Investment in regeneration versus asexual reproduction is resource-dependent in a freshwater annelid

The post-embryonic developmental processes of regeneration and asexual agametic reproduction are widespread and often co-occur in animals. These traits are of great ecological significance, but their physiological dynamics within species are not well understood. In naid annelids, regeneration and asexual reproduction via fission are evolutionarily related and mechanistically similar yet distinct, making these animals useful systems in which to study resource allocation strategies between the two processes.
How asexual reproductive investment varies as a function of somatic investment demands was tested in the naid Pristina leidyi by repeatedly amputating the heads of individual worms, allowing regeneration to proceed, and measuring reproductive output over time. Treatments were replicated under high and low food levels to determine to what extent the investment dynamic between regeneration and fission is affected by the resource pool.
Reproductive output was affected by injury and regeneration frequency in a resource-dependent manner, such that only worms with less food availability exhibited reproductive deficits; injury and regeneration did not affect reproductive output of worms under the high food condition. When reproductive output was decreased, this occurred not through a reduction in offspring quantity but a reduction in offspring quality. In the offspring of experimental animals, body size and fission speed were dependent on parental feeding level and to a lesser and inconsistent extent on parental injury history, but regeneration speed was unaffected by parental treatment.
These findings suggest that in a species capable of both regeneration and asexual reproduction: 1) the resource pool is a key factor mediating the resource investment pattern between regeneration and fission; 2) sacrificing per-offspring investment rather than fecundity may be an optimal strategy if resources are limiting; 3) regeneration and fission have evolved distinct resource allocation pathways. This work prompts further questions about the physiological dynamics between regeneration and asexual reproduction in animals, such as whether and to what extent these have evolved adaptively, including in response to injury and resource pressures.

Methods
Animal culture and material
The use of animals in this study did not require approval from an ethical committee. Established cultures of P. leidyi (see Bely & Wray, 2001) were cultured at room temperature (~23 ⁰C) in glass bowls (12 cm diameter) filled with ~150 ml of artificial spring water (1% artificial seawater) (ASpW). Strips of brown paper towels were provided as substrate. Cultures were fed once weekly with 10 mg powdered Spirulina. Half-volume water changes were administered weekly.

To generate experimental animals, 109 healthy-looking worms of similar size were pulled from a culture that had been established 3 weeks prior, to ensure that the culture was undergoing active growth. These worms were moved to individual wells of 24-well plates filled with 1.5 mL ASpW. Each individual worm was fed once weekly with 0.15 mg Spirulina. Individuals were monitored daily for fission (zooid release). P. leidyi typically forms fission zones (FZ) at approximately two-thirds the length of the body, within segments 14–16. Because it possesses the original head and represents the larger fission product, the anterior zooid is considered the “parent” and the posterior zooid, once released, is considered the “offspring”, by convention. The first posterior zooid released by each worm was moved to a new 24-well plate, and these 109 young worms were used as experimental animals (F0) as described below and illustrated in Fig. 1. All F0 worms were produced within 5 days of each other (Fig. 1a). Individual P. leidyi are estimated to have a lifespan of approximately 1 year (Bely, AE, unpublished data) and F0 worms born within days of each other were thus considered of approximately equivalent age.

Experiment One: Effect of injury and feeding on survival and reproduction 
An experiment with a 2×4 factor design, including two food levels and four injury levels, was conducted to test the effects of anterior amputation injury (and regeneration) frequency on reproductive output and how these are modulated by feeding.

F0 worms were randomly assigned to one of two feeding treatments: low food (LF), which received 0.15 mg Spirulina on the first day of each week (n = 55); and high food (HF), which received 0.15 mg Spirulina on the first and fourth days of each week (n = 54). Full water changes were administered for all worms on the first and fourth day of each week.

