Introduction

Threats to wildlife can drive small and isolated populations to extinction. In particular, oceanic island species have suffered greater extinctions than continental species, with endemic species having a higher extinction rate than non-endemic species. Endemic species are restricted to a geographic range, sometimes associated with specialised environmental niches, limited dispersal capacity, small populations and reduced adaptive capacity. These factors make them susceptible to threats such as habitat loss, the impact of invasive species, overexploitation or the effects of climate change, reaching a threshold where stochastic factors, including Allee effects, amplify extinction risks (Frankham 1998; Duncan and Blackburn 2007; Chichorro et al. 2019; Lucas et al. 2019; Staude et al. 2020). Furthermore, Manes et al. (2021) emphasize the dire extinction prospects for island and mountain endemics under climate change. As the climate shifts, many species relocate to higher altitudes seeking climates more conducive to their survival (Walther et al. 2002; Walther 2010; Couet et al. 2022).

Translocation programs have become important tools in the conservation and management of endangered species, but their success depends on rigorous evaluation and testing, and the accumulation of further experimental evidence with different endangered species in different environments. Intensive management, like reintroduction programs, offers conservation benefits to threatened species, often averting potential extinctions (Armstrong and Seddon 2008; IUCN 2013; Seddon et al. 2014; McGowan et al. 2017).

Paramount to the translocation process is the methodology employed during reintroductions, with hard- and soft-release strategies each offering distinct advantages that can influence post-release survival and adaptation. The choice between hard- and soft-release methods in translocation projects remains a central discussion in conservation biology. Soft-release, involving a confinement period at the release site, often yields superior survival and reproduction outcomes due to enhanced acclimatization, especially in species with strong site fidelity or predation risks (Armstrong and Seddon 2008). However, it might present logistical and financial challenges. Conversely, hard-release, which involves direct release without acclimatization, is less complex and typically cheaper, yet may lead to increased initial mortality from predation or unfamiliar habitats (Teixeira et al. 2007). For hard-released wild birds, issues like disorientation, higher predation risks, and food resource challenges might arise. Therefore, translocation success can be improved by testing different methods, as the most effective technique depends on several factors. For example, Attum and Cutshall (2015) found that soft-release of translocated turtles was effective in continuous lotic habitats and hard-release being effective in patchy wetland complexes. Similarly, Cabezas et al. (2011) concluded that the survival of translocated wild rabbits was significantly influenced by predation risk and physiological health, advocating for soft-release techniques to minimize predation threats and guarantee optimal physiological conditions upon release. Supporting this, Wanless et al. (2002), working with the flightless Aldabra Rail Dryolimnas aldabranus, provided evidence that the soft-release is recommended as the conservative and precautionary method of choice for avian reintroductions and translocations, by allowing individuals to acclimatise and providing additional energetic reserves for the period between release and self-sufficiency. Captive-reared Burrowing Owls Athene cunicularia hypugaea had higher site-affinity, survival, and reproductive performance when reintroduced using a soft-release (Mitchell et al. 2011). However, a uniform consensus remains elusive. Hardman and Moro (2006) argued that soft release techniques do not necessarily confer an advantage to the successful immediate establishment and survival of hare-wallaby species in the short term. On the same line, De Milliano et al. (2016) showed that soft-release did not confer a consistent and substantive advantage for captive-bred Eastern Barred Bandicoots Perameles gunnii. Therefore, the success of soft or hard release is contingent on the natural history of the translocated species or the habitat type in which it occurs.

The source of individuals, whether captive-bred or wild-caught, introduces another layer of complexity in translocation programs. Animals for translocation can originate from either their natural habitat in source or donor populations, or from captive breeding centres (Seddon 2010; IUCN 2013). The nuances in behaviour, adaptation capacity, and survival rates between these two groups necessitate a comprehensive understanding. In conservation efforts, the use of captive-bred animals for translocations is frequently favoured, especially in scenarios where the action might exacerbate the vulnerability of a dwindling wild population (Wilson and Price 1994). Historical data show that wild-to-wild translocations are generally more successful than releasing captive-bred animals (Griffith et al. 1990). Nonetheless, these approaches are not mutually exclusive. As exemplified by Kraus et al. (2017) in their work with Eastern Hellbenders Cryptobranchus alleganiensis, integration of both strategies can be successfully achieved in translocation programs.

