Data on the Southern Annular Mode (SAM) Index and the Multivariate El Niño Southern Oscillation Index (MEI) were obtained from online data repositories. As the two dominant modes of climate variability in the Southern Hemisphere, SAM and the El Niño Southern Oscillation (ENSO) alter atmospheric circulation patterns (e.g., winds) and can impact ecosystem functionality by changing the physical conditions that structure Antarctic food webs and ecosystems (Thompson and Wallace, 2000; Massom et al., 2006; Stammerjohn et al., 2008b).

Monthly SAM data were acquired from the online database established by the National Environmental Research Council (NERC) British Antarctic Survey (BAS) (https://www.bas.ac.uk/data/our-data/publication/the-southern-annular-mode-variability-trends-and-climate-impacts-across/) and monthly MEI data were taken from the National Oceanic and Atmospheric Administration (NOAA), Earth System Research Laboratory (ESRL), Physical Sciences Division (https://psl.noaa.gov/enso/mei/) (Zhang et al., 2019; Fogt and Marshall, 2020). Seasonal SAM indices and the MEI were calculated based on the average values observed during the preceding austral spring (September-November), summer (December-February) and autumn (March-May) leading up to the winter sampling season, and during the winter (June-August) sampling season. We note that in 2013 and 2014 some of the sampling took place in early-to-mid September, rather than in August ( Table 1 ). Winter values of the SAM index and the MEI during these years were calculated in the same way as the other sampling years for comparison purposes, averaging the monthly values from June to August to obtain climate values for the SAM index and for MEI during the sampling season. September sampling took place during the first few weeks of the spring season, thus the maturation state of females collected during September of 2013 and 2014 was likely impacted more by the seasonal climate and environmental variability leading up to the time of sampling, rather than the spring climate phases at the time of sampling. The seasonal SAM and ENSO anomalies for 2012 to 2016 were calculated by subtracting the seasonal values for a given year (e.g., 2012) from the average seasonal values of the full dataset (1957 to 2021 for the SAM Index and 1979 to 2021 for the MEI). The seasonal grouping of SAM and ENSO indices was done for two primary reasons (i) krill reproductive cycles happen on seasonal timescales (Ross and Quetin, 2000), and (ii) previous studies (e.g., Saba et al., 2014) have found that seasonal SAM and ENSO can impact ecological and food web dynamics at the Antarctic Peninsula.

A Seabird 911 CTD was used to measure salinity (PSU), temperature (°C) and chlorophyll-a (chl-a; µg L^-1; determined from fluorescence) in the water column at the sampling stations that corresponded with krill collected in each year. Salinity and temperature values used in our study were calculated as the average values over the top 100 meters of the water column at each station and are subsequently referred to in the text as AvgTemp and AvgSal. Integrated chl-a (IntChl, mg m^-2) was calculated using trapezoidal integration of chl-a over the top 100 meters of the water column.

Sea ice data were downloaded from the National Snow and Ice Data Center (https://nsidc.org/data/nsidc-0079/versions/3) and were derived from the Nimbus-7 SMMR and DMSP SSMI/S-SSMIS satellite temperature brightness data, which were binned into 25 by 25 km grid cells. Daily sea ice concentration was assessed as percent sea ice cover (Comiso, 2017; Stammerjohn and Maksym, 2017). Average sea ice coverage at the sampling station two weeks prior to each sampling event (SIC2Wks) and the number of days between the date of sea ice advance and the sampling date each year (DaysAdv) were determined. Sea ice advance was determined to occur when the sea ice in a study region was greater than 15% for at least 5 consecutive days (Stammerjohn and Maksym, 2017). After running linear models assessing the relationships between multiple sea ice variables and OMI, we chose to include SIC2Wks and DaysAdv as the sea ice variables in our models because they had stronger univariate relationships (i.e., lowest AIC scores) with female krill OMI relative to the other sea ice variables we tested (i.e., maximum sea ice coverage and the concentration of sea ice at the time of sampling).

