Introduction

Coastal ecosystems are deeply susceptible to changes caused by localized and direct human pressure (e.g., urban, industrial and agricultural development, pollution, overfishing; Crain et al. 2009; Brown et al. 2018), as well as due to global processes of sea-level rise and more frequent extreme weather events (Hoegh-Guldberg and Bruno 2010; Doney et al. 2012), often resulting in the loss of habitat and biodiversity. Consequently, species depending on coastal systems, such as terns nesting in low-lying areas, are particularly at risk of losing suitable sites for breeding and resting (e.g., Reynolds et al. 2015), but also important resources and foraging grounds.

Most tern species use coastal environments, where they forage mainly on fish (Shealer 2001). Terns are exceptional flyers, feeding on the wing and often performing long migratory journeys (e.g., Widen et al. 2014; Rueda-Uribe et al. 2021; Piro and Ornés, 2022), some of which are among the longest known in animals (Egevang et al. 2010). Several tern species are highly gregarious, frequently roosting and foraging together in large flocks all year-round (e.g., Erwin 1978; Bugoni et al. 2005; Correia et al. 2019). During the breeding season, terns often nest in dense colonies of thousands of pairs located on islands and sandbanks, and forage in the surrounding waters during the day (Cabot and Nisbet 2013), usually performing shorter but more frequent foraging trips while feeding chicks than while incubating (Oppel et al. 2018). Nocturnal foraging is considered infrequent in most tern species, as they are visual predators and low light levels may reduce their foraging efficiency (Cannell and Cullen 1998; Regular et al. 2011). However, this behaviour may, in some cases, be underestimated due to the difficulty of observing foraging birds at night.

The use of tracking devices has improved our understanding of bird ecology, mostly regarding migration and foraging behaviour and distribution (e.g., Catry et al. 2014; Hedenström et al. 2016; Alonso et al. 2018). Although to a lesser extent, tracking studies have also been used to investigate more accurately the frequency of nocturnal activity in seabirds (Phalan et al. 2007; Pratte et al. 2021). By understanding where and when seabirds go at sea, important sites for foraging have been identified for many seabird populations based on tracking data, allowing for the assessment of spatial overlap with potential threats (e.g., Le Corre et al. 2012; Gremillet et al. 2015). Tracking studies have, therefore, proven to be a critical tool in seabird conservation (Beal et al. 2021; Davies et al. 2021). Despite the current boom in seabird tracking (Bernard et al. 2021), those targeting tropical breeding terns are scarce (but see Soanes et al. 2015; Neumann et al. 2018), with virtually no published studies in West Africa. For this reason, knowledge on foraging habits and locations for these species are mostly based on direct, land-based (Catry et al. 2009a, b; Correia et al. 2019) and at-sea observations (e.g., Surman and Wooller 2003), or derived from dietary analysis (Correia et al. 2018). These methods, however, usually fail to provide accurate information on location and behaviour at the individual level, and do not cover the entire distribution range of individuals nor their nocturnal activity.

The West African Crested Tern (Thalasseus albididorsalis) is the most numerous breeding tern in West Africa with about 77 000 breeding pairs (Veen et al. 2022; Wetlands International 2022). As a recently described species, split from Royal Tern (Thalasseus maximus; Collinson et al. 2017; Clements et al. 2022), it has a restricted breeding distribution (from Mauritania to Guinea), with most individuals breeding in Senegal and Guinea-Bissau (Veen et al. 2022). Guinea-Bissau, in particular, holds up to 38% of the global breeding population, with colonies located on coastal sandbanks (Veen et al. 2022). West African Crested terns forage mostly on pelagic and, to a lesser extent, benthopelagic fish (Dunn 1972; Veen et al. 2018a, b; Correia et al. 2018). During the breeding season this species is a single-prey loader, i.e., carries only one prey per trip (e.g., Stienen et al. 2015), which may limit its foraging distribution while rearing chicks (Cairns et al. 1987). Yet, the foraging and breeding ecology of the species remains understudied, and is often regarded as similar to that of the Royal Tern (Billerman et al. 2021).

In spite of recognition as a separate species, the extinction risk of the West African Crested Tern has yet to be evaluated by the IUCN. The small global population of this species calls for an urgent assessment, and a prioritization of efforts to improve understanding of its ecological requirements, such as identify key foraging areas across the annual cycle, particularly for the most important breeding colonies. Such efforts are also essential to comprehend the potential threats faced by this species. While the most pressing threats to terns in West Africa have been generally identified (climate change, pollution, human disturbance and habitat loss, direct overexploitation, overfishing; Dodman et al. 2023), our efforts are essential to comprehend the potential threats faced by this species in particular.

Here we characterize, for the first time, the foraging behaviour of West African Crested Tern, at the second-most important colony of the species in the world, with ~ 19,000 breeding pairs (Veen et al. 2022). By tracking breeding individuals with GPS devices, we (1) analyse the variation in foraging trip characteristics from late incubation to early chick-rearing periods, (2) identify the most important foraging and resting sites for the population, and (3) investigate the importance of nocturnal foraging. Moreover, we evaluate the degree to which the existing marine Important Bird and Biodiversity Area (IBA) of the Bijagós Archipelago in Guinea-Bissau encompasses important foraging sites for this colony.

