Larval Survivorship and Settlement of Crown-of-Thorns Starfish ( Acanthaster cf . solaris ) at Varying Chlorophyll Concentrations

The dispersal potential of crown-of-thorns starfish (CoTS) larvae is important in understanding both the initiation and spread of population outbreaks, and is fundamentally dependent upon how long larvae can persist while still retaining the capacity to settle. This study quantified variation in larval survivorship and settlement rates for CoTS maintained at three different chlorophyll concentrations (0.1, 1.0 or 10.0 μg.L-1), achieved by varying densities of single-celled flagellate phytoplankton, Proteomonas sulcata. Based on the larval starvation hypothesis we expected that low to moderate algal concentrations would significantly constrain both survival and settlement. CoTS larvae were successfully maintained for up to 50 days post-fertilization, but larval survival differed significantly between treatments. Survival was greatest at intermediate (1.0 μg.L-1) chlorophyll concentrations, and lowest at highest (10.0 μg.L-1) chlorophyll concentrations. Rates of settlement were also highest at intermediate (1.0 μg.L-1) chlorophyll concentrations and peaked at 22 days post-fertilization. Peak settlement was delayed at low chlorophyll concentrations, probably reflective of delayed development, but there was no evidence of accelerated development at high chlorophyll concentrations. CoTS larvae were capable of settling 17-43 days post-fertilization, but under optimum conditions with intermediate chlorophyll concentrations, peak settlement occurred at 22 days post-fertilization. Moderate increases in nutrient concentrations and algal densities may increase the number of CoTS that effectively settle, but are unlikely to influence dispersal dynamics.


Introduction
Sessile and benthic marine invertebrates are fundamentally dependent on the larval phase of their lifecycle for dispersal away from natal reefs, which is important for enabling colonization of new habitats, recolonization following population depletion, and genetic exchange among sub-populations [1,2].Despite the short larval duration of most marine organisms (days to months), larvae may be dispersed over great distances [2].Importantly, ecologically and evolutionarily significant levels of genetic exchange occur over very large (even oceanic) scales (e.g., [3]).There is however, evidence for some species that most of the larvae (up to 60%) settling on a given reef are of local origin [4], implying that most larvae may not even travel beyond the confines of a single reef.Ultimately, there may be two distinct modes (short versus long, retention versus dispersal, or self-recruitment versus departure) in the range of dispersal distances for individual species [5].
Variation in dispersal within and among cohorts of larvae will have a critical influence on population dynamics and persistence for widespread species [5].
For crown-of-thorns starfish (Acanthaster spp.), hereafter referred to as CoTS, the proportion of larvae that are retained versus dispersed is important in understanding both the initiation and spread of population outbreaks [6][7][8].Strong larval retention is fundamental to the progressive accumulation of CoTS within a given location, which is increasingly viewed as the predominant mechanism that gives rise to primary outbreaks [6,7] On the Great Barrier Reef (GBR), Woolridge and Brodie [7] suggested that rates of larval retention vary among years with changes in ENSO driven ocean current velocities, and that initiation of outbreaks is coincident with periods of "strong local clustering", where there are high levels of retention at individual reefs or within tightly packed reef clusters.Once established, initial outbreak populations give rise to large numbers of larvae that invariably spread to other reefs.Moreover, dispersal to other downstream (generally southerly) reefs is facilitated by increases in ocean currents and relaxation of the strong local clustering [7].This larval retention hypothesis [7] is particularly tenable because it explains why major flood events and nutrient enrichment in near reef waters do not necessarily initiate outbreaks of CoTS, but may facilitate rapid proliferation and spread of outbreaks across the GBR.There are nonetheless important questions about developmental rates for crown-of-thorns starfish larvae and how far they can disperse, which are important for understanding the initiation and spread of population outbreaks.
