Assessing the effectiveness of rehabilitation for management of an endangered seabird, the Yellow-eyed Penguin

ABSTRACT Rehabilitation of wildlife can be a meaningful conservation technique if rehabilitated animals contribute to the breeding population. Endangered Yellow-eyed Penguins (Megadyptes antipodes) are declining on South Island, New Zealand, where modelling published in 2017 predicted their extirpation by 2043. Four management plans dating back to 1989 have been implemented in attempts to mitigate threats. The first three plans overlooked rehabilitation whereas the most recent, published in 2020, regarded rehabilitation as essential to save the South Island population. We assess the outcome of four decades of management of Yellow-eyed Penguins at Moeraki, southeast South Island, by Penguin Rescue, a volunteer conservation organisation. Here, nest numbers have fluctuated but overall increased at a long-term annual average of 5%. Their proportion of the southeast South Island total rose from 1% (six of about 453 nests) in 1982 to 26% (43 of about 166 nests) in 2021. Since 1986 our management has included rehabilitation of all juvenile or adult Yellow-eyed Penguins we encountered locally with life-threatening injuries, emaciation or sickness, with 590 of these marked before release from our rehabilitation facility. We accounted for the effect of rehabilitation on nest numbers by subtracting the number of rehabilitated female breeders and their female descendants from the total number of female breeders. Without rehabilitation nest numbers at Moeraki in 2021 probably would have remained similar to the initial six nests in 1982 instead of the seven-fold increase through four decades. We conclude that rehabilitation is an effective management technique for this species.


