عوامل موثر بر طول عمر در عنکبوت های پرنده خوار (آرچنیدا: مایگالومرفا،ترافسیدا) - یک روش چند گونه

Lifespan is a life-history trait being of utmost importance, as it is frequently closely related to individual fitness. However, interspecific comparisons are relatively rare, being hampered by the high effort to collect longevity data across taxa. We here compiled lifespan data for 85 species of bird-eating spiders (Theraphosidae) held in captivity, based on 2183 individual records from the animal record books of both zoological gardens of Berlin, Germany. Using a data-mining approach we sought for broad patterns of correlations between lifespan and an array of other variables as derived from the literature. We found that the subfamily Eumenophoriinae lived on average longest, followed by the Theraphosinae, Ornithoctinae, Grammostolinae, Selenocosmiinae, Ischnocolinae and finally the Avicularinae. Species inhabiting tropical, more humid and/or low-altitude environments lived longer, suggesting that more predictable environments facilitate the evolution of longer lifespans. Furthermore, large range size, low abundance, sub-terrestrial life-style, and aggressive behavior were all associated with longer lifespans. Evidence for resource allocation trade-offs was revealed as larger spiderling and prosoma size was negatively related to longevity. Our rather rough approach revealed several patterns worth of future investigations, and illustrate the value of zoo records for interspecific comparisons.

Lifespan is a life-history trait which is of utmost importance, as it is frequently closely related to individual fitness (Tatar, 2001, Zera and Harshman, 2001, Partridge and Gems, 2007 and Vogt, 2012). This is because a minimum adult lifespan needs to be realized in order to secure any reproductive success, and as lifespan frequently correlates in a positive manner with reproductive output (Fischer, 2007 and Bonduriansky et al., 2008). Despite its extraordinary importance, all sexually reproducing organisms are subject to senescence, eventually resulting in death (Williams et al., 2006 and Hulbert et al., 2007). Several hypotheses have been put forward to explain the prevalence of senescence, including e.g. the disposable soma theory of aging stating that the available energy needs to be divided among competing functions, namely reproduction, maintenance, and repair (Stearns, 1992 and Kirkwood and Austad, 2000). From a life-history perspective senescence can thus be understood as a consequence of resource allocation trade-offs, resulting in optimized rather than maximized phenotypes (Carey, 2001 and Boggs, 2009). Resource-allocation trade-offs result if a limited amount of energy needs to be divided among two or more competing functions. As somatic maintenance (and therefore longevity) requires substantial amounts of energy, lifespan is expected to be involved in trade-offs (Roff, 1992 and Stearns, 1992). For instance, the energy expenditure allocated to reproduction, being measureable as e.g. fecundity or offspring size, has been repeatedly found to be traded off against longevity (Roff, 1992, Stearns and Partridge, 2001 and Zera and Harshman, 2001). Other ‘classical’ trade-offs are those between present and future reproduction, which are also frequently found (Roff, 1992 and Stearns, 1992). As a consequence of such trade-offs, lifespan is typically positively related to body size and/or storage reserves (Roff, 1992 and Speakman, 2005a). As energy intake and metabolic rates depend strongly on environmental conditions, lifespan is generally strongly environment-dependent (Carey and Liedo, 1995, Carey, 2001 and Hulbert et al., 2007). In addition to environmental factors lifespan may also be subject to genetically determined variation. For instance, Hughes and Reynolds (2005) list a number of transgenic and mutant alleles of Drosophila melanogaster which affect longevity. Longevity also shows striking variation across species and higher order taxa, suggesting a ‘phylogenetic legacy’ ( Carey, 2001, Sgrò and Hoffmann, 2004 and Hulbert et al., 2007). In general though, the heritability of lifespan appears to be low as expected for a trait closely related to fitness ( Carey and Judge, 2001). Despite a strong interest in the evolution of lifespan and associated trade-offs, comparative data on this matter is exceedingly rare (Carey, 2001, Beck and Fiedler, 2009 and Vogt, 2012). Such potentially very valuable studies are likely hampered by the high effort needed to collect longevity data across taxa. Against this background we here present data on 85 species of bird-eating spiders (Theraphosidae), on which hardly any systematic data on lifespans and their ecological correlates was available thus far. Theraphosidae are a family of mygalomorph spiders (Araneae), which mainly inhabit tropical and subtropical regions in both Americas, Africa and Asia, using a wide variety of habitats such as rainforests, woods, savannahs, grasslands, semi-deserts and deserts (Smith, 1988 and Klaas, 2007). Data were compiled from the animal record books of both zoological gardens of Berlin, thus including exclusively animals which were kept under controlled environmental conditions in captivity. Using a data-mining approach we sought for factors affecting lifespan, using an array of predictor variables as derived from the literature (e.g. on climate, habitat, behavior, morphology and life history). Obviously, the evolution of lifespan across species is shaped by a large array of selective forces, which may well vary throughout a species’ range adding further complexity. Therefore, we are able here to examine some broad patterns of correlation only. Note though that this is a conservative approach. We will specifically test the hypothesis that more predictable environmental conditions should select for longer lifespans. This prediction rests on the assumption that random mortality (through e.g. predation, starvation, or catastrophic events) should be low in those environments, such that intrinsically longer lifespans may pay off (Sgrò and Hoffmann, 2004, Williams et al., 2006 and Pruitt and Riechert, 2012). Apart from the above hypothesis we further investigate whether (1) the different subfamilies of Theraphosidae differ in lifespan, whether (2) lifespan is traded off against reproductive effort, and whether (3) lifespan is positively related to body size.

