Ecology

With S1GBM’s characteristics as a global, PLIA-normalised, high-resolution C-band backscatter dataset, a direct validation experiment is not feasible since we lack matching reference backscatter data collected during airborne or ground based radar campaigns. Other existing global mosaics were generated based on different time-spans, polarisations17, frequencies18, or do not share the novel feature of the PLIA-normalisation20.On these grounds, we prefer to assess the characteristics of the S1GBM layers with respect to different land cover types on a global scale, and to incorporate the gained knowledge into an easy-to-use classification algorithm for permanent water bodies (PWB). This simple mapping experiment acts as an example and should on the one hand demonstrate the integrity and quality of the S1GBM mosaics (and document its limitations), and on the other hand, stimulate more advanced applications and ingestion-models by the remote sensing- and the wider user -communities. Our validation of the obtained PWB-map compares—over a representative and diverse set of eight world regions (see Fig. 1b)—the S1GBM mosaic with a reference water body map, as well as with true-colour imagery from the Sentinel-2 optical sensor. This arrangement should also portray the shape and texture of the S1GBM mosaic and help the audience with the interpretation of the SAR imagery, which as stated at the outset, allows a unique view on the Earth’s surface.In the following, 1) we examine in detail the appearance and spatial features of the S1GBM VV- and VH-mosaics over the region of Bordeaux, also investigating the effect of the PLIA-normalisation. Then, 2) we derive the characteristic C-band backscatter signature for global land classes. Finally, 3) we perform the PWB-experiment in eight world regions a) to evaluate the dataset’s integrity, b) to demonstrate its spatial information and exemplify its use, and c) to comment on the S1GBM’s assets and caveats.Detail example BordeauxFigure 2 gives an example of the land cover signal in the S1GBM VH and VV mosaics over Bordeaux, France. Comparing it with the recent PROBA-V-based Land Cover dataset of the Copernicus Global Land Service (CGLS LC10052), several surface features are apparent in the mosaics, including urban areas with varying density in both VV- and VH-channels. In the VH mosaic, a clear discrimination of forest areas (cf. with LC100’s broadleaf in brighter green, needle leaf in darker green) against crops (brighter yellow) and vineyards (darker yellow) is apparent. The cross-polarised VH-backscatter is more sensitive to vegetation-density, -structure, and -status, as multiple scattering between branches and volume scattering increases the share of backscattered microwaves with changed polarisation. Most prominent, in both VH and VV, is the very large contrast between land surfaces and open waters with significant lower backscatter signatures. This is the basis for our PWB-mapping experiment discussed in detail in the subsequent section.We would also like to draw the attention to the spatial detail carried by the S1GBM mosaics, with various features at deca- and hectometric scale shown for example in Fig. 2. For instance, one can see bridges, highways, railways, and airports in the Bordeaux metropolitan area in the south-west corner of the here displayed T1-tile (100 km extent). Also, in the west, from north to south, the shorelines of the Gironde estuary and its downstream rivers are clearly mapped, resolving small islands and narrow straits. Agricultural plots and forest sections may be differentiated especially in the VH mosaic, e.g. with particular structures in the north-west corner. For further exploration, users may visit the open web-based S1GBM viewer51 offering a pan-and-zoom exploration of the full S1GBM VV- and VH-mosaics.Figure 2b,d allows the comparison of the S1GBM VV backscatter mosaic (which underwent the PLIA-normalisation) against the mean of non-normalised Sentinel-1 VV backscatter from the same observation period (not part of the dataset publication; just for comparison). As discussed above, radar backscatter is strongly dependent to PLIA, and hence Sentinel-1 SAR images are subject to the observation geometry defined by the mission’s relative orbit configuration and the overlapping pattern (cf. global map in Fig. 1b). One can clearly see this impact in Fig. 2d, where data from all local orbits are averaged in their native orbit geometry (i.e. mean of σ0 (θro, t), resulting to characteristic linear artefacts of backscatter discontinuities along the limits of the (repeating) orbit footprints. The mini-map of the Bordeaux-T1-tile in Fig. 2d plots the number of input Sentinel-1 scenes, also reflecting the heterogeneous coverage pattern induced by the different number of overlapping relative orbits (from 2 to 4 in this area), each with a different local PLIA-range, generally. Notably, the triangular zone covered by only 2 orbits (yellow, 194 scenes) is a zone that features a PLIA-spread that is not large enough to reliably estimate the local PLIA-slope β. This zone is part of the pixel domain where we applied the static slope value of −0.13 dB/° to the S1GBM mosaic, with a resulting backscatter image that is free from orbit-related artefacts (Fig. 2b). We note that the sections covered by 3 or 4 orbits in this example are normalised with the regular regression slope, letting us conclude that our approach yields a smooth mosaicking impression in areas of mixed coverage density.Backscatter signature analysisDelving into above concept that SAR backscatter characteristics in the S1GBM are determined by land cover, we analysed the backscatter signature for the global land surface for each major land cover class (LCC). We globally aggregated data from the normalised S1GBM VV and VH mosaics per LCC and formed the backscatter distribution within each LCC, allowing the discrimination of typical SAR backscatter signatures for a specific land cover class.Land cover definitionsAs land cover dataset, we selected the above-mentioned PROBA-V-based CGLS LC100 for its full global coverage and the (for global datasets) relatively high spatial resolution with a pixel spacing of 100 m. To allow a fast pixel-by-pixel comparison, we resampled the CGLS LC100 to the Equi7Grid at 10 m using nearest-neighbour-downsampling. After a first inspection of backscatter signatures, we grouped the 23 LCC of the LC100 to 13 major LCC, accounting for the similarity between certain classes: Respective open and closed forest classes were aggregated to evergreen needle leaf forest, evergreen broad leaf forest, deciduous needle leaf forest, and deciduous broad leaf forest, and herbaceous wetland was grouped with herbaceous vegetation. Table 2 lists the main statistics per land cover and the group aggregations.Table 2 Sentinel-1 backscatter statistics per land cover class (LCC) of the CGLS LC100 dataset, mean and standard deviation, for the S1GBM mosaics in VV and VH polarisation.Full size tableC-band backscatter signaturesThe C-band backscatter signatures of our major 13 LCC are plotted for VV- and VH-polarisation as distribution-density-“heatlines” in the upper part of Fig. 3, illustrating the global average backscatter levels of each surface class, and the variance within. Forest and water-body classes have a very narrow distribution, whereas snow and ice and bare vegetation have a greater spatial backscatter variability. Snow and ice packs often have a heterogeneous structure from its complex genesis involving melting and freezing phases, leading to a mixture of surface- and volume-scattering when observed by radar. Likewise, the LCC bare vegetation comprises very different surfaces dominated by rocky, sandy, or mountainous surfaces, each governed by a distinct backscatter behaviour and hence create the wide spread within this LCC.Fig. 3Results from the S1GBM C-band backscatter signature analysis for major land cover classes, which are provided by the 100 m Land Cover Version 2.0 product of CGLS. The heatlines in (a) and (b) show the S1GBM’s normalised backscatter distribution within the total area of each major land cover class, for VV and VH, respectively. In preparation for the mapping of permanent water bodies (PWB), (c) and (d) show the distributions for the globally combined water- and land- surfaces, with the combined classes indicated by blue and brown bars in (a) and (b) legends. For the PWB-mapping, three land cover classes have been excluded due to the lack of clear separability against the water classes, i.e. due to largely overlapping distributions. The selected thresholds for VV and VH mosaics used in our PWB-mapping algorithm are indicated as red lines.Full size imageThe LCC-heatlines in Fig. 3a,b are approximately ordered by the mean backscatter value. On top, one can find the two water LCCs with a very low backscatter level that is caused by mirror-like-reflection away from the sensor, followed by bare and herbaceous vegetation LCCs that are dominated by dry conditions and hence are generally weak scatterer. The LCCs moss & lichen, shrubs, and agriculture feature medium backscatter and variation thereof. Higher backscatter levels are observed over the forest LCCs, where volume and multiple scattering become more dominant, as well as over the LCC urban & built up, where corner reflections acting as echo cause the strongest radar backscatter.When comparing VV and VH polarisation, the biggest difference is in the overall level of backscatter, with about 7 dB between both polarisations across all LCCs. The order of LCCs as a function of mean backscatter is mostly the same for VV and VH, except for the water and ice classes. Interestingly, the open sea class shows a steeper drop from VV to VH, whereas shrubs show a comparatively small drop. We found that the strongest changes in the backscatter distributions are apparent in the non-forest vegetation classes, e.g. for bare vegetation and agriculture, supporting our initial assumptions on the sensitivity of Sentinel-1 VH backscatter to complex vegetation dynamics and crop varieties.Permanent water body mappingFollowing up to what we have already seen along the rivers in Fig. 2, water bodies (represented by the LCCs open sea and permanent water bodies) show a most distinctive backscatter signature in relation to other land cover classes (cf. 3a-b). Effectively, water surfaces show in radar images a strong contrast with land surfaces. The reason for this are the different microwave scattering mechanism over water- and land-surfaces and the side-looking geometry of SAR systems. A specular reflection of the radar pulses by the water surfaces leads to backscatter intensities received by the sensor that are much lower than for most other land cover types. With the S1GBM VV- and VH-mosaics at hand, we exploited this discriminative feature of water bodies and employed a simple permanent water body mapping method. Unlike the backscatter mosaics of the S1GBM, the obtained PWB map can be validated directly, as we have available matching global water body maps as a reference. Moreover, the experiment should demonstrate the ease of realising a land cover mapping application in short time, exploiting the novel S1GBM data and its high-resolution radar imagery of the Earth’s land surfaces.Based on above insights from the Sentinel-1 backscatter signature analysis, our first step was to spatially merge all water- and all land-LCCs and build the combined backscatter signatures for VV and VH (Fig. 3c,d). The water distribution (all water classes; bright blue) is plotted for both polarisations next to the non-water distribution (all land classes, bright brown), already demonstrating an acceptable feature separation. However, as one can see in the heatlines above, water has still some significant overlap with some land LCCs, e.g. with bare vegetation, herbaceous vegetation, and moss & lichen. Naturally, this translates to a considerable overlap in the merged distributions below, especially in the VH case and for moss & lichen. We concluded that for these LCCs no robust separability against water bodies is given in the S1GBM data and excluded the three classes from further PWB-mapping. Also, we dropped the LCC open sea in further processing as we limit the PWB experiment to inland surfaces (that are also covered by the reference dataset). The backscatter distributions of the PWB LCC and the selected land LCCs are shown in dark blue and brown (permanent water bodies and selected land classes in Fig. 3c,d), with a noticeably improved separability, especially in VH polarisation.As a next step, evoking the theory of Bayesian inference with equal priors for binary classification, we obtained a statistically optimal global threshold for VV and VH, each. In this respect, we identified two thresholds, −15.0 dB for VV and −22.9 dB for VH polarisation, which we applied in a third step as an upper-bound backscatter-value on the complete S1GBM mosaics to map the global PWBs. Note again that the LCCs bare vegetation, herbaceous vegetation, moss & lichen, and open sea are not included in the PWB-mapping and are masked in all later results.Although the VV and VH mosaics are redundant to some degree, the consideration of both channels is most advantageous for the PWB-mapping. First, the classification based on Bayesian inference is more robust when resulting from two discriminations. Second, while the VH mosaic offers a better separability between water and non-water (having less overlap in the distributions and hence less false positive and negative classifications), and the heatline of the PWB-LCC is better defined in VH, the VV mosaic offers in general a higher spatial detail due to its stronger backscatter signal and hence more favourable signal-to-noise ratio.By applying the obtained thresholds to the normalised S1GBM mosaics as simple classification rules$${sigma }_{0}^{{rm{VV}}}(38)le -15.0;{rm{dB}}$$
(2)
$${sigma }_{0}^{{rm{VH}}}(38)le -22.9,{rm{dB}}$$
(3)
and through joining them with logical “AND”, we were able to produce a global PWB map in less than two hours, using 70 parallel cores on the VSC-3 supercomputer.Evaluation of S1GBM mosaics and PWB mapTo evaluate our S1GBM permanent water body (PWB) map, we chose as a reference dataset the Global Surface Water (GWS23) from the European Commission’s Joint Research Centre (JRC-EC). The GSW offers globally at a 30 m native sampling different variables on water bodies, e.g. annual seasonality, occurrence, recurrence, or maximum extent, and is based on 36 years of Landsat data in its newest version (GSW1_2). Although the annual seasonality for 2015 or 2016 was not accessible from version GSW1_2 at the time of writing this manuscript, we found the Seasonality 2015 dataset of the GSW1_0 version suitable as a reference. Pixels valued with seasonality “12” (i.e. all months) are labelled permanent water and constitute our reference PWB map, which we warped by means of bilinear resampling to the Equi7Grid at a 10 m pixel spacing.The evaluation presented in this paper was carried out on a representative and diverse set of eight world regions (see locations in Fig. 1b). For each region, classification results were assessed by a pixel-by-pixel comparison between the PWB map from S1GBM and from the GSW reference. Having such binary maps (water vs. non-water) it was straightforward to generate an “accuracy layer” representing the four elements of the commonly used confusion matrix, i.e. true positives, false positives, false negatives, and true negatives, to discuss the skill of the S1GBM to map PWBs. Areas belonging to the four excluded LCCs were masked in the result plots. Furthermore, to give some visual guidance in the evaluation regions, we acquired from the Copernicus Sentinel-2 Global Mosaic (S2GM) service the RGB-composite for the year 201953 (the mosaic for 2015 was available only over Europe).In the following, we present results for four large-scale regions (500 km × 500 km) in Fig. 4, and for four small-scale regions (120 km × 120 km) in Fig. 5. For each region, the S1GBM VV mosaic is displayed on the left panel (space-saving/omitting the VH mosaic, which contributes likewise to the PWB mapping), the accuracy maps showing the performance against the GSW reference in the centre panel, and the Sentinel-2 RGB-composite to aid visual interpretation on the right panel. The accuracy maps are annotated with the respective User’s Accuracy (UA) and Producer’s Accuracy (PA), as the percentage of the agreement between the two PWB-maps.Fig. 4For four example sites at the large scale (500 km extent), the S1GBM VV mosaic (left) is contrasted with classification results from the S1GBM PWB mapping against the PWB taken from JRC Global Surface Water (GSW) in 2015 (centre), and with the RGB-composite of the Copernicus Sentinel-2 Global Mosaic (S2GM) for the year 2019 (right). Box outlines are shown in global overview in Fig. 1b.Full size imageFig. 5For four detailed example sites (120 km extent), the S1GBM VV mosaic (left) is contrasted with classification results from the S1GBM PWB mapping against the PWB taken from JRC GSW in 2015 (centre), and with the RGB-composite of the Copernicus S2GM for the year 2019 (right). Box outlines are shown in global overview in Fig. 1b.Full size imageLarge-scale examinationsFigure 4a–c shows the southern part of Finland, an area accommodating a multitude of small and large post-glacial lakes. Those are clearly visible in dark colours representing low backscatter values in the S1GBM mosaic, while the other parts of the country (which is dominated by vast forests) shows rather uniform medium backscatter. The optical RGB-composite from Sentinel-2 does not feature the same accentuation of the lakes, troubled by remainders of cloud coverage in the yearly mosaic. The PWB accuracy map shows perfect agreement between S1GBM and GSW, with an UA and PA of 100% each. We identified two reasons for the excellent performance: First, the C-band backscatter signatures of the predominant land covers in Finland, such as forests, cities, agriculture, are well distinguishable against water bodies and hence allow an almost sterile PWB-mapping. Second, northern Europe is well covered by the Sentinel-1 mission and the S1GBM has been built with a high data density, letting us expect the best mosaic quality.Moving to the region of the Lake Superior Basin in Canada and USA presented in Fig. 4d–f, we encounter a very similar, cold-temperate environment, but with a substantial higher share in spacious inland water bodies. Also here, the accuracy map shows a perfect agreement between S1GBM and GSW, which, in our interpretation, is clearly because of good feature separability in the SAR image. Particularly remarkable is that North America is much less covered by Sentinel-1 than Europe and that the imperfect modelling of the PLIA-dependency over water surfaces (as apparent e.g. in the east section of Lake Superior) does not impair the S1GBM PWB-mapping. Generally, imperfect PLIA-normalisation of SAR images is prominent over water bodies, whose specular reflection regime is characterised by a very strong PLIA-gradient (i.e. the slope β). However, we note that also the Sentinel-2 mosaic has striping artefacts bound to orbit footprints, and additionally suffers from cloud cover. The latter is a common problem in optical observation of higher latitudes, but is without effect in SAR imagery.Figure 4g–i depicts the situation for a section of the Albertine Rift Valley in eastern Africa with its lake system. Reflecting to a great deal the region’s diverse flora, which is displayed in many green and brown tones in the RGB-composite, the S1GBM VV mosaic shows a much more heterogeneous pattern than in the above examples. The forested sections in the west show distinct higher backscatter values than the savanna sections in the east, and also other geomorphological features correspond well with the radar and optical mosaic. Concerning the PWB-mapping, we see again perfect agreement, but with one large exception: the eastern end of Lake Albert is entirely labelled in red as false land, suggesting that these water areas are missed in the S1GBM PWB map (what can be confirmed after a quick check with common thematic maps). In this area we see the impact of the relatively poor input data density of about only 50 Sentinel-1 scenes (cf. Figure 1b), and apparently, we overlooked the impact of a few images with outlying backscatter levels during the manual quality curation. Moreover, the three Sentinel-1 relative orbits covering this area create almost identical viewing angles and yield a very small PLIA-range, troubling our backscatter normalisation. As a result, striping artefacts appear not only over water bodies (cf. Canada example) but also over land (in north-west part Fig. 4g), while, however, the Sentinel-2 mosaic is likewise affected by striping issues (cf. Figure 4i), for other reasons, though.The last row in Fig. 4(j–m) is centred at Bangladesh and displays the confluence of the Ganges and Brahmaputra streams, which are joined downstream by the Meghna river and ultimately discharge into the Bay of Bengal. Also in this region, the geomorphological features perceivable in the RGB-composite are reflected well by strong textural patterns in the S1GBM mosaic, promoting its broader use in land cover applications (note also the zoom-in plotted in Fig. 5j–m). The PWB-mapping results are inconclusive, as rivers of all sizes are correctly mapped, but many pixels are labelled in yellow as false waters. We consider this disagreement between S1GBM and GSW to be most likely a result of the different temporal resolutions of the two datasets, as the S1GBM is a two-year data aggregation reduced to single layers, whereas the GSW allows monthly snapshots of water bodies. For example, the Hoar ecosystem—which appears as yellow bulb in the north-east of Fig. 4k—is a large monsoon-fed lagoon system that is labelled by the GSW with seasonality-values ranging from 9 to 12 months. In the S1GBM mosaics, which are built using temporally averaged backscatter, these areas are obviously dominated by the high occurrence of water surfaces and act therefore as “most-of-the-time water bodies”. Some more vindication comes from the Sentinel-2 yearly mosaic, which also draws the Hoar area with a water texture. We conclude on this matter that seasonal water bodies are not properly modelled by our simple approach with Eq. 3, and it would need additional inputs from variance measures like the backscatter standard deviation.Small-scale examinationsFigure 5 depicts the small-scale example regions with respect to the PWB-mapping experiment. The first row in a-c) zooms to the Swiss lakes in central Europe and both, the radar and the optical mosaic, feature a high level of heterogeneity and detail, with many individual forests, cities, valleys, rivers, alpine lakes, and with the airport north of Zurich resolvable (in the centre-left of the box). The results from the PWB-mapping are very good with high UA- and PA-values, but with two anomalies: First, the southern arm of Lake Lucerne (in the south-west) shows some red segments of false land along the mountain flanks reaching into the lake. After inspection of the S1GBM mosaics we can state that this is clearly an artefact from the terrain modelling with the rather coarse, 90 m-sampled VFP SRTM Digital Elevation Model (DEM) during the Sentinel-1 preprocessing. At the time of the project, we selected the VFP DEM35 for its complete global coverage and its manually-checked quality, and accepted the coarse resolution (with respect to the 10 m-sampled Sentinel-1 SAR data). The second small anomaly can be found in the Alps in the south of the image, with the west-end of the Klöntalersee labelled in yellow as false water. The S1GBM is artefact-free at this location, and after checking the GSW’s seasonality, we hypothesise that ice covers this mountain lake during winters and leads to the different interpretation.Figure 5d–f presents the area around the confluence of the Amazon and Tapajós streams in central Pará in Brazil. Here, the rivers ramify into a multitude of lagoons and channels at various sizes, forming a complex system of water bodies. Fortunately, while the Sentinel-2’s RGB mosaic appears impure and rugged from contamination with the frequent cloud coverage in the central tropics, the Sentinel-1 mosaic offers a clear image that fully resolves the capillary structure of the water bodies and its shorelines. We consider this a remarkable feature, also recognising the very low input data density of the S1GBM mosaics in this area (cf. Figure 1b). Concerning the PWB-mapping, we obtained a good agreement with the GSW’s reference, labelling most PWBs correctly and misclassifying only small sections of the lagoons and river-arms. The false-water deviations are bound again to the seasonality of those segments that are most of the time under water, much alike to the situation in Bangladesh discussed around Fig. 4j–m. The red-labelled areas highlight water bodies which are mapped by the GSW but not by the S1GBM, and are of particular interest, as they exemplify that water surfaces seen by optical sensors are not necessarily identical to those seen by radars54. Swamp-like structures and waters with out-growing vegetation show a completely different SAR signature and hence might be distinguishable from open waters within a SAR image.The third small-scale example is the Great Salt Lake in Utah, USA, as displayed in Fig. 5g–i. The S1GBM offers many details of Salt Lake City’s structures in the south-east, and of the mining facilities at the eastern shorelines of the lake, as also visible in the RGB-composite. Obviously, the radar image does not account for the difference in salinity between the north- and south-section of the Great Salt Lake that is visible in the optical image. However, our S1GBM PWB method maps correctly—contrary to the GSW reference—the east-west rail causeway splitting the lake, which one can see as a red line in the accuracy map in Fig. 5h. With its pronounced semi-arid climate, this region shows a different behaviour than above examples. The dry conditions and the sparse vegetation with its weak scattering trouble seriously the S1GBM PWB-mapping, with many false water pixel all around the area. Here, we see the weak performance of the simple threshold approach with Eq. 3 in regions with a general low backscatter from land, and hence small contrast to water bodies.Figure 5j–m zooms into the Sundarbans at the southern shorelines of Bangladesh, with its multifaceted surface and its complex river-deltas. Both, the true-colour image from Sentinel-2 and the VV-mosaic from Sentinel-1 produce a feature-rich image and highlight the mangrove forest in the southern section with strong green colour or high backscatter, respectively. Adjacent to the north, the rice and bean agriculture draws large contrast patterns in the satellite images. For the PWB-mapping, a similar result as from the larger view on this region (cf. Figure 4k) is obtained, with all rivers and channels correctly classified, but with a substantial overestimation of permanent water bodies in areas of high water seasonality. To what extent rice fields and its managed inundations play a role here is left unanswered by the data, though, as managed rice fields typically show significant jumps in seasonal backscatter time series. More

