Ecology

Field investigation
This study was performed between from May 2017 and October 2019 at Xinsheng Aquaculture Professional Cooperative (121° 0′ 56″ N, 30° 58′ 17″ E) in Qingpu District, Shanghai, Eastern China. This region has a subtropical monsoon climate with a mean monthly air temperature of 17.6 ± 2.3 °C and mean monthly precipitation of 126.9 ± 24.6 mm.
Each RC paddy (667 m2) had a rice-growing area (80% of the total area), aquaculture area (10%) and ridge area (10%; Fig. 4A). In the aquaculture area, a 1.2 m deep ditch was dug to provide a more comfortable habitat for the crayfish and eels. The ridge had a height of 40 cm, and it was covered with a high-density polyethylene film to prevent the aquatic animals from escaping. Every May, rice (Oryza sativa L., Qing-Xiang-Ruan-Geng) seedlings were transplanted from a nursery into the paddies at a planting density of 20 × 20 cm (one seedling on each hill). Moreover, the juvenile crayfish weighing 1.5 ± 0.3 g were released into the paddies according to the standard of 45,000 juveniles per hectare, and the crayfish were allowed to self-propagate inside the rice paddies. A total of nine RC paddies were divided into three groups according to the rearing density of the eels: control group (C), low-density group (LD) and high-density group (HD) with rearing densities of 0, 6000 and 12,000 ind. ha−1, respectively. The LD and HD groups were supplemented with juvenile eels at a density of 2000 and 4000 ind. ha−1 in June 2018 and 2019. The average weights of juvenile eels in 2017, 2018 and 2019 were 21.4 ± 1.8, 24.1 ± 0.9 and 26.8 ± 1.1 g, respectively. All juvenile crayfish and eels were purchased from Shanghai Xiangsheng Aquaculture Cooperative. In the aquaculture area, floating plants, such as duckweed (Lemna minor L.) and foxtail (Myriophyllum spicatum L.), covered one-third of the water surface. The soil contained 20.6–23.7 g kg−1 of organic matter, 0.7–1.2 g kg−1 of total N and 0.31–0.37 g kg−1 of total P.
Figure 4

Photographs of the paddies in the field investigation (A) and plots in the mesocosm experiment (B).

Full size image

Only basal fertilizer was used for rice cultivation, and it contained 587 kg ha−1 of urea (46.4% N), 625 kg ha−1 of superphosphate and 150 kg ha−1 of potassium chloride. Every day, 500 g of commercial fish diet (5.83% N) was applied, and no pesticides or herbicides were used in the paddies.
In late August, the mature crayfish and eels were collected using ground cages to measure the aquatic product yields. The immature crayfish and eels were returned to the paddy fields during the collection. After the rice was harvested, the rice grains were air-dried and weighed to estimate the rice yield. The N content of the rice grains and aquatic animals was determined using the semi-micro Kjeldahl method33. Before testing, rice grains, crayfish and eels were weighted, dried at 65 °C and ground. Then, all the samples were digested with concentrated sulphuric acid (H2SO4) and hydrogen peroxide.
Water samples were collected every month during the co-culture period. Three duplicate 500 mL water samples were collected from 0 to 10 cm below the surface in the aquaculture area; the three subsamples were combined to obtain one sample per paddy. In the laboratory, the total N content of the water was analysed using UV spectrophotometry after digestion by alkaline potassium persulfate oxidation.
Soil samples were collected after the rice-planting period. In each paddy, three samples were collected from a rice-planting area of 0.25 m × 0.25 m × 0.10 m. All the soil samples were air-dried, ground, passed through a 0.15 mm sieve and digested with K2SO4–CuSO4–Se solution. Then, the semi-micro Kjeldahl method was used to test the total N content of the soil.
The N2O flux rate was measured using the static chamber method34. The size of the chamber was 1.0 m × 1.0 m × 1.0 m. The N2O samples were collected every half month between 8:30 and 10:30 AM from June to October. In each paddy, four gas samples were collected using 40 mL vacuum tubes at 10 min intervals (0, 10, 20 and 30 min after chamber closure). All samples were analysed with gas chromatography (GC 2010; Shimadzu, Kyoto, Japan). The N2O flux rate was calculated using the following equation:

$$F = rho times h times left[ {{273}/left( {{273} + T} right)} right] times {text{d}}C{text{/d}}t$$
(1)

where F is the N2O flux rate (μg N m−2 h−1); ρ, density of N2O at the standard state (μg m−3); h, height of the chamber (m); T, average temperature in the chamber during gas collection and dC/dt, concentration variation rate of N2O.
The ammonia volatilization flux was measured with a continuous airflow enclosure method35. The NH3 flux was measured every half month from 09:00 to 11:00 AM during the rice-planting period. NH3 was absorbed using boric acid, and 0.01 M H2SO4 was used to titrate the solution to determine the rate of NH3 volatilization. The ammonia volatilization flux was calculated using the following equation:

$$F = 14 times V times C times A^{ – 1} times t^{ – 1}$$
(2)

where F denotes the ammonia volatilization flux (mg N m−2 h−1); V, volume of H2SO4 titrated (L); C, concentration of H2SO4 (mol L−1); A, area of the chamber base (m2) and t, continuous measurement time.
In this study, all the data were shown as mean ± standard error of the mean (SEM) values. One-way ANOVA and Tukey’s test (SPSS V.16.0) were used to compare the differences of the yields and total N content among the three groups and three investigated years.
Mesocosm experiment
Between May and October 2019, the mesocosm experiment was conducted at Shanghai Academy of Agricultural Sciences. Each mesocosm consisted of an experimental plot (1.2 m × 1.2 m × 0.6 m) covered with a high-density polyethylene film (Fig. 4B). In each experiment plot, 30 kg of soil from Xinsheng Aquaculture Professional Cooperative was used to construct a rice-planting platform and an aquaculture ditch (40 cm in depth). The platform area was about three-fourth of the cross-sectional area of the plot.
A total of six mesocosms were constructed: three experimental plots (RCE) and three control plots (RC). In each plot, the rice seedlings were planted in hills (one seedling per hill) within rows in May, with 20 cm between rows and 20 cm between hills in the same row for the experimental and control plots. The fertilizers used in each plot contained 84.5 g of urea (N content, 46.8%; 15N abundance, 10.15%), 90 g of superphosphate and 15 g of potassium chloride. The duckweed was planted in the aquaculture area, and it covered 30% of the aquaculture zone. Mudsnails (Cipangopaludina cathayensis, 500 g) were added to each plot. After a month, 12 crayfish were cultured in each simulated paddy, and two eels were reared in each experiment plot. The proportion of crayfish and eels was set according to that in the LD group of field investigation. The crayfish feed was supplied once every day, and the daily allowance was about 3% of the estimated crayfish weight in each mesocosm. The rice and aquatic products were harvested in October.
Rice, crayfish and eel samples were collected to measure the total N content and 15N abundance. The total N content of the soil and organism samples were measured using the semi-micro Kjeldahl method after digestion with concentrated H2SO4 and hydrogen peroxide. The 15N abundance was measured in all samples by using the MAT-271 isotope mass spectrometer (Finnigan MAT, California). The accumulation of N in rice, crayfish and eels from N fertilizer was calculated using the following equations:

$${text{Percentage of accumulated N from fertilizer NDFF}});( % ) = {text{A}}% ;{text{E of the organism sample/A}}% ;{text{E of the fertilizer sample}} times 100$$
(3)

$${text{Amount of accumulated N from fertilizer}} = {text{organism N accumulation amount}} times {text{NDFF}}$$
(4)

$${text{N use efficiency NUE}});(% ) = {text{amount of N accumulated by the organism accumulated from N fertilizer/total N content of the fertilizer}} times 100$$
(5)

where A% E is the difference between the 15N abundance of the samples or 15N-labelled fertilizers and natural abundance of 15N.
The independent-samples t-test was used to determine the differences in total N, N use efficiency and percentage of N derived from fertilizer between RCE and RC at 95% confidence level by using SPSS 16.0 (P value  More

1.
Marquet, P. A. Scaling and power-laws in ecological systems. J. Exp. Biol. 208, 1749–1769 (2005).
PubMed  Google Scholar 
2.
GarciaMartin, H. & Goldenfeld, N. On the origin and robustness of power-law species-area relationships in ecology. Proc. Natl. Acad. Sci. 103, 10310–10315 (2006).
ADS  Google Scholar 

3.
Taylor, L. R. Aggregation, variance and the mean. Nature 189, 732–735 (1961).
ADS  Google Scholar 

4.
Damuth, J. Population density and body size in mammals. Nature 290, 699–700 (1981).
ADS  Google Scholar 

5.
Carbone, C. & Gittleman, J. L. A common rule for the scaling of carnivore density. Science 295, 2273–2276 (2002).
ADS  CAS  PubMed  Google Scholar 

6.
White, C. R. & Seymour, R. S. Allometric scaling of mammalian metabolism. J. Exp. Biol. 208, 1611–1619 (2005).
CAS  PubMed  Google Scholar 

7.
da Silva, J. K. L., Garcia, G. J. M. & Barbosa, L. A. Allometric scaling laws of metabolism. Phys. Life Rev. 3, 229–261 (2006).
ADS  Google Scholar 

8.
Reich, P. B., Tjoelker, M. G., Machado, J.-L. & Oleksyn, J. Universal scaling of respiratory metabolism, size and nitrogen in plants. Nature 439, 457–461 (2006).
ADS  CAS  PubMed  Google Scholar 

9.
Eisler, Z., Bartos, I. & Kertész, J. Fluctuation scaling in complex systems: Taylor’s law and beyond. Adv. Phys. 57, 89–142 (2008).
ADS  CAS  Google Scholar 

10.
Reed, D. H. & Hobbs, G. R. The relationship between population size and temporal variability in population size. Anim. Conserv. 7, 1–8 (2004).
Google Scholar 

11.
Benton, T. G. & Beckerman, A. P. Population dynamics in a noisy world: lessons from a mite experimental system. in Advances in Ecological Research vol. 37, pp. 143–181 (Academic Press, Cambridge, 2005).

12.
Ramsayer, J., Fellous, S., Cohen, J. E. & Hochberg, M. E. Taylor’s law holds in experimental bacterial populations but competition does not influence the slope. Biol. Lett. 8, 316–319 (2012).
PubMed  Google Scholar 

13.
Kaltz, O., Escobar-Páramo, P., Hochberg, M. E. & Cohen, J. E. Bacterial microcosms obey Taylor’s law: effects of abiotic and biotic stress and genetics on mean and variance of population density. Ecol. Process. 1, 5 (2012).
Google Scholar 

14.
Anderson, R. M., Gordon, D. M., Crawley, M. J. & Hassell, M. P. Variability in the abundance of animal and plant species. Nature 296, 245–248 (1982).
ADS  Google Scholar 

15.
Ballantyne, F. I. The upper limit for the exponent of Taylor’s power law is a consequence of deterministic population growth. Evol. Ecol. Res. 8 (2005).

16.
Engen, S., Lande, R. & Sæther, B.-E. A general model for analyzing taylor’s spatial scaling laws. Ecology 89, 2612–2622 (2008).
PubMed  Google Scholar 

17.
Damuth, J. Interspecific allometry of population density in mammals and other animals: the independence of body mass and population energy-use. Biol. J. Linn. Soc. 31, 193–246 (1987).
Google Scholar 

18.
Blackburn, T. M. & Gaston, K. J. The relationship between animal abundance and body size: a review of the mechanisms. In Advances in Ecological Research (eds Fitter, A. H. & Raffaelli, D.) 181–210 (Academic Press, Cambridge, 1999).
Google Scholar 

19.
Jennings, S., Oliveira, J. A. A. D. & Warr, K. J. Measurement of body size and abundance in tests of macroecological and food web theory. J. Anim. Ecol. 76, 72–82 (2007).
PubMed  Google Scholar 

20.
Belgrano, A. & Reiss, J. The Role of Body Size in Multispecies Systems (Academic Press, Cambridge, 2011).
Google Scholar 

21.
Lawton, J. H. What is the relationship between population density and body size in animals?. Oikos 55, 429–434 (1989).
Google Scholar 

22.
Marquet, P. A., Navarrete, S. A. & Castilla, J. C. Scaling oopulation density to body size in rocky intertidal communities. Science 250, 1125–1127 (1990).
ADS  CAS  PubMed  Google Scholar 

23.
Silva, M. & Downing, J. A. The allometric scaling of density and body mass: a nonlinear relationship for terrestrial mammals. Am. Nat. 145(5), 704–727 (1995).
Google Scholar 

24.
Dunham, J. B. & Vinyard, G. L. Relationships between body mass, population density, and the self-thinning rule in stream-living salmonids. Can. J. Fish. Aquat. Sci. 54, 6 (1997).
Google Scholar 

25.
Enquist, B. J., Brown, J. H. & West, G. B. Allometric scaling of plant energetics and population density. Nature 395, 4 (1998).
Google Scholar 

26.
Hendriks, A. J. Allometric scaling of rate, age and density parameters in ecological models. Oikos 86, 293–310 (1999).
Google Scholar 

27.
Schmid, P. E. Relation between population density and body size in stream communities. Science 289, 1557–1560 (2000).
ADS  CAS  PubMed  Google Scholar 

28.
Morand, S. & Poulin, R. Body size–density relationships and species diversity in parasitic nematodes: patterns and likely processes. Evol. Ecol. Res. 12 (2002).

29.
Niklas, K. J., Midgley, J. J. & Enquist, B. J. A general model for mass–growth–density relations across tree-dominated communities. Evol. Ecol. Res. 5, 459–468 (2003).
Google Scholar 

30.
Makarieva, A. M., Victor, G. & Li, B.-L. Why do population density and inverse home range scale differently with body size?. Ecol. Complex. 2, 259–271 (2005).
Google Scholar 

31.
Reuman, D. C., Mulder, C., Raffaelli, D. & Cohen, J. E. Three allometric relations of population density to body mass: theoretical integration and empirical tests in 149 food webs. Ecol. Lett. 11, 1216–1228 (2008).
PubMed  Google Scholar 

32.
Reuman, D. C. et al. Allometry of body size and abundance in 166 food webs. in Advances in Ecological Research vol. 41, pp. 1–44 (Elsevier, 2009).

