Philippa Kaur

Study populationThe study is based on data from mothers and children participating in a longitudinal Southwest Finland cohort, Steps to Healthy development of Children (the STEPS Study) (described in detail in Lagström et al.31). The STEPS study is an ongoing population-based and multidisciplinary study that investigates children’s physical, psychological and social development, starting from pregnancy and continuing until adolescence. All Finnish- and Swedish-speaking mothers delivering a child between 1 January, 2008 and 31 March, 2010 in the Hospital District of Southwest Finland formed the cohort population (in total, 9811 mothers and their 9936 children). Altogether, 1797 mothers with 1805 neonates volunteered as participants for the intensive follow-up group of the STEPS Study. Mothers were recruited by midwives either during the first trimester of pregnancy while visiting maternity health care clinics, or after delivery on the maternity wards of Turku University Hospital or Salo Regional Hospital, or by a letter mailed to the mothers. The participating mothers differ slightly from the whole cohort population in some background characteristics (being older, with first-born child and higher socioeconomic status)31. The ethics committee of the Hospital District of Southwest Finland has approved the STEPS Study (2/2007) and all methods were performed in accordance with relevant guidelines and regulations. Written informed consent was obtained from all the participants and, for children, from one parent for study participation. Subjects have been and are free to withdraw from the study at any time without any specific reason. The STEPS Study have the appropriate government authorization to the use of the National birth register (THL/974/5.05.00/2017).Breastmilk collection and HMO analysisMothers from the STEPS Study were asked to collect breastmilk samples when the infant was approximately 3 months old. In total, 812 of the 1797 mothers (45%) provided a breastmilk sample. There were only slight differences in maternal and child characteristics between the participants providing breastmilk samples and the total STEPS Study cohort40. Altogether, 795 breastmilk samples were included in this study (excluding the duplicate observations and the 2nd born twins, samples with technical unclarity or insufficient sample quantity, one breastmilk sample collected notably later than the other samples, at infant age of 14.5 months (range for the other breastmilk samples: 0.6–6.07 months), one sample with missing information on the date of collection and six mothers missing data on residential green environment) (Supplementary Fig. 2). Mothers received written instructions for the collection of breastmilk samples: samples were collected by manual expression in the morning from one single breast, first milking a few drops to waste before collecting the actual sample (~ 10 ml) into a plastic container (pre-feed sample). The samples were stored in the fridge and the mothers brought the samples to the research center or the samples were collected from their homes on the day of sampling. All samples were frozen and stored at − 70 °C until analysis.High Performance Liquid Chromatography (HPLC) was used to identify HMOs in breastmilk as previously described40,57,58 at the University of California, San Diego (methods described in detail in Berger et al.58). Milk samples were spiked with raffinose (a non-HMO carbohydrate) as an internal standard to allow absolute quantification. HMOs were extracted by high-throughput solid-phase extraction, fluorescently labelled, and measured using HPLC with fluorescent detection (HPLC-FLD)58. Absolute concentrations for the 19 HMOs were calculated based on standard retention times and corrected for internal standard recovery. Quantified HMOs included: 2′-fucosyllactose (2′FL), 3-fucosyllactose (3FL), lacto-N-neotetraose (LNnT), 3′-sialyllactose (3′SL), difucosyllactose (DFlac), 6′-sialyllactose (6′SL), lacto-N-tetraose (LNT), lacto-Nfucopentaose (LNFP) I, LNFP II, LNFP III, sialyl-LNT (LST) b, LSTc, difucosyllacto-LNT (DFLNT), lacto-N-hexaose (LNH), disialyllacto-N-tetraose (DSLNT), fucosyllacto-Nhexaose (FLNH), difucosyllacto-N-hexaose (DFLNH), fucodisialyllacto-lacto-N-hexaose (FDSLNH) and disialyllacto-N-hexaose (DSLNH). HMOs were also summed up to seven groups based on structural features: small HMOs (2′FL, 3FL, 3′SL, 6′SL, and DFLac), type 1 HMOs (LNT, LNFP I, LNFP II, LSTb, DSLNT), type 2 HMOs (LNnT, LNFP III, LSTc), α-1-2-fucosylated HMOs (2’FL, LNFP I), terminal α-2-6-sialylated HMOs (6′SL, LSTc), internal α-2-6-sialylated HMOs (DSLNT, LSTb), terminal α-2-3-sialylated HMOs (3′SL, DSLNT). The total concentration of HMOs was calculated as the sum of the 19 oligosaccharides. HMO-bound fucose and HMO-bound sialic acid were calculated on a molar basis. The proportion of each HMO comprising the total HMO concentration was also calculated. HMO Simpson’s diversity was calculated as Simpson’s Reciprocal Index 1/D, which is the reciprocal sum of the square of the relative abundance of each of the measured 19 HMOs57,59. The higher the diversity value, the more heterogenous is the HMO composition in the sample.Properties of the residential green environmentThe selected residential green environment variables measure the properties of the green environments surrounding the homes of the participants and do not include any measures of the house characteristics, indoor environment or the actual use of green spaces by the participants. The residential green environment variables were selected due to their previously observed associations with residential microbiota and health33,34,35. The variables of the residential green environments were derived from multispectral satellite images series, with a 30 m × 30 m of spatial resolution (Landsat TM 5, National Aeronautics and Space Administration—NASA) and land cover data (CORINE). We used Landsat TM images obtained over the summertime (June–August, greenest months in Finland), to minimize the seasonal variation of living vegetation and cloud cover as well as to better identify vegetation areas and maximise the contrast in our estimated exposure. In each selected Landsat TM 5 images, the cloud was masked out, and the Normalized Difference Vegetation Index (NDVI)36 was calculated. The final NDVI map was the mean of NDVI images collected over three consecutive years (2008–2010), to make an NDVI map with non-missing values due to cloud cover for the study area. NDVI map measures the vegetation cover, vitality and density. The NDVI can get values ranging from − 1 to 1 where values below zero represent water surfaces, values close to zero indicate areas with low intensity of living vegetation and values close to one indicate high abundance of living vegetation. For the analyses, areas covered by water were removed and the value ranged from 0 to 1, to prevent negative values for underestimating the greenness values of the residential area like in some prior studies60. We assumed that summertime NDVI identified the green space and vegetation density well, but greenness intensity might vary seasonally.Second, we used calculated indicators related to the diversity and naturalness of the land cover from CORINE Land Cover data sets of the year 201261. The 12 land cover types include: (1) Residential area, (2) Industrial/commercial area, (3) Transport network, (4) Sport/leisure, (5) Agriculture, (6) Broad-leave forest, (7) Coniferous forest, (8) Mixed forest, (9) Shrub/grassland, (10) Bare surface, (11) Wetland, and (12) Water bodies. From this information, we calculated two vegetation cover indexes. The Vegetation Cover Diversity Index (Simpson’s Diversity Index of Vegetation Cover, VCDI)37, only includes vegetation classes from CORINE land cover types (categories 5–9 and 11). VCDI approaches 1 as the number of different vegetation classes increases and the proportional distribution of area among the land use classes becomes more equitable. Furthermore, because we were particularly interested in the natural vegetation cover in the residential area, we calculated the area-weighted Naturalness Index (NI)38. This is an integrated indicator used to measure the human impact and degree of all human interventions on ecological components. The index is based on CORINE Land Cover data but reclassified to 15 classes. Residential areas have been divided to two classes: Continuous residential area (High density buildings) and Discontinuous residential area (Low density, mostly individual houses area). Agricultural area has also been divided to two classes: Agricultural area (Cropland) and Pasture as well as class 9 (Shrub/grassland) has been separated to Woodland and Natural grassland. Assignment of CORINE Land Cover classes to degrees of naturalness has been made based on Walz and Stein 201438. The area-weighted NI range from 1 to 7, where low values represent low human impact (≤ 3 = Natural), medium values moderate human impact (4–5 = Semi-natural) and high values strong human impact (6–7 = Non-Natural). To ease the interpretation of results and to correspond to the same direction than the other residential green environment variables, we have reverse-scaled the NI values, so that higher values illustrate more natural residential areas.Background factorsAs genetics is strongly linked to HMO composition, maternal secretor status was determined by high abundance (secretor) or near absence (non-secretor) of the HMO 2’FL in the breastmilk samples. Mothers with active secretor (Se) genes and FUT2 enzyme produce high amounts of α-1-2-fucosylated HMOs such as 2′-fucosyllactose (2′FL), whereas in the breastmilk of non-secretor mothers these HMOs are almost absent. Beyond genetics, other maternal and infant characteristics may influence HMO composition. So far, several associations have been reported, including lactation stage, maternal pre-pregnancy BMI, maternal age, parity, maternal diet, mode of delivery, infant gestational age and infant sex22,40. Information on the potential confounding factors, child sex, birth weight, maternal age at birth, number of previous births, marital status, maternal occupational status, smoking during pregnancy (before and during pregnancy), maternal pre-pregnancy BMI, mode of delivery, duration of pregnancy and maternal diseases [including both maternal disorders predominantly related to pregnancy such as pre-eclampsia and gestational diabetes and chronic diseases (diseases of the nervous, circulatory, respiratory, digestive, musculoskeletal and genitourinary systems, cancer and mental and behavioral disorders, according to ICD-10 codes, i.e. WHO International Classification of Diseases Tenth Revision)], were obtained from Medical Birth Registers. Self-administered questionnaires upon recruitment provided information on family net income and maternal education level. Those who had no professional training or a maximum of an intermediate level of vocational training (secondary education) were classified as “basic”. Those who had studied at a University of Applied Sciences or higher (tertiary education) were classified as “advanced”. The advanced level included any academic degree (bachelor’s, master’s, licentiate or doctoral degree). Maternal diet quality was assessed in late pregnancy using the Index of Diet Quality (IDQ62) which measures adherence to health promoting diet and nutrition recommendations. The IDQ score was used in its original form by setting the statistically defined cut-off value at 10, with scores below 10 points indicating unhealthy diets and non-adherence and scores of 10–15 points indicating a health-promoting diet and adherence dietary guidelines. Lactation time postpartum (child age) and season were received from the recorded breastmilk collection dates. Lactation status (exclusive/partial/unknown breastfeeding) at the time of breastmilk collection were gathered from follow-up diaries. From partially breastfeeding mothers (n = 277) 253 had started formula feeding and 28 solids at the time of milk collection. Last, a summary z score representing socio-economic disadvantage in the residential area was obtained from Statistics Finland grid database for the year 2009 and is based on the proportion of adults with low level of education, the unemployment rate, and proportion of people living in rented housing at each participant’s residential area55.Statistical analysesTo harmonize the residential green environment variables we calculated the mean values for NDVI, VCDI and NI in 750 × 750 m squares (and 250 × 250 m) around participant homes in a Geographical Information System (QGIS, www.qgis.org). The same grid sizes were used to calculate residential socioeconomic disadvantage in the residential area55 at the time of child birth. The geographical coordinates (latitude/longitude) of the cohort participants’ home address (795 mothers) were obtained from the Population Register Centre at the time of their child birth.The background characteristics of the mothers and children are given as means and standard deviations (SD) for continuous variables and percentages for categorical variables. Due to non-normal distribution, natural logarithmic transformation was performed for all HMO variables (19 individual components, sum of HMOs, HMO-bound sialic acid, HMO-bound fucose and HMO groups (all in nmol/mL)) except for HMO diversity. Associations between each background factor and HMO diversity and 19 individual HMO components were analysed with univariate generalized linear models to identify factors independently associated with HMO composition. All factors demonstrating a significant association (p  More