Population genomicsDNA was sourced from fin clips or gill tissue sampled from 223 individuals of Siganus fuscescens from 2013 to 2017. From the northwest to the southwest of Australia, 40 individuals were sampled from the Kimberley, 36 from the Pilbara, nine from Exmouth Gulf, seven from Coral Bay, 40 from Shark Bay, 51 from Cockburn Sound, and 40 from Wanneroo Reef (Supplementary Data 3). However, following quality filtering of these DNA sequences, three rabbitfish individuals were excluded (see below), resulting in 220 rabbitfish individuals used in all remaining analyses (Supplementary Table S4). These tissue samples were extracted using the DNeasy Blood & Tissue Kit (Qiagen, Germany) based on a modified protocol, which included an in-house binding buffer, 1.4× volume of both wash buffers, and a partial automation of the extractions on a QIAcube (Qiagen) platform to minimize human handling and cross-contamination.SNP genotyping was conducted using the DArTseq protocol at the Diversity Arrays Technology (University of Canberra, Australia), which is a reduced representation genomic library preparation method that uses two restriction enzymes46,47. Genomic DNA was digested with the enzymes PstI–SphI and PstI–NspI and small fragments (0.75) or rare (allele frequency 1% and those 620. OTUs not assigned to bacterial or eukaryotic kingdoms were removed from the dataset and the accuracy of taxonomic assignment was assessed through the use of Australian databases for marine flora and diatoms25,26. This resulted in a table containing 86 OTUs, but we only retained OTUs with at least 10 read sequences given that these are less likely to be erroneous sequences that can arise from index-tag jumping. These 78 OTUs—used in downstream statistical analyses—corresponded to cyanobacteria (Cyanophyceae), unknown Eukaryota, dinoflagellates (Dinophyceae), diatoms (Coscinodiscophyceae and Fragilariophyceae), microalgae (microscopic algae of cell size ≤20 µm including Cryptophyceae, Haptophyceae, Mediophyceae, and Chlorarachniophyceae), green macroalgae (Chlorophyta with cell size >20 µm), red macroalgae (Rhodophyta with cell size >20 µm), and brown macroalgae (Ochrophyta with cell size >20 µm) and were represented by silhouettes from PhyloPic (http://phylopic.org/about/) on Figs. 4 and 5, and Supplementary Fig. S2. We then calculated the relative abundance of the 78 OTUs (based on the total number of sequence reads from each individual stomach content, which was visualized in the figure) using a circular plot that was generated with the R-package Circlize57. We also represented the 30% most abundant OTUs across all stomach content samples with a heatmap using a Bray–Curtis distance matrix, which was computed with the R-package phyloseq73 (Supplementary Fig. S2).To investigate differences in stomach contents between tropical residents and vagrants to temperate environments, we performed a non-metric multidimensional scaling ordination (nMDS) in two dimensions based on the Bray–Curtis dissimilarity of individuals. The nMDS plot, whose stress value was 0.12, was plotted using the R-package ggplot274. To further test the dissimilarity in diet composition among tropical residents and temperate vagrants, a permutational analysis of variance (PERMANOVA) was conducted on the same distance matrix with 100,000 permutations. We also tested the homogeneity of group dispersions using the PERMDISP2 procedure with 100,000 permutations as well. The nMDS plot, PERMANOVA, and PERMDISP2 were done with the R-package Vegan60. Finally, to highlight food sources that were unique or significantly associated to a single region or a combination of regions, we used the indicator species (IndVal) analysis in the R-package Indicspecies75 with 100,000 permutations and a significance level corrected with the Benjamini and Hochberg (BH) method76 (Supplementary Data 1 and 2). Significant results were illustrated using colored Venn diagrams on Fig. 5.The 23S rRNA sequence of the kelp species, Ecklonia radiata, from the Western Australian region was not available in the NCBI database, and so three samples were collected in November 2018 at Dunsborough (southwest Australia) and their DNA was extracted with the Miniplant Kit (Qiagen) according to manufacturer’s instructions. Prior to extraction, kelp tissues were rinsed with a continuous flow of tap water for 30 min, then soaked in a solution of 70% ethanol, and finally thoroughly rinsed with Milli-Q water. Tissues were also bead-bashed twice with the Tissue Lyzer II (Qiagen) for 30 s on each cycle. The optimal yield of template DNA was estimated with qPCR following the same method as described above. Each kelp sample was prepared for single‐step fusion‐tag library build using unique index tags following the methods of DiBattista et al.77 and pooled to form an equimolar library. Size selection was also conducted with a Pippin Prep instrument using the same size range as above, and cleaning was done with QIAQuick PCR purification kit (Qiagen). Final libraries were quantified using a Qubit 4.0 Fluorometer (Invitrogen) and sequenced on the Illumina Miseq platform using 500 cycles and V2 chemistry (for paired-end sequencing).Paired-end reads were stitched together using the Illumina Miseq analysis software (MiSeq Reporter V. 2.5) under the default settings. Sequences were assigned to samples using MID tag combinations in Geneious v.10.2.6 and reads strictly matching the MID tags, sequencing adapters, and template-specific primers were retained. Each of the three samples was dereplicated into unique sequences. The unique sequence with the highest number of reads (86,000–120,000) was identical in the three samples, and it did not match any 23S rRNA gene sequences available in the NCBI database based on BLASTn. This sequence was thus designated the 23S rRNA voucher sequence of Ecklonia radiata from southwestern Australia, blasted against all OTUs found in the stomach of rabbitfish individuals in this study, and deposited on GenBank (accession number MW752516).Past and current observations, and climate modelsHistorical sea surface temperature (SST) data were acquired from two sources, each with different temporal coverage and spatial resolution. The present-day (2008–2017) and 1900–1909 SST climatologies were calculated from HadISST78, which is resolved monthly and at 1° spatially. Additionally, the National Oceanic and Atmospheric Administration (NOAA) Coral Reef Watch “CoralTemp v1.0” (daily and 5-km resolution)79 was used to assess SST anomalies during the 2011 marine heatwave.Historical and projected SST data were extracted from outputs of a suite of Coupled Model Intercomparison Project Phase 5 (CMIP5) models. We used the monthly-resolution SST model outputs that included historical greenhouse gas (Historical GHG), and representative concentration pathways of 4.5 and 8.5 W m−2 forcings (“RCP4.5” and “RCP8.5”) runs of the r1i1p1 (designation of initial conditions) ensemble member80. These models included ACCESS, CanESM, CMCC, CNRM, CSIRO, GFDL, GISS-E2-H, INMCM, MIROC, MRI, and NorESM80. The model SST data for each run (historical GHG, RCP4.5, and RCP8.5) were converted to anomalies relative to a 2008–2017 base period, and these anomalies were added to the HadISST 2008–2017 climatology. This analysis was conducted separately for both mean annual and minimum monthly mean (MiMM). Finally, we calculated ensemble means by averaging the SST anomalies from the 11 models. Ensemble means are plotted in Fig. 1 as decadal averages (thick lines) and decadal ranges (shading) of the mean annual 20 °C contour and the MiMM 17 °C contour. The historical GHG run is used to compare the observed and GHG-forced rates of warming between 1900–1909 and 2018–2017, while the two RCP runs are used to project future (2090–2099) SST scenarios. The observed 1900–1909 contours (from HadISST) fall within the ranges of those from the CMIP5 historical GHG ensembles, indicating that anthropogenic emissions are responsible for warming in this region over the past century.Surface ocean currents during the 2011 heatwave were assessed using Simple Ocean Data Assimilation (SODA) v.3.3.181, a state-of-the-art ocean model constrained by observations when and where they are available. We calculated the near-surface (0–25 m) current anomalies (relative to 1980–2015 mean) for the austral summer (January, February, March, or “JFM”) of 2011, which was the peak of the 2010–2011 Western Australia marine heatwave7. These current anomalies are plotted on top of SST anomalies in Fig. 1b. All climate analyses were performed in MATLAB2012b.Reporting summaryFurther information on research design is available in the Nature Research Reporting Summary linked to this article. More

