Ecology

Affiliations

State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China
Kris A. G. Wyckhuys & Yanhui Lu

Fujian Agriculture and Forestry University, Fuzhou, China
Kris A. G. Wyckhuys

University of Queensland, Brisbane, Queensland, Australia
Kris A. G. Wyckhuys & Michael J. Furlong

Chrysalis Consulting, Hanoi, Vietnam
Kris A. G. Wyckhuys

Zhejiang University, Hangzhou, China
Wenwu Zhou

CABI, Wallingford, UK
Matthew J. W. Cock & Frances E. Williams

USDA ARS, Maricopa, AZ, USA
Steven E. Naranjo

Secretariat of the Pacific Community SPC, Suva, Fiji
Atumurirava Fereti

Authors
Kris A. G. Wyckhuys

Yanhui Lu

Wenwu Zhou

Matthew J. W. Cock

Steven E. Naranjo

Atumurirava Fereti

Frances E. Williams

Michael J. Furlong

Corresponding authors
Correspondence to Kris A. G. Wyckhuys or Yanhui Lu or Wenwu Zhou. More

Multiple statistical methods have been developed to estimate the effect on birds and bats as a result of wind energy during the last 20 years26,27,28,29,30,31. Some of these studies are focused on the conservation status of the species32, the incidence factor of the wind turbines19,33, demographic parameters34,35, behavioural12 and also morphological parameters of the species36,37. In any case, it is essential to group all types of affections in order to be able to establish a global quantification that can be adapted to each species and to each specific wind farm. In other words, it can be obtained from a mathematical algorithm that allows quantifying the effect on each species, taking into account the characteristics of each wind farm3.
Furthermore, the formula that reflects the effect on the species must consider aspects related to the wind farm itself (type and distribution of turbines, occupation of the territory, etc.) and those related to the species, both in terms of its degree of threat and social importance, as well as its special sensitivity to the presence of the wind farm. According to this, the affection to each species must respond to the following formula:

$${text{AS}}_{{text{I}}} = {text{WF}}left( {{text{SS}}_{{text{I}}} + {text{VS}}_{{text{I}}} } right)$$

where ASI = Affection to species i, WF = Constant derived from the characteristics of the wind farm, SSI = Sensitivity of the species i to the presence of the wind farm, VSI = Social value of species i.
The index of affection, therefore, will be the product of multiplying the obtained values by the wind farm with those of each species that is present in the area.
Wind farm value constant (WF)
The impact value of the wind farm will be determined by the characteristics of the wind farm and also be influenced by both the characteristics of the wind farm (VWF) and its location (UF). At the same time, the VWF will be determined both by the affection of each wind turbine (WT) and by the distribution within the wind farm (extension and lines of turbines).

$${text{WF}} = {text{VWF}} + {text{UF}}$$

Value of the wind farm (VWF)
To calculate the overall effect of the wind farm not only is necessary to know the effect of each turbine but also its distribution in space. It is relevant to assess the distances between the wind turbine and if they are or not operating because when the turbines are very close together, the risk of moving between them is greater than in wind farms with more separate wind turbines38 and to know the number of rows in which the turbine are distributed. Crossing a wind farm with a single line of wind turbines is easier than those wind farms with several consecutive rows39. For this reason, the global affection of the wind farm (VWF) is understood as:
1.
The individual value of each of the wind turbine (WT) multiplied by the number of existing turbines.

2.
The total area occupied by the wind farm (AWF); in this way, it is not only considered the whole area of affection but also is established the density of the wind turbines.

3.
The number of rows that are included in the wind farm.

Based on the preceding information, the proposed formula for assessing the characteristics of the wind farm is:

$${text{VWF}} = left( {left( {{text{Ni}}*{text{WTi}}} right)/{text{AWF}}} right)^{{{1}/{text{F}}}}$$

where Ni: Number of wind turbines. WTi: Incidence of each wind turbines (the WT value will be the same for all unless in the same wind farm there were different types of wind turbines with different affection areas). AWF: Total surface of the area of study understood as the area formed by the vertical rectangle created between the furthest wind turbines from the same front line and the height of them. In the case of wind farms with more than one row, the total surface area is calculated as the sum of the surfaces of each row. F = Number of lines forming the wind farm.
Of these variables considered in the previous formula, it is only necessary to develop the affection inherent to each wind turbine (WT) that has to be calculated considering both the area of the turbine’s affection and the rotation speed of the blades.

$${text{WT}} = {text{AFM}}*{text{BRS}}$$

where AFM: Area of affection of each wind turbines, BRS: Blade rotation speed.
The area of affection of each turbine is the surface of the circumference formed by the blades (a), plus the surface of the triangle formed by the blades with the ground when they form an angle of 60° with the support tower (b), minus the intersection of both surfaces (c) (Fig. 3):

$${text{AFM}} = {text{a}} + {text{b}}{-}{text{c}}$$

Figure 3

Scheme and values to calculate the area of affection of each wind turbine. The area of affection of each turbine is the surface of the circumference formed by the blades (a), plus the surface of the triangle formed by the blades with the ground when they form an angle of 60° with the support tower (b), minus the intersection of both surfaces (c).

Full size image

Figure 4

Zoning scheme of risk areas. ZONE I: Corresponds to the free height between the ground and the blades. ZONE II: This zone corresponds to the area of the circumference formed by the blades when turning. ZONE III: Corresponds to the free height above the blades so that this interval is above the previous interval.

Full size image

Being: a = πr2   b = (sen60°*r)(L − cos 60° * r), where r is the length of the blades and L the height of the support tower. c = ((πr2/3) − ((sen60° * r)(cos 60° * r))).
To calculate the affection of the rotation speed of the blades (SB) it is assumed that the greater the rotation speed, the greater the turbulence and the greater the risk for the fauna19,40. In any case, this incidence is not linear but exponential since from a certain speed the affection can be considered high. To calculate this value, we established the following formula:

$${text{BRS}} = {1} + {text{Log}}left( {{text{SB}}} right)$$

Given that the value of the wind Farm (VWF) is the quotient between the sum of the areas affected by each turbine and the total area occupied, the value generally will be less than 1. In cases where the value is greater (when the surface area of the turbine multiplied by the rotation speed of the turbine is greater than the total surface of the area) it will be equal to 1. i.e., the maximum surface area affected cannot be greater than the surface area occupied by the total wind farm.
Location in the natural environment (UF)
Many works show the importance of selecting the location of the wind farm to minimize its impact on birdlife. However, it is possible that wind farms may be authorized in sensitive areas or in areas with poor environmental conditions (predominance of fog) or that have synergistic effects with adjacent wind farms. In this sense and as indicated in the introduction, there are four factors that can influence this impact: low visibility, proximity to sensitive areas, location in migratory crossings and synergies with other wind farms. Therefore, the value of this variable should be at least the same as that established for the previous variable (VWF). In this regard, it is proposed that the maximum value of the variables used to compute this factor should also be 1.

Visibility (VI): This variable measures the frequency of days with low visibility (fog, intense rain, etc.) compared with the total number of observation days (total number of days with low visibility/total number of observation days). The maximum value is 0.25.

Proximity to sensitive areas (ZS): Sensitive areas are those in which occur high concentrations of individuals, either because they are breeding areas, feeding areas, resting areas or roosts. Protected areas such as IBAS or LICs may also be considered. Not all species have the same radius of action, so setting a minimum radius of affection can only be established randomly. For example, for some species a radius of influence of 10 km is small but for others can be large. In any case and for having a uniform criterion, it will be considered that a sensitive area is close to the wind farm when it is located less than 10 km4, in this case, the value of this variable will be 0.25 and if they are between 10 and 50 km the value will be 0.15 while if it is more than 50 km is considered that the location of the wind farm does not influence these areas (value 0).

Migratory passes (MP): Migratory passes are those areas used by avian fauna for their daily or migratory movements. If the wind farm is located in one of these Migratory passes, the effect will be high so it will be valued with a maximum value (0.25) and the value will be minimal (0) if this is not the case.

Proximity to other wind farms (PWF): It is relevant to include this variable because of the proximity of different wind farms cause negative synergistic effects on the species by limiting the length of possible free corridors of wind turbines. In this way, the location of another wind farm less than 3 km away is considered very negative (0.25), between 3 and 5 km (0.15), between 5 and 10 km (0.10) and more than 10 km (0), it does not affect. If there is more than one wind farm in the area, the value will increase 0.25 if it is between 3 and 5 km and 0.15 if it is between 5 and 10 km.

$${text{UF}} = {text{VI}} + {text{ZS}} + {text{MP}} + {text{PWF}}$$

where WT: Value related to the location of the wind farm. VI: Predominant visibility in the area. ZS: Presence of sensitive areas in the vicinity of the wind farm. MS = Incidence of the wind farm in migratory crossings. PWF: Proximity to wind farms.

The possible maximum value for the wind farm location will be 1.
Therefore, the possible maximum value inherent in the characteristics and location of the wind farm will be 2. Substituting the values in the proposed formula:

$${text{WF}} = {text{VWF}} + {text{UF}}$$

And considering the values obtained for each mill, the wind farm in general and its location, the result is the following formula:

$${text{WF}} = left( {left( {{text{Ni}}*{text{IMi}}} right)/{text{AWF}}} right)^{{{1}/{text{F}}}} + left( {{text{VI}} + {text{ZS}} + {text{MP}} + {text{PWF}}} right)$$

Affection on the species
Not all species have the same sensitivity to the presence of the wind farm, being some of them more sensitive than others (25). On the other hand, the incidence on endangered species is not the same as that on species with stable populations in the area so, it is necessary to differentiate two types of variables related to the species: those related to the special sensitivity of each species to the presence of these infrastructures (SS) and the one inherent to its degree of threat, conservation or socioeconomic interest (VS). The affection value of each species will be the sum of the values of each type of variable. Therefore, the value of this section will be:

$${text{Affection}};{text{to}};{text{the}};{text{species}} = left( {{text{SS }} + {text{ VS}}} right)$$

Sensitivity of the species to the wind farm (SS)
These variables will be considered as the impact of the wind farm on each species due to its morphological, ethological, historical and demographic characteristics, etc. It is the closest thing to what could be understood as collision risk since it assesses the different characteristics of each bird (morphological, ethological, demographic, etc.) based on the risk of colliding with wind turbines and, valuing more those characteristics that enhance the probability of collision.
Bird size will be considered in this variable. A higher percentage of affection is detected on large birds in the majority of the recorder monitoring of the incidence of wind farms. However, this value seems to be overestimated since the detectability of carcasses of small birds and bats is lower as they remain less time on the ground30,41,42.
On the other hand, small birds show much less resistance to wind flows generated by the blades so it seems logical to think that the affection on this group of birds and on bats is higher. For this reason, a greater impact on small species has been assessed.
As a reference size, those birds smaller than or equal to a turtledove have been considered as small birds; medium-sized birds are those whose sizes are between a turtledove and a heron while those larger than a heron are considered as large birds. Considering these aspects:
The behaviour of different species will influence their risk of collision increasing the possibility of being affected by wind turbines38, for example, species that tend to go in groups show a greater risk of collision. The phenological characteristics of species are also important, for example, those species that are only in passage (prenuptial and postnuptial) will be little time in the study area but as they are not accustomed to the presence of wind turbines, probability of collision is high and possibly increased by going in large groups. Breeding species in the area are more dangerous because the young ones, still inexperienced in flight, show high risks of collision1. In other words, variables reflected in this section are related to the time the species spends in the area38, its dexterity in flight and its gregariousness. Together with these variables, the type of flight carried out by each species has been also considered: direct flights avoid staying longer in the area while cycloid or indirect movements increase the possibility of collision.