Systematic paleontologySauropterygia Owen, 186038.Eosauropterygia Rieppel, 199439.Pachypleurosauroidea Huene, 195640.Pachypleurosauridae Nopcsa, 192841.Luopingosaurus imparilis gen. et sp. nov.EtymologyThe genus name refers to the Luoping County, at which the fossil site is located. Species epithet imparilis (Latin) means peculiar and unusual.HolotypeA ventrally exposed skeleton with a posterior part of the caudal missing, IVPP V19049.Locality and horizonLuoping, Yunnan, China; Second (Upper) Member of Guanling Formation, Pelsonian (~ 244 Ma), Anisian, Middle Triassic37.DiagnosisA pachypleurosaurid distinguishable from other members of this family by the following combination of features (those unique among pachypleurosaurids identified with an asterisk): snout (preorbital portion) long and anteriorly pointed, 55.0% of skull length (*); orbital length about one quarter of skull length; internal naris retracted, without contribution from premaxilla; nasal ending at level of anterior margin of prefrontal; dentary length 71.7% of mandibular length; hyoid length 9.7% of mandibular length; presence of entepicondylar foramen in humerus; 21 cervical and 27 dorsal vertebrae (*); distinct expansions of distal heads of posterior two sacral ribs; six pairs of caudal ribs; phalangeal formula 2–3-5–5-3 for manus and 2–3-4–6-4 for pes (*); Metatarsal I short and stout with expanded proximal end, 56.4% of Metatarsal V in length (*); and Metatarsal IV being longest phalange in pes.Comparative descriptionThe holotype and only currently known specimen of Luopingosaurus has a preserved length of 46.2 cm from the rostral tip to the 30th caudal vertebra (for measurements, see Table 1). The estimated total length of the body may have reached 64 cm, assuming similar tail proportions of pachypleurosaurids. As such, Luopingosaurus is longer than most of other pachypleurosauroids that are small-sized with a maximum total length rarely exceeding 50 cm4,9,10,11,12,14,15,16,18,23,25, although some pachypleurosaurids are notably larger (e.g., 88 cm in Diandongosaurus cf. acutidentatus22, ~ 120 cm in Neusticosaurus edwardsii8, and ~ 130 cm in Wumengosaurus delicatomandibularis13).Table 1 Measurements (in mm) of the holotype (IVPP V19049) of Luopingosaurus imparilis gen. et sp. nov. R, right.Full size tableThe pre-orbital portion, distinctly longer than the postorbital region, measures 55% of the total skull length (the premaxillary symphysis to the occipital condyle) and 51% of the mandibular length. The paired premaxillae form most of the snout anterior to the naris with a pointed anterior tip, similar to the conditions in Wumengosaurus13,30 and Honghesaurus23. By contrast, other pachypleurosauroids uniformly have a blunt rostrum4,6,7,8,9,10,11,12,14,15,16,18,22,25. The premaxilla bears a long posteromedial process inserting between the anterior parts of the elongate nasals (Fig. 3). The premaxillary teeth are homodont with a tall peduncle and a short, conical crown, but the tooth number is hard to estimate because of occlusion of jaws. The posterior parts of nasals contact each other medially, and posteriorly, they contact the frontals in an interdigitating suture at the level of the anterior margin of the prefrontal. In Honghesaurus23, Wumengosaurus30, Neusticosaurus8 and Serpianosaurus9, the even longer nasal extends posteriorly beyond this level and ends at the anterior portion of the orbit.Figure 3Skull and mandible of Luopingosaurus imparilis gen. et sp. nov., IVPP V19049. Head before (a) and after (b) dusted with ammonium chloride. (c), Line- drawing. (d, e), two computed laminography scanning slices. (f), reconstruction in ventral view. ac, acetabulum; an, angular; ar, articular; ax, axis; c, cervical vertebra; den, dentary; eo, exoccipital; f, frontal; hy, hyoid; in, internal naris; j, jugal; m, maxilla; n, nasal; p, parietal; par, prearticular; pof, postfrontal; prf, prefrontal; pt, pterygoid; q, quadrate; qj, quadratojugal; sa, surangular; sp, splenial; sq, squamosal; stf, supratemporal fossa; v, vomer.Full size imageThe orbit is oval and large, measuring 24.8% of the skull length (Fig. 3). The lateral margin of the frontal contacts the prefrontal anteriorly and the postfrontal posteriorly, and defines most of the medial border of the orbit. The L-shaped jugal, together with the posterolateral process of the maxilla, forms the lateral border of the orbit. No distinct lacrimal is discernable; the bone is probably absent as in other sauropterygians. The postfrontal contacts the dorsal process of the triradiate postorbital ventrally, and both bones define the posterior border of the orbit. Additionally, the posterior process of the postorbital contacts the anterior process of the squamosal, forming the bar between the supratemporal fossa and the ventrally open infratemporal fenestra. The jugal extends beyond the ventral margin of the postorbital and also contacts the anterior process of the squamosal, resembling the conditions in Wumengosaurus30, Honghesaurus23 and Diandongosaurus15. This contact is absent in other pachypleurosauroids4,6,7,8,9,10,11,12.A pair of vomers and pterygoids and a right palatine are discernable in the palate (Fig. 3a–c). The vomer is elongate and slender, extending anteriorly well beyond the nasal. The internal naris, partly covered by the detached splenial, is longitudinally retracted, corresponding to a retracted external naris (Fig. 3d–f). The medial margin of the naris is defined by the nasal, without contribution from the premaxilla. A retracted naris is otherwise present in Wumengosaurus13,30, Qianxisaurus16 and Honghesaurus23. Similar to the condition in Honghesaurus23, the retracted naris of Luopingosaurus is relatively short, having a longitudinal diameter distinctly less than half of the longitudinal diameter of the orbit. By contrast, other pachypleurosauroids4,6,7,8,9,10,11,12,25 generally have an oval-shaped naris. The elongate palatine has a slightly convex medial margin suturing with the pterygoid. Because of the coverage of the detached splenial, the lateral portion of the palatine is unexposed, and it is hard to know whether an ectopterygoid is present or not. The pterygoid is the largest and longest element of the palate, measuring 55.2% of the mandibular length. It has an anterior projection that contacts the vomer anteromedially, and does not participate in the margin of the internal naris. At the level of the posterior orbital margin, the pterygoid has a triangular lateral extension, which was termed as the ectopterygoid process of the pterygoid in Neusticosaurus8. The pterygoid extends back to the occipital condyle, and covers the basicranium and parietals in ventral view. Additionally, the bone has a broad posterolateral process that is set off from the palatal surface by a distinct ridge, resembling the conditions in Serpianosaurus9 and Neusticosaurus8. Posteriorly, the basioccipital is exposed in ventral view, showing the area for attachment to the right exoccipital.The left quadrate is exposed in lateral view with its dorsal process extending underneath the base of the descending process of the squamosal. The posterior margin of the quadrate is excavated, as in many other pachypleurosaurids (e.g., Serpianosaurus9 and Honghesaurus23). The quadratojugal is narrow and splint-like, flanking the anterior margin of the quadrate. A pair of hyoids are ossified. They are rod-like, slightly expanded at both ends. The dentary is wedge-shaped, being 71.7% of the mandibular length. Laterally, it bears a longitudinal series of pores and grooves parallel to the oral margin of the bone (Fig. 3a). The elongate angular tapers at both ends, contacting the dentary anterodorsally and the surangular dorsally in ventral view. The surangular, slightly shorter than the angular, contacts the articular posterodorsally, with a pointed anterior tip wedging into the notched posterior margin of the dentary. The retroarticular process of the articular is very short with a rounded posterior margin. Medially, the splenial and prearticular form most of the inner wall of the mandible. The splenial tapers at both ends and enters the mandibular symphysis anteriorly, having a length similar to the dentary. The relatively slender prearticular contacts the splenial anterodorsally, extends posteriorly and abuts the articular dorsally, measuring 41.1% of the mandibular length.The whole series of 21 cervical vertebrae (including the atlas-axis complex) is well exposed ventrally. The atlas centrum is oval, much smaller than the axis centrum (Fig. 3c). From the axis, the cervical vertebrae increase gradually in size toward the trunk vertebrae posteriorly. The bicephalous cervical ribs have typical free anterior and posterior processes as in other pachypleurosauroids8,9. The trunk is relatively long, including 27 dorsal vertebrae. The posterior dorsal ribs show certain pachyostosis (Fig. S1). Each gastralium consists of five elements (a short and more massive median element and two slender rods in line towards each side; Figs. 3, 4a, b, S1), similar to the conditions in most of other pachypleurosauroids9,11,18,25 (except Neusticosaurus8). Three sacral ribs are clearly revealed by X-ray computed microtomography (Fig. 4c–f). They are relatively short and stout, with the posterior twos bearing a distinct expansion on their distal heads. The distal expansion of the sacral rib is also present in Keichousaurus11, Prosantosaurus25, Qianxisaurus16 and Wumengosaurus13, but it is not pronounced in other pachypleurosauroids4,6,7,8,9,10. The caudal ribs are relatively few, six pairs in number. Additionally, several chevron bones are visible in the proximal caudal region, and they are gradually reduced in length posteriorly (Fig. 4d).Figure 4Girdles, limbs and vertebrae of Luopingosaurus imparilis gen. et sp. nov., IVPP V19049. Photo (a) and line-drawing (b) of pectoral girdle, forelimbs and anterior dorsal vertebrae. Photo (c), line-drawing (d) and two computed laminography scanning slices (e, f) of pelvic girdle, hind limbs and posterior vertebrae. as, astragalus; ca, caudal vertebra; cal, calcaneum; car, caudal rib; co, coracoid; d, dorsal vertebra; dltp, deltopectoral crest; enf, entepicondylar foramen; fe, femur; fi, fibula; h, humerus; il, ilium; int, intermedium; is, ischium; mc, metacarpal; mt, metatarsal; pu, pubis; s, sacral vertebra; sc, scapula; sr, sacral rib; ti, tibia; ul, ulna; uln, ulnare.Full size imageThe paired clavicles and the median interclavicle form a transverse bar at the 20th cervical vertebrae (Fig. 4a, b). The blade-like clavicle tapers posterolaterally with its distal projection overlapped by the scapula in ventral view. The left clavicle contacts the right one anterodorsally to the interclavicle. The interclavicle tapers laterally to a point at each end. The anterior margin of the interclavicle is convex and its posterior margin is slightly concave without a midline projection (contra the condition in Anarosaurus42). The scapula consists of a broad ventral portion and a relatively narrow and elongate dorsal wing that varies little through its length. The coracoid is hourglass-shaped with a slightly concave posterolateral margin and a conspicuously concave anteromedial margin. The medial margin is straight, along which the coracoids would articulate each other in the midline. The humerus is constricted at the middle portion with a nearly straight preaxial margin and a concave postaxial margin. A slit in the expanded distal portion of this bone indicates the possible presence of an entepicondylar foramen (Fig. 4a, b). The radius, slightly longer than the ulna, is more expanded proximally than distally. The ulna is straight with a slightly constricted shaft. In each forelimb, there is two nearly rounded carpals, ulnare and intermedium; the former is half the width of the latter. Five metacarpals are rod-like, slightly expanded at both ends. Among them, Metacarpal I is the shortest, 48% of the length of Metacarpal II. Metacarpal III is slightly shorter than Metacarpal IV, which is the longest. Metacarpal V is 71% of the length of Metacarpal IV. The phalangeal formula is 2–3–5–5–3 for the manus, indicating presence of hyperphalangy in Luopingosaurus (see Discussion below).In the pelvic girdle, the ilia, pubes and ischia are well exposed (Fig. 4c–f). The ilium is nearly triangular with a relatively long and tapering posterior process. The plate-like pubis is well constricted at its middle portion, with the medial portion nearly equal to the lateral portion. The obturator foramen is slit-like, located at the posterolateral corner of this bone (Fig. 4e). The ischium is also plate-like, having a relatively narrow lateral portion and an expanded medial portion that is notably longer than the medial portion of the pubis. The posterolateral ischial margin is highly concave. The posterior pubic margin and anterior ischial margin are moderately concave, and both together would enclose the thyroid fenestra. The femur is slightly longer than the humerus, with a constricted shaft and equally expanded ends (Fig. 4d). No internal trochanter is developed. The tibia is nearly equal to the fibula in length; the former is straight and thicker than the slightly curved latter. Two ossified tarsals, calcaneum and astragalus, are nearly rounded; the latter is significantly larger than the former. As in Honghesaurus23, the astragalus lacks a proximal concavity. The right metatarsals are well-preserved. Metatarsal I is the shortest and stoutest phalange with an expanded proximal end, and Metatarsal IV is the longest. Metatarsal II is nearly twice the length of Metatarsal I. Metatarsal III is slightly shorter than Metatarsal IV, and Metatarsal V is 76% of the length of Metatarsal IV. The phalangeal count is 2–3–4–6–4, which is complete judging from the appearance of the distal phalanges in the right pes (Fig. 4c). More

