Phloem sap in Cretaceous ambers as abundant double emulsions preserving organic and inorganic residues

Fossilized remains preserved in amber provide abundant data on the paleobiota surrounding the resin-producing plants, but relatively scarcer information about the resinous sources themselves. Here, dark pseudoinclusions in kidney-shaped amber pieces from the Early Cretaceous (Albian) amber from Spain are studied. This type of fossilized remain, abundant in Cretaceous ambers, was first interpreted as fossilized vacuole-bearing microorganisms, but later regarded as artifactual and probably secreted by the resinous trees, although their origin remained unclear. Using complementary microscopy (light, electron, confocal), spectroscopy (infrared, micro-Raman), mass spectrometry and elemental analysis techniques, we demonstrate that the pseudoinclusions correspond to droplets of phloem sap containing amber spheroids and preserving both organic and inorganic residues consistent with degraded components from the original sap. The amber pieces containing pseudoinclusions are fossilized, resin-in-sap-in-resin double emulsions, showing banding patterns with differential content of resin-in-sap emulsion droplets. Our findings represent the first time fossilized phloem sap, 105 million years old, has been recognized and characterized, and open new lines of paleontological research with taxonomic, taphonomic, physiological and ecological implications.

, it does so using 552 nm laser excitation (red pseudocolor; Fig. 7b,f). In reflection mode, all sub-inclusions reflect the 488 nm laser (green pseudocolor; Fig. 7c,g). Figure 7d,h shows the combination of amber and dark matter emissions.
LSCM measurements show a maximum emission intensity with 405 nm laser excitation for both the "only amber" and the "pseudoinclusions" experiments (see Materials and Methods section). It occurs at 437-438 nm (intensity of the maximum peak at 150 and 115 fluorescence arbitrary units, respectively), with one shoulder at approximately 460 nm (Fig. 8a). In the "only amber experiment", the emission spectrum using 488 nm laser excitation shows signal in the 500 to 600 nm region, with a peak at 531 nm (10 fluorescence arbitrary units), and weaker peaks at 621, 671 and 741 nm (Fig. 8b). In addition, with 552 nm laser excitation the amber emits in the 565 to 700 nm range, with a weak peak at approximately 570 nm (6 fluorescence arbitrary units) and a shoulder at 595 nm (Fig. 8b). Lastly, in the "pseudoinclusions experiment", the recorded fluorescence in green-yellow region is similar to that obtained in the "only amber experiment" using 488 and 552 nm laser excitation, but with some differences (Fig. 8c). The maximum emission with 488 nm of excitation occurs at 532 nm, although with another peak at 517 nm (27 fluorescence arbitrary units) and others with lower intensity at 597, 672, 702 (clearly absent in the "only amber experiment") and 737 nm (Fig. 8c). With 552 nm laser excitation, the maximum fluorescence peak was recorded at 576 nm with numerous sub-peaks at 586, 586, 606, 621 and 636 nm. Finally, the 638 nm laser excitation produces a small emission in the red region, between 650 and 720 nm (Fig. 8c). www.nature.com/scientificreports www.nature.com/scientificreports/ Mass spectrometric data. The Electrospray Ionization Mass Spectrometry (ESI-MS) spectrum from the analyzed pseudoinclusion-rich dark fraction shows two characteristic peaks at m/z 188.65 and m/z 359.23 (Fig. S1), which are absent in the analyzed pseudoinclusion-void light fraction. The molecular mass corresponding to the peak at m/z 188.65 is 374 g/mol, assuming the proton as an ionization mechanism, giving (M + 2H/2) 2+ . Likewise, the molecular mass corresponding to the peak at m/z 359 is 761 g/mol. The tandem mass spectrometry (MSMS) fragmentation spectrum of the peak at m/z 359 shows two consecutive water losses at m/z 350 and m/z 341, also with 2+ charge. (h) Undeformed pseudoinclusion of irregular morphology (from preparation 18711). (i-j) Moderately deformed pseudoinclusion of irregular morphology (from preparation 18711). Arrows in (b) and (j) point to one of the light spheroids/ellipsoids (respectively) composing the pseudoinclusions and surrounded by dark matter. (k-l) Moderately deformed, ovoid pseudoinclusions imaged at polished amber (from preparation 18068). Note the presence of electron-dense (mineralized) sub-inclusions located within the dark matter (arrows). (m) Moderately deformed, ovoid pseudoinclusion imaged on a broken amber surface (from sample 18614). Scale bars: 30 µm (a,c), 50 µm (b,d,f,g,h,i,j), 100 µm (e), 20 µm (k,l,m).