1.Holcapek, M., Liebisch, G. & Elcroos, K. Lipidomic analysis. Anal. Chem. 90, 4249–4257. https://doi.org/10.1021/acs.analchem.7b05395 (2018).CAS 
Article 

Google Scholar 
2.Schwudke, D., Shevchenko, A., Hoffmann, N. & Ahrends, R. Lipidomics informatics for life-science. J. Biotech. 261, 131–136. https://doi.org/10.1016/j.jbiotec.2017.08.010 (2017).CAS 
Article 

Google Scholar 
3.Hsu, F. F. Mass spectrometry-based shotgun lipidomics: A critical review from the technical point of view. Anal. Bioanal. Chem. 410, 6387–6409. https://doi.org/10.1007/s00216-018-1252-y (2018).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
4.Hu, T. & Zhang, J. L. Mass-spectrometry-based lipidomics. J. Separ. Sci. 41, 351–372. https://doi.org/10.1002/jssc.201700709 (2018).CAS 
Article 

Google Scholar 
5.Tang, C. H., Lin, C. Y., Tsai, Y. L., Lee, S. H. & Wang, W. H. Lipidomics as a diagnostic tool of the metabolic and physiological state of managed whales: A correlation study of systemic metabolism. Zoo. Biol. 37, 440–451. https://doi.org/10.1002/zoo.21452 (2018).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
6.Li, X. B. et al. Targeted lipidomics profiling of marine phospholipids from different resources by UPLC-Q-Exactive Orbitrap/MS approach. J. Chromatog. B 1096, 107–112. https://doi.org/10.1016/j.jchromb.2018.08.018 (2018).CAS 
Article 

Google Scholar 
7.Monroig, O., Tocher, D. R. & Navarro, J. C. Biosynthesis of polyunsaturated fatty acids in marine invertebrates: Recent advances in molecular mechanisms. Mar. Drugs 11, 3998–4018. https://doi.org/10.3390/md11103998 (2013).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
8.Kharlamenko, V. I. & Odintsova, N. A. Unusual methylene-interrupted polyunsaturated fatty acids of abyssal and hadal invertebrates. Prog. Oceanog. 178, 102132. https://doi.org/10.1016/j.pocean.2019.102132 (2019).Article 

Google Scholar 
9.Rezanka, T., Kolouchova, I., Gharwalova, L., Palyzova, A. & Sigler, K. Lipidomic analysis: From Archaea to mammals. Lipids 53, 5–25. https://doi.org/10.1002/lipd.12001 (2018).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
10.Lavarias, S., Dreon, M. S., Pollero, R. J. & Heras, H. Changes in phosphatidylcholine molecular species in the shrimp Macrobrachium borellii in response to a water-soluble fraction of petroleum. Lipids 40, 487–494. https://doi.org/10.1007/s11745-005-1408-y (2005).CAS 
Article 
PubMed 

Google Scholar 
11.Miniadis-Meimaroglou, S., Kora, L. & Sinanogiou, V. J. Isolation and identification of phospholipid molecular species in a wild marine shrimp Penaeus kerathurus muscle and cephalothorax. Chem. Phys. Lipids 152, 104–112. https://doi.org/10.1016/j.chemphyslip.2008.01.003 (2008).CAS 
Article 
PubMed 

Google Scholar 
12.Huang, M. X. et al. Growth and lipidomic responses of juvenile pacific white shrimp Litopenaeus vannamei to low salinity. Front. Physiol. https://doi.org/10.3389/fphys.2019.01087 (2019).Article 
PubMed 
PubMed Central 

Google Scholar 
13.Garofalaki, T. F., Miniadis-Meimaroglou, S. & Sinanoglou, V. J. Main phospholipids and their fatty acid composition in muscle and cephalothorax of the edible Mediterranean crustacean Palinurus vulgaris (spiny lobster). Chem. Phys. Lipids 140, 55–65. https://doi.org/10.1016/j.chemphyslip.2006.01.006 (2006).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
14.Rey, F. et al. Unravelling polar lipids dynamics during embryonic development of two sympatric brachyuran crabs (Carcinus maenas and Necora puber) using lipidomics. Sci. Rep. 5, 14549. https://doi.org/10.1038/srep14549 (2015).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
15.Rey, F., Alves, E., Domingues, P., Domingues, M. R. M. & Calado, R. A lipidomic perspective on the embryogenesis of two commercially important crabs, Carcinus maenas and Necora puber. Bull. Mar. Sci. 94, 1395–1411. https://doi.org/10.5343/bms.2017.1140 (2018).Article 

Google Scholar 
16.de Souza, L. M. et al. Glyco- and sphingophosphonolipids from the medusa Phyllorhiza punctata: NMR and ESI-MS/MS fingerprints. Chem. Phys. Lipids 145, 85–96 (2007).Article 
PubMed 
PubMed Central 

Google Scholar 
17.Zhu, S. et al. Lipid profile in different parts of edible jellyfish Rhopilema esculentum. J. Agric. Food Chem. 63, 8283–8291. https://doi.org/10.1021/acs.jafc.5b03145 (2015).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
18.Kostetsky, E. Y., Sanina, N. M. & Velansky, P. V. The thermotropic behavior and major molecular species composition of the phospholipids of echinoderms. Russ. J. Mar. Biol. 40, 131–139. https://doi.org/10.1134/s1063074014020059 (2014).CAS 
Article 

