The importance of fine-scale landscape characteristics for habitat selection in penguins

Investigating the ecology of a species and the spatial requirements needed for their survival within an environment can improve and help develop conservation measures. In this study, we reviewed the literature describing the importance of fine-scale landscape characteristics on the distribution of penguin species. We then investigated little penguin nest-site use across eleven colonies in South Australia, with a focus on nest type, vegetation cover, nest entrance orientation, proximity to the nearest active nest, and side of the island. We showed that both abiotic and biotic variables were important for nest-site use in penguins and that the specific variables varied between species and populations. Little penguins in South Australia did not appear to use nest sites randomly, and active nests were mostly found on the northern side of the island and facing east or west. Our study highlights the importance of gaining a better understanding of penguin nest-site use, and their fitness consequences for populations, to ensure effective conservation outcomes.


Introduction
Understanding the spatial requirements of a species and the factors that influence their distribution can be fundamental to develop appropriate conservation and management measures (e.g., Guderyahn et al. 2016;Jaworski et al. 2019;Louzao et al. 2006). Seabirds are considered bio-indicators of the marine environment (e.g., Croxall et al. 2002;Lascelles et al. 2012) and the most threatened group of birds (reviewed in Dias et al. 2019). Due to their status and environmental importance, their conservation is essential. Yet most landscape research to date on seabirds have primarily focused on their requirements at sea (such as food availability) with limited focus on how their land requirements affect their Communicated by Stephen Garnett.
* Diane Colombelli-Négrel diane.colombelli-negrel@flinders.edu.au 1 3 distribution (but see for example Eveillard-Buchoux et al. 2019;Paredes and Zavalaga 2001;Seddon and Davis 1989). While seabird conservation depends on many factors (e.g., Dias et al. 2019;Trathan et al. 2015), the importance of land-based fine-scale landscape characteristics, such as nest sites, has been overlooked relative to marine habitat features. Fine-scale landscape characteristics (also referred to as small natural features) can have a disproportionate ecological influence on species by providing resources that otherwise limit population size or distribution, similar to keystone species in ecology (Hunter Jr 2017;Hunter et al. Jr 2017). Nest sites are particularly important for breeding success in seabirds (e.g., Colombelli-Négrel 2019; Frere et al. 1992;Gandini et al. 1999;Kim and Monaghan 2005b;Michielsen et al. 2019;Paredes and Zavalaga 2001). For example, nest sites with high vegetation cover can offer more protection from conspecifics, climate, and/or predators than less covered nests, thereby increasing breeding success (Bukacińska and Bukaciński 1993;Burger 1977;Frere et al. 1992;Gandini et al. 1999;Kim and Monaghan 2005a, b). Different nest types also offer differing microclimate, which in turn influence hatching success and chick survival (Bull 2000;Renner and Davis 2001) as well as thermoregulation (Colombelli-Négrel 2019), and thus can have long-term impacts on individual fitness. Therefore, identifying nest-site characteristics that may limit species distribution and/or promote higher breeding success should constitute a key objective in seabirds' conservation (Eveillard-Buchoux et al. 2019;Rönkä et al. 2008).
Little penguins (Eudyptula minor) are a seabird species endemic to the southern coast of Australia and parts of New Zealand (Marchant and Higgins 1990). South Australia host approx. 100 little penguin colonies, with 20+ colonies suspected to have declined in the last two decades for reasons not well understood , 2020DEWNR 2016;Wiebkin 2011). The Granite Island colony, for example, experienced one of the strongest declines as the population went from > 1500 adult little penguins in 2001 to < 30 adults since 2012 (Colombelli-Négrel 2020). Very little is known regarding the importance of fine-scale habitat and landscape characteristics on the distribution of penguins and their breeding habitat. For little penguins, Marker (2016) showed in Tasmania that the proximity of their foraging site to their breeding site was important for their distribution at the regional level. More locally, their breeding distribution was influenced by the potential for human disturbances (Marker 2016), as well as the distance between neighbouring nests (Tasmania; Marker 2016), the amount and type of vegetation near their nests (Western Australia; Clitheroe 2021), the aspect of the nests (Western Australia; Clitheroe 2021), and the proximity of landing sites (Victoria and Western Australia; Clitheroe 2021; Weerheim et al. 2003).
In this study, we first reviewed the literature describing the importance of fine-scale landscape characteristics on the distribution of penguin species. We then asked which habitat characteristics predicted nest-site use (active/inactive) in little penguins. Based on the literature, we specifically focused on nest type, orientation, and location, as well as vegetation cover and proximity to other active nests. To determine if we were correct in our hypotheses, we measured nest-site characteristics at active and inactive nests during population surveys of 11 penguin colonies on islands off the South Australia coastline. We used colonies that varied in population trends (see methods) to also determine whether differences in activity, rather than topography or vegetation, influenced nest-site use. We predicted (1) that active nests will be clustered together, and thus distance between neighbouring nests would be an important predictor for the presence of other active nests (Marker 2016), and (2) that active little penguin nests will be found sheltered from the elements, either facing away from the prevailing winds (north and south winds in South Australia) or 1 3 with high vegetation cover to reduce high temperatures inside the nest, because little penguins breeding success can be influenced by nest type and local environmental conditions, such as temperature and wind (Colombelli-Négrel 2019; Johnson and Colombelli-Négrel 2021).