Discussion
From the morphological standpoint, the undeformed pseudoinclusions 12,16 (Fig. 3a-c) resemble emulsion droplets from double emulsions 28,29 . If an emulsion is a mixture of immiscible liquids, a double emulsion is an emulsion of emulsions, where droplets of the dispersed phase themselves contain smaller droplets 28 . More specifically, undeformed pseudoinclusions are most similar to the type C double emulsion droplets described by some authors, which contain a high number of the smaller, internal droplets 29 . Double emulsions have been intensively studied for decades due to their widespread use in the pharmaceutical, cosmetic, oil, agricultural and food industries 28 , and their manufacturing and demulsification are still a hot topic of research. The most common types of double emulsions are water-in-oil-in-water (w/o/w) and oil-in-water-in-oil emulsions (o/w/o). In the latter, a hydrophilic phase separates two hydrophobic ones: the primary dispersed phase in the inner emulsion is hydrophobic, the secondary dispersed phase in the outer emulsion (the intermediate phase) is hydrophilic (polar), and the final continuous phase is, again, hydrophobic 28 .
Unlike in the stalactite-shaped amber pieces, the inner layering of which is related to successive resin flows that covered previous ones dried by exposure to aerial conditions (e.g., sunlight, wind 3 ), the banding pattern in the kidney-shaped amber pieces can be explained by a different process. In each of these pieces, most of the new resin inputs became injected into a previously emitted resin accumulation 22 . This circumstance is shown by the abundant pseudoinclusion-bearing dark layers bent by ductile deformation (Fig. 2a,b) and the altered morphology of the pseudoinclusions themselves, which were stretched to different degrees in the same direction of the resin deformation ( Fig. 3a-g). Thus, although the original morphology of most of the pseudoinclusions was nearly spherical, in deformed layers they did not recover such shape because of resin viscosity. However, the resin must have remained relatively fresh to enable the injection of new inputs. In that regard, resin emission in unexposed conditions, such as on roots or in trunk pockets, would have slowed down the volatile loss and resin hardening process 22 , increasing the time in which resin remained ductilely deformable and, thus, favoring the accumulation of pseudoinclusion-bearing dark layers. Although most resin inputs were emitted into ductile resin masses, some inputs intruded into hardened resin bodies, in-filling small fractures (Fig. 2c). The dark coloration of the pseudoinclusions is not deemed to be diagenetic in origin, at least not entirely, as these inclusions present in modern resin are already darkened 16 .
During the Cretaceous, fungal mycelia commonly consumed the fresh, kidney shaped resin emissions from their external surface inwards, creating light brown to opaque cortices 22 . Note that the identification of this Cretaceous filamentous, heterotrophic microorganism is currently under debate 30 . In the studied kidney-shaped amber pieces, the mycelia grew towards the core of the resin emissions, preferably following the layers exhibiting a higher density of pseudoinclusions ( Fig. 1d-g). This strongly suggests that the pseudoinclusions were more nutritious for the fungi than the resin alone. Among the substances secreted by plants, phloem sap is rich in www.nature.com/scientificreports www.nature.com/scientificreports/ nutrients and generally free of deterrent substances and toxins, and has been shown to allow bacterial growth and multiplication 31 .
