Have whales returned to a historical hotspot of industrial whaling? The pattern of southern right whale Eubalaena australis recovery at South Georgia

Around 176 500 whales were killed in the sub-Antarctic waters off South Georgia (South Atlantic) between 1904 and 1965. In recent decades, whales have once again become summer visitors, with the southern right whale (SRW) the most commonly reported species until 2011. Here, we assess the distribution, temporal pattern, health status and likely prey of SRWs in these waters, combining observations from a summertime vessel-based expedition to South Georgia, stable isotope data collected from SRWs and putative prey and sightings reports collated by the South Georgia Museum. The expedition used directional acoustics and visual surveys to localise whales and collected skin biopsies and photo-IDs. During 76 h of visual observation effort over 19 expedition days, SRWs were encountered 15 times (~31 individuals). Photo-IDs, combined with publicly contributed images from commercial vessels, were reconciled and quality-controlled to form a catalogue of 6 fully (i.e. both sides) identified SRWs and 26 SRWs identified by either left or right sides. No photo-ID matches were found with lower-latitude calving grounds, but 3 whales had gull lesions supporting a direct link with Península Valdés, Argentina. The isotopic position of SRWs in the South Georgia food web suggests feeding on a combination of copepod and krill species. Opportunistic reports of SRW sightings and associated group sizes remain steady over time, while humpback whales provide a strong contrast, with increased sighting rates and group sizes seen since 2013. These data suggest a plateau in SRWs and an increasing humpback whale presence in South Georgia waters following the cessation of whaling.


INTRODUCTION
The sub-Antarctic island of South Georgia lies in the high latitudes of the South Atlantic, just south of the Polar Front and within the Antarctic Zone of the Antarctic Circumpolar Current (Fig. 1). To the east of the island, the front of the Southern Antarctic Circum-polar Current is deflected northwards, creating a mixing zone of water masses offshore to the north and east of South Georgia. The shelf waters around the island are characterised by very high productivity of phytoplankton and copepods and are globally recognised as a biodiversity hotspot (Hogg et al. 2011) containing a very large biomass of Antarctic krill Euphau sia superba and other zooplankton (Atkinson et al. 2001). A variety of species feed on krill in this ecosystem, including seals, ceta ceans, seabirds and fish (Trathan & Croxall 2004, Murphy et al. 2007). As early as 1775, when Captain James Cook first landed on the island, whales were anecdotally reported to be abundant near South Georgia. During the subsequent century, South Georgia was the site of significant sealing activity (Headland 1989, Hoffman et al. 2011). It was not, however, until the 20 th century that South Georgia be came the epicentre of modern whaling in the Southern Hemisphere, with ~176 500 whales killed within a day's transit from the coast (Allison 2013; Fig. 2, Table 1). The shore-based whaling stations closed in the 1960s and a moratorium on whaling from factory ships for all species other than minke whales was agreed by the International Whaling Commission (IWC) in 1979. Since then, 2 dedicated investigations into summer cetacean abundance in South Georgia waters have been carried out: a line transect survey in 1997 out to the 2000 m isobath along the north coast (Moore et al. 1999) and a research cruise by Brazilian investigators in 2006(Rossi-Santos et al. 2007. Cetacean observations were also made on other South Georgia expeditions, for exam-ple during dedicated krill surveys (Reid et al. 2000, Hedley et al. 2001, Reilly et al. 2004, Širović et al. 2006, Tarling et al. 2012) and as part of a cetacean and seabird survey along the north Scotia Ridge (Joiris et al. 2015). Most other sightings of whales in South Georgia come from opportunistic sources (tourist, fishery and land-based sightings; Moore et al. 1999, Richardson et al. 2012.
Taken together, the opportunistic sightings and survey data assembled to date suggest that the southern right whale Eubalaena australis (SRW) was the most commonly sighted whale in nearshore waters off South Georgia up to (Moore et al. 1999, Richardson et al. 2012. As with some other baleen whale species, SRWs show a seasonal pattern of summering in high latitude offshore feeding grounds and migrating to mid-latitude coastal wintering grounds in autumn, where mating and calving occurs. In the southwest Atlantic, wintering grounds for SRWs are found along the coast of Argentina, Uruguay and southern Brazil (see Harcourt et al. 2019 for a recent review). There, SRWs were a relatively easy target for small boat coastal whaling because they congregate to give birth and nurse their calves in calm sheltered waters, are slow swimmers and float when dead (IWC 2001  Hunting licences were first issued for the Brazilian SRW wintering grounds in the early 1600s by charter of King Philip III of Spain . This represented a substantial portion of the SRW southwest Atlantic wintering range, which currently spans between Bahia State in Brazil at 12°S (de Morais et al. 2017) and Golfo San Jorge in Argentina at 46°S (Iñiguez et al. 2003). The Brazilian fishery was a Basque whaling style operation that lasted until the 1820s (Reeves & Smith 2006, de Morais et al. 2017. Such shore-based whaling stations were usurped as the primary whaling industry by pelagic North American and British whalers operating in the 18 th and 19 th centuries (Townsend 1935, Tønnessen & Johnsen 1982, IWC 2001, Smith et al. 2012 whale species and were already significantly depleted by the time whaling began at South Georgia. Today, the SRW population in the southwest Atlantic is slowly recovering (with the Argentine calving ground estimated at ~4000 whales in 2010; IWC 2013). Whales are now regularly seen in bays along the coast of Argentina and southern Brazil during winter and spring. Pioneering photo-ID work, based on the unique pattern of callosities on the head of each whale, has enabled identification of individual right whales and the tracking of their activities through time (Payne et al. 1983, Kraus et al. 1986). These photo-ID studies began in Península Valdés, Argentina, in the 1970s and in Brazil in the 1980s (Payne 1986, Groch et al. 2005. The summer foraging destinations of whales wintering in Península Valdés have been investigated using photo-ID, stable isotope and satellite telemetry data, and directly link this calving ground to South Georgia (Best et al. 1993, Moore et al. 1999, Rowntree et al. 2001, Valenzuela et al. 2009, 2018, Zerbini et al. 2018. The stable isotope work led to the hypothesis that whales in the southwest Atlantic use 2 primary offshore feeding strategies during summer: one where they feed on copepods in lower latitude waters of the South Atlantic, including the Patagonian shelf, and another where they feed on Antarctic krill in the high latitudes of the South Atlantic, including South Georgia, the Scotia Arc and the South Sandwich Islands (Valenzuela et al. 2009(Valenzuela et al. , 2018. Feeding habitat quality is thought to affect SRW population dynamics by mediating prey availability. High latitude climate anomalies (e.g. mediated by El Niño and the Southern Annular Mode) impact South Atlan tic krill distribution and density (Murphy et al. 2007) and lead to variations in climate and krill densities that have been associated with reduced calving success (Leaper et al. 2006, Seyboth et al. 2016, suggesting that the feeding conditions at high latitudes affect calving success. Whether the overall signal of calving success within the population holds for lower latitude feeders or whales using a mixed strategy is unclear. The 1997 whale research expedition to South Georgia concluded that 18 yr after the end of commercial whaling at South Georgia, there was no evidence of major concentrations of whales returning to the island (Moore et al. 1999). Twenty-one years later, we conducted a new expedition to South Georgia waters, spanning the expected period of peak SRW occurrence in summer months (January−February 2018). This expedition was designed to investigate SRW genetic diversity, population connectivity with calving areas, health status, prey sources, acoustic behaviour and habitat use in relation to the krill fishing within the sustainable-use South Georgia and South Sandwich Islands Marine Protected Area. Here, we report the first results of this survey, and consider our findings in the context of past catches, recent sightings and the environmental context of whales feeding in the vicinity of South Georgia.

