1 INTRODUCTION

To maximise fitness, organisms time life cycle events to coincide with windows of favourable environmental conditions. For birds, such events include breeding and moulting which, in temperate regions, are timed to follow the predictable seasonal rise in food availability in spring arising from a strong yearly oscillation in photoperiod and temperature. Temporal mismatch between life cycle events and environmental resources has considerable costs for birds in these regions (Visser & Gienapp, 2019). In tropical areas, where photoperiod and temperature oscillate less strongly, the timing of life cycle events in birds can be much more variable (Jenni & Winkler, 2020; Nwaogu et al., 2019). Our understanding of how tropical birds time their life cycle events in relation to environmental conditions has improved over recent years but remains severely biased by the relative volume of research from temperate regions (Cohen et al., 2018; Grames et al., 2023). Given that most the world's bird species reside in the tropics, this is a bias that must be addressed to allow a more objective understanding of biological rhythms in birds.

A potential consequence of the temperate bias in avian life cycle studies is a focus on breeding as the most critically timed event in the life cycle. Breeding must be well timed because it is an energy expensive process that requires high food availability both to maintain the parents' body conditions and to feed chicks. In temperate birds, where the window of suitable breeding conditions occurs once per year, and where longevity is thought to be in general low (Hau et al., 2010), it is assumed that maximising breeding success through accurate synchronisation to food resources is under strong selection pressure. However, in tropical regions, priorities may be different; birds tend to live longer, and nest predation is often higher, and therefore fitness may be strongly mediated through survival, as bird must survive until the next breeding attempt (Class & Moore, 2013; Jetz et al., 2008; Martin et al., 2017; Muñoz et al., 2018; Stevens et al., 2013). To increase survival, birds may therefore prioritise self-maintenance processes such as moult (Class & Moore, 2013; Jenni & Winkler, 2020).

Moult can be a costly life cycle event, due to the direct energetic cost of regrowing feathers and accompanying tissues, as well as additional costs such as increased thermoregulation and less effective predator avoidance during this period (Jenni & Winkler, 2020). For successful moult, birds must have access to sufficient food, including certain important nutrients (Jenni & Winkler, 2020). Food limitation, or lack of nutrients such as protein, carotenoids, or certain amino-acids, can lead to depletion of body reserves, slower moult or the production of lower quality feathers (Hill, 2000; Murphy et al., 1988; Swaddle & Witter, 1997). To increase the chance of surviving until the next breeding attempt, it is logical that tropical species may synchronise moult, more than breeding, to favourable environmental conditions. However, so far, we know little about which environmental variables may be associated to higher probability of moulting in the tropics, and few studies have tested whether moult is indeed more seasonal than breeding in these environments (Class & Moore, 2013; Jenni & Winkler, 2020; Nwaogu et al., 2019).

One aspect that makes studying tropical bird life cycles complicated is the extent of variation between individuals, populations, regions and years. This has led to the impression that tropical species may lack seasonality (Baker, 1939). However, evidence from tropical regions demonstrates that species often do show seasonal life cycles that are distinct between populations even in close spatial proximity. One extreme example is a ~5 months difference in the timing of breeding of two populations of rufous-collared sparrows (Zonotrichia capensis) separated by 25 km (Moore et al., 2005). This stark difference is explained by shifted weather patterns between the sites given their position on either side of the Andean divide. A similar phenomenon occurs in Cameroon, where highland populations of the same species breed in a different season to lowland populations, possibly as a consequence of rainfall patterns (Tye, 1992). Some other studies from the tropics and subtropics show that timing of breeding and moult entrain with local rainfall patterns along environmental gradients (Nwaogu & Cresswell, 2020; Oschadleus & Underhill, 2006). These studies demonstrate that when tropical species are studied at broad spatial scales, local seasonal variation may be masked resulting in a deceptively non-seasonal pattern in life cycle events (Baker, 1939). Additionally, they demonstrate that seasonality in weather conditions, and consequently in lower trophic levels such as plants and insects, may also vary over small spatial scales. Overall, these studies suggest locally adapted timing of breeding and moulting of birds in the tropics—but offer little insight into the actual environmental factors that could influence such timings, aside from rainfall.