Methods

Study area

Fieldwork was carried out at Bantambur (Fig. 1; 11° 58′ 27.3″ N 16° 18′ 15.4″ W), a 0.04 km2 sandbank situated in the northwest of Guinea-Bissau. Guinea-Bissau is a country internationally recognized for its importance to coastal bird conservation, as it hosts thousands of migratory birds present during the non-breeding period (Robertson 2001; Dodman and Sá 2005; BirdLife International 2023). Bantambur is regularly used by a few seabird species for breeding and holds the second-largest colony of West African Crested Tern in the world (Veen et al. 2022), estimated at 19,182 breeding pairs, based on drone images in April 2021 (authors’ unpublished information), which represents 25% of the global breeding population of the species. The sandbanks of Bantambur lie within the Bijagós Archipelago marine IBA, the designation of which was based on its importance for the West African Crested Tern and Caspian Tern breeding populations, as well as for its relevance for the non-breeding population of Common Gull-billed Tern (Gelochelidon nilotica).

Fig. 1
figure 1

Movements of 18 West African Crested terns tracked during breeding from the colony on Bantambur (diamond), Guinea-Bissau. Each dot represents the furthest location from the colony of a foraging trip, where the colors indicate different individuals. Black lines indicate the full trajectory. Water depth (in meters) was obtained from ETOPO1 global relief model (Amante and Eakins 2009)

Fieldwork

Using a tent spring trap, 20 adult West African Crested Tern were captured in 20 different nests during late incubation, between 24 and 27 April 2021. All birds were ringed, weighed (to the nearest 10 g), and the maximum wing length (maximum chord method, to the nearest millimetre) and culmen (to nearest 0.1 mm) measured. A drop of blood was collected from the brachial vein from each bird and stored in 100% ethanol for molecular sexing. Thereafter, individuals were fitted with GPS devices (8 g nanoFix® GEO + RF Pathtrack) using leg-loop attachment technique (Mallory and Gilbert 2008; Thaxter et al. 2014). Deployed devices always represented less than 3% of the bird weight (Table S1). Two UHF base stations were set up in the colony, approximately 150 m from each other, to be able to retrieve GPS data from the entire colony area. GPS devices were programmed to collect a position every 10 min. To save battery power, only 6 loggers were programmed to be active 24 h per day, whereas the remaining 14 were programmed to be inactive from 00:00 to 05:00 a.m. each day.

The nests of trapped birds were numbered and marked with a small flag and subsequent visits were conducted one, eight and 14 days after the last captures to assess nest contents (classified as egg, chick or empty nest).

Molecular sexing

DNA was extracted from blood samples with E.Z.N.A Tissue DNA kit (Omega Biotek) following the manufacturer´s instructions. CHD-W and CHD-Z genes were amplified using the 2602F/2669R primers (van der Velde et al. 2017). PCR reactions were carried out in volumes of 10 µl that comprised 5 µl of Qiagen Multiplex PCR Master Mix, with 0.4 µl of each 10 pM amplification primer and 1 µl of DNA extract. PCR cycling conditions consisted of 10 min of initial denaturing at 95 °C, followed by 35 cycles of denaturing at 94 °C for 30 s, annealing at 62 °C for 60 s, extension at 72 °C for 60 s, and a final extension of 2 min at 72 °C. The amplified fragments were separated by capillary electrophoresis.

Tracking data analysis

Tracking locations were assigned, according to nests visits, as pertaining to one of three breeding stage categories: “incubation”, “chick-rearing”, or “either”. “Either” was used to classify periods which occurred between two colony visits, in which the nest contained an egg during the earlier visit and which on the second visit it was not possible to estimate the chick age. As tern chicks become mobile and leave their nests soon after hatching, it was only possible to confirm the presence of a hatched chick in four of the studied nests, to thereby estimate hatching date. For the remaining nests we considered the “chick-rearing” period as beginning at the first date of confirmed absence of an egg in the nest. Even though the absence of the egg may be the result of predation, all terns continued returning to the colony daily and even increased the number of foraging trips, we are confident with our assumption that they were indeed rearing a chick. The breeding stage categories were only used to explore and visualize data, as analyses were made taking into account the date rather than the breeding stage (see below).

For all tracking data, travel speed was calculated in the forward direction, as the quotient of the distance and time between two subsequent locations. Points with a calculated speed > 20 m/s were removed, as they represent biologically implausible flight speeds (Wakeling and Hodgson 1992). This filter resulted in the omission of 43 locations (0.1% of all locations). Locations were then classified as either “resting” or “moving” according to a speed threshold of 0.2 km/h, which was set by identifying a break in the histogram of travel speeds (Supplementary Fig. S1; see also Tables S2 and S3, where we show that no major conclusions would change with an adjustment of the speed threshold up or down by 0.1 km/h). Next, tracking data were split into distinct foraging trips, defined as periods of at least 20 min spent at ≥ 2 km from the colony centre. Trip splitting was performed using the R package track2KBA (Beal et al. 2021). Trips were further classified as either containing a period of resting behaviour or not, and whether the bird spent the night away from the colony (designated as “overnight”) or not.

