-lambda ) , b_1 + lambda , b_2) that both depend on the prey composition (lambda) (Fig. 1c).Table 1 Model parameters (before/after rescaling) and their meanings.Full size tableThe variable (lambda) can be considered an internal degree of freedom within the C population. In all of the models we study below, (lambda) dynamically stabilizes to a special value (lambda ^*) (a bifurcation point), as shown in Fig. 3a. Accordingly, a new equilibrium point (P_N) appears (on the line (mathscr {L}) in Fig. 2), at which all three species coexist. For comparison, Fig. 3b shows the equilibrium points if (lambda) is held fixed at any other values, which all result in the exclusion of one of the predators. Thus, heterogeneous interactions give rise to a new coexistence phase (see Fig. 4 below) by bringing the prey composition (lambda) to the value (lambda ^*), instead of having to fine-tune the interaction strengths. The exact conditions for this new equilibrium to be stable are detailed in “Methods”.Figure 3(a) Time series of (lambda) for systems with each kind of heterogeneity. All three systems stabilize at the same (lambda ^*) value, which is the bifurcation point in panel (b). (b) Equilibrium population of each species (X = A), B, or C, with (lambda) held fixed at different values. Solid curves represent stable equilibria and dashed curves represent unstable equilibria (see Eq. (9) in “Methods”). The vertical dashed line is where (lambda = lambda ^*), which is also the bifurcation point. Notice that the equilibrium population of C is maximized at this point (for (a_1 > a_2) and (b_2 > b_1)). Parameters used here are ((a_0, a_1, a_2, b_0, b_1, b_2, rho , eta _1, eta _2, kappa ) = (0.25, 0.5, 0.2, 0.4, 0.2, 0.6, 0.5, 0.05, 0.05, 50)).Full size imageInherent heterogeneityWe first consider a scenario where individuals of the prey species are born as one of two types with a fixed ratio, such that a fraction (rho) of the newborns are (C_2) and ((1-rho )) are (C_1). This could describe dimorphic traits, such as the winged and wingless morphs in aphids12 or the horned and hornless morphs in beetles13. We call this “inherent” heterogeneity, because individuals are born with a certain type and cannot change in later stages of life. The prey type given at birth determines the individual’s interaction strength with the predators. This kind of heterogeneity can be described by Eq. (4d) with (eta _1 = rho (1-C)) and (eta _2 = (1-rho ) (1-C)) (see “Methods”).Figure 4Phase diagrams showing regions of parameter space identified by the stable equilibrium points. Yellow region represents (P_C) (predators A, B both extinct), red represents (P_A) (A excludes B), blue represents (P_B) (B excludes A), and green represents (P_N) (A, B coexist). The middle point (black dot) is where the preferences of the two predators are identical, (a_2/a_0=b_2/b_0) and (b_1/b_0=a_1/a_0). The coexistence phase appears in all three kinds of heterogeneity modeled here. (a–d) Inherent heterogeneity: Individuals of the prey population are born in two types with a fixed composition (rho). In the extreme cases of (rho = 0) and 1, the prey is homogeneous and there is no coexistence of the predators. (e–h) Reversible heterogeneity: Individual prey can switch types with fixed switching rates (eta _1) and (eta _2). As the switching rates increase, the coexistence region shrinks because the prey population becomes effectively homogeneous (the occasional green spots are numerical artifacts because the time to reach the equilibrium becomes long in this limit). (i–l) Adaptive heterogeneity: The switching rates (eta _i) dynamically adapt to the predator densities, so as to maximize the growth rate of the prey. As the sharpness (kappa) of the sigmoidal decision function is increased, the prey adapts more optimally and the region of coexistence expands. Parameters used here are ((a_0, a_1, b_0, b_2) = (0.3, 0.5, 0.4, 0.6)).Full size imageThe stable equilibrium of the system can be represented by phase diagrams that show the identities of the species at equilibrium. We plot these phase diagrams by varying the parameters (a_2) and (b_1) while keeping (a_1) and (b_2) constant. As shown in Fig. 4a–d, the equilibrium state depends on the parameter (rho). In the limit (rho = 0) or 1, we recover the homogeneous case because only one type of C is produced. The corresponding phase diagrams (Fig. 4a, d) contain only two phases where either of the predators is excluded, illustrating the competitive exclusion principle. For intermediate values of (rho), however, there is a new phase of coexistence that separates the two exclusion phases (Fig. 4b, c). There are two such regions of coexistence, which touch at a middle point and open toward the bottom left and upper right, respectively. The middle point is at ((a_2/a_0 = b_2/b_0, b_1/b_0 = a_1/a_0)), where the feeding preferences of the two predators are identical (hence their niches fully overlap). Towards the origin and the far upper right, the predators consume one type of C each (hence their niches separate). The coexistence region in the bottom left is where the feeding rates of the predators are the lowest overall. There can be a region (yellow) where both predators go extinct, if their primary prey type alone is not enough to sustain each predator. Increasing the productivity of the system by increasing the birth rate ((beta _C)) of the prey eliminates this extinction region, whereas lowering productivity causes the extinction region to take over the lower coexistence region. Because the existence and identity of the phases is determined by the configuration of the equilibrium points (Fig. 2, see also section “Mathematical methods”), the qualitative shape of the phase diagram is not sensitive to changes of parameter values.The new equilibrium is not only where the predators A and B can coexist, but also where the prey species C grows to a larger density than what is possible for a homogeneous population. This is illustrated in Fig. 3b, which shows the equilibrium population of C if we hold (lambda) fixed at different values. The point (lambda = lambda ^*) is where the system with a dynamic (lambda) is stable, and also where the population of C is maximized (when A and B prefer different prey types). That means the population automatically stabilizes at the optimal composition of prey types. Moreover, the value of (C^*) at this coexistence point can even be larger than the equilibrium population of C when there is only one predator A or B. This is discussed further in section “Multiple-predator effects and emergent promotion of prey”. These results suggest that heterogeneity in interaction strengths can potentially be a strategy for the prey population to leverage the effects of multiple predators against each other to improve survival.Reversible heterogeneityWe next consider a scenario where individual prey can switch their types. This kind of heterogeneity can model reversible changes of phenotypes, i.e., trait changes that affect the prey’s interaction with predators but are not permanent. For example, changes in coat color or camouflage14,16,17, physiological changes such as defense15, and biomass allocation among tissues18,19. One could also think of the prey types as subpopulations within different spatial patches, if each predator hunts at a preferred patch and the prey migrate between the patches20,21. With some generalization, one could even consider heterogeneity in resources, such as nutrients located in different places, that can be reached by primary consumers, such as swimming phytoplankton22. We can model this “reversible” kind of heterogeneity by introducing switching rates from one prey type to the other. In Eq. (4d), (eta _1) and (eta _2) now represent the switching rates per capita from (C_1) to (C_2) and from (C_2) to (C_1), respectively. Here we study the simplest case where both rates are fixed.In the absence of the predators, the natural composition of the prey species given by the switching rates would be (rho equiv eta _1 / (eta _1 + eta _2)), and the rate at which (lambda) relaxes to this natural composition is (gamma equiv eta _1 + eta _2). Compared to the previous scenario where we had only one parameter (rho), here we have an additional parameter (gamma) that modifies the behavior of the system. Fig. 4e–h shows phase diagrams for the system as (rho) is fixed and (gamma) varies. We again find the new equilibrium (P_N) where all three species coexist. When (gamma) is small, the system has a large region of coexistence. As (gamma) is increased, this region is squeezed into a border between the two regions of exclusion, where the slope of the border is given by (eta _1/eta _2) as determined by the parameter (rho). However, this is different from the exclusion we see in the case of inherent heterogeneity, which happens only for (rho rightarrow 0) or 1, where the borders are horizontal or vertical (Fig. 4a,d). Here the predators exclude each other despite having a mixture of prey types in the population.This special limit can be understood as follows. For a large (gamma), (lambda) is effectively set to a constant value equal to (rho), because it has a very fast relaxation rate. In other words, the prey types exchange so often that the population always maintains a constant composition. In this limit, the system behaves as if it were a homogeneous system with effective interaction strengths (a_text {eff} = (1-rho ) , a_1 + rho , a_2) and (b_text {eff} = (1-rho ) , b_1 + rho , b_2). As in a homogeneous system, there is competitive exclusion between the predators instead of coexistence. This demonstrates that having a constant level of heterogeneity is not sufficient to cause coexistence. The overall composition of the population must be able to change dynamically as a result of population growth and consumption by predators.An interesting behavior is seen when we examine a point inside the shrinking coexistence region as (gamma) is increased. Typical trajectories of the system for such parameter values are shown in Fig. 5. As (gamma) increases, the system relaxes to the line (mathscr {L}) quickly, then slowly crawls along it towards the final equilibrium point (P_N). This is because increasing (gamma) increases the speed that (lambda) relaxes to (lambda ^*), and when (lambda rightarrow lambda ^*), (mathscr {L}) becomes marginally stable. Therefore, the attraction to (mathscr {L}) in the perpendicular direction is strong, but the attraction towards the equilibrium point along (mathscr {L}) is weak. This leads to a long transient behavior that makes the system appear to reach no equilibrium in a limited time23,24. It is especially true when there is noise in the dynamics, which causes the system to diffuse along (mathscr {L}) with only a weak drift towards the final equilibrium (Fig. 5). Thus, the introduction of a fast timescale (quick relaxation of (lambda) due to a large (gamma)) actually results in a long transient.Figure 5Trajectories of the system projected in the A-B plane, with parameters inside the coexistence region (by holding the position of (P_N) fixed). As (gamma) increases, the system tends to approach the line (mathscr {L}) quickly and then crawl along it. The grey trajectory is with independent Gaussian white noise ((sim mathscr {N}(0,0.5))) added to each variable’s dynamics. Noise causes the system to diffuse along (mathscr {L}) for a long transient period before coming to the equilibrium point (P_N). Parameters used here are ((a_0, a_1, a_2, b_0, b_1, b_2) = (0.2, 0.8, 0.5, 0.2, 0.6, 0.9)), chosen to place (P_N) away from the middle of (mathscr {L}) to show the trajectory drifting toward the equilibrium.Full size imageAdaptive heterogeneityA third kind of heterogeneity we consider is the change of interactions in time. By this we mean an individual can actively change its interaction strength with others in response to certain conditions. This kind of response is often invoked in models of adaptive foraging behavior, where individuals choose appropriate actions to maximize some form of fitness25,26. For example, we may consider two behaviors, resting and foraging, as our prey types. Different predators may prefer to strike when the prey is doing different things. In response, the prey may choose to do one thing or the other depending on the current abundances of different predators. Such behavioral modulation is seen, for example, in systems of predatory spiders and grasshoppers27. Phenotypic plasticity is also seen in plant tissues in response to consumers28,29,30.This kind of “adaptive” heterogeneity can be modeled by having switching rates (eta _1) and (eta _2) that are time-dependent. Let us assume that the prey species tries to maximize its population growth rate by switching to the more favorable type. From Eq. (4c), we see that the growth rate of C depends linearly on the composition (lambda) with a coefficient (u(A,B) equiv (a_1 – a_2) A + (b_1 – b_2) B). Therefore, when this coefficient is positive, it is favorable for C to increase (lambda) by switching to type (C_2). This can be achieved by having a positive switching rate (eta _2) whenever (u(A,B) > 0). Similarly, whenever (u(A,B) < 0), it is favorable for C to switch to type (C_1) by having a positive (eta _1). In this way, the heterogeneity of the prey population constantly adapts to the predator densities. We model such adaptive switching by making (eta _1) and (eta _2) functions of the coefficient u(A, B), e.g., (eta _1(u) = 1/(1+mathrm {e}^{kappa u})) and (eta _2(u) = 1/(1+mathrm {e}^{-kappa u})). The sigmoidal form of the functions means that the switching rate in the favorable direction for C is turned on quickly, while the other direction is turned off. The parameter (kappa) controls the sharpness of this transition.Phase diagrams for the system with different values of (kappa) are shown in Fig. 4i–l. A larger (kappa) means the prey adapts its composition faster and more optimally, which causes the coexistence region to expand. In the extreme limit, the system changes its dynamics instantaneously whenever it crosses the boundary where (u(A,B) = 0), like in a hybrid system31. Such a system can still reach a stable equilibrium that lies on the boundary, if the flow on each side of the boundary points towards the other side32. This is what happens in our system and, interestingly, the equilibrium is the same three-species coexistence point (P_N) as in the previous scenarios. The region of coexistence turns out to be largest in this limit (Fig. 4l).Our results suggest that the coexistence of the predators can be viewed as a by-product of the prey’s strategy to maximize its own benefit. The time-dependent case studied here represents a strategy that involves the prey evaluating the risk posed by different predators. This is in contrast to the scenarios studied above, where the prey population passively creates phenotypic heterogeneity regardless of the presence of the predators. These two types of behavior are analogous to the two strategies studied for adaptation in varying environments, i.e., sensing and bet-hedging33,34. The former requires accessing information about the current environment to make optimal decisions, whereas the latter relies on maintaining a diverse population to reduce detrimental effects caused by environmental changes. Here the varying abundances of the predators play a similar role as the varying environment. From this point of view, the heterogeneous interactions studied here can be a strategy of the prey species that is evolutionarily favorable. More

