Samples were collected on 24 January 2019 outside of Askö marine station in the Baltic Sea. Benthic sled drags were taken at two stations (Low OM station 6019: latitude 58 44 32 N, longitude 17 40 58 E, depth 42 m, 1.77% C, 0.26% N; and High OM station 6022: latitude 58 44 68 N, longitude 17 48 75 E, depth 41 m, 3.7% C, 0.52% N; Figure 1 ). The low OM station is characterized by sandy clay sediments (i.e. larger particle sizes) and good oxygenation indicative of a transport bottom, while the high OM stations has muddy clay sediments (i.e. smaller particle sizes) and periodic hypoxia, representative of an accumulation bottom. Sediment was sieved through 1 mm to obtain individual gravid Monoporeia affinis and remove additional macrofauna, and sediment and water were saved. Materials were transported to Stockholm University, where the experimental units were set up in a 4°C constantly dark room to simulate aphotic field conditions from where the samples were collected. The sieved sediment and water were added to each individual 50 ml beaker (n = 64), with sediment around 3 cm depth (41.6 cm³ sediment volume) and a water column of around 3 cm. Four treatments were set up as follows: high control (gravid females, water, and sediment from High OM station 6022), high transferred (gravid females from Low OM station 6019 in water and sediment from High OM station 6022), low transferred (gravid females from High OM station 6022 in water and sediment from Low OM station 6019), and low control (gravid females, water, and sediment from Low OM station 6019). The experimental setup can be found in Figure 2 . For comparisons between treatments, the sediment OM content is referred to as high or low, and the amphipod origin is described as control (same experimental treatment as origin) or transferred (not same experimental treatment as origin). See Figure 2 also for timeline and treatments. One gravid M. affinis was individually placed in each beaker, corresponding to 722 ind m^-2, within densities previously reported for M. affinis in the area (Ankar and Elmgren, 1975; Raymond et al., 2021). Sediment was allowed to settle for 24 hours before adding oxygenation by bubbling air through a hose into the water. Water (alternating brackish and fresh) was added to each beaker when required due to evaporation, and salinity (6.9 ± 0.9) was checked to ensure no physiological stress on the amphipods.

All initial female M. affinis samples (n = 15 from each station; initial) were collected and frozen within 24 hours of field sampling. A sediment sample was taken from each station for initial stable isotopes and preserved by freezing at -80°C until stable isotope analysis. Eggs from the initial gravid females were removed from the female and counted under a stereomicroscope (Stereo Zoom 2000, Leica) to assess individual fecundity (number of embryos per female). Embryos with aberrations were identified according to Sundelin and Eriksson (1998), and percent viable embryos (VE) were calculated as the percentage of eggs without aberrations. Number of malformed, arrested, and membrane-damaged embryos were not differentiated between. The female were frozen separately in -80°C until further processing.

The experiment ran for 10 weeks. At experiment termination (end), a subsample (1.5 ml) of sediment was taken from each microcosm and frozen at -80°C. The surviving females and hatched juveniles were collected by sieving the sediment on a sterilized 200 µm sieve with brackish water, and then washed onto a sterile petri dish with sterile MilliQ water for counting recovered juveniles under a stereomicroscope. The females and juveniles were frozen separately in -80°C until further processing (females: n = 44; juveniles: n = 40; sediment: n = 45; see Figure 2 for more details).

Animal and bulk sediment samples were analyzed for bulk C and N elemental and stable isotope analysis from the initial time point (initial; head of females and sediment) and the experiment termination (end; head of females, sediment and a subset of 13 randomly selected pooled juveniles from each replicate to reach sufficient biomass for analyses). Samples were dried in a 60°C oven overnight (or over 3 nights in the case of sediment). Each sample was weighed and packaged in a tin capsule before analyzed at the University of California-Davis Stable Isotope Facility on a PDZ Europa ANCA-GSL elemental analyzer interfaced to a PDZ Europa 20-20 isotope mass spectrometer (Sercon Ltd., Cheshire, UK). The stable isotope ratios (R) of ¹³C:¹²C and ¹⁵N:¹⁴N was calculated based on the equation:

and utilized international standards (Vienna Pee Dee Belemnite for C and atmospheric nitrogen gas for N) and are denoted by δ (‰). Standard deviation was ± 0.09 ‰ for C and ± 0.12 ‰ for N. Percent C and N were calculated, as well as C:N. Due to high lipid percentage in amphipods, δ¹³C values for the amphipods were corrected for high C:N ratio according to Post et al. (2007). It has been shown that percent lipid and C:N ratio correlates strongly for M. affinis (Lehtonen, 1996).