The overall quantity of dark matter compared to that of amber in the pseudoinclusion-rich layers is low, as indicated by the virtually identical FT-IR spectra between the analyzed samples ( Fig. 5a,b) and the similarity between both spectra and that of raw amber without pseudoinclusions from Rábago/El Soplao 19 (Fig. 5c). On the other hand, the micro-Raman spectrum obtained for the Rábago/El Soplao amber (Fig. 5d) is similar to that of amber or resin from any age due to their common chemical functional groups 26 . Remarkably, the main peaks in the 1200-700 cm −1 range (Fig. 5e,f) shown by the two micro-Raman spectra obtained from the dark matter partly constituting the pseudoinclusions coincide with those characteristic for sugars 32,33 (Fig. 5g-i), especially in the regions of δ(COH), δ(CCH) and δ(OCH) side group deformations 31 . Although these results do not allow to identify the exact molecular composition of the pseudoinclusion dark matter, it likely contains sugar residues that degraded during diagenesis. In the vascular system of terrestrial plants, whereas the xylem transports water and mineral nutrients taken up by the roots from the soil to the aerial part of the plant, the major role of the phloem is to transport the photosynthates from a photosynthetically active source to sink tissues 34 . Xylem sap contains carboxylates, hormones, amino acids, peptides and proteins; phloem sap is rich in nutrients and contains mostly sugars and amino acids, as well as organic acids, vitamins and inorganic ions 31,35 . Although sucrose is widely believed to be the most predominant sugar transported by the phloem in most plants, other sugars such as hexoses (including fructose and glucose) have been found in significant proportions in the phloem sap of certain www.nature.com/scientificreports www.nature.com/scientificreports/ plants 31 , yet their ability to be transported in the phloem remains controversial 36 . The data obtained using ESI-MS ( Fig. S1) show the presence of molecules with molecular masses of 374 and 716 g/mol, which could indicate transformation or polymerization processes of the original sugar. In any case, obtaining two consecutive water losses during MSMS fragmentation of 359 peak (m/z 350 and m/z 341) suggests that the substance of molecular mass 716 g/mol is polar and contains at least two OH groups 37 .
The organic chemical composition of the analyzed amber sample from Rábago/El Soplao (C, O and S; Table 1) is similar to that of other Cretaceous and Cenozoic fossil resins 38 . Regarding trace (inorganic) elements, results from the analyzed sample are coincident with previous analyses of fossil resins that show that Ca is the most abundant trace element in the amber, with presence of other elements in minor proportion 39 . More importantly, the chemical composition of the dark matter significantly differs from that of the amber matrix containing the pseudoinclusions (Table 1, Fig. 6b). Only Na and K were previously identified in pseudoinclusions using ToF-SIMS 16 .
Remarkably, most of the inorganic cations present in the dark matter occur as mineralized sub-inclusions, which become apparent due to their electron-dense appearance (Fig. 3k,l), variable elemental composition and distribution within the pseudoinclusion (Table 1, Fig. 6), as well as fluorescence reflectivity (green pseudocolor) (Fig. 7c,g). Even though the minute size of these mineralized aggregates has prevented a more detailed characterization, their mineral composition must be similar to calcite (C-Ca aggregates), dolomite (C-Ca-Mg aggregates) and apatite (Ca-P aggregates). The fact that the stoichiometric relationships of the elements are not as expected is probably due to contamination of the dark matter matrix that houses the mineralized aggregates. The most plausible scenario is that these mineralized sub-inclusions were formed by concentration of most of the ions originally dissolved in the dark matter during diagenesis. As a result, the inorganic chemical composition of the dark matter itself (Table 1) is not fully representative of the composition of the original fluid. Although our results do not show the presence of pyrite in the samples (note also that almost all Fe contents are below the detection limit), this mineral was abundantly detected within pseudoinclusions in previous studies 10,11 . These studies most probably analyzed different types of amber pieces (not necessarily kidney-shaped) or pseudoinclusions that had a more superficial position within the amber pieces, and were therefore more susceptible to cracks favoring typical allochthonous iron and sulfur infiltration.