Google Scholar 
19.Yin, F. W. et al. Identification of glycerophospholipid molecular species of mussel (Mytilus edulis) lipids by high-performance liquid chromatography-electrospray ionization-tandem mass spectrometry. Food Chem. 213, 344–351. https://doi.org/10.1016/j.foodchem.2016.06.094 (2016).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
20.Chen, Q. S. et al. Mechanism of phospholipid hydrolysis for oyster Crassostrea plicatula phospholipids during storage using shotgun lipidomics. Lipids 52, 1045–1058. https://doi.org/10.1007/s11745-017-4305-7 (2017).CAS 
Article 
PubMed 

Google Scholar 
21.Liu, Z. Y. et al. Characterization of glycerophospholipid molecular species in six species of edible clams by high-performance liquid chromatography-electrospray ionization-tandem mass spectrometry. Food Chem. 219, 419–427. https://doi.org/10.1016/j.foodchem.2016.09.160 (2017).CAS 
Article 
PubMed 

Google Scholar 
22.Chan, C. Y. & Wang, W. X. A lipidomic approach to understand copper resilience in oyster Crassostrea hongkongensis. Aquatic Toxicol. 204, 160–170. https://doi.org/10.1016/j.aquatox.2018.09.011 (2018).CAS 
Article 

Google Scholar 
23.Facchini, L., Losito, I., Cataldi, T. R. I. & Palmisano, F. Seasonal variations in the profile of main phospholipids in Mytilus galloprovincialis mussels: A study by hydrophilic interaction liquid chromatography-electrospray ionization Fourier transform mass spectrometry. J. Mass Spectrom. 53, 1–20. https://doi.org/10.1002/jms.4029 (2018).ADS 
CAS 
Article 
PubMed 

Google Scholar 
24.Tran, Q. T. et al. Fatty acid, lipid classes and phospholipid molecular species composition of the marine clam Meretrix lyrata (Sowerby 1851) from Cua Lo Beach, Nghe An Province, Vietnam. Molecules https://doi.org/10.3390/molecules24050895 (2019).Article 
PubMed 
PubMed Central 

Google Scholar 
25.Wu, Z. X. et al. Lipid profile and glycerophospholipid molecular species in two species of edible razor clams Sinonovacula constricta and Solen gouldi. Lipids 54, 347–356. https://doi.org/10.1002/lipd.12153 (2019).CAS 
Article 
PubMed 

Google Scholar 
26.Zhang, Y. Y. et al. Evaluation of lipid profile in different tissues of Japanese abalone Haliotis discus hannai Ino with UPLC-ESI-Q-TOF-MS-based lipidomic study. Food Chem. 265, 49–56. https://doi.org/10.1016/j.foodchem.2018.05.077 (2018).CAS 
Article 
PubMed 

Google Scholar 
27.Rey, F. et al. Coping with starvation: Contrasting lipidomic dynamics in the cells of two sacoglossan sea slugs incorporating stolen plastids from the same macroalga. Integr. Comp. Biol. 60, 43–56. https://doi.org/10.1093/icb/icaa019 (2020).CAS 
Article 
PubMed 

Google Scholar 
28.Imbs, A. B. & Grigorchuk, V. P. Lipidomic study of the influence of dietary fatty acids on structural lipids of cold-water nudibranch molluscs. Sci. Rep. 9, 20013. https://doi.org/10.1038/s41598-019-56746-8 (2019).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
29.Gold, D. A. et al. Lipidomics of the sea sponge Amphimedon queenslandica and implication for biomarker geochemistry. Geobiol. 15, 836–843. https://doi.org/10.1111/gbi.12253 (2017).CAS 
Article 

Google Scholar 
30.Imbs, A. B., Dang, L. P. T. & Nguyen, K. B. Comparative lipidomic analysis of phospholipids of hydrocorals and corals from tropical and cold-water regions. PLoS ONE 14, e0215759. https://doi.org/10.1371/journal.pone.0215759 (2019).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
31.Imbs, A. B., Ermolenko, E. V., Grigorchuk, V. P. & Dang, L. T. P. Seasonal variation in the lipidome of two species of Millepora hydrocorals from Vietnam coastal waters (the South China Sea). Coral Reefs 40, 719–734. https://doi.org/10.1007/s00338-021-02073-2 (2021).Article 

Google Scholar 
32.Sogin, E. M., Putnam, H. M., Anderson, P. E. & Gates, R. D. Metabolomic signatures of increases in temperature and ocean acidification from the reef-building coral, Pocillopora damicornis. Metabolomics 12, 71 (2016).Article 

Google Scholar 
33.Tang, C. H., Lin, C. Y., Lee, S. H. & Wang, W. H. Membrane lipid profiles of coral responded to zinc oxide nanoparticle-induced perturbations on the cellular membrane. Aquatic Toxicol. 187, 72–81. https://doi.org/10.1016/j.aquatox.2017.03.021 (2017).CAS 
Article 

Google Scholar 
34.Tang, C. H., Shi, S. H., Lin, C. Y., Li, H. H. & Wang, W. H. Using lipidomic methodology to characterize coral response to herbicide contamination and develop an early biomonitoring model. Sci. Total Environ. 648, 1275–1283. https://doi.org/10.1016/j.scitotenv.2018.08.296 (2019).ADS 
CAS 
Article 

Google Scholar 
35.Imbs, A. B., Dang, L. P. T., Rybin, V. G., Nguyen, N. T. & Pham, L. Q. Distribution of very-long-chain fatty acids between molecular species of different phospholipid classes of two soft corals. Biochem. Anal. Biochem. 4, 205. https://doi.org/10.4172/2161-1009.1000205 (2015).CAS 
Article 

Google Scholar 
36.Imbs, A. B. & Dang, L. T. P. The molecular species of phospholipids of the cold-water soft coral Gersemia rubiformis (Ehrenberg, 1834) (Alcyonacea, Nephtheidae). Russ. J. Mar. Biol. 43, 239–244. https://doi.org/10.1134/s1063074017030051 (2017).CAS 
Article 

Google Scholar 
37.Sikorskaya, T. V. & Imbs, A. B. Study of total lipidome of the Sinularia siaesensis soft coral. Russ. J. Bioorg. Chem. 44, 712–723. https://doi.org/10.1134/S1068162019010151 (2018).CAS 
Article 

Google Scholar 
38.Sikorskaya, T. V. Investigation of the total lipidoma from a zoantharia Palythoa sp. Chem. Nat. Comp. 56, 44–49. https://doi.org/10.1007/s10600-020-02940-4 (2020).CAS 
Article 

Google Scholar 
39.Garrett, T. A., Hwang, J., Schmeitzel, J. L. & Schwarz, J. Lipidomics of Aiptasia pallida and Symbiodinium: A model system for investigating the molecular basis of coral symbiosis. Faseb J. 25, 9382. https://doi.org/10.1096/fasebj.25.1_supplement.938.2 (2011).Article 

Google Scholar 
40.Schmeitzel, J. L., Klein, J., Smith, M., Schwarz, J. & Garrett, T. A. Comparative lipidomic analysis of the symbiosis between Aiptasia pallida and Symbiodinium. FASEB J. 26, 7891. https://doi.org/10.1096/fasebj.26.1_supplement.789.1 (2012).Article 

Google Scholar 
41.Garrett, T. A., Schmeitzel, J. L., Klein, J. A., Hwang, J. J. & Schwarz, J. A. Comparative lipid profiling of the cnidarian Aiptasia pallida and its dinoflagellate symbiont. PLoS ONE 8, e57975. https://doi.org/10.1371/journal.pone.0057975 (2013).ADS 
CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
42.Bosh, T. V. & Long, P. Q. A Comparison of the composition of wax ester molecular species of different coral groups (Subclasses Hexacorallia and Octocorallia). Russ. J. Mar. Biol. 43, 471–478. https://doi.org/10.1134/s1063074017060049 (2017).CAS 
Article 

Google Scholar 
43.Sogin, E. Development and application of metabolomics for reef-building corals Ph.D. – Zoology thesis, University of Hawaii at Manoa (2015).44.Sikorskaya, T. V., Ermolenko, E. V. & Imbs, A. B. Effect of experimental thermal stress on lipidomes of the soft coral Sinularia sp. and its symbiotic dinoflagellates. J. Exp. Mar. Biol. Ecol. 524, 151295. https://doi.org/10.1016/j.jembe.2019.151295 (2020).Article 

Google Scholar 
45.Sikorskaya, T. V. & Imbs, A. B. Coral lipidomes and their changes during coral bleaching. Russ. J. Bioorg. Chem. 46, 643–656. https://doi.org/10.1134/s1068162020050234 (2020).CAS 
Article 

Google Scholar 
46.Rosset, S. et al. Lipidome analysis of Symbiodiniaceae reveals possible mechanisms of heat stress tolerance in reef coral symbionts. Coral Reefs 38, 1241–1253. https://doi.org/10.1007/s00338-019-01865-x (2019).ADS 
Article 

Google Scholar 
47.Imbs, A. B. & Chernyshev, A. V. Tracing of lipid markers of soft corals in a polar lipidome of the nudibranch mollusk Tritonia tetraquetra from the Sea of Okhotsk. Polar Biol. 42, 245–256. https://doi.org/10.1007/s00300-018-2418-y (2019).Article 

Google Scholar 
48.Imbs, A. B., Latyshev, N. A., Dautova, T. N. & Latypov, Y. Y. Distribution of lipids and fatty acids in corals by their taxonomic position and presence of zooxanthellae. Mar. Ecol. Prog. Ser. 409, 65–75. https://doi.org/10.3354/meps08622 (2010).ADS 
CAS 
Article 

Google Scholar 
49.McIntyre, T. M., Snyder, F. & Marathe, G. K. Ether-linked lipids and their bioactive species. In Biochemistry of Lipids, Lipoproteins and Membranes (eds Vance, D. E. & Vance, J. E.) 245–276 (Elsevier, 2008).Chapter 

Google Scholar 
50.Vance, J. E. Phospholipid synthesis and transport in mammalian cells. Traffic 16, 1–18. https://doi.org/10.1111/tra.12230 (2015).CAS 
Article 
PubMed 

Google Scholar 
51.Sundahl, H., Buhl-Mortensen, P. & Buhl-Mortensen, L. Distribution and suitable habitat of the cold-water corals Lophelia pertusa, Paragorgia arborea, and Primnoa resedaeformis on the Norwegian continental shelf. Front. Mar. Sci. 7, 22. https://doi.org/10.3389/fmars.2020.00213 (2020).Article 

Google Scholar 
52.Vysotskii, M. V. & Svetashev, V. I. Identification, isolation and characterization of tetracosapolyenoic acids in lipids of marine coelenterates. Biochim. Biophys. Acta 1083, 161–165. https://doi.org/10.1016/0005-2760(91)90037-I (1991).CAS 
Article 
PubMed 

Google Scholar 
53.Imbs, A. B., Demidkova, D. A. & Dautova, T. N. Lipids and fatty acids of cold-water soft corals and hydrocorals: A comparison with tropical species and implications for coral nutrition. Mar. Biol. 163, 202. https://doi.org/10.1007/s00227-016-2974-z (2016).CAS 
Article 

Google Scholar 
54.Magnusson, C. D. & Haraldsson, G. G. Ether lipids. Chem. Phys. Lipids 164, 315–340 (2011).CAS 
Article 
PubMed 

Google Scholar 
55.Mann, J. & Skeaff, M. Triacylglycerols in Encyclopedia of Life Sciences, 1–9 (Nature Publishing Group, 2001).56.Imbs, A. B., Yakovleva, I. M., Latyshev, N. A. & Pham, L. Q. Biosynthesis of polyunsaturated fatty acids in zooxanthellae and polyps of corals. Russ. J. Mar. Biol. 36, 452–457. https://doi.org/10.1134/S1063074010060076 (2010).CAS 
Article 