Statistical tests were performed in R (version 3.5.0, R Core Team, 2018). When required by assumptions, data was tested for normality with a Shapiro–Wilk test, and for homogeneity of variance with a Levene’s test in the ‘car’ package. When data did not meet the assumptions of normality or homogeneity of variance, transformation was attempted, and if unsuccessful, non-parametric Kruskal-Wallis tests were performed instead of an analysis of variance (ANOVA) test. When significant ANOVA results were found, Tukey post-hoc tests were utilized to determine which variables differed significantly.

Three types of tests were here used: 1) The parametric t-tests and 2-way ANOVA to test for differences between groups and the interaction between independent variables when the data is normally distributed and the variances are homogeneous; 2) the Kruskal-Wallis tests when these assumptions are not met; 3) linear models or correlations to test the relationship between two continuous variables. A summary of the statistical tests and how they relate to the three hypotheses can be found in Table S1 . Hypothesis 1 was tested with unpaired t-tests between initial stations female fecundity (number of eggs per female), viable embryos (VE), and C:N ratios. To evaluate the effect of sediment OM content on amphipod reproduction, the number of recovered juveniles (square root transformed) at the end of the experiment were tested with a 2-way ANOVA as a response variable to sediment OM content (high or low) and female origin (control or transferred) as fixed factors with interaction allowed. Still related to H1, end of experiment female C:N (proxy for female body condition), juvenile biomass per individual (as a measure female reproductive output), and juvenile C:N ratios (proxy for juvenile body condition) were each tested as dependent variables with univariate Kruskal-Wallis test with sediment OM content (high or low) and female origin (control or transferred) as explanatory variables. Juvenile biomass per individual was calculated by dividing the total dry weight by the number of pooled individuals (n=13). Finally, multiple linear regressions were used to identify significant associations between amphipod female body condition, nutritional status (inferred from isotopes values) and reproductive output when assumptions were met: i) initial female C:N & VE, ii) initial female δ¹⁵N (higher δ¹⁵N may indicate higher trophic position or deteriorated body condition) & VE, and iii) end female C:N & number of recovered juveniles per female. When assumptions of multiple linear regressions were not met, Spearman’s rank correlation was run comparing between: i) end female δ ¹⁵N & number of recovered juveniles per female, and ii) end female δ¹³C & number of recovered juveniles per female. Hypothesis 2 was tested by comparing initial female δ¹³C and δ¹⁵N (separately) with unpaired t-tests between the two stations. Continuing with H2, end of experiment female and juvenile δ¹³C and δ¹⁵N were tested separately as dependent variables in a 2-way ANOVA in response to sediment OM content (high or low) and female origin (control or transferred) with interaction allowed, to assess if the transfer of females to new sediment altered their isotope values. Sediment stable isotopes were also statistically tested for differences between end of experiment treatments. To evaluate inter-individual variability in bi-variate isotope composition, ranges in female δ¹³C and δ¹⁵N were calculated in initial stations and end of experiment treatments. Additionally, euclidian distance (ED) was calculated within each treatment for end of experiment females by the following formula.

where δ¹³C_a and δ¹⁵N_a are δ¹³C and δ¹⁵N anomalies calculated by taking the difference of each individual replicate female δ¹³C and δ¹⁵N values from the mean δ¹³C and δ¹⁵N of each treatment (Karlson et al., 2018). Differences in δ¹³C_a between treatments at the end of experiment females were evaluated with a 2-way ANOVA in response to sediment OM content (high or low) and female origin (control or transferred) with interaction allowed. End of experiment female δ¹⁵N_a and ED were evaluated separately by Kruskal-Wallis test with sediment OM content (high or low) and female origin (control or transferred) as explanatory variables. Finally, hypothesis 3 (aimed at addressing if female amphipods ate during the experiment) was tested by multiple 2-way ANOVA tests using female δ¹³C and δ¹⁵N as response variables to time (initial or end of experiment) and treatment (low control, low transferred, high transferred, or high control), with interaction allowed. Female C:N was tested with a Kruskal-Wallis test as a response variable to time (initial or end of experiment) to assess changes in female amphipods regarding body condition. Finally to test hypothesis 4, juvenile and end of experiment female δ¹³C and δ¹⁵N were tested for differences with unpaired t-tests within each treatment.

