Extreme reproductive skew at the dawn of sociality is consistent with inclusive fitness theory but problematic for routes to eusociality

Abstract

To understand the earliest stages of social evolution, we need to identify species that are undergoing the initial steps into sociality. Amphylaeus morosus is the only unambiguously known social species in the bee family Colletidae and represents an independent origin of sociality within the Apoidea. This allows us to investigate the selective factors promoting the transition from solitary to social nesting. Using genome-wide SNP genotyping, we infer robust pedigree relationships to identify maternity of brood and intracolony relatedness for colonies at the end of the reproductive season. We show that A. morosus forms both matrifilial and full-sibling colonies, both involving complete or almost complete monopolization over reproduction. In social colonies, the reproductive primary was also the primary forager with the secondary female remaining in the nest, presumably as a guard. Social nesting provided significant protection against parasitism and increased brood survivorship in general. We show that secondary females gain large indirect fitness benefits from defensive outcomes, enough to satisfy the conditions of inclusive fitness theory, despite an over-production of males in social colonies. These results suggest an avenue to sociality that involves high relatedness and, very surprisingly, extreme reproductive skew in its earliest stages and raises important questions about the evolutionary steps in pathways to eusociality.

1. Introduction

The evolution from solitary living to complex sociality involving sterile worker castes represents a major evolutionary transition and continues to generate debate surrounding the conditions under which eusociality can evolve [1]. It is generally thought that the evolution from solitariness to eusociality has arisen through a series progressive evolutionary steps referred to as a ‘social ladder’ [2–4]. The underlying assumption of this thinking is that extant species presenting similar forms of eusocial behaviour have passed through similar ‘rungs on a social ladder’ and have undergone gradual behavioural and genetic changes overtime [3,4] (but see [5]). Any significant social jumps that are able to hurdle intermediate ‘rungs on the social ladder’ are considered unlikely, especially jumps from solitary nesting to complex sociality [6] (but see [7]). A key feature that might promote such a rapid rise to eusociality is the differentiation between totipotent nest-mates at an early stage of social evolution in combination with evolutionary drivers that might offset an individual's reproductive sacrifice [8].

Eusociality has received particular attention in social evolution studies because it typically involves extreme reproductive skew, entailing morphological castes and effective worker sterility [9]. It has even been suggested that worker sterility may release species from the constraints of gene-level inclusive fitness models [10] (but see [11]). When trying to understand the evolution of eusociality, a major and on-going question could be simplistically framed as a ‘chicken-or-egg’ problem: did worker castes arise before helper sterility, or did effective sterility allow worker morphologies to evolve later on? That issue cannot be readily addressed using taxa such as ants, termites or honeybees because they do not contain extant species that might approximate conditions when sociality was first evolving [4,12]. Instead, we need to examine species where sociality has only recently evolved and where altruistic behaviour is not obligate.

Many previous attempts to elucidate the evolutionary transition from primitive to complex forms of sociality have frequently used species that were supposedly in the early steps of acquiring social traits but have turned out to involve species that are derived from ancestors where social behaviour was more complex, and so represent ‘reversions’ to seemingly less complex behaviour [13–16]. Consequently, these species may not provide strong insights into the very earliest stages of social evolution. To understand the earliest steps into sociality, we need to identify species that are still in early phases of social evolution.

The Australian native bee Amphylaeus morosus (Hymenoptera: Colletidae) is the only known social bee in the hyper-diverse Colletidae bee family. Social behaviour in A. morosus is facultative, with solitary and social nests occurring within the same population, sometimes within centimetres of each other (figure 1c). Social nests are infrequent and generally contain two nest-mates, though they can rarely contain up to three individuals, but always re-using natal nests from the previous year [17]. In the montane populations of south-eastern Australia, A. morosus is strictly univoltine; however, Houston [18] suggested the potential for more than one generation to be produced in the subtropical populations of southern Queensland, Australia. Females mate before overwintering and may either disperse and establish new nests in early spring or remain in their natal nests which are then re-used (electronic supplementary material, figure S3). Brood cells are laid in a linear sequence, with mothers ovipositing onto semi-liquid provisions before sealing cells (figure 1b). In the Dandenong Ranges, Victoria, Australia, A. morosus is attacked by at least eight different parasitoid species with staggered windows of attack driven by host resource utilization [19]. These parasitoids provide severe ecological pressures throughout the reproductive season and have been shown to dramatically increase the benefits of having at least one defender in the nest at the end of the reproductive season [20].