During the COVID-19 pandemic, behavioral restrictions were imposed, after which various health problems were reported in many countries45,46. The pandemic has also increased food insecurity worldwide; consequently, panic buying has been observed in many countries, including Japan47. However, even in such situations, we found that diversity in local food access, ranging from self-cultivation to direct-to-consumer sales, was significantly associated with health and food security variables. Specifically, our results revealed the following five key discussion points.Urban agriculture in walkable neighborhoods bore fruit for health and food system resilience. However, the magnitude of its contribution differed depending on the type of urban agricultureThe results of this study showed that those who grew food by themselves at allotment farms and home gardens had significantly better subjective well-being and physical activity levels than those who did not. This result is in line with previous studies conducted during times free from the impact of infectious disease pandemics38,39,40. The use of direct sales was not related to subjective well-being but was significantly associated with physical activity. The reason might be that farm stand users tend to live in areas with farmland and travel to purchase fruits and vegetables at farm stands on foot or by bicycle. This result is consistent with that of a previous study demonstrating that the food environment in neighborhoods is an important component in promoting physical activity17.Our results also showed that those who grew food by themselves at allotment farms and those who purchased local foods at farm stands were significantly less anxious about the availability of fresh food both during the state of emergency and in the future than their counterparts. In contrast, home garden users showed significant differences only for the state of emergency. This result might be due to the differences in the size and yield of cultivation at allotment farms and home gardens. One lot in allotment farms in Tokyo can produce as much as or more than the average annual vegetable consumption per household in Japan48. However, home gardens are generally smaller and produce limited fresh foods for consumption, which may have influenced food security concerns.As in other countries, Japan imports much food from overseas and is deeply integrated into the large-scale global food system. However, as shown in this study, urban agriculture in Japanese suburbs forms small-scale, decentralized, and community-based local food systems. This multilayered food system can complement the disruptions and shortages of the global system when various problems occur for climatic, sociopolitical, or other reasons, such as pandemics. In fact, our empirical evidence suggests that urban agriculture in walkable neighborhoods, particularly allotment farms and direct-to-consumer sales at farm stands, contributed to the mitigation of food security concerns in neighborhood communities. This means that urban agriculture could enhance the resilience of the urban food system at a time when the global food system has been disrupted due to a pandemic. This validates recent discussions about the potential of urban agriculture to facilitate food system resilience10. Furthermore, our findings imply that the types of urban agriculture employed matter in determining the degree of contribution to food system resilience.To summarize the overall results, urban agriculture in walkable neighborhoods bore fruit for health and food system resilience during the COVID-19 pandemic. However, different types of urban agriculture exhibited varying associations with health and resilience. Allotment farms were positively related to all of the following: subjective well-being, physical activity, and food security concerns, both during the state of emergency and in the future. Home gardens were positively related to subjective well-being, physical activity, and food security concerns only during the state of emergency. Farm stands were positively related to physical activity and food security concerns both during the state of emergency and in the future.These differences may be due to the characteristics of the respective spaces. It is suggested that this diversity of urban agriculture has led to different types of people benefiting from various kinds of urban agriculture. Allotment farms were found to be associated with high subjective well-being, physical activity, and food security, but they may not be feasible for those who do not have enough physical strength because users are responsible for cultivating their lots, which measure 10–30 square meters40. In contrast, home gardens can be created even by those who are not confident in their physical strength. In fact, our study showed that women and older people engaged in home gardening more than men and younger people. In addition, direct-to-consumer sales at farm stands are the easiest way to obtain local fresh foods for those who do not have the time and space for allotment farms and home gardens. The need for urban agriculture has been argued in many countries2,3. However, little attention has been paid to its scale, accessibility, and diversity. Our study suggests that it is worthwhile to create diverse food production spaces within walkable neighborhoods while considering the diversity of people who access these spaces.Compared to other urban greenery and food retailers, the benefits of urban agriculture on subjective well-being and food security could be greaterCompared to the use of other urban green spaces, including urban parks, our results indicated that self-cultivation at allotment farms and home gardens was more strongly associated with subjective well-being. Previous studies have offered limited perspectives on the differences among various types of urban green spaces33. Our study further suggests that urban parks, allotment farms, and home gardens are differently associated with human health. However, as the reason was not determined, further research is needed.Furthermore, compared to other food retailers, such as supermarkets, convenience stores, and co-op deliveries, allotment farms and farm stands were more strongly associated with less anxiety about fresh food availability in the future. The availability of local fresh foods within walkable neighborhoods might have mitigated food security concerns because residents could grow food by themselves or directly observe farmers’ production processes, which may have made the difference from purchasing at places where the food systems were not visible.Flexibility in work style might promote urban agriculture in walkable neighborhoodsThere was an association between work style—working from home—and access to local food. According to the Ministry of Health, Labor and Welfare (https://www.mhlw.go.jp/english), 52% of Tokyo office workers worked from home during the first emergency declaration. Long commute times and high train congestion rates have been a problem in Tokyo suburbs, but remote workers have gained more time at and around their homes by reducing their commute times, increasing their opportunities to access local food in their walkable neighborhoods. Those who worked from home sought outdoor activities for refreshment and exercise and used a variety of urban green spaces during the pandemic49. Allotment farms and home gardens might be used as such urban green spaces. This result is consistent with previous studies assessing the characteristics of Canadian gardeners during the COVID-19 pandemic28,30.Until now, urban planners and policymakers have rarely taken work style into account. However, the flexibility of work styles and work hours may bring new insights; for example, those who work from home may become important players in urban agriculture. It has been pointed out that cities have a large hidden potential for urban agriculture by cultivating underused lands50. Our study suggests that such underused lands could be converted into productive urban landscapes for remote workers to engage in farming or gardening in between jobs as a hobby or as a side business.Food equity might be improved by urban agriculture in walkable neighborhoodsLocal fresh food is generally considered more expensive than junk food in high-income countries, creating social issues of food inequity. Therefore, past discussions on urban agriculture and food security have focused primarily on low-income households in socioeconomically disadvantaged areas24,25,26.In contrast, our study covered people from all income groups and found no statistically significant relationship between access to local food and income. This finding might be due to two urban cultural backgrounds regarding local food in Tokyo, that is, accessibility and affordability. First, residential segregation by income levels is not noteworthy in Tokyo and people from various income brackets live mixed in the same neighborhoods51. Therefore, most urban residents living in the suburbs have geographically equitable opportunities to access local foods. Second, local foods sold at farm stands are affordable. Prices are almost the same or cheaper than buying food at food retailers. While prices increase because of middleman margins related to shipping in the wholesale market, such increases are unnecessary when selling directly to consumers at farm stands. In addition, the allotment farm lots are not expensive to rent, particularly those operated by local municipalities (Supplementary Note 1).These two backgrounds make local fresh food physically and economically accessible to consumers of all income levels, resulting in food equity. This is particularly important because the concept of food system resilience includes the equitability perspective27.The integration of urban agriculture into walkable neighborhoods is a fruitful wayWhile the current discussion on walkable neighborhoods does not emphasize urban agriculture, our evidence indicated its effectiveness. The concept of walkable neighborhoods (e.g., the 15-min city model) stresses the decarbonization benefit of limiting vehicle travel, as well as the health benefits of promoting walking and cycling13,14,15,16. In addition, our research indicated that urban agriculture in walkable neighborhoods benefited health and well-being by increasing recreational outdoor opportunities to neighborhood communities, including remote workers. It also contributed to food system resilience by providing local foods to all people, including low-income households, when the global food system was disrupted due to the pandemic. Furthermore, recent studies on urban agriculture reported the decarbonization benefit of reducing carbon footprints in food production and distribution7,8. Small-scale and community-based urban agriculture in walkable neighborhoods might especially bring this benefit because neighborhood communities travel to farms on foot or by bicycle, which means almost no emission by distribution. While urban green spaces have various health benefits32,33,34,35, urban agriculture also contributes to food system resilience as well as carbon emission reduction, which makes it unique.Urban agriculture was once considered a failure of urban planning in Japan because it symbolized uncontrolled sprawl. This is analogous to the Western view, as urban agriculture was once considered the ultimate oxymoron1. However, our empirical evidence suggests that the urban‒rural mixture at neighborhood scales is a reasonable urban form that contributes to the resilience of the urban food system and to the health and well-being of neighborhood communities. It is no longer a failure of urban planning but a legacy of urban sprawl in the current urban context.Our study showed that integrating urban agriculture into walkable neighborhoods is a fruitful way of creating healthier cities and developing more resilient urban food systems during times of uncertainty. In cities where there is no farmland in intraurban areas, it would be considered effective to utilize underused spaces such as vacant lots and rooftops as productive urban landscapes. In growing cities where urban areas are still expanding, it would be advantageous to conserve agricultural landscapes within their urban fabrics. Our study could provide referential insights and robust evidence for urban policy to integrate urban agriculture into walkable neighborhoods.This study has potential limitations, including the timing of the survey and the measurement method that was utilized. We conducted the survey between June 4 and 8, 2020, just after the end of the first declaration of a state of emergency by the Japanese government. During this period, the main cultivation activities were planting and growing, and the harvest was just beginning. This seasonal constraint may have influenced the results. Because the survey was conducted during the pandemic, we used subjective methods to measure health and well-being status. However, the results might be different using objective methods52, thus further research is necessary. In addition, a longitudinal study is needed to determine whether the trends observed in this study were specific to the emergency period or whether they will persist after the COVID-19 pandemic. More