MATERIALS AND METHODS
Our area of focus in this study was the South Georgia marine ecosystem, spanning the coastal and shelf waters around South Georgia and the shelf waters of Shag Rocks to the west of the island (33.5°− 43°W, 53°− 56°S; Fig. 1). The voyage was conducted on the R/V 'Song of the Whale' (owned and managed by Marine Conservation Research 1 ), which departed from the Falkland Islands (Islas Malvinas) on Monday 22 January 2018. The vessel sailed to the north coast of South Georgia, where she remained for 19 d (28 January to 16 February 2018) before returning to the Falkland Islands (Islas Malvinas) on Wednesday 21 February (a total of 31 d sailing). There were 8 researchers and 3 crew members onboard. All data collection was carried out under permit RAP/ 2017/ 017 issued by the Government of South Georgia and the South Sandwich Islands (GSGSSI) following review and approval of all data collection ap proaches by the British Antarctic Survey (BAS) Animal Welfare and Ethics Review Board (review #1040).

Survey effort and image collection
Visual sightings were conducted by the research team, with multiple categories of visual effort based on both sighting conditions and the type of survey being conducted. For much of the time around South Georgia, the vessel track was chosen based on acoustic SRW detections (see Section 2.2). Thus, visual searching effort was based on some prior knowledge of SRW locations. This effort was categorised as 'acoustic bearing'. If no recent information on SRW location was available, effort was classified as 'random searching', with search locations chosen based on assessment of weather conditions and the pattern of prior sightings. With both effort types, visual observations were carried out during daylight hours when the boat was underway, and when conditions allowed, by 2 observers from the deck (3.5 m above sea level) and one on a raised platform (A-frame) near the stern (5.5 m above sea level). Surveys were carried out when the Beaufort sea state was 5 or less and visibility levels were good.
Vessel speed and heading, wind speed and direction, sea state, weather conditions and visibility were recorded using the software Logger 2010 2 . These data were updated every hour, or if the observer, sea state or visibility conditions changed. They were also recorded with every sighting.
When cetaceans were sighted, information on species identity, group size, bearing and distance from the ship were logged. When possible, photographs for species and individual identification (photo-ID) purposes were obtained for all baleen whales sighted (Payne et al. 1983, Kraus et al. 1986, Patenaude & Baker 2001. Specific information on photographs for photo-ID and health assessment of SRWs is described below. Sightings for which species identification was not possible were classified to the lowest taxonomic level possible. Sighting data were plotted using ArcMap 10.4.1 (ESRI), with a high resolution (100 m) marine landscape map produced by Hogg et al. (2016) for background context (see Fig. 3).
Skin and blubber biopsy samples were collected during the survey using small, stainless steel biopsy darts deployed from a crossbow (Lambertsen 1987) or a Paxarm modified veterinary capture system (Krutzen et al. 2002). After a biopsy attempt, darts and/ or samples were retrieved via tether system reel or a dip net. Skin biopsy samples were divided into subsamples for genetics, or stable isotope and hormone analysis, and stored in 95% ethanol or frozen at −80°C, respectively.