Refsnider, J. M. & Janzen, F. J. Putting eggs in one basket: Ecological and evolutionary hypotheses for variation in oviposition-site choice. Annu. Rev. Ecol. Evol. Syst. 41, 39–57 (2010).Article 

Google Scholar 
Rudolf, V. H. W. & Rodel, M. O. Oviposition site selection in a complex and variable environment: The role of habitat quality and conspecific cues. Oecologia 142, 316–325 (2005).Article 
ADS 

Google Scholar 
Blaustein, L. Oviposition site selection in response to risk of predation: Evidence from aquatic habitats and consequences for population dynamics and community structure. In Evolutionary Theory and Processes: Modern Perspectives (ed. Wasser, S. P.) 441–456 (Springer, 1999).Chapter 

Google Scholar 
Elsensohn, J. E., Schal, C. & Burrack, H. J. Plasticity in oviposition site selection behavior in drosophila suzukii (diptera: drosophilidae) in relation to adult density and host distribution and quality. J. Econ. Entomol. 114, 1517–1522 (2021).Article 

Google Scholar 
Kempraj, V., Park, S. J. & Taylor, P. W. Forewarned is forearmed: Queensland fruit flies detect olfactory cues from predators and respond with predator-specific behaviour. Sci. Rep. 10, 7297 (2020).Article 
ADS 

Google Scholar 
Damodaram, K. J. P. et al. Centuries of domestication has not impaired oviposition site-selection function in the silkmoth, Bombyx mori. Sci. Rep. 4, 1–6 (2014).
Google Scholar 
Hansson, B. S. & Stensmyr, M. C. Evolution of insect olfaction. Neuron 72, 698–711 (2011).Article 

Google Scholar 
Ghosh, E., Sasidharan, A., Ode, P. J. & Venkatesan, R. Oviposition preference and performance of a specialist herbivore is modulated by natural enemies, larval odors, and immune status. J. Chem. Ecol. 48, 670–682 (2022).Article 

Google Scholar 
Nielsen, R. A. & Brister, C. D. The greater wax moth: Adult behavior. Ann. Entomol. Soc. Am. 70, 101–103 (1977).Article 

Google Scholar 
Kwadha, C. A., Ong’Amo, G. O., Ndegwa, P. N., Raina, S. K. & Fombong, A. T. The biology and control of the greater wax moth, Galleria mellonella. Insects 8, 61 (2017).Article 

Google Scholar 
Kebede, E. Prevalence of wax moth in modern hive with colonies in Kafta Humera. Anim. Vet. Sci. 3, 132–135 (2015).Article 

Google Scholar 
Ellis, J. D., Graham, J. R. & Mortensen, A. Standard methods for wax moth research. J. Apic. Res. 52, 1–17 (2013).Article 

Google Scholar 
Hepburn, H. R. & Radloff, S. E. Honeybees of Africa 227–241 (Springer, 1998). https://doi.org/10.1007/978-3-662-03604-4.Book 

Google Scholar 
Fletcher, D. J. C. The African Bee, Apis mellifera adansonii, Africa. Annu. Rev. Entomol. 23, 151–171 (1978).Article 

Google Scholar 
Li, Y. et al. Losing the arms race: Greater wax moths sense but ignore bee alarm pheromones. Insects 10, 81 (2019).Article 
ADS 

Google Scholar 
Feng, B., Qian, K. & Du, Y. J. Floral volatiles from Vigna unguiculata are olfactory and gustatory stimulants for oviposition by the bean pod borer moth Maruca vitrata. Insects 8, 60 (2017).Article 

Google Scholar 
Janz, N. Evolutionary ecology of oviposition strategies. In Chemoecology of Insect Eggs and Egg Deposition (eds Hilker, M. & Meiners, T.) 349–376 (Willey, 2008). https://doi.org/10.1002/9780470760253.ch13.Chapter 

Google Scholar 
Renwick, J. A. A. & Chew, F. S. Oviposition behavior in lepidoptera. Annu. Rev. Entomol. 39, 377–400 (1994).Article 

Google Scholar 
Nakajima, Y. & Fujisaki, K. Fitness trade-offs associated with oviposition strategy in the winter cherry bug, Acanthocoris sordidus. Entomol. Exp. Appl. 137, 280–289 (2010).Article 

Google Scholar 
Murphy, P. J. Context-dependent reproductive site choice in a Neotropical frog. Behav. Ecol. 14, 626–633 (2003).Article 

Google Scholar 
Geoffrey, G. et al. Larviposition site selection mediated by volatile semiochemicals in Glossina palpalis gambiensis. Ecol. Entomol. 46, 301–309 (2021).Article 

Google Scholar 
Yao, F. L. et al. Oviposition preference and adult performance of the whitefly predator Serangium japonicum (Coleoptera: Coccinellidae): Effect of leaf microstructure associated with ladybeetle attachment ability. Pest Manag. Sci. 77, 113–125 (2021).Article 

Google Scholar 
Spieler, M. & Linsenmair, K. E. Choice of optimal oviposition sites by Hoplobatrachus occipitalis (Anura: Ranidae) in an unpredictable and patchy environment. Oecologia 109, 184–199 (1997).Article 
ADS 

Google Scholar 
Figiel, C. R. & Semlitsch, R. D. Experimental determination of oviposition site selection in the marbled salamander, Ambystoma opacum. J. Herpetol. 29, 452 (1995).Article 

Google Scholar 
Kotler, B. P. & Mitchell, W. A. The effect of costly information in diet choice. Evol. Ecol. 9, 18–29 (1995).Article 

Google Scholar 
Nylin, S. & Janz, N. Oviposition preference and larval performance in Polygonia c-album (Lepidoptera: Nymphalidae): the choice between bad and worse. Ecol. Entomol. 18, 394–398 (1993).Article 

Google Scholar 
Nagaya, H., Stewart, F. J. & Kinoshita, M. Swallowtail butterflies use multiple visual cues to select oviposition sites. Insects 12, 1047 (2021).Article 

Google Scholar 
Scolari, F., Valerio, F., Benelli, G., Papadopoulos, N. T. & Vaníčková, L. Tephritid fruit fly semiochemicals: Current knowledge and future perspectives. Insects 12, 408 (2021).Article 