When SAM and MEI are in the same phase (i.e., -SAM/La Niña and +SAM/El Niño), they offset each other with regard to their impact on storms, creating milder conditions at the nAP (Clem and Fogt, 2013; Stammerjohn et al., 2008a). Although decreased storm activity occurs during both -SAM/La Niña and +SAM/El Niño paradigms, they impact the ocean environment differently. Unlike +SAM, -SAM encourages the growth of larger phytoplankton, notably diatoms, at the Antarctic Peninsula creating a favorable feeding environment for diatom grazers, like krill (Montes-Hugo et al., 2008). And, in contrast to El Niño, La Niña, which typically corresponds with reduced sea ice extent and duration in the winter and spring around the nAP, is associated with an earlier sea ice retreat that could initiate an earlier onset of seasonal primary productivity at the nAP, potentially encouraging earlier maturation in krill (Reiss et al., 2020). Thus, a combination of -SAM/La Niña is more likely to prompt early diatom blooms, while +SAM/El Niño is more likely to result in milder conditions with no associated diatom bloom.

As already noted, we observed the highest OMI during the winter of 2014. Here, we will explore the climate and environmental conditions leading up to, and during that survey. In the spring preceding our 2014 survey (September to November 2013), the nAP was forced by a -SAM/La Niña. As a result, storm activity would have been minimal, with SAM and ENSO offsetting each other. La Niña conditions would have resulted in greater inflow of warmer, mCDW into the study region resulting in a shallower seasonal mixed layer (Walsh and Reiss, 2023). A shallower mixed layer would promote early spring phytoplankton growth, primarily constituted of diatoms (Schofield et al., 2017), which are the preferred phytoplankton prey source of krill (Haberman et al., 2003). Indeed, seasonally averaged satellite-derived ocean color data indicate a robust phytoplankton biomass in our study area that spring ( Supplementary Figure 2 ). That summer (December 2013 to February 2014), SAM switched to its positive phase and La Niña remained in effect through autumn (March to May 2014). This combination would have created warmer conditions, but because both indices were relatively neutral in the summer and autumn, storm activity was likely minimal. Although seasonally averaged satellite-derived ocean color data indicate phytoplankton biomass was relatively high through the summer and autumn of 2014 ( Supplementary Figure 2 ), Mendes et al. (2018) found that cryptophytes were the dominant phytoplankton in the neighboring Gerlache Strait during the summer that same year, and thus, were likely the dominant phytoplankton in our study region too. Cryptophytes are not typically consumed by krill due to their small cell size (Haberman et al., 2003).

Copepods, on the other hand, are more efficient foragers of small phytoplankton (Deppeler and Davidson, 2017), and could have provided an alternative food source to krill during the summer and autumn of 2014. When krill feed carnivorously, they tend to accumulate storage lipids, primarily in the form of triacylglycerols (TAGs) more readily than when feeding on phytoplankton (Hagen et al., 2007). Thus, the summer and autumn conditions preceding winter 2014 may have promoted a more carnivorous diet in the local krill, resulting in an elevation of their total body lipid contents and improving body condition leading into the winter. With higher body condition at the onset of winter, krill would likely be in better condition toward the end of winter, even if food had been scarce. Bernard et al. (2022) found that krill given a carnivorous diet in the autumn and then starved in the winter had higher total body lipid contents by the end of winter than those krill fed diatoms the entire time. Because early reproductive development requires metabolic resources, krill with lipid reserves toward the end of winter would be more likely to initiate early reproductive development if they encountered sufficient phytoplankton due to their ability to fuel the process. In the winter of 2014, phytoplankton concentrations were significantly higher than those measured in any of the other years of our study. Despite this, analysis of fatty acid biomarkers on krill collected during the same cruise showed that krill had a low diatom index (the ratio of 16:1n-7 and 18:4n-3) (Walsh et al., 2020). Walsh et al. (2020) observed significantly lower total lipid contents in krill collected in 2014 compared to those collected in 2016, suggesting that, with cryptophytes dominating the phytoplankton community the previous summer (Mendes et al., 2018), krill would have likely been feeding largely carnivorously in the summer and autumn, which would have allowed them to build up their lipid reserves prior to the onset of winter (Bernard et al., 2022). This may have enabled female krill to limit sexual regression (Kawaguchi, 2016), allowing early and quick maturation toward the end of winter and early spring when blooms were initiated. In both scenarios, OMI would have been high at the end of winter.