Passive acoustic monitoring and tracking of SRWs
Directional frequency analysis and recording (DIFAR) sonobuoys (HIDAR SSQ-955A, Ultra Electronics) were used to acoustically locate and track SRWs in real-time, and to record their vocalisations.
DIFAR sono buoys contain an omnidirectional acoustic pressure sensor and 2 orthogonal acoustic vector sensors that are directional in the horizontal plane (Greene et al. 2004, McDonald 2004. Sonobuoy signals were received by VHF radio onboard the research vessel, digitised, recorded, processed using PAMGuard (Gillespie et al. 2009), and the DIFAR bearings to calls resolved as described by Miller et al. (2015Miller et al. ( , 2016. Continuous recordings were made at a sample rate of 48 000 samples s −1 , and data from all buoys were monitored and processed both visually and aurally by an on-duty acoustician and processed in real-time as they were received for the full duration of each buoy deployment. VHF signals were received using a Procom CXL 2-3LW/s omnidirectional antenna tuned to the 137− 150 MHz frequency band giving a gain of 3 dBd. The antenna was initially mounted at the top of the mast at a height of 33 m, giving an effective reception range to the sonobuoys of around 20 km. Problems with the mast head fitting meant that the antenna had to be moved to a fiberglass pole mounted on the A-frame with a height to the base of the antenna of 9.4 m. This reduced the effective reception range to 12 km. Sonobuoys were deployed in wind speeds of up to 35 knots because background noise levels were considered too high for effective monitoring at higher wind speeds. Sonobuoys were deployed to a depth of either 30 m (in waters shallower than 200 m) or 140 m (in water deeper than 200 m). After each buoy was deployed, it was subsequently monitored with the vessel within VHF range in order to detect calls and their associated bearings. Buoys were usually deployed overnight to allow the vessel to follow bearings to areas where SRWs had been detected before the commencement of visual observations in the morning. Further buoys were deployed and the vessel course altered in order to determine a more precise whale location and enable visual detections. In addition to deploying sonobuoys to search for whales, some buoys were deployed close to whale groups in order to match visual and acoustic observations. Because the vocalisations of SRWs on their feeding grounds are not well described, it was necessary to characterise their acoustic repertoire, and to attempt to differentiate them from humpback whales Megaptera novaeangliae, which are also present near South Georgia. This was achieved when acoustic bearings could be followed until the animals were located visually, or when the acoustic bearings to received calls could be aligned with known whale locations from a visual sighting. Analyses of these data are ongoing.

2
The logger software at www. marineconservationresearch. co. uk was developed by the International Fund for Animal Welfare (IFAW) to promote benign and non-invasive research.

Photo-ID catalogue compilation and matching
Individual adult SRWs can be identified by their unique patterns of head callosities and skin pigmentation (Payne et al. 1983, Kraus et al. 1986). All South Georgia SRW photographs obtained in this study were evaluated for image suitability and were assigned a quality rating from 0 (lowest quality) to 3 (highest quality), following the approach taken by Wade et al. (2011). The ratings were based on the following criteria (listed in order of importance): oblique angle, focus, amount of visible callosity pattern and exposure. Only images of quality 2 or 3 were used for individual identification purposes, yet all images were analysed and retained for health assessment and future photo-ID comparisons.
SRW photographs taken from vessels often only capture the callosity patterns on the left or right sides of the head. Every effort was made in the field and during subsequent image analysis to link these images to an individual animal. All photographed individuals were given an ID number, but only SRWs with high-quality left and right head callosity images are considered definitely unique. Definitely unique animals do not match any other whales identified during this survey, whereas whales with only a left or only a right head photograph may represent the same individual and may appear in the catalogue twice.
Photographs provided from the South Georgia Museum's archive and other platforms of opportunity were collated, and high-quality photo-ID images selected for further analysis. These photographs were compared to the BAS South Georgia photo-ID catalogue by an experienced matcher, and if the head callosity pattern did not match with any other photoidentified whale, this photograph was added to the catalogue as a unique individual. The photo-ID catalogue was also compared against the 12 whales photographed during the 1997 South Georgia expedition (Moore et al. 1999).
The 2018 South Georgia photo-ID catalogue was compared to wintering ground catalogues from (1) the Península Valdés (6887 images, 3255 individuals) using the software package BigFish (Skadia); (2) Brazil (921 unique individuals; Groch 2018); and (3) South Africa (2084 individuals; Best 1990), the latter 2 catalogues using the Hiby-Lovell software system to assist with matching. Cross-catalogue matching was complicated because photographs from wintering grounds were collected from aerial platforms and those from South Georgia were obtained from boats, with the BigFish and Hiby-Lovell systems requiring different amounts of photographic coverage of the head. Consequently, from the South Georgia photo-ID catalogue, 13 individuals could be compared with the Península Valdés catalogue, and 7 individuals compared with the Brazilian catalogue and with the South African catalogue.