Google Scholar 
Haverkamp, A., Hansson, B. S. & Knaden, M. Combinatorial codes and labelled lines: How insects use olfactory cues to find and judge food, mates, and oviposition sites in complex environments. Front. Physiol. 9, 49 (2018).Article 

Google Scholar 
Ichinosé, T., Honda, H. & Honda, H. Ovipositional behavior of papilio protenor demetrius Cramer and the factors involved in its host plants. Appl. Entomol. Zool. 13, 103–114 (1978).Article 

Google Scholar 
Spangler, H. G. Functional and temporal analysis of sound production in Galleria mellonella L. (Lepidoptera: Pyralidae). J. Comp. Physiol. A 159, 751–756 (1986).Article 

Google Scholar 
Spangler, H. G. & Takessian, A. Sound perception by two species of wax moths (Lepidoptera: Pyralidae). Ann. Entomol. Soc. Am. 76, 94–97 (1983).Article 

Google Scholar 
Skals, N. & Surlykke, A. Hearing and evasive behaviour in the greater wax moth, Galleria mellonella (Pyralidae). Physiol. Entomol. 25, 354–362 (2008).Article 

Google Scholar 
Kwadha, C. A. Determination of Attractant Semio-Chemicals of the Wax Moth, Galleria mellonella L., in Honeybee Colonies. M.Sc. Thesis, University of Nairobi, Kenya (2017).Pickard, S. C., Quinn, R. D. & Szczecinski, N. C. A dynamical model exploring sensory integration in the insect central complex substructures. Bioinspir. Biomim. 15, 026003. https://doi.org/10.1088/1748-3190/ab57b6 (2020).Article 
ADS 

Google Scholar 
Kamala Jayanthi, P. D., Saravan Kumar, P. & Vyas, M. Odour cues from fruit arils of artocarpus heterophyllus attract both sexes of oriental fruit flies. J. Chem. Ecol. 47, 552–563 (2021).Article 

Google Scholar 
Anfora, G., Tasin, M., de Cristofaro, A., Ioriatti, C. & Lucchi, A. Synthetic grape volatiles attract mated Lobesia botrana females in laboratory and field bioassays. J. Chem. Ecol. 35, 1054–1062 (2009).Article 

Google Scholar 
Fand, B. B. et al. Bacterial volatiles from mealybug honeydew exhibit kairomonal activity toward solitary endoparasitoid Anagyrus dactylopii. J. Pest Sci. 93, 195–206 (2020).Article 

Google Scholar 
Kovats, E. Gas chromatographic characterization of organic substances in the retention index system. Adv. Chromotogr. 1, 229–247 (1965).
Google Scholar  More

Potapov, P. et al. Global maps of cropland extent and change show accelerated cropland expansion in the twenty-first century. Nat. Food 3, 19–28 (2022).Article 

Google Scholar 
Land Use Statistics and Indicators. Global, Regional and Country Trends 2000–2020 FAOSTAT Analytical Brief Series No 48 https://www.fao.org/food-agriculture-statistics/data-release/data-release-detail/en/c/1599856/ (FAO, 2022).FAO. Land Statistics. Global, Regional and Country Trends, 1990–2018 FAOSTAT Analytical Brief Series No. 15 https://www.fao.org/3/cb2860en/cb2860en.pdf (FAO, 2021).Summary for policymakers in: Special Report on Climate Change and Land (eds Shukla, P. R. et al.) https://www.ipcc.ch/site/assets/uploads/sites/4/2020/02/SPM_Updated-Jan20.pdf (WMO, in the press).Sustainable Development Goals Indicator 2.4.1 (FAO, accessed); https://www.fao.org/sustainable-development-goals/indicators/241/en/Eggleston, H. S., Buendia, L., Miwa, K., Ngara, T. & Tanabe, K. 2006 IPCC Guidelines for National Greenhouse Gas Inventories (IGES, 2006).Grassi, G. et al. Carbon fluxes from land 2000–2020: bringing clarity on countries’ reporting. Earth Syst. Sci. Data 14, 4643–4666 (2022).Article 
ADS 

Google Scholar 
Tubiello, F. N. et al. Measuring Progress Towards Sustainable Agriculture FAO Statistical Working Papers Series No. 21–24 https://www.fao.org/3/cb4549en/cb4549en.pdf (FAO, 2021).Conchedda, G. & Tubiello, F. N. Drainage of organic soils and GHG emissions: validation with country data. Earth Syst. Sci. Data 12, 3113–3137 (2020).Article 
ADS 

Google Scholar 
Hanson, C., Mazur, E., Stolle, F., Davis, C. & Searchinger, T. 5 takeaways on cropland expansion and what it means for people and the planet. WRI Insights https://www.wri.org/insights/cropland-expansion-impacts-people-planet (2022).Potapov, P. et al. The Global 2000–-2020 land cover and land use change dataset derived from the Landsat archive: first results. Front. Remote Sens. 3, 856903 (2022).Article 

Google Scholar 
Hansen, M. C. et al. Global land use extent and dispersion within natural land cover using Landsat data. Environ. Res. Lett. 17, 034050 (2022).Article 
ADS 

Google Scholar 
Tubiello, F. N. et al. Carbon emissions and removals from forests: new estimates, 1990–2020. Earth Syst. Sci. Data. 13, 1681–1691 (2021).Article 
ADS 

Google Scholar  More

FAO. Handbook on crop statistics: improving methods for measuring crop area, production and yield. (FAO, Rome, Italy, 2018).FAO. Land use statistics and indicators: global, regional and county trends 1990-2019. FAOSTAT Anal. Brief Ser. No 28 (2021).Potapov, P. et al. Global maps of cropland extent and change show accelerated cropland expansion in the twenty-first century. Nat. Food 1–10 (2021) https://doi.org/10.1038/s43016-021-00429-zFAO. A system of integrated agricultural censuses and surveys. (FAO, 2005).FAO. Land use statistics and indicators. Global, regional and country trends, 2000–2020. (FAO, Rome, Italy, 2022).Loveland, T. R. et al. Development of a global land cover characteristics database and IGBP DISCover from 1 km AVHRR data. Int. J. Remote Sens. 21, 1303–1330 (2000).Article 

Google Scholar 
Zanaga, D. et al. ESA WorldCover 10 m 2020 v100. (2021) https://doi.org/10.5281/zenodo.5571936Cochran, W. G. Sampling techniques. (Wiley, 1977).Stehman, S. V. Estimating area and map accuracy for stratified random sampling when the strata are different from the map classes. Int. J. Remote Sens. 35, 4923–4939 (2014).Article 

Google Scholar 
Tsujino, R., Kaijisa, T. & Yumoto, T. Causes and history of forest loss in Cambodia. Int. For. Rev. 21, 372–384 (2019).
Google Scholar 
Hu, Q. et al. Global cropland intensification surpassed expansion between 2000 and 2010: A spatio-temporal analysis based on GlobeLand30. Sci. Total Environ. 746, 141035 (2020).Grainger, A. Difficulties in tracking the long-term global trend in tropical forest area. Proc. Natl Acad. Sci. 105, 818–823 (2008).Article 
ADS 
CAS 