Figure 1.

(a) Visualization of the reproductive skew for all sequenced social nests. Blue cells represent the primary females brood and orange represents the secondary females' brood. Adults that were not present in the nest at the time of collection but laid brood in the nest are denoted with a strike through the circle. (b) Social nest with full clutch of offspring, showing sequential order of brood laying (image of nest 05). (c) Nesting habitat of Amphylaeus morosus in the Dandenong Ranges, Victoria. Yellow circles show close proximity of nesting sites. (Online version in colour.)

Using A. morosus, we demonstrate extreme reproductive skew at the dawn of sociality in the only known social species in the bee family Colletidae. We show that even in this very early stage of social evolution, extreme reproductive skew can evolve and be consistent with inclusive fitness theory. This kind of early extreme skew suggests an evolutionary landscape that is very different from previously hypothesized routes to eusociality.

2. Results

(a) . Pedigree assignments and intracolony relatedness

Amphylaeus morosus rears a single brood per year, with egg-laying commencing in spring and adults emerging in mid-to-late summer. Colony sizes during brood rearing were very small, with nests containing a maximum of two adult females. The mean pairwise relatedness, based on 950 genome-wide SNP loci, for adult nest-mates in social colonies was r = 0.589 ± 0.075 (n = 10). Of 20 genotyped colonies, comprising both solitary and social nests, SNP-based pedigree assignments revealed eighteen that contained a female who could be confidently assigned as mother to all the offspring in the nest. In two nests, the only adult present had not produced offspring, and in one of these, the offspring were entirely unrelated to the adult female present in the nest at the time of collection.

Maternity analyses and intracolony relatedness revealed the existence of both matrifilial and full-sibling colonies, both containing one ‘primary’ female who produced all or most brood, and another ‘secondary’ female who produced zero or only one brood (table 1). In matrifilial colonies, the adult daughter was the secondary and exhibited no observable damage to their wings, whereas mothers had highly worn wing margins, consistent with extended age and/or activity (electronic supplementary material, table S6). A similar pattern was seen for full-sibling colonies: full monopolization over reproduction was found in five full-sibling colonies and in only one nest was a brood cell laid by the secondary female (a single male offspring in last cell of the nest) (figure 1a). In all social nests, the reproductive primary was inferred to be the primary forager based on wing wear. Reproductive skew was not significantly different between matrifilial colonies and full-sibling colonies (t-test: t₇ = 1.07, p = 0.319) and had no significant relationship with pairwise relatedness between adult nest-mates (F_1,7 = 1.465, p = 0.265).

Table 1.

Reproductive output and intracolony relatedness between social nestmates of Amphylaeus morosus. Reproductive skew values are calculated using the binomial skew index (B) and sex ratios are calculated as the proportion of male offspring in the colony.

nest	reproductive strategy	reproductive output		intracolony relatedness	reproductive skew (B)	sex ratio
		primary	secondary
5	unrelated	3♀, 13♂	0	0a	0.469	0.813
13	matrifilial	1♀, 12♂	0	0.486	0.462	0.923
35	matrifilial	1♀, 9♂	0	0.511	0.444	0.889
36	matrifilial	1♀, 11♂	0	0.512	0.458	0.917
55	full-sibling	1♀, 5♂	0	0.699	0.417	0.857
59	full-sibling	2♀, 4♂	0	0.750a	0.417	0.667
60	full-sibling	2♀, 7♂	0	0.749	0.444	0.778
62	full-sibling	3♀, 10♂	0	0.785	0.462	0.769
77	full-sibling	3♀, 9♂	1♂	0.750a	0.319	0.769
145	full-sibling	2♀, 11♂	0	0.653	0.462	0.846
mean				0.589	0.435	0.827

^aIf only one adult female was present at the time of collection, intracolony relatedness was estimated by back-calculating pedigrees to infer the relationship between adult nestmates based on the relationship of the remaining female to the offspring in the nest.