Google Scholar 
57.Treignier, C., Tolosa, I., Grover, R., Reynaud, S. & Ferrier-Pages, C. Carbon isotope composition of fatty acids and sterols in the scleractinian coral Turbinaria reniformis: Effect of light and feeding. Limnol. Oceanog. 54, 1933–1940 (2009).ADS 
CAS 
Article 

Google Scholar 
58.Kabeya, N. et al. Genes for de novo biosynthesis of omega-3 polyunsaturated fatty acids are widespread in animals. Science Advances 4, eaar6849, https://doi.org/10.1126/sciadv.aar6849 (2018).59.Rybin, V. G., Imbs, A. B., Demidkova, D. A. & Ermolenko, E. V. Identification of molecular species of monoalkyldiacylglycerol from the squid Berryteuthis magister using liquid chromatography–APCI high-resolution mass spectrometry. Chem. Phys. Lipids 202, 55–61 (2017).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
60.Imbs, A. B., Dang, L. P. T., Rybin, V. G. & Svetashev, V. I. Fatty acid, lipid class, and phospholipid molecular species composition of the soft coral Xenia sp. (Nha Trang Bay, the South China Sea, Vietnam). Lipids 50, 575–589. https://doi.org/10.1007/s11745-015-4021-0 (2015).CAS 
Article 

Google Scholar 
61.Folch, J., Lees, M. & Sloane Stanley, G. H. A simple method for the isolation and purification of total lipides from animal tissues. J. Biol. Chem. 226, 497–509 (1957).CAS 
Article 
PubMed 

Google Scholar 
62.Svetashev, V. I. Mild method for preparation of 4,4-dimethyloxazoline derivatives of polyunsaturated fatty acids for GC–MS. Lipids 46, 463–467. https://doi.org/10.1007/s11745-011-3550-4 (2011).CAS 
Article 
PubMed 

Google Scholar 
63.Brügger, B. Lipidomics: Analysis of the lipid composition of cells and subcellular organelles by electrospray ionization mass spectrometry in Annual Review of Biochemistry (ed. Kornberg, R. D.). Vol. 83, 79–98 (Annual Reviews, 2014).64.Wagner, S. & Richling, E. LC-ESI-MS determination of phospholipids and lysophospholipids. Chromatographia 72, 659–664. https://doi.org/10.1365/s10337-010-1698-3 (2010).CAS 
Article 

Google Scholar 
65.Wang, R. et al. Identification of ceramide 2-aminoethylphosphonate molecular species from different aquatic products by NPLC/Q-Exactive-MS. Food Chem. 304, 10. https://doi.org/10.1016/j.foodchem.2019.125425 (2020).CAS 
Article 

Google Scholar 
66.Hsu, F. F. & Turk, J. Electrospray ionization with low-energy collisionally activated dissociation tandem mass spectrometry of glycerophospholipids: Mechanisms of fragmentation and structural characterization. J. Chromatog. B 877, 2673–2695. https://doi.org/10.1016/j.jchromb.2009.02.033 (2009).CAS 
Article 

Google Scholar  More

1.IPCC Climate Change 2007: Impacts, Adaptation and Vulnerability (eds Parry, M. L. et al.) (Cambridge Univ. Press, 2007).2.Kiehl, J. Lessons from Earth’s past. Science 331, 158–159 (2011).CAS 

Google Scholar 
3.Parmesan, C. Ecological and evolutionary responses to recent climate change. Annu. Rev. Ecol. Evol. Syst. 37, 637–669 (2006).
Google Scholar 
4.Davis, M. B. Range shifts and adaptive responses to quaternary climate change. Science 292, 673–679 (2001).CAS 

Google Scholar 
5.Chen, I.-C., Hill, J. K., Ohlemuller, R., Roy, D. B. & Thomas, C. D. Rapid range shifts of species associated with high levels of climate warming. Science 333, 1024–1026 (2011).CAS 

Google Scholar 
6.Williams, J. W. & Jackson, S. T. Novel climates, no-analog communities, and ecological surprises. Front. Ecol. Environ. 5, 475–482 (2007).
Google Scholar 
7.Root, T. L. et al. Fingerprints of global warming on wild animals and plants. Nature 421, 57–60 (2003).CAS 

Google Scholar 
8.Diez, J. M. et al. Forecasting phenology: from species variability to community patterns. Ecol. Lett. 15, 545–553 (2012).
Google Scholar 
9.Aubin, I. et al. Traits to stay, traits to move: a review of functional traits to assess sensitivity and adaptive capacity of temperate and boreal trees to climate change. Environ. Rev. 24, 164–186 (2016).
Google Scholar 
10.Moran, E. V., Hartig, F. & Bell, D. M. Intraspecific trait variation across scales: implications for understanding global change responses. Glob. Change Biol. 22, 137–150 (2016).
Google Scholar 
11.Hilborn, R., Quinn, T. P., Schindler, D. E. & Rogers, D. E. Biocomplexity and fisheries sustainability. Proc. Natl Acad. Sci. USA 100, 6564–6568 (2003).CAS 

Google Scholar 
12.Greene, C. M., Hall, J. E., Guilbault, K. R. & Quinn, T. P. Improved viability of populations with diverse life-history portfolios. Biol. Lett. https://doi.org/10.1098/rsbl.2009.0780 (2010).13.Moore, J. W., Yeakel, J. D., Peard, D., Lough, J. & Beere, M. Life-history diversity and its importance to population stability and persistence of a migratory fish: steelhead in two large North American watersheds. J. Anim. Ecol. 83, 1035–1046 (2014).
Google Scholar 
14.Fagan, W. F. Connectivity, fragmentation, and extinction risk in dendritic metapopulations. Ecology 83, 3243–3249 (2002).
Google Scholar 
15.Fausch, K. D., Torgersen, C. E., Baxter, C. V. & Li, H. W. Landscapes to riverscapes: bridging the gap between research and conservation of stream fishes. BioScience 52, 483–498 (2002).
Google Scholar 
16.Comte, L. & Grenouillet, G. Do stream fish track climate change? Assessing distribution shifts in recent decades. Ecography 36, 1236–1246 (2013).
Google Scholar 
17.Troia, M. J., Kaz, A. L., Niemeyer, J. C. & Giam, X. Species traits and reduced habitat suitability limit efficacy of climate change refugia in streams. Nat. Ecol. Evol. 3, 1321–1330 (2019).
Google Scholar 
18.Beechie, T., Buhle, E., Ruckelshaus, M., Fullerton, A. & Holsinger, L. Hydrologic regime and the conservation of salmon life history diversity. Biol. Conserv. 130, 560–572 (2006).
Google Scholar 
19.Barbarossa, V. et al. Impacts of current and future large dams on the geographic range connectivity of freshwater fish worldwide. Proc. Natl Acad. Sci. USA 117, 3648–3655 (2020).CAS 

Google Scholar 
20.FitzGerald, A. M., John, S. N., Apgar, T. M., Mantua, N. J. & Martin, B. T. Quantifying thermal exposure for migratory riverine species: phenology of Chinook salmon populations predicts thermal stress. Glob. Change Biol. 27, 536–549 (2021).
Google Scholar 
21.Schindler, D. E. et al. Population diversity and the portfolio effect in an exploited species. Nature 465, 609–612 (2010).CAS 

Google Scholar 
22.Brennan, S. R. et al. Shifting habitat mosaics and fish production across river basins. Science 364, 783–786 (2019).CAS 

Google Scholar 
23.Crozier, L. G. et al. Climate vulnerability assessment for Pacific salmon and steelhead in the California current large marine ecosystem. PLoS ONE 14, e0217711 (2019).CAS 

Google Scholar 
24.Yoshiyama, R. M., Fisher, F. W. & Moyle, P. B. Historical abundance and decline of Chinook salmon in the Central Valley region of California. North Am. J. Fish. Manag. 18, 487–521 (1998).
Google Scholar 
25.Gustafson, R. G. et al. Pacific salmon extinctions: quantifying lost and remaining diversity. Conserv. Biol. 21, 1009–1020 (2007).
Google Scholar 
26.McClure, M. M. et al. Evolutionary consequences of habitat loss for Pacific anadromous salmonids: salmonid habitat loss and evolution. Evol. Appl. 1, 300–318 (2008).
Google Scholar 
27.Quinn, T. P. The Behavior and Ecology of Pacific Salmon and Trout (Univ. Washington Press, 2018).28.Yoshiyama, R. M., Gerstung, E. R., Fisher, F. W. & Moyle, P. B. in Contributions to the Biology of Central Valley Salmonids (ed. Brown, R. L.) 71–176 (California Department of Fish and Game, 2001).29.Metcalfe, N. B. & Thorpe, J. E. Determinants of geographical variation in the age of seaward-migrating salmon, Salmo salar. J. Anim. Ecol. 59, 135–145 (1990).
Google Scholar 
30.Moyle, P. B., Lusardi, R. A., Samuel, P. & Katz, J. State of the Salmonids: Status of California’s Emblematic Fishes 2017 (Center for Watershed Sciences, 2017); https://doi.org/10.13140/RG.2.2.24893.9776131.Hedgecock, D. Microsatellite DNA for the Management and Protection of California’s Central Valley Chinook Salmon (Oncorhynchus tshawytscha) Final Report for the Amendment to Agreement No. B-59638 (Univ. of California, 2002).32.Woodson, L. et al. Size, growth, and origin-dependent mortality of juvenile Chinook salmon Oncorhynchus tshawytscha during early ocean residence. Mar. Ecol. Prog. Ser. 487, 163–175 (2013).
Google Scholar 
33.Satterthwaite, W. et al. Match-mismatch dynamics and the relationship between ocean-entry timing and relative ocean recoveries of Central Valley fall run Chinook salmon. Mar. Ecol. Prog. Ser. 511, 237–248 (2014).
Google Scholar 
34.Johnson, M. R. & Merrick, K. Juvenile Salmonid Monitoring Using Rotary Screw Traps in Deer Creek and Mill Creek, Tehama County, California Summary Report: 1994–2010 RBFO Technical Report No. 04-2012 (California Department of Fish and Wildlife, 2012).35.Beckman, B. R., Larsen, D. A., Lee-Pawlak, B. & Dickhoff, W. W. Relation of fish size and growth rate to migration of spring Chinook salmon smolts. North Am. J. Fish. Manag. 18, 537–546 (1998).
Google Scholar 
36.Myrick, C. A. & Cech, J. J. Temperature Effects on Chinook Salmon and Steelhead Bay-Delta Modeling Forum Technical Publication 01-1 (Bay-Delta Modeling Forum, 2001).37.Cogliati, K. M., Unrein, J. R., Stewart, H. A., Schreck, C. B. & Noakes, D. L. G. Egg size and emergence timing affect morphology and behavior in juvenile Chinook salmon. Oncorhynchus tshawytscha. Ecol. Evol. 8, 778–789 (2018).
Google Scholar 
38.Richter, A. & Kolmes, S. A. Maximum temperature limits for Chinook, coho, and chum salmon, and steelhead trout in the Pacific northwest. Rev. Fish. Sci. 13, 23–49 (2005).39.Johnson, R. C. & Lindley, S. T. in Viability Assessment for Pacific Salmon and Steelhead Listed Under the Endangered Species Act: Southwest NOAA Technical Memorandum NMFS-SWFSC-564 (eds Williams, T. H. et al.) 48–63 (U.S. Department of Commerce, 2016).40.Dettinger, M. D., Ralph, F. M., Das, T., Neiman, P. J. & Cayan, D. R. Atmospheric rivers, floods and the water resources of California. Water 3, 445–478 (2011).
Google Scholar 
41.Frölicher, T. L., Fischer, E. M. & Gruber, N. Marine heatwaves under global warming. Nature 560, 360–364 (2018).
Google Scholar 
42.Ullrich, P. A. et al. California’s drought of the future: a midcentury recreation of the exceptional conditions of 2012–2017. Earth’s Future 6, 1568–1587 (2018).CAS 