All the inorganic elements found in the dark matter and the mineralized sub-inclusions (Table 1), except Ti (the latter present in negligible amounts or nearly so), are found in the phloem sap of extant plants, with concentrations that vary depending on the plant species and environmental conditions: 35,40 Ca is an important element for phloem physiology and signaling; K promotes the load of sugar in leaves and is required for the transmission of electric signals; P is involved in the control of the phloem sap's pH; Mg, Na and Cl contribute to the leaf 's solute balance; Fe is one of the micronutrients involved in carrying the sap; and S is a critical macronutrient for plant growth related to the production of essential amino acids 35 . Although Al is toxic for most plants, for some it appears to be beneficial or necessary, and could be transported by the phloem sap 41 . The presence of traces of these inorganic elements in Cretaceous amber has been interpreted as a consequence of the interaction between the resin and the sap of the same tree, and thus indirectly related to soil-water composition 39 . In that regard, aqueous inclusion droplets from Baltic amber 42 could represent inclusions of sap, since they contain Na, K, Ca, Mg, S and Cl. These aqueous inclusions were primarily interpreted as being derived from water splashed from a saline environment into the resin, yet the authors also suggested that the sap of the producing tree could have also been involved in the formation of these inclusions due to presence of ammonium and acetate ions 42 . Lastly, it is important to note that, although amber is a relatively permeable substance, most likely the detected inorganic www.nature.com/scientificreports www.nature.com/scientificreports/ elements are not contamination originated from the amber-bearing rock during diagenesis. First, the amber pieces analyzed lack internal cracks or any external signs of alteration. Moreover, if contamination had occurred, the obvious compositional difference between the amber and the dark matter would be less likely.
Some ambers exhibit a characteristic bluish to violet color when exposed to long-wave ultraviolet light, including sunlight. This type of amber is known as "blue amber" and, in the Cenozoic, it is known from the Eocene of the Far East of Russia 43 , and the Miocene from the Dominican Republic and Indonesia 27,44 . A fluorescence microscopy study of Dominican amber showed that its maximum emission (449 nm) is very similar to that of perylene (443 nm 44 ); FT-IR and gas chromatography-mass spectrometry studies did not find a distinct signature of this compound in Russian blue amber, even though its autofluorescence is similar to that of the Dominican Republic amber 43,45 . The only Cretaceous blue amber reported to date is extracted from the Rábago/El Soplao www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ deposit 19 . The main autofluorescence from this amber, detected in the blue spectral region (peaking at 437 nm) after 405 nm laser excitation (Fig. 8b), is notably similar to that of the perylene. However, neither FT-IR nor gas chromatography-mass spectrometry analyses of Rábago/El Soplao blue amber found the presence of perylene; instead, the azulene derivative guaiazulene was detected, which could be responsible for the blue fluorescence 19,20 . The intense blue fluorescence exhibited after violet light excitation by both the amber matrix and the light matter spheroids/ellipsoids that give the pseudoinclusions their vacuolated appearance (Fig. 7a,e) indicates that the light matter represents amber. The possibility that the light matter could be empty is ruled out, because it is clearly filled in exposed amber surfaces and has the same appearance as the amber matrix under SEM (Fig. 3k-m). The lack of blue fluorescence in the dark matter (Fig. 7) demonstrates that it does not contain azulene-derived compounds, at least in significant amounts. Thus, the blue emission (peaking at 438 nm) detected in the "pseudoinclusions experiment" is identical to the signal present in the "only amber experiment" (yet slightly weaker) (Fig. 8c) and must correspond to the emission of the amber, both from the matrix containing the pseudoinclusions and from the "vacuolated" amber within the pseudoinclusions.