Analytical method validationThe results of the precision study with relative standard deviation (RSD), and accuracy are shown in Table 1. Through the precision study we found the value of RSD as less than 5%. Moreover, accuracy was done with percent recovery experiments. The results showed that the percentage recoveries for spiked samples were in the range of 95.7–103.7%.Table 1 Shows percent (%) recovery and relative standard deviation.Full size tablePhysicochemical properties and water quality indexThe investigations of the water quality properties of the Narora channel are shown in Table 2. The temperature, TDS, turbidity, and alkalinity were within the standards of the country18 and WHO19 (taken from UNEPGEMS). While pH and dissolved oxygen (D.O) were above the recommended standards indicating poor water quality. Moreover, the detected heavy metals were in the following order Ni  > Fe  > Cd  > Zn  > Cr  > Cu  > Mn. Among these heavy metals Mn, Cu, and Zn were within the recommended limits whereas Cr, Fe, Ni, and Cd were crossing the limits18 contributing to the poor quality. Furthermore, the WQI calculation will give more insights into the overall quality of water as it explains the combined effect of several physicochemical properties12. Its calculation is done simply by converting numerous variables of water quality into a single number12,20. In addition to this, WQI simplifies all the data and helps in clarifying water quality issues by combining the complex data and producing a score that shows the status of water quality2,12,21. The WQI classifies water quality status into five groups such as if WQI  Cu  > Zn  > Fe  > Zn  > Ni  > Cr from root to stalk; and Mn  > Cd  > Zn  > Cu  > Fe  > Ni  > Cr from stalk to leaves.Table 5 Heavy metal concentrations in Eichhornia crassipes (mg/kg.dw).Full size tableFigure 3MPI values in E. crassipes.Full size imageTable 6 Bioaccumulation factor (BAF), transfer factor (TF), and mobility factor (MF) in plant E. crassipes.Full size tableThese factors BAF, TF, and MF are utilized to monitor the level of anthropogenic pollution in plants and their surrounding medium2,15,32,34,35. BAF shows the concentrations of heavy metals bioaccumulated by plants from the water. If the BAF  > 1 it indicates hyperaccumulation36. So, in the present study, all the concerned heavy metals were hyperaccumulated in the plant. The TF elucidates the capability of the plant to translocate the accumulated metals to its other parts. The roots of E. crassipes showed the highest translocation capacity for Ni (1.57) as well as Zn (1.30) to other parts. If the value of TF exceeds 1, then it represents the high accumulation efficiency37,38, therefore, plants will be considered as the hyperaccumulators for the Ni and Zn. Although the Cd was the highest accumulated metal in the plant, it could have been because of its may be because of its low TF. Whereas, TF values lower than 1 for Cr, Mn, Fe, Cu, and Cd pointed out that this plant’s roots act as a non-hyperaccumulator for these heavy metals. Furthermore, the highest MF values were depicted for Mn in both cases which reflects that E. crassipes can suitably be used for phytoextraction of Mn as well as for Cd, Zn, Fe, Ni, and Cu. The BAF, TF, and MF of Cr are low in the present study, which implies that roots are limiting the Cr. Moreover, if the BAF ≤ 1.00 then it shows the capability of absorption only rather than accumulation36,37. In addition, if the values of BAF, TF, and MF exceed 1, plants can also work for phytoextraction. Furthermore, if the BAF  > 1 and TF  More