Foraging trip characteristics

Two devices transmitted for less than a week and were thereby excluded from subsequent analysis. Each trip was characterized in terms of duration and distance (total distance covered and maximum distance from the colony). Average trip distances and durations were calculated for each individual and separately for the incubation, chick-rearing, and “either” periods. Daily metrics of the number of trips taken and the total time spent away from the colony were also calculated. Individual-level values were then averaged together to give group-level estimates of trip characteristics for each breeding stage. Trips classified as “either” incubation or chick-rearing, were not included in these group-level estimates. Due to programmed gaps in location sampling between 00:00 and 05:00 a.m. for 12 individuals we identified 20 trips in which the individual did not return to the colony before midnight and was located in the colony the following morning at 05:00 a.m.. As it was impossible to know when the individual returned to the colony, the duration was removed for these trips.

To explain variation in foraging trip characteristics, a series of mixed-effects models were fitted using the R package lme4 (Bates et al. 2015). Linear mixed models (LMM) were fitted to the following response variables: trip duration, maximum distance travelled from the colony, total distance, daily time spent off colony, and the number of trips per day. Day of the year and sex were set as fixed factors, and bird identity was included as a random factor to account for individual variability. Log10 (duration, daily time off-colony) and square root (maximum distance and total distance) transformations were applied to response variables to approximate normality of residuals, except for the “trips per day” variable, for which the model was fitted with Poisson family (link = ”log”). Next, to investigate if the behaviour outside the colony changed across the season (as nest attendance patterns are likely tighter during chick-rearing), the probability of resting on a trip or spending the night off-colony were tested using binomial generalized mixed models (GLMM). The response variables in each GLMM were the binomial classification of trips as resting or not, and overnighting or not, with day of the year and sex as fixed factors and with individual bird as a random effect. Sex was excluded from the model of resting probability to allow for model convergence.

Finally, to test whether resting periods and overnighting have an effect on trip characteristics, two sets of additional LMMs were fitted to trip duration, maximum distance, and total distance. In the first set, the classification of trips as containing resting or not was used as the fixed factor, and in the second, the binomial classification of overnighting or not was used; individual was a random factor in both model sets. For the model of resting behavior, trips classified as overnighting were omitted to look at the effect of resting within single-day trips only.

Foraging trip consistency

To test whether terns show individually consistent and distinct foraging strategies, we estimated the repeatability for the following trip characteristics: duration, maximum distance, total distance travelled, latitude, and longitude of the distal location, and direction from the colony to the distal location. Repeatability measures range from 0 (low) to 1 (high) and test the null hypothesis that between-individual variance equals within-individual variance. We used the R package rptR to fit linear mixed-effects models, derive standard errors, p values and repeatabilities (Nakagawa and Schielzeth 2010). For each dependent variable, we defined day of the year (centered) as a fixed effect and estimated random slopes for each individual. As trip direction represents angles, this variable was modelled using a circular ANOVA; there is no method available for deriving p values for repeatability measures from circular ANOVAs, so we report the repeatability measure and the standard error (Patrick et al. 2014; Clay et al. 2019).

Water depth use

To estimate the water depths most commonly used by foraging West African Crested terns, we overlaid points identified as “moving” over the ETOPO1 global relief model (Amante and Eakins 2009), which has a resolution of 1 arc-minute (circa 1.81 km in the tropics; Fig. 1). Depth-use estimates were calculated for two data sets: for all “moving” locations occurring away from the colony, and for the furthest location from the colony of each foraging trip (n = 570).

Identification of resting sites

We aimed to identify sites often used as resting locations during foraging trips away from the colony, as these sites may be important for maintenance behaviours (e.g., sleeping, preening) and socializing (Cramp 1998; Billerman et al. 2021). To identify resting sites, we classified periods of “resting” behaviour lasting >  = 20 min (i.e., the bird was grounded during at least two consecutive GPS locations). To identify when certain resting periods constituted use (or re-use) of the same site we considered 500 m buffers around each resting location. Then, we combined all overlapping buffer areas to get resting sites. We then calculated how many individuals, and foraging trips included resting periods at each site, and mapped them to show their relative importance, in terms of the number of individuals and foraging trips occurring there, as well as their spatial configuration relative to the wider distribution of the population.

Identification of important foraging sites and current marine IBA coverage

To identify important foraging sites, we employed the method of Lascelles et al. (2016) and Beal et al. (2021). A central purpose of this method is translating tracking data from a sample of animals into delineated areas that can be assessed for their importance to global or regional conservation using standard criteria (e.g., areas used predictably by ≥ 1% of global population under ‘Demographic aggregations’ Criterion D1a; IUCN 2016). This method uses tracking data to (1) estimate core-use areas for each tracked individual, (2) estimate the representativeness of the sample for the wider population and (3) scale-up the pattern of overlap among tracked individuals to delineate areas of particular importance at the population level. As we were interested in identifying potential foraging sites, we excluded all “resting” locations from the analyses. Moreover, we used behavioural segmentation techniques (Hidden Markov Models and Expectation Maximization Binary Clustering) based on speed and turning angle to distinguish foraging from transiting behaviour. However, these analyses proved ineffective in identifying different behaviours, likely due to the fact that terns often forage in transit, therefore, requiring either higher resolution data or further data streams (e.g., accelerometer) to identify distinct periods of foraging. All “moving” locations were, therefore, included in the analyses.

