Diatoms are among the most important photoautotrophic organisms, driving food web dynamics in some of the most productive marine systems, particularly in areas of nutrient upwelling. The classical view of aquatic food webs generally considers diatoms as passive participants in energy transfer, being heavily grazed by their predators and exhibiting little or no anti-grazing capacity [1,2]. Based on this paradigm, little consideration has been given to the possible evolution of diatom defensive strategies, in particular chemical defences, and the consequent role of such defences in regulating energy flow. Diatoms do posses defensive capabilities, most notably in the form of mechanical protection conferred by the silica frustule [3]. It is now apparent that complex and highly evolved chemical defences may also be in operation in many species [4–6]; however the topic of diatom chemical ecology is not without controversy [7–12]. Somewhat paradoxically, other autotrophs, primarily dinoflagellates and cyanobacteria, are renowned for producing highly toxic biomolecules [13], some of which have been linked to reproductive failures. The ingestion of toxic dinoflagellates by female copepods has been suggested detrimental for nauplii hatching success [14]. Yan et al. [15] demonstrated that intact cells and cellular fragments of the saxitoxin producing dinoflagellate Alexandrium tamarense inhibited hatching success and larval survival of the scallop Chlamys farreri. Heterosigma carterae was found to suppress egg hatching rates when fed to the copepod Acartia tonsa [16] and severe reductions in copepod egg production rates were recorded during cyanobacterial blooms in the Baltic Sea [17]. Turner et al. [18] proposed that ingested phycotoxins may adversely affect successive generations of herbivores. Domoic acid is the most recognised and established diatom biotoxin and has been investigated in relation to anti-herbivory functionality [19–21]. However, from the perspective of reproductive impacts, it is the less-well-known oxylipins that are of prime interest. Oxylipins do not induce the same acute toxicity syndromes as the more commonly recognised algal biotoxins, and this has prompted their impacts on grazers to be referred to as ‘insidious’ [4]. They are however broadly cytotoxic with potential molecular targets associated with the cytoskeleton, calcium signalling and cell death pathways. These are unusual biological activities for microalgal-derived compounds which have also prompted some investigators to question whether or not the oxylipins are indeed toxins in the conventional sense. Toxins or not, it is precisely their bioactivity that has prompted current interest in exploring the scope and potential of these molecules for biotechnological and pharmacological development.

Oxylipin production in diatoms is wound-activated, meaning that synthesis is only initiated subsequent to loss of membrane integrity, typically as would occur during predation. Production and release of oxylipins, including polyunsaturated aldehydes (PUA) (which are the most intensively researched of the diatom-oxylipin family), hydroxyl-, keto-, and epoxyhydroxy fatty acid derivatives (see Figure 1) is extremely rapid. Production is by enzymatic oxidation of precursor polyunsaturated fatty acids and phospho- and galactolipids [22–26]. There is also evidence that some of the precursor compounds such as the galactolipid (2S)-1-O-3,6,9,12,15-octadecapentaenoyl-2-O-6,9,12,15-octa-decatetraenoyl-3-O-β-d-galactopyranosyl-sn-glycerol (1), may also be directly involved in cytotoxic reactions [27]. In addition, a further group of bioactive-molecules has recently been described [28] including a range of highly reactive oxygen species (ROS) and fatty acid hydroperoxides. They are formed from lipoxygenase-mediated oxidation of polyunsaturated fatty acids 2–4 including production from some non-PUA-producers. The resultant cocktail of ROS and oxylipins are highly damaging to invertebrate reproduction.

Of the oxylipins thus far described it is the PUA that have been the most comprehensively studied. This is partly due to PUA being the first group described [4], but also most PUA are commercially available, inexpensive and sufficiently stable to allow for a range of laboratory bioassays to be conducted. In contrast, many of the compounds described by Fontana et al. [28] and many of the intermediary compounds observed during PUA-synthesis are extremely unstable, require direct isolation from the algal source material and by default are neither readily available nor particularly amenable for biological testing. Of the PUA the C₁₀ decadienal has received most attention and has been treated almost as a model PUA, despite the fact that this compound would appear to be produced in rather smaller quantities by diatoms compared with other PUA. Whereas PUA-based investigations dominate the literature and will be the main focus of this article, they may not necessarily represent the primary chemical strategy employed by diatoms; for instance, in a survey of 51 diatom species only 38% were identified as PUA-producers [29]. In addition, the benthic diatom Phaeodactylum tricornutum has been identified as a producer of the oxo-acids 12-oxo-(5Z,8Z,10E)-dodecatrienoic acid (12-ODTE) (5) and 9-oxo-(5Z,7E)-nonadienoic acid (9-ONDE), which also inhibit invertebrate embryonic cleavage [30]. The mechanisms described by Fontana et al. [28] may also help to explain a number of observed reproductive failures in non-PUA-producing diatom species. Assumptions cannot be made in relation to the relative importance or prevalence of one oxylipin system versus another due to limited data particularly for ROS production. As such, the combined contribution of PUA, ROS and other bioactive oxylipin production to diatom chemical defence is unknown.