The success of a translocation program can be determined by two phases: the establishment phase and the long-term persistence phase. Studying the establishment phase is useful because reintroduced populations may fail in this initial phase under conditions that allow them to persist after the establishment phase (Armstrong and Seddon 2008; Kemp et al. 2015; Brockett et al. 2022). Reintroductions may fail in the establishment phase due to high dispersal or low survival while animals recover from the stress of translocation or acclimatise to the release site. Dispersal and survival of reintroduced birds in the immediate post-release period may depend on release protocols, particularly the acclimatisation time and number of birds released, as well as the provision of supplementary food and water in the release environment (Scott and Carpenter 1987; Armstrong and Seddon 2008; IUCN 2013; Bernardo et al. 2011; Armstrong and Wittmer 2011). Appropriate post-release monitoring provides the metrics and feedback mechanisms that inform the adaptive management of translocations, which will contribute to their ultimate success (IUCN 2013).

Therefore, in order to evaluate the reinforced programme in its establishment phase, it is necessary to monitor the translocated animals closely to know their movements and immediate survival. Animals must be identifiable to allow individual monitoring. For small birds, combinations of coloured rings are often used, although advances in radio telemetry have allowed continuous monitoring of smaller species, such as most passerines (Kays et al. 2015; Taylor et al. 2017). Telemetry has become an increasingly powerful and accessible means of collecting much of the detailed information needed to effectively evaluate translocation projects (Whitford and Klimley 2019), but the use of telemetry is a balance of costs and benefits where researchers must offset potential negative effects of capture and tagging by maximising the benefits of their study (Kays et al. 2015).

The Gran Canaria Blue Chaffinch Fringilla polatzeki is a mountain passerine endemic to Gran Canaria that inhabits Pinus canariensis forests (Lifjeld et al. 2016; Sangster et al. 2016). The main causes of the species' decline in the past have been the destruction and fragmentation of the pine forest as a result of unsustainable logging and the indiscriminate collection of specimens for natural history museums. Currently, the greatest threats to the species are considered to be the reduced extent of the pine forest, the effects of climate change and the destruction of habitat by large forest fires. At the beginning of the twenty-first century, the species was restricted to the Inagua Nature Reserve. In 2008, after a large forest fire in this area, a few birds were found breeding successfully in another pine forest, La Cumbre, possibly due to birds fleeing the fire and settling in a pine forest with optimal conditions for the species. Therefore, this site was chosen to develop a reinforcement programme (population augmentation), starting with birds bred in captivity and subsequently translocating individuals captured in Inagua as a source population (LIFE + PINZÓN 2019). Preliminary analyses give a positive assessment of the program, which started with birds bred in captivity and recorded survival and reproductive success values similar to those of natural populations (Delgado et al. 2016). Subsequently, Illera et al. (2023) found that the reinforced population at La Cumbre, from juvenile individuals from captive breeding and captured from the source population of Inagua, seemed to be viable in the short term, considering the similar values of biometric measurements, reproductive success, body condition, genetic diversity, and inbreeding of wild-born birds in both populations. As a result, the reinforced program led to a 16% growth in its population, highlighting the importance of such actions in safeguarding bird species with limited populations confined to a single location (Carrascal et al. 2022). The total population size of the Gran Canaria Blue Chaffinch is estimated at 430 birds, restricted to a very small area of 60 km2 (Carrascal et al. 2022), making it the forest passerine with the smallest population size in the Western Palearctic, and therefore, still listed as Endangered by the IUCN (BirdLife International 2021).

The aim of this paper is to evaluate the establishment phase of the reinforced program of the endangered Gran Canaria Blue Chaffinch. We investigated the home ranges of released juvenile chaffinches—both captive-bred at the Tafira breeding centre and wild-caught from the Inagua source population—using radio-tracking. Previously, we examined the impact of radio transmitters on the birds' short- and long-term survival rates. Finally, we estimated the contribution of both bird groups to the reinforcement of the translocated population, considering the birds found in La Cumbre during the subsequent breeding season after the release. The findings of this research could inform the species' management strategies and offer insights for other reintroduction efforts targeting endangered forest passerines on islands.