Literature search
We searched for studies that specifically investigated whether fine-scale habitat and landscape characteristics explained penguin habitat use or for studies that mentioned such characteristics in their results. Using Google Scholar, we conducted a literature search of all studies published until 2021 with the following terms: 'penguin', 'distribution', 'nest', (including 'nest site') and 'habitat'. We focused on papers published in refereed journals, but also included book chapters, reports and thesis that were available online and relevant (see Fig. 1).

Study sites
Between May 2016 and October 2018, we surveyed 11 little penguin colonies in South Australia. GPS coordinates, years studied and population trends for each colony are presented in Table 1. In this study, we used colonies that varied in their population trends and were either: (i) well-studied (i.e., information regarding population sizes and population trends was available across several consecutive years or even decades; Troubridge Island, Granite Island, and Emu Bay on Kangaroo Island), (ii) little studied (i.e., information regarding population sizes was available for some years but was patchy; English Island, Hareby Island, Spilsby Island) or (iii) data deficient (i.e., no or little information regarding population sizes was available Wardang Island, Smith Island, Rabbit Island, Goose Island, Louth Island) (see also DEWNR 2016). The locations of the colonies are presented in Fig. 2.

Study species
Little penguins in Australia are known to breed both on the mainland and on offshore islands (Dann 1991;Marchant and Higgins 1990;Stevenson and Woehler 2007). During their breeding season, they become central-place foragers, often foraging within 20-30 km of their breeding site (Collins et al. 1999;Hoskins et al. 2008). Little penguins are monogamous and asynchronous breeders; breeding occurs between May and February , 2018Johannesen et al. 2003;Reilly and Cullen 1981), with each breeding pair producing up to two clutches per breeding season (Colombelli-Négrel 2015; Johannesen et al. 2003;Kemp and Dann 2001;Reilly and Cullen 1981). During incubation and the first two weeks of the 'chick-rearing' period, parents take turns sitting on the eggs or chicks (Miyazaki and Waas 2003). Incubation lasts 33-44 days and chicks fledge 8-9 weeks after hatching (Chiaradia and Nisbet 2006;Kemp and Dann 2001;Numata et al. 2004).