All the materials followed relevant institutional and national guidelines and legislation.MitesWe used a T. kanzawai population collected from trifoliate orange trees (Poncirus trifoliata [L.] Raf.) in 2018 in Kyoto, Japan, and a T. urticae population collected from chrysanthemum plants (Chrysanthemum morifolium Ramat.) in 1998 in Nara, Japan. These populations were reared on adaxial surfaces of kidney bean (Phaseolus vulgaris L.) primary leaves, which were pressed onto water-saturated cotton in Petri dishes (90 mm diameter, 14 mm depth). The water-saturated cotton served as a barrier to prevent mites from escaping. The dishes were maintained at 25 °C, 50% relative humidity, and a 16L:8D photoperiod. All experiments were conducted under these conditions. We only used mated adult females (i.e., the dispersal stage) of T. kanzawai or T. urticae mites.CaterpillarsWe used caterpillars of four lepidopteran species: Bombyx mori L., P. Xuthus, Spodoptera litura Fabricius and T. oldenlandiae. We collected eggs and larvae of T. oldenlandiae from C. japonica in 2021 in Kyoto, Japan, and reared them on C. japonica leaves until pupation. Theretra oldenlandiae shares Vitaceae host plants with T. kanzawai and T. urticae8,15. We collected eggs and larvae of P. xuthus from Ptelea trifoliata in 2021 in Kyoto, Japan, and reared them on Citrus unshiu Markov. leaves until pupation. Papilio. xuthus and T. kanzawai share P. trifoliata as a host plant in Kyoto (Kinto, personal observation).We obtained commercial populations of the B. mori Kinshu × Showa strain (Ueda-sanshu Co., Ltd, Nagano, Japan) or the w1-pnd strain. We reared B. mori larvae on an artificial diet produced at the Kyoto Institute of Technology. Although T. kanzawai use Morus alba, a food plant for the B. mori strain, the mite and the strain never encounter one another in the wild, because the B. mori strain has been domesticated for hundreds of years.We obtained a sub-cultured population of S. litura from the Kyoto Institute of Technology. We reared first to fourth instars of S. litura on an artificial diet (Insecta LFM, Nosan Insect Materials, Kanagawa, Japan), while final instars were fed P. vulgaris leaves. Because S. litura feeds on various wild and cultivated plants22,23, it may share some host plants with T. kanzawai and T. urticae, both of which also feed on many host plant species8,9,10.We reared caterpillars of T. oldenlandiae, P. xuthus, and S. litura in 900 mL transparent plastic cups and caterpillars of B. mori in transparent plastic containers (140 × 220 × 35 mm). All caterpillars were maintained under the same laboratory conditions described above.PlantsWe used several parts of P. vulgaris plants in the following experiments. This species is a preferred food for both mite species16,17 and S. litura24, but the other three caterpillar species do not feed on it (Kinto, personal observation). We thus used P. vulgaris rather than shared host plants, because some caterpillars and mites (T. urticae and P. xuthus, for example) do not share any host plant.Avoidance of caterpillar traces on leaf surfaces by spider mitesTo examine whether spider mites avoid settling on host plant surfaces bearing caterpillar traces, we conducted dual-choice tests using paired adjacent leaf squares with and without caterpillar traces. We did not use whole plants because, in practice, it was difficult to induce caterpillar traces on whole plants. We used two spider mite species (T. kanzawai and T. urticae) and four caterpillar species (T. oldenlandiae, P. xuthus, B. mori, and S. litura). We cut a 10 × 20 mm leaf piece from a fully expanded primary kidney bean leaf and then cut the piece into two equal squares (10 × 10 mm). To introduce caterpillar traces to one square, we arranged them on a separate piece of paper towel on water-saturated cotton. This procedure was necessary because the caterpillars used were larger than individual leaf squares. Then we placed a fourth or final instar caterpillar on the squares and induced the caterpillar to walk across every leaf square three times (Fig. 1a). We carefully removed all caterpillar-produced silk threads from the squares. Within 30 min, we arranged the square (trace +) to touch against the other square (trace −) on water-saturated cotton in a Petri dish. Subsequently, a 2- to 4-day-old mated adult female of T. kanzawai or T. urticae was introduced onto a pointed piece of Parafilm in contact with both leaf edges using a fine brush (Fig. 1a). We recorded the leaf square onto which the mite had settled at 2 h after its introduction, as preliminary observations confirmed that all females would settle on a particular leaf within that period. Each female mite and pair of leaf squares were used only once. All tests described below were conducted between 13:00 and 17:00 h, when adult female spider mites actively disperse by walking. There were 14 replicates using traces of T. oldenlandiae, 48 of P. xuthus, 20 of B. mori, and 26 of S. litura for T. kanzawai, as well as 18, 32, 16, and 47, respectively, for T. urticae. Data were subjected to two-tailed binomial tests with the common null hypothesis that a spider mite would settle on the two squares with equal probability (i.e., 0.5).Figure 1(a) Procedure used to observe avoidance of caterpillar traces by spider mites. (b) Experimental setup used to observe avoidance of B. mori traces on plant stems by T. kanzawai. (c) Experimental setup used to observe avoidance of B. mori trace extracts by T. kanzawai.Full size imageDuration of B. mori trace avoidance by T. kanzawai
To examine whether the effects of caterpillar traces on spider mite avoidance decline over time, we used T. kanzawai mites and B. mori caterpillars. We used B. mori because populations can be easily maintained over many generations. We prepared bean leaf squares with B. mori traces in the same manner descried above and preserved the traced square on water-saturated cotton for 0 h (n = 30), 24 h (n = 29), 48 h (n = 28), or 72 h (n = 28). Then we arranged the square (trace +) to lie in close proximity to the control square (trace −) that had been preserved for the same periods of time. Then we compared the avoidance response of T. kanzawai females in the same manner described above.Avoidance of B. mori traces on plant stems by T. kanzawai
To examine whether T. kanzawai females avoid walking along plant stems bearing caterpillar traces, we used Y-shaped kidney bean stems (Fig. 1b). We cut symmetric bean plants ca. 15 days after sowing from their base and inserted them perpendicularly into a 5 mL glass bottle filled with water and wet cotton. To induce caterpillar traces on one branch of the stem, we allowed a silkworm to crawl from the branching point to the far end of one branch three times for each stem (n = 20). Then we introduced a T. kanzawai adult female at a release point 35 mm below the branch point (Fig. 1b). We recorded the branch along which the female walked to the far end. Each female mite and each Y-shaped stem were used only once. The numbers of females were compared using binomial tests in the same manner described above.Avoidance of B. mori trace extracts by T. kanzawai
To extract chemical traces of caterpillar, we introduced 10 third instar B. mori to a glass Petri dish (120 mm diameter, 60 mm depth). After 1 h, we removed all caterpillars and washed the inside bottom of the dish with 1.0 mL acetone. We replicated the procedure twice using different individuals to combine all extracts and to acquire enough extract for the following experiment.To examine avoidance of B. mori trace extracts by T. kanzawai females, we conducted dual-choice experiments using T-shaped pathways of filter paper (35 × 35 mm; width, 2 mm; Fig. 1c). Using disposable micropipettes (Drummond Scientific Co., PA, USA), 1.75 caterpillar equivalents (i.e., 60 µL) of acetone extract were applied to an alternately selected branch (17.5 mm long) of each pathway (i.e., 0.10 caterpillar equivalent/mm), with control acetone applied to the other branch. We applied each solution dropwise at the junction point to minimize mixing. After evaporating the solvent from those pathways, we perpendicularly suspended them (Fig. 1c) and introduced an adult female mite at 2 days post-maturation onto the bottom of each pathway using a fine brush and recorded the branch along which the female first walked to the far end. Each female mite and each T-shaped filter paper were used only once, with 19 replicates. Each female mite made a choice within 10 min. The avoidance response of T. kanzawai was analysed in the same manner described above.Indirect effects of B. mori traces on T. kanzawai via plantsTo determine whether B. mori traces on plants indirectly affect the performance of T. kanzawai on plants, we introduced 70–80 randomly selected quiescent female deutonymphs of T. kanzawai onto kidney bean leaf disks. Immediately after synchronized adult emergence, we introduced the same number of adult males to allow mating; the detailed procedure is described elsewhere25. After 24 h, we transferred the females singly onto 10 × 10 mm bean leaf squares with or without B. mori traces prepared as described above. Because the number of eggs laid within a certain period is considered the most sensitive performance index of spider mite females26,27, any plant-mediated indirect interaction, such as defence induction in response to caterpillar traces, should result in lower egg numbers laid by the test females. We counted the eggs laid on the leaf squares 24 h after their introduction. One female that laid no eggs during the 24 h period (n = 1, trace +) was excluded from the analysis. We obtained 33 and 36 replicates for the trail+ and trail– conditions, respectively. We compared the numbers of eggs laid on leaves with and without B. mori traces using a generalized linear model with a Poisson error distribution using the SAS 9.22 software (SAS Institute Inc., Cary, NC, USA).EthicsThis article does not contain any studies with human participants or animals. More