Seasonality: It considers the number of months in which the species is detected in the area. The maximum value is 1 if the species is sedentary (12 months) so each month is valued as 0.083.

Phenology: Marks the periods in which the species is present in the area. It is considered that species present in the breeding season or in passage show a greater risk than those that are only wintering. In this sense, if the species is in the breeding season will be valued with 0.75, only in winter 0.25 and 0.5 only in passage. When it appears in all seasons or in three of them, the value will be maximum (1). The value of the station will be also maximum if appears in two periods.

Flight height: In order to calculate flight height with risk for each species, the characteristics of each wind farm are considered. That is to say, they have to be adjusted to each wind farm since the interval of each zone will vary according to these ones. In this sense, for example, small birds that fly at lower altitudes can be located in the lower zone or not depending on the wind farm, just as large birds can be located in the area of the blades or above. In this sense, three zones have been established (Fig. 4):

ZONE I: Corresponds to the free height between the ground and the blades so, this interval goes from 0 m to the height resulting from subtracting the size of the blade from the length of the support tower. Value 0.5

ZONE II: This zone corresponds to the area with the greatest risk of collision since it is equivalent to the circumference formed by the blades when turning. Therefore, the interval will go from the previous height to its sum with the diameter of the circumference formed by the blades. Value: 1.

ZONE III: Corresponds to the free height above the blades so that this interval is above the previous interval. Value: 0.

When a species presents different flight heights, the one more frequent and that presents the greater risk will be selected.

Type of flight: Direct flights are considered to have a lower risk of collision than those that cause a longer stay in the area. The values will be 0.25 in direct flights and 0.5 in indirect flight.

Flock size: The risk of collision is considered higher when species show large groups so the following classification is established: One individual: 0.25; groups of 2–5 individuals: 0.5; groups of 6–10 individuals: 0.75; groups of more than 10 individuals: 1.

Historical variables (Maximum value 2).

A variable related to mortality detected in previous studies has also been included. Those species that are systematically detected in the mortality reviews of these infrastructures or exist high figures of mortality due to collision in specific wind farms should be considered.

Species with previous collision data (usual 2; medium 1; scarce 0.5; no record 0). This value is established at the discretion of the technician who performs the assessment, but as a habitual criterion, it can be considered as usual when the species appears in most of the studies (more than 30% of the studies), between 15 and 30% of the studies on average; and it will be classified as scarce if it only appears between 1 and 15% of the studies.

The last variables considered are related to the incidence on population parameters of each species. It has been considered that the species with reproductive strategy R suffer a lower incidence on the populations (although the mortality may be higher) since their reproductive efficiency partly solves this loss. However, species with K strategy suffer enormously when the mortality of young individuals is high. On the other hand, those species that frequently use the area where the wind turbines are located will show a higher probability of collision than those that are less common and those species that have high abundances in the area have also a higher probability of impact than those with few individuals16.

Survival-Fertility (type K or R) (K = 0.5; R = 0.2).

Frequency: This variable measures the frequency with which each species appears in the area in relation to the rest of the species present (total number of presences of the species/total number of presences detected). The maximum value is 1.

Abundance of the species in the area (number of individuals detected of the species i/total number of individuals detected of all species) (maximum value 1).

Species value (VS)
This value will include the conservation and socio-economic importance of the species (including the hunting value or social interest of some species). The affections on those species that are in a situation of greater risk of extinction must be considered in a relevant way, since the loss of a few individuals can represent the unfeasibility of the population. In this respect, both the degree of threat and the legal cataloguing of the different species have been considered.
The maximum value of this variable is much higher than the rest of variables since those species with the maximum protection value or degree of threat will have a value of 9. The cataloguing according to the Red Books will relate to the value established in Table 143. It has also been considered necessary to assess the socio-economic importance of some species. In this regard, it is taken into account not only the importance of hunting, which is relevant for some species of birds, but also its social importance, that is to say, those species which have conservation or recovery plans established in areas close to the different administrations or which are especially valued by the population, although their threat level is not very high (colonies of birds especially loved by the local population, etc.).
Table 1 Values given to the different classifications or threat level.
Full size table More

1.
Giljum, S., Dittrich, M., Lieber, M. & Lutter, S. Global patterns of material flows and their socio-economic and environmental implications: A MFA study on all countries world-wide from 1980 to 2009. Resources 3, 319–339 (2014).
Article  Google Scholar 
2.
IRP, U. Global Resources Outlook 2019: Natural Resources for the Future we Want. A Report of the International Resource Panel. Report No. DTI/2226/NA (United Nations Environment Programme, 2019).

3.
Krausmann, F., Schandl, H., Eisenmenger, N., Giljum, S. & Jackson, T. Material flow accounting: Measuring global material use for sustainable development. Ann. Rev. Env. Resour. 42, 647–675 (2017).
Article  Google Scholar 

4.
Calvo, G., Mudd, G., Valero, A. & Valero, A. Decreasing ore grades in global metallic mining: A theoretical issue or a global reality? Resources 5 (2016).

5.
Prior, T., Giurco, D., Mudd, G., Mason, L. & Behrisch, J. Resource depletion, peak minerals and the implications for sustainable resource management. Glob. Environ. Change 22, 577–587 (2012).
Article  Google Scholar 

6.
West, J. Decreasing metal ore grades. J. Ind. Ecol. 15, 165–168 (2011).
Article  Google Scholar 

7.
Mudd, G. M. Global trends in gold mining: Towards quantifying environmental and resource sustainability. Resour. Policy 32, 42–56 (2007).
Article  Google Scholar 

8.
Sonter, L. J., Moran, C. J., Barrett, D. J. & Soares-Filho, B. S. Processes of land use change in mining regions. J. Clean. Prod. 84, 494–501 (2014).
Article  Google Scholar 

9.
Werner, T., Bebbington, A. & Gregory, G. Assessing impacts of mining: Recent contributions from GIS and remote sensing. Extract. Indus. Soc. 6, 993–1012 (2019).
Article  Google Scholar 

10.
Kobayashi, H., Watando, H. & Kakimoto, M. A global extent site-level analysis of land cover and protected area overlap with mining activities as an indicator of biodiversity pressure. J. Clean. Prod. 84, 459–468 (2014).
Article  Google Scholar 

11.
Sonter, L. J., Ali, S. H. & Watson, J. E. M. Mining and biodiversity: key issues and research needs in conservation science. Proc. Biol. Sci. 285 (2018).

12.
Islam, K., Vilaysouk, X. & Murakami, S. Integrating remote sensing and life cycle assessment to quantify the environmental impacts of copper-silver-gold mining: A case study from laos. Resour. Conserv. Recy. 154, 104630 (2020).
Article  Google Scholar 

13.
Butt, N. et al. Biodiversity risks from fossil fuel extraction. Science 342, 425–426 (2013).
ADS  CAS  Article  Google Scholar 

14.
Murguía, D. I., Bringezu, S. & Schaldach, R. Global direct pressures on biodiversity by large-scale metal mining: Spatial distribution and implications for conservation. J. Eenviron. Manage. 180, 409–420 (2016).
Google Scholar 

15.
Endl, A., Tost, M., Hitch, M., Moser, P. & Feiel, S. Europe’s mining innovation trends and their contribution to the sustainable development goals: Blind spots and strong points. Resour. Policy 101440 (2019).

16.
Bruckner, M., Fischer, G., Tramberend, S. & Giljum, S. Measuring telecouplings in the global land system: A review and comparative evaluation of land footprint accounting methods. Ecol. Econ. 114, 11–21 (2015).
Article  Google Scholar 

17.
Schaffartzik, A. et al. Trading land: A review of approaches to accounting for upstream land requirements of traded products. J. Ind. Ecol. 19, 703–714 (2015).
Article  Google Scholar 

18.
USGS – United States Geological Survey. Mineral resources online spatial data, https://mrdata.usgs.gov/ (2018).

19.
S&P Global Market Intelligence. SNL metals and mining database, https://www.spglobal.com/marketintelligence/en/campaigns/metals-mining (2018).

20.
Murguía, D. I. & Bringezu, S. Measuring the specific land requirements of large-scale metal mines for iron, bauxite, copper, gold and silver. Prog. Ind. Ecol. 10, 264–285 (2016).
Article  Google Scholar 

21.
Werner, T. T. et al. Global-scale remote sensing of mine areas and analysis of factors explaining their extent. Glob. Environ. Change 60 (2020).

22.
Mountrakis, G., Im, J. & Ogole, C. Support vector machines in remote sensing: A review. ISPRS J. Photogramm. 66, 247–259 (2011).
Article  Google Scholar 

23.
Belgiu, M. & Dragu, L. Random forest in remote sensing: A review of applications and future directions. ISPRS J. Photogramm. 114, 24–31 (2016).
Article  Google Scholar 

24.
Zhu, X. X. et al. Deep learning in remote sensing: A comprehensive review and list of resources. IEEE Geosc. Rem. Sen. M. 5, 8–36 (2017).
Article  Google Scholar 

25.
Wulder, M. A., Coops, N. C., Roy, D. P., White, J. C. & Hermosilla, T. Land cover 2.0. Int. J. Remote Sens. 39, 4254–4284 (2018).
ADS  Article  Google Scholar 

26.
Zhu, Z. et al. Benefits of the free and open Landsat data policy. Remote Sens. Environ. 224, 382–385 (2019).
ADS  Article  Google Scholar 

27.
Petropoulos, G. P., Partsinevelos, P. & Mitraka, Z. Change detection of surface mining activity and reclamation based on a machine learning approach of multi-temporal Landsat TM imagery. Geocarto Int. 28, 323–342 (2013).
Article  Google Scholar 

28.
LaJeunesse Connette, K. J. et al. Assessment of mining extent and expansion in Myanmar based on freely-available satellite imagery. Remote Sens. 8 (2016).