Two fundamental parameters that influence the retention versus dispersal of marine larvae are i) the minimum pre-competency period, which is the time taken (in hours or days) for larvae to complete necessary development prior to being physiologically and morphologically capable of settling, and ii) the maximum competency period, which is the longest period that larvae can remain in the plankton (often dictated by their initial energy reserves and/ or capacity to sustain themselves) and still retain the ability to successfully settle.The minimum pre-competency period has a strong bearing on the extent to which larvae are likely to retained on their natal reef(s), whereas the maximum competency period is fundamental in establishing how far larvae could potentially spread.Marine fishes typically exhibit very narrow ranges in their planktonic larval duration (PLD) (ie., only a few days separate the maximum competency period and minimum pre-competency period), especially for species with short pre-competency periods [9].For fishes, therefore, individual variation in dispersal is largely attributed to traits that influence their physical capacity (e.g., swimming performance) to extend dispersal distances despite limited differences in larval duration [5].Marine invertebrates however, may be competent for highly protracted periods, such that individuals from the same cohort may settle anywhere from several hours to many months after fertilization [10,11] depending on their nutritional status and exposure to settlement cues.
Acanthaster spp.have been successfully cultured under laboratory conditions since the early 1970s [12], largely for the purposes of understanding their developmental biology.Initial rearing experiments, where larvae were maintained at 24°-25° C for the formative stages of development (until larvae reached late bipinnaria at day 21) resulted in PLD estimates of 30-47 days, with a mean of 38 days [12].However, CoTS generally spawn only when sea surfaces temperatures are ≥27 o C [13], and larvae develop more quickly at higher temperatures [14,15].Lamare et al. [14] showed that development rates of CoTS from Australia's Great Barrier were maximized at or above 28.7 o C, at which temperrature, larvae may complete development and be competent to settle within 12-14 days [16].
Aside from effects of temperature, variation in larval survivorship and rates of larval development for CoTS are also influenced by nutrient concentrations and food availability (e.g., [17]).One of the foremost hypotheses asserting that outbreaks of CoTS are exacerbated (if not caused) by anthropogenic activities, the larval starvation hypothesis, links the initiation and/ or spread of outbreaks to elevated nutrient concentrations caused by intensive agriculture in adjacent catchments [18,19].This idea is largely predicated on marked differences in the proportion of CoTS larvae that complete development over relatively moderate ranges of chlorophyll concentrations [19].Fabricius et al. [19] showed that developmental success is effectively zero at <0.25 µg.L -1 chlorophyll compared to 100% at 2-4 µg.L -1 chlorophyll.It is suggested therefore, that elevated chlorophyll concentrations caused by run-off from modified catchments during intensive storms will reduce constraints on larval survivorship and recruitment imposed by normally low-levels of nutrients and food [19].More recent studies have however, shown that very high levels of nutrients are deleterious to larval development for CoTS, while larval growth and development are maximized at close to normal background concentrations of chlorophyll [20].These studies [17,20] compare the size and fate of CoTS larvae at specific time intervals (e.g., at 4 days old) across different nutrient concentrations, but it is unknown to what extent nutrient concentrations may influence the competency periods, and therefore dispersal potential, of crown-of-thorns larvae.
In this study, we quantify variation in larval survivorship and settlement rates for CoTS maintained at three different food treatments.Based on the larval starvation hypothesis (e.g., [19]) we expected that starfish subject to very low nutrient regimes (0.1 µg.L -1 chl a) would develop more slowly and have lower survivorship compared to larval cultures maintained in higher concentrations of microalgae.Accordingly, we expected that increases in nutrients (and food availability) would increase rates of development, thereby reducing the minimum pre-competency period.High nutrient concentrations meanwhile, are also expected to greatly extend the maximum competency period, both by increasing rates of survivorship and provisioning larvae with resources necessary for metamorphosis at settlement.It is known however, that excess food can inhibit filter feeding and digestion by CoTS larvae, ultimately leading to arrested development and high levels of mortality [17].While previous studies (e.g., [17,20]) have compared the size, development and survival of larvae of CoTS at set intervals in their development (to explicitly test the effects of food availability among individuals of equivalent age), this study looks at overall longevity and competency of larvae to test whether increased food availability will expedite development, potentially leading to higher rates of larval retention.Alternatively, elevated nutrients and food availability may extend the maximum competency period for CoTS larvae and therefore, facilitate greater dispersal.