Miralto et al. [4] isolated three C₁₀ PUA (2E,4E-decadienal 6; 2E,4E,7Z-decatrienal and 2E,4Z,7Z-decatrienal) from the bloom forming diatoms Thalassiosira rotula, Skeletonema costatum (now reclassified as S. marinoi, [31]) and Pseudonitzschia delicatissima. Subsequently an expanded range of PUA were identified including 2E,4E-heptadienal; 2E,4Z-octadienal; 2E,4E-octadienal; 2E,4E-2,4,7-octatrienal and 2E,4Z-decadienal [22,23,30,32–34]. Production of low-molecular-weight aldehydes by microalgae has been reported before [35–38], however it was not until the Miralto et al. [4] study that interest in these compounds began to blossom. It has now become apparent that the 2E,4E isomer is detected as an artifact, and that the diatoms produce the 2E,4Z isomer. The double bond geometry has no influence on PUA-bioactivity [39], rather the activity is primarily due to the α,β,γ,δ-unsaturated aldehyde reactive Michael acceptor element 7, i.e., the conjugation of a carbon–carbon double bond to the aldehyde functional group. The Michael acceptor can form covalent adducts with nucleophiles such as amines and is thus associated with toxicity [40,41]. The side chain polarity is also important for bioactivity with longer-chain PUA being more toxic. Prior to identification from diatoms, decadienal was recognised as possessing a range of biological activities including anti-mitotic and pro-apoptotic properties [42,43]. PUA are also recognised for their antibacterial and allelopathic activity [44–47], are genotoxic [48–50], and have been reported as cytotoxic to mammalian cells forming endogenous DNA adducts [51,52]. PUA are also associated with carbon tetrachloride-induced liver cirrhosis in rats [53]. Decadienal also inhibits tumour-necrosis factor-alpha mRNA, so modifying cytokine secretion by macrophages [54] and possibly contributing towards atherosclerosis [55,56] and carcinoma [57,58].

Aside from toxic and disease related properties, PUA also have important biological and ecological functions. Examples include forming a component of the defensive glandular secretion of Hoplocampa sawfly larvae acting as an allomone against epigaeic predators such as ants [59]. Both (E,E) and (E,Z) isomers of decadienal have been identified from the marine insect Trochopus plumbeus and are speculated to function as a pheromone [60]. Other aldehydes such as 10E,12Z-hexadecadienal and 10E,12E-hexadecadienal are important sex pheromones of the pyralid lepidopteran, Notarcha derogata [61]. There is also some evidence that decadienal can function as an attractant in a bait-type insecticide against adults of the western corn rootworm (Coleoptera: Chrysomelidae) [62]. Citral, a methylated form of octadienal (3,7-dimethyl-2,6-octadienal), has been shown to block limb development in chick embryos [63] and is functional as an alarm pheromone in mites [64]. Heptadienal may have a pheromonal function in the swallowtail butterfly, Papilio machaon [65]. 12-ODTE has been proposed as a precursor of an algal sex pheromone [66]; is cytotoxic [67]; and induces a rapid increase in cytoplasmic free calcium in human blood neutrophils [68].

Oxyplins, and PUA in particular, have important biological and biochemical properties. PUA disrupt a number of salient stages in reproductive and developmental processes including; gametogenesis, gamete functionality, fertilization, embryonic mitosis, and larval fitness and competence [5,69–75]. It has been suggested that PUA-production, via reproductive interference, could facilitate bottom-up control of grazer populations by limiting recruitment [4,5,76]. The use of aldehydes as defensive compounds in terrestrial systems and by macroalgae is well known [77–82]. These defensive aldehydes are often biologically-active in a variety of different ways e.g., antimicrobial; embryotoxic; sperm motility inhibitors; larval-toxic; grazer-toxic, allelochemicals and feeding deterrents [81]. Latterly, there has been considerable attention devoted to documenting the effects of diatom-grazing on the population dynamics of copepods [5,12,26,83–85] and also on elucidating the various biosynthetic pathways of oxylipin formation [24,25,28,32,86]. Both lines of enquiry have also been the subject of recent and pertinent reviews [76,87–89], however, despite the apparent criticality of reproductive interference there has been comparatively little effort directed towards elucidating the modes of action of these chemicals at the physiological, cellular and molecular levels.