29.
Yu, L. et al. Monitoring surface mining belts using multiple remote sensing datasets: A global perspective. Ore Geol. Rev. 101, 675–687 (2018).
Article  Google Scholar 

30.
Vasuki, Y. et al. The spatial-temporal patterns of land cover changes due to mining activities in the darling range, western australia: A visual analytics approach. Ore Geol. Rev. 108, 23–32 (2019).
Article  Google Scholar 

31.
Mukherjee, J., Mukherjee, J., Chakravarty, D. & Aikat, S. A novel index to detect opencast coal mine areas from Landsat 8 OLI/TIRS. IEEE J-STARS 12, 891–897 (2019).
Google Scholar 

32.
Waldrop, M. M. News Feature: What are the limits of deep learning? PNAS 116, 1074–1077 (2019).
CAS  Article  Google Scholar 

33.
EOX IT Services GmbH. Sentinel-2 cloudless (contains modified Copernicus sentinel data 2017 and 2018), https://s2maps.eu (2018).

34.
Pebesma, E. Simple Features for R: Standardized Support for Spatial Vector Data. R J. 10, 439–446 (2018).
Article  Google Scholar 

35.
Gutschlhofer, J. & Maus, V. Web application for mining area polygonization version 1.2. Zenodo https://doi.org/10.5281/zenodo.3691743 (2020).

36.
Lesiv, M. et al. Characterizing the spatial and temporal availability of very high resolution satellite imagery in Google Earth and Microsoft Bing maps as a source of reference data. Land 7 (2018).

37.
Bradshaw, A. Restoration of mined lands—using natural processes. Ecol. Eng. 8, 255–269 (1997).
Article  Google Scholar 

38.
EUROSTAT. Countries, 2016 – administrative units – dataset (generalised dataset derived from eurogeographics and UN-FAO GI data), https://ec.europa.eu/eurostat/cache/GISCO/distribution/v2/countries/ (2018).

39.
Amatulli, G. et al. A suite of global, cross-scale topographic variables for environmental and biodiversity modeling. Sci. Data 5, 180040 (2018).
Article  Google Scholar 

40.
Maus, V. et al. Global-scale mining polygons (version 1). Pangaea https://doi.org/10.1594/PANGAEA.910894 (2020).

41.
Marazuela, M., Vázquez-Suñé, E., Ayora, C., García-Gil, A. & Palma, T. The effect of brine pumping on the natural hydrodynamics of the Salar de Atacama: The damping capacity of salt flats. Sci. Total Environ. 654, 1118–1131 (2019).
ADS  CAS  Article  Google Scholar 

42.
Liu, W., Agusdinata, D. B. & Myint, S. W. Spatiotemporal patterns of lithium mining and environmental degradation in the Atacama Salt Flat, Chile. Int. J. Appl. Earth Obs. 80, 145–156 (2019).
Article  Google Scholar 

43.
Hansen, K. Brazil’s Carajás mines, NASA Earth Observatory, https://earthobservatory.nasa.gov/images/144457/brazils-carajas-mines (2018).

44.
Mining Technology. Batu Hijau copper-gold mine, Indonesia, https://www.mining-technology.com/projects/batu/ (2020).

45.
Shen, L. & Gunson, A. J. The role of artisanal and small-scale mining in China’s economy. J. Clean. Prod. 14, 427–435 (2006).
Article  Google Scholar 

46.
Shen, L., Dai, T. & Gunson, A. J. Small-scale mining in China: Assessing recent advances in the policy and regulatory framework. Resour. Policy 34, 150–157 (2009).
Article  Google Scholar 

47.
Potere, D. Horizontal positional accuracy of Google Earth’s high-resolution imagery archive. Sensors 8, 7973–7981 (2008).
Article  Google Scholar 

48.
Vajsová B & Åstrand, P. J. New sensors benchmark report on Sentinel-2A sensor over Maussane test site for CAP purposes. Report No. EUR 27674EN (Publications Office of the European Union, 2015).

49.
Cochran, W. G. Sampling Techniques. Series in Probability and Statistics (Wiley, 1977), 3 edn.

50.
Olofsson, P. et al. Good practices for estimating area and assessing accuracy of land change. Remote Sens. Environ. 148, 42–57 (2014).
ADS  Article  Google Scholar 

51.
OGC – Open Geospatial Consortium. GeoPackage Encoding Standard, https://www.geopackage.org/ (2005).

52.
OGC – Open Geospatial Consortium. Geographic tagged image file format (GeoTIFF), https://www.ogc.org/standards/geotiff (2019).

53.
The Internet Society. RFC 4180: Common format and MIME type for comma-separated values (CSV). https://tools.ietf.org/html/rfc4180 (2005).

54.
Wang, J.-F., Zhang, T.-L. & Fu, B.-J. A measure of spatial stratified heterogeneity. Ecol. Indic. 67, 250–256 (2016).
Article  Google Scholar 

55.
Brunsdon, C., Fotheringham, A. S. & Charlton, M. E. Geographically weighted regression: A method for exploring spatial nonstationarity. Geogr. Anal. 28, 281–298 (1996).
Article  Google Scholar 

56.
Brunsdon, C., Fotheringham, S. & Charlton, M. Geographically weighted regression. J. R. Stat. Soc., Ser. D Stat. 47, 431–443 (1998).
Article  Google Scholar 

57.
QGIS Development Team. QGIS geographic information system, version 3.12.0. Open Source Geospatial Foundation, https://www.qgis.org (2020).

58.
R Core Team. R: A language and environment for statistical computing, version 3.6.1. Foundation for Statistical Computing, Vienna, Austria, https://www.R-project.org (2019).

59.
Python Core Team. Python: A dynamic, open source programming language, version 2.7.17. Python Software Foundation, https://www.python.org (2019).

60.
OGC – Open Geospatial Consortium. Web map service interface standard (WMS), https://www.ogc.org/standards/wms (2020).

61.
GNU general public license, version 3. Free Software Foundation, https://www.gnu.org/licenses/gpl-3.0.en.html (2019).

62.
GDAL/OGR contributors. GDAL/OGR geospatial data abstraction software library, version 2.4.2. Open Source Geospatial Foundation, https://gdal.org (2019).

63.
Chang, W., Cheng, J., Allaire, J., Xie, Y. & McPherson, J. Shiny: Web Application Framework for R, version 1.3.2, https://CRAN.R-project.org/package=shiny (2019)

64.
The PostgreSQL Global Development Group. PostgreSQl: an open source object-relational database system, version 11.6, https://www.postgresql.org/ (2019).

65.
PostGIS Team. PostGIS: a spatial database extender for PostgreSQL object relational database, version 2.5.4. Open Source Geospatial Foundation, https://postgis.net (2019). More

Adams HD, Collins AD, Briggs SP, Vennetier M, Dickman LT, Sevanto SA et al. (2015) Experimental drought and heat can delay phenological development and reduce foliar and shoot growth in semiarid trees. Glob Change Biol 21:4210–4220
Google Scholar 

Basler D, Körner C (2012) Photoperiod sensitivity of bud burst in 14 temperate forest tree species. Agric For Meteorol 165:73–81
Google Scholar 

Bhalerao R, Keskitalo J, Sterky F, Erlandsson R, Björkbacka H, Birve SJ et al. (2003) Gene expression in autumn leaves. Plant Physiol 131:430–442
PubMed  PubMed Central  Google Scholar 

Bradshaw HD, Stettler RF (1995) Molecular genetics of growth and development in populus. IV. Mapping QTLs with large effects on growth, form, and phenology traits in a forest tree. Genetics 139:963–973
CAS  PubMed  Google Scholar 

Brelsford CC, Nybakken L, Kotilainen TK, Robson TM (2019) The influence of spectral composition on spring and autumn phenology in trees. Tree Physiol 39:925–950
CAS  PubMed  Google Scholar 

Christensen K, Frederiksen H, Vaupel JW, McGue M (2003) Age trajectories of genetic variance in physical functioning: a longitudinal study of Danish twins aged 70 years and older. Behav Genet 33:125–136
PubMed  Google Scholar 

Chuine I (2010) Why does phenology drive species distribution? Philos Trans R Soc B 365:3149–3160
Google Scholar 

Class B, Brommer JE (2020) Can dominance genetic variance be ignored in evolutionary quantitative genetic analyses of wild populations? Evolution 74:1540–1550
PubMed  Google Scholar 

Cong N, Shen M, Piao S (2016) Spatial variations in responses of vegetation autumn phenology to climate change on the Tibetan Plateau. J Plant Ecol 10:744–752.
Google Scholar 

Costa e Silva J, Borralho NMG, Potts BM (2004) Additive and non-additive genetic parameters from clonally replicated and seedling progenies of Eucalyptus globulus. Theor Appl Genet 108:1113–1119
PubMed  Google Scholar 

Cronk QCB (2005) Plant eco‐devo: the potential of poplar as a model organism. N Phytologist 166:39–48
CAS  Google Scholar 

Evans LM, Slavov GT, Rodgers-Melnick E, Martin J, Ranjan P, Muchero W et al. (2014) Population genomics of Populus trichocarpa identifies signatures of selection and adaptive trait associations. Nat Genet 46:1089
CAS  PubMed  Google Scholar 

Fabbrini F, Gaudet M, Bastien C, Zaina G, Harfouche A, Beritognolo I et al. (2012) Phenotypic plasticity, QTL mapping and genomic characterization of bud set in black poplar. BMC Plant Biol 12:47
CAS  PubMed  PubMed Central  Google Scholar 

Falconer DS, Mackay TFC (1996) Introduction to quantitative genetics. 4th edn. Longman

Fracheboud Y, Luquez V, Björkén L, Sjödin A, Tuominen H, Jansson S (2009) The control of autumn senescence in European Aspen. Plant Physiol 149:1982–1991
CAS  PubMed  PubMed Central  Google Scholar 

Hadfield JD (2010) MCMC methods for multi-response generalized linear mixed models: the MCMCglmm R package. J Stat Softw 33:1–22.
Google Scholar 

Horvath DP, Anderson JV, Chao WS, Foley ME (2003) Knowing when to grow: signals regulating bud dormancy. Trends Plant Sci 8:534–540
CAS  PubMed  Google Scholar 

Howe GT, Aitken SN, Neale DB, Jermstad KD, Wheeler NC, Chen THH (2003) From genotype to phenotype: unraveling the complexities of cold adaptation in forest trees. Can J Bot 81:1247–1266
CAS  Google Scholar 

Howe GT, Saruul P, Davis J, Chen THH (2000) Quantitative genetics of bud phenology, frost damage, and winter survival in an F2 family of hybrid poplars. Theor Appl Genet 101:632–642
Google Scholar 

Ingvarsson PK, Garcia MV, Luquez V, Hall D, Jansson S (2008) Nucleotide polymorphism and phenotypic associations Within and Around the phytochrome B2 Locus in European Aspen (Populus tremula, Salicaceae). Genetics 178:2217
CAS  PubMed  PubMed Central  Google Scholar 

Jansson S, Douglas CJ (2007) Populus: a model system for plant biology. Annu Rev Plant Biol 58:435–458
CAS  PubMed  Google Scholar 