Offspring numerical sex ratio was significantly male biased in all sequenced social nests when offspring were pooled across nests (pooled numerical ratio = 0.827, ${\chi }_1^{\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}2}=48.89,$ p < 0.001; table 1), by contrast, the sex ratio for solitary nests did not differ from a 1 : 1 ratio (pooled numerical ratio = 0.517, ${\chi }_1^{\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}2}=0892,$ p = 0.345). When the numerical sex ratio was calculated for each social colony, the mean numerical colony ratio was also male biased and significantly different from 0.5 (mean numerical ratio r = 0.823, one sample t₉ = 12.81, p < 0.001). Sex ratio was significantly influenced by the relatedness between adult nest-mates (F_1,7 = 19.42, p = 0.003).

(b) . Nestmate morphological hierarchies

All of the dissected adult females were mated (n = 21 colonies, n = 30 females). To explore if any morphology-based hierarchies occurred between social nestmates, we compared individuals in two-female social colonies for differences in ovary size, body size and wing wear. Independent-samples t-tests determined there was no statistical difference in mean ovary size and mean body size between adult female nestmates (t₁₆ = 1.96, p = 0.067; t₁₆ = 1.73, p = 0.103), but mean wing wear was significantly different (t₁₆ = 3.49, p = 0.003). To further untangle these differences, we used Monte Carlo techniques to simulate social nests which confirmed these patterns (electronic supplementary material, appendix).

(c) . Benefits of social nesting

Social nests (n = 13 colonies, 4.3% of sampled nests) were rare throughout the Dandenong Ranges compared to solitary nests (n = 289, 95.7% of sampled nests) and only contained up to two adult females at the times of collection. However, genetic sequencing of colonies revealed three ‘hidden’ social nests where a mother could be inferred but was no longer present, making it likely that social colonies are more common than our sampling effort was able to capture. On average, social nests were more productive than solitary nests in regard to reproductive output (solitary: 5.11 ± 0.22, n = 289; social: 12.31 ± 0.56, n = 13; Mann–Whitney: U = 317, p < 0.001; figure 2a). Nests with one female present at the time of collection contained a similar number of provisioned brood cells to nests with no adult female (one adult female: 5.29 ± 0.27, n = 203; no adult female: 4.56 ± 0.38, n = 86; Mann–Whitney: U = 7669.5, p = 0.068). However, nests with one adult female had significantly more offspring survive to adulthood (one adult female: 3.41 ± 0.26, n = 203; no adult female: 1.24 ± 0.27, n = 86; Mann–Whitney: U = 4666.5, p < 0.001; figure 2b).

Figure 2.

Comparison of (a) total clutch size and (b) total offspring that survived to adulthood between orphaned nests (n = 86), solitary nests (n = 203) and social nests (n = 13) of Amphylaeus morosus. (c) Inclusive fitness of secondary females for each reproductive strategy. (d) Mean genetic investment for secondary females in a nest for each reproductive strategy. Statistical significances using a Mann–Whitney test between matrifilial and full-sibling strategies are given.

(d) . Indirect fitness for secondaries

Indirect fitness was estimated for social secondaries in matrifilial and full-sibling colonies (electronic supplementary material, appendix). On average, secondary females in full-sibling colonies had slightly higher mean indirect fitness (mean = 5.54 ± 0.995) compared to secondary females in matrifilial colonies but no statistical difference was detected (mean = 4.45 ± 0.441; Mann–Whitney: U = 7.00, p = 0.599; figure 2c). Inclusive fitness was estimated as the combined direct and indirect fitness gain for primaries and secondaries, respectively (table 2).

Table 2.

Comparison of colony productivity variables and inclusive fitness benefits between life-history strategies of Amphylaeus morosus. Parasitization rate is the proportion of nests with at least one parasitised brood cell.

reproductive strategy	sample size	intracolony relatedness		reproductive output	parasitisation rate	benefit of social nesting	inclusive fitness		genetic investment by secondary
		expected	observed				primary	secondary
solitary	289	n.a.	n.a.	5.11 ± 0.222	0.290 ± 0.027	n.a.	2.78 ± 3.49	n.a.	n.a.
matrifilial	3	0.5	0.503 (0.486–0.512)	12.33 ± 1.202	0	8.89	11.67 ± 0.882	4.45 ± 0.441	3.69 ± 0.51
full-sibling	6	0.75	0.722 (0.653–0.785)	10.33 ± 1.406	0	7.39	10.17 ± 1.327	5.54 ± 0.995	3.42 ± 0.22

The genetic investment of secondary females to the primary female's offspring was calculated for matrifilial and full-sibling reproductive strategies as follows: matrifilial: n_{(female offspring)} × 0.75 + n_{(male offspring)} × 0.25; full-sibling: (n_{(female offspring)} + n_{(male offspring)}) × 0.375. The genetic investment of secondary females in full-sibling colonies (mean = 3.69 ± 0.51) was not significantly higher than secondary females in matrifilial colonies (mean = 3.42 ± 0.22; Mann–Whitney: U = 8.00, p = 0.795; figure 2d).