33.
Carbone, C., Pettorelli, N. & Stephens, P. A. The bigger they come, the harder they fall: body size and prey abundance influence predator–prey ratios. Biol. Lett. 7, 312–315 (2011).
PubMed  Google Scholar 

34.
Cohen, J. E., Xu, M. & Schuster, W. S. F. Allometric scaling of population variance with mean body size is predicted from Taylor’s law and density-mass allometry. Proc. Natl. Acad. Sci. 109, 15829–15834 (2012).
ADS  CAS  PubMed  Google Scholar 

35.
Segura, A. M. & Perera, G. The metabolic basis of fat tail distributions in populations and community fluctuations. Front. Ecol. Evol. 7, 148 (2019).
Google Scholar 

36.
Agusti, S., Duarte, C. M. & Kalff, J. Algal cell size and the maximum density and biomass of phytoplankton1. Limnol. Oceanogr. 32, 983–986 (1987).
ADS  Google Scholar 

37.
Belgrano, A., Allen, A. P., Enquist, B. J. & Gillooly, J. F. Allometric scaling of maximum population density: a common rule for marine phytoplankton and terrestrial plants. Ecol. Lett. 5, 611–613 (2002).
Google Scholar 

38.
Brown, J. H., Gillooly, J. F., Allen, A. P., Savage, V. M. & West, G. B. Toward a metabolic theory of ecology. Ecology 85, 1771–1789 (2004).
Google Scholar 

39.
Barneche, D. R., Kulbicki, M., Floeter, S. R., Friedlander, A. M. & Allen, A. P. Energetic and ecological constraints on population density of reef fishes. Proc. R. Soc. B 283, 20152186 (2016).
PubMed  Google Scholar 

40.
Ghedini, G., White, C. R. & Marshall, D. J. Metabolic scaling across succession: do individual rates predict community-level energy use?. Funct. Ecol. 32, 1447–1456 (2018).
Google Scholar 

41.
Lagrue, C., Poulin, R. & Cohen, J. E. Parasitism alters three power laws of scaling in a metazoan community: Taylor’s law, density-mass allometry, and variance-mass allometry. Proc. Natl. Acad. Sci. 112, 1791–1796 (2015).
ADS  CAS  PubMed  Google Scholar 

42.
Xu, M. Ecological scaling laws link individual body size variation to population abundance fluctuation. Oikos 125, 288–299 (2016).
Google Scholar 

43.
Taylor, L. R. & Woiwod, I. P. Comparative synoptic dynamics. I. Relationships between inter- and intra-specific spatial and temporal variance/mean population parameters. J. Anim. Ecol. 51, 879 (1982).
Google Scholar 

44.
Cyr, H., Downing, J. A., Peters, R. H. & Cyr, H. Density-body size relationships in local aquatic communities. Oikos 79, 333 (1997).
Google Scholar 

45.
O’Connell, A. F., Nichols, J. D. & Karant, U. K. Camera traps in animal ecology methods and analyses (Springer, London, 2010).
Google Scholar 

46.
Anile, S. & Devillard, S. Study design and body mass influence RAIs from camera trap studies: evidence from the Felidae. Anim. Conserv. 19, 35–45 (2015).
Google Scholar 

47.
Anile, S. & Devillard, S. Camera-trapping provides insights into adult sex ratio variability in felids. Mamm. Rev. 48, 168–179 (2018).
Google Scholar 

48.
Wilson, E. E. & Wolkovich, E. M. Scavenging: how carnivores and carrion structure communities. Trends Ecol. Evol. 26, 129–135 (2011).
PubMed  Google Scholar 

49.
Ripple, W. J. et al. Status and ecological effects of the world’s largest carnivores. Science 343, 1241484 (2014).
Google Scholar 

50.
Van Valkenburgh, B., Hayward, M. W., Ripple, W. J., Meloro, C. & Roth, V. L. The impact of large terrestrial carnivores on Pleistocene ecosystems. Proc. Natl. Acad. Sci. 113, 862–867 (2016).
ADS  PubMed  Google Scholar 

51.
Albert, C., Luque, G. M. & Courchamp, F. The twenty most charismatic species. PLoS ONE 13, e0199149 (2018).
PubMed  PubMed Central  Google Scholar 

52.
Dickman, S. Felid conservation priorities. Conserv. Biol. (2015).

53.
Inskip, C. & Zimmermann, A. Human-felid conflict: a review of patterns and priorities worldwide. Oryx 43, 18 (2009).
Google Scholar 

54.
Macdonald, D. W. & Loveridge, A. J. The Biology and Conservation of Wild Felids (Oxford University Press, Oxford, 2010).
Google Scholar 

55.
Hunter, L. Wild Cats of the World (Bloomsbury Publishing, London, 2015).
Google Scholar 

56.
Rizzuto, M., Carbone, C. & Pawar, S. Foraging constraints reverse the scaling of activity time in carnivores. Nat. Ecol. Evol. 2, 247–253 (2018).
PubMed  Google Scholar 

57.
Karanth, K. U., Nichols, J. D., Samba Kumar, N., Link, W. A. & Hines, J. E. Tigers and their prey: predicting carnivore densities from prey abundance. Proc. Natl. Acad. Sci. 101, 4854–4858 (2004).
ADS  CAS  PubMed  Google Scholar 

58.
Jiang, G. et al. New hope for the survival of the Amur leopard in China. Sci. Rep. 15, 15475 (2015).
ADS  Google Scholar 

59.
Jedrzejewski, W. et al. Estimating large carnivore populations at global scale based on spatial predictions of density and distribution: application to the jaguar (Panthera onca). PLoS ONE 13, e0194719 (2018).
PubMed  PubMed Central  Google Scholar 

60.
Kitchener, A. C. et al. A revised taxonomy of the Felidae. The final report of the Cat Classification Task Force of the IUCN/SSC Cat Specialist Group. Cat News Special Issue, pp. 11–80 (2017).

61.
Sinclair, A. R. E. Mammal populations: fluctuation, regulation, life history theory and their implications for conservation. Front. Popul. Ecol. 1, 127–154 (1996).
Google Scholar 

62.
Santini, L., Isaac, N. J. B. & Ficetola, G. F. TetraDENSITY: a database of population density estimates in terrestrial vertebrates. Glob. Ecol. Biogeogr. 27, 787–791 (2018).
Google Scholar 

63.
Ceballos, G., Ehrlich, P. R. & Dirzo, R. Biological annihilation via the ongoing sixth mass extinction signaled by vertebrate population losses and declines. Proc. Natl. Acad. Sci. 114, E6089–E6096 (2017).
CAS  PubMed  Google Scholar 

64.
Ripple, W. J. et al. Extinction risk is most acute for the world’s largest and smallest vertebrates. Proc. Natl. Acad. Sci. 114, 10678–10683 (2017).
CAS  PubMed  Google Scholar 

65.
Wikramanayake, E. et al. A landscape-based conservation strategy to double the wild tiger population: landscape-based strategy for tiger recovery. Conserv. Lett. 4, 219–227 (2011).
Google Scholar 

66.
Stearns, S. C. The Influence of size and phylogeny on patterns of covariation among life-history traits in the mammals. Oikos 41, 173 (1983).
Google Scholar 

67.
Paemelaere, E. & Dobson, F. S. Fast and slow life histories of carnivores. Can. J. Zool. 89, 692–704 (2011).
Google Scholar 

68.
Peters, R. H. & Wassenberg, K. The effect of body size on animal abundance. Oecologia 60, 89–96 (1983).
ADS  PubMed  Google Scholar 

69.
Savage, V. M., Gillooly, J. F., Brown, J. H., West, G. B. & Charnov, E. L. Effects of Body size and temperature on population growth. Am. Nat. 163, 429–441 (2004).
PubMed  Google Scholar 

70.
Gamelon, M. et al. Influence of life-history tactics on transient dynamics: a comparative analysis across mammalian populations. Am. Nat. 184, 673–683 (2014).
PubMed  PubMed Central  Google Scholar 

71.
McCallum, J. Changing use of camera traps in mammalian field research: habitats, taxa and study types: camera trap use and development in field ecology. Mamm. Rev. 43, 196–206 (2013).
Google Scholar 

72.
Johnson, P. J. et al. Rensching cats and dogs: feeding ecology and fecundity trends explain variation in the allometry of sexual size dimorphism. R. Soc. Open Sci. 4, 170453 (2017).
ADS  CAS  PubMed  PubMed Central  Google Scholar 

73.
Efford, M. Density estimation in live-trapping studies. Oikos 106, 598–610 (2004).
Google Scholar 

74.
Luskin, M. S., Albert, W. R. & Tobler, M. W. Sumatran tiger survival threatened by deforestation despite increasing densities in parks. Nat. Commun. 8, 1783 (2017).
ADS  PubMed  PubMed Central  Google Scholar 

75.
Royle, J. A., Chandler, R. B., Sollmann, R. & Gardner, B. Spatial Capture-Recapture (Springer, New York, 2013).
Google Scholar 

76.
Jones, K. E. et al. PanTHERIA: a species-level database of life history, ecology, and geography of extant and recently extinct mammals: ecological Archives E090–184. Ecology 90, 2648 (2009).
Google Scholar 

77.
Huaranca, J. C. et al. Density and activity patterns of Andean cat and pampas cat (Leopardus jacobita and L. colocolo) in the Bolivian Altiplano. Wildl. Res. 47(1), 68–76 (2020).
Google Scholar 

78.
Tobler, M. W. & Powell, G. V. N. Estimating jaguar densities with camera traps: problems with current designs and recommendations for future studies. Biol. Conserv. 159, 109–118 (2013).
Google Scholar  More

1.
Reich, P. B., Walters, M. B. & Ellsworth, D. S. From tropics to tundra: Global convergence in plant functioning. Proc. Natl Acad. Sci. USA 94, 13730–13734 (1997).
ADS  CAS  Article  Google Scholar 
2.
Shipley, B. et al. Reinforcing loose foundation stones in trait-based plant ecology. Oecologia 180, 923–931 (2016).
ADS  Article  Google Scholar 

3.
Wright, I. J. et al. The worldwide leaf economics spectrum. Nature 428, 821–827 (2004).
ADS  CAS  Article  Google Scholar 

4.
Onoda, Y. et al. Physiological and structural tradeoffs underlying the leaf economics spectrum. N. Phytol. 214, 1447–1463 (2017).
CAS  Article  Google Scholar 

5.
Dray, S. et al. Combining the fourth-corner and the RLQ methods for assessing trait responses to environmental variation. Ecology 95, 14–21 (2014).
Article  Google Scholar 

6.
Diaz, S. et al. The global spectrum of plant form and function. Nature 529, 167–171 (2016).
ADS  CAS  Article  Google Scholar 

7.
Güsewell, S. Variation in nitrogen and phosphorus concentrations of wetland plants. Perspect. Plant Ecol. Evol. Syst. 5, 37–61 (2002).
Article  Google Scholar 

8.
Page, S. E. & Baird, A. J. Peatlands and global change: response and resilience. Annu. Rev. Environ. Resour. 41, 35–57 (2016).
Article  Google Scholar 

9.
Sutton-Grier, A. E. & Megonigal, J. P. Plant species traits regulate methane production in freshwater wetland soils. Soil Biol. Biochem. 43, 413–420 (2011).
CAS  Article  Google Scholar 

10.
Alldred, M. & Baines, S. B. Effects of wetland plants on denitrification rates: a meta-analysis. Ecol. Appl. 26, 676–685 (2016).
Article  Google Scholar 

11.
Moor, H. et al. Towards a trait-based ecology of wetland vegetation. J. Ecol. 105, 1623–1635 (2017).
Article  Google Scholar 

12.
Sutton-Grier, A. E., Wright, J. P. & Richardson, C. J. Different plant traits affect two pathways of riparian nitrogen removal in a restored freshwater wetland. Plant Soil 365, 41–57 (2013).
CAS  Article  Google Scholar 

13.
Pan, Y., Cieraad, E. & van Bodegom, P. M. Are ecophysiological adaptive traits decoupled from leaf economics traits in wetlands? Funct. Ecol. 33, 1202–1210 (2019).
Article  Google Scholar 

14.
Lambers, H., Chapin, F. S. & Pons, T. L. Plant Physiological Ecology. (Springer New York, 2008). https://doi.org/10.1007/978-0-387-78341-3.

15.
Colmer, T. D. & Voesenek, L. A. C. J. Flooding tolerance: suites of plant traits in variable environments. Funct. Plant Biol. 36, 665–681 (2009).
CAS  Article  Google Scholar 

16.
Douma, J. C., Bardin, V., Bartholomeus, R. P. & van Bodegom, P. M. Quantifying the functional responses of vegetation to drought and oxygen stress in temperate ecosystems. Funct. Ecol. 26, 1355–1365 (2012).
Article  Google Scholar 

17.
Ramsar Convention Secretariat. The Ramsar Convention Manual: a guide to the Convention on Wetlands (Ramsar, Iran, 1971). https://doi.org/10.1007/978-94-007-0551-7 (2013).

18.
Pan, Y. et al. Drivers of plant traits that allow survival in wetlands. Funct. Ecol. https://doi.org/10.1111/1365-2435.13541 (2020).

19.
Keddy, P. A. Wetland ecology: principles and conservation. (Cambridge University Press, UK, 2010).

20.
Ordoñez, J. C. et al. Leaf habit and woodiness regulate different leaf economy traits at a given nutrient supply. Ecology 91, 3218–3228 (2010).
Article  Google Scholar 

21.
Mommer, L., Lenssen, J. P. M., Huber, H., Visser, E. J. W. & De Kroon, H. Ecophysiological determinants of plant performance under flooding: a comparative study of seven plant families. J. Ecol. 94, 1117–1129 (2006).
Article  Google Scholar 

22.
Pierce, S., Brusa, G., Sartori, M. & Cerabolini, B. E. Combined use of leaf size and economics traits allows direct comparison of hydrophyte and terrestrial herbaceous adaptive strategies. Ann. Bot. 109, 1047–1053 (2012).
Article  Google Scholar 

23.
Reich, P. B., Ellsworth, D. S. & Walters, M. B. Leaf structure (specific leaf area) modulates photosynthesis-nitrogen relations: Evidence from within and across species and functional groups. Funct. Ecol. 12, 948–958 (1998).
Article  Google Scholar 

24.
Reich, P. B. et al. Relationships of leaf dark respiration to leaf nitrogen, specific leaf area and leaf life-span: a test across biomes and functional groups. Oecologia 114, 471–482 (1998).
ADS  Article  Google Scholar 

25.
Mommer, L., Pedersen, O. & Visser, E. J. W. Acclimation of a terrestrial plant to submergence facilitates gas exchange under water. Plant, Cell Environ. 27, 1281–1287 (2004).
Article  Google Scholar 

26.
Herzog, M. & Pedersen, O. Partial versus complete submergence: Snorkelling aids root aeration in Rumex palustris but not in R. acetosa. Plant, Cell Environ. 37, 2381–2390 (2014).
CAS  Google Scholar 

27.
Reich, P. B. The world-wide ‘fast-slow’ plant economics spectrum: a traits manifesto. J. Ecol. 102, 275–301 (2014).
Article  Google Scholar 

28.
Violle, C. et al. Plant functional traits capture species richness variations along a flooding gradient. Oikos 120, 389–398 (2011).
Article  Google Scholar 

29.
Colmer, T. D., Winkel, A. & Pedersen, O. A perspective on underwater photosynthesis in submerged terrestrial wetland plants. AoB Plants 11, 1–15 (2011).
Google Scholar 

30.
Westoby, M., Falster, D. S., Moles, A. T., Vesk, P. A. & Wright, I. J. Plant ecological strategies: some leading dimensions of variation between species. Annu. Rev. Ecol. Syst. 33, 125–159 (2002).
Article  Google Scholar 

31.
Cyr, H. & Face, M. L. Magnitude and patterns of herbivory in aquatic and terrestrial ecosystems. Nature 361, 148–150 (1993).
ADS  Article  Google Scholar 

32.
Hikosaka, K. & Shigeno, A. The role of Rubisco and cell walls in the interspecific variation in photosynthetic capacity. Oecologia 160, 443–451 (2009).
ADS  Article  Google Scholar 

33.
Pedersen, O., Colmer, T. D., Borum, J., Zavala-Perez, A. & Kendrick, G. A. Heat stress of two tropical seagrass species during low tides—impact on underwater net photosynthesis, dark respiration and diel in situ internal aeration. N. Phytol. 210, 1207–1218 (2016).
CAS  Article  Google Scholar 

34.
Armstrong, W. & Beckett, P. M. Experimental and modelling data contradict the idea of respiratory down-regulation in plant tissues at an internal [O2] substantially above the critical oxygen pressure for cytochrome oxidase. N. Phytol. 190, 431–441 (2011).
CAS  Article  Google Scholar 

35.
Colmer, T. D. & Pedersen, O. Underwater photosynthesis and respiration in leaves of submerged wetland plants: gas films improve CO2 and O2 exchange. N. Phytol. 177, 918–926 (2008).
CAS  Article  Google Scholar 

36.
Jackson, M. B. & Armstrong, W. Formation of aerenchyma and the processes of plant ventilation in relation to soil flooding and submergence. Plant Biol. 1, 274–287 (1999).
CAS  Article  Google Scholar 

37.
Reich, P. B. et al. Scaling of respiration to nitrogen in leaves, stems and roots of higher land plants. Ecol. Lett. 11, 793–801 (2008).
Article  Google Scholar 