4.1. Lifespan of Theraphosidae Compared to other spider families, Theraphosidae are considered to be long-lived, with maximal lifespans reaching up to 20–30 years (Bücherl, 1962, Costa and Pérez-Miles, 2002, Schmidt, 2003 and Klaas, 2007). Accordingly, all Theraphosidae are able to survive the winter in the adult stage, whereas temperature-zone spiders are often seasonal (Bellmann, 2005). In our study the oldest animal (Aphonopelma seemanni, Grammostolinae) got older than 17 years, thus reaching a similarly high age. The overall mean value though was only 2.6 years. Currently it is unclear which factors cause the rather large difference between mean and maximum lifespan, in any case indicating large individual variation. In particular it is unclear whether our mean lifespans are comparatively short, as in other studies only maximal values are presented which will always differ substantially from mean values as shown here. Relatively short mean lifespans in our study may have resulted from keeping the animals at relatively high temperatures and at ad libitum feeding conditions, both increasing metabolic rates thereby decreasing lifespan ( Hulbert et al., 2007, Bauerfeind et al., 2009 and Karl and Fischer, 2009). On the other hand, captive individuals usually reduce energy spending and avoid natural risks, which may even result in an overestimation of lifespan. Note in this context that comparative studies aim at revealing intrinsic (i.e. genetic) differences among the species involved. It is therefore crucial to control for environmental variation, which is evidently the case here by using the same temperature and feeding regimes for all species. In Theraphosidae, particularly long lifespans seem to be restricted to females, while most males die after their first reproductive period (Costa and Pérez-Miles, 2002 and Klaas, 2007). The latter is not related to sexual cannibalism, which has never been observed in Theraphosidae (Costa and Pérez-Miles, 2002). In contrast to females, males invest large parts of their energy supplies to mate location behavior, actively seeking for potential mates while feeding rarely (Costa and Pérez-Miles, 2002). Concomitantly, males may have much higher resting and locomotion metabolic rates, as has been shown in the Texas tarantula Aphonopelma anax (Grammostolinae), in which males additionally reach only about half the body mass of females ( Shillington and Peterson, 2002 and Shillington, 2005). Consequently, sexual differences in longevity may have a behavioral (enhanced resource allocation to mate location) as well as a physiological basis ( Shillington and Peterson, 2002 and Shillington, 2005). The latter may further result in the production of reactive oxygen species negatively affecting lifespan ( Criscuolo et al., 2010 and Speakman, 2005a). Against this background it is unfortunate that it was not possible to analyze the data separately for males and females (see above). However, as the sex ratio within species is expected to be equal, this should not affect any of the results reported here. Not considering sexes will obviously increase the variance within the data set, thus making it more difficult to reach the significance threshold. Thus, our approach is conservative with a very low risk of reporting false positive results. At the subfamily level Eumenophoriinae and Theraphosinae lived longest, while Ischnocolinae and Aviculariinae lived shortest. Short lifespans in the Aviculariinae might be associated with their arboreal lifestyle (Smith, 1988 and Klaas, 2007). As they seem to suffer from high predation rates (Costa and Pérez-Miles, 2002), selection may favor rapid reproduction rather than long life in this subfamily. However, whether predation is able to generally explain differences in lifespan among arboreal and ground-dwelling species is unknown. Note further that different subfamilies may differ in their ability to tolerate laboratory conditions in captivity, which might be lower in e.g. the Aviculariinae than in other taxa. Unfortunately, we cannot compare our data on variation among subfamilies with other findings, as only Costa and Pérez-Miles (2002) reported a few longevity records, while other comparative data do not yet exist. In our analysis across subfamilies the highest mean values were not accompanied by the highest maximal lifespans. This result was not unexpected as maximum values may be heavily influenced by chance effects (Speakman, 2005b). Below we will discuss the factors affecting Theraphosidae mean lifespan as indicated by our results. However, as we could not use phylogenetic independent contrasts and owing to potential problems of multiple testing, we will restrict ourselves to patterns that are at least largely consistent across subfamilies, thus likely reflecting reliable patterns. We will not discuss any results on maximal lifespans as no consistent patterns could be found. Because maximum values are heavily affected by single events, they are rather inappropriate measures of aging (Carey, 2001 and Speakman, 2005b). 