3. Discussion

Our use of genome-wide SNPs indicated extreme reproductive skew in both matrifilial and full-sibling colonies in a species that was previously thought to be egalitarian and consisting of only weakly related individuals [17]. Importantly, we find that this extreme skew is consistent with Hamilton's rule [20,21], yet hierarchies based on size or morphology are absent. On the other hand, we did find one nest containing an adult female that was unrelated to any of the brood, suggesting a low level of drifting behaviour [22,23]. Our data indicate that social nesting in Amphylaeus morosus is uncommon (less than 5% of the nests sampled), and this raises questions as to the benefits of nest sharing. In A. morosus, the indirect fitness benefits for social secondaries exceeded the benefits of reproducing alone (table 2) and demonstrated a significant advantage to social nesting. Amphylaeus morosus is host to a diverse suite of parasitoids that attack at different stages throughout the reproductive season, including severe risk of mortality from mutillid wasps at the end of the season when nests contain full broods [24]. This parasite pressure is likely to be a key driver promoting cooperative nest defence in this species [25].

Intracolony relatedness for A. morosus was previously estimated at r = 0.26 [17] and this discrepancy between our higher relatedness estimate is likely to be due to the resolution of the available technology (allozymes) used at the time. Matrifilial associations of adults during brood rearing were not considered as possibilities in that earlier study of A. morosus [17], but our pairwise relatedness estimates combined with wing wear patterns make this colony composition apparent in some nests (electronic supplementary material, appendix), indicating that some individuals in this species can survive across consecutive years. Extended longevity of adults, allowing adult mother–daughter associations, is widely regarded as a prerequisite of eusociality [2,26–28]. At the same time, we did not find that reproductive skew differed between matrifilial and full-sibling colonies. This is important because it indicates that extreme skew can arise regardless of the potential for this to subsequently evolve into queenworker systems that are based on mother–daughter relationships. Additionally, we found no evidence of ovarian development suppression in the secondary females, which does not support the de-coupling model for the evolution of morphologically identifiable castes in the early stages of social development [29]. This prompts the question of whether matrifilial associations and distinct reproductive phenotypes really are pre-conditions for the evolution of eusociality; instead, the existence of extreme skew could create selection for increased adult longevity that could exploit already-present queen-like roles. We note that under some definitions [30], some colonies of A. morosus could be regarded as eusocial based on reproductive monopolization by mothers in matrifilial colonies.

Using allodapine bees as an exemplar life cycle, Michener [7] proposed that eusociality could rapidly evolve from subsocial colonies if first-emerging daughters simply switch from producing their own offspring to rearing their younger siblings. However, subsequent studies have shown that allodapine and xylocopine species with high levels of reproductive skew can comprise matrifilial, full-sibling, as well as less related compositions of adult females, as seen in the carpenter bee Xylocopa sulcatipes, where initially subordinate helpers may inherit the nest and become reproductive [31,32]. Matrifilial and full-sibling colonies have often been associated with the subsocial and semi-social routes to eusociality, respectively, yet those hypothetical pathways operate under distinctly separate evolutionary dynamics [7,25,33,34]. Subsociality could potentially favour worker-like daughters because of high full-sister relatedness in haplodiploids, whereas the semisocial route is thought to be more strongly driven by mutualistic factors, with worker-like behaviour evolving once colonial life has become established [25]. In A. morosus social nests, sex ratios were heavily male-biased resulting in no significant difference in either the inclusive fitness benefits or the genetic investment of secondary females when comparing matrifilial and full-sibling colonies. In other words, the potential for helper daughters to exploit high sister–sister relatedness was negated by mothers producing more sons than daughters.

