Botany

Subdividing the spectrum—quantifying host specialization in

Journal: Botany

Manuscript ID cjb-2019-0207.R1

Manuscript Type: Article

Date Submitted by the 17-Mar-2020 Author:

Complete List of Authors: Milner, Kirsty; University of Technology Leigh, Andrea; University of Technology Sydney Gladstone, William; University of Technology Sydney Watson, David;Draft Charles Sturt University, Keyword: host preference, host spectrum, host turn-over, parasitic ,

Is the invited manuscript for consideration in a Special IUFRO 2019 Dwarf Mistletoes Symposium Issue? :

https://mc06.manuscriptcentral.com/botany-pubs Page 1 of 32 Botany

1

Subdividing the spectrum—quantifying host specialization in mistletoes

Kirsty V. Milner1, Andrea Leigh1, William Gladstone1, David M. Watson2,3

1 School of Life Sciences, University of Technology Sydney, Ultimo NSW 2007 Australia;

([email protected], [email protected], [email protected] )

2 Institute of Land, Water and Society, Charles Sturt University, NSW 2640 Australia

3 Corresponding author:

David M. Watson, School of Environmental Sciences, Charles Sturt University, P.O. Box 789,

Albury, 2640 Australia E: [email protected] Draft P: +61 2 6051 9621

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 2 of 32

2

Abstract

Parasites necessarily depend on their hosts, but the number of host species used by a parasite varies from one to hundreds. Estimating host range and identifying preferred host species that influence distributional boundaries and confer greater advantage to the parasite has proven elusive. As well as the confounding effects of sampling effort, characterising host specificity and preference has been hindered by considering host use without accounting for availability. We selected three mistletoe species— exocarpi, quandang, and Amyema lucasii— and sampled mistletoe-host interactions and host availability free from sampling bias. To quantify host specificity and identify preferred host species we applied specialist/generalist scores (G) and resource selection ratios (ω) respectively. Host specificity and preference were assessed at four scales. The generalist L. exocarpi was foundDraft to parasitise 31 plant species. Even at small scales G values and host species turnover were high, with eight preferred hosts identified.

Amyema quandang had a low G score with significant preference for half of its hosts.

Amyema lucasii significantly preferred one host, consequently having low G value at all scales.

By collecting potential host data and applying G scores and ω, parasite host spectrum can be quantitatively estimated rather than qualitatively described.

Keywords host preference, host spectrum, host turn-over, , mistletoe

https://mc06.manuscriptcentral.com/botany-pubs Page 3 of 32 Botany

3

Introduction

Parasites have a pervasive influence on ecosystems (Hudson et al. 2006) and, while their

combined effects extend well beyond pairwise interactions with individual hosts, the magnitude

and extent of their influence is often tied to the number of host species they infect (i.e., their host

range). The degree to which parasites are able to use different host species—host spectrum (Poulin

2011)—ranges between two divergent strategies: high specificity, the reliance upon very few hosts

vs generalism, the use of many hosts, often in proportion to availability. Although a primary

determinant of many aspects of parasite life history, demarcating any groupings along this

continuum is challenging.

Three aspects have confounded estimation of host range, and constrained our

understanding of the actual variation in hostDraft use among parasites. The first is sampling bias, where

greater sampling effort necessarily identifies greater numbers of hosts (Walther et al. 1995;

Downey 1998; Grenfell and Burns 2009;). Although ecological parasitologists have long

recognised that estimates of host spectrum are susceptible to sampling bias (Poulin 1992; Walther

et al. 1995; Guégan and Kennedy 1996; Walther and Morand 1998), recent work with mistletoes

has successfully applied results-based stopping rules to standardize sampling and generate

comparable estimates (Watson et al. 2017). The second issue relates to the quantification of host

spectrum. Simply listing all recorded host species for a parasite assumes that all hosts are equally

important (Krasnov et al. 2011; Poulin et al. 2011), the accidental infection of one host individual

just as relevant as the most frequently infected host species that confers reproductive advantages.

Host preference, on the other hand, considers the relative ecological importance of each host

species (Poulin et al. 2011) and the evolutionary time over which host-dependent fitness has

formed (Giorgi et al. 2004), rather than simply consider whether a host is used by a parasite. To

date, however, variation in sampling effort has confounded estimation of host preference,

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 4 of 32

4 preventing comparative studies. Finally, since spatial variation and relative abundance of hosts in the environment controls host specificity via encounter rate (Jaenike 1990; Norton and Carpenter

1998; Norton and De Lange 1999), reliable estimates of host preference rely on representative estimates of host availability (Norton et al. 1995; Mourao et al. 2016). This is further complicated when host use varies across the geographic range of a parasite (Poulin et al. 2011) which may reflect both spatial variation in host specificity and/or spatial variation in host availability, neither of which are considered in many assessments of host spectrum (Grenfell and Burns 2009; Blick et al. 2012; Mans et al. 2014). Given this inherent complexity coupled with methodological difficulties in obtaining comparable samples, specialisation, to our knowledge, has not been determined quantitatively for parasite-host relationships.