Thus, for a more comprehensive understanding of avian seasonality, data from tropical species are needed that link breeding and moult to fluctuations in key aspects of the environment, such as climate and underlying trophic levels (Visser & Gienapp, 2019). In temperate areas, where phenology of breeding and moult is associated with the yearly raising temperatures and longer days of spring, studies tend to investigate inter-annual variation in timing and its drivers. However, in tropical areas, breeding and moult may be spread throughout the year, and so it becomes natural to investigate intra-annual drivers of phenology. This is especially true given the paucity of tropical phenology data available, which is partly explained by the challenges of funding and logistics for collecting year-round data (Cohen et al., 2018). To help fill this gap, in this study, we provide comparative annual-cycle data on three trophic levels, recorded for one full year. The logistical trade-off we faced, given the demands of detailed tri-trophic coverage, meant we could only collect data simultaneously at two sites and in a single year. We cannot therefore make conclusions about the repeatability of observed patterns, but we are able to investigate relationships between short-term temporal variation in the environment and bird life cycles, and assess whether these patterns are consistent between sites.

We build on an avian system introduced by Nwaogu et al. (2019), in which the target species, the common bulbul (Pycnonotus barbatus; hereafter called “bulbul”), showed strong annual cycles in moult and weak cycles in breeding near Jos in Nigeria (9°52′ N, 08°58′ E). The common bulbul is a good model species for phenological study. It is a common and relatively easily observed year-round resident in the study region, while also being widespread throughout the African continent. Therefore, it faces very distinct conditions throughout its range and likely shows local variation in life cycle timing (Jenni & Winkler, 2020; Nwaogu & Cresswell, 2020). Bulbuls are omnivorous birds and may therefore meet the nutritional requirements of moult and breeding with different combinations of plant and arthropod foods, according to local conditions. They are known to consume a wide range of plant and arthropod taxa, including termites, ants, caterpillars, butterflies, mango, guava, African plum, and many other fruits (CJ, CJN & AJW, unpublished data; Nwaogu, 2019). Therefore, phenological variation at lower trophic levels can be assumed to be roughly reflected in the bulbuls' diet. For a more direct description of trophic relationships of bulbuls, we have recently used faecal metabarcoding to assess diet around the year at the study site in Cameroon (CJ, CJN & AJW, unpublished data). This study confirmed that bulbuls consume a wide range of plant and arthropod taxa, broadly overlapping with the taxa surveyed in this study. However, moulting bulbuls had a more diverse diet than non-moulting birds, both in terms of fruit and arthropods, leading to the hypothesis that bulbuls may synchronise moult to periods of plentiful resources (CJ, CJN & AJW, unpublished data). In contrast, a captive study on bulbuls showed that a fruit-only diet instigated earlier moult compared with an arthropod-only diet (Nwaogu et al., 2020). Nwaogu and Cresswell (2020) assessed moult along a latitude gradient in Nigeria, and found that with increasing latitude in Nigeria, moult occurs later, matching the pattern in the onset of rains. The onset of rains may in turn drive the emergence of plant and insect resources, exploited by bulbuls during moult. Based on these studies, it seems that bulbuls seek out a diverse diet during moult (perhaps with high fruit content), and these foods may become available after the onset of the rains. However, we still lack data from the wild that links timing of moult and breeding in bulbuls to fluctuations in environmental resources.