The majority of work addressing copepod secondary production has applied egg production as the metric of choice [90,91]. If taken in isolation egg production would have resulted in negative diatom effects being overlooked, as in general egg production is not consistently affected, although there are exceptions in the literature. An influential study by Ban et al. [92] presents strong evidence for the deleterious effects of diatoms on copepod reproduction including fecundity (eggs/female/day). The authors identified 4 copepod reproductive response categories in relation to diatoms: 1) reduced fecundity and hatching success; 2) reduced hatching success; 3) reduced fecundity; or 4) no negative effect. This paper will attempt to summarize current understanding of the impacts of oxylipins on invertebrate reproduction and development, but will not consider egg production further given the lack of clear and consistent relationships. This work will focus specifically on impacts on gametogenesis, gamete functionality, embryogenesis and larval fitness. The majority of studies cited are copepod orientated, however significant progress has been made using alternative model organisms such as echinoderms, ascidians and polychaetes. Evidence collected from all species will be used to explore common patterns of oxylipin bioactivity.

There have been a number of documented examples of direct impacts of oxylipin exposure and/or diatom ingestion on gametogenesis; primarily oogenesis and oocyte maturation. The relative paucity of literature in this area is principally due to the difficulties of studying gametogenic processes in copepods; however application of confocal laser scanning microscopy coupled with specific fluorescent probes has proven an effective and instructive tool for investigation [75,93,94]. The oogenic- and endocrine-cycles in copepods are particularly understudied in comparison to higher crustaceans and insects. Gonadal development stages have been described for a limited number of copepods including Calanus and Pseudocalanus species [95–97]. A degenerative response of ovarian tissue as a direct response to diet has recently been documented by Poulet et al. [98]. The study of gonad and gamete development is particularly important when considering the potential effects of dietary-sourced developmental inhibitors as contact with the gonads will most likely be the first association between inhibitory compounds and the reproductive system.

Fertilization is the union of two sexually-differentiated haploid gametes to form a diploid zygote. For sexually reproducing organisms this is a critical stage in the reproductive process. Fertilization is a complex multi-step process, which from the perspective of chemical intervention, presents many targeting opportunities.

Fertilization strategies may be broadly categorised as internal and external. For animals exhibiting internal fertilization sperm are transferred directly from male to female by a variety of mechanisms, with subsequent fertilization occurring inside the female body. Assuming gamete competency, any chemically-induced disruption (oxylipin or otherwise), would therefore largely be attributable to agents present either within the body or sequestered within the gametes. There would be limited opportunity for a direct, point-of-contact chemical intervention. In contrast, external fertilizers (e.g., broadcast spawners) release their gametes freely into the environment with fertilization occurring independent of further parental influence. The germ cells of external spawners would therefore be significantly more vulnerable to fertilization failure as a result of point-of-contact chemical exposure.

References to the effects of algal toxins on fertilization biology are relatively few. Examples include; toxins produced by the haptophyte Chrysochromulina polylepis that inhibit fertilization in ascidians and mussels [131] and the ciguatera poison, maitotoxin, is known to inhibit sea urchin fertilization by oocyte exocytosis [132]. Diabolin, a 120K protein from the kelp, Laminaria diabolica activates elevation of the fertilization membrane in unfertilized sea urchin eggs [133]. As discussed previously, Ianora et al. [121] and Laabir et al. [99] describe adverse effects on fertilization success due to poor sperm quality in the copepod Temora stylifera reared on certain dinoflagellate diets. Diatom-oxylipins are the most comprehensively researched in relation to fertilization effects. While feeding on the diatom Phaeodactylum tricornutum, eggs of Calanus finmarchicus were spawned which had not completed pronuclear fusion [134]. Buttino et al. [135] also observed failed pronuclear fusion in sea urchin eggs exposed to diatom extracts. Hansen et al. [123] observed reduced fertilization success in sea urchins exposed to extracts of Phaeocystis pouchetii - the active component subsequently recognised as decadienal. Caldwell et al. [71] demonstrated that fertilization success was reduced by exposure to decadienal, with the impact reduced by sequential washings. Oocytes incubated with decadienal over set times were not greatly affected when fertilized with untreated sperm. The fact that oocytes treated with decadienal and subsequently washed retained a high fertilization rate suggests that decadienal does not strongly associate with the oocyte surface-membrane where it may have affected sperm/egg binding. Sperm incorporation and the formation of the fertilization cone proceeded as normal in oocytes treated in this way, in contrast with fertilizations in the presence of decadienal above a concentration of 1 μg mL⁻¹. A similar observation was made by Tosti et al. [70] whereby decadienal interfered with actin microfilament reorganisation. The cytoskeleton appears to be a crucial target for PUA. The implications of this in relation to fertilization biology are profound. A coordinated sequence of motility events mediated by both microtubules and microfilaments are required for successful fertilization. Microtubules are the functional cytoskeletal component during pronuclear migrations and syngamy, whereas microfilaments mediate extrusion of the sperm acrosomal process, the formation of the fertilization cone and the block to polyspermy [136,137]. Schatten and Schatten [137] demonstrated that sperm incorporation was possible in oocytes treated pre-insemination with microtubule-inhibitors, whereas microfilament-inhibitors prevented incorporation.