Mistletoes are an ideal system forDraft testing theories in parasite ecology and developing novel approaches to collect empirical data to inform and test them. Mistletoes are hemiparasites, physically and physiologically connected to their host through haustoria, establishing a functional connection to the host xylem (Yan 1993) to acquire water, minerals, nutrients and some photosynthates (Mathiasen et al. 2008). The sessile nature of host and parasite and mistletoes’ macroparasitic nature mean infection is apparent in the environment (Reid 1989; Overton 1994).

These characteristics mean accurate estimates of abundance and density of mistletoes, hosts and potential hosts can be made, especially for vegetation types with more open canopies and lower trees. A primary aim of this study was to quantify host spectrum beyond qualitative descriptions.

Here, we use data on host use and host availability for three mistletoe species with divergent host ranges. We apply an index of specialization developed in plant-pollinator systems to derive quantitative estimates of host specificity. We established the primary or ecologically meaningful hosts for each of the three mistletoe species using resource selection ratios to identify host preference. Until now the host preference of Australian mistletoes has been considered at a highly

https://mc06.manuscriptcentral.com/botany-pubs Page 5 of 32 Botany

5

localised scale (Yan 1990; Ward 2005) but host spectrum varies with scale; a parasite could be

either a specialist or generalist at a given location and/or across its geographic distribution

(Krasnov et al. 2005). Therefore, a second aim was to sample in multiple habitats across bioregions

to explore how host spectrum varies with scale. To our knowledge, this is the first time these

indices have been applied to parasites.

Method

Study species

Based on qualitative descriptions of recorded hosts in floras and host lists derived from

herbarium specimens (Barlow 1984; Cunningham et al. 1992; Quirico 1992; Downey 1998; Watson 2019), three mistletoe species withDraft differing host spectrums were selected, all members of the family. The mistletoe with the narrowest host spectrum was leopardwood

mistletoe Amyema lucasii (Blakely) Danser, considered to be almost exclusively dependent on

leopardwood ( maculosa) hosts, but across its range has been recorded on Grevillea

striata and parviflora. Grey mistletoe (Lindl.) Tiegh. has been

recorded on 53 host species, of which 40 are in the Acacia genus, and was chosen to represent

intermediate host spectrum. The study species with the broadest host spectrum was harlequin

mistletoe Lysiana exocarpi (Behr) Tiegh. It has been recorded infecting 114 species in 45 genera

from 21 families, with the most common hosts in the genera Acacia, Senna, Casuarina,

Eremophila, , , Santalum and Amyema (epiparasitic on the latter three parasitic

genera in the ). Greater detail on mistletoe host species and distributions are given in

Watson et al. (2017).

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 6 of 32

6

Sampling mistletoe hosts

Field sampling was conducted on roadsides across , Australia in April–

September 2014, in six bioregions (, , Darling Riverine Plains, Cobar

Peneplain, and Complex). A spatially-explicit results-based stopping rule was used to estimate the number of host species infected by each mistletoe species, independent of sampling effort (see Watson et al. 2017 for details). Although standardized by estimated completeness to ensure comparability within and between vegetation types, we recognize that our roadside samples may not be representative of more continuous woodlands far from the influence of roads (Milner 2014; Watson et al. 2017). Sample plots commenced where a target mistletoe species was encountered and continued until 25 individual mistletoe-host pairs were recorded. Sampling occurred in at leastDraft three different habitats within a bioregion before host species richness was predicted with a sample coverage estimator (Chao and Lee 1992), using

EstimateS (v 9.1.0, Colwell 2013). If observed host richness was ≥80 % of predicted host richness sampling in the bioregion was considered complete. If not, sampling continued in the bioregion and predicted host richness was re-calculated after each sampling plot until the 80 % threshold was reached.

Nearest-neighbour sampling

For each target mistletoe species, the 10 woody closest to each of the 25 infected hosts in each sample plot were identified to species. These potential hosts reflect those individuals most likely to receive mistletoe seeds within the sample plot (after Rödl and Ward 2002;

Okubamichael et al. 2011;). Nearest neighbours were recorded if they had a diameter at breast height (DAB) >150 mm and were considered a single individual if multiple trunks were touching at the base. Nearest neighbour data for each woody plant species sampled was decomposed into

https://mc06.manuscriptcentral.com/botany-pubs Page 7 of 32 Botany

7

two variables: number of individuals infected (used resource) and number uninfected (unused

resource) with the target mistletoe species.

Analysis

Host specificity—specialist/generalist Index (G)

Host spectrum was calculated using Medan et al. (2006) specialisation/generalisation index

(G), which minimises the bias from the increased likelihood that generalist interactions will be

encountered more often than specialist interactions. Also, the index includes a measure of the

evenness of species—a measure of hosts used relative to their proportion in the environment—

giving weight to the more ecologically meaningful interactions. Specialisation/generalisation G index for species i is calculated as the productDraft of resource usage (R) and evenness (E):

(1) 퐺푖 = 푅푖 ∙ 퐸푖

where Gi takes a value from zero to one. A value close to zero indicates a highly specialised parasite,

while a value close to one is highly generalised parasite. The component Ri was calculated by:

(2) 푆푖 푅 = 푖 푁

where Si is the number of host species interacting with parasite species i and N is the total number

of potential host species. Ei, based on the Shannon evenness measure (Taki and Kevan 2007), was

calculated by:

∑ (3) ― 푗푝푗 ∙ 푙푛 푝푗 퐸푖 = 푙푛 푆푖

where pj is the proportion of hosts in the environment. When parasite species i has one partner, a

value of 1 is assigned because Ei cannot be calculated (lnSi = 0).