We calculated utilization distributions (hereafter UDs) using kernel density estimation for each individual bird. UDs were derived for three different data groups: first using all data across the study period for each individual, second using only data during the incubation phase (incubation), and finally only during chick-rearing (chick-rearing). UDs were derived separately for incubation and chick-rearing to allow for identifying important sites specific to each stage. The same smoothing factor was used for each data group, and was calculated as the log of the median foraging range (i.e., the average maximum distance travelled from the colony across all tracked individuals; h = 4.01 km). After deriving individual UDs, we estimated the representativeness of each data group for the wider population distribution (see Lascelles et al. 2016). The representativeness estimates were used to scale-up the among-individual overlap among UDs to estimate the number of individuals in the wider population regularly occurring throughout the study area during each breeding phase (see Beal et al. 2021 for details regarding the scaling-up procedure). Following Lascelles et al. (2016) we defined “potential important sites” as areas used by >  = 10% of the studied population (Beal et al. 2021; Lascelles et al. 2016). We then mapped the areas identified as important throughout the whole sampling period (i.e., all stages), and the important areas specific to incubation and chick-rearing. However, data from the whole period was highly representative of the respective breeding stage distributions (whole period = 94%, incubation = 92%, chick-rearing = 93%; Supplementary Fig. S2), and little difference was found between the spatial extent of the most important areas identified for each breeding stage (Supplementary Fig. S4). In fact, the most important area identified using data across the whole studied period covered 91% and 88% of the important areas identified with incubation-only and chick-rearing-only data, suggesting only small differences in spatial distribution across breeding stages (Fig. S4). Therefore, we present and discuss only the results concerning the whole period, and used this data to assess the spatial overlap between the identified important sites and the existing marine IBA (BirdLife International 2023).

Results

Tagged birds and trip characteristics

From the 20 birds captured and tagged with a GPS device, 15 were females and five were males. Females were on average smaller and lighter than males, but overlapping in all measurements (Table 1). A total of 570 foraging trips were identified for the 18 individuals that transmitted data for more than a week (Fig. 1; mean of 32 trips per individual, SD ± 14). Birds were tracked on average for 19 days (SD ± 4.2).

Table 1 Measurements (mean ± SD and range) of female and male West African Crested terns breeding on Bantambur, Guinea-Bissau

During the whole tracking period, terns took an average of 2 trips per day (min = 1, max = 4), travelling an average of 102 km per trip, up to 120 km from the colony (Table 2; Fig. 1). Foraging trips had an average duration of 3.4 h (IQR 1.7–5.6), with birds spending 30% of each day away from the colony, on average (Table 2).

Table 2 Foraging trip characteristics of breeding West African Crested terns by breeding stage. Values represent the mean of the medians calculated per individual bird, with values in parentheses indicating the interquartile range

As the season progressed from late incubation to early chick-rearing, several significant changes in foraging trip characteristics were identified. Terns took significantly shorter trips (β = − 0.04, p < 0.05, 95% CI [− 0.08, − 0.01]) in terms of duration, but with no significant change in the total distance travelled nor the maximum distance displaced from the colony (Fig. 2A, B; Supplementary Table S4). The probability of either resting on a trip or spending the night away from the colony decreased significantly across the season (β = 0.82, p < 0.05, 95% CI [0.72, 0.92]; β = 0.90, p < 0.05, 95% CI [0.81, 0.99], respectively; Fig. 2C, D; Supplementary Table S5). The number of trips taken per day increased significantly with advancing season (β = 1.03, p < 0.01, 95% CI [1.02, 1.04], while the total daily time spent away from the colony showed little change (Fig. 2E, F; Supplementary Table S6). No differences were found between sexes for any of these trip characteristics (Supplementary Tables S4, S5 and S6).

Fig. 2
figure 2

Predicted effects of seasonal advancement on foraging trip behavior of West African Crested Tern breeding on Bantambur, Guinea-Bissau. A The duration and B maximum distance travelled from the colony; C probability of resting during a foraging trip, and D spending the night away from the colony; E number of foraging trips performed per day, and F the total hours spent away from the colony per day. Data underlying AD are foraging trips (n = 570), and E, F are daily totals (n = 322). Dot color (in all panels) indicates breeding stage of the tracked bird (either represents a period of uncertainty between research visits to the colony, in which the nest could have either been in late incubation or early chick-rearing). Predictions (A, B, E, F) are from linear mixed-effect models, and binomial generalized mixed-effects models (C, D). Black solid lines, where present (A, C–E), indicate the trend of a significant effect of day of the year on each behavioral metric, while red lines indicate 95% confidence intervals

Overall, periods of rest (i.e., when the bird was stationary during at least 20 min) were detected during 40% of all foraging trips (see Table 2 for season-specific values of resting). Single-day trips (i.e., no overnighting outside the colony) including a period of rest were significantly longer in duration, total distance travelled, and maximum distance from the colony than trips with no rest (Fig. 3; Supplementary Table S7). Terns spent the night away from the colony on 10% of all trips, and such overnight trips were significantly longer in duration, total distance travelled, and maximum distance from the colony (Fig. 3; Supplementary Table S8).