Methods

Study site

The island of Gran Canaria (1560 km2) is located in the centre of the Canary Islands (Spain). The characteristics of the habitat occupied by the Blue Chaffinch are described in Carrascal et al. (2017, 2022). The pine forests of Inagua, in the west of the island, host the main population of the species, which served as the source population in the reintroduction programme (LIFE + PINZÓN 2019). It is a mature forest of Pinus canariensis that develops under semi-arid conditions. Most of these pine forests are protected in the Inagua Integral Nature Reserve (37.6 km2). The pine forests of La Cumbre (21 km2), where the birds were released, are located in the centre of the island and at a higher altitude than the previous one. It is an immature pine forest, the result of reforestation carried out in the middle of the last century in the central part of the island. The two pine forests are separated by 5 km and are poorly connected by forest stands (Fig. 1).

Fig. 1
figure 1

Location of Gran Canaria in the Canary Islands, showing the location of the Captive Breeding Centre of Tafira (star), the sites where wild birds were captured in Inagua (black dots) and the sites where releases took place in the translocation area of La Cumbre (white dots). Pine forests are shaded in green. The two lower panels of the figure show the location of 30-min units survey in search of colour-banded birds in Inagua and La Cumbre (colour figure online)

Full size image

Origin of translocated birds

The Cabildo de Gran Canaria managed a captive breeding centre in Tafira (Las Palmas de Gran Canaria), where there was a captive stock of breeding pairs in aviaries. The genotyped individuals used in the captive breeding programme had the same allelic diversity as the free-ranging birds in the wild source population, therefore, the released birds were genetically viable (Garcia-del-Rey et al. 2013). The captive breeding of blue chaffinches and release protocols are described in Díaz and Delgado (2021). In addition, wild birds were captured in Inagua, where the main population of the species remains (Carrascal et al. 2022). Captures were made using mist nets in late summer in places where the birds go to drink and bathe. Thus, between 2010 and 2019, 194 individuals of the Gran Canaria blue chaffinch were released in the pine forests of La Cumbre to strengthen the population. A total of 114 birds born in captivity and 80 translocated from the wild were released (see Supplementary Material S1). Ninety-three per cent of the birds released were juveniles born in the same year. The birds were released between the end of August and the beginning of October each year, when the young are able to feed themselves and are no longer dependent on their parents.

Wild-caught and captive-bred birds were released using different methods, informed by our primary focus on the conservation of this endangered species and the imperative to minimize juvenile mortality. Captive-bred individuals, which are costly to produce and accustomed only to aviary life, underwent acclimatization for 2–14 days in groups of 2–7 birds in 24–50 m3 aviaries located within the pine forest of their release site. This soft release approach allowed these birds to adapt gradually to their natural environment and the presence of other birds, while also providing supplementary food to ease their transition to the wild (De Milliano et al. 2016). Conversely, wild-caught birds were released using hard methods 9.5–14 km away from their capture site, after spending 24 h in transport boxes to mitigate stress. This method was chosen to minimize injury risks, such as damage to wings, beaks, and tails, which could occur in large aviaries during prolonged periods. Additionally, keeping them in darkness at night in small boxes allowed close monitoring of their condition, addressing the stress associated with capture and human contact. All released birds wore a combination of coloured rings that allowed individualised identification from a distance. Five supplementary feeding and water points were placed in the surroundings of the release site throughout the following five months.

Monitoring and telemetry

For this study, 49 individuals were fitted with radio transmitters before being released: 15 captive-bred birds (years 2013–2017) and 34 wild-caught birds (years 2015–2018). Model BD-2 devices (Holohil Systems, Ltd., Carp, Ontario, Canada) lasted approximately 20–42 days and weighed 0.75–0.82 g, in no case exceeding 3% of the birds' body weight (average juvenile weight = 27.4 g, sd = 1.73). A modification of the leg harness using 0.5 mm diameter absorbable surgical suture was used to attach the transmitters, so that the harness detached after a few months, minimising adverse effects on birds (Karl and Clout 1987). Two model R1000 receivers and a three-element Yagi antenna were used for monitoring. On a daily basis, two people carried out radio tracking and manually triangulated bird locations. We obtained at least one precise location of each bird per day and determined its status (alive or dead).