Arunrat, N., Sereenonchai, S., Chaowiwat, W. & Wang, C. Climate change impact on major crop yield and water footprint under CMIP6 climate projections in repeated drought and flood areas in Thailand. Sci. Total Environ. 807, 150741 (2022).ADS 
CAS 

Google Scholar 
Chandio, A. A., Shah, M. I., Sethi, N. & Mushtaq, Z. Assessing the effect of climate change and financial development on agricultural production in ASEAN-4: the role of renewable energy, institutional quality, and human capital as moderators. Environ. Sci. Pollut. Res. 29, 13211–13225 (2022).
Google Scholar 
Masood, N., Akram, R., Fatima, M., Mubeen, M., Hussain, S., Shakeel, M., Khan, N., Adnan, M., Wahid, A., Shah, A. N. and Ihsan, M. Z. (2022) Insect pest management under climate change. In Building climate resilience in agriculture. Springer, ChamOzdemir, D. The impact of climate change on agricultural productivity in Asian countries: A heterogeneous panel data approach. Environ. Sci. Pollut. Res. 29, 8205–8217 (2022).
Google Scholar 
Aidoo, O. F. et al. Climate-induced range shifts of invasive species (Diaphorina citri Kuwayama). Pest Manag. Sci. 78, 2534–2549 (2022).CAS 