Our aim was to test three main predictions: first, we expected breeding and moult to show seasonality at both sites, with moult showing a stronger seasonal signal than breeding (Class & Moore, 2013; Nwaogu et al., 2019). Second, we expected to see parallel phenology between sites, with earlier timings in Cameroon due to the earlier onset of the rains (Nwaogu & Cresswell, 2020), and with links between timing of moulting and fruit availability, following Nwaogu et al. (2020), and between timing of breeding and insect availability, following literature from temperate environments (Perrins, 1991; Visser & Gienapp, 2019). Finally, we expected the site closer to the equator to show weaker seasonality than the northern site, though with only two sites in the comparison this contrast cannot be conclusively attributed to latitudinal effects. As our aims involved linking differences in timing of breeding and moult of bulbuls to environmental phenology, we chose to focus on arthropods, fruit and precipitation as representatives of food availability in the environment.

2 METHODS

3 RESULTS

Timing of breeding was seasonal in Nigeria according to the GAMs, with the model including the seasonal term performing significantly better than the random intercept only model (Table 1). Results from circular statistics indicated otherwise, with p = .37 in Rayleigh test (Table 1). Probability of breeding was highest in Nigeria in December–January (Figure 1, Figure S3; Table 1). We did not detect a seasonal trend in timing of breeding in Cameroon (Table 1; Figure 1, Figure S3). Primary moult was seasonal in Nigeria and more strongly predicted by date compared with breeding (25% variance explained vs. 5%; Table 1). Moulting peaked in June in Nigeria (Figure 1, Figure S3; Table 1). In Cameroon, there was weak evidence for seasonality in timing of moult (∆AIC = 1.5) and stronger evidence from circular statistics (Table 1), with moult peaking in July but with high uncertainty associated to predictions (Figure 1, Figure S3; Table 1). Overall, therefore, moult was more seasonal than breeding in both sites, and both life cycle events were more seasonal in Nigeria than Cameroon, following predictions (Table 1). Interestingly, moult was similarly timed between the two sites, with the peak in Cameroon occurring ~6-weeks before the peak in Nigeria.

In lower trophic levels (fruit, arthropods), there was seasonal variation in both sites, following different timing (Table 1; Figure 1, Figures S4 and S5). In Nigeria, fruiting tree abundance was highest in September–October, and lowest in April, while arthropods were most abundant in July, and least abundant in March. In Cameroon, fruiting tree abundance was highest in April (according to GAMs) and August (circular statistics), and lowest in November–December, and strength of seasonality was higher compared with Nigeria. Arthropod abundance spiked dramatically in April–May, mostly due to high abundance of aphids, mealybugs and ants, and was lowest in August (Figure 1, Figure S5). The difference between Nigeria and Cameroon in time of peak fruiting tree abundance was therefore large (~5 months), but smaller for peak arthropod abundance (1.5 months), but in both cases abundance peaked earlier in Cameroon.

In Cameroon there was higher yearly rainfall (1415 mm) than in Nigeria (889 mm), and in both sites rainfall showed strong and distinct seasonal patterns (Figure 1; Table 1). The times of highest rainfall in Nigeria were July–August, and the driest period was November–December. In Cameroon, rainfall appeared to follow a bi-modal pattern, with high rainfall in March–April and September–October, and the lowest in December–January.

With regards to correlations between moult and breeding and environmental conditions, in Nigeria, probability of breeding was negatively correlated with arthropod abundance and precipitation, and positively with fruiting tree abundance (Figure 2; Appendix S3). In contrast, in Cameroon, the models containing fruit and precipitation performed similarly well; therefore, breeding was positively associated to higher fruit abundance or higher precipitation (Figure 2; Appendix S3). Probability of primary moult in Nigeria was highest during the period of maximum arthropod abundance, and was negatively correlated to fruiting tree abundance and precipitation. In Cameroon, moult was significantly positively correlated with precipitation, and there was no significant relationship with arthropod or fruit abundance (Figure 2; Appendix S3). Therefore, our expected links between arthropod abundance and breeding, and fruit abundance and moulting were not met at either site.