4.2. Lifespan and climate Several lines of evidence suggest that lifespan in Theraphosidae is longer under more predictable climatic conditions. First, tropical species lived longer than non-tropical ones, second relative humidity correlated positively with lifespan, and third lifespan was longer in species from lower as compared to higher altitudes. Additionally, there was some evidence for a negative association between lifespan and potential landscape evaporation. These findings match the prediction that selection for increased lifespan is expected for species inhabiting tropical and more humid, i.e. more predictable and probably prey-rich, environments in which random mortality should be relatively low (Zera and Harshman, 2001 and Williams et al., 2006). In contrast, environmental effects of temperature are typically opposite in direction (i.e. shorter lifespans at higher temperatures) based on an increase of metabolic rate with temperature (Bauerfeind et al., 2009 and Karl and Fischer, 2009). 4.3. Lifespan and range size, abundance and habitat Species inhabiting larger ranges showed longer lifespans than those inhabiting smaller ranges. This interesting pattern might be related to more widespread species showing a more generalist lifestyle, enabling them to exploit a wider variety of resources and thereby to fuel longer lifespans (Zera and Harshman, 2001). Why low-abundance species should show longer lifespans than high-abundance ones is currently unclear, but perhaps longer periods of time are needed to successfully locate mates (Stearns, 1992). Furthermore, we found that sub-terrestrial species lived longer than arboreal or terrestrial ones. Sub-terrestrial species could perhaps be better protected against predators, facilitating the evolution of long life spans (Keller and Genoud, 1997). As the above pattern could not be investigated at the subfamily level, caution is needed as we cannot exclude phylogenetic effects. Note that most arboreal spiders in our analysis belonged to the subfamily Aviculariinae, such that it is unclear whether the above pattern results from phylogenetic constraints or ecological factors such as differential predation pressures. 4.4. Lifespan and morphology, life history and behavior Species producing larger compared to smaller spiderlings lived shorter, which may reflect a resource allocation trade-off between somatic maintenance and reproductive investment (Djawdan et al., 1996 and Fox and Czesak, 2000). However, between lifespan and egg number per cocoon, traits that can also be assumed to be involved in a trade-off (e.g. Fowler and Partridge, 1989, for Drosophila), no significant association was found. These findings probably reflect the complicated case of life-history trade-offs, which may occur only in specific cases or only between individual traits ( Fox and Czesak, 2000, Zera and Harshman, 2001 and Bonduriansky et al., 2008). Resource-allocation trade-offs may also be responsible for the negative relationship between prosoma width and longevity. Thus, a higher investment into the prosoma may be traded-off against somatic maintenance and therefore lifespan (Zera and Harshman, 2001). This finding does not contrast with the general notion that lifespan should be positively related to body size in ectotherms (Ackermann et al., 2001), for which we also found some evidence. Overall body and opisthosoma size (thus the amount of storage reserves) tended to be positively related to lifespan as expected, although correlations were rather weak and not found in all cases. Additionally, older age at sexual maturity tended to prolong lifespans in both male and female Theraphosidae. Similar patterns have been commonly reported for other ectotherms (Rose, 1984 and Zera and Harshman, 2001). In contrast, it is currently unknown why more aggressive species should live longer than less aggressive ones, although this pattern was largely consistent across subfamilies in the present study. We speculate that aggressiveness may reduce random mortality due to predation, which typically promotes the evolution of longer lifespans (Williams et al., 2006 and Pruitt and Riechert, 2012). 4.5. Conclusions We found that, at the subfamily level, Eumenophoriinae lived longest followed by the Theraphosinae, Ornithoctinae, Grammostolinae, Selenocosmiinae, Ischnocolinae and finally the Avicularinae. As this study provides the first comprehensive study on Theraphosidae lifespans, the reasons underlying such differences are currently unclear and require further investigations. However, we were able to reveal several patterns that were relatively consistent across subfamilies. We found as expected that inhabiting more predictable environments (i.e. the humid tropics, low altitudes) facilitates the evolution of longer lifespans. Furthermore, large range size, low abundance, sub-terrestrial life-style, and aggressive behavior were all associated with longer lifespans, though the underlying reasons are currently largely unclear. Evidence for resource allocation trade-offs were revealed as larger spiderling and prosoma size were negatively related to longevity. Considering our necessarily rough approach, finding several largely consistent patterns is striking, and sets the stage for in-depth follow-up studies. Our study illustrates that zoo records may provide valuable data source for comparative studies, especially for taxa on which hardly any other data is available.