The dark matter emits in the green-red, yellow-orange, and red regions of the spectrum, using 488, 522 and 638 nm laser excitation, respectively (Fig. 8). Although amber also emits in the same regions using the same laser excitations, the green-red emission of the dark matter is much more intense than that of the amber (approximately three times the intensity), and it shows a somewhat different fluorescence profile (Fig. 8). On the one hand, the presence of autofluorescent substances in the amber and the dark matter emitting in the same regions strengthens the idea that both fluids were produced and emitted by the same tree individual 14,15 . It has been suggested, based on analyses of inorganic trace elements present in French Cretaceous amber, that a certain compositional relationship existed between the resin and the sap 39 . On the other hand, the fluorescence emission in the green-red region of the spectrum for both the dark matter and the amber can be interpreted as content of plant pigments. Although the data obtained are insufficient for a detailed assessment of the fluorescent substances, we suggest the presence of residues of carotenoids (emitting at the 520-580 nm region of the spectrum 46 ), anthocyanins (emitting at 520-610 nm region 46 ) and chlorophylls (emitting in the red region 47 ). These three pigment types are the most important for vascular plants 48 . Anthocyanins are soluble in water 48 and they are found in sap (probably having a deterring effect on phloem sap feeders) 49 . In contrast, carotenoids and chlorophylls are not soluble in water 48 . However, carotenoids are known to be hydrosoluble when forming complexes with proteins and sugars 48 , both of which are abundant in phloem sap 35 . In addition, carotenoids have been identified in the www.nature.com/scientificreports www.nature.com/scientificreports/ wood of various tree species 50 , suggesting that the in situ formation of carotenoids in the living cells occurs in the sapwood 48 . Regarding chlorophylls, minor residues could have been incorporated (traumatically?) into the dark matter from photosynthetic tissues, which could explain the obtained low-intensity fluorescent signal.
The sum of the evidence presented here proves that the pseudoinclusions, common in Cretaceous ambers and originally misidentified as microbes, are composed of amber equivalent to the amber matrix and of dark matter containing inorganic elements and organic compounds, the latter likely representing diagenetically altered remnants of sugars and plant pigments. Morphologically, taphonomically and compositionally, the dark matter can be most conservatively interpreted as phloem sap produced by the same resin-producing tree. This is the first time that fossilized sap has been recognized and characterized. Moreover, the pseudoinclusion-rich, kidney-shaped amber pieces represent fossilized, resin-in-sap-in-resin double emulsions, i.e., polydispersed resin within droplets of sap, the latter being polydispersed in a resin matrix themselves. A resin-in-sap-in-resin double emulsion is comparable with an oil-in-water-in-oil double emulsion, in that the outer continuous phase is hydrophobic, and the phase dispersed within is hydrophilic and, in turn, contains the hydrophobic phase dispersed within. The double emulsion was formed when the ducts through which the phloem sap circulated were affected by tree damage that promoted resin exudation, and variable quantities of phloem sap mixed with fresh resin. As both immiscible liquids became mechanically joined and were intruded into a budding kidney-shaped resin body, a variable number and size of proto-pseudoinclusions (in other words, resin-in-sap emulsion droplets) were created. As resin injection continued, the previously secreted resin layers suffered a ductile deformation that distorted the shape of the resin-in-sap emulsion droplets within. The differential content of such droplets in suspension caused dark-light banding patterns to be formed. While the resin was still soft, the phloem sap embedded in the resin was opportunistically and preferentially consumed by Cretaceous resinicolous fungi. The sugar originally contained in the phloem sap could have polymerized during diagenesis, forming polar molecules of a higher molecular weight than sugars. Diagenetically, an important part of the dissolved ions originally present in the phloem sap were selectively concentrated as aggregates, forming carbonate and phosphate minerals. The data presented here provide definitive evidence for ruling out the microbiological nature of the protist-/microbe-like inclusions, as was previously advocated based on morphological, distributional and ToF-SIMS analyses 16 . Interestingly, although pseudoinclusions have occasionally been described from Cenozoic ambers, the pseudoinclusion-bearing dark layers typical of kidney-shaped amber pieces from Cretaceous ambers worldwide have not. For that reason, it is possible that only small amounts of sap were incorporated into fresh resin by resin-producing plants during the Cenozoic. This could be due to multiple factors affecting resin production and secretion, although that issue is beyond the scope of this study. In any event, the study of the fossilized, resin-in-sap-in-resin double emulsions offers a promising new tool to characterize diverse amber types from different ages and deposits worldwide. Specifically, bringing the knowledge that industry has gained on double emulsions to the paleobiology and geochemistry of organic resins can shed light on many significant subjects. These include the identification of the resinous plant sources, addressing plant physiology in deep time (e.g., responses to pathogen-mediated diseases), inferring paleoenvironmental parameters, determining taphonomic features of the ancient resin production (location, timing, viscosity), or understanding physicochemical transformations during diagenesis.