4 DISCUSSION

Common bulbuls showed varying seasonality in breeding and moulting in our study sites, and the specific timing of events was different between sites. Phenology of arthropods and fruit in the environment also varied between sites, with peaks in abundance occurring earlier in Cameroon than Nigeria, as predicted. In bird life cycles, we found stronger seasonality in moult than breeding at both sites, and overall stronger seasonality in Nigeria, both according to expectations. In contrast, the specific timing of bird life cycle events, especially moult, were unexpected. These findings provide additional support to the growing evidence that tropical birds time their life cycles to environmental fluctuations, but that these adjustments may vary over small spatial scales.

In Nigeria, our results suggest a strong seasonal pattern in moult, likely associated to the seasonal rise in arthropods in the environment. Specifically, certain arthropod groups such as Lepidoptera, Hymenoptera and Hemiptera had high abundances coinciding with peak moult timing (Figure S5), suggesting that potentially birds time their moult to specific arthropod food resources. As moult involves not only the replacement of feathers themselves but also the accompanying tissues (sheaths, epidermal structures), birds need to consume sufficient protein during the moulting period (Jenni & Winkler, 2020). Insects contain high protein, and therefore birds may increase insect consumption during moult (Jenni & Winkler, 2020; Pap et al., 2008; Simon et al., 2023). Indeed, birds fed insufficient protein under experimental settings had delayed or slower moult, or produced feathers of lower quality (Jenni & Winkler, 2020; Pap et al., 2008). These findings linking arthropod abundance and moult match recent dietary analyses of bulbuls in Cameroon, which showed that moulting individuals consumed a higher diversity of arthropods than non-moulting birds, especially Lepidoptera (CJ, CJN & AJW, unpublished data). However, our moult model for Cameroon indicated no associations with arthropods or fruit; while the lack of correlation could be partly a consequence of small sample sizes, it could also reflect differing resource limitationexperienced by bulbuls.

In line with the moult-arthropod association, breeding in Nigeria was negatively associated with arthropod abundance, indicating that birds may synchronise their moult to peak arthropod abundance, and consequently breed at times when arthropods are less abundant (Class & Moore, 2013). Instead, breeding in Nigeria and Cameroon appeared to be associated with fruiting tree abundance. While it is unlikely that bulbuls feed fruit to their chicks, which in most omnivorous or insectivorous bird species are fed protein-rich easily digested insects such as caterpillars (Jarrett et al., 2020), it could be that adults consume fruit during the breeding period to increase calory intake. Fruits can constitute high energy foods, consumed for instance by birds prior to migration to increase energy reserves (Bairlein & Gwinner, 1994; Iwajomo et al., 2017). Therefore, breeding at the time of peak fruiting tree abundance could allow maintenance of body conditions while undertaking this energy-expensive life cycle event. Interestingly, for Cameroon, the model including precipitation performed equally well to the model including fruit at explaining timing of breeding, indicating that breeding could be associated to higher precipitation. Independent of food, associations to precipitation may reflect habitat requirements for nesting or links between precipitation and nest predation (Nwaogu, 2019).

Our results indicate that bulbuls in the Afrotropics show stronger seasonality in moult that in breeding. Additionally, for moult, birds in Nigeria and Cameroon showed close timing, with the peak in probability of primary moult occurring ~1.5 months earlier in Nigeria than in Cameroon. The stronger seasonality in moult and the closer timing of moult between sites follow from each other, as cycles with high amplitude are more robust and respond weakly to fluctuations in environmental cues (Tokuda et al., 2020). The high consistency supports the idea of moult as a more strongly seasonal fixed event, and perhaps as an anchor of the bulbul's annual cycle. In comparatively long-lived organisms experiencing strong seasonal constraints on the timing of moult, but with multiple breeding opportunities with limited output per year, there might be a weak selection to organise the annual cycle around breeding (Nwaogu et al., 2019). In contrast, there will be a stronger selection to organise the annual cycle around moult. However, the earlier moult in Nigerian birds goes against our expectations, given the findings in Nwaogu and Cresswell (2020), which show later moult at higher latitudes within Nigeria. Based on our study with only two sites, though, we cannot draw conclusions about latitudinally driven patterns as there may be many other factors driving such variation.