Introduction
Rehabilitation of compromised animals is a unique tool in wildlife management that targets individuals instead of populations and has become a standard response to the public expectation for action when wildlife populations are threatened following man-made disasters, especially oil spills (e.g.Williams 1990;Henkel and Ziccardi 2018).However, we believe that rehabilitation can only be justified as a meaningful conservation technique if rehabilitated animals contribute to the breeding population: otherwise, rehabilitation is primarily an emotive and unscientific exercise (e.g.Smith 1996;Ratz and Lalas 2010;Henkel and Ziccardi 2018).A review by Pyke and Szabo (2017) concluded there was little evidence to indicate that rehabilitation directly contributed to conservation of wildlife.However, rehabilitation has had a substantial positive effect on at least one seabird species: the African Penguin (Spheniscus demersus) population in 2002 was about 20% larger than it would have been without any rehabilitation of oiled birds (Ryan 2003).We follow the concept that most if not all successfully rehabilitated individuals would have died without intervention (Ress and Guyer 2004) and so each potentially represents an addition to the population.
Here we consider the history and efficacy of rehabilitation as a key technique in the recovery of an endangered species of seabird, the New Zealand endemic Yellow-eyed Penguin (Hoiho, Megadyptes antipodes), on South Island, the northern limit of its breeding distribution (Figure 1).Here the species faces threats from a combination of anthropogenic sources and extreme fluctuations in population size (BirdLife International 2000, 2022).Four decades of continual monitoring along the southeast coast of South Island showed there were about 465 nests in 1980, a peak of about 635 nests in 1996 (Seddon et al. 2013), and then a decline to 227 nests in 2018 (BirdLife International 2022).In 2000, the conservation status of Yellow-eyed Penguins in the IUCN Red List of Threatened Species was upgraded from Vulnerable to Endangered (BirdLife International 2022) in response to ongoing decreases attributed primarily to terrestrial habitat degradation; predation by introduced mammals; by-catch mortality in fisheries; die-off events and diseases; human disturbance; and climatic perturbations and climate change.Despite ongoing conservation measures, modelling of the southeast South Island population by Mattern et al. (2017) predicted extirpation by 2043.
The appropriateness of rehabilitation as a conservation tool for Yellow-eyed Penguins has had a chequered career through the sequence of four management plans for this species.(Henceforth 'rehabilitation' refers only to rehabilitation of individuals: management plans also applied the term 'rehabilitation' to restoration of terrestrial habitat.)The first plan did not mention rehabilitation (Department of Conservation 1989).The second plan mentioned rehabilitation only in an appendix entitled 'Penguin Ranching' that presented an alternative management concept contrary to the prescribed strategy (Department of Conservation 1991: 51; this page is reproduced in Figure S1).The third plan restricted mention of rehabilitation to 'The treatment and rehabilitation of injured birds is recognised as a means of sustaining public interest and support' (McKinlay 2001: 20).In contrast, the current fourth plan promotes rehabilitation as a key for the recovery of the species: 'Hoiho are becoming increasingly stressed and are in need of rehabilitation and direct treatment to reduce mortalities from illness, injury and starvation' (Te Rūnanga o Ngāi Tahu et al. 2020a: 18).Rehabilitation is now designated as 'Strategic Priority 1' and 'essential' in the 'Strategic direction' for saving the South Island population (Te Rūnanga o Ngāi Tahu et al. 2020b: 1-2).
Three scientific publications have addressed the outcome of rehabilitation of Yellow-eyed Penguins.Busch and Cullen (2009) modelled the efficacy and costeffectiveness of different management methods by comparing trends in nest numbers following three designated treatments: trapping (killing introduced mammalian predators that eat chicks), revegetation (creation of nesting habitat by revegetation and/or nest boxes) and intensive management (regular monitoring and rehabilitation in addition to trapping and revegetation).Trapping and revegetation (excluding nest boxes) were the key techniques prescribed for pastoral land by the contemporary management plan (McKinlay 2001) and intensive management matched penguin ranching, the alternative management concept proposed in the second management plan (Department of Conservation 1991; Figure S1).Modelling the application of trapping and revegetation either alone or in combination resulted in population decreases, whereas modelling for intensive management generated a 5% average annual increase in nest numbers (Busch and Cullen 2009).In the second assessment of the outcome for rehabilitation of Yellow-eyed Penguins, Ratz and Lalas (2010) found that rehabilitation increased average annual survival of female breeders by 9%.In contrast to the positive outcomes for rehabilitation from the first two assessments, Alden et al. (2021) found that survival after fledging was lower for rehabilitated chicks than for unrehabilitated chicks.
Adopting rehabilitation as the priority in Yelloweyed Penguin conservation management (Te Rūnanga o Ngāi Tahu et al. 2020a, 2020b) occurred without a comprehensive assessment of the impact of rehabilitation as a technique to increase population size.Intensive management (including rehabilitation) of Yellow-eyed Penguins occurred only at two locations assessed by Busch and Cullen (2009): one on Otago Peninsula undertaken by Penguin Place, a commercial ecotourism venture; and the other further north at Moeraki, North Otago, undertaken by Penguin Rescue, a volunteer organisation (Figure 1).Penguin Place was the source of data for Ratz and Lalas (2010) and Alden et al. (2021).Unfortunately, this population has been decimated through predation by New Zealand Sea Lions (Phocarctos hookeri) (Lalas et al. 2007;Augé et al. 2012) and is no longer a suitable location to evaluate the effect of post-fledging rehabilitation of Yellow-eyed Penguins.
Here we evaluate the effect of rehabilitation by quantifying Yellow-eyed Penguin nest numbers attributable to rehabilitated females and their female descendants through four decades at Moeraki.We are unable to prove that most if not all individuals rehabilitated due to life-threatening injuries, emaciation or sickness would have died without intervention.Instead, we begin with a comparison of trends in Yellow-eyed Penguin nest numbers for Moeraki with the totals for elsewhere on southeast South Island.This will indicate whether or not intensive management at Moeraki coincides with a different outcome to other forms of management.We then model trends in nest numbers at Moeraki with and without rehabilitation, under the assumption that all rehabilitated individuals would have died without intervention.