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 8 of 32

8

Preferred host analysis—Resource selection ratios

Host preference was estimated using resource selection ratios—a proportional comparison

of used resource (hosts) and unused resources (non-hosts). The sampling design fits the

assumptions of Manly et al.’s (2002) Design I: known proportions of available resources with the

appropriate resource selection ratio equation applied (Manly et al. 2002, equation 4.14):

(4) 표푖 휔푖 = 휋푖

where oi is the proportion of used resource in category i (i.e. number of infected individuals of

species i / infected + uninfected individuals of species i) and πi is the proportional availability of resource in category i in relation to all available resources (i.e. infected + uninfected individuals of species i / total number of individual Draftnearest neighbours). Therefore, ωi is the preference score for resource in category i (i.e. host species i).

A value of ωi > 1 suggests a parasite may prefer host species i; however, the significance

of the resource selection ratio must then be tested using the confidence limits for ωi. If the upper

confidence interval is <1 the host species was significantly avoided. If the lower confidence

interval is <1, there is no preference for the host species, i.e., it is used in proportion to its

availability. When the lower confidence interval is > 1, the host species is significantly preferred

(Manly et al. 2002). To allow for multiple comparisons between all resource categories (host

species) a Bonferroni correction was applied by dividing the significance level (α = 0.05) by the

number of host species.

Scale

Since we were evaluating patterns of host use at different scales, data were analysed at

multiple scales. Specialisation/generalisation (G) values were calculated at the sample plot level,

https://mc06.manuscriptcentral.com/botany-pubs Page 9 of 32 Botany

9

plot data were pooled into habitats within bioregion, and all data were pooled to generate a G value

over the entire study range (herewith referred to as state-level, so as not to be confused with species

geographic range). As with host spectrum data, the additional effect of scale on host preference

was analysed at two levels: habitat within a bioregion and across the state.

Spatial turnover in host species - β-specificity—Similarity index (Chao-Sørensen)

Having evaluated host use at multiple scales, patterns of host use across different areas of

their geographic range were explored. In free-living organisms, geographical diversity can be

recorded as α-diversity—local species diversity—and β-diversity—the difference in composition

among regions. Similarly, in parasitic organisms, α-specificity reflects host specificity of a parasite

in one location, while β-specificity reflects spatial turnover in host species used across regions

(Poulin et al. 2011). β-specificity becomesDraft a measure of local host specialisation and the extent to

which hosts can be exchanged in space (Krasnov et al. 2011). β-specificity allows for the pairwise

comparison of bioregions to reveal how similar the suite of hosts used in one bioregion are to the

suite of hosts used in the next. Spatial turnover in host species or β-specificity was calculated in

using the Chao-Sørensen similarity index. This index is a probabilistic approach extended from

the Sørensen index (previously used by Krasnov et al. 2011; Poulin et al. 2011), which uses

abundance data and corrects for unseen shared species (Chao et al. 2005, 2006). Abundance data

were entered as repeated measures within bioregion in EstimateS v 9.1.0 (Colwell 2013). A

parasite with high β-specificity will have a score close to one where all host species are shared

between bioregions, while, a generalist parasite will have a score close to zero, indicating

extremely low geographic specificity meaning that a parasite species uses entirely different hosts

from location to location (Fadini 2011; Okubamichael et al. 2011).

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 10 of 32

10

Results

A total of 11,825 plants were recorded as potential hosts in this study, with 2,516 infected with one of the three mistletoe species. A total of 1,339 individual hosts from 31 plant species were encountered as hosts for L. exocarpi. For A. quandang, 952 individual hosts were recorded, from 8 host species and A. lucasii was present on 225 individual hosts from a single species. A similar proportion of potential hosts were infected with L. exocarpi, A. quandang or A. lucasii

(0.19 ± 0.01, 0.27 ± 0.03 and 0.16 ± 0.04, respectively; mean ± SE).

Host specificity—specialist/generalist Index (G)

Mean specialist/generalist index G of L. exocarpi were greater than 0.2 (Fig. 1a), A. quandang were less than 0.5 (Fig. 1b)Draft and G did not exceed 0.11 for A. lucasii (Fig. 1c) at all spatial scales from plot to habitat to bioregion to state-level. From these values the degree of specialisation/generalisation (G) categorised the host spectrums of L. exocarpi and A. lucasii as having low specificity and very high specificity respectively, and A. quandang intermediate with high specificity. With further applications of our approach, these terms can be defined more explicitly with reference to specified index values.