Tosti et al. [70] made significant progress in explaining the potency of decadienal and similar PUA against early fertilization processes. Decadienal and decatrienal were observed to inhibit the fertilization-current generated at the point of sperm-egg interaction. Concomitant inhibition of plasma membrane voltage-gated calcium currents was also noted. Decadienal was identified as a specific fertilization-channel-inhibitor as it did not perturb either gap-junctional communication or plasma membrane steady state conductance. Decadienal appears to be the first documented fertilization-channel blocker in marine invertebrates [70].

The extent of in situ fertilization impacts as a result of diatom-oxylipins is unknown and there is currently insufficient data to even begin to develop projections. What is clear is that diatom-oxylipins have distinct biochemical properties which impact fertilization processes either by impairing gamete functionality and/or -competency or by disrupting salient stages during the early fertilization process. Environmental exposure to oxylipins may be dietary or by direct encounter with spawned gametes, particularly if spawning coincides with a diatom or Phaeocystis bloom (see [71] for further discussion). The enzymes involved in PUA-synthesis remain active for a considerable period of time after cell-wounding (see [23]), resulting in a ‘cloud’ of PUA associated with the dispersed algal cytoplasm. This may potentially exert a significant environmental check on fertilization success.

Development may be considered at a number of levels, ranging from the rapid mitotic cleavages of embryogenesis, the metamorphosis between larval stages, the formation of an adult body plan, the development to reproductive maturity and the continuous repair of somatic tissue during life. The life-cycle phases which are most susceptible to stress are the earliest (embryo and larvae) and the latest (senescent).

Arrested embryogenesis has frequently been reported in response to both diatom-feeding, exposure to diatom extracts and direct exposure to diatom-derived oxylipins. The initial report was in 1994 by Poulet [111] and co-workers who observed aborted development in embryos of the copepod Calanus helgolandicus. The point of arrest was variable but tended to be between just prior to pronuclear fusion or at various subsequent mitotic cleavages. The authors describe a number of cellular aberrations including globular cytoplasm and dispersed chromatin. Poulet et al. [134] describe cytological disruption in embryonic development during incubation in diatom extracts. Developmental abnormalities arose from an inability of the embryo to synchronise nuclear division with intercellular membrane formation during mitosis resulting in either developmental-arrest at the zygote stage or continued but abnormal development. Similarly, Buttino et al. [135] documented errors during embryogenesis resulting in hatching inhibition in sea urchins and ascidians incubated with extracts of the diatom Thalassiosira rotula. Extracts were observed to block tubulin organisation, promote microtubule depolymerisation and prevent chromatin condensation. Microtubule depolymerisation was observed from pronuclear fusion to telophase resulting in blockage of cell division. Nuclear fragmentation without cytokinesis was also observed. At high cell densities (5 × 10⁶ and 10⁷ cells mL⁻¹) the cell cycle was inhibited at the first mitotic division while at lower cell densities (2.5 and 1.25 × 10⁶ cells mL⁻¹) the majority of embryos completed first cleavage with a number developing to the eight blastomere stage. The pathology described by Buttino et al. [135] for Paracentrotus lividus embryos was very similar to that described for Calanus helgolandicus by Poulet et al. [134] suggesting a homogeneous mode of action.

Romano et al. [69], using a suite of techniques, confirmed that embryotoxicity was linked to apoptosis, however differences in response were noted between copepod and sea urchin models. Copepods appeared to follow a caspase-independent path whereas a caspase-3-like protease was activated in sea urchins. Using Calanus helgolandicus and Temora stylifera, Buttino et al. [75] observed apoptosis in embryos spawned from mothers that had ingested decadienal encapsulated in liposomes.