Population surveys
Population surveys were used to determine the location of active vs. inactive nests. We surveyed each colony during the second peak of the breeding season, either by inspecting the entire colony, or, in the case of large islands, by using transects or 30 × 30 m quadrats in areas where little penguins were present (see below). Such areas were determined using a combination of (1) random searches, (2) local knowledge from park rangers (and habitants for habited islands), (3) knowledge from previous surveys, and (4) helicopter surveys (when possible) to determine potential penguin habitat within an island. The peaks of the breeding season for each colony were determined using local knowledge from park rangers (and habitants for habited islands) and knowledge from previous surveys (see Table 1). Except on Troubridge, Goose and Rabbit islands, where nests were located all over the island, nests were only found along the coastline no 1 3 further than 120 m from the sea (see also Weerheim et al. 2003). Therefore, transects were positioned along the coast following the coastline, and all transects were marked with a Garmin GPS 62 or 64 s. Granite Island (2016-2018), English Island (2018) and Goose Island (2017-2018) were surveyed entirely each year. Troubridge Island (2016)(2017)(2018) and Emu Bay (2016-2017) were surveyed entirely in 2016 and using quadrats in 2017 (both colonies) and 2018 (Troubridge Island). Data extrapolated from the quadrat surveys on Troubridge Island were comparable to surveys conducted over the entire island , 2018. All the other colonies were surveyed using transects. All the islands in the lower Spencer Gulf (English Island, Hareby Island, Spilsby Island, Smith Island, Rabbit Island, and Louth Island) were surveyed in 2018.
Nests were searched within ~ 1 m of our transects on the land side and all the way to the water on the water side. The number and length of transects varied between colonies due to the size and shape of the islands and the vegetation present (number of transects ranging from 2 to 9 per colony, 3.5 on average; length of transects ranging from 400 m to 1 km, 700 m on average). For each colony, once a nest was found, we noted its status noted as active or inactive. Active nests were identified if they contained egg, chick or adult penguins or if they had clear evidence of recent (i.e., in the current year) penguin presence, such as fresh excrement, strong penguin smell, fresh digging or penguin feathers (Colombelli-Négrel 2015; Schumann et al. 2013). Deep nests were checked using a burrow-scope (Faunatech Austbat, Australia). A burrow was recorded as inactive if none of the above criteria was found or if it had cobwebs or overgrown vegetation at the Fig. 2 Distribution of the eleven islands surveyed for the presence of little penguin nests in South Australia. Islands marked with black triangles represent well-studied little penguin colonies, those with dark grey triangles represent the little-studied little penguin colonies, and those with light grey triangles are data deficient 1 3 entrance indicating that no large animal was or had been regularly entering/exiting the burrow. While little penguins are asynchronous breeders, they have two peaks during their breeding season in South Australia (Colombelli-Négrel, pers. obs.). As our surveys occurred during the second peak of breeding, all birds would have either bred in the first peak or started breeding in the second peak and, hence, we are confident that inactive nests remained inactive after our surveys. All nests were marked with a Garmin GPS 62 or 64 s.

Nest-site characteristics
While surveying each colony, the following nest-site characteristics were noted for each nest: (1) nest-site use (active or inactive as describe above), (2) nest type (categorized as either: surface (scrapes under an open bush), sand (nests dug in soft sand), bush (scrapes deep under a thick bush), rock (burrows under boulder or in rock crevices) or artificial (plastic boxes with rocks, metal drums or concrete structure) (Colombelli-Négrel 2019; see also Marker 2016), (3) vegetation cover (percentage of vegetation immediately above the nest (up to 0.5 m) estimated visually when standing 1 m away; 0-100% following Colombelli-Négrel 2019); (4) nest entrance orientation [recorded as a score between 1 (north facing) and − 1 (south facing)]; (5) proximity to the nearest active nest (in meters), and (6) location within the colony/side of the island (recorded in degrees as a value between 1 and 360 degrees in relation to north = 360 degrees; i.e., a north-east side would be recorded as 45 degrees and a southern side would be recorded as 180 degrees). Nest-site characteristics were collected once per nest site. To ensure independence of the data, when colonies were visited more than once, we only collected nest-site characteristics during the last survey.