The G for each of the three mistletoes differed with scale (Fig. 1). The ‘generalist’ mistletoe,

L. exocarpi, displayed progressively less specialisation with increasing spatial scale (Fig. 1a).

Nevertheless, even as mean G increased with scale at the habitat and bioregion scales, L. exocarpi showed a range of G values from a highly specialist parasite to a generalist parasite (G = 0.08–

0.85 and 0.12–0.91 respectively). With increasing scale, Amyema quandang initially followed a similar pattern to L. exocarpi, moving from a local specialist to less specialised-parasite at the regional scale. Unlike L. exocarpi, at the state-level scale, A. quandang showed high host specificity, similar to its specificity at the smallest scale (Fig. 1b). Despite showing high specificity

https://mc06.manuscriptcentral.com/botany-pubs Page 11 of 32 Botany

11

at most scales, in the Riverine Plain Woodlands habitat of the Riverina bioregion, A. quandang

infected a single host species (Acacia pendula) where all the neighbouring trees were of the same

species resulting in G = 1, the maximum possible value and the highest G (lowest specificity) we

recorded. When this extreme case is omitted, A. quandang showed a narrow spectrum of specificity

at the habitat (G = 0.09–0.25) and bioregion (G = 0.14–0.22) scales. Finally, A. lucasii showed the

opposite pattern of the generalist mistletoe, with its high local-scale specialisation becoming

progressively more specialised at increasing spatial scales (Fig. 1c). This mistletoe displayed the

highest average degree of specificity at all scales (Fig. 1c) within the narrowest of ranges at the

sample plot scale (G = 0.07–0.11).

Host preference

All mistletoes recorded had a significantDraft preference for at least some of their recorded hosts.

Preference was considered at two scales, habitat within bioregion and across the sampled state-

level range. The three species showed differences in host preference with scale: Lysiana exocarpi

significantly preferred 26 % of hosts at the habitat level and 13 % of hosts at the state-level, while

the proportional preference for hosts at different scales did not differ for A. quandang or A. lucasii.

Amyema quandang showed significant preference for 50 % of its hosts and A. lucasii had 100%

preference for its host at the habitat and state-level scales.

Lysiana exocarpi showed significant preference for eight hosts (of 31 species infected) at

the habitat level (Fig. 2, Supplementary Table S1) of which three species occurring in multiple

habitats were significantly preferred (Casuarina pauper (Fig. 2a-c), Pittosporum angustifolium

(Fig. 2a,b) and Allocasuarina luehmannii (Fig 2e)). Acacia stenophylla (Fig. 2b) and Casuarina

cristata (Fig. 2d) were significantly preferred in the one habitat in which they occurred. Alectryon

oleifolius (Fig. 2a–d), sturtii (Fig. 2a,b), and Acacia salicina (Fig. 2e) were

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 12 of 32

12

significantly preferred in half of the habitats in which they occurred. In two habitats L. exocarpi

showed no preference for any host species (Fig. 2b,c). At the whole-of-distributional-range scale,

the number of significantly preferred hosts was reduced to four: P. angustifolium, A. oleifolius,

A. salicina and E. sturtii (in order of preference; Fig. 3, Supplementary Table S2). Genera of

available hosts that were never recorded as hosting L. exocarpi include Brachychiton, ,

Chenopodium, Choretrum, Dodonaea, Eucalyptus, Geijera, Grevillea, Lycium, Maireana,

Myoporum, Notelaea, Pimelea, Prostanthera and Schinus (Supplementary Table S2).

At the habitat scale A. quandang significantly preferred four of eight hosts (Fig. 4,

Supplementary Table S3); two in all habitats in which they occurred (Acacia homalophylla (Fig.

4a,d) and A. loderi (Fig. 4c)) and two for most of the habitats in which they occurred (A. aneura

(Fig. 4a,c) and A. pendula (Fig. 4b,d)).Draft The remaining four hosts were infected in proportion to their relative abundance in the habitat (results for Acacia burkittii and Amyema linophylla, regarded as provisional as there were <5 individuals (Fig. 4a)). At the state-level scale of sampling,

A. quandang showed significant preference for the same four hosts (Fig. 5, Supplementary Table

S4). Aside from the single instance of an epiparasitic infection of Amyema linophylla, all recorded hosts were in the genus Acacia (Supplementary Table S4), despite the availability of 26 potential host species from 17 genera and 12 families.

The strict species-specialist A. lucasii showed preference for its sole host F. maculosa in both habitats sampled (North-west Plain Shrublands and Sand Plain Mulga Shrublands) and at the state-level scale (Supplementary Table S5). There were 26 potential host species (15 genera, 12 families) growing beside infected F. maculosa that were never recorded hosting A. lucasii

(Supplementary Table S6).