Google Scholar 
FAO. FAOSTAT. https://www.fao.org/faostat/en/#home (2021). More

Fully terrestrial eDNA sampling approaches offer a potentially powerful addition to biodiversity monitoring efforts23,24. However, protocols for using eDNA-based methods to characterize terrestrial biodiversity, and vertebrate communities in particular, are still nascent27,28,30. In this study, we show for the first time that an eDNA metabarcoding approach can be used to broadly characterize tree-dwelling mammal communities by sampling tree trunks and surrounding soil. Our findings add to recent work (e.g., for reptiles22,24) showing that surface eDNA collection methods, which are relatively untested compared with soil-based eDNA methods, can also be effective at detecting terrestrial vertebrates. Further, we demonstrate that supplementing metabarcoding detection with qPCR-based methods can greatly improve sensitivity, a potentially important consideration for monitoring schemes focused on rare taxa (e.g., Refs.11,12). Together, our results have significant implications for global biodiversity conservation as the broader guild of arboreal vertebrates includes highly threatened5,48,49, as well as invasive alien species50, that are often cryptic, inhabit inaccessible locations, and are therefore challenging to monitor.Our methods captured over 60% of the mammalian diversity expected at the sites, and a similar fraction of the subset of arboreal species, despite sampling only 21 trees. Species accumulation curves suggest that more species would likely have been added with increased sampling effort. These results broadly agree with those of Leempoel et al.23 who found that soil eDNA metabarcoding well characterized mammal communities in California chaparral. However, in both our study and that of Leempoel et al.23, some conspicuous absences were evident. Bats comprised all of the arboreal species that we expected but failed to detect at our sites using metabarcoding (Fig. 1A). Leempoel et al.23 also noted a lack of bat detections (2 of 14 possible taxa detected), which they suggested could be due to low efficiency of either the 12S primer set or of their soil sampling methods for that order. While both reasons could also apply to the lack of bats detected in our study (discussed further below), the performance of the 12S primer set very likely contributed to our lack of American black bear detections as MiMammal-U primers are known to be ineffective at amplifying bear DNA38. These challenges highlight the reality that false negatives and varying detectability among species are common issues to all survey approaches, including eDNA metabarcoding. Our study represents a rare example among metabarcoding studies in that it uses repeated sampling and community occupancy models to quantify false negative rates. This quantitative approach, coupled with continued experimentation with different molecular techniques and survey methods (e.g., Refs.23,27), will be vital to helping researchers decide how eDNA metabarcoding methods will fit into existing biodiversity monitoring efforts moving forward.Although our results suggest that sampling for tree-roosting bats using eDNA metabarcoding still requires further research and optimization, our approach likely has application to characterizing communities in a much broader range of arboreal species globally. Geographic regions with multiple elusive arboreal mammals of management interest—for example, gliders and tree kangaroos in Australasia, or primates in the global tropics—may be particularly suited for a metabarcoding approach for community-level assessments4,8,9,49. It may be especially useful for rapid biodiversity assessments (RBAs51) in remote forested environments, where the ability to collect multiple samples relatively rapidly without regard to time of day would be a key advantage27. Existing survey methods to monitor arboreal mammals tend to be optimized for particular groups of species, often segregated by body size and behavior, with no suitable single method available to characterize all members of the guild4,8,9,16,49. Diurnal and nocturnal species, for example, often require separate survey methods or timing8. While camera traps capture both diurnal and nocturnal species, they typically miss smaller species16,23. The need for multiple methods to survey for nocturnal and diurnal, or large and small, species separately raises the cost of sampling and can result in datasets that are difficult to compare across sites because of inherent sampling biases8. Excluding bats, we found encouraging results for both diurnal and nocturnal arboreal species of a broad range of body sizes, detecting all seven expected species (Fig. 1A). While more work is needed to assemble robust genetic reference libraries before global arboreal mammal monitoring with eDNA metabarcoding will be broadly feasible, a clear advantage of the technique remains the power to detect a broad swath of species, with widely varying morphologies and behaviors, with a single method23,27,28,51.The promise of eDNA metabarcoding approaches for at least some arboreal guilds is well illustrated by our results for southern flying squirrel, Glaucomys volans. Like other flying squirrels (Tribe: Pteromyini), this species is strictly nocturnal, highly arboreal, and tends to get injured in live traps, making it difficult to directly observe and monitor48,52. Yet G. volans eDNA was readily detectable using metabarcoding in our study, occurring in 19–26% of soil samples and 47–52% of roller samples across both sites. Our similarly encouraging results for detecting other squirrels (Sciuridae) also bode well for management applications. For example, the methods would enable fine-scale mapping of habitat use in places such as the United Kingdom where native red squirrels (Sciurus vulgaris) are outcompeted by eastern gray squirrels, or the Delmarva peninsula (USA) to support the conservation efforts for the Delmarva fox squirrel (Sciurus niger cinereus)53. Further research is needed to determine the extent to which our results for squirrels generalize to other taxa with similar active tree-climbing lifestyles (e.g., gliders4, primates49).Our finding that soil samples revealed fewer species, had lower detection probability, and had lower read counts than roller samples, even for some non-arboreal species like white-tailed deer, likely reflects multiple factors. First, soil and tree bark represent markedly different biological and chemical environments that likely differ in eDNA quantity by species, eDNA persistence rates54,55, and microorganism abundance. The latter may be especially pertinent to our study as we observed a relatively large drop in the number of reads after removal of microorganism reads, especially for soil samples. This suggests that performing additional purification steps prior to sequencing could boost the ability of both methods, and especially soil eDNA, to detect target species by increasing mammalian sequencing depth. Other in-lab factors, such as method of extraction23 or choice of primers, similarly have the potential to influence the recovery and amplification of target species’ DNA and should be the focus of future research.Next, our focal trees were not chosen to occur near any special attractants or areas of multi-species use, such as saltlicks or water sources, which has proven successful in other vertebrate eDNA studies18,25,31,32,56. It is possible that adding a broader range of soil sampling sites, including some targeted towards other guilds (e.g., burrow users32,56), would have yielded a more complete inventory. Nevertheless, both soil and surface methods have advantages over the much more commonly-used metabarcoding approaches that rely on natural water bodies for assessing mammal communities16,17,27,28,29,30,38 as they are not limited to where these features occur. Our study is the first to suggest that surface eDNA metabarcoding methods can be a powerful supplement to established soil-based methods of characterizing mammal communities, especially for arboreal species.As noted, bats were especially lacking from our eDNA metabarcoding results, with only two of six likely species detected. Notably, our metabarcoding species list lacked two of the bat species that our sampling scheme was designed around (eastern red bat and northern long-eared bat) and for which we had confirmed recent presence at the sites (Table 1). The lack of northern long-eared bat detections may directly relate to recent precipitous population declines (~ 99%) caused by white-nose syndrome57. However, the lack of eastern red bat detections was especially surprising as roosting of this species was suspected based on telemetry in 17 of our 21 target trees. Reasons for this omission may relate to the fact that eastern red bats roost singly on small twigs and in leaf clusters, and therefore may not leave much DNA on tree trunks. Another possibility is low efficiency of the 12S primer set for bats, although we were unable to find information about this in the literature. It is notable that Leempoel et al.23 had a similarly poor representation of bats with comparable soil-based methods. However, our metabarcoding results did indicate that we are capable of detecting even uncommon, or at least unexpected, bat species with our methods. Eastern small-footed bat, which is typically viewed as a rock-roosting species and is considered endangered by the International Union for Conservation of Nature (IUCN)58, was detected in both soil and surface eDNA samples from Morristown National Historic Park. This species was not otherwise confirmed as present at the site until a year later, in spring 2022, when it was caught in a mist net (BM, unpublished data). Our results with respect to bat detections, along with those of others23, underscore the need for further research to adapt eDNA metabarcoding methods to this vulnerable group, which could contribute much needed demographic and distribution information. This is especially urgent as 18% of bat species are listed as “data deficient” by the IUCN, while 57% lack basic population trend information5,58.Our comparison of qPCR to metabarcoding detection methods for big brown bat represents a hopeful result for the use of eDNA to monitor rare vertebrates that are of particular conservation interest. It is well-known that qPCR-based eDNA surveys targeted towards individual species return higher detection probabilities and have greater power at low abundance, than metabarcoding approaches59. Our results agree, showing for the first time that adding a qPCR step in the analysis of surface and soil eDNA samples can be effective for detecting bats in forested environments. The addition of a qPCR step opens the door for developing species-specific assays to increase detection power for endangered or elusive bat species, or other cryptic arboreal mammals49. Emerging molecular detection approaches such as droplet digital PCR have the potential to increase this sensitivity even further59. Like other eDNA-based tools and survey tools in general, careful consideration of sampling effort, the natural history of target species, and the configuration of different field and molecular methods will be key to optimizing our approach to characterize mammal communities, or to target a particular species, in different regions.Although eDNA surveys are not inexpensive given the need for both fieldwork and molecular analyses, they can be cheaper than conventional approaches, especially if such approaches require many hours of fieldwork or expensive equipment60. Thus, the relative cost-effectiveness of surface or soil eDNA surveys will depend heavily on the mammal communities of interest, the mix of methods that must be employed to effectively sample them, and the purpose of the sampling efforts. However, even if costs are increased, eDNA surveys can reduce field time to the extent that they can improve detection rates, either by replacing or supplementing conventional sampling methods (e.g., as a supplement to visual observations). With higher detection rates, fewer visits are required to achieve the same results. This operational efficiency would be especially advantageous when field conditions present safety risks, are intrusive to sensitive habitats, or are challenging to access. For example, adding surface eDNA sampling to existing visual surveys of eastern wood rats (Neotoma floridana), a cryptic mammal that inhabits steep, rocky slopes in the eastern US, could likely increase detection power, thereby reducing the need for additional risky and costly sampling visits. More studies involving direct comparisons among methods (e.g., Refs.23,24,30,60), in a variety of ecoregions, are needed to determine the extent to which incorporating our methods into existing vertebrate monitoring workflows would increase efficiency.Finally, we detected other vertebrates, including seven birds and one salamander, in soil and surface eDNA samples, despite our use of a mammal-specific primer set. This is similar to results from Leempoel et al.23 in California using the same primer set, in which six bird species were detected. We found that surface eDNA detected more bird species than soil, perhaps for the same reasons as for mammals (above). Our results provide evidence that surface eDNA surveys, with taxon-specific primers, could be used to survey bird communities, or used to target particularly rare species in forested ecosystems (e.g., Ref.61). Our detection of a salamander, coupled with recent promising research into reptile detection using surface eDNA methods22,24 suggests a broader potential for applications with other vertebrates as well. Finally, both surface and soil eDNA metabarcoding can be expanded beyond forests, providing insight into their effectiveness in other habitats (e.g., caves17 or talus slopes). Our study and others highlight that the potential of coupling surface and soil eDNA methods for detecting and monitoring mammalian biodiversity, and terrestrial organisms generally, has yet to be fully realized. More

A map of the Amazon River and its tributaries, as published in Alfred Russel Wallace’s 1853 book.Credit: Mary Evans/Natural History Museum

Dzoodzo Baniwa, a member of an Indigenous community in Brazil’s Amazonas state, has been collecting data on the region’s biodiversity for around 15 years. He lives in a remote village called Canadá on the Ayari River, a tributary of the Içana, which in turn feeds the Rio Negro, one of the main branches of the Amazon. The nearest city, São Gabriel da Cachoeira, is a three-day trip by motor boat.Dzoodzo (who goes by his Indigenous name but is also known as Juvêncio Cardoso) takes inspiration for his work from many cross-cultural sources. A perhaps unexpected one is a 170-year-old book by the British naturalist Alfred Russel Wallace, who visited the Amazon and Negro rivers on his expeditions in 1848–52. A Narrative of Travels on the Amazon and Rio Negro gives detailed accounts of the wildlife and people Wallace encountered near Dzoodzo’s home, including the Guianan cock-of-the-rock (Rupicola rupicola), a bright orange bird that Wallace describes as “magnificent … sitting amidst the gloom, shining out like a mass of brilliant flame”1.Dzoodzo’s passion for local biodiversity is reflected in his work at Baniwa Eeno Hiepole School, an internationally praised education centre for Indigenous people. He dreams of one day turning it into a research institute and university that might increase scientific understanding of the region’s species, including R. rupicola.
Alfred Russel Wallace’s first expedition ended in flames
Wallace, who was born 200 years ago, on 8 January 1823, is best known for spurring Charles Darwin into finally publishing On the Origin of Species, after Wallace sent Darwin his own independent discovery of evolution by natural selection in 1858. Most of Wallace’s subsequent work drew on observations from his 1854–62 expeditions in southeast Asia; his earlier work in Amazonia is much less well known.Yet there are lessons from Wallace’s time in Brazil that are especially relevant for conservationists and other scientists today — notably, what can come from paying attention to what local people say about their own territory.Barriers and boundariesWallace made two key contributions that still shape thinking about Amazonia, the world’s most biodiverse region, which covers parts of Bolivia, Brazil, Colombia, Ecuador, Peru, Venezuela, Guyana, Suriname and French Guiana.On 14 December 1852, Wallace read out his manuscript ‘On the monkeys of the Amazon’ at a meeting of the Zoological Society of London. In this study, which was later published2, Wallace relays observations that form the basis of the most debated hypothesis for how Amazonian organisms diversified: the riverine barrier hypothesis.His paper refers to the large Amazonian rivers as spatial boundaries to the ranges of several primate species. “I soon found that the Amazon, the Rio Negro and the Madeira formed the limits beyond which certain species never passed,” he writes. Since 1852, Wallace’s observations that large rivers could act as geographical barriers that shape the distribution of species have been corroborated, criticized and debated by many. The phenomenon he described clearly holds for some groups, such as monkeys and birds3,4, but not for other groups, such as plants and insects5.Subsequent researchers have explored whether the distribution patterns of species, such as those observed by Wallace, indicate that the evolution of the Amazonian drainage system has itself driven the diversification of species6. Work in the past few years by geologists and biologists show that this drainage system, which includes some of the largest rivers in the world, is dynamic7, and that its rearrangements lead to changes in the distribution ranges of species8. Current species ranges thus hold information about how the Amazonian landscape has changed over time.