Fig. 3
figure 3

Effect of resting on a trip or overnighting on (A) trip duration, (B) maximum distance travelled from the colony and (C) total distance travelled by trip. “Single-Day” represents trips beginning and ending within the same 24 h span (single-day trips), and “Overnight” denotes trips in which the tern spent the night off-colony. Single-day trips were further split between trips containing periods of resting and those without. Two separate linear mixed models were fit for each trip characteristic, one to test the effect of resting, the other to test overnighting. Both resting and overnighting showed significant effects on all three trip characteristics, making trips longer in distance and duration

Within overnight trips, 98% included nocturnal flights (i.e., between astronomical sunset—20:30 p.m.—and sunrise—06:15 a.m.) further than 2 km from the colony or 500 m from resting sites. Foraging trips were performed mostly during daylight hours; however, 10.7% of all trips departed from the colony during the nocturnal period, mostly between 05:00 and 06:00 a.m. (Supplementary Fig. S3). On average, 20.1% (SD 15.9%) of all foraging trips included nocturnal flight activity away from either the colony or resting sites. Nocturnal activity was more frequently associated with departures from resting sites than from the colony (Fig. 4). Most of this activity was detected around 1 h before the astronomic sunrise; however, some nocturnal trips were performed in the middle of the night (Fig. 4).

Fig. 4
figure 4

Nocturnal flight and resting activity of African Crested terns breeding on Bantambur (diamond), Guinea-Bissau. A Small, colored dots represent nocturnal flight locations of individual terns (each color represents a different individual) and large red dots depict nocturnal resting locations. The solid blue area represents the group-level 50% core utilization distribution of in-flight locations at night, and the dotted line is the 95% use area. The black diamond represents the breeding colony. Hourly distribution of nocturnal locations from trips which departed from resting sites (B) and from the colony (C) with the astronomical sunrise and sunset marked with dotted lines

Individual terns were significantly repeatable in their trip characteristics (Supplementary Table S9). Distal locations visited by terns were highly repeatable (latitude: R = 0.44 ± 0.09, p = 0.001; longitude: R = 0.48 ± 0.08, p = 0.001; direction from colony: R = 0.53 ± 0.09). Trip characteristics were less repeatable, but were still significantly different than expected at random (duration: R = 0.31 ± 0.08, p = 0.001; total distance: R = 0.35 ± 0.08, p = 0.001; max distance: R = 0.32 ± 0.07, p = 0.001).

Off-colony distribution

West African Crested terns tracked from Bantambur colony used shallow coastal waters in both Guinea-Bissau and Senegal (mean depth: 8.5 ± 4.2 m at the distal trip locations; 6.3 ± 3.0 m of all off-colony points; n = 570; Fig. 1). The most important foraging sites identified in the present study (areas used by >  = 10% of the colony population) occupy an area of 4006 km2, corresponding to 40% of the total area used by the terns (Fig. 5). We estimate that within these areas between 10 and 94% of the colony population predictably occur, translating to 3800 and 36,000 birds in total.

Fig. 5
figure 5

Identification of important foraging areas and resting sites of West African Crested Tern breeding on Bantambur (diamond), Guinea-Bissau. Filled-in area (in a gradient of black–blue–white) indicates sites estimated to be predictably used by 10–94% of the population breeding at Bantambur (i.e., 3800–36,000 individuals of the total ~ 38,000 birds breeding at the colony), from late incubation and early chick-rearing. The red line indicates the 95% used area of the colony. White-to-red circles indicate sites used for rest during foraging trips, where the size of the dot indicates the number of tracked terns using the site, and the color indicating the number of foraging trips with visits to each site. The dotted black lines represent (from south to north) the borders of the Bijagós Archipelago—marine (Guinea-Bissau), Casamance National Park (Senegal) and Kalissaye Avifaunal Reserve Important Bird and Biodiversity Areas (IBA)

The borders of the Bijagós Archipelago marine IBA captured 56% of the most important foraging areas, and 50% of the whole area used by the terns (Fig. 5). The foraging areas falling outside the marine IBA were largely situated along the coast of southern Senegal and in more distant, deeper waters of Guinea-Bissau (Fig. 5).

For the whole tracking period, we identified 51 resting sites (Fig. 5), which correspond to exposed sand banks and beaches. The most important site, used by 14 out of 18 individuals, was situated near Varela in the northwest corner of Guinea-Bissau, close to the border with Senegal, and more than 50 km away from the colony. Of the 51 resting sites identified, 46 (88%) were situated inside the limits of the existing marine IBA. Nocturnal foraging activity occurred in a restricted portion of the full foraging distribution of the population, with core areas located around the colony and northwest of the breeding colony (up to 115 km from the colony). The more distant foraging area used at night was also located in the vicinity of the most important resting site, located in northwest Guinea-Bissau (Fig. 4).

The “either” category represents a period of uncertainty between research visits to the colony, in which the nest could have either been in late incubation or early chick-rearing.