Management at Moeraki by Penguin Rescue
Monitoring and management of Yellow-eyed Penguins breeding on grazed farmland at Moeraki, North Otago (Figure 1), was begun in 1982 by Janice and Bob Jones (Gillies 2017).Chris Lalas began participation in 1983 while employed for 18 months by the Department of Lands and Survey (a predecessor of the New Zealand Department of Conservation) to survey seabirds on southeast South Island.Rosalie Goldsworthy replaced Janice at Katiki Point in 2002 and subsequently has led management.Hiltrun Ratz joined Penguin Rescue in 2013.
Initially, we intended to follow the strategy proposed by the first Yellow-eyed Penguin management plan (Department of Conservation 1989).Here, the key action for pasture farmland was the creation and protection of nesting habitat by removing stock and planting to create forest.The strategy envisaged that grass growth following stock removal would eliminate loss of chicks to introduced mammalian predators by creating an impenetrable vegetative barrier.We found that this concept did not work, and we implemented trapping to control predators.We gradually further diverged from the prescribed strategy in two ways, firstly with a change in the perception of ideal nesting habitat from old-growth forest to vegetation ecotones (Ratz and Murphy 1999) and grazed shrubland (McKay et al. 1999), and secondly with the implementation of techniques to enhance breeding success and survival, primarily through the provision of nest boxes (Lalas et al. 1999) and rehabilitation of compromised individuals (Ratz and Lalas 2010).The concept of our diversion from the prescribed strategy was added as an appendix to the 1991 species management plan (Department of Conservation 1991; Figure S1).
We formally registered as a non-government organisation in 2010 as Katiki Point Penguin Refuge Charitable Trust (New Zealand Charities Services registration number CC44085) and in 2014 we adopted for the Trust the simpler alias Penguin Rescue (https://www.penguinrescue.nz).Penguin Rescue is financed through donations and operated by volunteers.Our annual expenditure in 2021/2022 was NZ$37,500 with volunteers contributing the equivalent of three full-time positions.

Trends in nest numbers
The breeding season of Yellow-eyed Penguin in the southeast South Island extends for 6 months from the start of egg laying in September to the end of chick fledging the following March (Seddon et al. 2013).
Here, we allocate nest numbers to the calendar year of egg laying.Breeding in southeast South Island is divided into three regions: The Catlins, Otago Peninsula and North Otago.Nest numbers in each region were depicted for 1996-2016 by MacLean (2017).Moeraki is one of seven breeding locations in North Otago and is situated midway along the North Otago coast (Figure 1).Elsewhere on the South Island breeding occurs only on Banks Peninsula with a few nests (Seddon et al. 2013).
We compared counts for annual nest numbers we recorded through 40 years  of conservation management of Yellow-eyed Penguins at Moeraki by Penguin Rescue with the numbers for the southeast South Island excluding Moeraki.Annual numbers of Yellow-eyed Penguin nests at southeast South Island are collated by the New Zealand Department of Conservation.These data are considered estimates because they include numbers for locations not surveyed annually and numbers for some other locations are derived from imprecise or incomplete counts (Seddon et al. 2013;Mattern et al. 2017).In particular, some nests are missed and most annual numbers per location are derived from single counts with an average detection rate of 88% compared to double counts (Hegg et al. 2012).
Estimates for annual numbers of Yellow-eyed Penguin nests at southeast South Island date back to 1980 (Seddon et al. 2013).These data are now updated annually by the Department of Conservation which has declared a copyright on behalf of the contributors (including us, Penguin Rescue).We used published versions of these data, with data for 1980-2009 from Seddon et al. (2013);1996-2016from MacLean (2017)); 2017 from Lewis (2018); 2018 from Edwards (2019); 2019 and 2020 from MacLean (2020); and 2021 from White (2021).Annual data for nest numbers are presented in Table S1.