Visual health assessment
South Georgia SRW side-on photographs were also analysed to quantify certain aspects of individual health. The complete set of images was scored by one researcher with substantial past experience analys ing right whale photographs for photo-ID and body condition. Images were scored for the following 4 parameters: body fat condition, skin condition, rake marks near the blowhole and cyamids around the blowhole (Pettis et al. 2004; Table S1 in Supplement 2 at www. int-res. com/ articles/ suppl/ n043p323 _ supp2. pdf). Standard health assessment analyses re quire the scoring of rake marks on both sides of the blowhole (Pettis et al. 2004), yet due to the paucity of images available for this study, we assumed the same score for both left and right sides in individuals for which photographs of only one side were available.

Stable isotope analyses of SRWs and putative prey
Putative prey species including Antarctic krill Euphausia superba and other zooplankton were collected from western South Georgia waters during December 2017 and January 2018 on the BAS Western Core Box cruise (cruise no. JR17002), for the purpose of identifying potential SRW prey species via stable isotope analysis of carbon and nitrogen. To establish the trophic position of SRWs in the South Georgia food web compared to other pelagic krill predators, some myctophid species were also collected for analysis. Myctophids in this region feed predominantly on copepod and euphausiid species so were anticipated to have similar isotopic signatures to the SRWs (Saunders et al. 2019). Samples were stored frozen at −80°C. Whale skin tissues were sampled approximately 1 mo after the zooplankton samples were taken. Samples were prepared for stable isotope analysis at the BAS and sent to the Scottish Universities Environmental Research Centre (SUERC) isotope facility (East Kilbride, UK) for analysis. All zooplankton and fish samples were lipid-extracted prior to carbon isotope analysis (following Reid et al. 2012). Whale skin samples were both lipid-extracted and gelatinised following the standard laboratory protocol of the Department of Archaeology, University of Cambridge (as detailed by O'Connell et al. 2001).

Sightings from platforms of opportunity
To both augment our data and place our SRW expedition work in the broader context of whale recovery at South Georgia, we undertook analyses of sightings from platforms of opportunity. Prior to and during the expedition, we promoted the collection of SRW photo-ID images within the International Association of Antarctic Tour Operators tourist and naturalist cruise-going community via informative lectures, circulation of leaflets and through social media outlets. As part of this campaign, we requested that visitors submit SRW photo-ID images to www. happy whale .com (hereafter referred to as Happywhale). Additional sighting data have been collated by the South Georgia Museum from other mariners (e.g. Richardson et al. 2012), and sightings recorded by GSGSSI Fisheries Protection vessels were provided directly to the authors (see Table S3).
South Georgia Museum sightings reports from 1992−2018 were examined and collated by year and species. These reports were mostly contributed by expedition ship naturalists and mariners and provided the location of each species sighted and an estimate of the numbers of animals seen during the sighting. Sighting reports were limited to the area 50°−58°S and 33.5°−43°W, which includes Shag Rocks and the South Georgia region and continental shelf waters. Only baleen whale and sperm whale sightings were summarised, providing annual sightings rates for all large whales. Due to uncertainties as to whether English or American date notation were used in original reporting, sightings were binned by year rather than season. A number of caveats must be considered in relation to these data; sightings reported to the museum are opportunistic, are a small snapshot of the whales sighted locally, are biased towards coastal areas and the quality of reporting can vary considerably. Here, we focus on the sightings reported for humpback and SRWs, because these are both commonly reported and are particularly visually distinctive species relative to other rorqual whales sighted in the region.
The annual sightings for humpback and SRWs were collated, as well as total numbers of animals reported. The relative frequency of sightings of each species in relation to all whale species sighted was also calcu-lated, in order to provide a sightings measure that is independent of increasing visitor numbers through time. Sighting reports from GSGSSI fishery protection vessels the 'Dorada ' (2003−2004) and the 'Pharos' (2016−2018) were summarised separately. Counts were sometimes reported to the museum as a range, or as a minimum number (i.e. x+). For ranges, the lower value was taken, and for minimum numbers only the minimum value was used, to provide conservative estimates from these opportunistic sightings.

Survey effort and visual observations
Weather and sea conditions were suitable for visual observations for 76 h (within 12 d out of 19 d spent in South Georgia waters) and spanned 327 nautical miles of survey. Cruise tracks during observation periods are illustrated in Fig. 3. A total of 36 cetacean sightings were recorded during the expedition (Table 2). Fifteen sightings of 31 SRWs were documented (Fig. 3); 11 of these encounters were of singletons or pairs, and the largest group sighted was estimated at 5 whales. No surface feeding behaviour by SRWs was observed during the expedition. The most notable behavior corresponded to long, repeated dives, estimated at > 9 min in duration, performed by singletons (at least 6 different individuals). Biopsy sampling was attempted 7 times on 6 SRWs and 1 humpback whale. Four biopsy samples were collected (from 3 SRWs and 1 humpback whale). The SRWs were subsequently sexed by genetic analysis; 2 were females and 1 was male (details of genetic analyses are summarised in Carroll et al. 2020). Two of these samples (1 female, 1 male) provided sufficient tissue for stable isotope analysis.