Discussion

This study provides novel information on the foraging movements of West African Crested terns in the second-largest breeding colony for the species, situated in northwestern Guinea-Bissau. The tracked terns foraged mostly in shallow waters (average depth of all off-colony points 6.3 ± 3 m), along the coasts of Guinea-Bissau and Senegal, but also travelling towards western deeper waters. We estimated that the most important foraging areas, i.e., areas used by 10–94% of this population (ca. 3800 to 36,000 individuals), cover approximately 3500 km2, within 100 km from the colony. Since Bantambur represents 25% of the global breeding population of the species (i.e., around 19,000 of the global 77,000 breeding pairs, Veen et al. 2022; Wetlands International 2022), we may assume that the important foraging areas identified in the present study are predictably used by 2.5–23.4% of all breeding West African Crested terns, suggesting these are highly relevant for conservation. During incubation, half of all foraging trips included a resting period and in a fifth of the trips, terns spent the night outside the colony. During chick-rearing terns were nearly four times less likely to rest or spend the night off colony.

Trip characteristics and foraging distribution

From late incubation to early chick-rearing, West African Crested terns decreased the duration of the trips, but foraged at similar distances from the colony. Terns did not increase the total amount of time spent off-colony per day during chick-rearing, instead increasing the number of trips taken per day. Although tracking studies comparing foraging movements between the incubation with chick-rearing periods in terns are rare, single-prey loaders, as West African Crested terns, are expected to limit their foraging range while feeding chicks (Cairns et al. 1987) and compensate by increasing the number of trips taken per day. The median of two trips per day recorded here may seem rather low when translated to a rate of food provisioning (four feeding events per day by the two parents), and although our ad-hoc observations at the colony suggest that foraging trips during chick-rearing are considerably longer than 10 min, we cannot completely rule out the hypothesis that a few short trips were missed due to the 10 min interval in GPS fixes. Nevertheless, our results were similar to that recorded for the Royal Tern in coastal Louisiana (although chick-rearing and incubation trips were pooled together in that study; Rolland et al. 2020). Surprisingly, this relevant metric can hardly be found in tern tracking studies. Food provisioning rates in small to medium-sized terns during early chick-rearing are variable but overall higher (e.g., Little Tern Sternula albifrons, ~ 0.6–2 events/chick/h in broods of more than one chick, Paiva et al. 2006; Common Tern Sterna hirundo and Arctic Tern Sterna paradisea, ~ 4 events/hour in broods of one chick, Robinson and Hamer 2000 or Sandwich Tern Thalasseus sandvicensis, ~ 0.25–0.5 events/chick/h in broods of one or two chicks, Stienen et al. 2000), indicating a larger number of daily trips. Radio-tracked Caspian terns (Hydroprogne caspia) were shown to perform on average six trips per day while feeding two-chick broods, and the provisioning rate of these broods were more than twice the rate of single-chick broods (Anderson et al. 2005). Therefore, the number of daily trips for single-chick brood Caspian tern parents are quite similar to the value we found for West African Crested terns.

Our study shows that West African Crested terns perform longer foraging trips than other Thalasseus species; i.e., the foraging distance from the colony and the total distance travelled were at least two times longer as compared to Royal Tern, Sandwich Tern and Great Crested Tern (Thalasseus bergii; McLeay et al. 2010; Fijn et al. 2017; Gatto et al. 2019; Rolland et al. 2020). Among terns, only the tropical and pelagic Sooty Tern (Onychoprion fuscatus) typically perform longer trips than the ones recorded in our study, both during incubation and chick-rearing (Soanes et al. 2015; Neumann et al. 2018). The studied colony, at Bantambur, holds a much higher number of breeding pairs as compared to those of the other previously tracked Thalasseus populations (Fijn et al. 2017; Rolland et al. 2020; McLeay et al. 2010). This may imply stronger competition for food in the waters around the colony, forcing terns to forage further away, as shown in several other seabird studies (Ashmole 1963; Lewis et al. 2001; Patterson et al. 2022). Nonetheless, as the breeding season advanced, we did not find a concomitant increase in the maximum distance terns travelled from the colony, which is the expected response to a competition-induced decrease in prey availability near the colony. The differences found in the length of foraging trips between our study and previous work, may, however, be linked to differing habitat availability and environmental characteristics around the colony. Royal terns in Louisiana rely largely on marsh habitats close to the colony (Rolland et al. 2020), and Sandwich and Crested terns in the Netherlands and South Australia, respectively, forage in temperate, highly productive waters (Fijn et al. 2017; McLeay et al. 2010). For the tropical West African Crested terns studied here, areas of enhanced productivity are mostly located further away from the colony (see below).

While it could be expected that terns skip foraging in areas close to the colony to avoid potential competition, instead flying directly to forage in more distant yet profitable areas (Fijn and Gyimesi 2018; Lieber et al. 2022), we were unable to distinguish between commuting and active foraging behavior (see methods). These results suggest that, as with many other terns (Rueda-Uribe et al. 2021), West African Crested Terns forage while in transit, possibly along the whole trip, and therefore, most of the range used would represent true foraging area.