Marked and unmarked individuals
The natal location of penguins is known only for those marked as chicks and age is known for those marked as chicks or juveniles but not as adults.Since 1983 most Yellow-eyed Penguin chicks and older unmarked individuals encountered at Moeraki have been marked either with a stainless-steel flipper band  or an injected passive integrated transponder tag (since 2008).To document the fate of emigrants from Moeraki that were recorded nesting elsewhere we used the Department of Conservation Yellow-eyed penguin database (Seddon et al. 2013;Mattern et al. 2017) that hold data updated to the 2019 season.
Yellow-eyed Penguins encountered in the wild at Moeraki included unmarked individuals.An idealised marked population has all breeders and their descendants identified.Consequently, the only unmarked breeders encountered are immigrants from an unmarked population.Unfortunately, nothing was this simple at Moeraki.Some chicks fledged unmarked (Table S2) and so could not be differentiated from unmarked immigrants if encountered later.Similarly, some individuals were released unmarked after rehabilitation and so remained unrecognisable.To further complicate matters, we occasionally made unresolved errors in records of band or transponder numbers that generated anomalous data.We accounted for these cumulative imprecisions by presenting the outcomes of rehabilitation as ranges instead of definite values.

Quantifying the effects of rehabilitation
Our analyses encompassed Yellow-eyed Penguins admitted to our Penguin Rescue rehabilitation facility at Moeraki.We released rehabilitated individuals at abutting breeding habitat along the southern half of a 5 kmlong peninsula at Moeraki.Most admissions were birds we found during our monitoring of the North Otago coast (Figure 1), with others found by locals or brought by Department of Conservation staff.We restricted analyses to individuals that we assessed would have likely died if not rehabilitated.We designated these individuals as compromised either by life-threatening injury, or by life-threatening emaciation or sickness.Here, we do not give any details of methods involved with symptoms, treatment, convalescence or release of rehabilitated birds.Instead, we follow the fate of rehabilitated females after their release and quantify their contribution to the number of females breeding at Moeraki.
We restricted analyses to rehabilitations from two age classes depicted in Seddon et al. (2013): either juveniles (youngsters from fledging up to midway through their first annual moult at age 1.5 y) or adults (older birds).Our analyses excluded chicks because we could not assume that all would have died without rehabilitation.In particular, all chicks diagnosed with diphtheritic stomatitis (e.g.Saunderson et al. 2021) were treated regardless of the severity of their symptoms.Also excluded here are 40 juveniles or adults that were relocated from sites regularly frequented by people and dogs.These individuals were not in need of rehabilitation but were at risk of death if not removed, particularly from dog attacks.Another two individuals were excluded from analysis because they were admitted with injuries that were not life-threatening, with both released 2 days after admission.
Analyses of annual numbers rehabilitated begin with 1986, the first year that we treated all compromised individuals encountered locally.Our only rehabilitations prior to 1986 were three marked individuals, one in 1984 and two in 1985, none of which subsequently bred at Moeraki.Some individuals were rehabilitated more than once within one year and/or in multiple years, and so the sum of sample sizes from the 36 years 1986-2021 was greater than the respective total number of rehabilitated individuals.We assessed records in cases where one individual was admitted more than once in order to avoid duplication in the form of multiple admissions for the same reason.
We found that we could not reliably sex pre-breeders but within breeding pairs males had larger heads than their female partner (Setiawan et al. 2004).We largely restricted our analyses to the fate of females rather than males because breeding populations contain a surplus of males and the number of nests is controlled by the number of females of breeding age; i.e. females aged ≥ 2 y (Ratz and Lalas 2010).
We designated female breeders either as 'affected by rehabilitation' or as 'unaffected by rehabilitation' in order to assess the effect on nest numbers at Moeraki if rehabilitation had not been applied.'Affected by rehabilitation' means that a female was rehabilitated as a juvenile or adult (so would otherwise be dead) or she was derived from a lineage that post-dates rehabilitation in a female ancestor (so would not have existed without rehabilitation of the ancestor).'Unaffected by rehabilitation' means that a female had not been rehabilitated as a juvenile or adult and was derived from a lineage that precedes any rehabilitation events in all female ancestors.In the case of 'unaffected' individuals being rehabilitated, their status swapped to 'affected' after their rehabilitation.
Lifetime reproductive success, here the number of chicks fledged in the lifetime of a female breeder, is a parameter that combines survival and fecundity into one value (Newton 1989).We calculated lifetime reproductive success of known-age female breeders at Moeraki that recruited and died during the study period, 1982-2021.We compared lifetime reproductive success among female breeders using Mann-Whitney U tests following Zar (1999).