If selection acts more strongly on the timing of moult, then seasonality in breeding may simply arise because of seasonal moult, but not because breeding is the event that is timed primarily to specific environmental conditions. This idea is backed up by the coinciding peak timing for primary moult in Nigeria between this study and a previous study from the same site and sites on higher latitudes (Nwaogu et al., 2019; Nwaogu & Cresswell, 2020). This contrasts with temperate-zone biased suggestions of relatively fixed breeding phenology as an anchor for avian cycles, while moult functions as a more flexible buffer (Helm & Gwinner, 2006). Indeed, we found mixed support for seasonality in timing of breeding in Nigeria, with results from GAMs and circular statistics disagreeing with each other, and in Cameroon, we found no evidence of seasonal breeding.

Overall, our study provides novel evidence of the seasonal yet spatially variable phenology of tropical birds entrained to local environmental seasonality. However, given the limited temporal span of our data, we cannot be certain about the consistency of such patterns between years, and urge longer-term year-round studies across latitudinal gradients to tease apart the drivers of such variation. To ensure robust results from our data, we used both GAMs and circular statistics to detect seasonal trends. Importantly, we found that results from GAMs and circular statistics were consistent, with a few exceptions: moult in Nigeria was seasonal according to GAMs but not according to the Rayleigh test, and timing of peak fruiting in Cameroon was April according to GAMs and August according to circular statistics. The former inconsistency may be due to a higher flexibility of the GAM to accommodate seasonal patterns, and to the presence of the temporal random effect in the GAM that may absorb some of the noise in the data. The latter is likely due to the seasonal plateau in fruit abundance in Cameroon between April and August (Figure 1).

In conclusion, our results suggest that over small spatial scales, timing of resources may vary considerably, leading to different life cycle timing in birds, but with differential effects for life cycle stages. Understanding bird phenology is important as it can reveal vulnerabilities of species in the face of continued land-use and climate change (Cohen et al., 2018; Fotso, 1996; Helm et al., 2013; Samplonius et al., 2016; Thomas et al., 2001). Incorporating a temporal dimension to conservation efforts, for instance in the seasonal control of burning in the Afrotropics, could encourage win-win scenarios for humans and wildlife in transformed landscapes (Fotso, 1996). With our knowledge of tropical bird life cycles lagging behind that of temperate counterparts, our study reveals the importance of considering local context in the study of tropical birds, and provides additional evidence of the strong seasonal nature of tropical environments.

AUTHOR CONTRIBUTIONS

Crinan Jarrett: Conceptualization (equal); formal analysis (lead); investigation (equal); methodology (equal); project administration (equal); visualization (lead); writing – original draft (lead); writing – review and editing (equal). Ojodomo Godday Simon: Data curation (equal); investigation (equal); methodology (equal); writing – review and editing (equal). Christian N. Tchana: Data curation (equal); investigation (equal); methodology (equal); writing – review and editing (equal). Thaddeus Apezan Pev: Data curation (equal); investigation (equal); methodology (equal); writing – review and editing (equal). Michelle Fany Meigang Kamkeng: Data curation (equal); investigation (equal); methodology (equal); writing – review and editing (equal). Alain Christel Wandji: Data curation (equal); investigation (equal); methodology (equal); writing – review and editing (equal). Shiiwua A. Manu: Project administration (equal); supervision (equal); writing – review and editing (equal). Mélanie Adèle Tchoumbou: Methodology (equal); project administration (equal); supervision (equal); writing – review and editing (equal). Barbara Helm: Conceptualization (equal); formal analysis (equal); funding acquisition (equal); investigation (equal); methodology (equal); project administration (equal); supervision (equal); writing – review and editing (equal). Luke L. Powell: Conceptualization (equal); funding acquisition (equal); methodology (equal); project administration (equal); writing – review and editing (equal). Chima J. Nwaogu: Conceptualization (equal); formal analysis (equal); investigation (equal); methodology (equal); project administration (equal); supervision (equal); writing – review and editing (equal).

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no conflicts of interest.