Google Scholar 
Hebbar, K. B. et al. Predicting the Potential Suitable Climate for Coconut (Cocos nucifera L.) Cultivation in India under Climate Change Scenarios Using the MaxEnt Model. Plants. 11, 731 (2022).
Google Scholar 
Martín-Vélez, V. & Abellán, P. Effects of climate change on the distribution of threatened invertebrates in a Mediterranean hotspot. Insect Conserv. Divers. 15, 370–379 (2022).
Google Scholar 
Williams, J. J., Freeman, R., Spooner, F. & Newbold, T. Vertebrate population trends are influenced by interactions between land use, climatic position, habitat loss and climate change. Glob. Chang. Biol. 28, 797–815 (2022).CAS 

Google Scholar 
Aidoo, O. F. et al. Lethal yellowing disease: insights from predicting potential distribution under different climate change scenarios. J. Plant Dis. Prot. 128, 1313–1325 (2021).
Google Scholar 
Sofaer, H. R. et al. Development and delivery of species distribution models to inform decision-making. Bioscience 69, 544–557 (2019).
Google Scholar 
Mead FW, The Asiatic citrus psyllid, Diaphorina citri Kuwayama (Homoptera: Psyllidae). Florida Department of Agriculture Conservation Service, Division of Plant Industry Entomological Circular No. 180.Bové, J. M. Huanglongbing: A destructive, newly-emerging, century-old disease of citrus. Plant Pathol. J. 1, 7–37 (2006).
Google Scholar 
Li, S., Wu, F., Duan, Y., Singerman, A. & Guan, Z. Citrus greening: Management strategies and their economic impact. HortScience 55, 604–612 (2020).
Google Scholar 
Jia, H. et al. Genome editing of the disease susceptibility gene Cs LOB 1 in citrus confers resistance to citrus canker. Plant Biotechnol. J. 15, 817–823 (2017).CAS 

Google Scholar 
Ehsani, R., Dewdney, M. & Johnson, E. Controlling HLB with thermotherapy: What have we learned so far?. Citrus Ind. News 9, 26–28 (2016).
Google Scholar 
Spreen, T. H., Baldwin, J. P. & Futch, S. H. An economic assessment of the impact of Huanglongbing on citrus tree plantings in Florida. J. Hortic. Sci. 49, 1052–1055 (2014).
Google Scholar 
Djeddour, D., Pratt, C., Constantine, K., Rwomushana, I. and Day, R., (2021) The Asian citrus greening disease (Huanglongbing). Evidence note on invasiveness and potential economic impacts for East Africa. CABI Working Paper, 24, 94Hu, J., Jiang, J. & Wang, N. Control of citrus Huanglongbing via trunk injection of plant defense activators and antibiotics. Phytopathology 108, 186–195 (2018).CAS 

Google Scholar 
Fan, G. C. et al. Evaluation of thermotherapy against Huanglongbing (citrus greening) in the greenhouse. J. Integr. Agric. 15, 111–119 (2016).
Google Scholar 
Nguyen, V. A., Bartels, D. & Gilligan, C. Modelling the spread and mitigation of an emerging vector-borne pathogen: citrus greening in the US. Biorxiv https://doi.org/10.1101/2022.05.04.490566 (2022).Article 

Google Scholar 
Milosavljević, I. et al. Post-release evaluation of Diaphorencyrtus aligarhensis (Hymenoptera: Encyrtidae) and Tamarixia radiata (Hymenoptera: Eulophidae) for biological control of Diaphorina citri (Hemiptera: Liviidae) in Urban California, USA. Agronomy 12, 583 (2022).
Google Scholar 
Maluta, N., Castro, T. & Lopes, J. R. S. Entomopathogenic fungus disrupts the phloem-probing behavior of Diaphorina citri and may be an important biological control tool in citrus. Sci. Rep. 12, 1–10 (2022).
Google Scholar 
Hall, D. G., Richardson, M. L., Ammar, E. D. & Halbert, S. E. Asian citrus psyllid, Diaphorina citri, vector of citrus huanglongbing disease. Entomol. Exp. Appl. 146, 207–223 (2013).
Google Scholar 
Vázquez-García, M. et al. Insecticide resistance in adult Diaphorina citri Kuwayama1 from lime orchards in central west Mexico. Southwest. Entomol. 38, 579–596 (2013).
Google Scholar 
Naeem, A., Freed, S., Jin, F. L., Akmal, M. & Mehmood, M. Monitoring of insecticide resistance in Diaphorina citri Kuwayama (Hemiptera: Psyllidae) from citrus groves of Punjab Pakistan. Crop Prot. 86, 62–68 (2016).CAS 

Google Scholar 
Hulme, P. E. et al. Grasping at the routes of biological invasions: A framework for integrating pathways into policy. J. Appl. Ecol. 45, 403–414 (2008).
Google Scholar 
Oke, A. O., Oladigbolu, A. A., Kunta, M., Alabi, O. J. & Sétamou, M. First report of the occurrence of Asian citrus psyllid Diaphorina citri (Hemiptera: Liviidae), an invasive species in Nigeria. West Africa. Sci. Rep. 10, 1–8 (2020).
Google Scholar 
Tang, Y.Q. (1990) On the parasite complex of Diaphorina citri Kuwayama (Homoptera: Psyllidae) in Asian-Pacific and other areas. In proceedings 4th international conference on citrus rehabilitation, Chiang Mai, Thailand. 4: 240 245Chien, C. C., Chiu, S. C. & Ku, S. C. Biological control of Diaphorina citri in Taiwan. Fruits 44, 401–407 (1989).
Google Scholar 
Hoddle, M. S. Foreign exploration for natural enemies of Asian citrus psyllid, Diaphorina citri (Hemiptera: Psyllidae), in the Punjab of Pakistan for use in a classical biological control program in California USA. Pakistan Entomol. 34, 1–5 (2012).
Google Scholar 
Étienne, J., Quilici, S., Marival, D., Franck, A. & Gonzalez Fernandez, C. Biological control of Diaphorina citri (Hemiptera: Psyllidae) in Guadeloupe by imported Tamarixia radiata (Hymenoptera: Eulophidae). Fruits 56, 307–315 (2001).
Google Scholar 
Qureshi, J. A., Rogers, M. E., Hall, D. G. & Stansly, P. A. Incidence of invasive Diaphorina citri (Hemiptera: Psyllidae) and its introduced parasitoid Tamarixia radiata (Hymenoptera: Eulophidae) in Florida citrus. J. Econ. Entomol. 102, 247–256 (2009).
Google Scholar 
Chen, X., Triana, M. & Stansly, P. A. Optimizing production of Tamarixia radiata (Hymenoptera: Eulophidae), a parasitoid of the citrus greening disease vector Diaphorina citri (Hemiptera: Psylloidea). Biol. Control. 105, 13–18. https://doi.org/10.1016/j.biocontrol.2016.10.010 (2017).Article 