Karacic A, Verwijst T, Weih M (2003) Above-ground woody biomass production of short-rotation populus plantations on agricultural land in Sweden. Scand J For Res 18:427–437
Google Scholar 

Kennedy BW, Schaeffer LR (1989) Genetic evaluation under an animal model when identical genotypes are represented in the population. J Anim Sci 67:1946–1955
Google Scholar 

Keskitalo J, Bergquist G, Gardeström P, Jansson S (2005) A cellular timetable of autumn senescence. Plant Physiol 139:1635–1648
CAS  PubMed  PubMed Central  Google Scholar 

Lagercrantz ULF (2009) At the end of the day: a common molecular mechanism for photoperiod responses in plants? J Exp Bot 60:2501–2515
CAS  PubMed  Google Scholar 

Leuchner M, Menzel A, Werner H (2007) Quantifying the relationship between light quality and light availability at different phenological stages within a mature mixed forest. Agric For Meteorol 142:35–44
Google Scholar 

Luquez V, Hall D, Albrectsen BR, Karlsson J, Ingvarsson P, Jansson S (2008) Natural phenological variation in aspen (Populus tremula): the SwAsp collection. Tree Genet Genomes 4:279–292
Google Scholar 

Lynch M, Walsh B (1998) Genetics and analysis of quantitative traits. Sinauer Sunderland

MacKenzie CM, Primack RB, Miller-Rushing AJ (2018) Local environment, not local adaptation, drives leaf-out phenology in common gardens along an elevational gradient in Acadia National Park, Maine. Am J Bot 105:986–995
Google Scholar 

Marron N, Storme V, Dillen SY, Bastien C, Ricciotti L, Salani F et al. (2010) Genomic regions involved in productivity of two interspecific poplar families in Europe. 2. Biomass production and its relationships with tree architecture and phenology. Tree Genet Genomes 6:533–554
Google Scholar 

McKown AD, Guy RD, Klápště J, Geraldes A, Friedmann M, Cronk QC et al. (2014a) Geographical and environmental gradients shape phenotypic trait variation and genetic structure in Populus trichocarpa. N. Phytologist 201:1263–1276
CAS  Google Scholar 

McKown AD, Klápště J, Guy RD, Geraldes A, Porth I, Hannemann J et al. (2014b) Genome‐wide association implicates numerous genes underlying ecological trait variation in natural populations of Populus trichocarpa. N. Phytologist 203:535–553
CAS  Google Scholar 

McKown AD, Klápště J, Guy RD, El-Kassaby YA, Mansfield SD (2018) Ecological genomics of variation in bud-break phenology and mechanisms of response to climate warming in Populus trichocarpa. N. Phytologist 220:300–316
CAS  Google Scholar 

Michelson I, Ingvarsson Pär K, Robinson Kathryn M, Edlund E, Eriksson Maria E, Nilsson O et al. (2017) Autumn senescence in aspen is not triggered by day length. Physiol Plant 162:123–134
PubMed  Google Scholar 

Olson M, Levsen N, Soolanayakanahally Raju Y, Guy Robert D, Schroeder William R, Keller Stephen R et al. (2013) The adaptive potential of Populus balsamifera L. to phenology requirements in a warmer global climate. Mol Ecol 22:1214–1230
CAS  PubMed  Google Scholar 

Pliura A, Suchockas V, Sarsekova D, Gudynaitė V (2014) Genotypic variation and heritability of growth and adaptive traits, and adaptation of young poplar hybrids at northern margins of natural distribution of Populus nigra in Europe. Biomass Bioenergy 70:513–529
Google Scholar 

Polgar CA, Primack RB (2011) Leaf‐out phenology of temperate woody plants: from trees to ecosystems. N Phytologist 191:926–941
Google Scholar 

Porth I, El‐Kassaby YA (2015) Using Populus as a lignocellulosic feedstock for bioethanol. Biotechnol J 10:510–524
CAS  PubMed  Google Scholar 

Porth I, Klápště J, McKown AD, La Mantia J, Guy RD, Ingvarsson PK et al. (2015) Evolutionary quantitative genomics of Populus trichocarpa. PLoS ONE 10:e0142864
PubMed  PubMed Central  Google Scholar 

Porth I, Klápště J, McKown AD, La Mantia J, Hamelin RC, Skyba O et al. (2014) Extensive functional pleiotropy of REVOLUTA substantiated through forward genetics. Plant Physiol 164:548–554
CAS  PubMed  Google Scholar 

Rohde A, Bastien C, Boerjan W (2011b) Temperature signals contribute to the timing of photoperiodic growth cessation and bud set in poplar. Tree Physiol 31:472–482
PubMed  Google Scholar 

Rohde A, Storme V, Jorge V, Gaudet M, Vitacolonna N, Fabbrini F et al. (2011a) Bud set in poplar – genetic dissection of a complex trait in natural and hybrid populations. N Phytologist 189:106–121
CAS  Google Scholar 

Sannigrahi P, Ragauskas AJ, Tuskan GA (2010) Poplar as a feedstock for biofuels: a review of compositional characteristics. Biofuels Bioprod Bioref 4:209–226
CAS  Google Scholar 

Singh RK, Svystun T, AlDahmash B, Jönsson AM, Bhalerao RP (2017) Photoperiod- and temperature-mediated control of phenology in trees—a molecular perspective. N Phytologist 213:511–524
CAS  Google Scholar 

Slavov GT, Leonardi S, Burczyk J, Adams WT, Strauss Sh, DiFazio SP (2009) Extensive pollen flow in two ecologically contrasting populations of Populus trichocarpa. Mol Ecol 18:357–373

Triozzi PM, Ramos-Sánchez JM, Hernández-Verdeja T, Moreno-Cortés A, Allona I, Perales M (2018) Photoperiodic regulation of shoot apical growth in poplar. Front Plant Sci 9:1030–1030
PubMed  PubMed Central  Google Scholar 

Vitasse Y, Delzon S, Dufrêne E, Pontailler J-Y, Louvet J-M, Kremer A et al. (2009) Leaf phenology sensitivity to temperature in European trees: Do within-species populations exhibit similar responses? Agric For Meteorol 149:735–744
Google Scholar 

Walsh B, Lynch M (2018) Evolution and selection of quantitative traits. Oxford University Press

Weih M (2004) Intensive short rotation forestry in boreal climates: present and future perspectives. Can J For Res 34:1369–1378
Google Scholar 

Wilson AJ, Kruuk LEB, Coltman DW (2005) Ontogenetic patterns in heritable variation for body size: using random regression models in a wild ungulate population. Am Naturalist 166:E177–E192
Google Scholar 

Yu Q, Tigerstedt PMA, Haapanen M (2001) Growth and phenology of hybrid aspen clones (Populus tremula L. x Populus tremuloides Michx.)

Zanewich KP, Pearce DW, Rood SB (2018) Heterosis in poplar involves phenotypic stability: cottonwood hybrids outperform their parental species at suboptimal temperatures. Tree Physiol 38:789–800
CAS  PubMed  Google Scholar 

Zhou L, Bawa R, Holliday JA (2014) Exome resequencing reveals signatures of demographic and adaptive processes across the genome and range of black cottonwood (Populus trichocarpa). Mol Ecol 23:2486–2499
CAS  PubMed  Google Scholar  More

1.
Schminke, H. K. Entomology for the copepodologist. J. Plankton Res. 29, i149–i162, https://doi.org/10.1093/plankt/fbl073 (2007).
Article  Google Scholar 
2.
Le Borgne, R. Equivalences between the measures of biovolumes, dry weight, ashfree dry weight, carbon, nitrogen and phosphorus of the mesozooplankton of the tropical Atlantic. Cah. O.R.S.T.O.M. Ser. Oceanogr. 13, 179–196 (1975).
Google Scholar 

3.
Corral Estrada, J. Contribucion al conocimiento del plancton de Canarias. Estudio cuantitativo, sistematico y observaciones ecologicas de nos Copépodos epipelagicos en la zona de Santa Cruz de Tenerife en el curso de un ciclo anual. Publnes Fac. Cienc. Madrid, Seccion de Biol., (A) 129 Doct. thesis, Univ. Madrid (1970).

4.
Lovegrove, T. In Some contemporary studies in marine science (ed Barnes, H.) 429-467 (Allen and Unwin, 1966).

5.
Harris, R., Wiebe, P., Lenz, J., Skjoldal, H. R. & Huntley, M. ICES zooplankton methodology manual. (Elsevier, 2000).

6.
McEnnulty, F. R. et al. The Australian Zooplankton Biomass Database (1932–2019). Australian Ocean Data Network https://doi.org/10.26198/5c4170d42ab24 (2019).

7.
Davies, C. H. et al. Over 75 years of zooplankton data from Australia. Ecology 95, 3229–3229, https://doi.org/10.1890/14-0697.1 (2014).
Article  Google Scholar 

8.
Davies, C. H. et al. A database of marine phytoplankton abundance, biomass and species composition in Australian waters (vol 3, 160043, 2016). Sci. Data 3, https://doi.org/10.1038/sdata.2016.111.

9.
Davies, C. H. et al. A database of chlorophyll a in Australian waters. Sci. Data 5, https://doi.org/10.1038/sdata.2018.18

10.
Skerratt, J. H. et al. Simulated nutrient and plankton dynamics in the Great Barrier Reef (2011–2016). J. Mar. Syst. 192, 51–74, https://doi.org/10.1016/j.jmarsys.2018.12.006 (2019).
Article  Google Scholar 

11.
Bernardi, D. In A manual on methods for the assessment of secondary productivity in freshwaters (eds Downing, J.A & Rigler, F.H.) 59–86 (Blackwell Scientific, Oxford, 1984).

12.
Omori, M. & Ikeda, T. Methods in marine zooplankton ecology. (Wiley, 1984).

13.
Richardson, A. J. et al. Using continuous plankton recorder data. Prog. Oceanogr. 68, 27–74, https://doi.org/10.1016/j.pocean.2005.09.011 (2006).
ADS  Article  Google Scholar 

14.
Everett, J. D. et al. Modeling What We Sample and Sampling What We Model: Challenges for Zooplankton Model Assessment. Front. Mar. Sci. 4, https://doi.org/10.3389/fmars.2017.00077 (2017).

15.
Herman, A. W., Beanlands, B. & Phillips, E. F. The next generation of Optical Plankton Counter: The Laser-OPC. J. Plankton Res. 26, 1135–1145, https://doi.org/10.1093/plankt/fbh095 (2004).
Article  Google Scholar 

16.
Bivand, R. & Lewin-Koh, N. maptools: Tools for Handling Spatial Objects. R package version 0.9–5, https://CRAN.R-project.org/package=maptools. (2019).

17.
Amante, C. & Eakins, B. W. ETOPO1 1 Arc-Minute Global Relief Model: Procedures, Data Sources and Analysis. NOAA Technical Memorandum NESDIS. NGDC-24, https://doi.org/10.7289/V5C8276M.19 (NOAA., National Geophysical Data Center, 2009).