Again, using a sea urchin model, Hansen et al. [107] characterised a range of mitotic cellular events that were disrupted due to PUA exposure, including; inhibition of pronuclear migration; lack of DNA replication; tubulin depolymerisation; and importantly the failure to activate the G₂-M promoting-complex cyclin B-Cdk1. This final observation is potentially very important. M-phase promoting-factor (MPF) and cyclin-dependent kinases are essential for mitotic progression in the embryo. Interference with the synthesis or activity of either will have dire consequences, for example see Abraham et al. [117] who observed a range of effects associated with the inhibition of cyclin-dependent kinases in a range of eukaryotes. Interestingly, that study is, as far as I am aware, the only published work on the role of cyclins in copepod mitotic progression. It is unclear as yet whether PUA are acting directly on this complex or are affecting upstream cellular processes, perhaps one that is governed by cytoskeletal function which may have been compromised.

Cellular extracts from Skeletonema costatum have been shown to inhibit tumour cell proliferation through inhibition of normal cell division at G1 phase [138]. Both decadienal and decatrienal inhibited proliferation in Caco2 human colon adenocarcinoma cell lines [4]. The production of anti-mitotic compounds by plants is widely recognized, many of which have subsequently been exploited by the pharmaceutical industry, including some of the most valuable cancer treatments. Examples include: colchicine from the autumn crocus (Colchicum autumnale); podophyllotoxin from the lapacho tree (Tabebuia sp.); vinblastine and vincristine from the Madagascar periwinkle (Catharanthus roseus); and paclitaxel (Taxol®) from the Pacific yew (Taxus brevofolia). Colchicine interferes with the binding of tubulin-Guanosine triphosphate, which prevents microtubulin formation [139]. Colchicine is also an aneugen, a substance causing numerical chromosomal aberrations, frequently resulting in cell death [140]. It is therefore unsurprising to learn of potential anti-mitotic compounds from microalgal sources.

Interference with microtubule function in the activating oocyte and early embryo can affect localized RNA transcription which may in turn affect polarity and patterning [141]. Laabir et al. [99] suggest that cytoskeletal disruption may interfere with protein and amino acid biosynthesis, therefore affecting organogenesis in copepods. The authors noted sudden sharp variation in several amino acids coinciding with abortion of early embryos. Damage to DNA and RNA is potentially disastrous for rapidly dividing embryos, many of which appear to lack the repair checkpoints of somatic cells until the mid-blastula transition [142]. As discussed in relation to sperm motility and fertilization, the cytoskeleton is intrinsic to virtually all cellular functions, including chromosome separation, intracellular vesicle transport and determination of cell shape. The thiol-groups contained within tubulin are important for its function and are suspected targets for PUA-reactivity [143]. A compound that targets and inhibits the dynamic functioning of the cytoskeleton can essentially bring all cellular processes to a halt. To use a rather crude metaphor, it would be like trying to play Handel’s Concerto in B flat on a harp that has had all the strings cut. Therefore, the production of compounds that target your predator’s ability to function on a cellular level can be a highly effective anti-grazing strategy.

Caldwell et al. [71] noted that exposure to decadienal at sublethal concentrations slowed the rate of embryonic development. In effect, exposure prolonged the duration of the embryonic and larval phases. Egg viability is a crucial concept in relation to reproductive and population fitness. This has been demonstrated in fish [144–146] where it was observed that poor egg quality reduced the survival rate of hatched larvae by extending the length of the larval phase and hence the succeeding generation. Frangópulos et al. [147] reared Acartia clausi on combinations of toxic and non-toxic dinoflagellates, and noted that nauplii hatched from toxin-fed mothers required more time to moult to the copepodite stages. Sellem et al. [148] described abnormal development in embryos and larvae of the sea urchin Paracentrotus lividus incubated with the fatty acid octadecapentaenoic acid derived from the dinoflagellate Gymnodinium cf. mikimotoi. Development rate was retarded during the differentiation stage. The time delay was compensated for during the proliferating stages allowing the larvae to synchronously metamorphose to the pleuteus stage. Similar observations were made by Gentien et al. [149] on Mytilus sp. embryos incubated with Gymnodinium cultures.