38.
Pagter, M., Bragato, C. & Brix, H. Tolerance and physiological responses of Phragmites australis to water deficit. Aquat. Bot. 81, 285–299 (2005).
Article  Google Scholar 

39.
Ellenberg, H. H. Vegetation ecology of central Europe. (Cambridge University Press, 1988).

40.
Warton, D. I., Duursma, R. A., Falster, D. S. & Taskinen, S. smatr 3- an R package for estimation and inference about allometric lines. Methods Ecol. Evol. 3, 257–259 (2012).
Article  Google Scholar 

41.
Warton, D. I., Wright, I. J., Falster, D. S. & Westoby, M. Bivariate line-fitting methods for allometry. Biol. Rev. Camb. Philos. Soc. 81, 259–291 (2006).
Article  Google Scholar 

42.
Wasserstein, R. L., Schirm, A. L. & Lazar, N. A. Moving to a World Beyond “p < 0.05”. Am. Stat. 73, 1–19 (2019). MathSciNet  Article  Google Scholar  43. Nakagawa, S. & Cuthill, I. C. Effect size, confidence interval and statistical significance: a practical guide for biologists. Biol. Rev. 82, 591–605 (2007). Article  Google Scholar  44. R. Core Team. R: A language and environment for statistical computing. R Foundation Statistical Computing, Vienna, Austria. URL https://www.r-project.org/ (2018). More

1.
Dressler, R. L. In Phylogeny and Classification of the Orchid Family (ed. Dressler, R. L.) 7–13 (Cambridge University Press, Cambridge, 1994).
Google Scholar 
2.
Delforge, P. In Orchids of Europe, Nord Africa and the Middle East (ed. Delforge, P.) 67–68 (A & C Black Publishers, London, 2001).
Google Scholar 

3.
Barman, D. & Devadas, R. Climate change on orchid population and conservation strategies: a review. J. Crop Weed. 9(12), 1–12 (2013).
Google Scholar 

4.
Fay, M. F. Orchid conservation: how can we meet the challenges in the twenty-first century. Bot. Stud. 5, 1–6 (2018).
Google Scholar 

5.
Brovkin, V. Climate–vegetation interaction. J. Phys. IV FRANCE 12, 57–72 (2002).
Google Scholar 

6.
Thomas, C. D. et al. Extinction risk from climate change. Nature 427, 145–148. https://doi.org/10.1038/nature02121 (2004).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

7.
Anderson, M. G. & Ferree, C. E. Conserving the stage: climate change and the geophysical underpinnings of species diversity. PLoS ONE 5(7), e11554 (2010).
ADS  PubMed  PubMed Central  Google Scholar 

8.
Bálint, M. et al. Cryptic biodiversity loss linked to global climate change. Nat. Clim. Change. 1, 313–318. https://doi.org/10.1038/nclimate1191 (2011).
ADS  Article  Google Scholar 

9.
Kolanowska, M. Niche conservatism and the future potential range of Epipactis helleborine (Orchidaceae). PLoS ONE 8(10), e77352. https://doi.org/10.1371/journal.pone.0077352 (2013).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

10.
Kolanowska, M. The naturalization status of African Spotted Orchid (Oeceoclades maculata) in Neotropics. Plant Biosyst. 148(5), 1049–1055. https://doi.org/10.1080/11263504.2013.824042 (2014).
Article  Google Scholar 

11.
Kolanowska, M. & Konowalik, K. Niche conservatism and future changes in the potential area coverage of Arundina graminifolia, an invasive orchid species from Southeast Asia. Biotropica 46(2), 157–165. https://doi.org/10.1111/btp.12089 (2014).
Article  Google Scholar 

12.
Konowalik, K. & Kolanowska, M. Climatic niche shift and possible future spread of the invasive South African Orchid Disa bracteata in Australia and adjacent areas. PeerJ. 6, e6107. https://doi.org/10.7717/peerj.6107 (2018).
Article  PubMed  PubMed Central  Google Scholar 

13.
Kolanowska, M. et al. Global warming not so harmful for all plants – response of holomycotrophic orchid species for the future climate change. Sci. Rep. 7, 12704. https://doi.org/10.1038/s41598-017-13088-7 (2017).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

14.
Naczk, A. & Kolanowska, M. Glacial refugia and future habitat coverage of selected Dactylorhiza representatives (Orchidaceae). PLoS ONE 10(11), e0143478. https://doi.org/10.1371/journal.pone.0143478 (2015).
CAS  Article  PubMed  PubMed Central  Google Scholar 

15.
Kolanowska, M. & Rykaczewski, M. From the past to the future – glacial refugia, current distribution patterns and future potential range changes of Diodonopsis (Orchidaceae) representatives. Lankesteriana. 17(2), 315–327 (2017).
Google Scholar 

16.
Wang, H. H. et al. Species distribution modelling for conservation of an endangered endemic orchid. AoB Plants 7, plv039 (2015).
PubMed  PubMed Central  Google Scholar 

17.
Tsiftsis, S., Djordjević, V. & Tsiripidis, I. Neottia cordata (Orchidaceae) at its southernmost distribution border in Europe: threat status and effectiveness of Natura 2000 Network for its conservation. J. Nat. Conserv. 48, 27–35 (2019).
Google Scholar 

18.
Vollering, J., Schuiteman, A., de Vogel, E., van Vugt, R. & Raes, N. Phytogeography of New Guinean orchids: patterns of species richness and turnover. J. Biogeogr. 43(1), 204–214 (2016).
Google Scholar 

19.
Reina-Rodríguez, G. A., Rubiano Mejía, J. E., Castro Llanos, F. A. & Soriano, I. Orchid distribution and bioclimatic niches as a strategy to climate change in areas of tropical dry forest in Colombia. Lankesteriana 17(1), 17–47 (2017).
Google Scholar 

20.
Gogol-Prokurat, M. Predicting habitat suitability for rare plants at local spatial scales using a species distribution model. Ecol. Appl. 21, 33–47. https://doi.org/10.1890/09-1190.1 (2011).
Article  PubMed  Google Scholar 

21.
Dudley, T. L. & Bean, D. W. Tamarisk biocontrol, endangered species risk and resolution of conflict through riparian restoration. Biocontrol 57, 331–347. https://doi.org/10.1007/s10526-011-9436-9 (2012).
Article  Google Scholar 

22.
Antúnez, P. et al. The potential distribution of tree species in three periods of time under a climate change scenario. Forests 9(10), 628. https://doi.org/10.3390/f9100628 (2018).
Article  Google Scholar 

23.
Wilson, C. D., Roberts, D. & Reid, N. Applying species distribution modeling to identify areas of high conservation value for endangered species: a case study using Margaritifera margaritifera (L.). Biol. Cons. 144, 821–829 (2011).
Google Scholar 

24.
Koch, R., Almeida-Cortez, J. S. & Kleinschmit, B. Revealing areas of high nature conservation importance in a seasonally dry tropical forest in Brazil: combination of modelled plant diversity hot spots and threat patterns. J. Nat. Conserv. 35, 24–39 (2017).
Google Scholar 

25.
Spiers, J. A., Oatham, M. P., Rostant, L. V. & Farrell, A. D. Applying species distribution modelling to improving conservation based decisions: a gap analysis of Trinidad and Tobago’s endemic vascular plants. Biodivers. Conserv. 27(11), 2931–2949 (2018).
Google Scholar 

26.
Ramírez, S. R., Gravendeel, B., Singer, R. B., Marshall, C. R. & Pierce, N. E. Dating the origin of the Orchidaceae from a fossil orchid with its pollinator. Nature 448, 1042–1045 (2007).
ADS  PubMed  Google Scholar 

27.
Conran, J. G., Bannister, J. M. & Lee, D. E. Earliest orchid macrofossils: early Miocene Dendrobium and Earina (Orchidaceae: Epidendroideae) from New Zealand. Am. J. Bot. 96(2), 466–474 (2009).
PubMed  Google Scholar 

28.
Kenny, J. 2008. Orchids of Trinidad and Tobago (ed. Kenny, J.) 1–127 (Prospect Press, 2008).

29.
Swarts, N. D. & Dixon, K. W. Terrestrial orchid conservation in the age of extinction. Ann. Bot. 104(3), 543–556 (2009).
PubMed  PubMed Central  Google Scholar 

30.
Teketay, D. History, botany and ecological requirements of coffee. Walia 20, 28–50 (1999).
Google Scholar 

31.
Tupac, O. J., Ackerman, J. D. & Bayman, P. Diversity and host specificity of endophytic Rhizoctonia-like fungi from tropical orchids. Am. J. Bot. 89(11), 1852–1858 (2002).
Google Scholar 

32.
Pellegrino, G., Luca, A. & Bellusci, F. Relationships between orchid and fungal biodiversity: mycorrhizal preferences in Mediterranean orchids. Plant Biosyst. 150(2), 180–189 (2016).
Google Scholar 

33.
Suárez, J. P. & Kottke, I. Main fungal partners and different levels of specificity of orchid mycorrhizae in the tropical mountain forests of Ecuador. Lankesteriana 16(2), 299–305 (2016).
Google Scholar 

34.
Senthilkumar, S. Mycorrhizal fungi of endangered orchid species in Kolli, a part of eastern ghats, South India. Lankesteriana 7, 15–156 (2003).
Google Scholar 

35.
Pereira, O. L., Rollemberg, C. L., Borges, A. C., Matsuoka, K. & Kasuya, M. C. M. Epulorhiza epiphytica sp. nov. isolated from mycorrhizal roots of epiphytic orchids in Brazil. Mycoscience 44, 153–155 (2003).
Google Scholar 

36.
Tedersoo, L. Biogeography of mycorrhizal symbiosis (Springer, Cham, 2017).
Google Scholar 

37.
Waud, M., Brys, R., Van Landuyt, W., Lievens, B. & Jacquemyn, H. Mycorrhizal specificity does not limit the distribution of an endangered orchid species. Mol. Ecol. 26(6), 1687–1701 (2017).
CAS  PubMed  Google Scholar 

38.
van der Cingel, N. A. An atlas of orchid pollination: America, Africa, Asia and Australia (A.A. Balkema Publishers, Rotterdam, 2001).
Google Scholar 

39.
Pansarin, E. R. & Maria do Carmo, E. A. Biologia reprodutiva e polinização de duas espécies de Polystachya Hook. no Sudeste do Brasil: evidência de pseudocleistogamia em Polystachyeae (Orchidaceae). Rev. Bras. Bot. 29(3), 423–432 (2006).

40.
Chakraborty, D. et al. Selecting populations for non-analogous climate conditions using universal response functions: the case of Douglas-Fir in Central Europe. PLoS ONE 10(8), e0136357 (2015).
PubMed  PubMed Central  Google Scholar 

41.
Broennimann, O. & Guisan, A. Predicting current and future biological invasions: both native and invaded ranges matter. Biol. Lett. 4, 585–589 (2008).
PubMed  PubMed Central  Google Scholar 

42.
Abrams, M. D. Adaptations of forest ecosystems to air pollution and climate change. Tree Physiol. 31, 258–261 (2011).
PubMed  Google Scholar 

43.
Atwater, D. Z., Ervine, C. & Barney, J. N. Climatic niche shifts are common in introduced plants. Nat. Ecol. Evol. 2, 34–43 (2018).
PubMed  Google Scholar 

44.
Konowalik, K. & Kolanowska, M. Climatic niche shift and possible future spread of the invasive South African Orchid Disa bracteata in Australia and adjacent areas. PeerJ 6, e6107 (2018).
PubMed  PubMed Central  Google Scholar 

45.
Early, R. & Sax, D. F. Climatic niche shifts between species’ native and naturalized ranges raise concern for ecological forecasts during invasions and climate change. Global Ecol. Biogeogr. 23, 1356–1365 (2014).
Google Scholar 

46.
Baranow, P. & Mytnik-Ejsmont, J. Two new species of Polystachya Hook. (Orchidaceae) from Africa. Plant Syst Evol. 281, 11–16 (2009).
Google Scholar 

47.
Mytnik-Ejsmont, J. & Baranow, P. Taxonomic study of Polystachya Hook. (Orchidaceae) from Asia. Plant Syst. Evol. 290, 57–63 (2010).
Google Scholar 

48.
Russell, A. et al. Phylogenetics and cytology of a pantropical orchid genus Polystachya (Polystachyinae, Vandeae, Orchidaceae): Evidence from plastid DNA sequence data. Taxon 59(2), 389–404 (2010).
Google Scholar 

49.
McCartney, C. African affinities, part I: the surprising relationship of some of Florida’s wild orchids. Orchids 69(2), 130–139 (2010).
Google Scholar 

50.
Mytnik-Ejsmont, J. A monograph of the subtribe Polystachyinae Schltr. (Orchidaceae) (Wydawnictwo Uniwersytetu Gdańskiego, Gdańsk, 2011).
Google Scholar 

51.
GBIF Occurrence Download; https://doi.org/10.15468/dl.ks410t (2018).

52.
Phillips, S. J., Dudík, M. & Schapire, R. E. A maximum entropy approach to species distribution modeling. In: ICML ’04. Proceedings of the Twenty-First International Conference on Machine learning. 655–662 (ACM, New York, 2004).

53.
Phillips, S. J., Anderson, R. & Schapire, R. E. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190, 231–259 (2006).
Google Scholar 

54.
Elith, J. et al. A statistical explanation of MaxEnt for ecologists. Divers. Distrib. 17, 43–57 (2011).
Google Scholar 

55.
Barve, N. et al. The crucial role of the accessible area in ecological niche modeling and species distribution modeling. Ecol. Model. 222, 1810–1819 (2011).
Google Scholar 

56.
Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G. & Jarvis, A. Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978 (2005).
Google Scholar 

57.
WorldClim (version 1.4) www.worldclim.org

58.
Warren, D. L., Glor, R. E. & Turelli, M. ENMTools: a toolbox for comparative studies of environmental niche models. Ecography 33, 607–611 (2010).
Google Scholar 

59.
Chung, M. Y. et al. Comparison of genetic variation between northern and southern populations of Lilium cernuum (Liliaceae): Implications for Pleistocene refugia. PLoS ONE 13(1), e0190520. https://doi.org/10.1371/journal.pone.0190520 (2018).
CAS  Article  PubMed  PubMed Central  Google Scholar 

60.
Kim, S. H. et al. Phylogeography and ecological niche modeling reveal reduced genetic diversity and colonization patterns of skunk cabbage (Symplocarpus foetidus; Araceae) from Glacial Refugia in Eastern North America. Front. Plant Sci. 9, 648. https://doi.org/10.3389/fpls.2018.00648 (2018).
Article  PubMed  PubMed Central  Google Scholar 

61.
Moss, R. et al. Towards New Scenarios for Analysis of Emissions, Climate Change, Impacts, and Response Strategies. (Intergovernmental Panel on Climate Change, 2008)

62.
Weyant, J. et al. Report of 2.6 Versus 2.9 Watts/m2 RCPP Evaluation Panel (IPCC Secretariat, 2009).

63.
Sohel, S. I., Akhter, S., Ullah, H., Haque, E. & Rana, P. Predicting impacts of climate change on forest tree species of Bangladesh: evidence from threatened Dysoxylum binectariferum (Roxb.) Hook.f. ex Bedd. (Meliaceae). Forest 10(1), 154–160 (2016).
Google Scholar 

64.
Sony, R. K., Sen, S., Kumar, S., Sen, M. & Jayahari, K. M. Niche models inform the effects of climate change on the endangered Nilgiri Tahr (Nilgiritragus hylocrius) populations in the southern Western Ghats, India. Ecol. Eng. 120, 355–363 (2018).
Google Scholar 

65.
Mason, S. J. & Graham, N. E. Areas beneath the relative operating characteristics (ROC) and relative operating levels (ROL) curves statistical significance and interpretation. Q. J. R. Meteorol. Soc. 128, 2145–2166 (2002).
ADS  Google Scholar 

66.
Evangelista, P. H. et al. Modelling invasion for a habitat generalist and a specialist plant species. Divers. Distrib. 14, 808–817 (2008).
Google Scholar 

67.
Allouche, O., Tsoar, A. & Kadmon, R. Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). J. Appl. Ecol. 43, 1223–1232 (2006).
Google Scholar 

68.
Hijmans, R. J., Phillips, S., Leathwick, J. & Elith J. Dismo: Species Distribution Modeling. R package version 1.1-4. https://cran.r-project.org/package=dismo (2017)

69.
Phillips, S. B., Aneja, V. P., Kang, D. & Arya, S. P. Modelling and analysis of the atmospheric nitrogen deposition in North Carolina. Int. J. Glob. Environ. Issues 6, 231–252 (2006).
Google Scholar 

70.
Warren, D. L., Glor, R. E. & Turelli, M. Environmental nicheequivalency versus conservatism: quantitative approaches toniche evolution. Evolution 62, 2868–2883. https://doi.org/10.1111/evo.2008.62.issue-11 (2008).
Article  PubMed  Google Scholar 

71.
Schoener, T. W. The Anolis lizards of Bimini: resource partitioning in a complex fauna. Ecology 49, 704–726. https://doi.org/10.2307/1935534 (1968).
Article  Google Scholar 

72.
Heibl, C. & Calenge, C. Phyloclim: integrating phylogenetics and climatic Niche modeling. R package version 0.9-4 https://cran.rproject.org/web/packages/phyloclim/phyloclim.pdf (2015).