Between spring and summer of 2011–2020, we conducted intensive surveys and searches in the two populations to resight the released birds in 2010–2019, monitor their reproduction in subsequent breeding seasons, and observe other aspects outside the scope of this study. Surveys were conducted on foot along paths and cross-country tracks. In La Cumbre, surveys were throughout the whole pine forests, while in Inagua, due to its vast size, areas with less suitable habitat for the species were excluded (following Carrascal et al. 2017). From 2017 onwards, a more standardized survey approach was adopted, segmenting the effort into 30-min units and recording both initial and final geographic locations, providing a detailed spatial understanding of the survey effort: 2139 30-min survey periods were conducted in Inagua and 1526 in La Cumbre (Fig. 1; in 2011–2016, the sampling effort was similar although not georeferenced in detail).

Data analyses

Survival estimates were made with MARK (White and Burnham 1999) using “Live Recaptures (CJS)” and “Nest Survival” analysis. Spatial analysis of radio-tagged birds was performed with R software version 4.2.1. (R Core Team 2022) using the sp, AdehabitatHR and maptools packages (Pebesma and Bivand 2005; Bivand and Lewin-Koh 2023; Calenge and Fortmann-Roe 2023).

First, we examined the visual capture-resight history of coloured-ringed wild-caught birds from the springs and summers following liberation to test the effect of radio transmitters. We compared the annual survival of wild-caught birds released in La Cumbre during those years: 34 birds with transmitters and 28 birds without transmitters. Annual survival was estimated using MARK considering the effect of group (birds with and without transmitter), year of monitoring and the interaction between these two factors on survival (S) and resight probability (p). The Akaike Information Criterion for small sample sizes (AICc) was used to assess the fit of the models to the data, with those models with a difference in AIC (∆AIC) of less than two units considered to have a similar fit to the data (Burnham and Anderson 1998).

Second, we estimated the daily survival of the released birds with the radio-tracking data in La Cumbre for a maximum period of 42 days. In this analysis, we examined 15 captive-bred juveniles from Tafira and 34 wild-caught juveniles from Inagua, all released in La Cumbre between 2013 and 2018. The MARK models considered the effect of the origin of the birds (captivity or wild) and the time (days) after liberation. AIC values were also used to assess the fit of the models to the data.

Home range

The area occupied by radio-marked individuals during the study period was estimated using minimum convex polygons (MCP) for 95% of the locations, excluding 5% of the most distant locations. To obtain information on activity centres, home ranges were also estimated using kernel analysis with 95% encounter probabilities. The reference and least square cross validation (LSCV) smoothing parameter (h-bandwidth) was calculated to find the best kernel fit to the data (Seaman and Powell 1996; Fieberg 2007; Baíllo and Chacón 2021). To estimate home ranges, five birds with less than five locations were excluded (Pebesma and Bivand 2005). To avoid bias in the kernel activity centres due to heterogeneity in the number of contacts of each bird per day, the first location of each bird per day was used in cases where there were several locations per day. See Supplementary Material S2 for the home range areas of the studied birds.

Contribution of translocated birds to the reinforced population

Survival alone should not be used as a parameter for assessing the establishment phase of a translocation programme, as there may be birds that survive but do not contribute breeding individuals to the translocated population. This occurs when birds disperse widely or return to the area of origin. Therefore, as a surrogate measure of establishment success, we measured the number of translocated birds that remained in La Cumbre release area during the following breeding season, thereby contributing to the translocated population. The percentage of released birds sighted again in La Cumbre during the following breeding season was estimated for those coming from the captive breeding centre in Tafira or wild-caught in Inagua (using 10,000 bootstraps for each group of birds). The prospection of the pine forests of Inagua and La Cumbre in search of translocated birds (see Monitoring and telemetry section) allowed resightings and determined the dispersal/return rate of released birds during the reproductive period of the species. We also recorded whether the birds exhibited breeding behaviour (mating and territoriality).

Results

Survival analyses

The fate of six birds could not be determined because either the transmitter became detached, some birds disappeared and were not found, or the transmitters were emitting motionlessly from an inaccessible point, making it impossible to know whether the birds died or the transmitter became detached. Of the remaining 43 birds, 16 were found dead during radio-tracking. Causes of death were predation by a bird of prey (7 wild-caught birds; probably by Eurasian Sparrowhawk Accipiter nisus) and feral cats (2 captive-bred birds). Additionally, separate incidents involving wild-caught birds resulted in one death due to a collision with a vehicle on a road and another due to entanglement in a pine needle caused by a plastic ring. For the remaining birds (5), the cause of death could not be determined (Table 1).