18.
Pante, E. & Simon-Bouhet, B. marmap: A Package for Importing, Plotting and Analyzing Bathymetric and Topographic Data in R. Plos One 8, https://doi.org/10.1371/journal.pone.0073051 (2013).

19.
Wiebe, P. H. & Benfield, M. C. From the Hensen net toward four-dimensional biological oceanography. Prog. Oceanogr. 56, 7–136, https://doi.org/10.1016/s0079-6611(02)00140-4 (2003).
ADS  Article  Google Scholar 

20.
Smith, P. E., Counts, R. C. & Clutter, R. I. Changes in the filtering efficiency of plankton nets due to clogging under tow. Cons. Perm. Int. p. l’Expl. d. l. Mer 32, 232–248 (1968).
Article  Google Scholar 

21.
Jackson, C. J., Rothlisberg, P. C. & Pendrey, R. C. Role of larval distribution and abundance in overall life-history dynamics: a study of the prawn Penaeus semisulcatus in Albatross Bay, Gulf of Carpentaria, Australia. Mar. Ecol. Prog. Ser. 213, 241–252, https://doi.org/10.3354/meps213241 (2001).
ADS  Article  Google Scholar 

22.
Hernroth, L. Sampling and filtration efficiency of 2 commonly used plankton nets – A comparative study of the Nansen net and the UNESCO WP-2 net. J. Plankton Res. 9, 719–728, https://doi.org/10.1093/plankt/9.4.719 (1987).
Article  Google Scholar 

23.
Verheye, H. M., Richardson, A. J., Hutchings, L., Marska, G. & Gianakouras, D. Long-term trends in the abundance and community structure of coastal zooplankton in the southern Benguela system, 1951–1996. S. Afr. J. Mar. Sci. 19, 317–332 (1998).
Article  Google Scholar 

24.
Colton, J. B., Green, J. R., Byron, R. R. & Frisella, J. L. Bongo net retention rates as effected by towing speed and mesh size. Can. J. Fish. Aquat.Sci. 37, 606–623, https://doi.org/10.1139/f80-077 (1980).
Article  Google Scholar 

25.
Pillar, S. C. A comparison of the performance of four zooplankton samplers. S. Afr. J. Mar. Sci. 2, 1–18 (1984).
Article  Google Scholar 

26.
McKinnon, A. D. et al. Zooplankton Growth, Respiration and Grazing on the Australian Margins of the Tropical Indian and Pacific Oceans. Plos One 10, e0140012, https://doi.org/10.1371/journal.pone.0140012 (2015).
CAS  Article  PubMed  PubMed Central  Google Scholar 

27.
Beers, J. R. Determination of zooplankton biomass. Monogr. Oceanogr. Methodol. 4, 35–84 (1976).
Google Scholar 

28.
Postel, L. et al. 83–192 (Elsevier, 2000).

29.
Motoda, S. Plankton sampler for collecting uncontaminated materials from several zones by a single vertical haul. Rapp. P.-V. Réun. Cons. Int. Explor. Mer. 153, 55–58 (1962).
Google Scholar 

30.
Motoda, S. & Osawa, K. Filteration ratio, variance of samples and estimated distance of haul in vertical hauls with Indian Ocean standard net. Inf. Bull. Planktol. Japan 11, 11–24 (1964).
Google Scholar 

31.
Wiebe, P. H., Boyd, S. & Cox, J. L. Relationships between zooplankton displacement volume, wet weight, dry weight, and carbon. Fish. Bull. (Wash. D. C.) 73, 777–796 (1975).
Google Scholar 

32.
Wiebe, P. H. Functional regression equations for zooplankton displacement volume, wet weight, dry weight, and carbon: a correction. U S Fish. Wildl. Serv. Fish. Bull. 86, 833–835 (1988).
Google Scholar 

33.
Moriarty, R. & O’Brien, T. D. Distribution of mesozooplankton biomass in the global ocean. Earth Syst. Sci. Data 5, 45–55, https://doi.org/10.5194/essd-5-45-2013 (2013).
ADS  Article  Google Scholar 

34.
Omori, M. Some factors affecting on dry-weight, organic weight and concentrations of carbon and nitrogen in freshlt prepared and in preserved zooplankton. Int. Rev. Gesamt. Hydrobiol. 63, 261–269, https://doi.org/10.1002/iroh.19780630211 (1978).
Article  Google Scholar 

35.
Balvay, P. G. Equivalence entre quelques paramètres estimatifs de l’abondance du zooplancton total. Swiss J. Hydrol. 49, 75–84, https://doi.org/10.1007/BF02540381 (1987).
Article  Google Scholar 

36.
Tranter, D. J. Comparison of zooplankton biomass determinations by Indian Ocean Standard Net, Juday Net and Clarke-Bumpus sampler. Nature 198, 1179, https://doi.org/10.1038/1981179a0 (1963).
ADS  Article  Google Scholar 

37.
DeVries, D. R. & Stein, R. A. Comparison of three zooplankton samplers: a taxon-specific assessment. J. Plankton Res. 13, 53–59, https://doi.org/10.1093/plankt/13.1.53 (1991).
Article  Google Scholar 

38.
Skjoldal, H. R. et al. Intercomparison of zooplankton (net) sampling systems: Results from the ICES/GLOBEC sea-going workshop. Prog. Oceanogr. 108, 1–42, https://doi.org/10.1016/j.pocean.2012.10.006 (2013).
ADS  Article  Google Scholar 

39.
Eriksen, R. S. et al. Australia’s Long-Term Plankton Observations: The Integrated Marine Observing System National Reference Station Network. Front. Mar. Sci. 6, https://doi.org/10.3389/fmars.2019.00161 (2019).

40.
O’Brien, T. D. COPEPOD: The Global Plankton Database. An overview of the 2014 database contents, processing methods, and access interface. U.S. Dep. Commerce, NOAA Tech. Memo., NMFS-F/ST-37. 29 p. (NOAA, 2014).

41.
O’Brien, T. D. COPEPOD: The Global Plankton Database. An overview of the 2010 database contents, processing methods, and access interface. U.S. Dep. Commerce, NOAA Tech. Memo., NMFS-F/ST-36. 28 p. (NOAA, 2010).

42.
O’Brien, T. D. COPEPOD: The Global Plankton Database. A review of the 2007 database contents and new quality control methodology. U.S. Dep. Commerce, NOAA Tech. Memo., NMFS-F/ST-34. 28 p. (NOAA, 2007).

43.
O’Brien, T. D. COPEPOD: A Global Plankton Database. U.S. Dep. Commerce, NOAA Tech. Memo. NMFS-F/SPO-73. 136 p. (NOAA, 2002).

44.
O’Brien, T. D., Wiebe, P. H. & Falkenhaug, T. ICES Zooplankton Status Report 2010/2011. ICES Cooperative Research Report. 318, 208 (2013).
Google Scholar 

45.
Clark, R. A., Frid, C. L. J. & Batten, S. A critical comparison of two long-term zooplankton time series from the central-west North Sea. J. Plankton Res. 23, 27–39 (2001).
Article  Google Scholar 

46.
John, E. H., Batten, S. D., Harris, R. P. & Hays, G. C. Comparison between zooplankton data collected by the Continuous Plankton Recorder survey in the English Channel and by WP-2 nets at station L4, Plymouth (UK). J. Sea Res. 46, 223–232 (2001).
ADS  Article  Google Scholar 

47.
Pitois, S. G. & Fox, C. J. Long-term changes in zooplankton biomass concentration and mean size over the Northwest European shelf inferred from Continuous Plankton Recorder data. ICES J. Mar. Sci. 63, 785–798, https://doi.org/10.1016/j.iccsjms.2006.03.009 (2006).
CAS  Article  Google Scholar 

48.
Richardson, A. J., John, E. H., Irigoien, X., Harris, R. P. & Hays, G. C. How well does the continuous plankton recorder (CPR) sample zooplankton? A comparison with the Longhurst Hardy Plankton Recorder (LHPR) in the northeast Atlantic. Deep Sea Res. Part I. 51, 1283–1294, https://doi.org/10.1016/j.dsr.2004.04.002 (2004).

49.
Herman, A. W. Design and calibration of a new optical plankton counter capable of sizing small zooplankton. Deep Sea Res. Part I. 39, 395–415, https://doi.org/10.1016/0198-0149(92)90080-D (1992).
ADS  Article  Google Scholar 

50.
Espinasse, B. et al. Conditions for assessing zooplankton abundance with LOPC in coastal waters. Prog. Oceanogr. 163, 260–270, https://doi.org/10.1016/j.pocean.2017.10.012 (2018).
Article  Google Scholar 

51.
Baird, M. E. et al. Biological properties across the Tasman Front off southeast Australia. Deep Sea Res. Part I. 55, 1438–1455, https://doi.org/10.1016/j.dsr.2008.06.011 (2008).
Article  Google Scholar 

52.
Everett, J. D., Baird, M. E. & Suthers, I. M. Three-dimensional structure of a swarm of the salp Thalia democratica within a cold-core eddy off southeast Australia. J. Geophys. Res. 116, https://doi.org/10.1029/2011jc007310 (2011).

53.
Baird, M. E., Everett, J. D. & Suthers, I. M. Analysis of southeast Australian zooplankton observations of 1938–42 using synoptic oceanographic conditions. Deep Sea Res. Part II. 58, 699–711, https://doi.org/10.1016/j.dsr2.2010.06.002 (2011).
ADS  Article  Google Scholar 

54.
Brady, G. S. Report on the Copepoda collected by H.M.S. ‘Challenger’ during the years 1873–76. Rep. Sci. Results, Voy. of H. M. S. “Challenger,” Zool. viii, 1–142 (1883).
Google Scholar 

55.
Knox, G. A. Biology of the Southern Ocean. Second edition (CRC Press/Taylor & Francis Group, 2007).

56.
Scott, A. The Copepoda of the Siboga Expedition, Part I. Free swimming, littoral and semi-parasitic Copepoda Siboga expeditie. 29a, 323 p. (1909).

57.
Farran, G. P. Crustacea, Part X. Copepoda. Natural History Report 8 no. 3. (Trustees Brit. Mus., London, 1929).

58.
Brady, G. S. Copepoda. AAE 1911-1914 Ser C. 5, 1–48 (1918).
Google Scholar 

59.
Russell, F. S. & Colman, J. S. Great Barrier Reef expedition 1928-1929. The zooplankton I. Gear, methods and station list. 2, 2–35 (The Bristish Museum, London, 1931).