73.
Kremen, C. et al. Aligning conservation priorities across taxa in Madagascar with high-resolution planning tools. Science 320, 222–226 (2008).
ADS  CAS  PubMed  Google Scholar 

74.
Leps, J. & Smilauer, P. Multivariate Analysis of Ecological Data Using CANOCO (Cambridge University Press, Cambridge, 2003).
Google Scholar 

75.
Peterson, A. T. et al. Ecological Niches and Geographic Distributions (MPB-49) (Princeton University Press, Princeton, 2011).
Google Scholar  More

In-flight dosing system
The system used in this work is detailed in Fig. 1. It can be broken down into three primary modules: (1) a “coarse tracking” system that uses a pair of stereoscopic cameras to identify the three dimensional location of a target subject, which is passed on to (2) the “fine tracking” system that uses a single higher speed camera and a fast scanning mirror (FSM) to keep the target in the middle of the field of view (FOV) of the camera using a proportional-integral-derivative (PID) control loop, and (3) the laser dosing system that fires the laser pulse, which is co-aligned with the fine tracking system to ensure the laser pulse is accurately applied to the subject even while it is moving. For both the coarse and fine tracking systems, subjects are identified by the size of their silhouettes generated from near infrared LED back-illumination or reflection. As demonstrated in Fig. 1b, the subject cages were 20 cm cubes constructed from clear acrylic, but with the side facing the tracking and dosing systems made of borosilicate glass, placed two meters away from the FSM. A high-speed video camera (Vision Research Phantom) was set up to record a small subset of experiments at 2000 frames per second as well. For all experiments, each cage contained only a single subject to be illuminated, or “dosed,” with the laser, and conditions were set such that a subject could be dosed only when it was at least 2.5 to 5 cm away from any wall of the cage to ensure the subject was flying normally, rather than taking wing or alighting.
Figure 1

Schematics and images of in-flight dosing setup. (a) Schematic (not to scale) of primary dosing system components and communication lines among them. Note that the mirror control and 3D position subsystems were run on separate kernels within the same PC. (b) Image of complete system setup including the test cage location, LED backlighting, and control electronics. Cameras for coarse and fine tracking, the fast scanning mirror (FSM), and dosing laser path are contained in the white circle and better seen in (c) close-up image of core tracking and dosing system components and alignment among them. Red line represents fine tracking image path, green line the dosing laser path (laser not shown), and yellow line the combined fine tracking and dosing laser path.

Full size image

Prior to commencing the laser dosing experiments, a number of parameters were defined to analyze the results, as detailed in Table 1 and Fig. 2. The key outcome, in line with our previous study15 and with WHO guidelines on insecticide treatments17, was whether the subjects were alive or disabled (i.e. dead or moribund) 24 h after the treatment. To characterize how well the subjects were tracked during the laser pulse, the tracking error was defined as how far the insect’s centroid was from the center of the fine tracking camera’s FOV. Other parameters defined in Table 1 relate to the “occlusion factor,” or how much of the laser beam’s energy (assuming a Gaussian profile) overlapped with the target’s outline, as demonstrated in Fig. 2. From the coarse tracking system’s output of xyz position of the target, we could also define velocity, speed, and both linear and angular acceleration of the subjects before, during, and after the laser pulse.
Table 1 Definitions of all parameters tracked or calculated for each in-flight dosing event.
Full size table

Figure 2

Representative fine tracking camera images of A. stephensi silhouettes. In each frame, the approximate outline of the thorax and abdomen is drawn (thick black lines) according to a set pixel intensity threshold. The centroid of this region is then calculated (intersection of red crosshairs) and compared with the center of the camera’s field of view (green dot) to determine the current tracking error and provide input to the fine tracking PID loop controlling the direction of the scanning mirror. The green circle around the green dot represents the spot size of the laser (2.5 mm diameter for all images shown here); occlusion factor represents how much of this laser spot (assuming a Gaussian profile) overlaps with the body of the subject. The images in (a) demonstrate a typical time course for a single subject in 1 ms intervals, and those in (b) show representative depictions of tracking errors ranging from 1 to 5 pixels for various subjects.

Full size image

Initial results with 532 nm laser
Initial experiments were conducted with A. stephensi subjects and a 532 nm laser (Verdi, Coherent) using parameters for power (3 W), pulse duration (25 ms), and laser spot diameter (2.5 mm) that were defined in our previous work15 as an optimal combination of potential cost and mortality performance. As seen in Supplemental Video 1, the system could be quite effective in disabling a flying mosquito with these parameters. Of note from the video, along with flight tracking data for numerous trials (not shown), the 25 ms pulse duration appeared to be short enough that the mosquitoes did not perceptibly alter their flight pattern during the pulse to throw off the tracking algorithm. From Fig. 3a,b, though, the initial system did not have consistent enough tracking performance, such that survival of the subjects was almost entirely dictated by how well the fine tracking system kept the target near the center of its FOV. From this initial experiment on a sample of 80 subjects, we set limits for mean and maximum tracking error during the laser pulse as 2 and 3 pixels, respectively, where each pixel represents ~ 250 μm. Figure 3c,d demonstrates that the system was able to achieve and maintain this performance after modifying the PID loop parameters, and that there was no longer any association between tracking performance and mortality at these conditions. These criteria were evaluated and assured for all mortality data reported in this manuscript (i.e. a Kruskal–Wallis test indicated no significant differences in tracking errors among subjects that survived or were disabled by the laser pulses). Further, Supplemental Fig. S1 shows that flight behavior, in particular speed and linear acceleration, did correlate with tracking accuracy, but that the system was able to meet the noted tracking requirements even at the extremes of flight behavior that could be observed in this study given the restricted flight volumes (though the typical values seen here largely align with available data for other Anopheles mosquitoes18,19).
Figure 3

Influence of mean and maximum tracking errors on mortality outcomes. Initial mortality outcomes using LD90 conditions with a 532 nm laser from previous anesthetized work showed strong correlation with (a) mean and (b) maximum tracking errors over the 25 ms pulse duration. After setting and achieving new tracking error targets of less than 2 pixels mean and 3 pixels max (see Fig. 2b for depictions of overlap between the laser spot and the subject for various error magnitudes), identical experiments no longer showed any correlation with (c) mean or (d) maximum tracking errors. Moribund and dead outcomes were grouped in (c) and (d), labeled “Disabled,” and subsequent experiments since they both represent functional kills and are grouped as such in WHO guidelines for insecticide trials.

Full size image

Mortality results were analyzed as in Fig. 4. Each data point represents the mortality outcomes (i.e. percentage of targets that were dead or moribund at 24 h, assuming acceptable control mortality  2 × that of the other experiments at 445 or 532 nm. Note that this greater LD90 fluence value for the smaller spot still represented a lower pulse energy than required for the larger spots given the spot size differential.
Table 2 List of in-flight dosing experimental conditions and resulting LD90 fluence values for A. stephensi subjects.
Full size table

Figure 5

Dose–response curves for all experiments from Table 2, other than the two constant pulse energy tests. The x-axis is plotted on a log scale to allow clear visualization of the curves at both low and high fluence values. Individual data points and error bars not shown for the sake of clarity. The intersections of the dashed horizontal line at 90% mortality with the dose–response curves indicate the LD90 points, which are also presented in Table 2, along with pseudo R2 values for the logistic regression fits.

Full size image

Dosing with near infrared wavelengths
Two near infrared (NIR) laser sources were examined as well—a custom constructed 1,064 nm (1 μm) fiber laser with a maximum output of 30 W and a commercial 1,570 nm (1.5 μm) fiber laser (IPG Photonics) with a maximum deliverable output of 11 W. The right-hand side of Fig. 5 shows the dose–response curves for these laser sources in addition to those using visible wavelengths. All of the infrared experiments resulted in significantly higher LD90 fluence levels compared with the visible lasers, and given the log scale of Fig. 5, varying experimental conditions with the 1 μm source led to comparatively larger changes in LD90 compared with the visible sources. As with the blue diode, the small spot (1.5 mm) had the highest LD90 fluence, but unlike the visible lasers, the larger spot (4 mm) offered a lower LD90 fluence compared with the baseline 2.5 mm. For the 1.5 μm source, there was insufficient mortality at the maximum power available from this source with a 2.5 mm spot and 25 ms pulse duration, so Fig. 5 shows a curve using slightly longer pulse durations to make up the additional fluence required. From the available data, it appears that the LD90 fluence levels for the 1 μm and 1.5 μm sources are at least at reasonably comparable levels.
Effects of longer pulse durations with lower power
We then further explored the effects of increasing the pulse duration to determine whether this could relax the optical power required from the laser source, given that laser cost is correlated with output power. Figure 6a,b show the results for the 532 nm and 1,064 nm sources, respectively, with spot diameters of 2.5 mm and pulse duration set to 25, 50, or 100 ms. In all cases, the optical power was adjusted according to the pulse duration to supply a constant pulse energy, and therefore fluence at approximately the LD75 level determined from previous experiments, which was chosen to ensure a low likelihood of any data point being 100% mortality (i.e., a saturated signal). As evidenced from Fig. 6a and Table 3, at 532 nm there was a significant drop off in mortality at the longer pulse durations, although the effect is smaller going from 50 to 100 ms compared with the initial drop from 25 to 50 ms. Similar, but smaller magnitude effects were seen with the 1 μm source in Fig. 6b and Table 3. Supplemental Fig. S2 shows that the tracking accuracy for these experiments somewhat mirrors the mortality performance, with a notable decrement from 25 to 50 ms but no significant change from 50 to 100 ms.
Figure 6

Mortality results for experiments with constant pulse energy achieved by proportionally adjusting the power according to pulse duration (25, 50, and 100 ms). Dosing was performed with both the (a) 532 nm and (b) 1,064 nm lasers, both using a 2.5 mm spot size, and set to the approximate LD75 fluence value from the dose–response curves (solid curves with dashed line 95% confidence intervals of logistic regression fit) from the corresponding 25 ms experiments. Error bars on individual points represent 95% confidence intervals of exact binomial probabilities. Statistical differences among mortality at the various pulse durations summarized in Table 3. The slight offsets on the x axis in (a) stem from slight changes in the beam spot size as a function of power.

Full size image

Table 3 Results of χ2 tests for mortality equivalence among experiments with constant pulse energy but varied amounts of power and pulse duration (25, 50, and 100 ms).
Full size table

Comparing dosing performance across other insects
Using the same system and procedures, experiments were also run on SWD and ACP subjects as a way of cursorily exploring laser-insect interaction for species of interest besides A. stephensi. For SWD, a full dose–response curve was created for the 1 μm laser with typical exposure settings (2.5 mm spot diameter and 25 ms pulse duration). The LD90 fluence from this work was 8.6 J/cm2, compared with 12.9 J/cm2 for A. stephensi under the same conditions. Tracking performance on SWD subjects was comparable to that for A. stephensi, given that flight parameters such as speed and acceleration were comparable as well. With ACP subjects, inducing them to fly in the cages was very difficult. As such, we could not process a sufficient number of subjects to create a proper dose–response curve. Based on the limited data available, and as demonstrated in Supplemental Video 2, the system as set for the mosquito LD90 condition with the 1 μm laser, 2.5 mm spot diameter and 25 ms pulse duration was effective in tracking and disabling the ACP subjects, despite their very different flying behavior (greater speeds and accelerations, limited durations) relative to the other test species.
Longer range dosing system demonstration
As a final proof of principle test, a long-range version of the system from Fig. 1 was configured. This system worked at a distance of 30 m instead of 2 m, facilitated primarily by modifying the optics used for the coarse and fine tracking systems. Given the reduced flexibility of this system, it was only used with the 1 μm laser with a 2.5 mm spot diameter and 25 ms pulse duration. Rather than acquiring new dose–response curves, which was impractical given the setup’s location in a building ~ 30 km away from the insect-housing chamber, we verified the efficacy of the system using the LD90 conditions established by short-range dosing. Supplemental Video 3 shows this system dosing an A. stephensi subject from the same stock as used for short-range testing, and Supplemental Video 4 shows the same for a wild-sourced Culex pipiens mosquito obtained in a local pond. Note that the wild C. pipiens was substantially larger in size than the lab-reared A. stephensi. In all of the limited number of tests of this system for both species (10 A. stephensi and 5 C. pipiens), the mortality rate was 100%. More

1.
Yang, J., Kloepper, J. W. & Ryu, C. M. Rhizosphere bacteria help plants tolerate abiotic stress. Trends Plant Sci. 14, 1–4 (2009).
ADS  CAS  PubMed  Google Scholar 
2.
Kearl, J. et al. Salt-tolerant halophyte rhizosphere bacteria stimulate growth of alfalfa in salty soil. Front. Microbiol. 10, 1849. https://doi.org/10.3389/fmicb.2019.01849 (2019).
Article  PubMed  PubMed Central  Google Scholar 

3.
Khan, N. et al. Comparative physiological and metabolic analysis reveals a complex mechanism involved in drought tolerance in Chickpea (Cicer arietinum L.) induced by PGPR and PGRs. Sci. Rep. 9, 2097. https://doi.org/10.1038/s41598-019-38702-8 (2019).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

4.
Compant, S., Samad, A., Faist, H. & Sessitsch, A. A review on the plant microbiome: ecology, functions, and emerging trends in microbial application. J. Adv. Res. 19, 29–37 (2019).
CAS  PubMed  PubMed Central  Google Scholar 

5.
Bruto, M., Prigent-Combaret, C., Muller, D. & Moënne-Loccoz, Y. Analysis of genes contributing to plant-beneficial functions in plant growth-promoting rhizobacteria and related Proteobacteria. Sci. Rep. 4, 6261. https://doi.org/10.1038/srep06261 (2014).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

6.
Timmusk, S. Perspectives and challenges of microbial application for crop improvement. Front. Plant Sci. 8, 49. https://doi.org/10.3389/fpls.2017.00049 (2017).
Article  PubMed  PubMed Central  Google Scholar 

7.
Sessitsch, A., Pfaffenbichler, N. & Mitter, B. Microbiome applications from lab to field: facing complexity. Trends Plant Sci. 24, 194–198 (2019).
CAS  PubMed  Google Scholar 

8.
Olanrewaju, O. S., Glick, B. R. & Babalola, O. O. Mechanisms of action of plant growth promoting bacteria. World J. Microb. Biot 33, 197. https://doi.org/10.1007/s11274-017-2364-9 (2017).
CAS  Article  Google Scholar 