Spatial turnover in host species - β-specificity—Similarity index (Chao-Sørensen)

https://mc06.manuscriptcentral.com/botany-pubs Page 13 of 32 Botany

13

Pairwise analysis of bioregions gives β-specificity values that reveal host turnover across

biogeographic regions by comparing similarity of host-use. As expected, based on its extreme high

specificity (Fig. 1), A. lucasii scored maximum similarity (1) between the

and Cobar Peneplain, resulting from high β-specificity. By contrast, host-use between bioregions

for L. exocarpi ranged from high similarity of host species used for adjacent regions (0.85 for

Broken Hill Complex and Mulga Lands) to no shared host species for the most distant regional

pair (Cobar Peneplain and Sydney Basin) (Table 1). Host species present in the Sydney Basin were

the most distinct suite of species for L. exocarpi. Amyema quandang used similar hosts in the

Darling Riverine Plains and Cobar Peneplain (0.90) and also in the Darling Riverine Plains and

Riverina (0.93) (Table 2). Although the Darling Riverine Plains shared many species with both the Cobar Peneplain and Riverina, these twoDraft bioregions have low similarity to one another (0.17) (Table 2). For this species, Broken Hill Complex supported the most distinct suite of hosts.

Discussion

Given the confounding effects of sampling effort, previous work has been unable to

disentangle observed patterns of host use nor develop robust hypotheses concerning their

mechanistic basis. By using a quantitative approach to evaluate mistletoe infection relative to host

availability, we were able to test descriptions of host specificity and identify host species

significantly preferred by each mistletoe species. In addition to enabling a more thorough

exploration of the multiple determinants of a parasite’s host-spectrum, our application of results-

based stopping rules allows this work to be replicated efficiently at multiple scales, from localised

infections to across the distributional range of the parasite.

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 14 of 32

14

Findings from each of the quantitative tests in this study support the qualitative descriptions

of L. exocarpi as generalist, A. quandang as genus specialist and A. lucasii as a strict specialist

(Watson 2019 and references therein). Lysiana exocarpi showed a low degree of specificity (high

G), with significant preference for only a small proportion of its hosts at the state-level scale and

an ability to switch hosts (low β-specificity) between bioregions. These findings suggest that this

species is an opportunistic parasite that can use a multitude of hosts (Krasnov et al. 2005; Takken

and Verhulst 2013). In stark contrast, Amyema lucasii had the highest degree of specificity (low

G) and β-specificity (spatial host turnover), displaying a strong preference for a single host species at all scales covered in this study. This finding confirms A. lucasii as one of the most specialised

mistletoes, occurring almost exclusively on F. maculosa (Barlow 1984; Cunningham et al. 1992). Along the host spectrum Amyema quandangDraft is intermediate, having a relatively high specificity and showing significant preference for half its Acacia hosts. Tests of β-specificity suggest

A. quandang is indeed a genus specialist; it has the ability to switch to other Acacia hosts when

needed, for example, when dispersed into a new environment where its source Acacia host does

not occur. However, qualitative descriptions were most applicable at the state-level range of the

study as host-specificity varied with scale.

As host use by a parasite is constrained at the smallest scales (localised infection at the

stand scale) by both adaptation to local conditions (Giorgi et al. 2004) and availability of hosts

(Jaenike 1990; Norton and Carpenter 1998) it was expected that all mistletoes would have high

specificity at the local scale. Amyema lucasii always recorded high specificity, however in both

A. quandang and L. exocarpi instances of high and low specificity were recorded at small scales.

Generally, A. quandang displayed low G within bioregions, but in Riverine Plain Woodlands, a

stand of A. pendula was infected by the mistletoe, resulting in the lowest specificity recorded for

any mistletoe (G = 1). This apparent expression as a generalist at the local scale was because there

https://mc06.manuscriptcentral.com/botany-pubs Page 15 of 32 Botany

15

was no choice of host and aggregated infections are common due to disperser behaviour (Mourao

et al. 2016; Messias et al. 2014) and host compatibility (Fonturbel et al. 2017; Fonturbel et al.

2019). In the case of L. exocarpi, it is a specialist (G = 0.18) in Riverine Sandhill Woodlands

habitat of the Riverina, whereas in a sample plot in the Sand Plain Mulga Shrublands habitat of

the Broken Hill Complex it showed a low level of specificity (G = 0.57), suggesting it can be a

generalist at even the smallest scale. These variations in host spectrum for L. exocarpi result, in

part, from availability of hosts (Jaenike 1990; Norton and Carpenter 1998) but also highlight that

simple host lists do not inform us on the importance of host species to parasites. If we consider the

importance of hosts species, despite conflicting G values, in both these habitats A. oleifolius was

the only significantly preferred host. Host preference is a far more important measure of host use because preference can inform about anDraft organism’s survival requirements (Manly et al. 2002).

All study mistletoe species showed preference for at least some of their hosts, as recorded

in past studies (Norton et al. 1995; Giorgi et al. 2004; Fadini 2011; Blick et al. 2012). It was

expected that a specialist parasite would show preference for its hosts in a particular habitat and

continue to show this same preference at a larger scale, as seen for A. lucasii and A. quandang,

where both species recorded the same proportion of preferred hosts at the scales of habitat and

state-level. Generalist parasites, however, may show preference for certain hosts at the local scale

but use hosts in proportion to their availability at a larger scale (Krasnov et al. 2005; Takken and

Verhulst 2013). Lysiana exocarpi significantly preferred a higher proportion of its hosts at the

habitat scale (0.26) than over the state-level range (0.13). The four species preferred as hosts across

the state-level range were present in two or more habitats, suggesting that these hosts are

encountered sufficiently frequently to develop preference. Given that mistletoes on occasional host

species are smaller than those on main hosts (Yan 1990) host preference reflects the trade-off

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 16 of 32

16

between benefits of high reproductive rates and costs associated with infecting more host species

(Tripet et al. 2002).