Table 1 Summary of the fate of radio-tagged birds
Full size table

Table 2 shows the model selection for survival of wild-caught birds with and without transmitters. There was a slight overdispersion in the global model explaining survival and resight probability, so a c-hat adjustment was made. The null model, excluding the effect of transmitters and year of capture, had the highest weight (0.372). The mean annual survival and resight probability obtained by averaging the values of the three most plausible models was similar between the two groups: 0.459 (SE: 0.074, 95% CI 0.321–0.602, N = 34) for birds with transmitters and 0.488 (SE: 0.075, 95% CI 0.346–0.633, N = 28) for birds without transmitters. Therefore, the wearing of a transmitter had no effect on the long-term survival of birds.

Table 2 Selection of models to estimate annual survival (S) and resight probability (p) of wild translocated birds with and without radio-transmitters
Full size table

A daily survival analysis was performed for captive-bred and wild-caught radio-monitored birds. The time model was the most plausible with a weight of 0.574 (Table 3). Daily survival of both groups of birds during radio tracking, estimated by model averaging of the two best models with ΔAICc < 2, is shown in Fig. 2. Mortality occurred in the first few weeks after release, so daily survival increased with time. For example, on day 10, the estimated survival 95% confidence intervals were 0.967–0.992 for captive-bred juveniles and 0.969–0.993 for wild-caught juveniles, indicating comparable survival rates between the two groups during the initial post-release period. Therefore, the survival of transmitter-bearing birds, during the first 42 days of radio-tracking, was not affected by their origin, whether they were captive-bred or captured from the source population at Inagua.

Table 3 Selection of models to estimate the survival of translocated birds of both origins during the 42 days of radio-tracking
Full size table
Fig. 2
figure 2

Daily survival of captive-bred (A) and wild-caught (B) birds during the 42 days of radio-tracking immediately after release. Error bars show the 95% confidence interval

Full size image

Home range

A total of 993 locations were obtained from 14 captive-bred birds (372 locations) and 30 wild-caught birds (621 locations). The mean locations per captive-bred and wild-caught bird were 26.6 (range 5–38) and 20.70 (range 5–43), respectively. None of the individuals translocated from Inagua returned to the site of capture during radio tracking. The size of the pooled MCP95 was 142.25 ha and 907.10 ha for captive-bred and wild-caught birds, respectively (Fig. 3).

Fig. 3
figure 3

Home ranges during the 42 days of radio-tracking for the set of released birds. 95% Minimum Convex Polygons of released birds (A), where the larger polygon corresponds to wild-caught birds and the smaller polygon corresponds to captive-bred birds. For the 95% kernel estimation area (B), wild-caught birds are represented by the red area and captive-bred birds by the black area. The triangle indicates the point of release and the crosses indicate points of contact with released birds (colour figure online)

Full size image

Home range (in logarithm), estimated by the MCP95 method (Minimum Convex Polygon 95%), was significantly related to the number of locations of each individual (b = 0.102, se = 0.020, P < < 0.001) and was higher for wild birds from the Inagua population than for captive-bred birds (P = 0.00015). This ANCOVA model explained 46.8% of the variance in home range (in logarithm) of the 44 birds studied. The adjusted mean of the home range for captive-bred birds was 2.4 ha (95% confidence interval 0.9–6.1 ha; total range 0.2–69.9 ha), while for wild-caught birds, it was 25.8 ha (95% CI 13.6–49.4; total range 0.1–549.1 ha). Repeating the analysis for those individuals with 18 or more records (12 captive-bred birds, 15 wild-caught birds), the effect of number of records per individual no longer reached significance (P = 0.236), while the effect of bird origin remained significant (P < < 0.001), with a mean of 5.5 ha for captive-bred birds (95% CI 2.3–13.2 ha) and 83.9 ha for wild-caught birds (95% CI 38.4–183.1 ha).

For kernel analysis, the most appropriate bandwidth was h-ref, as h-LSCV was unstable for estimating the home range of some birds due to its low value (see Supplemental Material S1). The 95% kernel home ranges were significantly correlated with the 95% minimum convex polygon estimates for those radio-tagged individuals with 18 or more locations (r = 0.82, P < < 0.001). Repeating the analyses with the 95% kernel estimates yields nearly the same results regarding significance patterns (lack of significance of the number of records per individual—P = 0.738—and strong effect of bird origin—P < 0.001—), and home areas: adjusted means of 6.8 ha for captive-bred birds (95% CI 2.7–16.8 ha) and 85.6 ha for wild-caught birds (95% CI 38.1–194.4 ha).