Passive acoustic monitoring
A total of 27 sonobuoys were deployed between 25 January and 17 February. Six were deployed in the western approaches to South Georgia, and 21 were deployed in locations around the island's northern coastline and shelf. All of the 27 deployed sonobuoys functioned correctly. A total of 85.5 h of recordings were made, and around 3500 calls were detected and processed to calculate their bearings. SRWs were confidently detected on 13 buoys and detected with less certainty on a further 6, for a total of 19 buoys with detections (Fig. S1). The most commonly heard call types were upcalls < 200 Hz; some gunshot calls were also detected. These 2 call types have previously been described for North Pacific and North Atlantic right whales on calving and feeding grounds and for SRWs on calving grounds; upcalls (but not gunshots) have also been described from SRWs on feeding grounds (Širović et al. 2006, Mellinger et al. 2007, Munger et al. 2008, Parks et al. 2011, Matthews et al. 2014, Soldevilla et al. 2014, Webster et al. 2016, Crance et al. 2017, Dombroski et al. 2017. Call rates were generally low, with periods of silence of several hours, making triangulation using multiple buoys challenging. Acoustic detections of Antarctic blue whales Balaenoptera musculus intermedia, fin whales Balaenoptera physalus, humpback whales, sperm whales Physeter microcephalus and killer whales Orcinus orca were also documented on the sonobuoys (see Table S2).

Photo-ID catalogue compilation and matching
Of the approximately 1880 images collected during SRW encounters, 1032 images were of sufficient quality and angle to be retained for photo-ID and health assessment analysis. An additional 64 oppor- tunistic photographs taken during commercial cruises to the region were submitted to Happywhale and have been added to the photo-ID catalogue. Ana lysis of these images resulted in 35 assigned ID numbers (photo-IDs of medium or high quality). There were 11 individuals with high-quality left photos, 15 with high-quality right photos and 6 definitely unique individuals with high-quality left and right photos). No matches were found in a comparison of images from the 2018 South Georgia expedition and the wintering ground catalogues from South Africa, Brazil and Península Valdes or between the 2018 South Georgia expedition and the 12 images collected during the 1997 South Georgia expedition (Moore et al. 1999).

Health assessment
A total of 26 of the 35 photo-identified whales had adequate photographic coverage to support a complete health assessment (Table 3). Four of the 35 whales identified in this study could not be assessed for any of the 4 health parameters due to a lack of adequate imagery of the associated body region. An additional 5 whales could only be partially scored. The assessment included animals with photos of only one side of the head, as rake mark scores were assumed to be the same for both sides in this study.
The cumulative scaled health score for individual animals ranged from 12 (the lowest, or 'healthiest' animal) to 21 with an overall mean (±SD) of 15.4 ± 2.9. The maximum possible score (i.e. the poorest possible health assessment) was 30. More than half of the SRWs analysed here received a 'Good' rating for each of the 4 parameters overall (Table 3), and no SRW received a 'Poor' body condition or rake marks rating. However, 41.9% of animals analysed for body condition were characterized as 'Medium', indicating some level of body fat loss in the cervical region immediately caudal to the blowholes, and 35.7% were given a 'Poor' skin condition rating due to patches of sloughed skin, cyamids or lesions on the body. Rake marks were somewhat prevalent (40% left side and 35.3% right side) near the blowholes as well. No whales with appropriate imagery displayed any significant cyamid coverage around the blowholes. There were no obvious signs of current or previous entanglement observed. Examples of the image assessment are shown in Fig. S2A−F. Inspection of the South Georgia catalogue also revealed the presence of kelp gull lesions on some animals (e.g. Fig. S2G). Lesions were positively identified on 4 whales and were possibly identified on an other 3 animals.

Stable isotope analysis
Samples of 10 zooplankton species and 4 midwater fish species were collected during the Western Core Box cruise (JR17002). Stable isotope data were also obtained from 2 South Georgia SRWs for comparison. Zooplankton and fish species showed a wide variety of both δ 13 C and δ 15 N values within and between species (Fig. 4). Stable isotope δ 13 C values ranged from −26.5 ‰ in particulate organic matter to −16.4 ‰ in the copepod Calanoides acutus. A general enrichment was found from primary (e.g. Salpa spp.,) to secondary consumers (e.g. Themisto gaudichaudii, Paraeucha eta spp.) with considerable overlap in values among taxa (Table S3). Values of δ 15 N ranged from 1.9 ‰ in filter feeding salps to 9.9 ‰ in the predatory myctophid fish Gymnoscopelus piabilis. As for carbon isotope ratios, nitrogen values showed a continuous rather than stepwise increase from filter and suspension feeders to secondary consumers. SRWs were enriched by 2.3−4.0 ‰ in δ 15 N relative to euphausiid species and by 1.5−2.7 ‰ relative to various copepod species. Both whale specimens showed similar δ 5 N and δ 13 C values to the majority of myctophid fish species analysed, and Krefft ichthys anderssoni and Gymnoscopelus fraseri in particular (Fig. 4). Of the 2 specimens analysed, the tissue of the male specimen (EAU-SG01) was slightly en riched in δ 15 N (8.7 ‰) and depleted in δ 13 C (−21.2 ‰) compared to the female (EAU-SG02, 7.6 and −20.7 ‰).