Can these rate delays be rationalised? With regard to what is known of the effects of PUA on embryogenesis, perhaps consideration of cell cycle checkpoints may prove fruitful. It is known that decadienal can induce significant cellular abnormalities such as aneuploidy [150] and can inhibit DNA synthesis and replication [107]. Cellular mechanisms have evolved to prevent further cell division until any genetic damage has been repaired [151,152]. Ordinarily, apoptotic pathways would also be activated to remove non-viable cells or those beyond repair [153]. These checkpoints can have the effect of delaying development until the cell or embryo is competent to continue. However, it is generally regarded that checkpoints do not begin to function until the mid-blastula transition in rapidly dividing embryos [154]. Earlier mitotic divisions express an abridged cell cycle with highly compressed G₁ and G₂ phases, thereby allowing continuous cycles between S-phase (DNA replication) and M-phase (nuclear division). DNA repair would normally occur during the G phases, therefore cellular surveillance would not operate until numerous cycles of DNA replication and mitotic cleavage had occurred, with or without genetic damage. There is some suggestion that this generalisation may not hold entirely true for sea urchins [155,156], however most sources are in support of the critical stage of the mid-blastula transition [157,158] which leaves a considerable portion of early embryogenesis without an apparent mechanism for genetic surveillance. Le Bouffant et al. [156] suggest that the extent of DNA damage is an important factor in whether cell cycle delay or arrest is triggered. Relatively minor damage will cause a delay in cycle progression thereby allowing DNA repair to occur; more extensive damage initiates cell cycle arrest leading to an apoptotic response. Such observations would appear to hold true for PUA-toxicity; higher exposure is pro-apoptotic whereas lower exposure results in delayed cell cycle progression. Figure 6 illustrates what may be regarded as a series of gateway checkpoints through which the oocyte/embryo must pass during development. It would appear that this is not an either/or process in PUA-exposed embryos; severe exposure during early embryogenesis does result in developmental arrest and abortion, low-level exposure may allow development to progress unaffected with perhaps only minor rate delays. However, a potentially intermediary response with apoptosis restricted to badly damaged cells, the extent of which may or may not affect subsequent larval development and fitness, may explain teratogenic observations. An argument may therefore be made for potentially non-lethal, ‘selective’ or sacrificial apoptosis induced by environmental stressors. For this to be a realistic option, the larva must have cells or tissues that can be sacrificed. It would therefore seem logical that as the larva grows in size the potential to survive apoptotic events increases; in other words the larva has an increased capacity for non-mortal tissue sacrifice. Such a response may be important for the evolution of counter defences assuming the animal can then develop to reproductive maturity. However, as will been seen from the following section, surviving cell checkpoint apoptosis in the short term does not necessarily equate to survival in the longer term.

Exposure to harmful environmental agents during embryogenesis and larval development can affect subsequent development, growth and organism fitness. The degree of exposure to harmful compounds can be broadly classified as acute (rapidly lethal) or chronic (sublethal). In relation to environmental exposure, chronic exposure is the most frequently encountered situation. Exposure to sublethal levels of harmful agents can have significant effects on exposed animals including reduced immunological status, slower growth and abnormal development.

Evidently, the oxylipin impacted embryo must circumvent a number of critical developmental challenges if it is to survive to ‘larvalhood’. The embryo makes the transition to larva upon hatching; however, the hatching experience in itself represents a considerable hurdle. After the metabolic demands of embryogenesis the larva is then required to breach the confines of the oocyte membrane. Many species have evolved specialised appendages to aid hatching, such as the egg tooth in birds. Although the oocyte membrane in aquatic invertebrates will not provide the same degree of resistance as the calcium carbonate of a bird’s eggshell [159], hatching will nevertheless exert a metabolic toll. Hatching, therefore, represents a potential developmental bottleneck that may be further complicated by a weakened disposition due to toxin exposure. Indeed, Caldwell et al. [71] observed numerous fully formed larvae that were apparently morphologically normal but were unable to hatch.

Assuming successful hatching, the larva is then propelled into a world where it must face a new set of challenges; it must avoid predation; and, depending on species, be required to feed; undergo further development; potentially metamorphose; locate a suitable habitat; compete for space and resources; and ultimately recruit to the adult population. Considering that the larval phase can last from a few brief hours to a number of years depending on the species in question, it is fair to say that the traumas of embryogenesis are more than matched by the harshness of larval life. For larvae that embark upon this journey already carrying the burden of embryonic stress, the prospect of recruitment success is woefully slim.