Larval culture and settlement experiments
Adult Acanthaster cf.solaris [21] were collected from the GBR near Cairns, Australia (16°55'S, 145°46'E) by control divers employed by the Association of Marine Park Tourism Operators (AMPTO) and air freighted to the National Marine Science Centre, Southern Cross University at Coffs Harbour, NSW, Australia.Eggs from three female starfish were collected, pooled and fertilized using sperm from three males, following Kamya et al. [22].Fertilized embryos were reared for the two days in an aerated 300 L cylindro-conical tank at ~28 °C at a density of 2.5 larvae.mL - .After 2 days (coinciding with the onset of feeding) larvae were distributed among replicate 1 L cylindro-conical containers and subject to one of three different feeding/ nutrient treatments.Larvae were reared in 1 L cylindro-conical containers in filtered (5 µm) seawater at a density of 1.2 larvae.mL - and fed Proteomonas sulcata at three chl a levels; 0.1, 1.0 or 10.0 µg.L -1 , equivalent to 1, 10 and 100 × 10 3 cells.mL - , respectively (Wolfe et al. 2015).Proteomonas sulcata is a single-celled flagellated phytoplankton that is 7-10 µm long, and within the size range of food particles readily consumed by CoTS [23].There were 10 replicates for each distinct algal density and chlorophyll concentration.Containers were maintained in a climate-controlled room at 28.5 °C (±0.2 °C).Larvae were fed daily after a 100% water exchange.Larval cultures were moved to new containers every 2-3 days.
To follow survival, the number of larvae in each container was quantified every 2-3 days by extrapolating the number of larvae counted in 20 mL subsamples taken from each container.However, when larval density was < 0.1 larvae.mL - all of the larvae were counted.Survival rates were calculated as the number of larvae remaining out of the total added to the container at the start of the experiment, while also adjusting for larvae removed from the containers for settlement assays and for larvae that spontaneously settled in the containers, from 17 days onwards.
Settlement assays were undertaken from 17 days post-fertilization, which was the first day at which late-stage brachioloaria with well-developed rudiments were observed across all treatments, thereby enabling comparisons of settlement rates across treatments.Settlement assays were conducted in 36 mm plastic petri dishes (Sarstedt) containing 4 mL of fresh seawater, and a 1 cm 2 crustose coralline algae (CCA) encrusted polycarbonate plate to induce settlement [24].For each assay, there were three replicates from each larval container with 10 larvae per replicate, although this was reduced to one replicate when there were less than 100 larvae in a container.Settlement rates were quantified after 48 h.All newly settled, 5-arm juveniles were photographed in seawater, aboral side up, and flat to the plane of focus using a camera (Olympus DP26) mounted on a stereomicroscope.Area of the juveniles was measured from the photographs using Image-J image analysis software (NIH, USA).Settlement assays were conducted 17, 22, 29, 36 and 43 days post-fertilization.Thereafter, there were too few larvae in any of the treatments to effectively test for settlement.

Statistical analyses
Larval survival was estimated from sub-sampling rather than a complete census, so traditional methods of survival analysis were not appropriate.Instead, we modeled counts using non-linear splines (GAMMs; [25]) assuming Poisson error and a log link function.More specifically, we examined counts of surviving larvae using models with a categorical treatment effect, and treatment specific cubic regression splines; splines were estimated with shrinkage, meaning that smoothness selection can zero a term completely [25], and the number of knots was set to 4 for all smooths to prevent over-fitting.Additionally, replicate was fitted as a random effect.All models were estimated using the gamm4 package [26].Overall goodness-of-fit was checked using standard techniques (e.g., residuals plotted against all explanatory variables and fitted values from the models).
Settlement assay data were reduced to observations where the numbers of larvae available per settlement assay were known before analysis.This resulted in a data set of 292 observations for all age-treatment combinations, with replication at each combination ranging from 1-30.Due to this unbalanced design, we modeled the probability of successful settlement using a mixed effects model with binomial error and a logit link function.Age (continuous) and treatment (categorical: three levels) were modeled as fixed effects; replicate and assay were modeled as random effects, with assay (up to 3 per replicate) nested within replicate (up to 10 replicates per age-treatment combination); age was centered by subtracting the mean age from all observed ages before model fitting.To examine potential nonlinearities in settlement probability with larval age and age-by-treatment interactions, we fit a model with a second-order term for age, and an age-treatment interaction term.Initial model fits suggested that data were over-dispersed, and the beta-binomial was supported over the binomial distribution (likelihood ratio test: Χ 2 = 84.62,df = 1, p < 0.001).Models were estimated using maximum likelihood in the glmmADMB package [27,28] in R 3.3.1 [29].We assessed the significance of fixed effects using type II Wald's chi-square tests, and the fit of models and their conformity to statistical assumptions were visually inspected using standard techniques (i.e., plots of residuals versus fitted values, and all covariates).