Analyses of temporal trends in numbers
We estimated temporal trends with exponential regressions in the form N t ∝ e λt , where N t = number in year t, λ = average exponential annual rate of change, and average arithmetic rate of change μ = e λ − 1.These analyses were either stochastic (derived from all annual data points) or deterministic (derived from only the first and last data points).Graphs depicting stochastic analyses show lines of best fit bounded by 95% confidence limits (the likely spread of the line of best fit) and 95% prediction limits (the likely spread of data).Exponential regressions and statistical tests followed Zar (1999) and we generated 3-parameter growth curves in Curve Expert Professional 2.7.3.

Trends in nest numbers
We began conservation management of Yellow-eyed Penguins with six nests in 1982 (Figure 2 S3. For the same period, nest numbers for southeast South Island excluding Moeraki began with 447 nests in 1982 (Table S1).Nest numbers subsequently fluctuated more wildly than at Moeraki and with moredistinct peaks and troughs.The maximum recorded was 619 nests in 1996 and the minimum was 123 nests in 2021.Overall, nest numbers averaged a 2% stochastic annual decrease (Figure S2A).In common with the same relationship for Moeraki, this rate presently appears unduly optimistic because recent years have been well below the line of best fit.Consequently, perhaps a more realistic representation is the 3% average annual decrease generated by deterministic growth rate from 1982 to 2021 (Figure 2(b)).Moeraki accounted for 1% (6/453) of Yellow-eyed Penguin nests at southeast South Island in 1982, rising to 26% (43/166) in 2021.The consequence of the long-term overall increase in nests at Moeraki and decrease in nests elsewhere has been an average 7% annual increase contributed by Moeraki towards the southeast South Island total number of Yellow-eyed Penguin nests, a proportion unaltered by recent decreases at Moeraki (Figure S2B).

Trends in number of individuals rehabilitated
We began systematic rehabilitation of injured, emaciated or sick juvenile or adult Yellow-eyed Penguins at the start of the 1986 calendar year, with a total 975 admissions through 36 years, 1986-2021.Most admissions (705 = 72%) were collected locally at Moeraki (along the 5 km of shoreline from Moeraki village to Katiki Point), with practically all the others (245 = 25%) collected elsewhere within North Otago.The remainder (25 = 3%) consisted of 13 collected further north, 11 collected further south and one with collection site unrecorded.
The annual number of admissions to our rehabilitation facility fluctuated wildly but overall increased at a stochastic average about 5% (Figure 3(a): n = 36 y, λ = 0.052, μ = 0.053, r 2 = 0.577, P < 0.001), a rate similar to the average annual increase in local nest numbers (n = 36 y, λ = 0.055, μ = 0.057, r 2 = 0.647, P < 0.001).The relationship between these two chronological trends indicated that the annual number of admissions to our rehabilitation facility (y) averaged the equivalent of 86% of the number of nests in the same year (x) (Figure 3(b): n = 36 y, y = 0.857x, r 2 = 0.821, P < 0.001).
Most importantly, these data show that the occurrence of injured, emaciated or sick Yellow-eyed Penguins is not solely a recent event.Instead, we recorded a major occurrence over 30 years ago when numbers admitted matched or exceeded nest numbers at Moeraki through our first 5 years of comprehensive rehabilitation, 1986-1990 (Figure 3(a) (Figure 3(a); Table S1 for nest numbers and Table S3 and Table S4 for numbers admitted).Also, the proportion of the total 975 admissions attributed to emaciation or sickness (593 = 61%) greatly exceeded the proportion of admissions attributed to injury (382 = 39%; comparison of deviation of proportions from 0.5: Z (2) = 9.511, P < 0.001).