Opportunistic whale sightings
The distribution of opportunistic SRW and humpback whale sightings reported to the South Georgia Museum since 1992 is shown in Fig. 5A   Relative sighting frequencies indicate that the SRW was the most commonly seen species in South Georgia waters between 1999 and 2010 (Fig. 6, Table S4). Since 2010 the pattern is more variable, with humpback whales being more regularly sighted from 2013− 2018, with the exception of 2015 (Fig. 6A,B).
The fishery protection vessel data shows a similar pattern, with more SRW sightings during January 2003 and 2004 and more humpback whale records in January 2017 and 2018. Species counts show fairly stable numbers of both species reported over time until 2013. After this year, total counts of humpback whales have increased by 3−10× (from < 40 yr −1 to > 250; Fig. 6C). This reflects an increase in the group sizes reported rather than the total numbers of sightings (Fig. 6B). In contrast, higher counts of SRWs between 2004 and 2007 reflected an increase in total sightings at that time, with no change in the group sizes reported.

SRW habitat use
SRWs were frequently encountered in South Georgia during the 2018 survey, supporting previous conclusions by Moore et al. (1999) and Richardson et al. (2012) that they occur regularly off the north coast in summer. SRWs were also regularly detected acoustically, as they were heard on at least 13 of the 21 sonobuoys deployed in South Georgia waters. However, the small estimated SRW group sizes observed during this survey (median group size = 2, SD = 1.2) and in past reports to the South Georgia Museum (from 2014−2019, median group size = 2, SD = 2.1) indicate perhaps small group foraging activity or travelling. During the survey, some whales were observed making repeated dives of > 9 min duration. Their target prey were most likely to be krill or copepods, based on previous studies (Matthews 1938, Valenzuela et al. 2009, 2018 and the current isotope data. Similarly long dive profiles of ~10 min duration and 50 m depth have also been recorded in Argentine Patagonia (Argüelles et al. 2016); in that case it was suggested that those whales were feeding on lobster krill (the Grimothea post-larvae of Munida gregaria), consistent with previous observations by Matthews (1932). Historical stomach contents data suggests whales at South Georgia fed on krill rather than copepods (Matthews 1938, Tormosov et al. 1998), but sample sizes were very small as very few SRW diets were analysed. Active acoustic analysis of the prey field in the vicinity of SRWs exhibiting this diving behaviour could provide more insights into the target prey.
Stable isotope analyses of whale feeding patterns at Península Valdés provide a general indication of whether SRWs are feeding at high or low latitudes (Rowntree et al. 2008, Valenzuela et al. 2009, 2018. Stable isotope signatures were available from just 2 SRWs at South Georgia, so we can only make limited interpretation from these data. These 2 SRWs had carbon and nitrogen isotopic values within the same range as the 'high latitude' SRW feeding group previously identified by Valenzuela et al. (2018). Since the isotope signature of food is thought to take 1−3 mo to be incorporated into whale skin tissue (Borrell et al. 2012) and the South Georgia samples were collected in January, the pattern suggests that these whales were likely feeding at high latitudes during late spring, and so may have been present in the vicinity of South Georgia for 2 mo or more. If we assume a tissue discrimination factor in nitrogen of 2.8 ‰ between whale skin and their candidate zooplankton prey (Borrell et al. 2012), these data suggest that these SRWs were likely to be feeding on local copepods and/or euphausiids, but these prey sources could not be distinguished as they have overlapping stable isotope signatures (Fig. 4). The 2 myctophid fish species closest in isotopic values to the SRW (Krefftichthys anderssoni and Gymnoscopelus fraseri) feed on a mixture of Thysanoessa krill and copepod species (Metridia spp. and Rhincala nus gigas, respectively) (Saunders et al. 2019). Myctophid and SRW isotopic signatures suggest both predators feed at a similar trophic level and potentially have similar prey niches in South Georgia waters. Further analyses of fatty acids and compound-specific isotopes of whales and their candidate zooplankton prey could help to better distinguish the species of prey being consumed at South Georgia. Habitat modelling of SRW feeding patterns using 19 th and 20 th century offshore whaling data suggests that SRWs forage in proximity to oceanographic features that act to aggregate prey patches, such as thermoclines and currents (Torres et al. 2013, González Carman et al. 2019). In the southwest Atlantic, SRWs have a broad summer feeding range, which includes the Patagonian Shelf (Rowntree et al. 2001), Scotia Sea (Moore et al. 1999, Reilly et al. 2004 and historically also offshore waters 30°−45°S (Smith et al. 2012). Stable isotope studies conducted on their main calving ground at Península Valdés suggest that the majority of whales feed south of the Polar Front before returning to the calving grounds (Valenzuela et al. 2009(Valenzuela et al. , 2018, but recent satellite tracking studies indicate that this area spans much of the Scotia Arc (Zerbini et al. 2016(Zerbini et al. , 2018. South Georgia is therefore one of a number of sites where SRWs feed during the summer. Patterns of SRW sightings by visitors to South Georgia are very coastal, reflecting the areas where the vessels are most concentrated and spend most time. However, fisheries pro-tection vessels travel further offshore and routinely survey out to the shelf break. Sightings reports from these vessels show a number of reports due westnorthwest of the island within shelf waters, and further west offshore at Shag Rocks (Fig. 4D). Shelf waters west of the island also appear important to SRWs satellite tracked from their Argentine calving areas. For example, 'area restricted searching' (ARS) behaviours (locations where whales stop migrating and remain for extended periods) have been seen in South Georgia shelf waters just off the western edge of the island (Zerbini et al. 2018), and to the east of South Georgia, further offshore (beyond the 1000 m contour). Tracking of krill-feeding Antarctic fur seals in 2005−2006 also showed seals concentrating their feeding dives in the same 2 locations (Staniland et al. 2011), suggesting that these areas may be a predictable, or important, source of krill for both species. This westerly area may have also been a historically important site for SRWs; Matthews (1938) quotes Hinton (1925, p. 155) as saying 'In 1914 Barrett Hamilton was told that SRW were taken when found with other species, but that they usually kept themselves to the north-west of South Georgia and were not worth hunting, especially there'. Higher levels of chlorophyll a productivity and iron are found at the northwest end of the island compared to the northeast, supporting a higher biomass of krill -and, as a corollary, krill predators -in the western part of South Georgia, possibly also enhanced by nitrogen feedbacks from the krill predators (Whitehouse et al. 1999, Korb & Whitehouse 2004. Further studies to track SRWs in this region and analyse habitat use patterns will help to better establish the significance of these locations, and of the different prey sources that SRWs could be targeting.