Larval survival
Larval crown-of-thorns starfish were successfully maintained in 1 L cylindro-conical containers for up to 50 days post-fertilization, though most larvae had died or settled within 30-40 days.Larval survival differed significantly between treatments (Χ 2 =470, df = 2, p < 0.001), and was greatest at intermediate (1.0 µg.L -1 ) chlorophyll concentrations, followed by the low and the high chlorophyll concentrations (Table 1; Figure 1).There were initially high rates of larval mortality across all treatments, but survivorship increased (between approximately 15 and 22 days) at low to intermediate chlorophyll concentrations, before increased rates of mortality after 30 days (Figure 1).Larval survival initially decreased at similar rates for the high and low food treatments, but the Peer-reviewed version available at Diversity 2017, 9, 2; doi:10.3390/d9010002initial loss of larvae was slower in the medium food treatment.At high algal concentrations, high rates of early post-fertilization mortality continued in an exponential-like decay throughout the experiment, such that the maximum longevity of larvae was only 36 days.

Probability of successful settlement
Although larvae were maintained for up to 50 days post-fertilization, too few larvae persisted beyond 43 days post-fertilization to enable effective tests of competency.Even at 43 days post-fertilization, few of the larvae tested were capable of settling, and mainly from the low algal food treatment.The probability of a larva successfully settling depended on both age and food treatment (Figure 1).We found strong evidence for settlement probability being a non-linear function of age (age 2 term was significant: Χ 2 =6.35, df = 1, p = 0.01), and there was strong support for an interaction between food treatment and age (Χ 2 =15.45, df = 2, p < 0.001).Settlement probability was highest for the medium food treatment (1μg/L) and peaked at around 22 days.Settlement probability was also greatest around the 22-day point for the high food (10 μg/L) treatment, but declined precipitously thereafter and none of the larvae from this treatment settled on days 36 and 43.For larvae from the low food treatment (0.1μg/L chlorophyll A), settlement probability peaked at around 30 days post-fertilization, and some larvae successfully settled at day 43.However, peak settlement probabilities in the low food treatment were much lower than those observed in the intermediate food treatment (Figure 1; see Table 1 for parameter estimates).
The algal concentrations to which larvae were exposed appeared to affect the size of CoTS that settled at 17 days, whereby starfish from the low (0.1 µg.L -1 ) chlorophyll concentration were one-third of the size (0.13mm 2 ± 0.03 SE; n = 2) of the settled starfish from intermediate (0.43mm 2 ± 0.60 SE; n = 10) and high (0.44mm 2 ± 0.56 SE; n = 8) algal concentrations.However, too few larvae settled in the low food treatment to enable robust statistical comparisons of size at settlement, and any effect of algal treatment on size at settlement disappeared after 22 days post-fertilization.

Non-linear effects of increasing food availability on larval survivorship and settlement
This study shows that variation in algal densities and corresponding phytoplankton concentrations has a significant effect on both larval survivorship and settlement rates for CoTS, as would be expected for planktotrophic larvae [17,20].However, survivorship and settlement rates were maximized at intermediate (1.0 µg.L -1 chlorophyll; reflective of 10 × 10 3 cells.mL - of Proteomonas sulcata) algal concentrations, and lowest at high (10 µg.L -1 chlorophyll reflective of 100 × 10 3 cells.mL - of Proteomonas sulcata) algal concentrations.There is little doubt that very low algal concentrations (0 -0.01 µg.L -1 chlorophyll) constrain food intake by CoTS [17,20], and thereby potentially constrain growth, development and survival.However, CoTS larvae may be very resilient to limited food availability (Olson 1987) and/ or derive nutrients from prior maternal provisioning [30] or from other exogenous food sources [31,32], especially in the wild.In the current study, high algal concentrations (10 µg.L -1-chlorophyll) were more deleterious for larval survival and settlement than the lowest algal concentration (0.1 µg.L -1-chlorophyll) tested.