Statistical analysis
All statistical analyses were performed in SPSS version 25.0 for Windows (SPSS Inc., Chicago, IL, USA). Prior to analysis, we assessed collinearity between the predictor variables using the variance inflation factors (VIF) analysis; all VIF values were well < 2, confirming no collinearity (Zuur et al. 2009). We then modelled penguin nest-site use (active vs. inactive) using a General Linear Model with a binary distribution and six fixed factors (colony, nest type, vegetation cover, nest entrance orientation, proximity to the nearest active nest, and side of the island). We investigated differences in nest-site characteristics between colonies using MANOVAs and Bonferroni post hoc pairwise comparisons.

Literature review
The type (biotic and abiotic) and number (between one and nine) of habitat variables measured varied between studies ( Fig. 1; Table 2). Most studies focused on habitat use at the nest site (i.e., presence of active nests; 71% − 27 studies out of 38), while 8% (3/38) focused on habitat use at the scale of general breeding habitat (i.e., colony size) and 21% (8/38) focused on breeding success; only one study (3%) focused on both presence of active nests and colony size and four studies (11%) investigating habitat use at the nest site also examined breeding success and its association with habitat characteristics ( Fig. 1;   Trathan et al. (1996)          sites and in open grasslands. Ratz and Murphy (1999)    associated with headlands or adjoining beaches (Dann 1994) Table 2). Only 60% of the studies (23 out of 38 studies) statistically evaluated the influence of habitat characteristics on habitat use in penguins and only two studies (5%) compared nest-site characteristics between active and inactive nests ( Table 2). The remaining 40% simply described penguin habitat use or mentioned nest-site characteristics in their results but did not statistically evaluate their importance for habitat or nest-site use.
Habitat variables influencing nesting site use or colony size varied between studies, regions, and species ( Fig. 3; Table 2). Penguins in Antarctica were commonly reported nesting on snow-free or ice-free surfaces as well as stable sea ice platforms, up to several kilometres from the sea. Penguins in South America and Southern Africa nested in sheltered areas, often under shrubs or bushes, up to 1 km from the sea. Penguins in Australasia nested in rock crevices, burrows (man-made or artificial) or under thick vegetation, less than 500 m from the sea. In addition, habitat variables important for nest use and breeding success differed between cavity vs. surface nesting species. For surface nesting species, abiotic and topographical variables as well as the proximity of conspecifics were important factors determining nest use, while both abiotic (slope, aspect, location within the colony) and biotic (vegetation cover) variables were important for breeding success ( Fig. 3; Table 2). For burrow nesting species, both abiotic (location within the colony, distance to shore and neighbours) and biotic (vegetation cover, habitat type) variables were important for both nest use and breeding success ( Fig. 3; Table 2).

Little penguin population surveys and nest distribution
A total of 394 nests were found across nine islands, of which 238 were active. For 78% of the surveyed islands (excluding English and Smith Islands where no penguins were Fig. 3 Habitat variables found to be important for (a) nestsite selection and (b) breeding success in 39 studies across 16 species of penguins. The data are presented for surface nesting species (black) and burrow nesting species (grey) 1 3 present), active nests were found at higher densities on the northern side of the island. For the remaining 22%, nests were found towards the southern side of the island. The best predictors for the presence of active little penguin nests were the side of the island and nest entrance orientation (GLM; Table 3): active nests were mostly found on the northern side of the island, including northeast and northwest sides, and facing east or west. Hareby Island significantly had more inactive nests than the other colonies (GLM; Table 3). All colonies significantly differed in their nest-site characteristics: nest type (MANOVA: F 8, 393 = 41.10; P < 0.0001), vegetation cover (F 8, 393 = 14.87; P < 0.0001), nest entrance orientation (F 8, 393 = 9.70; P < 0.0001), proximity to the nearest active nest (F 8, 393 = 8.20; P < 0.0001), and side of the island (F 8, 393 = 70.17; P < 0.0001; Table S1). We present in Table S2 our sample sizes and a summary of the nest-site characteristics measured in this study in relation to active and inactive nests.