Worms in each feeding group were randomly assigned to one of four injury treatments: uninjured controls (0X) and worms injured once (1X), twice (2X), or three times (3X) during the experiment (Fig. 1a). Worms were divided approximately evenly among these four treatments (low food: n = 14, 14, 12, 14; high food: n = 13, 14, 12, 16, respectively). Injuries were all anterior amputations, inflicted by cutting at a consistent position at the anterior end of F0 worms. Specifically, at each amputation time point, all worms (regardless of treatment) were anesthetized in 0.05 mM nicotine and worms designated for injury were amputated at the junction of segments six and seven using a scalpel (Fig. 1b). The excised anterior portion was discarded and the remainder of the worm, representing the posterior-most ~3/4 of the worm body, was retained for the experiment.  Note that worms can continue fissioning even after amputation of the anterior part of the body. Three rounds of amputation occurred across the experiment: at week 3 (day 19 of the experiment, beginning with the birth of the first F0 worm), week 6 (day 40 of the experiment), and week 9 (day 61 of the experiment). At each of these time points, all worms were anesthetized and worms designated for injury at that time were amputated. All worms were then transferred to clean 24-well plates in fresh ASpW and all worms (including those that were not amputated) were not fed for one week, while injured animals regenerated. Designated feeding schedules resumed after that week. At the first amputation time point (week 3 of the experiment, ~2 weeks following the “birth” (release) of the last F0 worm), worms in the 1X, 2X, and 3X treatments were amputated. At the second amputation time point (week 6 of the experiment), worms in the 2X and 3X treatments were amputated. At the third amputation time point (week 9 of the experiment), only worms in the 3X treatment were amputated. The experiment was concluded on day 100 of the experiment.

Each F0 worm was scored daily for survival and fission (release of a posterior zooid, designated F1) across the duration of the experiment. The first F1 produced by each F0 worm, and the first F1 produced following each round of amputation (regardless of whether the individual was in an injury treatment or not), was removed, imaged, and assigned to experimental treatments as described below (Fig. 1c). All other offspring were discarded upon discovery. Throughout this paper, the following shorthand is used to refer to time intervals during which offspring production was recorded: t0 - from F0 birth until the day of first injury; t1 - from 1 week after first injury until the day of second injury; t2 - from 1 week after second injury until the day of third injury; t3 - from 1 week after third injury until the conclusion of the experiment. These intervals are collectively referred to as the F1 “birth periods”.

Experiment One: Offspring quality assessments 
During Experiment One, the first F1 worm produced by each F0 worm at the beginning of the experiment and following each regeneration period, or at each equivalent time point for uninjured worms, was collected. ­Offspring quality was assessed by measuring offspring volume, reproductive speed, and regeneration speed, as described further below.

Offspring volume was assessed on the day of discovery. On that day, offspring were anesthetized and imaged under a 2.5x objective using a Zeiss Axioplan 2 microscope running AxioVision 4.8 imaging software. Body dimensions were measured with ImageJ using the Fiji package (v1.47f ) (Schindelin et al., 2012) and volume was approximated by using the formula for a cylinder (πW2L) (Fig. 1d). To calculate total body volume, worm length and width were measured. Whole worm length (L) was measured by tracing from the base of the proboscis to the tip of the pygidium along the anteroposterior axis. Whole worm width (W) was averaged across three roughly equally spaced positions along the body, as worms vary slightly in diameter along the body length. Especially under higher food conditions, P. leidyi can develop multiple FZs along the anteroposterior body axis simultaneously, such that new offspring sometimes already possess one or occasionally even two FZs when they are released from the parent. The total volume contained within all FZs (if present) in an offspring was separately calculated (if more than one FZ was present in an individual, the volumes of all its FZs were pooled to produce a single measure of FZ volume). Note that FZs, even early stage ones, are easily identified by differences in opacity and slight narrowing of body width relative to older tissues. Total FZ volume was calculated by measuring, for each FZ, the length and width of the FZ, with width measured around the center of the FZ. Note that imaging for body size measurements did not begin until partway through t0, leading the sample size for F0 body volume to be slightly smaller than the sample size for F0 fission speed.

Following imaging, worms were randomly assigned to one of two assay groups: an “offspring fission speed” assay and an “offspring regeneration speed” assay (Fig. 1c).

For the fission speed assay, F1 worms were placed individually in fresh 24-well plates filled with ASpW and maintained on the same feeding regimen as their respective F0 worm. F1 worms were scored daily to determine both the date of a first FZ appearance and the completion of the first fission (release of a “F2” offspring). Time (# days) from “birth” to formation of a first FZ and from formation of the first FZ to zooid release (fission speed) were recorded and their sum calculated (Fig. 1c). After fission, the F1 worm and all of its offspring were discarded.

For the regeneration speed assay, within an hour following imaging, F1 worms were amputated in the manner described previously and transferred individually to fresh 24-well plates filled with spring water. Worms were not fed during this experiment. Worms were checked daily for two indicators of full regeneration: the emergence of visible chaetae in the new anterior segments (four in total) and the emergence of the proboscis. Time from injury to full regeneration (# days) (regeneration speed) was recorded (Fig. 1c). At the completion of regeneration, worms were discarded. If amputated F1 worms produced offspring, which was common within 1-2 days following amputation if a fission zone was present at birth (itself common in F1 derived from HF F0), the date of that offspring release was noted, and such offspring were discarded.