The Guianan cock-of-the-rock (Rupicola rupicola), which Wallace likened to a “brilliant flame”.Credit: Hein Nouwens/Getty

The second crucial observation made by Wallace, also in his 1852 paper, was that the composition of species varies in different regions. He describes how “several Guiana species come up to the Rio Negro and Amazon, but do not pass them; Brazilian species on the contrary reach but do not pass the Amazon to the north. Several Ecuador species from the east of the Andes reach down into the tongue of land between the Rio Negro and Upper Amazon, but pass neither of those rivers, and others from Peru are bounded on the north by the Upper Amazon, and on the east by the Madeira.” From these observations, he concluded that “there are four districts, the Guiana, the Ecuador, the Peru and the Brazil districts, whose boundaries on one side are determined by the rivers I have mentioned.”
Evolution’s red-hot radical
Even though Amazonia is presented as a single, large, green ellipse in most world maps, it is actually a heterogeneous place, with each region and habitat type holding a distinct set of species9,10. The four districts proposed by Wallace are bounded by the region’s largest rivers: the Amazon, Negro and Madeira. But further studies of species ranges since then have revealed more districts, now called areas of endemism, some of which are also bounded by these and other large Amazonian rivers, such as the Tapajós, Xingu and Tocantins9,11.This recognition of spatial heterogeneity in Amazonian species distributions — first accomplished by Wallace — is essential for today’s research, conservation and planning10. Each area of endemism includes species that occur only in that area. And different areas of endemism are affected differently by anthropogenic impacts, such as deforestation, fires and development10. More than half of Amazonia is now within federal or state reserves or Indigenous lands — territories that are recognized by current governments as belonging to Indigenous people. But nearly half of the region’s areas of endemism are located in the south of the region, close to the agricultural frontier, and the species they contain are severely threatened by habitat loss10 (see also www.raisg.org/en).Local knowledgeAlthough Wallace’s writings indicate that in many ways he admired most of the Indigenous people he met, especially those from the upper Rio Negro basin, he still viewed Indigenous people through the European colonial lens of his time. In A Narrative of Travels on the Amazon and Rio Negro1, Wallace describes the Indigenous communities he encountered as “in an equally low state of civilization” — albeit seemingly “capable of being formed, by education and good government, into a peaceable and civilized community”.Yet he did better than many of his contemporaries when it came to respecting local knowledge. In his 1852 paper, for example, Wallace notes that his fellow European naturalists often give vague information about the locality of their collected specimens, and fail to specify such localities in relation to river margins. By contrast, he writes, the “native hunters are perfectly acquainted” with the impact of rivers on the distribution of species, “and always cross over the river when they want to procure particular animals, which are found even on the river’s bank on one side, but never by any chance on the other.” Likewise, in his 1853 book1, Wallace frequently corroborates his findings with information he has obtained from Indigenous people — for example, about the habitat preferences of umbrellabirds (Cephalopterus ornatus) or of “cow-fish” (manatees; Trichechus inunguis).Considering the vastness and complexity of Amazonia, it is hard to see how Wallace could have gained the insights he did after working in the region for only four years, had he not paid close attention to local knowledge.
The other beetle-hunter
Amazonian Indigenous peoples have had to endure invasion of their lands, enslavement, violence from invaders and the imposition of other languages and cultures. Despite this, numerous Indigenous researchers wish to expand their knowledge about Amazonia by combining Indigenous and European world views. Meanwhile, a better understanding of how the Amazonian socio-ecological system is organized, and how it is being affected by climate change and local and regional impacts12, hinges on the ability of researchers worldwide to learn from and to be led by Indigenous scientists.The 98 Indigenous lands in the Rio Negro basin cover more than 33 million hectares (see go.nature.com/3wkkftu). If the hopes of Dzoodzo and others to build a research institute and university for the region are met, school students will no longer have to leave their homeland to pursue higher education. The community would have a way to document its own knowledge and that of its ancestors in a more systematic way. And the legitimization of Indigenous people’s research efforts in the legal and academic frameworks recognized by non-Indigenous scientists — such as through the awarding of degrees — would make it easier for Indigenous researchers to partner with other organizations, both nationally and internationally.Indigenous people in the Rio Negro basin today are no longer objects of observation — they have taken charge of their own research using tools from different cultures. Indeed, Dzoodzo is turning to Wallace’s writings, in part, to learn more about how his own ancestors lived.Perhaps the thread between Wallace and Dzoodzo, spanning so many years and such disparate cultures, could seed new kinds of partnership in which learning is reciprocal and for the benefit of all. More

EU Commission. Achieve Good Environmental Status. EU Commission Web site https://ec.europa.eu/environment/marine/good-environmental-status/index_en.htm (2008).Olenin, S. et al. Marine strategy framework directive. Task Group 2 (2010).Long, R. The marine strategy framework directive: a new european approach to the regulation of the marine environment, marine natural resources and marine ecological services. Journal of Energy & Natural Resources Law 29, 1–44 (2011).Article 
ADS 

Google Scholar 
Borja, A. et al. Good environmental status of marine ecosystems: what is it and how do we know when we have attained it? Marine Pollution Bulletin 76, 16–27 (2013).Article 

Google Scholar 
Froese, R., Demirel, N., Coro, G. & Kleisner, K. M. & Winker, H. Estimating fisheries reference points from catch and resilience. Fish and Fisheries 18, 506–526 (2017).Article 

Google Scholar 
Coro, G. Open science and artificial intelligence supporting blue growth. Environmental Engineering & Management Journal (EEMJ) 19 (2020).JRC. EU data collection Web site https://datacollection.jrc.ec.europa.eu/ – Accessed June 2022 (2021).EcoScope Consortium. The EcoScope EU project Web site https://ecoscopium.eu/ – Accessed June 2022 (2021).Pikitch, E. K. et al. Ecosystem-based fishery management. Science 305, 346–347 (2004).Article 

Google Scholar 
McLeod, K. L. & Leslie, H. M. Why ecosystem-based management. Ecosystem-based management for the oceans 3–12 (2009).Coro, G., Magliozzi, C., Ellenbroek, A. & Pagano, P. Improving data quality to build a robust distribution model for architeuthis dux. Ecological modelling 305, 29–39 (2015).Article 

Google Scholar 
Coro, G., Ellenbroek, A. & Pagano, P. An open science approach to infer fishing activity pressure on stocks and biodiversity from vessel tracking data. Ecological Informatics 64, 101384 (2021).Article 

Google Scholar 
Elith, J. & Leathwick, J. R. Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution and Systematics 40, 677–697 (2009).Article 

Google Scholar 
Coro, G., Bove, P. & Ellenbroek, A. Habitat distribution change of commercial species in the adriatic sea during the covid-19 pandemic. Ecological Informatics 101675 (2022).Stanton, J. C., Pearson, R. G., Horning, N., Ersts, P. & Reşit Akçakaya, H. Combining static and dynamic variables in species distribution models under climate change. Methods in Ecology and Evolution 3, 349–357 (2012).Article 

Google Scholar 
Coro, G., Magliozzi, C., Ellenbroek, A., Kaschner, K. & Pagano, P. Automatic classification of climate change effects on marine species distributions in 2050 using the aquamaps model. Environmental and ecological statistics 23, 155–180 (2016).Article 
MathSciNet 

Google Scholar 
Coro, G., Pagano, P. & Ellenbroek, A. Detecting patterns of climate change in long-term forecasts of marine environmental parameters. International Journal of Digital Earth 13, 567–585 (2020).Article 
ADS 

Google Scholar 
Wayte, S. E. Management implications of including a climate-induced recruitment shift in the stock assessment for jackass morwong (nemadactylus macropterus) in south-eastern australia. Fisheries Research 142, 47–55 (2013).Article 

Google Scholar 
Tanaka, K. R. Integrating environmental information into stock assessment models for fisheries management. Predicting Future Oceans 193–206 (2019).Szuwalski, C. S. & Hollowed, A. B. Climate change and non-stationary population processes in fisheries management. ICES Journal of Marine Science 73, 1297–1305 (2016).Article 

Google Scholar 
Bevilacqua, A. H. V., Carvalho, A. R., Angelini, R. & Christensen, V. More than anecdotes: fishers’ ecological knowledge can fill gaps for ecosystem modeling. PLoS One 11, e0155655 (2016).Article 

Google Scholar 
Heymans, J. J. et al. Best practice in ecopath with ecosim food-web models for ecosystem-based management. Ecological Modelling 331, 173–184 (2016).Article 

Google Scholar 
Piroddi, C. et al. Historical changes of the mediterranean sea ecosystem: modelling the role and impact of primary productivity and fisheries changes over time. Scientific reports 7, 1–18 (2017).Article 
ADS 

Google Scholar 
Campana, E. F., Ciappi, E. & Coro, G. The role of technology and digital innovation in sustainability and decarbonization of the blue economy. Bulletin of Geophysics and Oceanography 123 (2021).Van Vuuren, D. P. et al. The representative concentration pathways: an overview. Climatic change 109, 5–31 (2011).Article 
ADS 