Discussion
Nest-site in birds is critical for courtship, pairing, and reproductive success (e.g., Colombelli-Négrel and Kleindorfer 2009;Mainwaring et al. 2014;Stokes and Boersma 1998). Several factors may influence nest-site selection and use, especially those that offer protection from predators and environmental extremes and thus increase breeding success and survival of nesting birds Colombelli-Négrel and Kleindorfer 2009;Frere et al. 1992;Michielsen et al. 2019;Stokes and Boersma 1998). Our literature review demonstrated that both abiotic and biotic variables were important for nest-site use in penguins, but that the specific variables varied between studies, regions, and species 1 3 (Table 2), maybe due to adaption to local environment. Our study on South Australian little penguins further supports these finding and showed that little penguins in this region did not appear to use nest sites randomly, but instead likely based their nest-site use on abiotic factors such as the side of the island and the orientation of the nest entrance. Our study highlights that a clearer understanding of nest-site use in penguins is required to better manage their breeding habitat (see also Eveillard-Buchoux et al. 2019;Hunter Jr et al. 2017). Similar to other studies in penguins (Bried and Jouventin 2001;Challies and Burleigh 2004;Long and Litchwark 2017;Morandini et al. 2021;Ratz and Murphy 1999), the location of a nest within a colony was an important factor for nest activity in South Australian little penguins. However, in our study such location was not defined as centre vs. edge, as found in most studies, but rather as a northerly side. Indeed, we found that active little penguin nests were more likely to be found on the northern side of the island. A study on Tasmanian little penguins also found that active nests were more likely to be found on north-facing slopes (Marker 2016), suggesting that a northerly side may be important for the establishment of little penguin nests within a colony/island. This may be due to the potential impact that the winter prevailing winds (when little penguins start breeding in South Australia) could have on the nest micro-climate and/or on the coastal waters, which remains to be tested.
Additionally, we found that active little penguin nests were more likely to face an eastern or western aspect, similar to what was observed in Western Australia by Klomp et al. (1991) and Clitheroe (2021). Again, this may be due to the prevailing winds, and how winds influence nest microclimate (i.e., temperature and humidity) (Ropert-Coudert et al. 2004), which would suggest that nest microclimate (and not orientation per se) may be driving nest use in little penguins. Numerous studies in birds have highlighted the importance of the nest microclimate for habitat use (e.g., Colombelli-Négrel 2019; Lei et al. 2014;Rhodes et al. 2009), even sometimes suggesting that it may play a larger role than vegetation or topography. In penguins, nest temperature is important for thermoregulatory behaviours and chick development (Frere et al. 1992;Frost et al. 1976;Lei et al. 2014), while nest humidity is important for egg shell integrity and hatching success (Colombelli-Négrel 2019). Additional studies separating the effects of microclimate, winds, and orientation on nest use may help answer this question.
Surprisingly, and contrary to other studies on little penguins (Clitheroe 2021;Dann 1994;Marker 2016), vegetation cover was not an important factor explaining the presence of active nests in our study. Vegetation cover (either adjacent to the nest or directly above) is an important variable for both surface and burrow nesting penguin species (Fig. 3; Table 2), likely to minimise predation risk (Frere et al. 1992;Stokes and Boersma 1998) and/or exposure to extremes environments (Kim and Monaghan 2005a). It should be noted however, that most colonies in this study did not have terrestrial predators, which may have impacted our results. The impact of vegetation may also have been mitigated by interaction with other variables (not measured in this study), such as substrate. Indeed, substrate composition can influence vegetation growth and structure (see Borboroglu et al. 2002) as well as help maintain nests' microclimate and stability (Stokes and Boersma 1991). It should also be noted that most previous studies investigated the importance of vegetation cover in relation to the presence vs. absence of penguin nests (Clitheroe 2021;Dann 1994;Marker 2016) rather than nest-site use (this study), which may also explain the observed difference in results.