Individual foraging specialization has been shown in several seabird species, including terns (e.g., Nisbet 1983; Beal et al. 2022). As individuals specialize in space use and foraging behaviour, this can have important implications for how populations respond to environmental changes (Bolnick et al. 2003; Geary et al. 2019). Individual consistency in foraging behaviour is thought to be less prevalent in tropical marine environments (Weimerskirch 2007; Soanes et al. 2021), as a consequence of lower resource predictability, compared to temperate and polar regions (Longhurst and Pauly 1987; Ballance and Pitman 1999). Consequently, in tropical waters some seabird species rely largely on associations with sub-surface marine predators to feed (Veit and Harrison 2017; Correia et al. 2019). However, this does not seem to be the case for West African Crested terns, which are rarely observed foraging in such associations in the coastal waters of Guinea-Bissau (pers obs., Correia et al. 2018). The fact that the studied terns seem to show individual foraging site fidelity, as depicted by the repeatability of foraging trip distances, durations, latitude, longitude, and distal point directions, suggests that individuals may have preferred sites for foraging (Patrick et al. 2014).

Water depth must play an important role in determining the foraging distribution of West African Crested terns during the breeding season, since they foraged primarily in shallow waters, usually not exceeding 10 m in depth. Many tern species show preference to forage in shallow waters (Rock et al. 2007; McLeay et al. 2010; Beal et al. 2022), though this is not a universal rule within coastal tern species or populations (e.g., Surman and Wooller 2003; Fijn et al. 2017). West African Crested terns forage mostly on pelagic fish (Correia et al. 2018), which are more available near the water surface in shallow waters, as this habitat provides forage, shelter, spawn and nursery opportunities for small fish (e.g., Blaber et al. 1995). Indeed, several fish species use the coastal shelf ecosystems of West Africa as nursery grounds (Fréon 1988; Brainerd 1991; Correia et al. 2021). Surprisingly, the tracked terns did not exploit the shallow waters of the Bijagós Archipelago (to the south of the colony), which are largely acknowledged as a key nursery area for fish within the region (Lafrance 1994; Arkhipov et al. 2015) and are within the range of trip distance recorded. Terns may be, however, avoiding competition with the conspecifics breeding in the south of the Bijagós Archipelago, even though that colony is much smaller (Veen et al. 2022), or else with other tern species that are very abundant in those waters (Dodman and Sá 2005; van Roomen et al. 2018). Instead, terns foraged to the north of the colony, particularly in the coastal waters of northern Guinea-Bissau and southern Senegal, and towards the west, which may coincide with the higher productivity areas of the Canary Current upwelling during the studied period (April–May; Roy 1998; Valdés and Déniz-González 2015). Thus, both shelf and upwelling areas seem to be favored by West African Crested Tern to find prey. The intensive fishing activity towards the north of the colony may also attract terns to feed on fishery discards (e.g., Oro and Ruiz 1997), which may not be so frequent in the Bijagós Archipelago, where fishery is much less intensive (Campredon and Cuq 2001).

Resting behaviour and nocturnal activity during foraging trips

Our results show that terns often roost on sandbanks and beaches. During single-day trips, terns that stopped to rest took longer trips, travelled greater distances, and foraged further away from the colony. Still, some terns were able to perform trips of a similar length without resting, suggesting that resting may not always be physiologically necessary for terns to achieve longer trips. Nevertheless, as the likelihood of resting on a trip decreased from incubation to chick-rearing, this suggests that with the increased demand to feed chicks, adult terns spend less time resting, resulting in more trips per day of a shorter duration. The most important resting sites identified for West African Crested terns, are most likely communal roosts used by a large number of individuals which may serve as information centres, where birds gain information on prey distributions (Ward and Zahavi 1973; Wright et al. 2003). Despite the opportunity to exchange information within the breeding colony, the position of each nest and the density of the colony may limit these interactions, which can be better achieved in communal roosts (Evans et al. 2016). Resting sites can further provide opportunities for self-maintenance activities, such as preening or bathing, which is also considered a social activity in some species (Gochfeld and Burger 2020). Our results suggest that the presence of suitable sandbanks and beaches within 100 km of the colony may be highly important for West African Crested Tern during the breeding season. Thus, protecting these resting sites is likely critical for a successful management of this breeding population.

Some tern species are known to overnight outside the colony (e.g., Black Tern Chlidonias niger, van der Winden 2005; Least Tern Sternula antillarum, Atwood 1986) as a strategy for reducing predation of eggs and chicks if the adult presence attracts predators, or to enhance survival if predators threaten adults. West African Crested terns nest in large and highly conspicuous colonies, and on Bantambur they breed along with several hundred Gray-hooded gulls Chroicocephalus cirrocephalus and Slender-billed gulls Chroicocephalus genei, which are the main predators of their eggs and small chicks (pers. obs.; Veen et al. 2022). Given the absence of other important predators, it is unlikely that adults overnight outside the colony to increase clutch survival. We show that the median distance from the colony for overnight trips was on average up to three times longer than for single-day trips (during chick-rearing). Therefore, overnight trips may enhance foraging further away from the colony, but also allow birds to exploit a wider foraging area, by travelling overall longer distances. Moreover, terns may spend the night outside the colony as a result of less successful foraging days, choosing to sleep closer to the foraging areas and avoiding the costs of returning to the colony, as shown in other seabird species (Zavalaga et al. 2012).