Google Scholar 
43.Beamish, R. J. & Mahnken, C. A critical size and period hypothesis to explain natural regulation of salmon abundance and the linkage to climate and climate change. Prog. Oceanogr. 49, 423–437 (2001).
Google Scholar 
44.Sturrock, A. M. et al. Reconstructing the migratory behavior and long-term survivorship of juvenile Chinook salmon under contrasting hydrologic regimes. PLoS ONE 10, e0122380 (2015).
Google Scholar 
45.Michel, C. J., Notch, J. J., Cordoleani, F., Ammann, A. J. & Danner, E. M. Nonlinear survival of imperiled fish informs managed flows in a highly modified river. Ecosphere 12, e03498 (2021).
Google Scholar 
46.Isaak, D. J., Young, M. K., Nagel, D. E., Horan, D. L. & Groce, M. C. The cold-water climate shield: delineating refugia for preserving salmonid fishes through the 21st century. Glob. Change Biol. 21, 2540–2553 (2015).
Google Scholar 
47.Ebersole, J. L., Quiñones, R. M., Clements, S. & Letcher, B. H. Managing climate refugia for freshwater fishes under an expanding human footprint. Front. Ecol. Environ. 18, 271–280 (2020).
Google Scholar 
48.Klemetsen, A. et al. Atlantic salmon Salmo salar L., brown trout Salmo trutta L. and Arctic charr Salvelinus alpinus (L.): a review of aspects of their life histories. Ecol. Freshw. Fish. 12, 1–59 (2003).
Google Scholar 
49.Kovach, R. P., Gharrett, A. J. & Tallmon, D. A. Genetic change for earlier migration timing in a pink salmon population. Proc. R. Soc. B 279, 3870–3878 (2012).
Google Scholar 
50.Miettinen, A. et al. A large wild salmon stock shows genetic and life history differentiation within, but not between, rivers. Conserv. Genet. 22, 35–51 (2021).CAS 

Google Scholar 
51.Birnie-Gauvin, K. et al. Life-history strategies in salmonids: the role of physiology and its consequences. Biol. Rev. 96, 2304–2320 (2021).
Google Scholar 
52.Seebacher, F., White, C. R. & Franklin, C. E. Physiological plasticity increases resilience of ectothermic animals to climate change. Nat. Clim. Change 5, 61–66 (2015).
Google Scholar 
53.Jonsson, B. & Jonsson, N. A review of the likely effects of climate change on anadromous Atlantic salmon Salmo salar and brown trout Salmo trutta, with particular reference to water temperature and flow. J. Fish. Biol. 75, 2381–2447 (2009).CAS 

Google Scholar 
54.Zillig, K. W., Lusardi, R. A., Moyle, P. B. & Fangue, N. A. One size does not fit all: variation in thermal eco-physiology among Pacific salmonids. Rev. Fish. Biol. Fish. 31, 95–114 (2021).
Google Scholar 
55.Thorstad, E. B. et al. Atlantic salmon in a rapidly changing environment-facing the challenges of reduced marine survival and climate change. Aquat. Conserv. 31, 2654–2665 (2021).
Google Scholar 
56.5-Year Review: Summary and Evaluation of Central Valley Spring-run Chinook Salmon Evolutionarily Significant Unit (National Marine Fisheries Service, 2016); https://repository.library.noaa.gov/view/noaa/1701857.Barnett-Johnson, R., Grimes, C. B., Royer, C. F. & Donohoe, C. J. Identifying the contribution of wild and hatchery Chinook salmon (Oncorhynchus tshawytscha) to the ocean fishery using otolith microstructure as natural tags. Can. J. Fish. Aquat. Sci. 64, 1683–1692 (2007).
Google Scholar 
58.Barnett-Johnson, R., Ramos, F. C., Grimes, C. B. & MacFarlane, R. B. Validation of Sr isotopes in otoliths by laser ablation multicollector inductively coupled plasma mass spectrometry (LA-MC-ICPMS): opening avenues in fisheries science applications. Can. J. Fish. Aquat. Sci. 62, 2425–2430 (2005).CAS 

Google Scholar 
59.Barnett-Johnson, R., Pearson, T. E., Ramos, F. C., Grimes, C. B. & MacFarlane, R. B. Tracking natal origins of salmon using isotopes, otoliths, and landscape geology. Limnol. Oceanogr. 53, 1633–1642 (2008).CAS 

Google Scholar 
60.Hobson, K. A., Barnett-Johnson, R. & Cerling, T. in Isoscapes (eds West, J. B. et al.) 273–298 (Springer, 2010).61.Ingram, B. L. & Weber, P. K. Salmon origin in California’s Sacramento–San Joaquin river system as determined by otolith strontium isotopic composition. Geology 27, 851–854 (1999).
Google Scholar 
62.Phillis, C. C., Sturrock, A. M., Johnson, R. C. & Weber, P. K. Endangered winter-run Chinook salmon rely on diverse rearing habitats in a highly altered landscape. Biol. Conserv. 217, 358–362 (2018).
Google Scholar 
63.Fielding, A. Cluster and Classification Techniques for the Biosciences (Cambridge Univ. Press, 2007).64.Ramsay, J. & Silverman, B. W. Functional Data Analysis (Springer-Verlag, 1997).65.R Core Team R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2017).66.Fraley, C. & Raftery, A. E. Model-based clustering, discriminant analysis, and density estimation. J. Am. Stat. Assoc. 97, 611–631 (2002).
Google Scholar 
67.Scrucca, L., Fop, M., Murphy, T. B. & Raftery, A. E. mclust 5: clustering, classification and density estimation using gaussian finite mixture models. R J. 8, 289–317 (2016).
Google Scholar 
68.Schick, R. S., Edsall, A. L. & Lindley, S. T. Historical and Current Distribution of Pacific Salmonids in the Central Valley, CA NOAA-TM-NMFS-SWFSC-369 (NOAA-NMFS, 2005).69.Isaak, D. J. et al. The NorWeST summer stream temperature model and scenarios for the western U.S.: a crowd-sourced database and new geospatial tools foster a user community and predict broad climate warming of rivers and streams. Water Resour. Res. 53, 9181–9205 (2017).
Google Scholar 
70.Bjornn, T. C. & Reiser, D. W. in Influences of Forest and Rangeland Management on Salmonid Fishes and Their Habitats Special Publication 19 (ed. Meehan, W. R.) 83–138 (American Fisheries Society, 1991).71.Cordoleani, F. et al. Threatened salmon rely on a rare life history strategy in a warming landscape. Github https://github.com/floracordoleani/MillDeerOtolithPaper (2021).72.Cordoleani, F. et al. Threatened salmon rely on a rare life history strategy in a warming landscape. Dryad Digital Repository https://doi.org/10.5061/dryad.bk3j9kdc9 (2021). More

1.Quinn, T. P. The Behavior and Ecology of Pacific Salmon and Trout 2nd edn (Univ. Washington Press, 2018).2.Cordoleani, F. et al. Nat. Clim. Change https://doi.org/10.1038/s41558-021-01186-4 (2021).Article 

Google Scholar 
3.Schindler, D. E. et al. Nature 465, 609–612 (2010).CAS 
Article 

Google Scholar 
4.Tilman, D. Ecology 77, 350–363 (1996).Article 

Google Scholar 
5.Norberg, J. et al. Proc. Natl Acad. Sci. USA 98, 11376–11381 (2001).CAS 
Article 

Google Scholar 
6.Moore, J. W., Yeakel, J. D., Peard, D., Lough, J. & Beere, M. J. Anim. Ecol. 83, 1035–1046 (2014).Article 

Google Scholar 
7.Aulus-Giacosa, L., Aymes, J.-C., Gaudin, P. & Vignon, M. Mar. Freshw. Res. 70, 1828–1837 (2019).Article 

Google Scholar 
8.Barnett-Johnson, R., Ramos, F. C., Grimes, C. B. & MacFarlane, R. B. Can. J. Fish. Aquat. Sci. 62, 2425–2430 (2005).CAS 
Article 

Google Scholar 
9.Miller, J. A., Geary, A. & Merz, J. Mar. Ecol. Prog. Ser. 408, 227–240 (2010).Article 

Google Scholar 
10.FitzGerald, A. M., John, S. N., Apgar, T. M., Mantua, N. J. & Martin, B. T. Glob. Change Biol. 27, 536–549 (2021).Article 

Google Scholar  More

Data collectionBased on the cropland extent, we first introduced a cropland distribution template, the Self-adapting Statistics Allocation Model of Global Cropland (SASAM-GC)16, as shown in Fig. S1. The global cropland extent map used herein was a global cropland synergy map with a 500-m spatial resolution representing approximately the year 2010, developed by the Smart Agriculture Innovation Team of the Key Laboratory of Agricultural Remote Sensing (AGRIRS) of the Chinese Academy of Agricultural Sciences in cooperation with the International Food Policy Research Institute (IFPRI) and the International Institute of Applied Systems Analysis (IIASA). The overall accuracy of the SASAM-GC products is 90.8%, which is higher than those of existing global farmland products such as the Climate Change Initiative Land Cover (CCI-LC), GlobeLand30, and Medium Resolution Imaging Spectrometer (MODIS) products.Vegetation indices are often used to depict crop growth, such as the ratio vegetation index (RVI)17, normalized difference vegetation index (NDVI), and enhanced vegetation index (EVI)18. Among them, the EVI is the most sensitive to high-biomass regions and less susceptible to atmospheric and soil interference19,20. MODIS vegetation index datasets are generated every 8 days or 16 days at spatial resolutions of 250 m, 500 m, and 1000 m. The 250-m spatial resolution is considered the best resolution for detecting crops21,22. Here, we used EVI time series with a spatial resolution of 250 m reported every 16 days in the MODIS product MOD13Q1 as the primary data to calculate the global cropping intensity; these data can be accessed at https://lpdaac.usgs.gov/products/mod13q1v006/.$${rm{EVI}}=2.5times frac{{{rm{rho }}}_{{rm{NIR}}}-{{rm{rho }}}_{{rm{Red}}}}{{{rm{rho }}}_{{rm{NIR}}}+6times {{rm{rho }}}_{{rm{Red}}}-7.5times {{rm{rho }}}_{{rm{Blue}}}+1}$$
(1)
In formula 1, ρNIR, ρRed, and ρBlue represent the reflectivity of the near-infrared band, the red band, and the blue band, respectively.The Food and Agricultural Organization of the United Nations statistical data (FAOSTAT) provides long-time-series cropland-related statistical data at the country level and can be accessed at http://www.fao.org/faostat/en/#data. FAOSTAT cropland data have been widely used in a variety of studies evaluating food security and hindcasting historical land-use changes23,24. Here, we defined the cropland area as the sum of areas characterizing arable land (land under temporary crops, temporary meadows used for mowing or pasture, market and kitchen gardens, and land that is temporarily fallow; abandoned land resulting from shifting cultivation was excluded) and permanent croplands (land cultivated with long-term crops that do not have to be replanted for several years). Additionally, the harvested area is used; this value refers to the area from which a crop is gathered. If the crop under consideration is harvested more than once during a year due to successive cropping (i.e., the same crop is sown or planted more than once in the same field during the same year), the area is counted as many times as the crop is harvested. Therefore, the sown area is recorded only for the crop reported. The statistical cropping intensity is defined as the harvested area divided by the cropland area and is used to validate the consistency between the cropping intensity obtained herein and that reported in the FAOSTAT product at the country level.Brief algorithmIn this study, a sixth-order polynomial function was used to reconstruct EVI time series for brief and rapid calculations at the global scale25. As the global cropping intensity ranges from 0 to 3 and the sixth-order polynomial function can have 3 maxima at most, we chose the sixth-order polynomial function (formula 2) in this study:$${rm{EVI}}({rm{t}})={{rm{a}}}_{0}+{{rm{a}}}_{1}{rm{t}}+{{rm{a}}}_{2}{{rm{t}}}^{2}+cdots +{{rm{a}}}_{6}{{rm{t}}}^{6}$$
(2)
where t is the time-series length referring to the beginning of the time series, EVI(t) is the fitted EVI time series, and a0, a1, … an are the fitted parameters of the sixth-order polynomial function. Then, the first derivative EVI′(t) and the second derivative EVI″(t) were calculated to find the real numerical solution of the equation’s maxima when EVI′(t) = 0and EVI″(t)  More

1.Caro, T. The adaptive significance of coloration in mammals. Bioscience 55, 125 (2005).Article 