Fates of rehabilitated marked individuals
Nine hundred and seven (93%) of the 975 admissions were released from our rehabilitation facility.Marked individuals accounted for 734 (81%) of these 907 releases.These marked individuals were identifiable if encountered again.The other 173 (19%) releases were unidentifiable if encountered again and consisted of 153 unmarked and another 20 anomalous, with recorded band or transponder numbers either mis-recorded or undocumented in databases (Table S5).
After accounting for 91 penguins rehabilitated in more than one calendar year a total of 591 individuals were released marked from our rehabilitation facility from 1986 to 2021.A total of 167 (28%) of the 591 rehabilitated individuals released marked were recorded as breeders after release (Table 1).Eighty-nine were female of which 77 remained to breed at Moeraki and the other 13 were recorded breeding elsewhere; and 78 were male, of which 71 remained to breed at Moeraki and the other seven were recorded breeding elsewhere (Table 1).
Average lifetime reproductive success appeared unaffected by rehabilitation as juveniles or adults among the 112 known-age female breeders at Moeraki that recruited and died during the study period, 1982-2021.The difference between the 38 rehabilitated (mean = 7.13 chicks fledged, SD = 6.90) and 74 not rehabilitated (mean = 6.24 chicks fledged, SD = 6.68) was not statistically significant (Mann-Whitney U test: U = 1264, P > 0.05).

Rehabilitation status of female breeders
We divided the 240 Yellow-eyed Penguin female breeders we recorded nesting at Moeraki through 40 years, 1982-2021, into nine categories in order to deduce whether or not they were affected by rehabilitation.Sixty-nine (29%) were confirmed unaffected because these individuals and all their female ancestors were confirmed not rehabilitated.Ninety-four (39%) were confirmed affected, divided into the 79 (33%) individuals confirmed rehabilitated (Table 1) and another 15 (6%) confirmed not rehabilitated but with female ancestor(s) confirmed rehabilitated.The rehabilitation status of the remaining 77 (32%) individuals was unknown because either individuals or any of their female ancestors were unmarked when first encountered at Moeraki.

Effect of rehabilitation on nest numbers
Beginning in 2014, we have marked every breeder at Moeraki found previously unmarked, with the most recent unmarked female breeder encountered in 2017.Having all breeders marked combined with ongoing death or rehabilitation of previously unmarked breeders has had a profound impact on the precision of outcomes.Of the 43 female breeders in 2021, 37 (86%) had been rehabilitated, and an additional two had not been rehabilitated but had a confirmed rehabilitated ancestor, totalling 39 (91%) confirmed affected by rehabilitation.
Of the remaining four individuals, three (7%) were unaffected by rehabilitation and the status of only one (2%) individual remained unknown.Consequently, without rehabilitation, there may have been only three or four nests at Moeraki instead of the actual 43 nests in 2021 (Figure 4).This estimated number of nests if rehabilitation had not been undertaken is less than the number of nests through the early years of our conservation management four decades earlier (Figure 4).