The recognition, documentation and certainly quantification of abnormal copepod larval forms are very much a contemporary development. Uye [160] found that nauplii hatched from diatom-fed females were generally deformed and expressed marked morphological asymmetry. Most deformed nauplii died after hatching, whereas those that survived displayed greatly impaired swimming behaviour and died shortly afterwards. Miralto et al. [161] recorded that 81.8% of spawned eggs of Calanus simillimus from the Straits of Magellan underwent abnormal development resulting in either abortion or production of asymmetrical larvae. Similar observations have been noted in a range of laboratory and field studies [28,85,98,113,134,162,163]. Poulet et al. [134] document the severity of swimming behaviour change as a direct consequence of abnormal asymmetric development whereby the nauplii could not maintain a linear heading but tended to turn in the direction opposite to the most developed swimming appendages. Such a change in locomotory capacity will impair feeding efficiency and escape behaviours [164,165]. Many of these studies report concomitant poor hatching success and larval survival with patterns of abnormal production often reversed if mothers were switched to non-diatom-diets. The study by Ianora et al. [5] highlights the complexity of this issue as, using Calanus helgolandicus, they observed arrested and abnormal development with high mortality rates in larvae spawned from diatom-fed mothers and for which the larvae were similarly diatom-fed. Larval survival was not improved even when the larvae were switched to a non-diatom-diet. Survival was improved when the maternal diet was changed even when the resultant larvae were diatom-reared, suggesting that teratogens consumed maternally are more critical that those consumed by the progeny. Impaired larvae were shown to be undergoing apoptotic events. Similarly, Caldwell et al. [71] described polychaete and echinoderm embryos that hatched following PUA-exposure which tended to display anatomical malformations. These included incomplete ciliary band formation in Nereis virens trochophore larvae, and stunted and asymmetrical larval arms in Psammechinus miliaris echinopluteus larvae. Developmental abnormalities were also carried into subsequent larval forms, for example, damage accrued during the trochophore stage was still traceable in the nechtochaete stage (see Figure 7). Phenotypic malformations resulting from environmentally realistic toxin exposures will impact upon larval fitness, subsequent recruitment success and population growth. Morphological abnormalities acquired during embryogenesis may be maintained by the individual throughout the remaining ontogenetic process. There is no evidence yet for this in copepods; however there does exist some circumstantial support. Williams and Wallace [166] identified a deformation rate of 1–8% of the fifth limbs in adult female Clausocalanus sp. but the origin of this was unknown. Such malformations may impair the adult’s ability to mate successfully.

The morphometric changes that larvae undergo during development are critical to normal functioning such as feeding capability and metabolic activity. The sea urchin echinopleuteus larval stage is dependent upon morphological integrity for continued successful development. The ciliated bands located on the larval arms and circumoral surfaces are essential in creating and maintaining feeding currents [167]. Metabolic demands require that the ciliary band length increase in accordance with larval size. The larva accomplishes this by increasing larval arm length and developing additional pairs of larval arms. If the larva’s physical, physiological and metabolic capacity to satisfy these demands are compromised, then the larva will be at a competitive disadvantage – its larval fitness has been compromised.

Lewis et al. [74] and Caldwell et al. [73] developed a system based on fluctuating developmental asymmetry to precisely quantify very-low-dose sublethal effects of PUA on invertebrate development, thereby providing a useful proxy for larval fitness. Their approach demonstrated harmful effects at PUA concentrations previously deemed safe. The sublethal impacts of PUA during development may be described as altering developmental stability. Developmental stability can be defined as a suite of processes that tend to resist or buffer the disruption of precise development. Developmental instability on the other hand, incorporates a suite of processes that tend to disrupt precise development, such as small differences in rates of cell division, cell growth and cell shape change [168]. A number of techniques have been developed to assess the effects of developmental instability on organisms. Bilaterally symmetrical organisms are regarded as ideal candidates for developmental stability studies. It is accepted that the development of bilaterally symmetric traits are controlled by homologous genes. Therefore the ability of an organism to develop symmetrically while under stress conditions should reflect its propensity to resist environmental or genotoxic stress [169,170] and as such, act as a useful measure of organismal fitness.

The most commonly used descriptor for deviations from perfect symmetry is frequency distributions of right minus left (R-L) trait size. Three general patterns of developmental instability have been described: 1) directional asymmetry which is a pattern of bilateral variation in a sample of individuals, where a statistically significant difference exists between sides, but where the larger side is generally the same; 2) anti-symmetry is a pattern of bilateral variation in a sample of individuals, where a statistically significant difference exists between sides, but where the side that is larger varies at random among individuals; and 3) fluctuating asymmetry is a pattern of bilateral variation in a sample of individuals where the mean of right minus left is zero and variation is normally distributed about that mean [168].

Fluctuating asymmetry studies are based on random fluctuations from the bilaterally symmetrical state of particular characteristics or appendages of an organism [171]. Often fluctuating asymmetry measurements are combined with observations for survival and developmental rate. The measurement of morphological parameters during ontogeny of aquatic organisms is an accepted technique to evaluate the effects of chronic xenobiotic exposure [172] and the use of fluctuating asymmetry measurements as a test of larval fitness is gaining increasing acceptability [173–175].