Deleterious effects of high algal concentrations on planktotrophic larvae of CoTS were first apparent based on sustained declines in filtration rates with increasing algal concentrations above 100 cells.mL - [17].At very high algal concentrations (> 5 × 10 3 cells.mL - ), Lucas [17] noted that ingestion rates tended to plateau and only a fraction of algal cells were actually digested.Accordingly, CoTS larvae exposed to algal concentrations of > 5 × 10 3 cells.mL - for Dunaliella primolecta and > 1 × 10 4 cells.mL - for Phaeodactylum tricomutum had arrested development and poor survival [17].Larvae may be overfed at extremely high algal concentrations, and the alimentary canal is so congested with cells that digestion is impeded and rates of assimilation are substantially reduced [17].In bivalve cultures, overfeeding of larvae has resulted in increased incidence of bacterially related diseases [33].However, it was considered very unlikely that CoTS larvae would ever be exposed to supra-optimal algal concentrations, and much more likely that larval development and survival would be constrained by low algal concentrations and limited food availability [17].Similarly, the highest algal concentrations used in this study (10 µg.L -1 chlorophyll reflective of 100 × 10 3 cells.mL - of Proteomonas sulcatq) are well beyond what would normally be expected on mid-shelf reefs of the GBR [34].
Chlorophyll concentrations recorded on or adjacent to coral reefs generally range from 0.2 to 0.6 µg.L -1 [32,34], which are below optimum levels for growth and survival of CoTS larvae (1.0-4.0 µg.L -1-chlorophyll; [19,20]).At mid-shelf reefs on the GBR, for example, chlorophyll concentrations generally range from 0.2 to 0.5 µg.L -1 (e.g., [34,35]) except on inshore (coastal) reefs or during extreme flood events.However, these normal background chlorophyll concentrations are still greater than minimum thresholds for survival and effective settlement of CoTS larvae.In our study, rates of survival for larvae up until 29 days post-fertilization (which is well beyond the period of peak settlement) at low algal concentrations (0.1 µg.L -1-chlorophyll) was still >50% of that for 1.0 µg.L -1-chlorophyll (Figure 1).This suggests that even during "normal" conditions a significant portion of CoTS larvae could effectively settle on mid-shelf reefs on the GBR.While moderate increases in algal concentrations may enhance survival and settlement of CoTS, thereby potentially contributing to the initiation [19] and/ or spread of population outbreaks [13], it seems unlikely that normally low nutrient concentrations and algal densities would be sufficiently low to actually prevent any CoTS larvae from completing development and effectively settling.
During extreme flood events, which are often considered to be potential triggers for initiating outbreaks on the GBR [7,19], chlorophyll concentrations on inshore and mid-shelf reefs can ≥10 µg.L -1 [36].In the extreme, chlorophyll concentrations ≥18 µg.L -1 have been recorded along the leading edge of flood plumes [36], where conditions are optimal for phytoplankton production.Given increasing evidence of deleterious effects of high algal densities and chlorophyll concentrations, such events may actually constrain larval development and settlement for Acanthaster spp, rather than initiate outbreaks.Major flood events will also reduce salinity and increase levels of suspended sediment, which are also deleterious for development and survival of CoTS larvae (Caballes et al., Unpublished manuscript).It is clear however, that survival and settlement of CoTS larvae increases with moderate increases in chlorophyll concentrations (though limited data to resolve the optimum levels for performance of CoTS larvae) and given their exceptional reproductive potential [37], even small changes in the proportion of larvae that complete development and effectively settle will significantly affect the incidence, if not severity, of population outbreaks.