60.
Marshall, S. M. The production of microplankton in the Great Barrier Reef region. Sci. Rpts. Gt. Barrier Reef Exped. London 1928–29 2, 111–157 (1933).
Google Scholar 

61.
Jesperson, P. Quantative investigation of the distribution of the macroplankton in different oceanic regions. Rep. “Dana” Exped. 2, 1–44 (1935).
Google Scholar 

62.
Sheard, K. Plankton of the Australian-Antarctic quadrant. Rep. BANZ Antarct. Res. Exped. ser. B 6, 1–120 (1947).
Google Scholar 

63.
Dakin, W. J. & Colefax, A. The marine plankton of the coastal waters of New South Wales. I. The chief planktonic forms and their seasonal distribution. Proc. Linn. Soc. N.S.W. 58, 186–222 (1933).
Google Scholar 

64.
Dakin, W. J. & Colefax, A. The plankton of the Australian coastal waters off New South Wales Part I. Publ. Univ. Syd. 210 (1940).

65.
Thompson, H. Pelagic Tunicates in the plankton of South-eastern Australian Waters, and their place in Oceanographic studies. Bull. CSIR. Aust. 153, 1–51 (1942).
Google Scholar 

66.
Foxton, P. The distribution of the standing crop of zooplankton in the Southern Ocean. Discovery Reports 28, 191–236 (1956).
Google Scholar 

67.
Smith, J. A. et al. A database of marine larval fish assemblages in Australian temperate and subtropical waters. Sci. Data 5, 180207, https://doi.org/10.1038/sdata.2018.207 (2018).
Article  PubMed  PubMed Central  Google Scholar 

68.
Brewer, D. T. et al. Impacts of gold mine waste disposal on a tropical pelagic ecosystem. Mar. Pollut. Bull. 64, 2790–2806, https://doi.org/10.1016/j.marpolbul.2012.09.009 (2012).
CAS  Article  PubMed  Google Scholar 

69.
Dennis, D. et al. Baseline assessment of the status of the tropical marine benthic and pelagic communities adjacent to and distant from the nickel/cobalt refinery site at Basamuk, PNG. Final Report for Ramu NiCo Management Ltd.Commericial-in-confidence (pp. 208. CSIRO Marine and Atmospheric Research, Australia, 2009).
Google Scholar 

70.
Duggan, S., McKinnon, A. D. & Carleton, J. H. Zooplankton in an Australian tropical estuary. Estuar. Coasts 31, 455–467 (2008).
Article  Google Scholar 

71.
McKinnon, A. D., Duggan, S., Holliday, D. & Brinkman, R. Plankton community structure and connectivity in the Kimberley-Browse region of NW Australia. Estuar. Coast Shelf Sci. 153, 156–167, https://doi.org/10.1016/j.ecss.2014.11.006 (2015).
ADS  Article  Google Scholar 

72.
McKinnon, A. D., Duggan, S., Boettger-Schnack, R., Gusmaoa, L. F. M. & O’Leary, R. A. Depth structuring of pelagic copepod biodiversity in waters adjacent to an Eastern Indian Ocean coral reef. J. Nat. Hist. 47, 639–665 (2013).
Article  Google Scholar 

73.
McKinnon, A. D. Planning tools for environmentally sustainable tropical finfish cage culture in Indonesia and northern Australia. Final Report, ACIAR Project FIS 2003/027 (Australian Centre for International Agricultural Research, Canberra, 2009).

74.
Strzelecki, J., Koslow, J. A. & Waite, A. Comparison of mesozooplankton communities from a pair of warm- and cold-core eddies off the coast of Western Australia. Deep Sea Res. Part II. 54, 1103–1112, https://doi.org/10.1016/j.dsr2.2007.02.004 (2007).
ADS  Article  Google Scholar 

75.
McKinnon, A. D., Duggan, S. & De’ath, G. Mesozooplankton dynamics in nearshore waters of the Great Barrier Reef. Estuar. Coast Shelf Sci. Science 63, 497–511, https://doi.org/10.1016/j.ecss.2004.12.011 (2005).
ADS  CAS  Article  Google Scholar 

76.
AIMS. Biological Oceanographic Reconnaissance of the Arafura Sea. https://apps.aims.gov.au/metadata/view/7b5b2d16-052e-054a098-a392-e051a835e64746 (AIMS Data Centre, Australia, 2019).

77.
McKinnon, A. D. & Duggan, S. Summer copepod production in subtropical waters adjacent to Australia’s North West Cape. Mar. Biol. (Berl) 143, 897–907, https://doi.org/10.1007/s00227-003-1153-1 (2003).
Article  Google Scholar 

78.
Taw, N. Studies on the zooplankton and hydrology of south-eastern coastal waters of Tasmania. PhD thesis, Univ. Tasmania (1975).

79.
Terauds, A. In Tasmanian Slope Trophodynamics: Final Report. FRDC Project 91/7 (ed Parslow, J. et al.) 213 (CSIRO Division of Fisheries, Australia, 1995).

80.
Pausina, S. et al. In Moreton Bay Quandamooka & Catchment: Past, present and future. (eds Rothlisberg, P. C. et al.) The Moreton Bay Foundation. Brisbane, Australia., pp. 335–360 (The Moreton Bay Foundation, Australia, 2019).

81.
Lynch, T. P. et al. In Oceans 2008, Vols 1-4 Oceans-Ieee 367–374 (Ieee, 2008).

82.
Armstrong, A. O. et al. Prey Density Threshold and Tidal Influence on Reef Manta Ray Foraging at an Aggregation Site on the Great Barrier Reef. Plos One 11, https://doi.org/10.1371/journal.pone.0153393 (2016).

83.
Furnas, M. J., Mitchell, A. W., Gilmartin, M. & Revelante, N. Phytoplankton biomass and primary production in semi-enclosed reef lagoons of the central Great Barrier Reef, Australia. Coral Reefs 9, 1–10, https://doi.org/10.1007/bf00686716 (1990).
ADS  Article  Google Scholar 

84.
CSIRO. Oceanographical observations in the Indian Ocean in 1959. H.M.A.S. Diamantina cruises Dm1/59 and Dm2/59. Oceanographical Cruise Report 1 (Division of Fisheries and Oceanography, CSIRO Australia, 1962).

85.
CSIRO. Oceanographical observations in the Indian Ocean in 1960. H.M.A.S. Diamantina cruise Dm1/60. Oceanographical Cruise Report 2 (Division of Fisheries and Oceanography, CSIRO Australia, 1962).

86.
CSIRO. Oceanographical observations in the Indian Ocean in 1961. H.M.A.S. Diamantina cruise Dm2/61. Oceanographical Cruise Report 9 (Division of Fisheries and Oceanography, CSIRO Australia, 1963).

87.
CSIRO. Oceanographical observations in the Indian Ocean in 1961. H.M.A.S. Diamantina cruise Dm1/61. Oceanographical Cruise Report 7 (Division of Fisheries and Oceanography, CSIRO Australia 1963).

88.
CSIRO. Oceanographical observations in the Indian Ocean in 1960. H.M.A.S. Diamantina cruise Dm2/60. Oceanographical Cruise Report 3 (Division of Fisheries and Oceanography, CSIRO Australia, 1963).

89.
CSIRO. Oceanographical observations in the Indian Ocean in 1961. H.M.A.S. Diamantina cruise Dm3/61. Oceanographical Cruise Report 11 (Division of Fisheries and Oceanography, CSIRO Australia, 1964).

90.
CSIRO. Oceanographical observations in the Pacific Ocean in 1960. H.M.A.S. Gascoyne cruises G1/60 and G2/60. Oceanographical Cruise Report 5 (Division of Fisheries and Oceanography, CSIRO Australia, 1962).

91.
CSIRO. Oceanographical observations in the Pacific Ocean in 1961. H.M.A.S. Gascoyne cruise G1/61. Oceanographical Cruise Report 8 (Division of Fisheries and Oceanography, CSIRO Australia, 1963).

92.
CSIRO. Oceanographical observations in the Pacific Ocean in 1961. H.M.A.S. Gascoyne cruise G4/61. Oceanographical Cruise Report 12 (Division of Fisheries and Oceanography, CSIRO Australia, 1967).

93.
CSIRO. Oceanographical observations in the Pacific Ocean in 1962. H.M.A.S. Gascoyne cruise G1/62. Oceanographical Cruise Report 13 (Division of Fisheries and Oceanography, CSIRO Australia, 1967).

94.
Crooks, A. D. Coastal hydrological investigations in the New South Wales tuna fishing area, 1960. Oceanographical Station List 53 (Division of Fisheries. CSIRO Australia, 1963).

95.
CSIRO. Investigations by F.R.V. Derwent Hunter on the eastern Australian tuna grounds in 1961. Oceanographical Station List 54 (Division of Fisheries. CSIRO Australia, 1968).

96.
CSIRO. Investigations by F.R.V. Derwent Hunter on the eastern Australian tuna grounds in 1962. Oceanographical Station List 59 (Division of Fisheries, CSIRO Australia, 1968).

97.
CSIRO. Coastal investigations at Port Hacking, New South Wales, 1961. Oceanographical Station List 81 (Division of Fisheries, CSIRO Australia, 1961).

98.
CSIRO. F.R.V. “Derwent Hunter”. Scientific report of Cruises 13/57-16/57. Report of Division of Fisheries and Oceanography 21 (Division of Fisheries and Oceanography, CSIRO Australia, 1959).

99.
CSIRO. F.R.V. Derwent Hunter. Scientific report of Cruises 5/57-8/57. Report of Division of Fisheries and Oceanography 19 (Division of Fisheries and Oceanography, CSIRO Australia, 1958).

100.
CSIRO. F.R.V. Derwent Hunter. Scientific report of Cruises DH 9/57-12/57. Report of Division of Fisheries and Oceanography 20 (Division of Fisheries and Oceanography, CSIRO Australia 1959).

101.
Rothlisberg, P. C. & Jackson, C. J. Temporal & spatial variation of plankton abundance in the Gulf of Carpentaria, Australia 1975–1977. J. Plankton Res. 4, 19–40, https://doi.org/10.1093/plankt/4.1.19 (1982).
Article  Google Scholar 

102.
CSIRO. RV Southern Surveyor dropnet data files: C2012/7164, C2012/7171, C2012/7166. (CSIRO Oceans & Atmosphere DataCentre, Australia, 2012).

103.
Ikeda, T. et al. Biological, chemical and physical observations in inshore waters of the Great Barrier Reef, North Queensland 1975–1978. AIMS Techn. Bull. AIMS-OS-80-1. 56 (AIMS, Australia, 1980).

104.
Tranter, D. J. Zooplankton abundance in Australasian waters. Aust. J. Mar. Freshwater Res. 13, 106–142 (1962).
MathSciNet  Article  Google Scholar 

105.
Young, J. W., Bradford, R. W., Lamb, T. D. & Lyne, V. D. Biomass of zooplankton and micronekton in the southern bluefin tuna fishing grounds off eastern Tasmania, Australia. Mar. Ecol. Prog. Ser. 138, 1–14 (1996).
ADS  Article  Google Scholar 

106.
CSIRO. Australian Equatorial JGOFS Dataset. CSIRO Division of Marine Research, http://www.marine.csiro.au/datacentre/JGOFSweb/inventory/index.htm (2000).