Google Scholar 
Kistner, E. J., Amrich, R., Castillo, M., Strode, V. & Hoddle, M. S. Phenology of Asian citrus psyllid (Hemiptera: Liviidae), with special reference to biological control by Tamarixia radiata, in the residential landscape of southern California. J. Econ. Entomol. 109, 1047–1057. https://doi.org/10.1093/jee/tow021 (2016).Article 

Google Scholar 
Ramos Aguila, L. C. et al. Temperature-dependent biological control effectiveness of Tamarixia radiata (Hymenoptera: Eulophidea) under laboratory conditions. J. Econ. Entomol. 114, 2009–2017 (2021).
Google Scholar 
Ramos Aguila, L. C. et al. Temperature-dependent demography and population projection of Tamarixia radiata (Hymenoptera: Eulophidea) reared on Diaphorina citri (Hemiptera: Liviidae). J. Econ. Entomol. 113, 55–63 (2020).
Google Scholar 
Ashraf, H. J. et al. Comparative microbiome analysis of Diaphorina citri and its associated parasitoids Tamarixia radiata and Diaphorencyrtus aligarhensis reveals Wolbachia as a dominant endosymbiont. Environ. Microbiol. 24, 1638–1652 (2022).CAS 

Google Scholar 
Chow, A. & Sétamou, M. Parasitism of Diaphorina citri (Hemiptera: Liviidae) by Tamarixia radiata (Hymenoptera: Eulophidae) on residential citrus in Texas: Importance of colony size and instar composition. Biol. Control 165, 104796 (2022).
Google Scholar 
Ajene, I. J. et al. Habitat suitability and distribution potential of Liberibacter species (“Candidatus Liberibacter asiaticus” and “Candidatus Liberibacter africanus”) associated with citrus greening disease. Environ. Microbiol. 26, 575–588 (2020).
Google Scholar 
Shabani, F., Kumar, L. & Ahmadi, M. A comparison of absolute performance of different correlative and mechanistic species distribution models in an independent area. Ecol. Evol. 6, 5973–5986 (2016).
Google Scholar 
Kearney, M. & Porter, W. Mechanistic niche modelling: Combining physiological and spatial data to predict species’ ranges. Ecol 12, 334–350 (2009).
Google Scholar 
Byeon, D. H., Jung, S. & Lee, W. H. Review of CLIMEX and MaxEnt for studying species distribution in South Korea. J. Asia-Pac. Biodivers. 1, 325–333 (2018).
Google Scholar 
Kriticos, D. J., Yonow, T. & McFadyen, R. E. The potential distribution of Chromolaena odorata (Siam weed) in relation to climate. Weed Res 45, 246–254 (2005).
Google Scholar 
Wharton, T. N. & Kriticos, D. J. The fundamental and realized niche of the Monterey pine aphid, Essigella californica (Essig) (Hemiptera: Aphididae): implications for managing softwood plantations in Australia. Divers. Distrib. 10, 253–262 (2004).
Google Scholar 
Sutherst, R., Maywald, G. and Kriticos, D., CLIMEX version 3: user’s guide. (2007).Ramirez-Cabral, N. Y., Kumar, L. & Shabani, F. Global alterations in areas of suitability for maize production from climate change and using a mechanistic species distribution model (CLIMEX). Sci. Rep. 7, 1–3 (2017).CAS 

Google Scholar 
McCalla, K. A., Keçeci, M., Milosavljević, I., Ratkowsky, D. A. & Hoddle, M. S. The influence of temperature variation on life history parameters and thermal performance curves of Tamarixia radiata (Hymenoptera: Eulophidae), a parasitoid of the Asian citrus psyllid (Hemiptera: Liviidae). J. Econ. Entomol. 112, 1560–1574 (2019).
Google Scholar 
Gonzalez-Cabrera, J., Moreno-Carrillo, G., Sanchez-Gonzalez, J. A. & Bernal, H. C. Natural and augmented parasitism of tamarixia radiata (Hymenoptera Eulophidae) in Urban Areas of western Mexico. Entomol. Sci. 53, 486–492. https://doi.org/10.18474/JES17-112.1 (2018).Article 

Google Scholar 
Chavez, Y. et al. Tamarixia radiata (Waterston) and Cheilomenes sexmaculata (Fabricius) as biological control agents of Diaphorina citri Kuwayama in Ecuador. Chil. J. Agric. Res. 77, 180–184. https://doi.org/10.4067/S0718-58392017000200180 (2017).Article 

Google Scholar 
Flores, D. & Ciomperlik, M. Biological control using the ectoparasitoid, Tamarixia radiata, against the Asian citrus psyllid, Diaphorina citri, in the lower Rio Grande valley of Texas. Southwest. Entomol. 42, 49–59. https://doi.org/10.3958/059.042.0105 (2017).Article 

Google Scholar 
Parra, J. R., Alves, G. R., Diniz, A. J. & Vieira, J. M. Tamarixia radiata (Hymenoptera: Eulophidae) × Diaphorina citri (Hemiptera: Liviidae): Mass rearing and potential use of the parasitoid in Brazil. J. Integr. Pest. Manag. https://doi.org/10.1093/jipm/pmw003 (2016).Article 

Google Scholar 
Diniz, A. J. F., Otimização da criação de Diaphorina citri Kuwayama, 1908 (Hemiptera: Liviidae) e de Tamarixia radiata (Waterston, 1922) (Hymenoptera: Eulophidae), visando a produção em larga escala do parasitoide e avalliação do seu estabelecimento em campo. Tese (Doutorado em Entomologia)—Escola Superior de Agricultura “Luiz de Queiroz”, Universidade de São Paulo, São Paulo. (2013)Hoddle, M. S. & Pandey, R. Host range testing of Tamarixia radiata (Hymenoptera: Eulophidae) sourced from the Punjab of Pakistan for classical biological control of Diaphorina citri (Hemiptera: Liviidae: Euphyllurinae: Diaphorinini) in California. J. Econ. Entomol. 107, 125–136. https://doi.org/10.1603/EC13318 (2014).Article 

Google Scholar 
Gómez-Torres, M. L., Nava, D. E. & Parra, J. R. Thermal hygrometric requirements for the rearing and release of Tamarixia radiata (Waterston) (Hymenoptera, Eulophidae). Rev. Bras. Entomol. 58, 291–295. https://doi.org/10.1590/S0085-56262014000300011 (2014).Article 

Google Scholar 
Gómez-Torres, M. L., Nava, D. E. & Parra, J. R. Life table of Tamarixia radiata (Hymenoptera: Eulophidae) on Diaphorina citri (Hemiptera: Psyllidae) at different temperatures. J. Econ. Entomol. 105, 338–343 (2012).
Google Scholar 
Chong, J. H., Roda, A. L. & Mannion, C. M. Density and natural enemies of the Asian Citrus Psyllid, Diaphorina citri (Hemiptera: Psyllidae), in the residential landscape of Southern Florida. J. Agric. Urban Entomol. 27, 33–49. https://doi.org/10.3954/11-05.1 (2010).Article 

Google Scholar 
Pluke, R. W., Qureshi, J. A. & Stansly, P. A. Citrus flushing patterns, Diaphorina citri (Hemiptera: Psyllidae) populations and parasitism by Tamarixia radiata (Hymenoptera: Eulophidae) in Puerto Rico. Florida Entomol. 91, 36–42 (2008).
Google Scholar 
Ashraf, H. J. et al. Genetic diversity of Tamarixia radiata populations and their associated endosymbiont Wolbachia species from China. Agronomy 11, 2018 (2021).CAS 