Google Scholar 
Intergovernmental Panel on Climate Change https://www.academia.edu/download/60673993/climate_change_emission_Special_scenarios20190922-59363-1j1i98f.pdf – Accessed October 2022. IPCC Special Report (2000).Scarcella, G. et al. The potential effects of covid-19 lockdown and the following restrictions on the status of eight target stocks in the adriatic sea. Frontiers in Marine Science 1963 (2022).Wikipedia. ESRI-GRID formats description. Wikipedia https://en.wikipedia.org/wiki/Esri_grid (2022).QGIS. Qgis software version 3.20.0. QGIS Web site https://www.qgis.org/en/site/ (2022).ESRI. Arcgis software version 10.7. ArcGIS Web site https://www.esri.com/it-it/arcgis/products/arcgis-desktop/overview (2022).American Museum of Natural History. Maxent software for modelling species distributions. AMNH Web site https://biodiversityinformatics.amnh.org/open_source/maxent/ (2022).Christensen, V. et al. Ecopath with ecosim: a user’s guide. Fisheries Centre, University of British Columbia, Vancouver 154, 31 (2005).
Google Scholar 
Coll, M., Bundy, A. & Shannon, L. J. Ecosystem modelling using the ecopath with ecosim approach. In Computers in fisheries research, 225–291 (Springer, 2009).Colléter, M. et al. Global overview of the applications of the ecopath with ecosim modeling approach using the ecobase models repository. Ecological Modelling 302, 42–53 (2015).Article 

Google Scholar 
VanDerWal, J., Falconi, L., Januchowski, S., Shoo, L. & Storlie, C. Sdmtools: Species distribution modelling tools: Tools for processing data associated with species distribution modelling exercises. R package version 1, 1 (2014).
Google Scholar 
US National Institutes of Health. ImageJ software for image analysis with Java and the Terrain Cartography plugin for reading ASC files. ImageJ Web site https://imagej.nih.gov/ij/index.html (2018).GDAL. Translator library for raster and vector geospatial data, version 3.5.0. GDAL Web site https://gdal.org/ (2022).Claus, S. et al. Marine regions: towards a global standard for georeferenced marine names and boundaries. Marine Geodesy 37, 99–125 (2014).Article 

Google Scholar 
VLIZ. World marine regions definitions and geospatial data. Marine Regions Web site www.marineregions.org (2022).Coro, G. The ASCFileManagement GitHub repository. GitHub https://github.com/cybprojects65/ASCFileManagement (2022).Ready, J. et al. Predicting the distributions of marine organisms at the global scale. Ecological Modelling 221, 467–478 (2010).Article 

Google Scholar 
Selig, E. R. et al. Global priorities for marine biodiversity conservation. PloS one 9, e82898 (2014).Article 
ADS 

Google Scholar 
O’hara, C. C., Afflerbach, J. C., Scarborough, C., Kaschner, K. & Halpern, B. S. Aligning marine species range data to better serve science and conservation. PLoS One 12, e0175739 (2017).Article 

Google Scholar 
Scarponi, P., Coro, G. & Pagano, P. A collection of aquamaps native layers in netcdf format. Data in brief 17, 292–296 (2018).Article 

Google Scholar 
CMEMS. Copernicus Marine Service ocean products data. Copernicus Marine Service Web site https://marine.copernicus.eu/ (2022).E.U. Copernicus Marine Service Information. Global Ocean 1/12° Physics Analysis and Forecast updated Daily. Copernicus Marine Service Web site https://doi.org/10.48670/moi-00016 (2021).Article 

Google Scholar 
E.U. Copernicus Marine Service Information. Global Ocean Biogeochemistry Analysis and Forecast. Copernicus Marine Service Web site https://doi.org/10.48670/moi-00015 (2021).Article 

Google Scholar 
Hijmans, R. J. et al. Terra: Spatial data analysis. R Spatial Data Science Web site https://rspatial.org/terra/ (2022).MacLeod, C. D. Habitat representativeness score (hrs): a novel concept for objectively assessing the suitability of survey coverage for modelling the distribution of marine species. Journal of the Marine Biological Association of the United Kingdom 90, 1269–1277 (2010).Article 

Google Scholar 
Abdi, H. & Williams, L. J. Principal component analysis. Wiley interdisciplinary reviews: computational statistics 2, 433–459 (2010).Article 

Google Scholar 
Coro, G., Pagano, P. & Ellenbroek, A. Combining simulated expert knowledge with neural networks to produce ecological niche models for latimeria chalumnae. Ecological modelling 268, 55–63 (2013).Article 

Google Scholar 
Coro, G., Pagano, P. & Ellenbroek, A. Automatic procedures to assist in manual review of marine species distribution maps. In International Conference on Adaptive and Natural Computing Algorithms, 346–355 (Springer, 2013).Magliozzi, C., Coro, G., Grabowski, R. C., Packman, A. I. & Krause, S. A multiscale statistical method to identify potential areas of hyporheic exchange for river restoration planning. Environmental Modelling & Software 111, 311–323 (2019).Article 

Google Scholar 
Coro, G. & Bove, P. Global-Scale Parameters for Ecological Models, FigShare, https://doi.org/10.6084/m9.figshare.c.6039275.v4 (2022).Coro, G. Means, standard deviations, geometric means, and log-normal standard deviation of the data produced for the present publication. D4Science distributed storage system https://data.d4science.net/foLS (2022).Mann, M. E., Bradley, R. S. & Hughes, M. K. Global-scale temperature patterns and climate forcing over the past six centuries. Nature 392, 779–787 (1998).Article 
ADS 

Google Scholar 
Biskaborn, B. K. et al. Permafrost is warming at a global scale. Nature communications 10, 1–11 (2019).Article 

Google Scholar 
Nemani, R. R. et al. Climate-driven increases in global terrestrial net primary production from 1982 to 1999. science 300, 1560–1563 (2003).Article 
ADS 

Google Scholar 
Huang, Y., Zhang, W., Sun, W. & Zheng, X. Net primary production of chinese croplands from 1950 to 1999. Ecological Applications 17, 692–701 (2007).Article 

Google Scholar 
Sunlu, U., Aksu, M., Buyukisik, B. & Sunlu, F. S. Spatio-temporal variations of organic carbon and chlorophyll degradation products in the surficial sediments of izmir bay (aegean sea/turkey). Environmental monitoring and assessment 146, 423–432 (2008).Article 

Google Scholar 
Kubryakov, A., Mikaelyan, A., Stanichny, S. & Kubryakova, E. Seasonal stages of chlorophyll-a vertical distribution and its relation to the light conditions in the black sea from bio-argo measurements. Journal of Geophysical Research: Oceans 125, e2020JC016790 (2020).ADS 

Google Scholar 
Gamo, T. Global warming may have slowed down the deep conveyor belt of a marginal sea of the northwestern pacific: Japan sea. Geophysical Research Letters 26, 3137–3140 (1999).Article 
ADS 

Google Scholar 
Mahaffey, C., Palmer, M., Greenwood, N. & Sharples, J. Impacts of climate change on dissolved oxygen concentration relevant to the coastal and marine environment around the uk. MCCIP Science Review 2002, 31–53 (2020).
Google Scholar 
Zhang, W., Dunne, J. P., Wu, H., Zhou, F. & Huang, D. Using timescales of deficit and residence to evaluate near-bottom dissolved oxygen variation in coastal seas. Journal of Geophysical Research: Biogeosciences 127, e2021JG006408 (2022).ADS 

Google Scholar 
Helm, K. P., Bindoff, N. L. & Church, J. A. Changes in the global hydrological-cycle inferred from ocean salinity. Geophysical Research Letters 37 (2010).Ren, L., Speer, K. & Chassignet, E. P. The mixed layer salinity budget and sea ice in the southern ocean. Journal of Geophysical Research: Oceans 116 (2011).Mahmuduzzaman, M. et al. Causes of salinity intrusion in coastal belt of bangladesh. International Journal of Plant Research 4, 8–13 (2014).
Google Scholar 
Podymov, O., Zatsepin, A. & Ocherednik, V. Increase of temperature and salinity in the active layer of the north-eastern black sea from 2010 to 2020. Physical Oceanography 28, 257–265 (2021).Article 

Google Scholar 
Mizyuk, A. & Puzina, O. Sea ice modeling in the sea of azov for a study of long-term variability. In IOP Conference Series: Earth and Environmental Science, vol. 386, 012023 (IOP Publishing, 2019).Pärn, O., Friedland, R., Rjazin, J. & Stips, A. Regime shift in sea-ice characteristics and impact on the spring bloom in the baltic sea. Oceanologia 64, 312–326 (2022).Article 

Google Scholar 
Lundesgaard, Ø., Sundfjord, A. & Renner, A. H. Drivers of interannual sea ice concentration variability in the atlantic water inflow region north of svalbard. Journal of Geophysical Research: Oceans 126, e2020JC016522 (2021).ADS 

Google Scholar 
Schwegmann, S. & Holfort, J. Regional distributed trends of sea ice volume in the baltic sea for the 30-year period 1982 to 2019. Meteorologische Zeitschrift 33–43 (2021).Simon, S. Interpretation of the correlation coefficient. PMean Web site http://www.pmean.com/definitions/correlation.htm (2020).Yacobi, Y. et al. Chlorophyll distribution throughout the southeastern mediterranean in relation to the physical structure of the water mass. Journal of Marine Systems 6, 179–190, https://doi.org/10.1016/0924-7963(94)00028-A (1995).Article 
ADS 

Google Scholar 
Kucuksezgin, F., Balci, A., Kontas, A. & Altay, O. Distribution of nutrients and chlorophyll-a in the aegean sea. Oceanologica Acta 18, 343–352 (1995).
Google Scholar 
Villate, F., Aravena, G., Iriarte, A. & Uriarte, I. Axial variability in the relationship of chlorophyll a with climatic factors and the north atlantic oscillation in a basque coast estuary, bay of biscay (1997–2006). Journal of Plankton Research 30, 1041–1049 (2008).Article 

Google Scholar 
Iriarte, A. et al. Dissolved oxygen in contrasting estuaries of the bay of biscay: effects of temperature, river discharge and chlorophyll a. Marine Ecology Progress Series 418, 57–71 (2010).Article 
ADS 

Google Scholar 
Stanev, E. V. Black sea dynamics. Oceanography 18, 56–75 (2005).Article 

Google Scholar 
Tsimplis, M. N. & Rixen, M. Sea level in the mediterranean sea: The contribution of temperature and salinity changes. Geophysical research letters 29, 51–1 (2002).Article 