Origin and body condition of South Georgia SRWs
At least 3 individuals photo-identified at South Georgia had wounds consistent with gull lesions, which link them directly to Península Valdés in Argentina because it is the only calving area where gulls currently harass whales (IWC 2011(IWC , 2016. This is consistent with linkages directly identified by past photo-ID matching (Rowntree et al. 2001) and recent satellite tracking of whales directly between these locations (Zerbini et al. 2018). However, no individually identified whales from the South Georgia catalogue were directly connected with any lower latitude calving grounds in this study, despite com -parison with the sizeable photo-ID catalogues held in Argentina, Brazil and South Africa. This is likely because in South Georgia, photo-ID images were taken from the side, and could not easily be compared with the calving grounds because the calving ground photo-IDs are taken overhead. The Península Valdés catalogue represents a large proportion of all whales using Argentine waters, with 3034 individuals present in the catalogue in 2015 compared to a population abundance estimated at ~4245 in 2010 (IWC 2013). Thus, if South Georgia whales all visit this region, the chance of a match should be high. However, the likelihood of matching is reduced if the whales sighted at South Georgia were young, as this cohort is the least well represented in the calving ground catalogue (the average age for first reproduction in SRWs is 8 yr old; Best et al. 2001). Some of the whales on the South Georgia survey were recorded as being small in size, and therefore may have been juveniles (they were not small enough to be definitely categorised as calves); therefore, they may have been less likely to be present in the Península Valdés catalogue. Collection of overhead images from South Georgia is challenging due to the remoteness of the region (which prohibits aerial surveys) and local weather conditions which limit data collection via unmanned aerial vehicle (UAV). Only South Georgia images which show both sides of the head could be used in comparisons with the calving grounds. More collection of side-on images from the calving grounds could improve the likelihood of photo-ID matching with South Georgia in the future.
Kelp gull wounds seen on 3 whales confirm the connection with Argentina, since this gull behaviour is widespread on the Península Valdés calving area and is also unique to the region (Marón et al. 2015). This connection is corroborated by regional genetic patterns. Genetic data collected from the expedition was assembled along with earlier genetic data collected by Moore et al. (1999) to assess the population origins of the South Georgia whales. Mitochondrial DNA and nuclear microsatellites supported a putative link with the southwest Atlantic, showing that SRWs from South Georgia are genetically more closely linked to the southwest Atlantic calving grounds (Brazil and Argentina) than to the southeast (South Africa) (Carroll et al. 2020), consistent with the migratory movements suggested by satellite tracking studies (Mate et al. 2011, Zerbini et al. 2018. Our body condition analyses provide a baseline data set for assessing the seasonal health condition of SRWs in South Georgia waters, for comparison with other regions, including their associated calving grounds, and for measuring local changes in body condition over time. None of the South Georgia SRWs analysed received a 'Poor' score for all 4 health parameters, but ~36% of whales were assessed as having 'Poor' skin condition and ~42% as 'Medium' body condition due to a lack of fat in the cervical region caudal to the blowhole (i.e. they are thin). In the North Atlantic, Pettis et al. (2004) reported a mean body condition score for known cows without a calf of 1.08 and cows with a calf of 1.67. These values are similar to our mean body condition results (1.15 ± 1.49) Overall, North Atlantic right whale health parameter scores were higher (reflecting poorer condition) during calving years and for presumed dead whales (Pettis et al. 2004), underscoring the fact that visual health assessment methods are useful for measuring individual fitness when individuals are assessed over multiple seasons. A similar pattern has been seen using overhead images (Christiansen et al. 2020), with juvenile, adult and lactating female SRWs across 3 calving grounds found to be in better body condition than North Atlantic right whales at equivalent stages.
The overall health of the surveyed animals (assessed using the scoring system of Rolland et al. 2016) was good, with average values of 0.81 (SD = 0.09); higher than those of all the demographic cohorts of North Atlantic right whales assessed by Rolland et al. (2016). At present, these results cannot be compared with other SRW populations, as body condition assessment using side-on images has not been published for any other SRW populations. This is because calving ground studies routinely use overhead, rather than side-on, images. For example, body condition analyses carried out by Argentina and South Africa both use size-calibrated overhead images of whales (photogrammetry) which can provide detailed information on whale growth rates and body fat variation during the calving period (e.g. Miller et al. 2012, Christiansen et al. 2018, 2020. Whales on their feeding grounds would be expected to gain in mass over the season (for an example with gray whales, see Bradford et al. 2012), but no photogrammetry work has been conducted on right whale feeding grounds to date to assess this in detail. Photogrammetry data collected from the feeding ground could provide quantitative information on SRW health which can be directly compared with the associated calving ground. There are difficulties in conducting this kind of research at South Georgia given the weather conditions, but as UAVs improve in terms of their stability and battery life, such data collection may prove more possible in the coming years.