The ability for workers to assess relatedness asymmetries and augment their indirect fitness by skewing sex allocation towards sisters has been reported for some eusocial Hymenoptera [35] but it is unclear if such biased alloparental care could promote origins of eusociality rather than simply be a response by already-existing worker castes [9,36]. However, such worker-controlled sex ratios are not possible in A. morosus, given that killing a mother's male egg would result in a provisioned cell without any brood, or else the secondary female would have to replace a primary's egg with her own and neither of these outcomes were detected in our sequenced colonies. The over-production of male brood may also limit the extent of social nesting within the population: if the presence of a guard increases the number of male brood, the increased number of males entering the population will lower their mean reproductive success, leading to diminishing indirect fitness returns for secondaries as they become more frequent in a population.

The potential for secondary females to occasionally survive into a second year of brood rearing and then assume reproductive dominance raises the question of reproductive queuing. Our estimates of inclusive fitness for secondary females in A. morosus did not explore this possibility, but reproductive queuing that entails alternation between initial indirect fitness gains with subsequent direct fitness has been shown for the large carpenter bee, Xylocopa sulcatipes [32], and has been posited for multiple allodapine bee species where it can eventually result in the evolution of permanent worker castes [37]. The costs of dispersal and social contests between nestmates may also be important factors determining why secondaries choose to stay in a social nest and relinquish direct reproduction [38]. While nesting sites do not appear to be limiting in the Dandenong Ranges [17] the costs of constructing new nests has not been estimated, though Spessa et al. [17] showed that brood provisioning begins earlier in social re-used nests than newly constructed ones.

High intracolony relatedness and reproductive altruism are characteristics underlying the most advanced forms eusociality and while these characteristics are sometimes present in small totipotent societies of extant halictine, ceratinine, allodapine bees and polistine wasps, those are not truly representative of the earliest stages of social evolution as they arise from lineages with very long histories of prior social behaviour [4,7,39]. Using A. morosus, a species in very early stages of social evolution, we have shown that very early forms of sociality can entail extreme reproductive skew in the absence of morphological castes, but this altruism is consistent with Hamilton's inclusive fitness theory. These two findings suggest that effective worker sterility can arise very early on in social evolution (an idea also raised in [7] and [29]), yet it remains a feature that is not accounted for in most proposed pathways to eusociality. Furthermore, the indirect fitness gains from altruist sterility are not significantly different between matrifilial and full-sibling colonies. When combined, these features do not fit neatly into proposed early steps in any of the hypothesized routes to eusociality and, we argue, those routes need to be re-evaluated.

4. Material and methods

(a) . Nest collection

Nests of Amphylaeus morosus were collected from the Dandenong Ranges, Victoria, Australia. Nests were sampled from seven separate collections during austral spring and summer across 5 years from 2017 to 2021. Within the Dandenong Ranges, A. morosus nests in dead abscised fronds of the rough fern tree, Cyathea australis. Nests were opened longitudinally up to the first brood cell so that mothers and additional contents preceding the first cell could be removed. Immature brood were left in the nest to develop to adult eclosion prior to extraction, at which point they were sexed. Upon nest opening, the contents including number of brood, parasitized brood and surviving brood were recorded. Brood and their presumptive mothers/alloparents were removed from the nest and placed in 99% ethanol for genetic sequencing.

(b) . Genotyping

Single nucleotide polymorphisms (SNPs) for 197 brood and 31 potential mothers collected from 20 nests over three collection periods (electronic supplementary material, table S1) were assayed using high-throughput microarray sequencing following the methods described in [40] using the DArTseq (Diversity Arrays Technology sequencing) proprietary platform, completed by Diversity Arrays Technology (Canberra, ACT, Australia). DNA was extracted from the head or thorax of each specimen to optimize DNA quality and minimize potential DNA contamination from gut microbiomes in metasomal tissue. Preliminary sequences revealed no differences in the sequencing depth between head and thoracic tissues.

(c) . Loci quality filtering

A total of 17 194 SNP loci were called for the 228 assayed individuals of A. morosus. Quality filtering of SNP markers was performed using the dartR package v. 1.9.9.1 [41], implemented in R v. 4.0.4 [42]. SNPs were filtered for minor allele frequencies below 5%, SNPs that share a sequence tag, monomorphic loci, loci with a repeatability less than 99%, and loci with a call rate less than 99%. Loci showing apparent linkage disequilibrium (LD) were filtered using the R package SNPRelate v. 1.28 [43] at an r² threshold of 0.8, resulting in the retention of 950 loci.