Google Scholar 
2.Caro, T. The colours of extant mammals. Semin. Cell Dev. Biol. 24, 542–552 (2013).Article 

Google Scholar 
3.Schaller, G. B., Jinchu, H., Wenshi, P. & Jing, Z. The Giant Pandas of Wolong (University of Chicago Press, 1985). https://doi.org/10.1086/414647.Book 

Google Scholar 
4.Schaller, G. B. The Last Panda (University of Chicago Press, 1994).
Google Scholar 
5.Morris, R. & Morris, D. Men and Pandas (McGraw-Hill Book Company, 1966).
Google Scholar 
6.Morris, R. & Morris, D. The Giant Panda (Penguin Books, 1982).
Google Scholar 
7.Lazell, J. D. J. Color Patterns of the ‘Giant’ Bear (Ailuropoda melanoleuca) and the True Panda (Ailurus fulgens) (Mississippi Wildlife Federation, 1974).
Google Scholar 
8.Cott, H. B. Adaptive Coloration in Animals (Methuen & Co., Ltd., 1940).
Google Scholar 
9.Endler, J. A. On the measurement and classification of colour in studies of animal colour patterns. Biol. J. Linn. Soc. 41, 315–352 (1990).Article 

Google Scholar 
10.Stevens, M. & Merilaita, S. Animal camouflage: Current issues and new perspectives. Philos. Trans. R. Soc. B Biol. Sci. 364, 423–427 (2009).Article 

Google Scholar 
11.Caro, T., Walker, H., Rossman, Z., Hendrix, M. & Stankowich, T. Why is the giant panda black and white?. Behav. Ecol. 28, 657–667 (2017).Article 

Google Scholar 
12.Endler, J. A. The color of light in forests and its implications. Ecol. Monogr. 63, 1–27 (1993).Article 

Google Scholar 
13.Merilaita, S. Crypsis through disruptive coloration in an isopod. Proc. R. Soc. B Biol. Sci. 265, 1059–1064 (1998).Article 

Google Scholar 
14.Cuthill, I. C. et al. Disruptive coloration and background pattern matching. Nature 434, 72–74 (2005).ADS 
CAS 
Article 

Google Scholar 
15.Stevens, M. & Merilaita, S. Defining disruptive coloration and distinguishing its functions. Philos. Trans. R. Soc. B Biol. Sci. 364, 481–488 (2009).Article 

Google Scholar 
16.Ruxton, G., Allen, W., Sherratt, T. & Speed, M. Avoiding Attack: The Evolutionary Ecology of Crypsis, Aposematism, and Mimicry (Oxford University Press, 2019).
Google Scholar 
17.Troscianko, J. & Stevens, M. Image calibration and analysis toolbox—A free software suite for objectively measuring reflectance, colour and pattern. Methods Ecol. Evol. 6, 1320–1331 (2015).Article 

Google Scholar 
18.van den Berg, C. P., Troscianko, J., Endler, J. A., Marshall, N. J. & Cheney, K. L. Quantitative colour pattern analysis (QCPA): A comprehensive framework for the analysis of colour patterns in nature. Methods Ecol. Evol. 11, 316–332 (2020).Article 

Google Scholar 
19.Troscianko, J., Skelhorn, J. & Stevens, M. Quantifying camouflage: How to predict detectability from appearance. BMC Evol. Biol. 17, 7 (2017).Article 

Google Scholar 
20.Caves, E. M. & Johnsen, S. AcuityView: An r package for portraying the effects of visual acuity on scenes observed by an animal. Methods Ecol. Evol. 9, 793–797 (2018).Article 

Google Scholar 
21.Marshall, N. J. Communication and camouflage with the same ‘bright’ colours in reef fishes. Philos. Trans. R. Soc. B Biol. Sci. 355, 1243–1248 (2000).CAS 
Article 

Google Scholar 
22.Barnett, J. B., Cuthill, I. C. & Scott-Samuel, N. E. Distance-dependent aposematism and camouflage in the cinnabar moth caterpillar (Tyria jacobaeae, erebidae). R. Soc. Open Sci. 5, 171396 (2018).ADS 
Article 

Google Scholar 
23.Barnett, J. B., Cuthill, I. C. & Scott-Samuel, N. E. Distance-dependent pattern blending can camouflage salient aposematic signals. Proc. R. Soc. B Biol. Sci. 284, 20170128 (2017).Article 

Google Scholar 
24.Stoner, C. J., Caro, T. M. & Graham, C. M. Ecological and behavioral correlates of coloration in artiodactyls: Systematic analyses of conventional hypotheses. Behav. Ecol. 14, 823–840 (2003).Article 

Google Scholar 
25.Caro, T., Walker, H., Santana, S. E. & Stankowich, T. The evolution of anterior coloration in carnivorans. Behav. Ecol. Sociobiol. 71, 177 (2017).Article 

Google Scholar 
26.Melin, A. D., Kline, D. W., Hiramatsu, C. & Caro, T. Zebra stripes through the eyes of their predators, zebras, and humans. PLoS ONE 11, e0145679 (2016).Article 

Google Scholar 
27.Land, M. F. & Nilsson, D.-E. Animal Eyes (Oxford University Press, 2012).Book 

Google Scholar 
28.Phillips, G. A. C., How, M. J., Lange, J. E., Marshall, N. J. & Cheney, K. L. Disruptive colouration in reef fish: Does matching the background reduce predation risk?. J. Exp. Biol. 220, 1962–1974 (2017).Article 

Google Scholar 
29.Li, Y. et al. Giant pandas can discriminate the emotions of human facial pictures. Sci. Rep. 7, 1–8 (2017).ADS 
Article 

Google Scholar 
30.Stevens, M., Párraga, C. A., Cuthill, I. C., Partridge, J. C. & Troscianko, T. S. Using digital photography to study animal coloration. Biol. J. Linn. Soc. 90, 211–237 (2007).Article 

Google Scholar 
31.Lind, O., Milton, I., Andersson, E., Jensen, P. & Roth, L. S. V. High visual acuity revealed in dogs. PLoS ONE 12, 1–12 (2017).
Google Scholar 
32.Pasternak, T. & Merigan, W. H. The luminance dependence of spatial vision in the cat. Vis. Res. 21, 1333–1339 (1981).CAS 
Article 

Google Scholar 
33.Clark, D. L. & Clark, R. A. Neutral point testing of color vision in the domestic cat. Exp. Eye Res. 153, 23–26 (2016).CAS 
Article 

Google Scholar 
34.Caves, E. M., Brandley, N. C. & Johnsen, S. Visual acuity and the evolution of signals. Trends Ecol. Evol. 33, 1–15 (2018).Article 

Google Scholar 
35.Vorobyev, M. & Osorio, D. Receptor noise as a determinant of colour thresholds. Proc. R. Soc. B Biol. Sci. 265, 351–358 (1998).CAS 
Article 

Google Scholar 
36.Nokelainen, O., Brito, J. C., Scott-Samuel, N. E., Valkonen, J. K. & Boratyński, Z. Camouflage accuracy in Sahara-Sahel desert rodents. J. Anim. Ecol. https://doi.org/10.1111/1365-2656.13225 (2020).Article 
PubMed 

Google Scholar 
37.Nokelainen, O., Stevens, M. & Caro, T. Colour polymorphism in the coconut crab (Birgus latro). Evol. Ecol. 32, 75–88 (2018).Article 

Google Scholar 
38.Nokelainen, O., Maynes, R., Mynott, S., Price, N. & Stevens, M. Improved camouflage through ontogenetic colour change confers reduced detection risk in shore crabs. Funct. Ecol. https://doi.org/10.1111/1365-2435.13280 (2019).Article 
PubMed 
PubMed Central 