107.
Motoda, S., Kawamura, T. & Taniguchi, A. Differences in productivities between the Great Australian Bight and the Gulf of Carpentaria, in summer. Mar. Biol. (Berl) 46, 93–99, https://doi.org/10.1007/bf00391524 (1978).
CAS  Article  Google Scholar 

108.
Sheard, K. Plankton characteristics at the Cronulla Onshore Station, New South Wales 1943-46. Bull. CSIR. Aust. 246, 1–23 (1949).
Google Scholar 

109.
Thompson, P., Lourey, M., Strzelecki, J., Wild-Allen, K. & McLaughlin, J. In Final report. Southwest Australian Coastal Biogeochemistry (ed J. Keesing) (CSIRO, 2011).

110.
Davis, T. L. O. & Clementson, L. A. Data report on the vertical and horizontal distribution of tuna larvae in the east Indian Ocean, January-February 1987. CSIRO Marine Laboratories Report (CSIRO Australia, 1990).

111.
van Ruth, P. D. Spatial and temporal variation in primary and secondary productivity in the easten Great Australian Bight PhD thesis, Univ. Adelaide (2009).

112.
Stevens, J. D., Hausfeld, H. F. & Davenport, S. R. Observations on the biology, distribution and abundance of Trachurus declivis, Sardinops neopilchardus and Scomber australasicus in the Great Australian Bight. CSIRO Marine Laboratories Report 27 (CSIRO Australia, 1984).

113.
Clementson, L. A., Harris, G. P., Griffiths, F. B. & Rimmer, D. W. Seasonal and inter-annual variability in chemical and biological parameters in Storm Bay, Tasmania. 1. Physics, chemistry and the biomass of components of the food chain. Aust. J. Mar. Freshwater Res. 40, 25–38 (1989).
CAS  Article  Google Scholar 

114.
Swadling, K. M., Eriksen, R. S., Beard, J. M. & Crawford, C. M. Marine currents, nutrients and plankton in the coastal waters of south eastern Tasmania and responses to changing weather patterns. (FRDC Project No 2014/031) 99 (University of Tasmania, Australia, 2017). More

Heterogeneous and affluence-influenced footprints of APAC
The amount of natural resource extractions, environmental emissions, and socio-economic influences associated with satisfying the final demand of average person vary significantly among the eight APAC countries/regions (Fig. 1). The variations correlate with differences in affluence levels (Supplementary Table 2), which is consistent with the findings of previous footprint research focusing on GHG emissions4, and blue water consumption42 across countries worldwide. While the footprints of lower-income countries (e.g., India, Indonesia, RoAP, and China) have also increased with poverty alleviation, yet, most of them are still below the global averages by 2015. In addition, based on the varying environmental footprints (i.e., blue water, energy, and GHG) of the region’s six countries assessed annually from 1995–2015, we find some evidence for the Environmental Kuznets Curve (EKC) hypothesis, i.e., environmental pollution first rises and then falls as economic development proceeds (Supplementary Fig. 1 and Supplementary Fig. 2). Previously, few studies have addressed the socio-economic implications driven by final demand, such as the labor inputs required (employment) and revenues generated (value added)1. Here we find that a country/region’s employment and value-added footprints are both positively correlated with its affluence level.
Fig. 1: The environmental-social-economic footprints of APAC countries/regions in 1995 and 2015.

The six footprint indicators fall into three categories and are presented in a natural resource, b local and global environmental threats, c socio-economic effects. The eight countries/regions are aligned on the y-axis following a descending order by average income per capita in 2015. Footprints are further broken down to the impacts occurred within the country/region, abroad in other APAC countries/regions, and abroad in non-APAC countries/regions. In each subplot, the dashed line indicates the world average level. GHG (CO2, N2O, and CH4) are measured in CO2 equivalent based on the 100-year Global Warming Potential (GWP100). Note, in the plots for value added, we use the insets to show the footprint compositions of the last four countries/region. China refers to Chinese mainland, and we call Taiwan, China as Taiwan for short in the rest.

Full size image

Affluence levels also affected the geo-compositions of the footprint indicators in the APAC region. Richer APAC economies showed high and increasing reliance on outsourcing blue water consumption, PM2.5 emissions, and labor abroad, especially within the APAC region. Specifically, 58, 56, and 52% of the blue water footprints of Japan, South Korea and Taiwan were traced to water consumed in other APAC economies in 2015 (Fig. 1), primarily through the intra-APAC trade of agricultural products, such as sugar cane/sugar beet and paddy rice (Supplementary Fig. 3). In contrast, the final demand of middle- to low-income APAC economies has been largely satisfied by domestic natural and labor resources over the two decades. Moreover, during the same period, the PM2.5 footprint of high-income APAC economies (Australia, Japan, South Korea, and Taiwan) that occurred in other APAC economies increased from ~17 to 40%. Yet, for the less wealthy countries (e.g., China, India, and Indonesia), 75−99% of PM2.5 footprints were indigenous, a considerable fraction of which are attributed to direct household emissions (Supplementary Fig. 3). The consumption of traditional fossil fuels and biomass (e.g., fuelwood and agriculture wastes) for home cooking and heating, and the open-air combustion of biomass, especially in rural areas of India and China has been considered a significant source of PM2.5 emissions43,44,45.
Among all the indicators, we find the value-added footprints of the APAC economies, i.e., the contributions of their final demand to global economic growth (domestic + intra-APAC + non-APAC) experienced the most significant increases. Such increases are especially significant for the middle- and low-income APAC economies. For example, China’s per capita value-added footprint increased by eight times during 1995−2015, followed by India (217%), Indonesia (113%), and RoAP (82%). The rates are much higher than those experienced by the high-income economies (28% on average).
The APAC economies have grown more interdependently linked during the past two decades. Depending on the footprint indicators, the foreign footprint shares traced to APAC countries (Abroad, APAC in Fig. 1) increased by 4–27% while the domestic shares decreased by 3–33%. Given that the compositional changes within the aggregated RoAP cannot be estimated, the variation of RoAP is not included in the ranges. The increased intra-APAC interdependencies are primarily attributable to the strengthened linkages among the six major APAC countries. In contrast, India’s environmental-social-economic footprints remained predominantly domestic (90–99% in 1995 and 82–97% in 2015). Previously, researchers highlighted that resource constraints have already become a bottleneck for India’s social and economic development, such as fresh water scarcity46 and energy deficiency47. Air quality deterioration caused by PM2.5 emissions has also made India under severe health burden48. Our results here confirm that India remains highly dependent on local resource use and labor-intensive production activities to sustain its socio-economic growth. Engaging in international trades has the potential of reducing the depletion of local scare resources (e.g., blue water and energy) through import while adding employment and added-value through exporting goods produced with abundant labor resources and low environmental impacts locally1,30. The trade-environment relationship is primarily rooted in the economic principle of competitive advantages among countries for international trade. Therefore, adopting the strategy of reducing trade barriers rationally, increasing the openness to the outside world to actively promote international economic cooperation may be one solution to alleviating the pressure of domestic resource depletion (i.e., water and energy) and environmental damages (i.e., carbon and air pollutions) during India’s economic growth.
Footprint outsourcing and disparity through intratrade
The linkages and imbalances among the APAC countries/regions, through the virtual flows of environmental-social-economic resources embedded in intra-APAC trade, are highlighted in Fig. 2 (2015) and Supplementary Fig. 4 (1995). The intra-APAC trade patterns observed over the past two decades confirm that developed economies (Japan, Australia, South Korea, and Taiwan) import natural resources and labor from less-developed regions where resources and labors are cheaper (China, India, and RoAP). In 2015, for the 37 billion m3 of net bilateral virtual water trade in the region, nearly 80% was associated with the exports from RoAP and India, while ~50% was driven by the final demand of Japan and South Korea, mainly embedded in a range of water-intensive agricultural crops and products. Yet, as mentioned above, India has already been threatened by serious water crises with low availabilities of safe drinking water49. For the 15 EJ net bilateral energy flow embedded in the 2015 intra-APAC trade, China was the main net supplier to the rest of the region, contributing 43%, followed by South Korea (27%), which possesses strong petrochemical and steel industries. China’s net energy outflows were predominately embedded in energy-intensive products (e.g., coal power, steel, and gasoline), 73% of which ultimately served RoAP’s final demand. Of the 488 Tg net bilateral flows of GHG emissions, 76% was supplied by China, while the final demand of RoAP contributed more than half of it (286 Tg), and the rest is attributed to the final demand of four higher-income economies (Australia, Japan, South Korea, and Taiwan, 123 Tg in total). 81% of the 677 thousand tons of net bilateral PM2.5 flows occurred in China and India. And 38% of the net flows were driven by the final demand of four higher-income economies. The social and economic effects of the intra-APAC trade are more nuanced than those related to natural resources and emissions. The Intra-APAC trade resulted in a net bilateral outpouring of 99 million people, i.e., labor resources, in 2015. RoAP was the largest net labor supplier, contributing 77%, while Australia, Japan, South Korea, and Taiwan had net inflows of labor to satisfy their final demands, equivalent to employing 11.4, 27.6, 9.3, and 5.6 million people from the rest of APAC, respectively. In terms of net flows of value added, the intraregion pattern appears significantly different and almost opposite from those of other indicators. By net outpouring value added to other APAC economies, China turned from the second trade surplus country in 1995 to the largest one within APAC in 2015, followed by the developed economies. Yet, on the per capita basis, South Korea, Australia, Japan and Taiwan achieved the most prominent economic gains through the intra-APAC trade.
Fig. 2: Net environmental-social-economic virtual flows of the intra-APAC trade in 2015.

The top five net flows are shown for each footprint indicator, and fall into three panels a natural resource, b local and global environmental threats, c socio-economic effects. The width of the arrows in each panel represents the magnitude of the net flow within the APAC region. The background colors indicate the specific net footprint (import-export) per capita of each region/country. The negative net footprint indicates net displacements (of resource use, emissions, labors, and economic value added) to other APAC countries/regions. Note, the background color and flows of Taiwan, China could not be shown on the map.