9.
Gupta, A. et al. Whole genome sequencing and analysis of plant growth promoting bacteria isolated from the rhizosphere of plantation crops coconut, cocoa and arecanut. PLoS ONE 9, e104259. https://doi.org/10.1371/journal.pone.0104259 (2014).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

10.
Pérez-Jaramillo, J. E., Carrión, V. J., de Hollander, M. & Raaijmakers, J. M. The wild side of plant microbiomes. Microbiome 6, 143. https://doi.org/10.1186/s40168-018-0519-z (2018).
Article  PubMed  PubMed Central  Google Scholar 

11.
Song, J. Y. et al. Genome sequence of the Plant Growth-Promoting Rhizobacterium Bacillus sp. strain JS. J. Bacteriol. 19, 14. https://doi.org/10.1128/JB.00676-12 (2012).
CAS  Article  Google Scholar 

12.
Duan, J., Jiang, W., Cheng, Z., Heikkila, J. J. & Glick, B. R. The complete genome sequence of the Plant Growth-Promoting bacterium Pseudomonas sp. UW4. PLoS ONE 8, e58640. https://doi.org/10.1371/journal.pone.0058640 (2013).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

13.
Li, S. et al. Complete genome sequence of Paenibacillus polymyxa SQR-21, a plant growth-promoting rhizobacterium with antifungal activity and rhizosphere colonization ability. Genome Announc. 2, e00281-e314. https://doi.org/10.1128/genomeA.00281-14 (2014).
Article  PubMed  PubMed Central  Google Scholar 

14.
Fierer, N. Embracing the unknown: disentangling the complexities of the soil microbiome. Nat. Rev. Microbiol. 15, 579–590 (2017).
CAS  PubMed  Google Scholar 

15.
Penrose, D. M. & Glick, B. R. Methods for isolating and characterizing ACC deaminase-containing plant growth-promoting rhizobacteria. Physiol. Plantarum 118, 10–15 (2003).
CAS  Google Scholar 

16.
Lo, K. J., Lin, S. S., Lu, C. W., Kuo, C. H. & Liu, C. T. Whole-genome sequencing and comparative analysis of two plant-associated strains of Rhodopseudomonas palustris (PS3 and YSC3). Sci. Rep. 8, 12769. https://doi.org/10.1038/s41598-018-31128-8 (2018).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

17.
Liu, W. et al. Whole genome analysis of halotolerant and alkalotolerant plant growth-promoting rhizobacterium Klebsiella sp. D5A. Sci. Rep. 6, 20–22 (2016).
Google Scholar 

18.
Andrés-Barrao, C. et al. Complete genome sequence analysis of Enterobacter sp. SA187, a plant multi-stress tolerance promoting endophytic bacterium. Front. Microbiol. 8, 2023. https://doi.org/10.3389/fmicb.2017.02023 (2017).
Article  PubMed  PubMed Central  Google Scholar 

19.
Esmaeel, Q. et al. Draft genome sequence of plant growth-promoting Burkholderia sp. strain BE12, isolated from the rhizosphere of maize. Genome Announc. 6, e00299-18. https://doi.org/10.1128/genomeA.00299-18 (2018).
Article  PubMed  PubMed Central  Google Scholar 

20.
Wang, Z. et al. Draft genome analysis offers insights into the mechanism by which Streptomyces chartreusis WZS021 increases drought tolerance in sugarcane. Front. Microbiol. 10, 3262. https://doi.org/10.3389/fmicb.2018.03262 (2019).
Article  Google Scholar 

21.
Matteoli, F. P. et al. Genome sequencing and assessment of plant growth-promoting properties of a Serratia marcescens strain isolated from vermicompost. BMC Genomics 19, 750. https://doi.org/10.1186/s12864-018-5130-y (2018).
CAS  Article  PubMed  PubMed Central  Google Scholar 

22.
Yu, Z. et al. Complete genome sequence of the nitrogen-fixing bacterium Azospirillum humicireducens type strain SgZ-5T. Stand. Genomic Sci. 13, 28. https://doi.org/10.1186/s40793-018-0322-2 (2018).
CAS  Article  PubMed  PubMed Central  Google Scholar 

23.
Westerman, R. L. Soil Testing and Plant Analysis. SSSA Book Series 3 3rd edn. (Madison, SSSA, 1990).
Google Scholar 

24.
Bharti, N., Pandey, S. S., Barnawal, D., Patel, V. K. & Kalra, A. Plant growth promoting rhizobacteria Dietzia natronolimnaea modulates the expression of stress responsive genes providing protection of wheat from salinity stress. Sci. Rep. 6, 34768. https://doi.org/10.1038/srep34768 (2016).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

25.
Sharma, S., Kulkarni, J. & Jha, B. Halotolerant rhizobacteria promote growth and enhance salinity tolerance in peanut. Front. Microbiol. 7, 1600. https://doi.org/10.3389/fmicb.2016.01600 (2016).
Article  PubMed  PubMed Central  Google Scholar 

26.
Gupta, S. & Pandey, S. ACC Deaminase producing bacteria with multifarious plant growth promoting traits alleviates salinity stress in French Bean (Phaseolus vulgaris) plants. Front. Microbiol. 10, 1506. https://doi.org/10.3389/fmicb.2019.01506 (2019).
Article  PubMed  PubMed Central  Google Scholar 

27.
Pan, J. et al. The growth promotion of two salt-tolerant plant groups with PGPR inoculation: a meta-analysis. Sustainability 11, 378. https://doi.org/10.3390/su11020378 (2019).
CAS  Article  Google Scholar 

28.
Vurukonda, S. S. K. P., Vardharajula, S., Shrivastava, M. & SkZ, A. Multifunctional Pseudomonas putida strain FBKV2 from arid rhizosphere soil and its growth promotional effects on maize under drought stress. Rhizosphere 1, 4–13 (2016).
Google Scholar 

29.
Marasco, R. et al. Salicornia strobilacea (synonym of Halocnemum strobilaceum) grown under different tidal regimes selects rhizosphere bacteria capable of promoting plant growth. Front. Microbiol. 7, 1286. https://doi.org/10.3389/fmicb.2016.01286 (2016).
Article  PubMed  PubMed Central  Google Scholar 

30.
Mukhtar, S. et al. Impact of soil salinity on the microbial structure of halophyte rhizosphere microbiome. World J. Microb. Biot. 34, 136. https://doi.org/10.1007/s11274-018-2509-5 (2018).
CAS  Article  Google Scholar 

31.
Patten, C. L. & Glick, B. R. Role of Pseudomonas putida indoleacetic acid in development of the host plant root system Appl. Environ. Microbiol. 68(3795), 3801 (2002).
Google Scholar 

32.
Etesami, H., Alikhani, H. A. & Hosseini, H. M. Indole-3-acetic acid (IAA) production trait, a useful screening to select endophytic and rhizosphere competent bacteria for rice growth promoting agents. MethodsX 2, 72–78 (2015).
PubMed  PubMed Central  Google Scholar 

33.
Abbamondi, G. R. et al. Plant growth-promoting effects of rhizospheric and endophytic bacteria associated with different tomato cultivars and new tomato hybrids. Chem. Biol. Technol. Agric. 3, 1. https://doi.org/10.1186/s40538-015-0051-3 (2016).
CAS  Article  Google Scholar 

34.
Gupta, S. & Pandey, S. Unravelling the biochemistry and genetics of ACC deaminase-An enzyme alleviating the biotic and abiotic stress in plants. Plant gene 18, 100175. https://doi.org/10.1016/j.plgene.2019.100175 (2019).
CAS  Article  Google Scholar 

35.
Khamna, S., Yokota, A., Peberdy, J. F. & Lumyong, S. Indole-3-acetic acid production by Streptomyces sp. isolated from some Thai medicinal plant rhizosphere soils. EurAsian J. BioSciences 4, 23–32 (2010).
CAS  Google Scholar 

36.
Spaepen, S. & Vanderleyden, J. Auxin and plant-microbe interactions. C. S. H. Perspect. Biol. 3, a001438. https://doi.org/10.1101/cshperspect.a001438 (2011).
CAS  Article  Google Scholar 

37.
Patten, C. L. & Glick, B. R. Bacterial biosynthesis of indole-3-acetic acid. Can. J. Microbiol. 42, 207–220 (1996).
CAS  PubMed  Google Scholar 

38.
Majeed, A., Abbasi, K. M., Hameed, S., Imran, A. & Rahim, N. Isolation and characterization of plant growth-promoting rhizobacteria from wheat rhizosphere and their effect on plant growth promotion. Front. Microbiol. 6, 198. https://doi.org/10.3389/fmicb.2015.00198 (2015).
Article  PubMed  PubMed Central  Google Scholar 

39.
Apine, O. A. & Jadhav, J. P. Optimization of medium for indole-3-acetic acid production using Pantoea agglomerans strain PVM. J. Appl. Microbiol. 110, 1235–1244 (2011).
CAS  PubMed  Google Scholar 

40.
Checcucci, A. et al. Role and regulation of ACC deaminase gene in Sinorhizobium melilotr: Is it a symbiotic, rhizospheric or endophytic gene?. Front. Genet. 8, 6. https://doi.org/10.3389/fgene.2017.00006 (2017).
CAS  Article  PubMed  PubMed Central  Google Scholar 

41.
Belimov, A. A. et al. Cadmium-tolerant plant growth-promoting bacteria associated with the roots of Indian mustard (Brassica juncea L. Czern.). Soil Biol. Biochem. 37, 241–250 (2005).
CAS  Google Scholar 

42.
Khan, N. A., Khan, M. I. R., Ferrante, A. & Poor, P. Editorial: ethylene: a key regulatory molecule in plants. Front. Plant Sci. 8, 1782. https://doi.org/10.3389/fpls.2017.01782 (2017).
Article  PubMed  PubMed Central  Google Scholar 

43.
Glick, B. R. Bacteria with ACC deaminase can promote plant growth and help to feed the world. Microbiol. Res. 169, 30–39 (2014).
CAS  PubMed  Google Scholar 

44.
Wang, Z. et al. Isolation and characterization of a phosphorus-solubilizing bacterium from rhizosphere soils and its colonization of Chinese Cabbage (Brassica campestris ssp. chinensis). Front. Microbiol. 8, 1270. https://doi.org/10.3389/fmicb.2017.01270 (2017).
Article  PubMed  PubMed Central  Google Scholar 

45.
Dinesh, R. et al. Isolation, characterization, and evaluation of multi-trait plant growth promoting rhizobacteria for their growth promoting and disease suppressing effects on ginger. Microbiol. Res. 173, 34–43 (2015).
PubMed  Google Scholar 

46.
Niu, X., Song, L., Xiao, Y. & Ge, W. Drought-tolerant plant growth-promoting rhizobacteria associated with foxtail millet in a semi-arid agroecosystem and their potential in alleviating drought stress. Front. Microbiol. 8, 2580. https://doi.org/10.3389/fmicb.2017.02580 (2018).
Article  PubMed  PubMed Central  Google Scholar 

47.
Singh, V. K., Singh, A. K., Singh, P. P. & Kumar, A. Interaction of plant growth promoting bacteria with tomato under abiotic stress: a review. Agricult. Ecosyst. Environ. 267, 129–140 (2018).
CAS  Google Scholar 

48.
Grover, M., Bodhankar, S., Maheswari, M. & Srinivasarao, C. Actinomycetes as mitigators of climate change and abiotic stress. In Plant Growth Promoting Actinobacteria: A New Avenue for Enhancing the Productivity and Soil Fertility of Grain Legumes (eds Subramaniam, G. et al.) 203–212 (Springer, Berlin, 2016).
Google Scholar 

49.
Sang, M. K., Jeong, J. J., Kim, J. & Kim, K. D. Growth promotion and root colonisation in pepper plants by phosphate-solubilising Chryseobacterium sp strain ISE14 that suppresses Phytophthora blight. Ann. Appl. Biol. 172, 208–223 (2018).
CAS  Google Scholar 

50.
Xiao, X., Fan, M., Wang, E., Chen, W. & Wei, G. Interactions of plant growth-promoting rhizobacteria and soil factors in two leguminous plants. Appl. Microbiol. Biot. 101, 8485–8497 (2017).
CAS  Google Scholar 

51.
Abdelkrim, S. et al. Effect of Pb-resistant plant growth-promoting rhizobacteria inoculation on growth and lead uptake by Lathyrus sativus. J. Basic Microb. 58, 579–589 (2018).
CAS  Google Scholar 

52.
Liu, Y. et al. Characterization of Lysobacter capsici strain NF87–2 and its biocontrol activities against phytopathogens. Eur. J. Plant Pathol. 155, 859–869 (2019).
CAS  Google Scholar 

53.
Edgar, R. C. Accuracy of taxonomy prediction for 16S rRNA and fungal ITS sequences. PeerJ. https://doi.org/10.7717/peerj.4652 (2018).
Article  PubMed  PubMed Central  Google Scholar 

54.
Lee, Z. M., Bussema, C. III. & Schmidt, T. M. rrnDB: documenting the number of rRNA and tRNA genes in bacteria and archaea. Nucleic Acids Res. 37, D489–D493 (2008).
PubMed  PubMed Central  Google Scholar 

55.
Shen, X., Hu, H., Peng, H., Wang, W. & Zhang, X. Comparative genomic analysis of four representative plant growth-promoting rhizobacteria in Pseudomonas. BMC Genomics 14, 271. https://doi.org/10.1186/1471-2164-14-271 (2013).
CAS  Article  PubMed  PubMed Central  Google Scholar 

56.
Niazi, A. et al. Genome analysis of Bacillus amyloliquefaciens subsp. Plantarum UCMB5113: a rhizobacterium that improves plant growth and stress management. PloS ONE 9, e104651. https://doi.org/10.1371/journal.pone.0104651 (2014).
ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

57.
Riggs, P. J., Chelius, M. K., Leonardo, I. A., Kaeppler, S. M. & Triplett, E. W. Enhanced maize productivity by inoculation with diazotrophic bacteria. Funct. Plant Biol. 28, 829–836 (2001).
Google Scholar 

58.
Saleem, M., Arshad, M., Hussain, S. & Bhatti, A. S. Perspective of plant growth promoting rhizobacteria (PGPR) containing ACC deaminase in stress agriculture. J. Ind. Microbiol. Biot. 34, 635–648 (2007).
CAS  Google Scholar 

59.
Dorjey, S., Gupta, V. & Razdan, V. K. Evaluation of Pseudomonas fluorescens isolates for the management of Fusarium oxysporum f.sp. lycopersici and Rhizoctonia solani causing wilt complex in tomato. Indian Phytopathol. 70, 127–130 (2017).
Google Scholar 

60.
Dardanelli, M. S. et al. Changes in flavonoids secreted by Phaseolus vulgaris roots in the presence of salt and the plant growth-promoting rhizobacterium Chryseobacterium balustinum. Appl. Soil Ecol. 75, 31–38 (2012).
Google Scholar 

61.
Ogut, M., Er, F. & Brohi, A. Excessive phosphorus fertilization does not increase cadmium concentrations in soil or carrots (Daucus carota L.) grown in Konya (Turkey). Acta Agr. Scand B-S.P. 60, 420–426 (2010).
CAS  Google Scholar 

62.
Rodríguez, H., Fraga, R., Gonzalez, T. & Bashan, Y. Genetics of phosphate solubilization and its potential applications for improving plant growth-promoting bacteria. Plant Soil 287, 15–21 (2006).
Google Scholar 

63.
Kalayu, G. Phosphate solubilizing microorganisms: Promising approach as biofertilizers. Int. J. Agron. 2019, 4917256. https://doi.org/10.1155/2019/4917256 (2019).
CAS  Article  Google Scholar 