Stable isotope signature of the female amphipods changed during the duration of the experiment, and this change was different among the experimental treatments. Female δ¹³C was not significantly different from the start of the experiment to the end (Fig 5; 2-way ANOVA; F_1,68 = 0.8, p = 0.37), but was significantly different between treatments (F_3,68 = 4.7, p = 0.005), with low transferred females being significantly lower compared to high control and low control. There was no significant interactive effect between treatment and time (F_1,68 = 0.44, p = 0.51) on female δ¹³C. In addition, female δ¹⁵N was significantly higher in low OM treatments compared to high OM treatments (F_3,68 = 3.4, p = 0.02), and significantly lower at the end of the experiment in comparison with the initial females ( Figure 6 ; F_1,68 = 6.9, p = 0.01), but there was no significant interactive effect between treatment and time (F_1,68 = 0.8, p = 0.37). As expected, end of experiment females had significantly lower C:N ratios than initial females (χ² = 39.52, df = 1, p < 0.001).

Juveniles were significantly depleted in ¹³C compared to end females across all treatments ( Figure 5 ; Two-sided unpaired Student’s t-test; high control: t = 6.3, df = 16.8, p < 0.001; high transferred: t = 6.5, df = 21.3, p < 0.001; low control: t = 6.4, df = 11.5, p < 0.001; low transferred: t = 4, df = 19.4, p < 0.001). For ¹⁵N, juveniles were significantly enriched compared to end females in high OM treatments ( Figure 7 ; high control: t = 3.4, df = 15.5, p = 0.004; high transferred: t = -5.05, df = 18.2, p < 0.001), but were not significantly different in low OM treatments ( Figure 7 ; low control: t = 2.1, df = 13.2, p = 0.05; low transferred: t = 0.05, df = 16.02, p = 0.96).

We found no support for a positive effect of higher sediment OM content on fecundity and viable embryos (VE) (hypothesis one), as there were no significant differences between the initial stations in terms of fecundity or VE, nor between number of hatched juveniles in the different end treatments. This could however simply be an effect of low sample size. We found similar fecundities and percent VE to previous studies (fecundity this study: 39 & 43; fecundity previous studies 29-45 in the Askö area; VE this study 91 & 94%; VE previous studies 96.4-99.7%; Elmgren et al., 1983; Sundelin, 1983; Ledesma et al., 2020). Adverse effects on fecundity at very high OM levels of more than 14% have been reported (Sundelin and Eriksson, 1998; Löf et al., 2016). However, in previous transplant experiments, Eriksson Wiklund et al. (2008) found that 10-fold higher sediment OM in the Baltic proper (5% C content compared to 0.5% in the Bothnian Sea) was not only beneficial, but critical for these amphipods to maintain fecundity. Indeed, Bothnian Sea amphipods from very low OM sediments significantly improved their fecundity when transplanted to the much higher OM Baltic proper sediment. Our results, that investigated more subtle and realistic increases in OM between two nearby stations (2-fold difference in OM and removing potential effects from differences in salinity and temperature), partly supports the positive effects of organic enrichment for offspring. Although there was no significant difference in number of offspring among our treatments, the individual mean biomass for the offspring whose mother was transplanted from low to high OM was indeed higher than in low OM. We cannot exclude that this effect is due to a somewhat earlier hatching in the high OM treatment allowing more time for feeding. Eriksson Wiklund et al. (2008) found however no effect from sediment OM on hatching time in their transplant experiment. Higher food availability in the OM treatment has therefore likely favored growth for juveniles. Another explanation could be that female amphipods can have the ability to adjust their offspring investment according to outside factors, such as food availability or exposure to contaminants (Sundelin and Eriksson, 1998; Eriksson Wiklund et al., 2008). Contaminants like polyaromatic hydrocarbons and heavy metals can accumulate and reach relatively high concentrations in some areas of the Baltic Sea Proper (Raymond et al., 2021), with reported effects on the fecundity of M. affinis (Sundelin and Eriksson, 1998; Eriksson Wiklund et al., 2008). Nevertheless, it is possible that in some cases, positive effects of higher resource availability related to moderate increases in sediment OM on fecundity may be masked by increased energy requirements for detoxification strategies (Sundelin and Eriksson, 1998; Eriksson Wiklund et al., 2008).