Full size image

Our results further demonstrate that the economic and environmental inequity owing to the intraregional trade worsened along time. From 1995 to 2015, the majority of the environmental externalities (i.e., more than 90% of the virtual water flows) were shifted from higher-income to lower-income countries, with a considerable worsening trend for PM2.5 (Supplementary Table 3). As for the economic gains associated with the intraregional trade, higher-income countries’ share increased from 38 to 59% from 1995 to 2015. At country level, China experienced the most significant transformation in intra-APAC trade, especially in the virtual trade of water and labor. For virtual blue water flows, China turned from the second largest net exporter of the region in 1995 to the third largest net importer in APAC (5.3 billion m3), largely due to the reverse of the import-export relationship with RoAP. Over the same period, the net virtual water export of India surged by 161%, despite the aggravating water stress concerns. China turned from the largest net supplier of employment (Supplementary Fig. 4) to the second net demander in APAC after two decades, while RoAP showed an opposite trend. For example, the labor- and resource-intensive textile and apparel production industries have been shifted from China to other less-developed countries, such as Vietnam and Bangladesh50. Such a role switching stemmed from the rising labor costs in China, which is expected to drive more low-end manufacturers to low-cost foreign economies in the coming years51,52. Overall, based on the consistently calculated virtual flows of multiple indicators, we highlight a growing environmental-socio-economic disparity within the APAC region, owing to the intraregional trade. The undesirable environmental externalities are primarily and increasingly shifted from higher to lower-income economies, while higher-income economies achieved more economic gains.
The role of APAC and intra-APAC trade in globalization
APAC is becoming an important player in the globalization process: as natural resource suppliers and manufacturers for the rest of the world, for managing global environmental emissions, and in the labor and monetary markets (Fig. 3). By 2015, APAC-related share of these categories had surpassed 50% (the red, yellow, and blue parts in Fig. 3). In contrast, the international trade without APAC countries (gray in Fig. 3) shrank for all the indicators. Moreover, for all the footprint indicators, the intra-APAC trade and non-APAC’s outsourcing to APAC (red and yellow parts in Fig. 3, respectively) account for a considerable and increasing fraction of the global trade. The former grew from 17 to 20% on average and the latter from 23 to 27% on average. APAC’s outsourcing to non-APAC countries (blue in Fig. 3) only grew slowly, with an average of 14 and 16% in 1995 and 2015, respectively. Earlier studies have proven that the world’s dominating embedded labor flows originate from developing countries, predominately to satisfy the final demand of wealthier economies23. Here, we further elucidate that non-APAC economies became more dependent on offshoring the resources, emissions and labor-intensive industries to APAC over the investigation period.
Fig. 3: Natural resources, environmental emissions, and socio-economic factors embedded in exports.

a natural resource, b local and global environmental threats, c socio-economic effects are shown in three panels. Note, the red and blue sections correspond to the same components in Fig. 1.

Full size image

Among the six indicators, we find APAC’s share in global value added was the smallest. In particular, APAC’s economic gains from exports (red + yellow) were significantly smaller (27–32%) than APAC’s resources, emissions, and labor embedded in its exports (40–48%). Intuitively, this may be due to the fact that the APAC region is dominated by population from developing countries, thus resource/labor-intensive and low value-added products dominate the region’s export. In comparison, non-APAC countries are relatively more skilled in producing products and with higher value added and lower intensities of resource and emissions4. Also, APAC’s relatively small share in the value-added dimensions of the global supply chains can be a result of the overall low resource efficiency of the region12. As resource extractions and environment emissions become increasingly outsourced to the region, where environmental regulations and efficiency measures are only emerging, more environmental impacts will likely be resulted on the global level, offsetting or even reversing the resource efficiency gains and climate change mitigation efforts achieved in developed countries.
On the positive note, from 1995 to 2015, APAC’s resource and environmental intensities declined substantially, both from the perspective of footprint, i.e., footprint per final expenditure, and from the perspective of trade, i.e., direct impacts/gross trade (see Materials and Methods). Such improvement is most noticeable for labor (Table 1). More specifically, over the past two decades, APAC improved faster than the world averages in blue water, PM2.5, and labor requirements per expenditure of final consumer, which decreased by 30, 39, and 35%, respectively, while lagging behind the global averages in reduction pace of energy and GHG intensity from footprint perspectives. The intra-APAC trade achieved a reduction in the intensities of energy, GHG and labor by 22, 42, and 34%, respectively. Relatively, this was higher than the world averages over the period. However, the PM2.5 embedded in intra-APAC trade per traded values even grew by 7%, whereas the same indicator decreased by 34% at the global scale. This can be explained primarily by the PM2.5 trade magnitude in APAC nearly tripling in 2015, much higher than that of the global average (0.8 time). This indicates that the PM2.5 issue has become graver for the APAC trade after two decades. The ratio of traded value added to total intratrade values measures the economic gains of trade. Value added is a general indicator of economic performance, yet, it could not evaluate all the advantages of trade, which may cause misleading results in some cases. Therefore, more comprehensive indicators need to be incorporated to assess its advantages.
Table 1 Intensity comparison between APAC and the global average from the footprint and trade perspectives.
Full size table

Despite APAC’s improvement, APAC’s resource and environmental intensities remain higher in general than the world averages in 2015. The comparison of intensity changes at the national and regional levels was also shown in Supplementary Fig. 5. This can be attributed to the characteristics of the APAC region and may not be altered soon. Specifically, the socio-economic development is supported by resource- and emission-intensive productions in primary industries, especially those linked with capital development and those to satisfy the export demand. The APAC region has a long way to go to achieve a greener and more sustainable trade pathway.
Our result further reveals that, China played a crucial role in APAC’s transformation to greener trade and greener consumption. Generally, without China’s improvement, all the footprint intensities of APAC would exhibit a further increase from 1995–2015, at 8% on average. By contrast, the value is −20% with China in the picture. For the trade intensities, the decline ranges would be much smaller (−1% on average) without China as opposed to with China (−24% on average). Another noteworthy finding is that when eliminating China from APAC, the intensities of footprint and trade in 1995 would be 20–30% lower, implying that China had turned from the lagger to the leader of efficiency improvement in the APAC region. More

Study area
This study was conducted at the Sierra Morena mountain range in Southwestern Spain (38° 22′ 50.64″ N, 4° 45′ 27.69″ W; Pozoblanco, Córdoba). This area has a continental-Mediterranean climate characterised by cold winters and hot, dry summers. Annual precipitation is 439 mm year−1 and mean annual temperature is 15.2 °C. The studied ecosystem is a “Mediterranean dehesa”, a savannah-type forest managed as a traditional sylvo-pastoral system, with an herbaceous stratum of native pasture and a tree layer of scattered oak trees (Quercus ilex L.). Tree density is 14.5 ± 1.3 trees ha−1, and the herbaceous layer is dominated by a diverse community of annual native species such as: Hordeum murinum (L.), Senecio vulgaris (L.), Bromus madritensis (L.), and Sinapis alba (L.), with a mean species richness of 9.6 ± 0.3 species m−2.
Experimental design
In September 2016, we selected three sites separated by 4.3 ± 1.9 km with similar tree density, topography, slope, orientation, and soil type. At each site, we sampled two areas: a partially shaded site under Q. ilex canopy (under tree canopy) and a nearby open grassland site (open grassland). These two habitat types are characteristic of the dehesa ecosystem in the region. Then, within each area (i.e. habitat type) at each site we established six permanent 4 × 6 m (1.2 m high) fenced plots (N = 36 plots). Each plot was divided into two subplots (N = 72 subplots), one subplot was subjected to a rainfall reduction treatment whereas the other was exposed to ambient rainfall. Rainfall was reduced by placing a 2.5 × 2.5 m rain-exclusion shelter over one of the subplots (Fig. 2), resulting in a 30% reduction in annual rainfall22 as predicted by the IPCC for Mediterranean regions23. In addition, within each subplot we sampled two ca. 1-m2 quadrats (144 quadrats in total, i.e. 2 quadrats × 72 subplots): one was subjected to a warming treatment and the other was not manipulated (i.e. ambient temperature). Warming was achieved by placing a 40 × 50 × 32 cm (ca. 0.65 m2) hexagonal open top chamber of methacrylate material without UV filter (Faberplast, Madrid) at the centre of the quadrat (Fig. 2), increasing air temperature by ca. 2 °C compared to ambient temperature (17.5 ± 0.1 vs. 15.7 ± 0.1 °C, respectively). This matches the temperature increase for the study region by climatic forecasting models (SRES A-2 model by the Intergovernmental Panel on Climate Change23). Ambient air temperature was slightly lower (approx. 0.5 °C) in the shaded habitat than in the open habitat, but the experimental temperature increase was of similar magnitude in both cases. Temperature in each quadrat was measured with data loggers. Based on the above set-up, the experiment followed a randomized split–split plot design replicated across three sites, with habitat type as the whole factor (plots as replicates), rainfall manipulation as the split factor (subplots), and temperature manipulation as the split–split factor (quadrats).
Figure 2

Experimental design. Picture showing open top chambers and rain-exclusion shelters used for our climate manipulations. Photo credit: Ignacio M. Pérez-Ramos.

Full size image

Sampling, measurements and chemical analyses
In April 2017 and April 2018, i.e. 1 and 2 years after establishment of climatic treatments, we identified all plant species and estimated their frequency within each quadrat. Then, in April 2018, after the second survey of species frequencies, we collected 3–4 fully expanded leaves from each plant for chemical analyses which were pooled to obtain a single sample per species and quadrat. Plant sampling was restricted to a central portion of 0.65 m2 in each quadrat (within chamber in the case of warming quadrats) as a function of the surface covered by and sampled within the hexagonal chamber placed in quadrats subjected to warming (see above). Due to lack of plant material, we only sampled 127 out of the 144 quadrats. In total, we collected 436 samples from 28 annual plant species (Table S1). Immediately after collection, we oven-dried leaves for 48 h at 40 °C, ground them with liquid nitrogen, and stored them at room temperature before conducting chemical analyses.
We determined total phenolic content colorimetrically by the Folin-Ciocalteu method24,25. Briefly, we extracted phenolics from 20 mg of plant tissue with 70% methanol in an ultrasonic bath for 15 min, followed by centrifugation. We determined total phenolic content colorimetrically by the Folin–Ciocalteu method in a Biorad 650 microplate reader (Bio-Rad Laboratories, PA, USA) at 740 nm, using tannic acid as standard. We performed three technical replicates of each sample in order to estimate variations due to the experimental procedure, and expressed concentrations based on dry weights (d.w.). We decided to use total phenolics as these could be measured across all species in order to address the goal of testing for community-level effects of climatic stressors on plant secondary chemistry.
Statistical analyses
For each quadrat, we calculated the CWMphenolics using the weighted.mean function in R software26. For this, we used phenolic values at the species level weighed by the abundance (i.e. frequency) of each species. We then performed linear mixed models using PROC MIXED in SAS 9.4 (SAS Institute, Cary, NC)27 to test for the effects of habitat type, rainfall manipulation, temperature manipulation, and their two-way and three-way interactions (all fixed factors) on CWMphenolics. We included site, the site × habitat type interaction, and the site × habitat type × rainfall manipulation interaction as random factors. The two- and three-way interaction interactions tested the main effect of habitat type and rainfall manipulation (respectively) with the appropriate error terms according to a split-split plot design. We log-transformed original values to achieve normality of residuals.
We also assessed whether climatic manipulations influenced the community-level expression of phenolic compounds through changes in plant composition with a PERMANOVA (‘vegan’ package in R) that included the effects of climatic treatment, year, and their interaction. 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