Google Scholar 
Schneider, A., Wallace, D. W. & Körtzinger, A. Alkalinity of the mediterranean sea. Geophysical Research Letters 34 (2007).Sara, G., Porporato, E. M., Mangano, M. C. & Mieszkowska, N. Multiple stressors facilitate the spread of a non-indigenous bivalve in the mediterranean sea. Journal of Biogeography 45, 1090–1103 (2018).Article 

Google Scholar 
Soto-Navarro, J. et al. Evolution of mediterranean sea water properties under climate change scenarios in the med-cordex ensemble. Climate Dynamics 54, 2135–2165 (2020).Article 
ADS 

Google Scholar 
Dietze, H. & Löptien, U. Retracing hypoxia in eckernförde bight (baltic sea). Biogeosciences 18, 4243–4264 (2021).Article 
ADS 

Google Scholar 
Ulses, C. et al. Oxygen budget of the north-western mediterranean deep-convection region. Biogeosciences 18, 937–960 (2021).Article 
ADS 

Google Scholar 
Jaskulak, M., Sotomski, M., Michalska, M., Marks, R. & Zorena, K. The effects of wastewater treatment plant failure on the gulf of gdansk (southern baltic sea). International Journal of Environmental Research and Public Health 19, 2048 (2022).Article 

Google Scholar 
Mihanović, H. et al. Observation, preconditioning and recurrence of exceptionally high salinities in the adriatic sea. Frontiers in Marine Science 8, 834 (2021).Article 

Google Scholar 
De Leo, F., Besio, G. & Mentaschi, L. Trends and variability of ocean waves under rcp8. 5 emission scenario in the mediterranean sea. Ocean Dynamics 71, 97–117 (2021).Article 
ADS 

Google Scholar 
Omar, A. M. et al. Trends of ocean acidification and pco2 in the northern north sea, 2003–2015. Journal of Geophysical Research: Biogeosciences 124, 3088–3103 (2019).Article 

Google Scholar 
Kröncke, I. et al. Comparison of biological and ecological long-term trends related to northern hemisphere climate in different marine ecosystems. Nature Conservation (2019).Bonnet, D. et al. Comparative seasonal dynamics of centropages typicus at seven coastal monitoring stations in the north sea, english channel and bay of biscay. Progress in oceanography 72, 233–248 (2007).Article 
ADS 

Google Scholar 
Borja, Á. et al. Implementation of the european marine strategy framework directive: A methodological approach for the assessment of environmental status, from the basque country (bay of biscay). Marine Pollution Bulletin 62, 889–904, https://doi.org/10.1016/j.marpolbul.2011.03.031 (2011).Article 

Google Scholar 
Coro, G. An open-source re-implementation of the habitat representativeness score. GitHub https://github.com/cybprojects65/HabitatRepresentativenessScore (2022).Coro, G. An OGC-WPS compliant interface to calculate Habitat Representativeness Score. D4Science RPrototypingLab VRE https://services.d4science.org/group/rprototypinglab/data-miner?OperatorId = org.gcube.dataanalysis.wps.statisticalmanager.synchserver.mappedclasses.transducerers.HABITAT_REPRESENTATIVENESS_SCORE (2022).Assante, M. et al. Enacting open science by d4science. Future Generation Computer Systems 101, 555–563 (2019).Article 

Google Scholar 
Assante, M. et al. The gcube system: delivering virtual research environments as-a-service. Future Generation Computer Systems 95, 445–453 (2019).Article 

Google Scholar 
Assante, M. et al. Virtual research environments co-creation: The d4science experience. Concurrency and Computation: Practice and Experience e6925 (2022).Coro, G., Candela, L., Pagano, P., Italiano, A. & Liccardo, L. Parallelizing the execution of native data mining algorithms for computational biology. Concurrency and Computation: Practice and Experience 27, 4630–4644 (2015).Article 

Google Scholar 
Coro, G., Panichi, G., Scarponi, P. & Pagano, P. Cloud computing in a distributed e-infrastructure using the web processing service standard. Concurrency and Computation: Practice and Experience 29, e4219 (2017).Article 

Google Scholar 
Gačić, M., Borzelli, G. E., Civitarese, G., Cardin, V. & Yari, S. Can internal processes sustain reversals of the ocean upper circulation? the ionian sea example. Geophysical research letters 37 (2010).Grilli, F. et al. Seasonal and interannual trends of oceanographic parameters over 40 years in the northern adriatic sea in relation to nutrient loadings using the emodnet chemistry data portal. Water 12, 2280 (2020).Article 

Google Scholar 
Cozzi, S. et al. Climatic and anthropogenic impacts on environmental conditions and phytoplankton community in the gulf of trieste (northern adriatic sea). Water 12, 2652 (2020).Article 

Google Scholar 
Ducrotoy, J.-P. & Elliott, M. The science and management of the north sea and the baltic sea: Natural history, present threats and future challenges. Marine pollution bulletin 57, 8–21 (2008).Article 

Google Scholar 
Dupont, N. & Aksnes, D. L. Centennial changes in water clarity of the baltic sea and the north sea. Estuarine, Coastal and Shelf Science 131, 282–289 (2013).Article 
ADS 

Google Scholar 
Dippner, J. W., Möller, C. & Hänninen, J. Regime shifts in north sea and baltic sea: a comparison. Journal of Marine Systems 105, 115–122 (2012).Article 
ADS 

Google Scholar 
Sisma-Ventura, G. et al. Post-eastern mediterranean transient oxygen decline in the deep waters of the southeast mediterranean sea supports weakening of ventilation rates. Frontiers in Marine Science 1202 (2021).Mavropoulou, A.-M., Vervatis, V. & Sofianos, S. Dissolved oxygen variability in the mediterranean sea. Journal of Marine Systems 208, 103348 (2020).Article 

Google Scholar 
Tyberghein, L. et al. Bio-oracle: a global environmental dataset for marine species distribution modelling. Global ecology and biogeography 21, 272–281 (2012).Article 

Google Scholar 
Assis, J. et al. Bio-oracle v2. 0: Extending marine data layers for bioclimatic modelling. Global Ecology and Biogeography 27, 277–284 (2018).Article 

Google Scholar 
Coro, G. & Bove, P. A high-resolution global-scale model for covid-19 infection rate. ACM Transactions on Spatial Algorithms and Systems (TSAS) 8, 1–24 (2022).Article 

Google Scholar 
Inness, A. et al. The cams reanalysis of atmospheric composition. Atmospheric Chemistry and Physics 19, 3515–3556 (2019).Article 
ADS 

Google Scholar 
Karger, D. N., Schmatz, D. R., Dettling, G. & Zimmermann, N. E. High-resolution monthly precipitation and temperature time series from 2006 to 2100. Scientific data 7, 1–10 (2020).Article 

Google Scholar 
Kesner-Reyes, K. et al. AquaMaps Environmental Dataset: Half-Degree Cells Authority File (HCAF ver. 7, 10/2019). AquaMaps Web site https://www.aquamaps.org/main/envt_data.php (2019).Kesner-Reyes, K. et al. AquaMaps Environmental Dataset: Half-Degree Cells Authority File (HCAF ver. 6, 08/2016). AquaMaps Web site https://www.aquamaps.org/main/envt_data.php (2016).NASA-NEX. NASA Earth Exchange data. NASA-NEX Web site https://www.nasa.gov/nex/data – data were publicly accessible up to 2020 (2020).Coro, G. A global-scale ecological niche model to predict sars-cov-2 coronavirus infection rate. Ecological modelling 431, 109187 (2020).Article 

Google Scholar 
CAMS. Global inversion-optimised greenhouse gas fluxes and concentrations. Copernicus Atmosphere Web site https://ads.atmosphere.copernicus.eu/cdsapp#/dataset/cams-global-greenhouse-gas-inversion?tab=doc (2020).NOAA. ETOPO2 Topography and Bathymetry. NOAA Web site https://sos.noaa.gov/catalog/datasets/etopo2-topography-and-bathymetry-natural-colors/ (2010).Coro, G. & Trumpy, E. Predicting geographical suitability of geothermal power plants. Journal of Cleaner Production 267, 121874 (2020).Article 

Google Scholar 
NOAA. World Vector Shorelines. NOAA Web site https://shoreline.noaa.gov/data/datasheets/wvs.html (2019).Tozer, B. et al. Global bathymetry and topography at 15 arc sec: Srtm15+. Earth and Space Science 6, 1847–1864, https://doi.org/10.1029/2019EA000658 (2019).Article 
ADS 

Google Scholar 
Ramesh, R. et al. Land–ocean interactions in the coastal zone: Past, present & future. Anthropocene 12, 85–98 (2015).Article 

Google Scholar 
Spalding, M. et al. World atlas of coral reefs (Univ of California Press, 2001).Laske, G. A global digital map of sediment thickness. Eos Trans. AGU 78, F483 (1997).
Google Scholar 
Davies, J. H. Global map of solid earth surface heat flow. Geochemistry, Geophysics, Geosystems 14, 4608–4622 (2013).Article 
ADS 

Google Scholar 
Rybach, L. & Muffler, L. J. P. Geothermal systems: principles and case histories. Chichester, Sussex, England and New York, Wiley-Interscience, 1981. 371 p. (1981).Glassley, W. E. Geology and hydrology of geothermal energy. Power Stations Using Locally Available Energy Sources: A Volume in the Encyclopedia of Sustainability Science and Technology Series, Second Edition 23–34 (2018).Barbier, E. Geothermal energy technology and current status: an overview. Renewable and sustainable energy reviews 6, 3–65 (2002).Article 
ADS 

Google Scholar 
Engdahl, E. R., van der Hilst, R. & Buland, R. Global teleseismic earthquake relocation with improved travel times and procedures for depth determination. Bulletin of the Seismological Society of America 88, 722–743 (1998).
Google Scholar 
Engdahl, E. R. Global seismicity: 1900–1999. International handbook of earthquake and engineering seismology 665–690 (2002).Richts, A., Struckmeier, W. F. & Zaepke, M. WHYMAP and the groundwater resources map of the world 1: 25,000,000. In Sustaining groundwater resources, 159–173 (Springer, 2011).Warszawski, L. et al. Center for international earth science information network—ciesin—columbia university.(2016). gridded population of the world, version 4 (gpwv4): Population density. palisades. ny: Nasa socioeconomic data and applications center (sedac). Atlas of Environmental Risks Facing China Under Climate Change 228, https://doi.org/10.7927/h4np22dq (2017). More