Contribution of translocated birds to the reinforced population

All translocated birds from 2010 to 2019 found in La Cumbre during the following springs and early summers showed breeding behaviour typical of the species (mating, singing males, fidelity and territorial defence, etc.). The contribution of captive-bred birds to the La Cumbre population was 52% (95% CI 42–61%; N = 114), higher than that of wild-caught birds, which contributed with only 21% (95% CI 12–30%; N = 80). This was partly due to the fact that 18% of the wild-caught individuals returned to the area of capture (i.e. they were detected in Inagua during the following breeding season). Very similar results are obtained when repeating the analysis with birds released during the entirely overlapping period of 2015–2019; captive-bred: mean = 47%, 95% CI 36–59%, N = 59; wild-caught: mean = 21%, 95% CI 12–30%, N = 80.

Captive-bred birds released in La Cumbre during early autumn nested, on average, 0.82 km away the following spring in the same area (95% CI 0.70–0.96 km; n = 51, all well-documented cases). In contrast, wild-caught juveniles from Inagua, released in La Cumbre, settled an average of 2.12 km from the release site in the subsequent spring in La Cumbre (95% CI 1.46–2.86 km; n = 13). The non-overlapping confidence intervals between these two groups of birds, each with clear indications of nesting behaviour, is consistent with the previous results regarding larger home ranges for Inagua's wild birds compared to captive-bred counterparts.

Discussion

Our research provides several key insights into the survival, movements, and contribution of translocated birds belonging to an endangered species. These findings can have important implications for the design and implementation of translocation programs, particularly for philopatric endangered woodland species that inhabit islands. Captive-bred birds liberated with soft-release techniques made a higher contribution to the reinforced population compared to wild-caught birds with hard-release techniques. This is a pivotal insight for translocation programs, as it might suggest the advantages of releasing captive-bred individuals over wild-caught ones in certain contexts. Wild-caught birds displayed larger home ranges compared to captive-bred birds. However, the origin of translocated birds (captive-bred or wild-caught) did not influence the survival during the initial 42 days of radio-tracking. The post-release survival of the birds was most critical during the initial week. A noteworthy secondary finding is that radio transmitters did not impact the birds' long-term survival, with survival rates remaining nearly identical between birds equipped with transmitters and those without.

Hard-released wild juveniles from the Inagua source population occupied an area 12–15 times larger than captive-bred juveniles released by soft release. This variation in home range may stem from several factors. Individuals bred in captivity, having become accustomed to regular human interaction and consistent provision of food and water, may possess limited experience in resource-seeking behaviours. Life within the confined space of indoor aviaries, characterized by limited environmental variation such as the absence of predators, simple habitat structures, and a lack of thermal sun-shade mosaic, might hinder these birds' exploration abilities in the more complex habitat into which they are released, especially given their early-stage development in such simplified conditions. In contrast, wild-caught birds, raised in natural environments, exhibit distinct advantages over their captive counterparts. They had experienced wider movement ranges, developed defense mechanisms against predators, and demonstrated proficiency in foraging within structurally complex habitats and under more varied thermal conditions. In addition, they were not dependent on humans for a steady supply of food and water. When released as solitary individuals in an unknown new area with younger pine forests, they lack social connections in an unrecognized home range. Their inherent mobility, experienced during their early life phases, may lead to larger movements in comparison to captive-bred birds. Volpe et al. (2014) have shown that movement behaviour of translocated birds reflects natural behaviour of un-manipulated birds. Thus, the differences in movement and home range size between captive-bred and wild blue chaffinches surely reflect the fact that the birds' behaviour during translocation is similar to their previous movement behaviour. A pivotal consideration in translocation is the stress experienced by both captive-bred and wild-caught birds. Captive‐bred birds may be less sensitive to capture and transportation than wild-caught birds (Jenni et al. 2015). Such stress can be exacerbated in wild-caught birds due to their unfamiliarity with human handling, possibly perceived as predatory threats, with consequences on their mobility and energy budget (Carrascal and Polo 1999). Measures like soft releases, where birds are kept in large cages for several days before being released, have been proposed as mitigation strategies to allow recovery from such induced stress. In line with this, our observations suggest that wild-caught birds, while benefiting from a broader range of experiences in their natural environments, may still be at a disadvantage in hard-release, particularly as solitary individuals in unfamiliar territories.