Changing patterns of SRW and humpback whale occurrence
Humpback whales were regularly detected by sonobuoys in South Georgia waters, although they were only sighted twice during the survey. Interestingly, sightings data from the South Georgia Mu seum suggest an increase in sighting rates of humpbacks relative to SRWs in recent years, with 5 out of the last 6 years reporting more humpback sightings than SRWs. This is in contrast to 2000− 2010, when SRWs were the most commonly encountered species. The temporal pattern may be a consequence of the relative abundance of the 2 species as they recover in the southwest Atlantic; recent trends in abundance of the 2 species have been similar at 6.5 ± 0.2% yr −1 for SRWs between 2000 and 2010 (Cooke et al. 2015) and ~6.1% yr −1 between 2008 and 2012 for humpback whales ). However, while SRW abundance in Península Valdés was estimated at 4245 whales in 2010 and abundance off Brazil was likely less than 1000 animals in that year (197 mature females counted; IWC 2013), humpback whale abundance on their Brazilian wintering ground was estimated at 20 389 whales (CV = 0.07) in 2012 . In 2019, this population was pre dicted to be approaching 25 000 (95% probability interval = 22 300− 27 000) and to be very close to full recovery from exploitation (Zerbini et al. 2019). Southwest Atlantic humpbacks feed throughout the Scotia Arc in summer, across a broad feeding area that includes the South Sandwich Islands and areas to the east of that region (Zerbini et al. 2006, Horton et al. 2020. In South Georgia, museum sightings reports suggest humpback whale sightings were much less frequently reported prior to 2012. Their occurrence in South Georgia waters only recently suggests that as the population has increased, it has expanded from other areas of the Scotia Arc into South Georgia waters. The recent 'jump' in sightings may also support the hypothesis advanced by Clap ham et al. (2008) that the southwest Atlantic humpback population lost the 'cultural memory' of South Georgia as an important feeding destination following prolonged hunting at South Georgia, wiping out the animals which had long-term fidelity to this region and therefore slowing the species' return.
The temporal pattern of SRW sightings does not suggest that SRW numbers are increasing on their South Georgia feeding ground (Fig. 5), despite a significant increase in cruise ships and associated opportunistic sighting effort since 1995 (IAATO 2020). This is in contrast to the steady SRW increases reported on their associated calving grounds (Groch et al. 2005, Cooke et al. 2015). Since SRWs use multiple feeding areas in the southwest Atlantic (Valenzuela et al. 2009, 2018, Zerbini et al. 2018, prey availability may limit the number of SRWs that feed in each location each year. However, while the relative sightings rates point towards a shift in the frequency of occurrence of SRWs and humpback whales over time, it must be cautioned that most of the sightings reports to the museum are from a narrow spatial area within South Georgia (the coastal waters and western approach to the island) and are not related to sighting effort, so caution must be applied when drawing biological interpretations from the relative sightings rates. South Georgia also has notoriously rough seas, particularly in deep water. SRWs display less of their back at the surface than humpback whales and have no dorsal fin, so may be less visible in rough conditions, particularly to vessels that do not have dedicated or experienced whale observers. SRW group sizes are also smaller than those of humpbacks, making them harder to detect. Line transect sightings surveys by experienced observers, spanning shelf and coastal waters, are required to quantitatively evaluate the relative densities of the 2 species.

CONCLUSIONS
The local ecosystem and climate into which baleen whales are recovering is very different from that of a century ago, when krill-feeding fur seals were absent and krill densities may have been much higher. Temporal trends in whale abundance over the last 20 yr cannot be strongly interpreted from opportunistic sightings data, since this depends on annually variable effort from visitors who follow similar coastal transit routes in South Georgia waters and who are not experts at cetacean identification. However, using these data to compare the fortunes of 2 easy-to-recognise species, and building on the literature reviewed in this study, we can develop some hypotheses for further investigation. For example, fisheries protection vessel data show a con cen tra tion of sightings of both species at the western shelf break between South Georgia and Shag Rocks (Fig. 4C,D), coincidentally also an area where satellite tracked SRWs have conducted ARS (e.g. foraging activities; Zerbini et al. 2018), suggesting that this area is important for whales. We have also seen large increases in humpback whale counts and group sizes since 2011 (Fig. 5). We expect that opportunistic whale sightings will continue to increase in the coming years and hypothesise that the western shelf break region is likely to be a hotspot for such sightings.