Interestingly, Carotenuto et al. [163] suggest that the influence of diatoms on larval fitness is not necessarily limited to developmental instability during embryogenesis. The authors found that healthy copepod nauplii were unable to complete development to adulthood while feeding on diatoms resulting in high mortality. The larvae showed no evidence of anatomical malformations in contrast to nauplii hatched from diatom-fed mothers. These observations suggest that diatoms, aside from being teratogenic, also exert checks on development as a consequence of larval-feeding. In terms of the potential ecological significance, larval copepods therefore must contend with potential embryogenic errors and, assuming survival, also apparent direct toxicity from ingestion (Figure 8). Dinoflagellate controls supported successful development. C. helgolandicus nauplii fed Prorocentrum minimum with the addition of decadienal were unable to complete development beyond the sixth naupliar stage [5]. This raises the important question of whether diatom-derived toxins can affect somatic growth and hence larval fitness via ingestion by the nauplii.

Clearly, PUA have teratogenic and potentially mutagenic properties [107]. Identifying precisely the point or points during early ontogeny that critical embryonic disturbances occur is difficult given the broad cytotoxic activities of PUA. It would seem, however, that the embryo must cope with attack during significant phases of development. The production of teratogens by plants is common, and indeed the presence of teratogens in vegetables and food-associated microbes has been suggested as the primary reason for morning sickness in human pregnancy. Periods of morning sickness tend to coincide with key stages in embryo development, particularly the formation of the nervous system. Eliminating, or avoiding potentially damaging chemicals from the maternal diet during this phase of embryogenesis would clearly be beneficial. One feature of plant-derived teratogens is that they may otherwise be harmless to the grazer, and indeed the plant source may be highly beneficial for somatic growth, e.g., consumption of Veratrum californicum by sheep [176]. However, consumption during pregnancy can result is severe congenital defects. This pattern would seem to hold true for diatom-oxylipins. A diatom-diet appears to support good and reliable copepod somatic growth (however see 163) yet clearly has implications for reproduction. The parallel between terrestrial examples and the diatom-oxylipin scenario would appear to be appropriate.

The activity of diatom-oxylipins against reproductive and developmental processes is extensive and varied, affecting steps from gametogenesis through to larval competence and the establishment of the larval-body-plan (Figure 8). Despite this broad range of activities it may be possible to rationalise a common molecular and physiological origin. It has already been noted that PUA affect calcium signalling during fertilization [70]. Importantly, the change in intracellular free-calcium concentration is a common intracellular signal in the vast majority of eukaryotic organisms, particularly in relation to reproductive and developmental processes. Aside from the role that calcium plays in the fertilization reaction, it is also functional during embryonic mitoses [177]. Calcium signals have been detected just prior to entry into mitosis and just before initiation of anaphase [178,179]. Interestingly, it is also known that calcium depolymerizes polymerized tubulin in vitro [180]. Taken together, these observations are suggestive that the cellular impacts of diatom-PUA may ultimately be traced to interference with intracellular calcium signalling. There are of course other potential mechanisms that may explain the biological responses to PUA including; DNA degradation and transcription errors, decoupling of protein phosphorylation and kinase activity or even simply a loss of essential fatty acids as suggested by Wichard et al. [12]. This, of course, remains entirely speculative but verification would seem an appropriate avenue to explore in an attempt to fully understand the activities and ecological functions of this interesting family of biomolecules.

Microorganisms offer an exceptionally rich source of biologically active metabolites and marine microbes in particular are a vastly underexploited resource. There has been growing interest over the last 25 years in microalgal sources of novel compounds with potential pharmacological applications [181]. Diatoms have rarely been screened for biologically active compounds, apart from those expressing antibiotic activity or enzyme inhibition. As such, the production by diatoms of compounds that express such a varied range of reproductive toxicities, and anti-mitotic activities in particular, offers many avenues for future research; both fundamental and applied. Many of the stalwart drugs used in cancer chemotherapy share common properties with diatom-oxylipins including anti-mitotic and teratogenic activities [182]. The search for candidate drugs from marine algae is ripe for expansion; however the traditional bioprospecting approach is both laborious and extremely expensive. An efficiency saving may be found by combining knowledge of algal chemical ecology with pharmacological screening. There may be much to gain by taking a more targeted approach to bioprospecting rather than the traditional shotgun approach. It is important to gain a more thorough and relevant understanding of diatom compounds within a larger ecological and evolutionary framework. To begin to exploit knowledge of marine chemical ecology to inform drug bioprospecting, will require a concerted multidisciplinary approach, studying physiochemical phenomena at the molecular, cellular, organismal, population, and community levels of organization. In this respect, the framework suggested by Moore [183] would seem an effective model to follow. The three key themes to consider are: 1) understanding the mechanisms of molecular and subcellular interactions, including genomic and proteomic aspects; 2) the development of predictive simulation models of effects on complex cellular and physiological processes; and 3) linking molecular, cellular and patho-physiological ‘endpoints’ with higher level ecological consequences. Could ‘chemical ecology guided bioprospecting’ deliver the next generation of marine drugs? Only time will tell.