Discussion
We have demonstrated that rehabilitation is an effective method to increase numbers of Yellow-eyed Penguins.Our evidence came from two independent sets of data through four decades.First, there were opposite longterm trends in nest numbers at southeast South Island with and without rehabilitation: nest numbers increased at Moeraki in contrast to an overall decrease elsewhere.Nest numbers at Moeraki through the duration of conservation management by Penguin Rescue have increased as a proportion of the southeast South Island total from 1% in 1982 to 26% in 2021.Second, the increase in nest numbers at Moeraki is largely attributable to rehabilitation.Without rehabilitation of injured, emaciated or sick juvenile and adult Yellow-eyed Penguins, Moeraki probably would have had about 3-4 nests in 2021, lower than the initial numbers in the early 1980s.Instead, in 2021 there were 43 nests, the culmination of a deterministic average 5% annual increase through the 40 years since we began conservation management of Yellow-eyed Penguins at Moeraki.
The success of rehabilitation of compromised wildlife typically has been judged by the release rate of admissions to rehabilitation facilities (e.g.Parsons and Underhill 2005;Pyke and Szabo 2017;Dessalvi et al. 2021).Here the 93% release rate from 975 admissions to our rehabilitation facility appears impressive but is uninformative when attempting to quantify effect on population size.Our approach towards judging the success of rehabilitation of Yellow-eyed Penguins targeted effect on population size, a quantitative approach pioneered by Ryan (2003) for African Penguins.We found  (Barham et al. 2007).More specific to our study, rehabilitated Yelloweyed Penguin chicks had lower survival post-release than unrehabilitated chicks (Alden et al. 2021).We suggest that the post-release success of rehabilitation should be judged by effect on population size and not on individual performance.
The only caution against rehabilitation of Yelloweyed Penguins that we could find in a peer-reviewed publication was gains in numbers attributed to rehabilitation must not be at the expense of depletions elsewhere (Busch and Cullen 2009).Here we have described the outcomes from rehabilitation of juveniles and adults we encountered with likely life-threatening injuries, emaciation, or sickness.Without rehabilitation most if not all would have died and so any subsequent breeding at Moeraki cannot be causing depletion elsewhere.Recruitment of a rehabilitated individual cannot deplete the source population when the individual would have died without rehabilitation.Instead, most if not all the 79 rehabilitated females and their 15 unrehabilitated female descendants that bred at Moeraki were simply additions to the species population.Busch and Cullen (2009) defined our management strategy of Yellow-eyed Penguins at Moeraki as 'intensive management' because it involves a high level of support for the population.In addition, our management strategy neatly matches the least scientifically rigorous of the three types of adaptive management described in a review by Fleming et al. (2014:106): 'evolutionary (or trial-and-error), where a best guess is implemented in a haphazard way and later choices are chosen from those with the best effects'.This may sound unflattering, but it is realistic and it worked.In contrast, the most scientifically rigorous type of adaptive management is a formal experiment involving treatments and controls (Fleming et al. 2014).However, we regard treatment (rehabilitation) and control (no rehabilitation) experiments as inappropriate to judge the effect of penguin rehabilitation for two reasons.First, criteria to divide individuals into treatment and control groups are fraught with difficulties in judging and comparing the severity of symptoms.Second, the creation of a valid control group raises ethical issues and we regard deliberately leaving individuals to die for the sake of an experiment to be repugnant.
Rehabilitation facilities typically target animals compromised by injury (Dessalvi et al. 2021;Paterson et al. 2021), including mass injuries attributed to anthropogenic causes (e.g. oil spills; Ryan 2003;Henkel and Ziccardi 2018).In contrast, most Yellow-eyed Penguins admitted to our rehabilitation facility were emaciated or sick individuals (61% of 960).The current management plan for Yellow-eyed Penguins acknowledges emaciation and sickness are important sources of mortality in addition to injury and recommends rehabilitation of all compromised individuals (Te Rūnanga o Ngāi Tahu et al. 2020aTahu et al. , 2020b)).Consistent with the prerequisites specified by Dessalvi et al. (2021) and Paterson et al. (2021), this rehabilitation as a management tool does not jeopardise or limit other vital conservation measures.
Yellow-eyed Penguin nest numbers at Moeraki increased at a long-term average 5% annually through four decades.This long-term trend of increase has been interrupted by periods of decrease, including the recent decrease that may signal a permanent end to increases.We suggest that our rehabilitation of compromised individuals has succeeded by reducing losses through periods of decreasing nest numbers and enhancing gains through periods of increasing nest numbers.
Success is not guaranteed: here we evoke Gould (1989) who emphasised the importance of acknowledging contingency in natural events and noted that the future cannot be predicted with certainty.We are particularly apprehensive when facing an increasing prevalence of endemic diseases (e.g.Saunderson et al. 2021) and the appearance of novel diseases (e.g.Wierenga et al. 2023).Also, we emphasise that the successful rehabilitation of compromised penguins is a highly skilled task that cannot be condensed into following a sequence of bullet points.Ideally the health of penguins in the field should be judged from passive observation and not through systematic capture and handling that causes undue stress.We emphasise that Yelloweyed Penguins are sentient beings, not inanimate objects, and they cannot be expected to cooperate in procedures intended to improve their welfare.