Dispersal over wide areas is often correlated with high mortality (Payevsky 2016; Dale 2001). This might disproportionately affect translocated wild birds as compared to captive-bred ones, especially during extensive post-release movements or when returning to their capture sites in the subsequent months. However, our data indicate that the mortality rate during the first 42 days post-release was consistent between the two groups, despite their distinct home range sizes. Moreover, mortality fell sharply within 7 days of release, showing that this is the most critical period in blue chaffinch translocations (see also Letty et al. 2000; Teixeira et al. 2007; Dickens et al. 2010). A possible compensatory dynamic might be at play: the extensive experience of wild-caught birds with natural habitats aids them in exploiting larger home ranges, which could offset the restricted mobility observed in captive-bred birds due to their unfamiliarity with the wild.

The main cause of mortality in the weeks following release was from natural predators (raptors), with normal mortality levels for juvenile birds (Cox et al. 2014). Although the sample size is too small to draw conclusions, mortality from raptors and collisions with cars on the road were exclusively associated with wild-caught juveniles from Inagua. This finding is consistent with the greater mobility and dispersal of wild-caught juveniles released one day after capture. In two cases, there was clear evidence of predation by a non-native species, feral cats. The negative impact of feral cats on wildlife, particularly in island environments, is well documented and widely recognised as a major concern for endangered species (Smucker et al. 2000; Woods and McDonald 2003; Medina and Nogales 2009; Medina et al. 2011; Palmas et al. 2017; Mori et al. 2019). Therefore, since this is an exotic species that preys on and threatens other endemic species (e.g., Gran Canaria Giant Lizard Gallotia stehlini, we recommend the implementation of a feral cat trapping programme in the current and potential distribution areas of the blue chaffinch, as well as in areas where future releases will take place (e.g., pine forest of Tamadaba).

There exists a notable caveat in our observational design: the interplay between wild-caught birds from the source population and captive-bred birds has not been effectively crossed with the hard-release versus soft-release strategies for translocated birds. This design constraint stems from a bird release strategy aimed at minimizing stress for wild birds, which were released a day after capture. Moreover, there was a pronounced concern regarding the potential loss of captive-bred birds. These birds, representing a significant financial and labor investment, are inherently more naive because of their controlled rearing environment. Utilizing a hard-release approach without prior acclimation to the release zone could jeopardize the substantial resources invested in their breeding and maintenance. Nevertheless, the contribution of wild translocated birds to the population was notably less than that of captive-bred individuals. A significant reason for this was the return of 18% of the wild translocated birds to their original location, preventing them from establishing connections between populations at La Cumbre. Delgado et al. (2020) highlight that very few captive-bred birds moved to Inagua pine forests (< 2%). Of the wild translocated birds that survived and returned to their original pine forests in Inagua from La Cumbre, less than 5% were subsequently observed again in La Cumbre. As a result, they did not contribute to the new population nor foster connections between the two populations. For future actions, we recommend implementing measures to better establish wild-caught birds at release sites. Utilizing soft releases can aid in their recovery from the stress of capture, handling, and transportation, while simultaneously offering protection from predators the first week after release (Beck et al. 1994; Letty et al. 2000; Wanless et al. 2002; Teixeira et al. 2007; Dickens et al. 2010). Mitchell et al. (2011) demonstrated the improved release success with soft-releases in species with a high tendency for dispersal or those likely to experience significant predation pressure. Such an approach also promotes a sense of familiarity with the new landscape and habitat they can see through the cage walls (Stamps and Swaisgood 2007) and encourages the development of social bonds with other birds in the same setting that can enhance future foraging activity while reducing the risk of predation (Carrascal and Moreno 1992; Lima 1995; Beauchamp 1998; Webber et al. 2023).

In conclusion, our research indicates that well-designed translocation programs can bolster or stabilize small populations of species at-risk, at least in the short term. Translocated birds exhibited impressive post-release survival rates and successfully established at the release site, supporting the population growth. Captive-bred birds contributed more significantly than wild-caught birds, primarily because a notable portion of the wild-caught birds returned to their original capture sites. Thus, the reinforced programme for the Gran Canaria Blue Chaffinch has been successful during its establishment phase. We advocate for continuous population monitoring of birds translocated to La Cumbre, putting emphasis on soft-release approaches to ensure the long-term efficacy of the programme.