Google Scholar  More

The arthritic degeneration visualized in the pathological Smilodon specimens could have arisen from one of three etiologies: traumatic, infective or degenerative arthritis. Findings on the specimens make infective or traumatic arthritis less likely. In the case of infective arthritis, the presupposition is that the animal developed typically before an insult that led to infection and subsequent obliteration of the hip joint. This assumption also holds true for the case of traumatic arthritis following an injury or fracture. However, the anatomical distortions of the right femoral head, in conjunction with the obliteration of the right acetabulum, suggest chronic changes that led to degeneration over time (Figs. 3, 4). The degeneration of the femoral head would not be expected if the degenerative change in the hip joint were due to infection or trauma, as the development of the pelvis and femur presumably would have been complete before the insult or injury occurred during the adult cat’s life.The condition of the right acetabulum and right femoral head demonstrates anatomy consistent with developmental distortion. Typically, the head of the femur develops in conjunction with the acetabulum of the pelvis16. The spherical femoral head fits into the concentric-shaped acetabulum to form a ball-and-socket joint that enables a four-legged mammal to ambulate, lie down, sit down, stand up, and generally function normally16. In developmental hip dysplasia, however, the acetabulum does not develop appropriately, and the articulation between the femoral head and acetabulum is lost. An elliptical (as opposed to concentric-shaped) acetabulum causes progressive subluxation (dislocation) of the femoral head17, which can result in coxa plana, or necrosis of the bony nucleus of the femoral head16. This subsequent coxa plana produces flattening and degeneration of the normally spherical femoral head18.Proper anatomical development and ossification of the hip joint rely on continuous and symmetrical pressure of the femoral head on the acetabulum, and dysplasia results from improper positioning of the femoral head within the acetabulum16. Dysplastic hips are characterized by pathological restructuring and accelerated remodeling of the joint in response to abnormal forces and tensions that create stress. This produces formation of new bone in some areas and resorption of bone in others, ultimately causing degenerative joint disease16. Dysplastic hips have varying degrees of deformity and malformation, but typically the acetabula are hypoplastic and deficient in various planes and dimensions (Supplementary Fig. S10).Further inspection of LACMHC 131 demonstrates anatomical changes consistent with chronic degeneration throughout the right hip joint and pelvis. The obliteration of the right innominate likely occurred over many years and progressively resulted in significant bony destruction and remodeling. These findings of a flattened femoral head in LACMHC 6963 in conjunction with a shallow acetabulum in LACMHC 131 are consistent with changes observed with mechanical instability of the hip joint and bony destruction secondary to dysplasia. Repeated subluxation events due to the dysplastic hip likely accelerated the destruction of the cartilage and joint, altering the biomechanical stresses through the joint. This increased stress along with cartilage loss likely led to a progressively hypertrophic and aberrant bone response with subchondral sclerosis and osteophyte formation in the acetabulum and pelvis. The external, anatomical deformities in these specimens are consistent with changes that have occurred over the animal’s lifespan and subsequently resulted in the gross morphology observed, with destruction of the hip joint on both the acetabular and femoral side.This type of pathology starts to impact movement at the time of first walking, although minimal pain tends to ensue at this time because of the animal’s flexibility at its early age19,20. As the joint cartilage wears out, however, bone begins to rub on bone. The resulting forces make the bone stiffer, producing osteophytes or bone spurs as well as sclerosis that manifests on CT imaging as increased bone density (Figs. 3 and 4; Supplementary Video S2; Supplementary Data S1–S2). At this point, loading the limb would cause pain, and range of motion would be limited. Therefore, the animal examined in this study would have spent as little time as possible on its right hindlimb, needing to compensate for the handicap by increasing the load on its left hindlimb. This compensation would explain the exostoses on the left ilium anterodorsal to the non-pathological acetabulum (Fig. 1; Supplementary File S1; Supplementary Video S1), indicating abnormal pulling of the quadriceps femoris muscles originating in this area.Hip dysplasia is a heritable, polygenic condition that affects a range of mammal species16, including humans17. Canine hip dysplasia (CHD) is one of the most prevalent orthopedic diseases in domestic dogs (Canis lupus familiaris)21 and is very well studied, in part because it is similar to developmental dysplasia of the human hip22. Feline hip dysplasia (FHD) has received less clinical attention than CHD, possibly because its functional impairment is less overt or because domestic cats (Felis catus) are able to compensate for the resulting lameness better than dogs20,23. The overall results of physiological changes from dysplasia are mechanical imbalance and instability in the hip joint causing displacement due to opposing forces from the acetabulum and femoral head, and osteophytes in the acetabulum to compensate for cartilage loss16.Embryologically, articular joints differentiate from skeletal mesenchyme in situ with the support of surrounding tissues that sustain mechanical and physiological forces that tend to pull on the joints16. In dogs, hip joints are normal at birth, as teratological factors and mechanical stresses that could displace the femoral head are rare at this time16. Epiphyseal ossification normally begins by 12 days of age; in dogs that eventually develop CHD, anatomical changes of the femoral head and pelvic socket begin before week three24. In dysplastic hips, the teres ligament, which is crucial for holding the femoral head in place, is too short; this produces luxation, or dislocation, of the top of the femoral head, beginning at around seven weeks16. This luxation increases throughout development, degrading the articular cartilage that surrounds the femoral head, delaying ossification of the femur and acetabulum16, and shortening the affected limb, as the femoral head becomes positioned higher in the acetabulum.In clinical reports of hip dysplasia in domestic cats, osteoarthritis (i.e., degenerative joint disease, DJD) of the hip secondary to FHD is well known19. For example, osteoarthritis was recorded in 43 of 45 (95.6%) of cats with FHD25. As well, in 5 of 13 (38.5%) cases of hip osteoarthritis with a radiographically or historically identifiable cause, hip dysplasia was pinpointed as the cause, with the remaining cases resulting from trauma or equivocal between trauma and dysplasia23. A recent study of FHD in Maine Coons—a large-bodied domestic cat breed in which hip dysplasia is known to be common—calculated a prevalence of 37.4%, finding severity to increase with age and body mass26. The same study further highlighted a genetic correlation between FHD and large body size within the Maine Coon26, inviting inquiry into how FHD impacts other breeds and non-domestic felid species across a range of body sizes.Reports of FHD in non-domestic large cats are rarer than in domestic cats. Captive snow leopards have exhibited hip dysplasia; across 14 zoos, seven cases were classified as moderate to severe, and at least two individual snow leopards needed total hip replacement before being able to breed27,28. Accounts of hip functional impairment in other captive large cats have tended to report osteoarthritis, which can be associated with FHD though may also stem from trauma and increased age29,30,31.For wild-caught large cats, the only comprehensive study of which we are aware is a survey of 386 individuals (283 wild-caught) across three felid genera mounted as exhibit skeletons in multiple North American natural history museums30. Though not focusing on hip dysplasia, the study tracked degenerative joint disease, which may be associated with dysplasia23,25. The sample recorded DJD in 9.7% of 31 tigers, 2.3% of 88 African lions, and 5.1% of 59 mountain lions (Puma concolor), and none in five other species of big cat. These frequencies are low compared to domestic cats, perhaps owing to differences in body size, diet, and lifestyle between large wild cats and domestic cats, as well as selective breeding constraining genetic variation in domestic animals. Furthermore, selection against hip dysplasia would be expected in the wild because hip dysplasia would compromise hunting19. Though this study identified instances of non-inflammatory osteoarthritis in the shoulder, elbow, and stifle joint, it found none in the hip. However, 4% of all joints afflicted by spondyloarthropathy—a form of inflammatory arthritis—included the hip30.What is the significance of Smilodon, an extinct Pleistocene predator, having the same congenital defect as living domestic cats and dogs? Previous workers have inferred social behavior from paleopathologies in fossil carnivorans ranging from the extinct Eurasian steppe lion7 to Pleistocene wolf-like canines from Italy8 and China9, interpreting signs of healing as evidence of survival after injury12. Given the severity of many injuries, authors have argued, the animal would have starved to death had it not operated within a social structure. The present hip dysplasia having manifested from a young age—hindering this animal’s ability to hunt prey and defend a home range over the course of its life—supports this assertion, although other inferences are possible.Sociality, the degree to which individuals live with conspecifics in groups32, is difficult to infer in Smilodon given that it has no living analogues or closely related taxa. Estimated to have weighed between 160 and 350 kg (3,14, this study), Smilodon was at least the size of the Amur tiger (Panthera tigris altaica), the largest living cat; some estimates reach 369 to 469 kg, placing Smilodon in the range of the largest extant ursids15,33. No living felid has Smilodon’s elongate, knife-like canines or stocky, powerful build. As well, Smilodon (of the extinct felid lineage Machairodontinae) is only distantly related to extant large felids (Felinae), introducing further uncertainty. Based on its robust morphology (e.g.,34,35) and on evidence from stable isotopes (e.g.,4), it likely stalked and ambushed prey; therefore, it may have been comparable to the African lion (Panthera leo), which has a similar hunting strategy and is the only truly social extant felid36. Yet sociality varies across felid species, including within a genus; for example, other extant pantherines like tigers (P. tigris) show incipient sociality37, while jaguars (P. onca) are solitary except for females with cubs. Social strategies also can vary within species, e.g., between sexes. For instance, African lion females are philopatric and social throughout their lives, while adult males are often nomadic and solitary until joining a gregarious pride, which itself usually lasts for only a few years38. This social variation complicates behavioral inferences based on ancestral reconstructions.Advocates of the solitary-cat hypothesis39,40 have cited Smilodon’s small relative brain size determined using endocranial casts as support for solitary behavior, because sociality exerts high cognitive demands. However, in 39 species across nine carnivoran families, larger relative brain size was found to correlate with problem-solving capabilities rather than social behavior41. Rather than analyses of overall encephalization across carnivoran families, studies of relative regional brain volume within families and species have been more informative regarding sociality42,43. In both African lions and cougars (Puma concolor, a solitary species), total relative endocranial volume was not sexually dimorphic; however, relative anterior cerebrum volume was significantly greater in female African lions than males, a difference absent in cougars38.Though regional endocranial studies have yet to be performed on Smilodon, the gregarious-cat hypothesis has drawn support from multiple lines of evidence. One is the abundance of Smilodon relative to prey at RLB10,11,34, although detractors have pointed out that some extant large cats aggregate at carcasses despite otherwise being solitary40. A full range of ages is present among RLB Smilodon; in contrast, animals interpreted to be solitary, such as the American lion Panthera atrox, are represented largely by adult individuals44. As well, the proportions of social and solitary species at RLB parallel those drawn to audio recordings of herbivore distress calls in the African savanna, suggesting that RLB Smilodon sample sizes are more consistent with it having been social rather than solitary45,46. The lack of size sexual dimorphism in Smilodon is more typical of modern solitary cats47 but could also be reflective of monogamy within a gregarious species, like modern wolves. Most relevant to the current study, the existence of healed injuries in Smilodon also has been interpreted as evidence for social behavior, with the assumption that surviving long after serious injury would be difficult if not impossible without cooperative sociality12. We now revisit this interpretation considering the novel diagnosis of hip dysplasia in this study.Smilodon’s large body size necessitated preying on megaherbivores for adequate sustenance3. To do so, like most large cats today, it would have used its hindlimbs for propulsion and acceleration48,49, a pounce behavior enabled by its morphology. Smilodon’s ratio of total forelimb to hindlimb length is greater while its ratio of tibia to femur length ranks lower than in living felids34. The shorter hindlimbs lacking the distal limb elongation in cursorial animals suggest that Smilodon was an ambush predator surpassing the ability of felids today50. Hunting large prey is dangerous51. After the initial hindlimb-powered leap, Smilodon would have grappled with its struggling prey, as evidenced by traumatic injuries in the rotator cuff and radiating from the ventral midline dorsolaterally to where the ribs articulate with the spine5. As it subdued prey with robust forelimbs35,48 under enough torque to injure the lumbar vertebrae5, Smilodon would have needed to leverage itself against the ground using its hindlimbs. Therefore, the pelvis and femur would have been critical to multiple phases of its hunting strategy.A dysplastic individual would have encountered much difficulty hunting in this manner. Yet, as evidenced by the complete fusion of its pelvic and femoral epiphyses (Figs. 1, 2) as well as its large body size (Figs. 6, 7), the individual in this study had reached adult age. (Studies of the detailed timing of epiphyseal fusion in large wild cats are lacking, but distal femoral epiphyses fuse at around the same time as or soon after proximal femoral epiphyses in domestic cats and dogs52,53. Given this, the broken distal femur likely had a fused epiphysis, as on its intact proximal end.) Limbs in African lions completely fuse between 4.5 and 5.5 years54,55,56, so it is reasonable to assume that adulthood in Smilodon likely started at around four years old. This estimate is reinforced by bone histological work quantifying at least four to seven lines of arrested growth (LAGs; one per growth year) in limb bones with fused epiphyses belonging to Smilodon fatalis from the Talara asphaltic deposits in Peru57. Some LAGs in the Talara histological specimens likely have been masked by secondary bone remodeling, which may be more extensive in larger-bodied taxa57, making these specimens possibly older than the number of visible LAGs suggest. Therefore, four years represents a likely minimum age for this individual, although it could have been much older.Ontogenetic growth patterns in teeth and bone further support inferences of sociality. In Smilodon, teeth appear to mature earlier than when sutures and long-bone epiphyses fuse, suggesting delayed weaning, prolonged juvenile dependence, and extended familial care until the adult hunting morphology—saber canines and robust limbs—was complete47. At RLB, most sampled Smilodon specimens show significant pulp cavity closure of the lower canine (14 of 19 specimens over approximately 80% closure), a sign of dental maturation58. This contrasts with RLB pantherine pulp cavities, which are more evenly distributed across the closure percentage range, suggesting that teeth mature earlier in Smilodon than in pantherines. (Other age assessments have ruled out the possibility that Smilodon juveniles were underrepresented relative to pantherines44). At Talara, age determination by dentition yields low estimates of juveniles (zero based on skulls; 8% based on dentaries), but age determination based on limb epiphyseal fusion yields higher estimates (41% juveniles)57. Histology of Talara Smilodon long bones reinforces this mismatch, as an apparent adult femur with fused epiphyses and seven LAGs was found to lack avascular and acellular subperiosteal lamellar bone57, suggesting that it had not yet finished growing. Further, prolonged parental care was interpreted in a recent description, from Pleistocene deposits in Corralito, Ecuador, of two subadult Smilodon fatalis individuals inferred to have been siblings and associated with an adult that was likely their mother59. This scenario of prolonged parental care, like that in the social African lion, would help explain how the individual in this current study survived to adulthood given its debilitating handicap.Novel application of CT visualization to an old question of paleopathology has enabled diagnosis of hip dysplasia, a lifelong condition, in an individual Smilodon fatalis saber-toothed cat. This individual was likely not the only Smilodon afflicted with hip dysplasia: multiple RLB Smilodon pelvic specimens, especially that described by Shermis11, exhibit gross morphology similar to the pathological pelvis examined in this study (Supplementary Figs. S6–S9). The individual examined in this study reached adulthood (at least four to seven years of age) but could never have hunted nor defended territory on its own, given its locomotor impairment that would have been present since infancy. As such, this individual likely survived to adulthood by association with a social group that assisted it with feeding and protection.Further conclusions are limited by the lack of a comprehensive and systematic comparative dataset comprising pathological post-crania from extant species, a persistent limitation of paleopathological studies5. Natural history museums may acquire cranial remains from zoos or similar institutions but often lack storage to accommodate postcranial skeletons, especially for large mammals. As well, while radiographic studies on domestic cats and dogs illustrate the nature of hip dysplasia, these studies tend to examine pathological bones in situ, still embedded in a muscular framework (e.g., Supplementary Fig. S10). This is opposed to the bones-only, flesh-free context of paleopathological specimens. Computed tomography and digital data may be key to building a comparative paleopathology dataset in the future.Within the scope of this study, we cannot rule out the hypothesis that the pathological animal was a scavenger and may have obtained food outside the context of a social structure. It is also possible that, regardless of its disability, its large size and fearsome canines made it a strong interference competitor. However, the pathological specimens examined here are consistent with the predominance of studies supporting a spectrum of social strategies in this extinct predator. In many extant carnivorans, sociality offers the benefits of cooperative hunting and rearing of young (e.g.,60): benefits that likely also applied to Smilodon in the late Pleistocene. As Smilodon coexisted with a rich megafaunal carnivore community including dire wolves (Aenocyon dirus), American lions (Panthera atrox), and short-faced bears (Arctodus simus), cooperative sociality may have aided its success as a predator in a crowded field. More