Host quality may also confer advantages that result in preference of certain host species.

Only a single taxon was preferred host to more than one mistletoe study species. Various Acacia

species were hosts of either or both A. quandang and L. exocarpi. are regarded as high- quality hosts due to nitrogen fixing abilities (Dean et al. 1994; Watson 2008). are also nitrogen fixers (Reddell et al. 1986) a that trait may explain L. exocarpi preference for three species in this family. Additionally, the open canopies of Acacias and Casuarinaceae allow high light to mistletoes, a necessary resource for these hemiparasites (Fonturbel et al. 2017; Sultan et al. 2018; Fonturbel et al. 2019).

In measuring nearest neighbours,Draft it was possible to identify those hosts that show high susceptibility and those that show high resistance to mistletoe infection. Host compatibility, which relates to how susceptible a host is to infection, plays a role in host preference. Specificity and preference develop as local adaptions to host availability (Blick et al. 2012). For example, seedling establishment is greater when seeds are from mistletoes infecting the same host species (Ramírez and Ornelas 2012) or growing in the same location (Reid and Stafford Smith 2000). Therefore, mistletoe survival rather than initial seed dispersal can explain aggregations (Fonturbel et al. 2017;

Fonturbel et al. 2019). The aggregation of A. quandang in a stand of pure A. homalophylla and

L. exocarpi using clonally-growing Casuarinaceae species highlights that low genetic diversity of hosts may increase host compatibility leading to host preference. These host-mistletoe interactions confirm that when hosts are abundant and predictable, parasites should show preference (Giorgi et al. 2004).

https://mc06.manuscriptcentral.com/botany-pubs Page 17 of 32 Botany

17

Although our data relate to only a subset of the geographic range of the three mistletoes

studied, we observed some noteworthy patterns regarding susceptibility of particular host species.

Of particular interest, there were some host species that are apparently resistant to infection. In this

study Callitris glaucophylla was often recorded as a nearest neighbour but not once as a host. As

a , tracheids are used for water transport rather than the wider xylem vessels of an

angiosperm, perhaps making the vascular system incompatible. Note that this species has been

previously recorded as a host, but most records relate to the genus-specific mistletoe Muellerina

bidwillii. Muellerina is an early branch of the Loranthaceae, endemic to Australia, with the recently

described Muellerina flexialabastra dependent on cunninghamii as primary host

suggesting that the association between these mistletoes and conifers may be a very old one (Watson 2019). Within the family Scrophulariaceae,Draft the genus Myoporum was resistant to infection but Eremophila species were not. This is an interesting observation as parasites often use

closely related hosts (Poulin and Mouillot 2005; Poulin et al. 2011) and Myoporum has been

recorded as a host of both A. quandang and L. exocarpi previously (Downey 1998). This finding

requires further work to evaluate the basis of differential susceptibility, highlighting the value of

unbiased measures of host preference to identify patterns and frame hypotheses to stimulate further

work.

Until now, the G index has only been applied to mutualistic associations (Taki and Kevan

2007), the data presented here representing the first application of Medan et al.’s (2006) approach

to quantify host specificity in a parasite. To inspire further work on other mistletoe assemblages

and host-parasite interactions more broadly, we suggest the following nominal thresholds. For

resultant G values less than 0.2, parasites are strict specialists; G values between 0.2 and 0.4 are

specialists, and G values greater than 0.4 are generalists. The use of Manly et al. (2002) preference

scores which includes proportional availability of potential hosts is a robust way to assess the

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 18 of 32

18 importance of a host to a parasite and identify species that are resistant to parasite infection. In addition to quantifying patterns of host specialisation and preference, our analyses revealed the importance of spatial scale as a moderating influence, patterns of preference changing as data from multiple locations, habitats and bioregions are incorporated.

Acknowledgements

We thank the following people for assistance with field work: Karen Avery, Adrian Boyd, Melinda

Cook, David Roach and Wendy Scott, and to Maggie Watson for logistical support during field work. We acknowledge the constructive input of two anonymous referees on an earlier version of this manuscript. Draft References

Barlow, B. 1984. Loranthaceae. In : Rhizophorales to Celastrales. Edited by A.

George. Bureau of Flora and Fauna, Publishing Service,

Canberra. pp. 68-131.

Blick, R., Burns, K., and Moles, A. 2012. Predicting network topology of mistletoe–host

interactions: do mistletoes really mimic their hosts? Oikos, 121(5): 761-771.

Chao, A., and Lee, S.-M. 1992. Estimating the number of classes via sample coverage. J. Am.