Experiment Two: Direct effect of feeding on regeneration speed
To provide additional context for interpreting offspring regeneration speed, a follow-up experiment was run to assess how feeding level directly affects regeneration speed of individual worms (rather than that of their offspring). For this experiment, 24 worms were pulled from a culture (maintained as above), and 12 worms were each randomly assigned to LF and HF levels, as described above. Worms were maintained individually and fed under LF and HF conditions for two weeks and then amputated in the same manner as described above. Worms were scored daily for regeneration completion, as described above. Three LF worms and one HF worm died following amputation and were excluded from analysis.

Statistical analysis
All statistical analyses were performed in the R computing environment (R Development Core Team, 2019). Most analyses were performed using generalized linear (mixed) models (GLMMs) via the lme4 package (Bates et al., 2015), and the 95% confidence intervals (CIs) for fitted values were estimated via the ciTools package (v0.6.1; Haman & Avery, 2020). The significance of explanatory variables in all models was determined with a Wald test. Pairwise comparisons were accomplished using the emmeans package (v1.8.9; Lenth, 2023), when appropriate. Analyses of F0 worm data included tests of the effects of feeding and injury on survival and fecundity. A GLMM with a binomial error distribution (logit link) was constructed to test differences in F0 survival over the course of Experiment One between feeding and injury treatments, with the response variable being whether or not a worm survived following a particular injury (and lived to the next injury, if applicable). GLMMs with a Poisson error distribution (log link) were constructed to test differences in fecundity (# of offspring produced) between feeding and injury treatments for different periods of time: t0, t1, t2, t3, and total duration of the experiment. Worms were randomly assigned to plates based on injury treatments but not feeding, such that some plates only held high-food worms and others only low-food worms, and worms were maintained in the same relative well position across the experiment. Therefore, inclusion of plate as a random effect in a mixed model created a singularity with feeding. A nested model using plate as a fixed effect to substitute for feeding was constructed to test whether there were significant differences between plates of the same feeding group. As no significant contribution of plate distinguishable from a feeding effect was apparent, and identical results were obtained between the singular mixed-effects models and reduced GLMs without the plate variable, these latter models were used for the final analyses.

Analyses of F1 worm (offspring) data included tests on the effect of parent feeding and injury treatments on offspring body volume, fission speed and associated measures, and regeneration speed. Offspring birth period (t0, t1, t2, or t3) was additionally modeled as a fixed effect in order to facilitate comparisons between offspring from recently- and not recently-injured F0 worms of the same approximate age as a control. A GLMM with a gamma error distribution (log link) was constructed to test differences in body volume, with the volume of any FZs present included as a fixed effect. A binomial GLMM was used to test the probability of F1 being born with a FZ already present. Time from birth to FZ detection in the fission speed assay (fission latency period), time from FZ detection to fission completion in the fission speed assay (fission speed), and time from injury to regeneration completion in the regeneration speed assay (regeneration speed), were each analyzed via the construction of GLMMs with a Poisson error distribution (log link). Inclusion of the identity of the F0 worm that a F1 worm was born from as a random effect in these models, with the exception of those for body volume, led to singular fits, as worm identity was not sufficiently replicated across levels of feeding. Therefore, F0 identity was modeled as a fixed effect to check for outlier contributions of individual worms to these various responses. As no outliers were detected through this method, and identical results were obtained between the singular mixed-effects models and reduced GLMs without F0 identity, these latter models were used for the final analyses with the exception of those for body volume, which was analyzed via a GLMM as described above.

In Experiment One, although each worm was assigned to a particular injury treatment at the start of the experiment, amputations were performed sequentially over time such that at any particular time point in the experiment, a worm may have only experienced a subset of its planned amputations (Fig. 1). For example, at the end of t1, the maximum number of anterior amputations that a worm could have experienced during the experiment is one; therefore, the real injury history of all worms assigned to the 1X, 2X, and 3X injury treatment groups at this point is effectively identical. Given this experimental design, data were analyzed in two different ways: 1) with worms strictly assigned to their initial injury treatment group, and 2) with worms pooled based on their actual injury history at the given birth period in the experiment (thus maximizing effective sample sizes). For the latter approach, for each birth period, the actual injury history of an F0 worm was considered to be its injury treatment. The data and results of these pooled analyses are presented in the main text and figures. The data and results of unpooled analyses, in which worms are analyzed at each time point based on the initially assigned injury treatment, are presented in the Supporting Information and mentioned in the main text when conclusions differ from those of the pooled analyses.

The effect of food level on regeneration speed in Experiment Two was tested using a Welch’s t-test.