Measure of allelic diversity was conducted using all adult females from each nest and four additional adults independent from the samples collected (n = 31). The population inbreeding coefficient, F_IS and departures from Hardy–Weinberg equilibrium were calculated in the R package SNPRelate (electronic supplementary material, appendix). To avoid any confounding influences from genetic correlations among related individuals, calculations of population allele frequencies were weighted by the inverse of colony size.

(d) . Pedigree assignments and relatedness estimates

To explore patterns in parentage, pedigree relationships were reconstructed for the brood and potential mothers of all genotyped individuals using the program Colony v. 2.0.6.6. [44]. A maximum-likelihood approach was used to infer putative pedigrees based on population allele frequency. This approach accounts for genotyping errors, where the genotyping error rate was set to 0.001. Details of the parameters used for these analyses are given in electronic supplementary material, appendix.

Pairwise relatedness and mean intracolony relatedness for each nest group were estimated using the program Kingroup V2 [45]. The putative pedigrees from Colony (based on highest log probability) were used to partition individuals into groups. The pairwise relatedness between both adult females-to-offspring and offspring-to-offspring in a given colony was calculated using the relatedness estimator in Kingroup V2 [46] based on population allele frequencies where colonies, rather than individuals, were given equal weight.

(e) . Measuring reproductive skew

Reproductive skew was calculated using the software Skew Calculator 2003 [47] for all genotyped social nests where maternity of brood could be confidently evaluated. We used the binomial skew index (B) [48] which tests within each group if the observed variance in reproduction significantly differs from the expected variance if all group members have an equal probability to reproduce (electronic supplementary material, appendix).

(f) . Dissection data

Adult females from all sequenced nests (n = 21 females) and adult females from an additional three solitary (n = 3 females) and three social nests (n = 6 females) were placed in 70% ethanol for 24 h prior to dissection. Individuals were dissected under a Leica MS5 stereomicroscope and measurements were taken for wing length, wing wear, ovary size and mated status. To avoid problems with some occasionally very worn wing margins, wing length was measured as the distance from the axillary sclerites to the apex of the marginal cell on the forewing. In A. morosus, wing length was found to have a linear relationship with body weight and intertegular distance of brood that had just reached adulthood and was therefore used as a proxy for body size (electronic supplementary material, figure S4). Wing wear was assessed as the number of nicks in the distal forewing margins and was used as a proxy of individual age and foraging activity [49]. Ovary size was measured as the arithmetic mean length of the three largest oocytes. Mated status was determined by observing the presence of sperm in the spermatheca. Monte Carlo resampling procedures were used to determine whether morphological hierarchies existed within social nests (electronic supplementary material, appendix).

(g) . Fitness estimation

Indirect fitness estimates were calculated by multiplying the expected pairwise relatedness between social nestmates (i.e. mother–daughter = 0.5; full-sisters = 0.75) by the benefit of social nesting, which was estimated as the difference between the mean number of offspring that survived to adulthood in social nests for matrifilial and full-sibling strategies and the mean number of offspring that survived to adulthood in solitary nests.

(h) . Statistical analyses

All statistical analyses were performed in R v. 4.0.4 [42]. We used Mann–Whitney and Kruskal–Wallis tests to assess differences in total brood output, surviving brood and parasitization rates between solitary and social nests. Colonies were grouped based on the number of adult females present at the time of collection. For some analyses, we classified colonies where the social dynamics were ambiguous at the time of collection as follows—nests with no adult females at the time of collection (n = 86 colonies) as solitary and genome-inferred multi-female nests (colonies where one mother had abandoned the nest prior to collection; n = 3 colonies) as social nests. Unless otherwise stated, values are presented as mean ± s.e.

Data accessibility

The analyses reported in this article can be reproduced using the dataset provided by Hearn et al. [50].

The data are provided in electronic supplementary material [51].

Authors' contributions

L.R.H.: conceptualization, data curation, formal analysis, funding acquisition, investigation, writing—original draft, writing—review and editing; O.K.D.: conceptualization, formal analysis, funding acquisition, writing—review and editing; M.P.S.: conceptualization, investigation, supervision, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

The authors declare no competing interests.

Funding

This project was supported by the Australian Federal Governments Research Training Program Scholarship and the Holsworth Wildlife Research Endowment & The Ecological Society of Australia grants awarded to L.R.H. and O.K.D.