Social interactions or the close presence of conspecifics may also influence selection/use of nest-sites in penguins (Ainley et al. 1995;Marker 2016;Santora et al. 2020;Volkman and Trivelpiece 1981). However, in South Australian little penguins, we found no correlation between the presence of active nests and the distance to the next active nest. Similarly, we found that only Hareby Island significantly differed in the percentage of active vs. inactive nests from the other colonies. It is possible that little penguins experience reduced breeding success when nesting in high density -as found in African (Sherley et al. 2014) and Magellanic (Stokes and Boersma 2000) penguins (see also Colombelli-Négrel 2015, in little penguins) -due to competition for food, and thus could be avoiding nesting too close to conspecifics. Alternatively, it may simply be because little penguins are highly philopatric, returning to the same part of their colony (and often re-using the same nest) year after year (Reilly and Cullen 1981), regardless of the presence of conspecifics. This idea is supported by our nest survey on Rabbit Island, where a single nest was found in the middle of the island.
Nest type has been shown to be important for breeding success in several penguin species (Frere et al. 1992;Paredes and Zavalaga 2001;Seddon and Van Heezik 1991;Sherley et al. 2012), likely due to its impact on the thermodynamic characteristics of the nest (Colombelli-Négrel 2019; Frost et al. 1976;Lei et al. 2014;Seddon and Davis 1989). However, in little penguins, the importance of nest type for breeding success varied between studies, with some studies finding a positive influence (Bull 2000;Renner and Davis 2001) and others not (Boyer 2010;Braidwood et al. 2011;Geurts 2006). In this study, we found that nest type did not influence the likelihood of finding an active nest vs. an inactive nest. Nest type use may be influenced by other factors, such as proximity to prey supply, predator presence and/or frequency of extreme weather events. The most advantageous combination of landscape characteristics, frequency and intensity of adverse factors, and beneficial factors that allows for the survival of individuals and population, is thus expected be unique for different species and populations (see also Table 2). This reinforces the importance of sitespecific investigations for each species to ensure that more effective conservation measure are put in place.
To assess the fitness benefits of habitat selection, studies should also demonstrate increased fitness in used habitats (Jones 2001;reviewed in Chalfoun and Schmidt 2012). Indeed, breeding success in penguins can be influenced by both abiotic and biotic variables, such as slope, aspect, location within the colony, vegetation cover or habitat type (Bried and Jouventin 2001;McKay et al. 1999;Morandini et al. 2021;Schmidt et al. 2021;Stokes and Boersma 1998;Trivelpiece and Fraser 1996). Yet very few studies in our review investigated nest attributes for both nest-use and breeding success ( Fig. 1; Table 2). In addition, the variables that correlated with nest-site use sometimes differed to those that correlated with improved breeding success (Fig. 3). Studies in other seabirds have suggested that habitat characteristics may be less important for breeding success than other factors, such as parental condition or experience (e.g., Pugesek and Diem 1983;Velando and Freire 2001), which may explain these contradictory results. Additionally, most studies assessing nest-site use in animals simply assume that nests or sites are voluntarily chosen without considering that they may be in fact selected randomly (see Goodenough et al. 2009). Additional studies experimentally testing nest selection in penguins are clearly needed.

Conclusion
Climate change is predicted to reduce the suitability or availability of breeding habitats for penguin species by altering the coastal environment (i.e., by melting nesting ice-surfaces or increasing sea-levels and thereby reducing sizes of low-lying islands 1 3 for example; see Chambers et al. 2011;Dann and Chambers 2009;Schumann et al. 2013),. While some species may be able to adjust their distribution, the persistence of other species or populations may depend on well-planned habitat management such as timely prioritisation of conservation measures for species that will be most affected by a reduction in the suitability or availability of their breeding habitats (see Grémillet and Boulinier 2009). Our study and literature review emphasise that factors important for nest-site use in penguins varied between species and even populations; thereby highlighting the importance of gaining a better understanding of penguin habitat use, and their fitness consequences for populations, to ensure effective habitat management in the light of climate change (see also Eveillard-Buchoux et al. 2019;Hunter Jr et al. 2017).