Stat. Assoc. 87(417): 210-217.

Chao, A., Chazdon, R.L., Colwell, R.K., and Shen, T.-J. 2005. A new statistical approach for

assessing similarity of species composition with incidence and abundance data. Ecol.

Lett. 8(2): 148-159. doi:10.1111/j.1461-0248.2004.00707.x.

https://mc06.manuscriptcentral.com/botany-pubs Page 19 of 32 Botany

19

Chao, A., Chazdon, R.L., Colwell, R.K., and Shen, T.-J. 2006. Abundance-Based Similarity

Indices and Their Estimation When There Are Unseen Species in Samples. Biometrics,

62(2): 361-371. doi:10.1111/j.1541-0420.2005.00489.x.

Colwell, R.K. 2013. EstimateS: Statistical estimation of species richeness and shared species

from samples. Department of Ecology & Evolution, University of Connecticut, USA.

Cunningham, G.M., Mulham, W., Milthorpe, P.L., and Leigh, J.H. 1992. Plants of western New

South Wales. Inkata Press, Melbourne.

Dean, W., Midgley, J., and Stock, W. 1994. The distribution of mistletoes in South Africa:

patterns of species richness and host choice. J. Biogeogr. 21: 503-510.

Downey, P.O. 1998. An inventory of host species for each aerial mistletoe species (Loranthaceae and Viscaceae) in Australia. Cunninghamia,Draft 5: 685-720. Fadini, R.F. 2011. Non-overlap of hosts used by three congeneric and sympatric loranthaceous

mistletoe species in an Amazonian savanna: host generalization to extreme specialization.

Acta Bot. Bras. 25(2): 337-345.

Fonturbel, F.E., Salazar, D.A., and Medel, R. 2017. Why mistletoes are more aggregated in

disturbed forests? The role of differential host mortality. For. Ecol. Manage. 394: 13-19.

doi:10.1016/j.foreco.2017.03.028.

Fonturbel, F.E., Bruford, M.W., Salazar, D.A., Cortes-Miranda, J., and Vega-Retter, C. 2019.

The hidden costs of living in a transformed habitat: Ecological and evolutionary

consequences in a tripartite mutualistic system with a keystone mistletoe. Sci. Total

Environ. 651(Pt 2): 2740-2748. doi:10.1016/j.scitotenv.2018.10.125.

Giorgi, M.S., Arlettaz, R., Guillaume, F., Nusslé, S., Ossola, C., Vogel, P., and Christe, P. 2004.

Causal mechanisms underlying host specificity in bat ectoparasites. Oecologia, 138(4):

648-654. doi:10.1007/s00442-003-1475-1.

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 20 of 32

20

Grenfell, M., and Burns, K.C. 2009. Sampling effects and host ranges in Australian mistletoes.

Biotropica, 41(6): 656-658.

Guégan, J.-F., and Kennedy, C. 1996. Parasite richness/sampling effort/host range: the fancy

three-piece jigsaw puzzle. Parasitology Today, 12(9): 367-369.

Jaenike, J. 1990. Host specialization in phytophagous insects. Annu. Rev. Ecol. Syst. 21: 243-

273.

Krasnov, B.R., Poulin, R., Shenbrot, G.I., Mouillot, D., and Khokhlova, I.S. 2005. Host

specificity and geographic range in haematophagous ectoparasites. Oikos, 108(3): 449-

456.

Krasnov, B.R., Mouillot, D., Shenbrot, G.I., Khokhlova, I.S., and Poulin, R. 2011. Beta- specificity: The turnover of hostDraft species in space and another way to measure host specificity. Int. J. Parasitol. 41(1): 33-41.

Manly, B., McDonald, L., Thomas, D., McDonald, T., and Erickson, W. 2002. Resource

selection by animals: statistical analysis and design for field studies. 2nd ed. Nordrecht,

The Netherlands: Kluwer. Kluwer Academic Publishers, Dordrecht.

Mans, B.J., de Klerk, D.G., Pienaar, R., and Latif, A.A. 2014. The host preferences of

Nuttalliella namaqua (Ixodoidea: Nuttalliellidae): a generalist approach to surviving

multiple host-switches. Exp. Appl. Acarol. 62(2): 233-240. doi:10.1007/s10493-013-

9737-z.

Mathiasen, R.L., Nickrent, D.L., Shaw, D.C., and Watson, D.M. 2008. Mistletoes: pathology,

systematics, ecology, and management. Plant Dis. 92(7): 988-1006. doi:10.1094/PDIS-

92-7-0988.

Medan, D., Basilio, A., Devoto, M., Bartoloni, N., Torretta, J., and Petanidou, T. 2006.

Measuring generalization and connectance in temperate, long-lasting systems. In Plant-

https://mc06.manuscriptcentral.com/botany-pubs Page 21 of 32 Botany

21

pollinator interactions. From specialization to generalization. Edited by N.M. Waser and

J. Ollerton. University of Chicago Press, Chicago, USA. pp. 245-259.

Messias, P.A., Vidal, J.D., Koch, I., and Christianini, A.V. 2014. Host specificity and

experimental assessment of the early establishment of the mistletoe Phoradendron

crassifolium (Pohl ex DC.) Eichler () in a fragment of Atlantic Forest in

southeast Brazil. Acta Bot. Bras. 28(4): 577-582. doi:10.1590/0102-33062014abb3523.