64.
Mellidou, I. et al. Silencing S-adenosyl-L-methionine decarboxylase (SAMDC) in Nicotiana tabacum points at a polyamine-dependent trade-off between growth and tolerance responses. Front. Plant Sci. 7, 379. https://doi.org/10.3389/fpls.2016.00379 (2016).
Article  PubMed  PubMed Central  Google Scholar 

65.
Michael, A. J. Biosynthesis of polyamines and polyamine-containing molecules. Biochem. J. 473, 2315–2329 (2016).
CAS  PubMed  Google Scholar 

66.
Nascimento, F. X., Hernández, A. G., Glick, B. R. & Rossi, M. J. Plant growth-promoting activities and genomic analysis of the stress-resistant Bacillus megaterium STB1, a bacterium of agricultural and biotechnological interest. Biotechnol. Rep. 25, e00406. https://doi.org/10.1016/j.btre.2019.e00406 (2020).
Article  Google Scholar 

67.
Goswami, R. et al. Optimization of growth determinants of a potent cellulolytic bacterium isolated from lignocellulosic biomass for enhancing biogas production. Clean Technol. Envir. 18, 1565–1583 (2016).
CAS  Google Scholar 

68.
Vokou, D., Giannakou, U., Kontaxi, C. & Vareltzidou, S. Axios. Aliakmon and gallikos delta complex, Νorthern Greece. In Encyclopedia of Wetlands, Vol. 4 World Wetlands (eds Finlayson, M. et al.) (Springer, Berlin, 2016).
Google Scholar 

69.
Mellidou, I., Keulemans, J., Kanellis, A. & Davey, M. W. Regulation of fruit ascorbic acid concentrations in high and low vitamin C tomato cultivars during ripening. BMC Plant Biol. 12, 239–258 (2012).
CAS  PubMed  PubMed Central  Google Scholar 

70.
Bouyoucos, G. J. Hydrometer method improved for making particle size analysis of soils. Agron. J. 54, 464–465 (1962).
Google Scholar 

71.
Sparks, D. L. Methods of Soil Analysis – Part 3 – Chemical Methods, SSSA Book Series 5 (ASA. Madison, WI, SSSA, 1996).
Google Scholar 

72.
Karamanoli, K., Bouligaraki, P., Constantinidou, H. I. & Lindow, S. E. Polyphenolic compounds on leaves limit iron availability and affect growth of epiphytic bacteria. Ann. Appl. Biol. 159, 99–108 (2011).
CAS  Google Scholar 

73.
Eevers, N. et al. Optimization of isolation and cultivation of bacterial endophytes through addition of plant extract to nutrient media. Microb. Biotechnol. 8, 707–715 (2015).
CAS  PubMed  PubMed Central  Google Scholar 

74.
Louden, B. C., Haarmann, D. & Lynne, A. M. Use of blue agar CAS assay for siderophore detection. J. Microbiol. Biol. Educ. 12, 51. https://doi.org/10.1128/jmbe.v12i1.249 (2011).
Article  PubMed  PubMed Central  Google Scholar 

75.
García, C. A., De Rossi, B. P., Alcaraz, E., Vay, C. & Franco, M. Siderophores of Stenotrophomonas maltophilia: detection and determination of their chemical nature. Rev. Argent. Microbiol. 44, 150–154 (2012).
PubMed  Google Scholar 

76.
Dworkin, M. & Foster, J. Experiments with some microorganisms which utilize ethane and hydrogen. J. Bacteriol. 75, 592–601 (1958).
CAS  PubMed  PubMed Central  Google Scholar 

77.
Hontzeas, N., Zoidakis, J., Glick, B. R. & Abu-Omar, M. M. Expression and characterization of 1-aminocyclopropane-1-carboxylate deaminase from the rhizobacterium Pseudomonas putida UW4: a key enzyme in bacterial plant growth promotion. Biochim. Biophys. Acta 1703, 11–19 (2004).
CAS  PubMed  Google Scholar 

78.
Bradford, M. A rapid and sensitive method for the quantitation of micro gram quantities of protein utilizing the principle of protein—dye binding. Anal. Biochem. 72, 248–258 (1976).
CAS  Google Scholar 

79.
Nautiyal, C. S. An efficient microbiological growth medium for screening phosphate solubilizing microorganisms. FEMS Microbiol. Lett. 170, 265–270 (1999).
CAS  PubMed  Google Scholar 

80.
Luziatelli, F., Ficca, A. G., Colla, G., Švecová, E. B. & Ruzzi, M. Foliar application of vegetal-derived bioactive compounds stimulates the growth of beneficial bacteria and enhances microbiome biodiversity in lettuce. Front. Plant Sci. 10, 60. https://doi.org/10.3389/fpls.2019.00060 (2019).
Article  PubMed  PubMed Central  Google Scholar 

81.
Herlemann, D. P. R. et al. Transitions in bacterial communities along the 2000 km salinity gradient of the Baltic Sea. ISME J. 5, 1571–1579 (2011).
CAS  PubMed  PubMed Central  Google Scholar 

82.
Quast, C. et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 41, D590–D596 (2013).
CAS  PubMed  Google Scholar 

83.
Afgan, E. et al. The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2018 update. Nucleic Acids Res. 2, W537–W544 (2018).
Google Scholar 

84.
Li, D. et al. MEGAHIT v.10: a fast and scalable metagenome assembler driven by advanced methodologies and community practices. Methods 1, 3–11 (2016).
Google Scholar 

85.
Seeman, T. Prokka: rapid prokaryotic genome annotation. Bioinformatics 15, 2068–2069 (2014).
Google Scholar 

86.
Thompson, J. D., Gibson, T. J. & Higgins, D. G. Multiple sequence alignment using ClustalW and ClustalX. Curr Protoc Bioinform. https://doi.org/10.1002/0471250953.bi0203s00 (2002).
Article  Google Scholar 

87.
Hall, T. A. BioEdit: A user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl. Acid S. 41, 95–98 (1999).
CAS  Google Scholar 

88.
Kumar, S., Stecher, G., Li, M., Knyaz, C. & Tamura, K. MEGA X: molecular evolutionary genetics analysis across computing platforms. Mol. Biol. Evol. 35, 1547–1549 (2018).
CAS  PubMed  PubMed Central  Google Scholar 

89.
Felsenstein, J. Confidence limits on phylogenies: an approach using the Bootstrap. Evolution 39, 783–791 (1985).
PubMed  PubMed Central  Google Scholar 

90.
Wu, S., Zhu, Z., Fu, L., Niu, B. & Li, W. WebMGA: a customizable web server for fast metagenomic sequence analysis. BMC Genomics 12, 444. https://doi.org/10.1186/1471-2164-12-444 (2011).
Article  PubMed  PubMed Central  Google Scholar 

91.
Kamou, N. N. et al. Isolation screening and characterisation of local beneficial rhizobacteria based upon their ability to suppress the growth of Fusarium oxysporum f. sp. radicis-lycopersici and tomato foot and root rot. Biocontrol Sci. Technol. 25(8), 928–949 (2015).
Google Scholar  More

Origin of hydrate-bound hydrocarbons
A relationship between C1/(C2 + C3) and C1 δ13C has been applied to identify the sources of hydrocarbons in submarine seeps24. Recently, this diagram was revised based on a large dataset25. As shown in Fig. 4a, hydrate-bound hydrocarbons at the Kedr MV have thermogenic and/or secondary microbial origins, whereas those of other gas hydrate sites (Malenky, Bolshoy, Malyutka, Peschanka P-2, Kukuy K-0, Kukuy K-2 and Goloustnoe; Fig. 1) in Lake Baikal demonstrate microbial or early mature thermogenic origins. The hydrate-bound C1 from all locations except those at the Kedr MV were interpreted to be of microbial origin via methyl-type fermentation23 according to Whiticar’s old diagram26; however, the revised diagram25 suggests early mature thermogenic gases (Fig. 4b). Those of the Kedr MV plot at the boundary of the thermogenic and secondary microbial origin zones. Low C1 and C2 δ13C at the Peschanka P-2 MV indicated that C1 and C2 are of microbial origin27,28, whereas Kedr MV shows high C1 and C2 δ13C indicating their thermogenic origin (Fig. 4c). At other sites, C1 and C2 δ13C suggested that gases are mainly of microbial origin (in terms of C1) with some thermogenic component (13C rich and higher concentration in C2).
Stable isotopes in hydrate-bound C1 at the Kedr-1 and Kedr-2 areas suggested its thermogenic origin. However, it is close to the field of secondary microbial C1 in Fig. 4b, and the data are plotted in the overlap between the fields of thermogenic and secondary microbial in Fig. 4a. Milkov29 mentioned that secondary microbial C1 is characterised by C1-rich dry gas, large C1 δ13C (between − 55‰ and − 35‰) and large CO2 δ13C (more than + 2 ‰). Although hydrate-bound and sediment gases in the Kedr MV were not C1 rich and contained 3%–15% of C2, C1 δ13C was around − 45‰, which agrees with the secondary microbial C1. Because some data of secondary microbial gas are plotted outside the field on the original graph25, we could include the gas data in the category of secondary microbial C1 in Fig. 4b.
Figure 6 shows the relationship between C1 δ13C and CO2 δ13C in the sediment gas obtained using headspace gas method. According to the genetic diagram25, gas hydrate cores are plotted at the zones of the thermogenic and secondary microbial origins, whereas the cores at the peripheral area are primary microbial. The headspace gas data of the hydrate-bearing cores in Fig. 6 seem to be plotted in the field of thermogenic gas (low CO2 δ13C), but the effect of light CO2 produced by methane oxidation in the subsurface layer also decreased CO2 δ13C as shown in Fig. 5. These results suggested that secondary microbial C1 mixes into thermogenic gas. Coal-bearing sediments exist around the Kedr area21,22, and secondary microbial C1 can also form from coal beds30. Hydrate-bound C1 of secondary microbial origin has been only reported at the Alaska North Slope31. This study is another case for it.
Figure 6

A diagram of headspace gases. CO2 δ13C plotted against C1 δ13C, based on the classification of Milkov and Etiope25.

Full size image

Formation process of the sII gas hydrates
As stated before, the crystallographic structure of gas hydrates at the Kedr MV is mainly due to the composition of thermogenic C2 in the volatile hydrocarbons. The concentration of C3, which is one of the sII-forming components, was two to three orders of magnitude smaller than that of C2, because biodegradation occurs and this preferentially reduces C3−5 of n-alkanes19,32, 33. The concentration of n-C4 was smaller than that of i-C4, whereas that of n-C5 was not detected (Table 1). C3 δ13C was around − 10‰, suggesting that light C3 is consumed by microbial activity. Assuming that sediment gas C3+ can be ignored, sediment gas ratio C1/C2 at the study area was 30 ± 17 (mean and standard deviation), and the concentration of C2 was ~ 3%. Such a composition of thermogenic gas is, therefore, considered to be supplied from a deep sediment layer, forming sI gas hydrates composed of mainly C1 and C211,12 in the lake floor sediment.
In the cases where sI gas hydrates plug and block migration pathways, upward fluid flow becomes more focused in other areas16. Once gas supply stops locally, gas hydrates begin to decompose, with the gas dissolving into gas–poor sediment pore water. In the system of C1 and C2, C2 is prone to be encaged in gas hydrate and decreases the equilibrium pressure of mixed-gas hydrate. Therefore, C2-rich gas hydrate forms in parallel with the decomposition of sI gas hydrate. The Colorado School of Mines Hydrate (CSMHYD) program34 showed that C2-rich sII gas hydrate (C2 concentration 17%) forms from mixed gas composed of C1 and C2 (C2 concentration 3%). The C2 concentration of hydrate-bound gas at the Kedr MV was ~ 14%, agreeing fairly well with the results of the CSMHYD program. Such secondary generation of gas hydrates can produce compositions and crystallographic structures that are different from the original crystals. A calorimetric study of synthetic C1 and C2 mixed-gas hydrate revealed that double peaks of heat flow correspond to the dissociation process of C1 and C2 mixed-gas hydrate, suggesting that C2-rich gas hydrate forms simultaneously from dissociated gas and showed that the second heat flow peak correspond to the dissociation of C2-rich gas hydrate18. The PXRD and solid-state 13C nuclear magnetic resonance techniques demonstrated that C2-rich sI gas hydrate forms in the dissociation process of C1 + C2 sII gas hydrate35.
Among twenty hydrate-bound cores in the Kedr area, four cores contained sI only, seven cores had sII only, and seven cores showed sII at the upper layer and sI at the lower layer, as observed at the Kukuy K-2 MV13,16,17. Furthermore, in the cores 2015St1GC15 and 2016St18GC2, gas hydrate structure had sI at the upper and lower layer, and sII at the middle layer. These results suggested that complex gas hydrate layers are composed of sI and sII in subsurface sediments as shown in the schematic illustration in Poort et al.16.
Depth profiles of C2 δ2H of gas hydrate cores from the Kedr MV are shown in Fig. 7. C2 δ2H of hydrate-bound gases varied between − 227‰ and − 206‰, with a grouping around − 210‰. C2 δ2H of sediment gases was also around − 210‰, indicating that C2 δ2H of the original thermogenic gas is − 210‰. As stated above, C2 δ2H of some cores showed low values at their base. Based on the isotopic fractionation of hydrogen in C2 during the formation of sI C2 hydrate36, δ2H of hydrate-bound C2 was 1.1‰ lower than that of residual C2. However, this is too small to explain the wide distribution in C2 δ2H shown in Fig. 7. On the other hand, Matsuda et al.37 reported that isotopic fractionation of hydrogen in C2 is dependent on the crystallographic structure: 1‰–2‰ for sI and ~ 10‰ for sII. Gas hydrates plotting around − 220‰ in C2 δ2H can be explained as a secondary generation of sII from dissociated gas hydrates, of which C2 δ2H was around − 210‰. However, some sII samples showed high C2 δ2H (around − 210‰), whereas some sI samples showed low C2 δ2H (around − 220‰). These results indicated that formation and dissociation processes of gas hydrates produce complicated isotopic profiles in C2 δ2H under non-equilibrium conditions.
Figure 7

Depth profiles of C2 δ2H of hydrate-bound and sediment gases. cmblf, centimetres below lake floor.

Full size image

Characteristics of hydrate-bound gases in sII
C3, i-C4, n-C4 and neo-C5 can be encaged in the larger hexadecahedral cages of sII1. n-C4 and neo-C5 can be encaged using a help gas (e.g. C1) to fill in the smaller dodecahedral cages of sII, because they cannot form pure n-C4 and neo-C5 hydrates, respectively. Figure 8 shows the concentration of C3, i-C4, n-C4, neo-C5 and i-C5 plotted against C2 concentration. The figure illustrates a clear division between sI (3–4%) and sII (14%) C2 concentrations. Data points between C2 concentrations of 5% and 13% were considered to have a mixture of sI and sII. Concentrations of C3, i-C4, n-C4 and neo-C5 had a positive correlation with the concentration of C2, and these concentrations in sII were 1 or 2 orders of magnitude larger than those in sI, suggesting that C3, i-C4, n-C4 and neo-C5 are encaged with C2 in the sII formation process.
Figure 8

Concentration of C3–5 against C2 concentration in the hydrate-bound gases.