Materials and Methods
A total of 32 sections, 0.5-1 mm thick, and ranging in size from 2 × 3 mm to 15 × 30 mm were prepared from different amber pieces from the Rábago/El Soplao outcrop. Although these were mostly fragmentary when unearthed, they were attributed to kidney-shaped pieces based on their irregular morphology (not stalactite-or flow-like), surface characteristics 19,22 , and their inner banding, which is consistent with complete kidney-shaped amber pieces (see below). From the total number of sections, 22 were mounted together on four glass slides using epoxy resin, i.e., preparations 18068 (ten sections), 18069 (six sections), 18070 (five sections), and 18711 (one section). A full section of the same thickness (preparation 18067), 12 × 8 cm, was taken from a complete kidney-shaped amber piece and prepared on a glass slide; this sample was previously embedded in epoxy resin for consolidation before cutting. The remaining nine sections, 18071-18076 and 18614-18616, were unmounted. All the studied samples and preparations are deposited at the Museo Geominero (Instituto Geológico y Minero de España, Madrid, Spain).
Light and scanning electron microscopy (SeM). All samples were examined using a Leica DMLP and an Olympus BX51 optical microscopes. The size distribution of pseudoinclusions in two non-deformed layers from two amber sections (mounted on preparation 18068) was calculated in sectors of 1.45 ×1.20 mm using the computer software FIJI for Windows (67 bits; https://fiji.sc/). Gold coated, freshly broken surfaces of samples 18614, 18615 and 18616 were observed using a JEOL 6400 Scanning Electron Microscope (SEM) equipped with an energy-dispersive X-ray microanalyzer and a back-scattered electron detector (BSE) at the Centro Nacional de Microscopía Electrónica (CNME) of the Universidad Complutense de Madrid (UCM).
fourier transform infrared spectroscopy (ft-iR). Two amber fractions from different kidney-shaped amber pieces that had been washed in order to remove surface finger grease and dust were selected: a pseudoinclusion-rich, dark fraction, and pseudoinclusion-void, light fraction. From them, solid amber samples were pulverized and analyzed using the KBr tablet technique (1 wt% sample) on a BRUKER IFS66v spectrometer, at the Servicio Interdepartamental de Investigación (SIdI) of the Universidad Autónoma de Madrid (UAM), Spain. A total of 250 scans were taken to improve the signal to noise ratio in the 7000-550 cm −1 range. The normal resolution was 4 cm −1 .
Micro-Raman Spectroscopy (MRS). Two sections obtained from different kidney-shaped amber pieces and mounted together on preparation 18068 were analyzed. Sections underwent cleaning with distilled water (2020) 10:9751 | https://doi.org/10.1038/s41598-020-66631-4 www.nature.com/scientificreports www.nature.com/scientificreports/ and compressed air to remove surface grease and dust. The micro-Raman spectra of the spot samples were performed with a Thermo-Fischer DXR Raman Microscope. The system consists of an Olympus BX-RLA2 Microscope, a CCD (1024 × 256 pixels) detector, a monitored XY stage, an auto-focus, and microscope objectives Olympus UIS2 series, all controlled with the software OMNIC 1.0 (https://www.thermofisher.com/), and is located at the Laboratorio de Técnicas No Destructivas of the Museo Nacional de Ciencias Naturales (Consejo Superior de Investigaciones Científicas), Madrid. Light at 780 nm of a frequency-doubled solid laser (maximum power 22 mW) was used for excitation. The average spectral resolution in the Raman shift ranging from 100 to 3600 cm −1 was 4 cm −1 , i.e., grating 900 lines/mm and 2 µm spot sizes. The system was operated under OMNIC 1.0 software fitting working conditions such as pinhole aperture of 25 µm, bleaching time 30 s and four exposures of 10 s each on average. Even though a total of eight spectra were obtained from the dark matter within pseudoinclusions (with five pseudoinclusions with diameters close to 300 µm; Fig. 3e), only two spectra from two different inclusions did not show interference by the light spheroids/ellipsoids (amber). The laser substantially deteriorated the sample's surface. electron microprobe analysis (eMp). The sections analyzed with this technique were the same two utilized for the MRS study, plus an additional section (all three mounted on preparation 18068) (Fig. 