Organic enrichment can alter organism condition. Indeed, Eriksson Wiklund et al. (2008) found an increase in certain essential fatty acids when M. affinis were moved from low OM to high OM sediment, and a decrease in certain essential fatty acids when moved from high OM to low OM sediment. Contradicting this pattern, females in our experiment had significantly higher C:N ratios (a proxy for lipid content) in low OM sediment treatments than high OM sediment treatments, with higher C:N ratios generally indicating better body condition (Lehtonen, 1996; Drylie et al., 2020). We found that, in general, number of recovered juveniles was explained by female C:N. This agrees with theory that parents with greater resources are able to contribute more energy to reproduction (van Noordwijk and de Jong, 1986; McNamara and Houston, 1996; Marshall et al., 2010; González-Ortegón et al., 2018). Indeed, previous studies have also found positive correlations between female lipid content and number of eggs produced in amphipods (Glazier, 2000), as well as positive correlations between female shrimp trophic position and fecundity (González-Ortegón et al., 2018). The reciprocal transfer design applied here allowed a novel way to explore potential relationships between body condition and reproductive output. Females from high OM when transferred to low OM sediment showed a higher C:N ratio but lower reproductive output ( Figure 2 ), suggesting a possible higher allocation of resources to maintain body condition in females (high C:N ratio) at the expense of lower reproductive output (biomass of juveniles). Such tradeoffs in situations of low resource availability have been found in marine crustaceans (Guisande et al., 1996) but would require further investigation for M. affinis populations.

Eutrophication has been shown to simplify feeding interactions (Nordström and Bonsdorff, 2017) and increase the size of the population stable isotopic niche (Wedchaparn et al., 2016). Thus, we expected an increased niche of M. affinis in sediments with higher sediment OM content. Indeed δ¹³C ranges were larger in females in high OM sediment treatments, with females in our low OM treatments showing a 31% lower δ¹³C range compared to high sediment OM females, indicating less available resources in low OM sediment, especially when females were moved from high OM sediment to low OM sediment. Accordingly, Dolédec et al. (2021) and Karlson et al. (2014) have found a positive relationship between sediment OM and niche size in freshwater benthic invertebrates and M. affinis, respectively. However, more severe levels of OM enrichment than those tested here can also lead to a temporal homogenization of prey communities, which would lead to a narrower isotopic niche of consumers (Cook et al., 2018). This was found by Karlson et al. (2018), in which M. affinis showed a narrower isotopic niche while consuming high quality diatoms as food. The ecological consequences at a population level of these changes in δ¹³C range are however hard to predict. Despite the larger δ¹³C range (indicative of greater resource diversity), females from high OM did not show higher C:N ratios, our proxy for body condition. Similarly Karlson et al. (2015), found no relationship between niche size (also measured by δ¹³C ranges) and C:N ratios in M. affinis, indicating that increased resources would not necessarily lead to better body condition. Additionally, the significantly more enriched ¹⁵N values of females in low OM sediment treatments than in high OM sediment treatments likely indicate a greater diversity of available resources, higher trophic position of available prey in low OM sediment, or that limited resource availability could translate into starvation effects (Gorokhova, 2017; Karlson et al., 2018). We believe either of the first two explanations are more likely, as the low OM sediment shows significantly higher δ¹⁵N than high OM sediment (Fig S3) and that females in low OM sediment had the higher C:N ratios ( Figure 4 ).