Terns are mostly considered visual predators foraging primarily during daylight hours. Nocturnal foraging seems, therefore, rare in this bird group, and it has been mostly observed in Sooty Tern (Gould 1966; Brown 1975), and to a much lesser extent in Roseate Tern Sterna dougallii (Pratte et al. 2021). Still, some species may start or end their foraging activity during the nocturnal period, just before sunrise or after sunset, respectively (e.g., Royal Tern Buckley and Buckley 1974; Black Tern, van der Winden 2005). Nocturnal movements of West African Crested terns were most frequently related to trip departures up to 1h30 before the sunrise rather than returning after the sunset. Yet, our results suggest that a large part of the nocturnal movements detected were not associated with the colony but instead with birds departing from resting sites. Although our sample size during the entire nocturnal period is rather small, since only six tracking devices were collecting data during the whole night (the remainder stopped collecting data between midnight and 05:00 a.m.), almost one quarter of all trips included nocturnal flight activity. These results suggest that nocturnal activity may be frequent in West African Crested terns, and should be further confirmed and better understood in future tracking studies. Among other things, the effect of the lunar cycle on the nocturnal foraging of terns could be investigated, since moonlight brightness can enhance visibility for travelling and foraging. Indeed, the full moon coincided with the beginning of the fieldwork (April 27), and therefore, the first half of the tracking period included moonlit nights.

Adequacy of the recently designated Bijagós Archipelago marine IBA for the conservation of the West African Crested Tern

The world population of the West African Crested Tern is estimated at less than 77,000 breeding pairs, mostly restricted to a few colonies in Senegal and Guinea-Bissau (Veen et al. 2008, 2022; Wetlands International 2022). Despite the low numbers and narrow distribution during breeding season, the species lacks a recognized conservation status, as a result of the recent split from the Royal Tern (Collinson et al. 2017; Clements et al. 2022). Likewise, the species lacks a conservation action plan, despite the recent efforts which have resulted in the designation of marine IBAs along the West African coast (Dias et al. 2016). Our results show that the existing marine IBA in Guinea-Bissau covers only half of the important foraging areas used by the population of West African Crested Tern breeding on Bantambur, highlighting the importance of population-level studies for informing spatial conservation measures (Jovani et al. 2015; Soanes et al. 2013). Moreover, we found that the important foraging areas identified may regularly hold up to 23.4% of the global breeding population of the West African Crested Tern, and, therefore, are relevant for the global persistence of the species, meeting the Criterion D1 for a Key Biodiversity Area (KBA) qualification (IUCN 2016, 2022).

The present study allowed for the identification of new sites that should be considered in a re-evaluation of the current marine IBA borders. Among the sites identified is a hitherto unknown area of importance in southern Senegal, highlighting the need for national and international management efforts for the conservation of this species. Still, the designation of an enlarged marine IBA may be insufficient to ensure the conservation of the Bantambur colony of West African Crested Tern. Existing sandbanks in the coastlines of Guinea-Bissau and Senegal risk complete inundation and disappearance in face of the combined effects of large tidal ranges and sea-level rise. Precipitation instability may also lead to limited terrestrial sediment discharge which is insufficient to replenish active sandbanks (Morawski 2019). The climate change prospects for Guinea-Bissau are particularly concerning as the ongoing changes have already caused significant habitat loss and few new potential breeding sites have been identified (Morawski 2019). The Bantambur bank itself has suffered a reduction of almost 40% of its area since 2007 (Morawski 2019; Insambé 2022). While at sea, West African Crested terns may also suffer with the increasing fishing pressure in West Africa, both from prey declines but also from the increase in interactions with fishery gear (Gremillet et al. 2015). In fact, high declines in fish biomass have been observed in the Eastern Central Atlantic (Campredon and Cuq 2001; Ecoutin et al. 2010; Belhabib et al. 2017), where 40% of the assessed fish stocks are considered to be exploited to unsustainable levels (FAO 2022). This assessment is certainly underestimated, as this region is highly affected by illegal, unreported and unregulated fishing activities (Doumbouya et al. 2017; Welch et al. 2022). Moreover, small pelagic fish, the most important prey for West African Crested terns (Dunn 1972; Veen et al. 2018a, b; Correia et al. 2018), are the most fished group in the West Africa region (Failler 2014; FAO 2022), with some species showing declines over 30% of their populations in the last decade (Polidoro et al. 2016). While local action will not suffice to reverse the current trend of disappearing sandbanks and increasing fishing pressure in Guinea-Bissau and Senegal, studies like the present one, are vital to ascertain the ecological requirements of the West African Crested Tern. Such information, on distribution and behaviour, is critical to the design of conservation action plans to ensure the future of this recently described seabird species and other coastal species inhabiting the region. Indeed, this region has unquestionable value for many coastal seabird species, including the populations of Caspian Tern (breeding) and Common Gull-billed Tern (non-breeding) that triggered the criteria for marine IBA proposal (BirdLife International 2023), but also for other non-breeding tern species, as Little, Common, Black and Sandwich terns (Dodman and Sá 2005; van Roomen et al. 2018). Most of these species strongly overlap their ecological requirements with the West African Crested Tern, and will, therefore, certainly benefit from conservation management targeting this species.