Google Scholar 
Jung, J. M., Lee, W. H. & Jung, S. Insect distribution in response to climate change based on a model: Review of function and use of CLIMEX. Entomol. Res. 46, 223–235 (2016).
Google Scholar 
Kriticos, D. J. et al. CLIMEX Version 4, 184p (2015).
Google Scholar 
Gomez-Marco, F., Gebiola, M., Baker, B. G., Stouthamer, R. & Simmons, G. S. Impact of the temperature on the phenology of Diaphorina citri (Hemiptera: Liviidae) and on the establishment of Tamarixia radiata (Hymenoptera: Eulophidae) in urban areas in the lower Colorado Desert in Arizona. Environ. Entomol. 48, 514–523 (2019).
Google Scholar 
Vieira, J. M. Biologia em temperaturas alternantes e exigências térmicas de Diaphorina citri Kuwayama, 1908 (Hemiptera: Liviidae) e Tamarixia radiata (Waterston, 1922) (Hymenoptera: Eulophidae) visando ao seu zoneamento em regiões citrícolas do estado (Doctoral dissertation, Universidade de São Paulo).Castillo, J., Jacas, J. A., Peña, J. E., Ulmer, B. J. & Hall, D. G. Effect of temperature on life history of Quadrastichus haitiensis (Hymenoptera: Eulophidae), an endoparasitoid of Diaprepes abbreviatus (Coleoptera: Curculionidae). Biol. Control. 36, 189–196 (2006).
Google Scholar 
McFarland, C. D. & Hoy, M. A. Survival of Diaphorina citri (Homoptera: Psyllidae), and its two parasitoids, Tamarixia radiata (Hymenoptera: Eulophidae) and Diaphorencyrtus aligarhensis (Hymenoptera: Encyrtidae), under different relative humidities and temperature regimes. Fla. Entomol. 84, 227–233 (2001).
Google Scholar 
Fauvergue, X. & Quilici, S. Etude de certains parametres de la biologie de Tamarixia radiata (Waterston, 1992)(Hymenoptera: Eulophidae), ectoparasitoide primaire de Diaphorina citri Kuwayama (Hemiptera: Psyllidae) vecteur du greening des agrumes. Paris Fruits 46, 179–179 (1991).
Google Scholar 
Araújo, F. H. et al. Modelling climate suitability for Striga asiatica, a potential invasive weed of cereal crops. Crop Prot. 1(160), 106050 (2022).
Google Scholar 
Silva, D. A. & RS, Kumar L, Shabani F and Picanço MC,. Potential risk levels of invasive Neoleucinodes elegantalis (small tomato borer) in areas optimal for open-field Solanum lycopersicum (tomato) cultivation in the present and under predicted climate change. Pest Manag. Sci 73, 616–627 (2017).
Google Scholar 
Kumar, S., Neven, L. G. & Yee, W. L. Evaluating correlative and mechanistic niche models for assessing the risk of pest establishment. Ecosphere 5, 1–23. https://doi.org/10.1890/ES14-00050.1 (2014).Article 
CAS 

Google Scholar 
Kriticos, D. J. et al. CliMond: global high-resolution historical and future scenario climate surfaces for bioclimatic modelling. Methods Ecol. Evol. 1, 53–64 (2012).
Google Scholar 
Santana Júnior PA, Worldwide spatial distribution of Tuta absoluta (Lepidoptera: Gelechiidae) and its natural enemies under current and future climatic change conditions through modelling. 136 f 2019 (Tese (Doutorado em Fitotecnia) – Universidade Federal de Viçosa, 2019).
Google Scholar 
Kriticos, D. J., Maywald, G. F., Yonow, T., Zurcher, E. J., Herrmann, N. I. and Sutherst, R. W., CLIMEX Version 4: Exploring the effects of climate on plants, animals and diseases. CSIRO, Canberra.156, (2015)Ramos Aguila, L. C. et al. Temperature-dependent demography and population projection of Tamarixia radiata (Hymenoptera: Eulophidea) reared on Diaphorina citri (Hemiptera: Liviidae). J. Econ. Entomol. 113, 55–63 (2019).
Google Scholar 
Oliveira, R. C., Modelagem de nicho ecológico para Helicoverpa punctigera (Wallengren, 1860) (Lepidoptera: Noctuidae) no mundo: Potencial invasão e riscos diante das mudanças climáticas. (2021). http://www.repositorio.ufc.br/handle/riufc/61961Bazzocchi, G. G., Lanzoni, A., Burgio, G. & Fiacconi, M. R. Effects of temperature and host on the pre-imaginal development of the parasitoid Diglyphus isaea (Hymenoptera: Eulophidae). Biol. Control 26, 74–82 (2003).
Google Scholar 
Hondo, T., Koike, A. & Sugimoto, T. Comparison of thermal tolerance of seven native species of parasitoids (Hymenoptera: Eulophidae) as biological control agents against Liriomyza trifolii (Diptera: Agromyzidae) in Japan. Appl. Entomol. Zool. 41, 73–82 (2006).
Google Scholar 
Duale, A. Effect of temperature and relative humidity on the biology of the stem borer parasitoid Pediobius furvus (Gahan) (Hymenoptera: Eulophidae) for the management of stem borers. Environ. Entomol. 34, 1–5 (2005).
Google Scholar 
Ashraf, H. J. et al. Comparative transcriptome analysis of Tamarixia radiata (Hymenoptera: Eulophidae) reveals differentially expressed genes upon heat shock. Comp. Biochem. Physiol. D: Genom. Proteom. 41, 100940 (2022).CAS 

Google Scholar 
van Doan, C. et al. Natural enemies of herbivores maintain their biological control potential under short-term exposure to future CO2, temperature, and precipitation patterns. Ecol. Evol. 11, 4182–4192 (2021).
Google Scholar 
Thomson, L. J., Macfadyen, S. & Hoffmann, A. A. Predicting the effects of climate change on natural enemies of agricultural pests. Biol. Control. 52, 296–306 (2010).
Google Scholar 
Rosenblatt, A. E. & Schmitz, O. J. Climate change, nutrition, and bottom-up and top-down food web processes. Trends Ecol. Evol. 31, 965–975 (2016).
Google Scholar 
Aidoo, O. F. et al. A machine learning algorithm-based approach (MaxEnt) for predicting invasive potential of Trioza erytreae on a global scale. Ecol. Inform. 71, 101792 (2022).
Google Scholar 
Aidoo, O. F. et al. The Impact of Climate Change on Potential Invasion Risk of Oryctes monoceros Worldwide. Front. Ecol. Evol. https://doi.org/10.3389/fevo.2022.895906 (2022).Article 

Google Scholar 
Hao, M. et al. Global potential distribution of Oryctes rhinoceros, as predicted by Boosted Regression Tree model. Glob. Ecol. Conserv. 1(37), e02175 (2022).
Google Scholar 
Aidoo, O. F. et al. Model-based prediction of the potential geographical distribution of the invasive coconut mite, Aceria guerreronis Keifer (Acari: Eriophyidae) based on MaxEnt. Agric. For. Entomol. 24, 390–404 (2022).
Google Scholar  More