Full size image

C3 values of 0.001%–0.01%, ~ 0.0001% of n-C4, and 0.0001%–0.01% of neo-C5 were also detected in sI hydrate-bound gas (Fig. 8), despite these hydrocarbons being unable to be encaged in sI. This can be explained by gases being adsorbed with sediments and gas hydrate crystals, which are then trapped in the grain boundary of polycrystalline gas hydrate crystals, and the gases are encaged if a small amount of sII crystals are present. For example, Uchida et al.38 examined natural gas hydrate retrieved at the Mackenzie Delta (onshore Canada) and detected C3 encaged in sII using Raman spectroscopy, although PXRD results suggested that the sample was sI and the major component of hydrate-bound gas was C1 (more than 99%).
neo-C5 is considered to form from the decomposition of gem-dimethylcycloalkanes derived from the terpenes of terrestrial organic matter39. It is easily enriched by preferential diffusion due to the nearly spherical molecules and its diffusion coefficient, which is higher than that of less branched isomers40. The sII hydrates retrieved at the Kukuy K-2 MV (central Baikal basin) contained 0.026–0.064% of neo-C5 in the volatile hydrocarbons13,14, and those at the Kedr MV had a maximum value of 0.054% of neo-C5 (Supplementary Information Table S1). On the contrary, in the case of natural gas hydrates retrieved at the Joetsu Basin (Japan Sea), neo-C5 was excluded and remained in sediment during the formation of sI gas hydrates from C1-rich gas41. The molecular size of i-C5 is considerably large to be encaged in the large cages of sII. Maximum concentration of i-C5 in the hydrate-bound gases was in several parts per million in both the fields of sI and sII (Fig. 8), indicating that i-C5 is not a hydrate-bound hydrocarbon and adsorbed with gas hydrate crystals and/or trapped in their grain boundary. More

General scheme
We reviewed the literature published until April 2018 on the economic impacts of invasive species. For reasons of feasibility (linguistic skills of the review team, restriction to a reasonable scale of the review), we conducted all searches in the English language assuming that a large body of knowledge (mostly from international peer-reviewed papers and reports) is written in English. The dates of each search process were systematically recorded. We used the following strategy for all repositories (Fig. 1), while also taking into consideration the specificity of their algorithms.
First, a literature search was performed using three online bibliographic sources successively to minimize the risk of omitting relevant materials (Fig. 1, step 1a): ISI Web of Science platform (https://webofknowledge.com/), Google Scholar database (https://scholar.google.com/) and the Google search engine (https://www.google.com/). We carefully composed appropriate search strings that were consensually retained as the most efficient among a set of potential candidates. A decision was taken following preliminary tests based on a handful of relevant articles provided by consulted subject experts on some taxonomic groups (amphibians, reptiles, fishes and ants). Final selection of search strings comprised those considered to have the largest potential to identify key references. Each search string was set to include a combination of two search terms, related to ‘invasive’ and ‘economics’. For both terms, we used a range of synonyms or related words. For example, for ‘invasive’ we used invasi*, invader or exotic; for ‘economics’, we used econom*, cost or monetary. In addition, the search string included exclusion terms to omit mismatches, for example, with studies from the field of medicine that are focused on pathologies or procedures that can be ‘invasive’ for patients. We complemented this search with documents gathered opportunistically (Fig. 1, step 1b). The potentially relevant materials derived from all these sources were combined in a single file and screened for duplicates. Second, retrieved documents were individually assessed at progressive levels (titles, then abstracts, keywords, and finally full text when abstracts were missing; Fig. 1, step 2) based on three criteria. Hence, materials were deemed relevant if (i) they matched with the linguistic competencies of the review team (i.e. written in English, or French where English language was restricted only to the title and/or abstract) for allowing reliable assessment, (ii) they contained at least one cost estimate (studies exclusively providing benefit estimates from direct use or exploitation of invasive species were excluded), and (iii) that this cost estimate is exclusively associated with invasive species (estimates merging non-invasive and invasive species, without the possibility of distinguishing the respective contribution of each group to the overall cost, were excluded). To ensure transparency and validity, each document was checked by two reviewers and in case of a disagreement between assessors, a third reviewer was involved. However, it was often difficult to judge from the topic whether the content of an article was relevant and so consequently many more articles were conservatively kept when final agreement was lacking among assessors.
Finally, relevant materials were scrutinized for data on economic costs (Table 1; Fig. 1, step 3). During this step, additional relevant materials were found as cited by the analysed materials. Obtained cost data were collated in a database and the costs were converted to a common and up to date currency (2017 US$), and then depicted by different descriptors. Categories extracted from relevant materials allow search of the database and data pre-selection to facilitating analysis of costs based on taxonomic groups, geographical areas, impacted sectors, types of costs, or other categories. The reliability of cost estimates and all associated information recorded in the final InvaCost database was systematically checked at least twice, and every ambiguous element was discussed to reach a consensus. We also checked all entries in the database to ensure that there were no obvious duplicate reports (i.e. multiple documents reporting the same cost estimate) or mistakes.
Hereafter, we specifically describe each of the steps made to generate InvaCost.
Literature search
Web of Science
We used the Web of Science (hereafter called WoS) to conduct a search for potentially relevant materials on 7 December 2017 (Fig. 1, step 1a). We applied the following search string: (econom* OR cost OR monetary OR dollar OR euro OR “sterling pound”) AND (invasi* OR alien OR non-indigenous OR nonindigenous OR nonnative OR non-native OR exotic OR introduced OR naturali* OR invader) NOT (cancer* OR cardio* OR surg* OR carcin* OR engineer* OR rotation OR ovar* OR polynom* OR purif* OR respirat* OR “invasive technique” OR carbon OR fuel OR therap* OR vehicle OR cell* OR drug OR fitness OR “operational research” OR banking OR liberalization). The terms were searched in the field code “Topic” which includes title, abstract and keywords, and which also comprises ‘Keywords Plus’ that are generated by WoS through an automatic computer algorithm, based on words and phrases that appear frequently in the titles of article’s bibliographic references and not necessarily in the main text of the article itself. To limit the search to relevant fields of research, we used the function ‘refine’ to exclude subject areas not related to economics and/or invasion biology.
We exported all records (n = 16,875) into an Excel worksheet30 (Table 1) to identify the relevant materials by a two-step procedure. First, we excluded the references identified only based on ‘Keywords Plus’, which were shown to be poor specific descriptors of the content of articles31. We also excluded references identified based on the presence of only a single search term in the topic, as we assumed that words related to both search terms (‘invasive’ and ‘economics’) should be mentioned at least once in the title, abstract and/or keywords of a relevant material. To identify these irrelevant materials within the references collected, we developed a script (see Code Availability) in the R programming language (R v.3.4.3)32. Subsequently, 10,592 references were kept for the next screening step based on the described criteria.
In the second step, the topic of every reference selected was checked manually to ensure potential relevance of its contents. This allowed the elimination of documents incorrectly identified as relevant, such as studies without a true monetary assessment, or those focusing on economic estimates not directly attributable to invasive species only. Finally, 1,333 documents were judged as relevant materials (Table 1) and moved to the final data collation step.
Google scholar
The Google Scholar database is a large source of grey as well as peer-reviewed literature. Nevertheless, we had to modify our approach in order to address inherent limitations of this database as a search tool (see Haddaway et al.33 for a comprehensive analysis). Typically, Google Scholar allows limited Boolean operators (no nesting using parentheses permitted) and search strings are limited to 256 characters. Additionally, only the first 1,000 search results can be viewed and the order in which results are returned is not disclosed. We also wanted to maximize novel information by avoiding too much overlap between the references collected with WoS and those gathered here.
In light of the above, we adapted our search string to generate the most efficient outcome, i.e. sufficiently pertinent to bring the most relevant items to the top of the result list while not unnecessarily large so as to limit the host of non-viewable results. Thus, the following search string was applied on 26 April 2018, using the advanced search facility to search for selected words anywhere in the article (see https://scholar.google.se/intl/en/scholar/help.html#searching for further details): dollars OR euros OR “USD” OR “EUR” OR “NZD” OR “AUD” OR “CAD” OR “GBP” OR “economic cost” OR “economic impact” OR “estimated cost” invasive species. We specified currencies for prioritising materials with monetary data in the top of the resulting list. These currencies were chosen as they were the most often used to express economic costs in the literature collected from the WoS. Nevertheless, any reference evoking economic costs in other currencies was expected to be also captured by some specific combinations of ‘economic’ terms in our search string that we would expect to be mentioned at least once in the full-text of relevant papers. In addition, we included the concomitant presence of ‘invasive’ and ‘species’ terms to restrict the outcomes to papers within the scope of our synthesis. Subsequently, we collected all viewable results (100 pages, n = 992 references of the 668,000 generated), thus going beyond the traditional and arbitrary sample size of first 50–100 results, which is frequently selected in many systematic reviews. We used a web-scraping programme (https://www.webscraper.io/) to extract all the titles’ references returned by the search in an Excel spreadsheet. Because we could not efficiently export the abstract for every reference, we screened them online to assess their potential relevance.
As a result of a search and relevance assessment within Google Scholar, the references, abstracts and specific bibliographic details of 432 documents were added to the sample for further analysis. After excluding duplicates with WoS retrieved references, 310 additional documents were included in the sample as potentially relevant materials (Table 1).
Google
We used the Google search engine to complete the standardised literature search. As when searching with Google Scholar, we took into account specific constraints related to the use of this search engine. Moreover, browsing through Google search results can be overwhelming due to the vast amount of information of highly variable quality. We attempted to implement a search strategy that could allow overcoming these limitations as much as possible. We used the following search string: economic species invasive OR nonnative OR alien OR exotic OR nonindigenous -disease -surgery -fungus -respiratory. We added four exclusion terms (disease, surgery, fungus, respiratory) identified during preliminary tests to restrict the number of irrelevant studies, associated with medical research. We did not use a range of economics-related terms, such as impact or cost, as they returned overly large numbers of mismatches.
The web search was conducted on 8 May 2018 by searching for specified terms within page titles of each document, in order to maximize the likelihood of identifying grey literature. We especially targeted grey literature because searches by the other two platforms mainly led to peer-reviewed publications. We assumed that documents published online by various governmental and non-governmental organisations (NGO), research centres and academic institutes are more likely to contain relevant data than other types of documents such as blogs and catalogues29. Therefore, we restricted our search to the documents located on governmental, academic and NGO webpages to ensure that explicit, traceable and expertise-based information was retrieved. We conducted independent searches for each type of webpage by specifying the type of web extension in the advanced search facility (.gov for governmental,.edu for academic, and.org for organisational webpages).
361 search hits were collected (document name, publishing year and URL of the main website homepage, if available) and stored in the database with the same host of dedicated information (Table 1). If the item analysed was a website homepage, we conducted on-line searches of potentially relevant materials within the website database(s), by filters if available, or by using the search bar with combinations of keywords. Websites that did not contain a database or search bar were searched manually. We then eliminated all duplicates resulting from references being listed on multiple websites, or due to typographical mistakes and/or incomplete records when reporting a reference within different repositories. A total of 119 potentially relevant materials was finally obtained (Table 1).
Targeted collection
Finally, we sourced other potentially relevant materials that did not originate from the above-described processes (Fig. 1, step 1b). On one side, we dedicated specific efforts on gathering cost estimates for particular taxa or areas for which data previously obtained seemed scarce. First, we made sure that some key species were adequately covered; for example, costs associated with invasive mosquito species responsible for much of the burden of mosquito-borne viral diseases worldwide (Aedes aegypti that mainly invaded the intertropical zone from the 15th-17th centuries, and Aedes albopictus for which the global dissemination was more recent34) were searched in a specific way using WoS and PubMed (https://www.ncbi.nlm.nih.gov/pubmed/) repositories (see supplementary file 1 for details on search strings and matching with PRISMA statements). Second, materials were also retrieved following requests to specialists (e.g. Aliens mailing list, https://list.auckland.ac.nz/sympa/info/aliens-l) to bridge gaps identified for Russia and China, two of the five largest countries for which available on-line data were particularly scarce. A typical message first summarized the objectives of our research project and second, requested recipients to provide relevant material and/or suggest further contacts in this regard. On the other side, we also compiled additional materials when establishing the methodology for the project (e.g. when testing different search string combinations at initial stages of the work), from the bibliographic alerts set up by the review team. All 1417 documents obtained from this process were entered in the database, with information on the person providing the document (Table 1;30). Subsequently, 150 documents identified as not previously retrieved were considered relevant for further, full-text screening (Table 1).
Extraction of cost estimates
The Online-only Table 1 comprises all the information of InvaCost that we mention further in this article, using simple quotation marks for ‘Columns’ of the database and italic letters for the different categories within each column. The full-text of each relevant material was scrutinized for any cost estimate that could be incorporated into InvaCost30. The final stage of inclusion/exclusion took place during this data extraction. When the screened documents reported cost estimates by citing sources that were not retrieved by our literature search, whenever possible we assessed the original sources of data in order to better characterize the reported cost. These novel information sources not initially captured by our literature search were then added to the collection list (Table 1). In such cases, we provided information on all documents that were consulted to trace back the original source (‘Previous materials’). In contrast, if no original cost data were found in the cited source, the document was discarded. For all reported costs where the original source was not available or accessible, we emphasized this in a dedicated column (‘Availability’).
Then, we first extracted raw cost data, i.e. how they appear in the material in local currency (‘Raw cost estimate local currency’). When multiple cost estimates were provided for a single instance, we calculated median values (e.g. different cost estimates according to several management scenarios dedicated to the same invasive population) and collated the minimum and maximum estimates provided (columns ‘Min/Max raw cost estimate local currency’). When costs were estimated at different time and/or spatial scales in the same material, we opted to choose – when possible – those estimate(s) that summarise(s) as effectively as possible the figure(s) shown in the study. If such an estimate was not obvious to identify throughout the full-text, we extracted every relevant cost estimate. In these latter cases where several cost estimates were provided in a single study, we also collated the minimum and maximum estimates provided.
Temporal information on the costs were also retrieved: the ‘Period of estimation’ as stated in the material and hence, when possible, the ‘Probable starting/ending year’ of the period of estimation and the ‘Time range’ (year if the estimate is given yearly or for a period up to one year, period if the estimate is given for a period exceeding a year). The ‘Occurrence’ column gives the status of the cost estimate as potentially ongoing (if the cost can be expected to continue beyond the period of estimation) or one-time (if the cost was deemed as unlikely to continue). For cost estimates provided without a clear indication on the timeframe considered, or covering periods shorter than a year, we considered them with a year ‘Time Range’ and a one-time ‘Occurrence’ to avoid the risk of overestimating the duration of collated costs. The ‘Raw cost estimate’– with complementary information on the ‘Time range’, ‘Period of estimation’ and ‘Occurrence’ – can be used to estimate total costs over a given period of time. We then transformed the raw cost estimates to cost estimates per year (‘Cost estimate per year’) by dividing the raw costs with a period ‘Time Range’ by the duration of the ‘Period of estimation’ (obtained from the difference between the ‘Probable ending year’ and ‘Probable starting year’). The raw costs with a year ‘Time Range’ were reported as they are, because they are already considered at the scale of a year.
Description of cost estimates in InvaCost
Each of the cost estimates recorded was characterized by a number of information, including (a) the reference from which the cost was extracted, (b) the taxonomy of the associated species, (c) the spatial and temporal coverage of the study, (d) the typology of each cost estimate and (e) the evaluation of the reliability of the estimation method(s). For most of the variables considered in InvaCost, a non-negligible part of the cost estimates was not attributable to a single existing category due to the lack of precise information provided by the authors or because they simultaneously belong to multiple categories. In such cases, we respectively reported them as either Diverse/Unspecified or as slash-separated lists of categories (e.g. Artiodactyla/Carnivora for the ‘Order’).
Details about the nature of the information retrieved as well as the choices made to characterize each cost are synthesized in Online-only Table 1:
(a) We provided bibliographic information on each reference (e.g. ‘Reference title’, ‘Authors’, ‘Publication year’). Others specific details (e.g. abstract, journal, download link) are given in a dedicated file30 with which the columns ‘Repository’ and ‘Reference ID’ of InvaCost allow correspondence of information.
(b) We normalised and harmonised all taxonomic information on the invasive species (‘Kingdom’ to ‘Species’ level) using the GBIF.org Backbone Taxonomy35. At this stage, spelling and other taxonomic errors were corrected. While each cost extracted was generally associated with a single invasive alien species, in some cases the data was related to multiple species without the possibility of disentangling species-specific costs. In this case, we mentioned either all species concerned if explicitly indicated by the author(s), or Diverse/Unspecified if not.
(c) We dedicated seven columns to describing the impacted area according to its environment (terrestrial and/or aquatic habitats), the temporal extent as mentioned earlier (e.g. ‘Period of estimation’, ‘Time range’) and the spatial coverage from the ‘Geographic region’ (e.g., Central America, South America, Oceania-Pacific Islands) – rather than the official continent for better accuracy – down to the exact site (‘Location’) when possible. Each area was related to its country of attachment, leading to some mismatches between the ‘Geographic region’ and ‘Official country’ columns due to the existence of countries with non-contiguous overseas territories. For instance, costs found from invaders in La Réunion (a French oversea department) were attributed to Africa as ‘Geographic region’ and France as ‘Country’, while France obviously belongs to European continent.
(d) We characterised the typology of each cost mainly based on the following descriptors. The ‘Implementation’ at the moment of the cost evaluation states whether the reported cost was observed (i.e. cost actually incurred by an invasive species within its invasive distribution area) or potential (i.e. not incurred but expected cost for an invasive species beyond its actual distribution area and/or predicted over time within or beyond its actual distribution area). The ‘Acquisition method’ provides information on how the cost data was obtained, i.e. report/estimation directly obtained or derived (using inference methods) from field-based information, or extrapolation relying on computational modelling. The ‘Impacted sectors’ indicates which activity, societal or market sectors were related to the cost estimate (see Table 2 for details). The ‘Type of cost’ ranges from the economic damages and losses incurred by an invasion (e.g. value of crop losses, damage repair) to different levels of means dedicated to the management of biological invaders (e.g. control, eradication, prevention).
Table 2 The different market and/or activity sectors mentioned in InvaCost.
Full size table