Figure 1 .
Figure 1.Map of South Island, New Zealand, showing locations mentioned in the text.
(a)).Nest numbers subsequently fluctuated but with a long-term trend of increase to a maximum of 58 nests in 2013 and 2014 followed by a decrease to a minimum of 38 nests in 2019, and most recently rising to 43 nests in 2021.Overall, nest numbers averaged a 6% stochastic annual increase (Figure 2(a): n = 40 y, λ = 0.058, μ = 0.060, r 2 = 0.686, P < 0.001).This rate of increase presently appears unduly optimistic because recent years have been well below the line of best fit, with the 2019, 2020 and 2021 seasons approaching the lower 95% prediction interval.Consequently, perhaps a more realistic representation is the 5% average annual increase generated by deterministic growth rate through the 40 years from 1982 to 2021 (Figure 2(b): λ = 0.050, μ = 0.052).The decrease in nest numbers since 2014 raises the possibility that nest numbers at Moeraki are heading towards long-term stability or decrease, respectively depicted by a logistic curve and a bell curve in Figure 2(b) with details presented in Table

Figure 2 .
Figure 2. Options for trends in annual number of Yellow-eyed Penguins nests at Moeraki, southeast South Island, through 40 years, 1982-2021.Stochastic exponential growth, where the stochastic exponential line of best fit is bounded by 95% confidence interval (solid lines) and 95% prediction interval (dashed lines).(b) Three other interpretations for lines of best fit: a deterministic exponential curve (black dashed line) plotting a continual increase; a sigmoid growth curve plotting towards a plateau of 50 nests (grey solid line); and a bell curve plotting an increase to a maximum in 2014 followed by a sustained decrease in nest numbers (grey dashed line).

Figure 3 .
Figure 3. Trends in the annual number of juvenile or adult Yellow-eyed Penguins released from our rehabilitation facility at Moeraki through 36 years, 1986-2021.(a) Number of penguins each calendar year (black solid line) and number nests at Moeraki in the breeding season that started in the same year (solid grey line), with respective stochastic exponential growth rates (dotted lines).(b) Relationship between annual number of rehabilitated penguins and number of nests plotted in (a), delineated by line of best fit (linear through origin) bounded by 95% confidence interval (solid lines) and 95% prediction interval (dashed lines).

Figure 4 .
Figure 4. Effect of rehabilitation of juveniles and adults on Yellow-eyed Penguin nest numbers at Moeraki through 40 years, 1982-2021.Black line delineates actual number of nests produced by 240 females and grey area delineates likely range in nest numbers after subtracting input from 79 females breeding after rehabilitation and their 15 unrehabilitated female descendants.

Table 1 .
Destinations of the 167 breeders derived from the 591 marked Yellow-eyed Penguins released from our rehabilitation facility at Moeraki through 36 years, 1986-2021.Results here are for up to the 2021 breeding season for Moeraki and up to the 2018 breeding season for other locations.Total here is 89 instead of 90 because one female moved to breed at Otago Peninsula after first breeding at Moeraki. that lifetime reproductive success was similar for rehabilitated and un-rehabilitated female breeders.Postrelease outcomes following rehabilitation have varied among other penguin species; e.g.breeding success following rehabilitation of oiled birds was unaffected in Little Penguins (Eudyptula minor, Sievwright et al. 2019) but reduced in African Penguins a