In addition to the impacts that SAM and ENSO have on ocean circulation patterns and sea ice dynamics, when SAM and ENSO are in opposing phases, they amplify the climate conditions at the nAP (i.e., +SAM/La Niña = amplified warming, -SAM/El Niño = amplified cooling) (Stammerjohn et al., 2008b). In the spring preceding 2013, the nAP was dominated by an El Niño and a negative SAM. This resulted in amplified stormy conditions that created an unfavorable environment for phytoplankton growth ( Supplementary Figure 2 ), with a late spring phytoplankton bloom causing a potential delay in krill physiological processes, such as growth, reproductive development, and spawning that year (Schmidt et al., 2012; Saba et al., 2014). However, in the summer of 2013, SAM shifted toward its positive phase and El Niño remained constant, resulting in neutral conditions, reduced winds, and favorable conditions for diatom blooms ( Supplementary Figure 2 ). These conditions continued through the autumn of 2013, with high chl-a observed throughout the study region ( Supplementary Figure 2 ). With plentiful phytoplankton in the water during the summer and autumn, female krill may have fed more herbivorously in the seasons preceding the winter sampling period, which may have limited their ability to accumulate lipid reserves prior to that winter.

Following the La Niña of 2014, ENSO shifted into its positive phase, El Niño, gaining strength through 2015 and reached a maximum in the spring (September to November 2015) prior to our winter survey in 2016. This was accompanied by a strongly positive SAM. With the nAP being dominated by a +SAM/El Niño from spring to autumn the region experienced little-to-no storm activity, cooler weather (primarily in response to the strong El Niño), and, based on seasonally averaged satellite-derived ocean color data, an intensive spring phytoplankton bloom at the retreating edge of the marginal ice zone ( Supplementary Figure 2 ). Although a -SAM is more highly associated with diatom presence than a + SAM, the strong El Niño prevailing over the system in 2016 resulted in phytoplankton blooms from spring to autumn ( Supplementary Figure 2 ) with the most intense blooms dominated by the large centric diatom, Odontella weissflogii (Costa et al., 2020). Interestingly, the large size (>70 µm) of O. weissflogii probably made it hard for krill to feed on, reducing their foraging efficiency during these blooms (Opaliński et al., 1997). Indeed, Costa et al. (2020) observed low grazing indices and a healthy, rapidly sinking diatom community. Therefore, although phytoplankton biomass was high, krill and indeed other smaller grazers like copepods, could not consume it. Thus, it is likely that copepods were less abundant given the paucity of edible phytoplankton resources and krill may have fed preferentially on protozoans, including dinoflagellates, which were a dominant part of the nanoplankton community and the second-most representative taxonomic group in the region (Costa et al., 2020). Through a process known as trophic upgrading, dinoflagellates can improve the biochemical composition of low-quality nano- and picophytoplankton, providing important polyunsaturated fatty acids that krill need for growth, eicosapentanoic acid (EPA) and docosahexaenoic acid (DHA) (Tang and Taal, 2005; Schmidt et al., 2006). However, krill obtain significantly less storage lipids when feeding solely on dinoflagellates compared to feeding on zooplankton, such as copepods, that are rich in TAGs (Dalsgaard et al., 2003). Thus, these krill may have entered the winter with relatively lower lipid reserves than those in 2014.

In the winter of 2016, ENSO shifted to a La Niña, which combined with the +SAM to result in warmer conditions, delayed sea ice formation, and reduced overall sea ice cover in the study region. Winter phytoplankton biomass was the lowest observed across all five years of our study ( Table 4 ). Interestingly, fatty acid profile analysis on krill collected in the same survey indicated that post-larval individuals had been feeding primarily on diatoms in the weeks-to-months leading up to collection in August 2016 because they had a high diatom index (the ratio of 16:1n-7 and 18:4n-3) (Walsh et al., 2020). It is possible that high grazing rates by krill could have led to a dramatic decrease in phytoplankton biomass, but adult krill do not tend to increase their grazing rates in the winter, even if food is plentiful (Atkinson et al., 2002). Bernard et al. (2019) measured winter grazing rates of adult krill using the gut fluorescence technique and found that grazing rates in August 2016 were negligible. Thus, an alternate explanation is that the fatty acid biomarkers measured were more strongly representative of the autumn diet of krill rather than the winter diet as the turnover rates of these biomarkers would likely have been reduced by the low winter feeding rates of adult krill (Atkinson et al., 2002) and diminished food availability (Graeve et al., 2020). This notion supports our theory that, like the krill measured in 2013, the krill measured in 2016 were likely feeding herbivorously in the months leading up to winter. Without the resources necessary to initiate early reproductive development, female krill may have delayed reproductive maturation in favor of conserving their limited lipid reserves. It is possible that in doing so, their total lipid content would have been higher at the end of winter than those female krill that underwent early maturation in 2014, which was observed by Walsh and Reiss (2023).