(e) Lastly, we evaluated the level of ‘Reliability’ of the methodology reported by the authors to provide cost estimates (Fig. 2). Prejudging the relevance of each cost estimate is not straightforward and could suffer from a high level of subjectivity. Here, we rather aimed to evaluate in the most objective manner whether the approach used for cost estimation was documented and traceable. Hence, materials that could not be accessed for full-text investigation were conservatively considered as of low reliability. Alternatively, each cost estimate recorded from any accessible material was qualitatively assessed as of high or low reliability following a procedure depending on the ‘Type of material’ analysed (peer-reviewed article or grey material; Fig. 2). Peer-reviewed articles and official documents (e.g. institutional or governmental reports) are likely validated by experts before publication. We assumed therefore that all cost estimates collected from these materials may likely be of high reliability. Conversely, for grey materials other than official reports, the attribution to one or other of these categories (high vs low reliability) was based on specific analysis of each cost estimate. We checked whether the method estimation was fully described, independently of its comprehensiveness, i.e. if the original sources or potential assumptions were properly documented or justified, and/or the calculation methodology was explicitly demonstrated. Here, we opted for a conservative strategy that might be not optimal, as depending mostly on the nature of the publication.
Fig. 2

Decision tree approach for assessing the reliability of the method used for estimating each cost. The colour of the boxes indicates which decision was taken: green when material was deemed as of high reliability, red when material was deemed as of low reliability, blue when taking any decision needs further investigation. The intended purpose of this process was not to evaluate the quality, relevance or realism of the studies performed for providing cost estimates, but rather to assess if the methodology (i) has been reviewed and validated by peers or experts prior any publication, or (ii) if not, whether this methodology was clearly stated and demonstrated.

Full size image

Beyond the factual elements included in the descriptors from (a) to (c), those presented in (d) and (e) (to which we can add the descriptor ‘Spatial scale’) are the result of a conceptual and analytical framework created based on our own experience. This experience was gained when collecting and getting acquainted with the diversity and complexity of situations one can find behind the “economic costs” linked to biological invasions, as well as the strategies used for estimating them. We think that the different subcategories identified therein (e.g. observed vs potential costs within the descriptor ‘Implementation’) should not be aggregated to limit potential confusions in future analysis. Also, we acknowledge that the possible sub-categories of these descriptors might be improved and adapted according to the scope of future analyses made using InvaCost. We are convinced that the descriptors thus defined and categorised may strongly help in this perspective.
Standardisation of cost data
Using definitions, data and indicators provided by the World Bank Open Data and the Organisation for Economic Cooperation and Development (OECD), we expressed all retrieved costs (raw costs and costs per year) in US dollars (US$) for the year 201730 using a multi-step procedure. We provided here two ways for standardising cost estimates according to the conversion factor: one based on the market exchange rate (local currency unit per US$, calculated as an annual average), and another based on the Purchasing Power Parity (PPP, local currency unit per US$, calculated as an annual average) that is the rate of currency conversion that standardises the purchasing power of different currencies by eliminating the differences in price levels between countries. Opting for one strategy or the other for further investigation or discussion is beyond the scope of this paper and will befall on the author(s) of future analyses made using InvaCost.
We first converted the cost estimates from local currencies to US$, by dividing the cost estimate with the official market exchange rate (https://data.worldbank.org/indicator/PA.NUS.FCRF?end=2017&start=1960) corresponding to the year of the cost estimation (‘Applicable year’, that is the year of the ‘Currency’ value, but not necessarily the year of the cost occurrence). The cost obtained in US$ of that year was then converted in 2017 US$ using an inflation factor that takes into account the evolution of the value of the US$ since the year of cost estimation. The inflation factor was computed by dividing the Consumer Price Index (CPI, which is a measure of the average change over time in the prices paid by consumers for a market basket of consumer goods and services; https://data.worldbank.org/indicator/FP.CPI.TOTL?end=2017&start=1960) of 2017 by the CPI of the year of the cost estimation.
As an alternative, we also converted costs to 2017 US$ value based on PPP instead of the classical market exchange rates in the initial conversion step. PPP values were primarily collected from data provided by the World Bank (https://data.worldbank.org/indicator/PA.NUS.PPP?end=2017&start=1990), or by the OECD (https://data.oecd.org/conversion/purchasing-power-parities-ppp.htm) when information was not retrievable through the World Bank database. For this purpose, we had to deal with published costs that were expressed in currency that was different from the country where the costs were estimated (e.g. published cost in African countries expressed in US or Canadian $). Thus, prior to using PPP as a conversion index, we had to perform a retro-conversion by multiplying the original cost estimate by the official market exchange rate (local currency unit per currency unit used). For PPP-based standardisation, it was not possible to perform the process for all cost estimates as PPP data do not exist for all countries and/or specific periods (we mentioned NA in the database when such information was missing).
In summary, we used the following formula to convert and standardise each cost estimate:

$${C}_{e}=left({{boldsymbol{M}}}_{{boldsymbol{V}}}/{{boldsymbol{C}}}_{{boldsymbol{F}}}right),times ,{{boldsymbol{I}}}_{{boldsymbol{F}}}$$

with Ce = Converted cost estimate (to 2017 US dollars based on exchange rate or Purchase Power Parity), MV = Cost estimate (either the ‘Raw cost estimate local currency’ extracted from analysed paper or the ‘Cost per year local currency’ transformed by us), CF = Conversion factor (either the official market exchange rate or the purchasing power parity, in US dollars), IF = Inflation factor since the year of cost estimation, calculated as CPI2017/CPIy with CPI corresponding to the Consumer Price Index and y corresponding to the year of the cost estimation (‘Applicable year’).
We thus provided four columns with the raw cost estimates or the cost estimates per year, expressed in 2017 USD based on the exchange rate or PPP.
Data summary
InvaCost currently contains 2419 cost estimates (1215 from peer-reviewed articles, 1204 from grey materials), collected from 849 references, of which 1769 estimates were deemed as of high reliability. In total, twenty currencies are reported in our database, the majority being US dollars, n = 1348 cost estimates. Not all cost estimates were successfully converted to 2017 US$ as (i) conversion data from official sources are available only since 1960 (cost estimates range from 1945 to 2017 in InvaCost) or simply not found for some years and countries, and/or (ii) cost data are sometimes simultaneously associated with several countries, constraining the PPP-based standardisations. Hence, respectively 2416 and 2126 estimates were successfully converted using market exchange rates and PPPs. Cost estimates are either direct reports/estimations (n = 2127) or values gathered from extrapolative computations (n = 292). At a taxonomic level, these estimates are associated with 343 species belonging to six kingdoms (Animalia, Bacteria, Chromista, Fungi, Plantae, Riboviria). InvaCost has global coverage (90 countries) and includes continental, insular and overseas territories. Data are associated with terrestrial as well as aquatic (freshwater, brackish and marine) environments. Costs were estimated at different spatial scales (continental (n = 35), country (n = 1111), global (n = 17), intercontinental (n = 9), regional (n = 67), site (n = 836), unit (n = 329)). The Table 3 summarises quantitative data and information reported in InvaCost for each geographic region considered (see also Supplementary file 2).
Table 3 Quantitative summary of information recorded in InvaCost according to the ‘Geographic region’ of the cost estimates.
Full size table

Possible applications
InvaCost is expected to help bridge the gap between a growing scientific understanding of invasion impacts and still inadequate management actions. This work is thus in line with the aims of a panel of decisions recently adopted by the Convention on Biological Diversity (Decision XIII/13, https://www.cbd.int/doc/decisions/cop-13/cop-13-dec-13-en.pdf) advocating the incorporation of invasion science knowledge into management planning. In addition to offer unique opportunities for future research, InvaCost will provide a strong quantitative and evidence-based support for impacts of invasive species reported in other databases such as the Global Register of Introduced and Invasive Species (GRIIS)20, helping refine information in this database. Also, invasive populations recorded in InvaCost but data deficient in the GRIIS should be ultimately classified in that database.
Additionally, InvaCost could be considered as another data-based component, adding novel and significant information on invader impacts categorised by the Socio-Economic Impact Classification of Alien Taxa (SEICAT)36. The latter is a classification system, applicable across a broad range of taxa and spatial scales, providing a consistent procedure for translating the broad range of measures and types of impacts into ranked levels of socio-economic impacts, assigning alien taxa on the basis of the best available evidence of their documented deleterious impacts. Quantitative support provided by InvaCost will strongly contribute to impact classification. Ultimately, integrating data from these diverse sources could allow a complete description of the overall impacts of biological invasions at regional and global scales.
Caveats and directions for further database improvement
Rather than claiming exhaustiveness of data collated, we highlight that InvaCost should be considered as the most current, standardised, accurate and globally representative repository of various economic losses and expenditures documented for the largest possible set of invaders. We are aware that our database can be improved in at least three ways.
First, InvaCost mostly does not include publications and reports not yet available in electronic format and/or using non-English language, leaving open the possibility of increasing data comprehensiveness and limiting potential biases. Indeed, local reports as well as research results from some countries (e.g., China, Russia) are likely to be published in non-English language37. Again, accessing grey literature is challenging as it is not systematically digitalised and/or included in well-curated bibliographic databases29. We strongly encourage future users of InvaCost to help gathering this currently unreachable information when possible. Furthermore, some mistakes might have occurred despite our best efforts when constructing InvaCost. In this regard, we advocate for regular public updates of InvaCost in order to improve it both quantitatively (by adding currently inaccessible or missed information) and qualitatively (if errors are identified).
Second, as the distribution and impacts of invaders are inherently dynamic for a number of reasons38, InvaCost should further consider the status of the species recorded for their economic impacts in order to improve both the relevance and the usefulness of the database. As an illustration, InvaCost likely includes invasive populations currently extirpated from particular areas after successful eradication campaign(s) as well as those still established but for which impacts are locally reduced as a result of management efforts. Attempting to obtain and integrate such information into InvaCost was beyond the scope of this work. Nonetheless, it should be reciprocally beneficial to establish connections between InvaCost and other databases such as the GRIIS that provides a harmonised, open source, multi-taxon database including verified information on the continued presence of introduced and invasive species for most countries20. In light of such additional information, the value of InvaCost will be its application for policy purposes, such as identification of exotic invaders that are currently associated with economic losses in particular areas. Also, crossing information between databases may allow the refinement of the descriptor ‘Spatial scale’ we propose here.
Third, we would recommend, for a future updated version of InvaCost that would require screening back all the materials, to improve the ‘Acquisition method’, ‘Implementation’ and ‘Reliability’ descriptors, to pay attention to the specificity of “avoided costs” and to create a new descriptor for ‘non-market values’. We detail these possibilities below.
Improving descriptors
An improved version of the ‘Acquisition method’ could lead to a subdivision of the extrapolation category into spatial, temporal and spatio-temporal extrapolation. This would allow simultaneous refinement from the currently binary ‘Implementation’ descriptor (observed vs potential) into several levels of certainty regarding the incurred cost (e.g. taking into consideration the temporality (past/current or predicted) of the onset of the cost and of the status of the invasive species in the study area). The next step for deeming the ‘Reliability’ of the cost estimates recorded in InvaCost would consist of assessing the repeatability of the methodology used, by adapting the approach previously developed by Bradshaw et al.14. The latter evidenced that assuming the reproducibility of published methods should not rely only on the nature of the materials and recognized the qualitative nature of the procedure, although applying this approach to InvaCost was constrained by the large sample size and high diversity in our database (Bradshaw et al.’s study focused on a single taxonomic class). Also, because InvaCost involves several collaborators and potential future contributors, consistent and objective criteria should be further defined to cope with the large array of materials, methods and situations encountered.
Avoided costs
Introducing certain actions against biological invasions leads to avoided costs. Such avoided costs are sometimes evaluated, for instance to examine the relevance of different potential actions or to assess the effectiveness of an action that was taken. However, avoided costs cover a great variety of situations and require a careful consideration for future analysis, even if they do not have to be analysed separately from the other economic costs gathered in InvaCost. For instance, in the case of hypothetical actions, avoided costs can be considered as minimum estimates of the “real” costs (if they are unknown). However, in the case of completed or planned actions, the reported data should be the original costs (if known) minus the avoided costs, because the latter do no longer exist. Some avoided costs are probably already included in InvaCost but they are likely underestimated because keywords such as “savings” or “benefits” were not included in the search strings. Also, even if they are sometimes mentioned as “benefits” in the literature, care should be taken not to confuse these avoided-costs with the benefits incurred by direct use or exploitation of invasive species. The latter have been ignored in InvaCost since they were relatively few (and beyond of the scope of this database), but might constitute a twin project.
A new ‘Non-market values’ descriptor
The means dedicated to preventing or managing an invasion (e.g. manual removal of invasive plants) and certain economic losses and damage due to an invasion (e.g. the value of crop losses or the repair costs of damaged infrastructures due to an invasive insect) are observable on markets. However, some costs are not observable on markets but can be translated in monetary terms using several valuation methods – for instance, the willingness to pay for the conservation of a native species that is impacted by an invasive species is considered as the value given by a group of people to preserving the native species (i.e. the value that would be lost if this native species was impacted). We recognize the importance of informing the public about “non-market values”, as giving an economic value to ecosystems or biodiversity can be a way of recognising and taking them into account in public decision-making processes39, but attention should be paid to the issues linked to their assessment40,41. Among others, the different methods for assessing non-market values do not necessarily capture the same aspects of the values, so the resulting estimates might be different. Moreover, the very principle of giving a value to “benefit from nature” through economic valuation is not necessarily acknowledged by the entirety of scientific and civil communities39,42. For future analysis, the ‘non-market values’ should not be systematically aggregated with the other economic costs gathered in InvaCost. It is to note that while some non-market values are probably already included in InvaCost within the losses and damage ‘Type of cost’, the loss of non-market values is probably largely underestimated in the database because they were not the primary focus of InvaCost and therefore the related keywords were not included in the search strings.
These possible ways of improvement call for completion and/or refinement of existing entries as well as integration of newly published or acquired data by future contributors in InvaCost, with the aim to consolidate its long-term relevance (cf. Usage Note paragraph). More