3e). Sections were polished a few tens of microns to access areas unaltered by the Raman laser and surfaces were cleaned using distilled water. Chemical analyses of amber, pseudoinclusions and mineralized sub-inclusions present within the latter were performed on chrome-coated polished sections using a JEOL JXA 8900 Electron Microprobe (EMP), operating at 15 kV, 20 nA, and 5-1 µm beam diameter, at the CNME (UCM). Detection limits are approximately 1000 ppm for C, 2200 ppm for O, 150 ppm for S, 100 ppm for Ca, 90 ppm for Mg, 120 ppm for K, 100 ppm for Na, 1000 ppm for Fe, 320 ppm for Ti, 150 ppm for P, 130 ppm for Al, and 80 ppm for Cl. Carbon and oxygen were measured using a LDE1 multilayer diffracting crystal, yielding a statistical precision similar to heavier elements. Measurements of N with the same channel were unsuccessful because amber deteriorated rapidly when the voltage necessary for proper measurements was applied. The content in H was estimated by subtracting the average analytic Total from 100. Only 60 out of the 207 EMP measurements taken from the pseudoinclusion constituents (i.e., light spheroids/ellipsoids, dark matter, and mineralized sub-inclusions) were used as the remaining showed interference with one another due to their small size. The amber surface deteriorated rapidly due to the emission of energy produced by the BSE setting.
Laser-scanning confocal microscopy (LScM). The sections analyzed for this technique were the same two utilized for the MRS and EMP studies (mounted on preparation 18068) and came from different kidney-shaped amber pieces. The sections were further polished to access areas unaltered by the previously used techniques, and their surfaces were cleaned again with distilled water. LSCM was used to study the morphology of the pseudoinclusions, the distribution of mineralized sub-inclusions, and to calculate the emission spectra of both the amber matrix and the dark matter. We carried out two experiments at room temperature (20-22 °C) using a Leica SP-2 AOBS Confocal Microscope at the Centro de Apoyo a la Investigación of the UCM: a sample of amber lacking pseudoinclusions (named "only amber experiment") and a sample containing pseudoinclusions but also including amber (named "pseudoinclusions experiment"). For the latter, a 180 × 180 µm area within a large pseudoinclusion (ca. 200 × 300 µm in diameter; Fig. 3e) was selected to obtain the emission wavelengths. Image analysis was performed with the computer software FIJI. Images were taken at different focal depths in two different modes: fluorescence, where the microscope collects the signal that is emitted from the sample through autofluorescence, and reflection, where the microscope collects the laser signal that is reflected by the sample. In fluorescence mode, emission excitation was done using diode lasers operating at 405, 488, 552 and 638 nm (although the latter wavelength was only used for the "pseudoinclusions experiment"). In reflection mode, the diode laser operating at 488 nm and 552 nm was used to distinguish the mineralized sub-inclusions within pseudoinclusions. electrospray ionization mass spectrometry (eSi-MS). The same pseudoinclusion-rich, dark fraction and pseudoinclusion-void, light fraction used in the FT-IR analysis were used here. Mass spectra were obtained using an ultra-high-resolution QTOF instrument (QSTAR Pulsar i, ABSciex) at the SIdI (UAM). The extraction was carried out with water and the samples were dissolved in methanol for their ionization and introduced in the mass spectrometer to 20 ul/min with an infusion pump. Electrospray ionization source in positive mode was used for all the analyses and the parameters were adjusted as follows: capillary voltage 5500 V, focusing Potential 210 V, declustering potential 30 V. Nitrogen was used as nebulizer gas (pressure of 10 Bar) and the scans of MS spectra were conducted in the m/z range of 50 to 2000. For accurate high resolution mass spectrometry (HRMS) internal calibration was performed after analysis. A tandem mass spectrometry (MSMS) fragmentation spectrum of m/z 359.23 peak has been performed to determine parts of the molecule. The accurate masses obtained were processed using the elemental composition calculator incorporated in the Data Analysis Software (ABSciex) (https://sciex.com/products/software/analyst-software).