Minimum and maximum competency periods
This study extends the maximum-recorded longevity (to >50 days) for CoTS larvae, and provides the first explicit evidence of extended competency for Acanthaster spp., up until 43 days post-fertilization.However, previous studies have tended to focus on the minimum pre-competency period [15] or minimum time taken for larvae to develop into late-stage brachiolaria [12], and generally do not consider the maximum competency period.In the current experiment, declines in the number of larvae after 30 days were caused by sustained settlement of competent larvae in experimental containers, as well as also punctuated mortality events ("crashes") likely caused by instabilities in conditions.Therefore, under ideal conditions and with better constraints on spontaneous settlement, larval longevity for CoTS might extend well beyond 50 days.Prolonged survival and competency will theoretically extend the maximum distances over which larvae can be dispersed [38], though maximum rates of settlement will nonetheless occur after relatively finite periods and probably close to natal reefs.At the medium (optimal) algal concentrations, after accounting for both larval survival and settlement rates, we found that peak settlement for A. cf.solaris tended to occur between 17-22 days.It is possible, therefore, that few larvae would ever travel beyond the confines of individual reef systems, especially on the GBR, where there is high water retention and extensive reef habitat [39].Rather, relatively rapid development of CoTS may promote high levels of self-recruitment, or at least significantly constrain dispersal among reefs.This study had limited capacity to resolve differences in the minimum pre-competency period for CoTS larvae exposed to three different chlorophyll concentrations.Rather, competency was established only when late-stage brachiolaria larvae were observed across all treatments, and thereby underestimates of the minimum pre-competency period (17 days).However, very few larvae in any treatment had completed development by 16 days post-settlement so few if any, CoTS larvae would have effectively settled if assays had been conducted earlier.Previous studies have shown that CoTs larvae can reach the late-stage brachiolaria stage within 9-11 days [40], though settlement has only ever been documented ≥14 days post-fertilization [16].This minimum pre-competency period is short relative to other echinoderms with planktotrophic larvae [41], which will enhance larval retention and relatively short-distance dispersal.On the GBR, the step-wise southerly progression of outbreaks suggests that dispersal is limited to 1-2 degrees of latitude (~100-200 km) per generation [42,43], potentially representing the maximum geographic extent of dispersal for individual larvae.
Genetic differentiation apparent when comparing among populations of CoTS in different geographic regions (1,000's of kilometers apart) suggest there are definitive constraints on larval dispersal at oceanic scales [44,45].In the extreme, it is now clear that there are several distinct species of Acanthaster [46,47], which must be maintained by limited larval and genetic exchange at these ocean scales.Within regions however, there is often very limited genetic structure [44,[48][49][50], indicative of extensive genetic exchange via widespread dispersal of larvae.Timmers et al. [48] investigated genetic connectivity among CoTS populations in the northern Pacific, and revealed extensive gene flow along the 2500km length of the Hawaiian archipelago, as well as between Hawai'i and Johnston Atoll separated by 865km.Similarly, Harrison et al. [50] found no genetic structure (using 27 optimized microsatellite loci) for outbreak populations of CoTS along the length of Australia's GBR.Distinct differences in levels of genetic structure within versus among regions, suggests that larvae are sufficiently long-lived so as to be effectively dispersed among reefs within regions, but largely incapable of dispersal among regions.
Existing connectivity models for CoTS [7,8] use idealized particle dispersal periods (generally equivalent to PLDs) ranging from 1 to 28 days.While these particle dispersal periods were largely determined by logistical constraints of the models, rather than the larval biology of the study species, our study suggests that most CoTS larvae will settle within 17-22 post-fertilization.Low productivity of near-reef waters [34] may however, extend larval development and peak settlement periods.It is also possible, though improbable, that some larvae could settle after >43 days in the plankton.It will be interesting to test whether refining these connectivity models to take account of more constrained estimates of settlement windows will alter predictions of connectivity and patterns in the spread of outbreaks.

Conclusions
This study revealed significant differences in both larval survivorship and settlement rates for CoTS across two orders of magnitude variation in chlorophyll concentrations (0.1-10 µg.L -1 ).As expected, based on the larval starvation hypothesis, CoTS larvae exposed to very low algal concentrations (equivalent to 0.1 µg.L -1 chl a) had lower survivorship and delayed development compared to larvae maintained in intermediate algal concentrations (equivalent to 1.0 µg.L -1 chl a).Conversely, very high algal concentrations (equivalent to 10 µg.L -1 chl a) did not accelerate development of CoTS larvae, nor extend their maximum competency period.Rather, very high algal concentrations had an overwhelmingly negative effect on CoTS larvae, resulting in much higher mortality rates from very early in larval development, compared to low (0.1 µg.L -1 chl a) and intermediate (1.0 µg.L -1 chl a) algal concentrations.This study reaffirms that moderate increases in chlorophyll concentrations (up to 1.0-4.0µg.L -1-chlorophyll; [19,20]) will lead to increases in rates of development, survival and settlement for CoTS larvae.However, low chlorophyll concentrations (0.1 µg.L -1 chl a) will still enable significant rates of population replenishment and settlement, while the greatest constraints on growth and survival of CoTS larvae occurred at very high chlorophyll concentrations and algal densities.

Table 1 .
Estimates of the a) parametric coefficients and b) the approximate significance of the smooth terms of the survival model (Poisson error distribution, log link functions).Parametric intercept term represents the low food treatment.