M. affinis females die soon after the offspring hatch, and it is believed that they stop eating for some time before they die (Lehtonen and Andersin, 1998; Sundelin and Eriksson, 1998), thus we predicted enriched ¹⁵N at the experiment termination. We found, however, support for females feeding, as discussed above. Female δ¹⁵N significantly decreased from initial to end of experiment samples, indicating that females were not starving, leading us to reject hypothesis 3. This feeding seems to be more visible for females from low OM sediment. While female stable isotope composition did not differ significantly between initial stations, the δ¹⁵N values from females in low OM sediment treatments were significantly enriched compared to females in high OM treatments, indicating that they actively fed on the low OM sediment which had higher δ¹⁵N values. This is further indicated by the significantly depleted ¹³C in low transferred females, which suggests feeding from low OM sediments with lower δ ¹³C values. This possible plastic response of M. affinis to sediment conditions would result in the benefit of a less severe reduction in body condition (C:N), but we observed a significantly lower C:N ratio in end females compared to initial females indicating a loss of lipids, as found previously in Lehtonen (1996). It is possible that females originally from sites with high OM require more resources to maintain their body condition, therefore increasing their feeding rate in the new sediment when transferred to low OM conditions. This could explain the more pronounced change in δ¹³C and δ¹⁵N seen in females in the low transfer treatment than in females in the high transfer treatment.

We found that females and their hatched juveniles differed significantly in terms of δ¹³C across all treatments, and in δ¹⁵N in both high OM treatments. Our findings of juveniles having enriched ¹⁵N compared to females support results by Rodil et al. (2020), who found that juveniles collected in the field showed ¹⁵N enrichment and ¹³C depletion from consuming reworked OM compared to adult M. affinis. However, experimental studies with radiolabeled fresh algae has shown that juveniles feed more on fresh algae than aged sediment material (Byrén et al., 2006), and other studies on crustaceans have found depleted ¹⁵N in larvae compared to adults (González-Ortegón et al., 2018), likely because quick growth and molting leads to comparatively lower fractionation (Schimmelmann, 2011; Gorokhova, 2017). Physiological effects confounding diet interpretation can therefore not be excluded; adults have higher lipid content than juveniles with will results in depleted ¹³C values, and food limitation can result in increased δ¹⁵N values (Doi et al., 2017; Gorokhova, 2017; Karlson et al., 2018). Furthermore, we isolated M. affinis and measured their response to sediment OM enrichment while removing the effects of interspecific competition (in the case of females) and predation, as well as preventing introduction of new food sources. It is also important to recognize the variability in isotope fractionation between life stages, nutritional status of individuals, temperature, chemical exposure, and parasites. Deciphering the mechanism behind the enriched ¹⁵N of juveniles will require more experiments. Our findings are also confounded by several other factors that can affect differences between adult and juvenile stable isotope fractionation – such as food limitation. For example, lower δ¹⁵N values of starved crustaceans can be related to re-assimilation of the nitrogen in the exoskeleton, which is isotopically light (Passano, 1960; Gorokhova, 2017), and could be an explanation for lower female δ¹⁵N. In addition, we must be cautious interpreting differences between juveniles and females in isotopic signature, as we utilized different tissues for females and juveniles. Juvenile samples were composed of 13-pooled individuals, where we utilized the entire body, while female samples consisted only of the head of a single individual. Furthermore, the difference in hatching times translates to disparities in feeding time, and potential variation for stable isotope endpoints, especially when comparing to the progenitor mothers. Hence, isotope niche comparisons cannot be reliably performed between juveniles and adults in this study.

Our study provides information on the effects of OM content on amphipod diet plasticity and population dynamics, namely that M. affinis’ diet varies depending on available resources and there are potential benefits of OM enrichment for the offspring. Our results are in agreement with previous literature showing stimulation to biodiversity and increases in biomass having an effect on M. affinis at low levels of increased OM content, from 2 to 5% (Cederwall and Elmgren, 1980; Widbom and Elmgren, 1988). We found no differences in niche size that could be linked to reproductive success but our results may suggest a tradeoff between female condition and reproductive success. The results here indicate that a moderate increase of sediment OM content is potentially beneficial for reproductive success in M. affinis. Additional studies testing realistic ranges of OM enrichment and lipid analyses are required to confirm such beneficial effects, and quality of OM should also be considered in future experiments. As coastal zones face increasing OM inputs from land use changes and climate change, experiments such as this one will be important for the predictions of how populations will react, and how key species and their ecosystem functions will be subsequently impacted.