Although the SAM and MEI anomalies are large-scale phenomena, they have been previously used to effectively forecast environmental conditions at the Antarctic Peninsula (e.g., Loeb, 2007; Stammerjohn et al., 2008b; Clem et al., 2016). For organisms that are highly mobile, and for which predicting their precise location nine months prior to sampling is impossible, the use of these indices is an appropriate way to broadly describe the environmental conditions they would have been faced with preceding the period of sampling. When SAM is in its negative phase, the westerly wind belt is geographically farther from Antarctica, maintaining colder and calmer conditions at the nAP (Clem et al., 2016). In its positive phase, westerly winds shift closer to the Peninsula, bringing more storms, warmer air and precipitation to the nAP (Clem et al., 2016). Positive SAM conditions are also associated with less sea ice, a deeper mixed layer, a shift in the phytoplankton community toward smaller cryptophytes, and a reduction in krill recruitment at the Antarctic Peninsula (Montes-Hugo et al., 2009; Atkinson et al., 2019).

During La Niña (negative MEI) conditions, the Amundsen Sea Low (ASL) is stronger than average, bringing warmer northerly winds to the nAP, resulting in less sea ice. During La Niña, there is also an enhanced delivery of warm, modified Circumpolar Deep Water (mCDW) into the Bransfield Strait (Ruiz Barlett et al., 2018). During El Niño (positive MEI) conditions, the ASL is weaker than average and Weddell Sea Shelf Water (WSSW) outflow into the Bransfield Strait is enhanced, resulting in cooler conditions, more extensive sea ice coverage, and potentially more sea-ice-associated food sources, in the region (Yuan, 2004; Clem et al., 2016; Nie et al., 2022; Walsh and Reiss, 2023). Further, WSSW has a higher salinity content and the PCR results showing the effects of salinity on OMI in this study may indicate the presence or absence of WSSW in the nAP region. A higher salinity content in the winter may also be observed when there is minimal sea ice and stratification in the upper water column (Venables et al., 2013; Saba et al., 2014). Despite being a year dominated by El Niño and cooler surface temperatures ( Supplementary Figure 1 ), there was anomalously low sea ice at the nAP in 2016 (Stuecker et al., 2017; Wang et al., 2019), which could have contributed to high salinity and low stratification during the winter of 2016 (Walsh and Reiss, 2023). A lack of stratification due to extremely low sea ice coverage makes potential prey items for krill, including pelagic phytoplankton, zooplankton, and organic debris, more widely dispersed through the water column, rather than being restricted to the surface ocean by a strong thermocline where they would constitute a concentrated food source for krill (Montes-Hugo et al., 2009; Kirillin et al., 2012).

Sexual maturation in female krill requires substantial nutritional input from their diet (Quetin and Ross, 2001). While diatoms are a critical food source for krill, krill are omnivorous and also feed on organisms such as copepods and dinoflagellates (Ericson et al., 2018). Feeding carnivorously provides krill with key lipids and fatty acids that they use to survive the winter and to aid in the progress of certain physiological processes, such as growth and reproduction (Hagen et al., 2007). Notably, Yoshida et al. (2011) found that the hatching success of embryos increased when the female krill that released them had fed omnivorously, transferring favorable ratios of key fatty acids that promote survival and growth to their embryos. Variability in seasonal climate and environmental conditions may impact the quality and quantity of food available for female krill to feed on throughout the year, encouraging them to be flexible in their foraging behavior (Walsh et al., 2020).

While our study reveals important seasonal climate and environmental relationships with krill reproductive development, it is important to be aware of the assumptions and limitations of this study when using these results to forecast how krill populations may change in the future. Due to the logistical limitations of winter research in Antarctica, we had a relatively small sample size for our study, with an uneven spread of samples throughout the study region and over time. Only three of our five survey years yielded sufficient sample sizes to allow for statistical comparisons between years. In these years, we observed strikingly different environmental and climate conditions, resulting in different OMIs of female krill. However, the number of potential unique combinations of seasonal SAM and ENSO as well as other environmental parameters is large, and our study has only just scraped the surface of these. Thus, while we can discuss implications of climate and environmental variability within the timeframe of our study, further work is needed to allow for the prediction of krill reproductive development in the future. In addition, while we have explored the seasonal impacts of climate and environmental variability on krill reproduction, we fully acknowledge that there are likely many more impacts that climate has on environmental conditions and food availability than we were able to discuss within the scope of this paper.