Milner, K.V. 2014. Optimising estimates of host spectrum: Australian mistletoe as a model

system. Honours Thesis, Faculty of Science, University of Technology Sydney, Sydney,

Australia.

Mourao, F.A., Pinheiro, R.B.P., Jacobi, C.M., and Figueira, J.E.C. 2016. Host preference of the hemiparasite Struthanthus flexicaulisDraft (Loranthaceae) in ironstone outcrop plant communities, southeast Brazil. Acta Bot. Bras. 30(1): 41-46. doi:10.1590/0102-

33062015abb0177.

Norton, D., and De Lange, P. 1999. Host specificity in parasitic mistletoes (Loranthaceae) in

New Zealand. Funct. Ecol. 13(4): 552-559.

Norton, D.A., and Carpenter, M.A. 1998. Mistletoes as parasites: Host specificity and speciation.

Trends Ecol. Evol. 13(3): 101-105.

Norton, D.A., Hobbs, R.J., and Atkins, L. 1995. Fragmentation, disturbance, and plant

distribution: mistletoes in woodland remnants in the Western Australian wheatbelt.

Conserv. Biol. 9(2): 426-438.

Okubamichael, D.Y., Griffiths, M.E., and Ward, D. 2011. Host specificity, nutrient and water

dynamics of the mistletoe Viscum rotundifolium and its potential host species in the

Kalahari of South Africa. J. Arid Environ. 75(10): 898-902.

doi:10.1016/j.jaridenv.2011.04.026.

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 22 of 32

22

Overton, J.M. 1994. Dispersal and infection in mistletoe metapopulations. J. Ecol. 82(4): 711-

723.

Poulin, R. 1992. Determinants of host-specificity in parasites of freshwater fishes. Int. J.

Parasitol. 22(6): 753-758. doi:10.1016/0020-7519(92)90124-4.

Poulin, R. 2011. The many roads to : A tale of convergence. Adv. Parasitol. 74: 1-40.

doi:10.1016/B978-0-12-385897-9.00001-X.

Poulin, R., and Mouillot, D. 2005. Combining phylogenetic and ecological information into a

new index of host specificity. J. Parasitol. 91(3): 511-514. doi:10.1645/GE-398R

Poulin, R., Krasnov, B.R., and Mouillot, D. 2011. Host specificity in phylogenetic and

geographic space. Trends Parasitol. 27(8): 355-361. doi:10.1016/j.pt.2011.05.003. Quirico, A.L. 1992. Loranthaceae. In FloraDraft of New South Wales. Edited by G.J. Harden. New South Wales University Press, Sydney. pp. 46-55.

Ramírez, M.M., and Ornelas, J.F. 2012. Cross-infection experiments of Psittacanthus

schiedeanus: Effects of host provenance, gut passage, and host fate on mistletoe seedling

survival. Plant Dis. 96(6): 780-787. doi:10.1094/pdis-06-11-0509.

Reddell, P., Bowen, G., and Robson, A. 1986. Nodulation of Casuarinaceae in relation to host

species and soil properties. Aust. J. Bot. 34(4): 435-444. doi:10.1071/BT9860435.

Reid, N. 1989. Dispersal of misteltoes by honeyeaters and flowerpeckers: Components of seed

dispersal quality. Ecology, 70(1): 137-145. doi:10.2307/1938420.

Reid, N., and Stafford Smith, D.M. 2000. Population dynamics of an arid zone mistletoe

(Amyema preissii, Loranthaceae) and its host Acacia victoriae (Mimosaceae). Aust. J.

Bot. 48(1): 45-58. doi:10.1071/bt97076.

https://mc06.manuscriptcentral.com/botany-pubs Page 23 of 32 Botany

23

Rödl, T., and Ward, D. 2002. Host recognition in a desert mistletoe: early stages of development

are influenced by substrate and host origin. Funct. Ecol. 16(1): 128-134.

doi:10.1046/j.0269-8463.2001.00592.x.

Sultan, A., Tate, J.A., de Lange, P.J., Glenny, D., Ladley, J.J., Heenan, P., and Robertson, A.W.

2018. Host range, host specificity, regional host preferences and genetic variability of

Korthalsella Tiegh. (Viscaceae) mistletoes in New Zealand. N.Z. J. Bot. 56(2): 127-162.

doi:10.1080/0028825x.2018.1464476.

Taki, H., and Kevan, P.G. 2007. Does habitat loss affect the communities of plants and insects

equally in plant–pollinator interactions? Preliminary findings [journal article]. Biodivers.

Conserv. 16(11): 3147-3161. doi:10.1007/s10531-007-9168-4. Takken, W., and Verhulst, N.O. 2013. HostDraft preferences of blood-feeding mosquitoes. Annu. Rev. Entomol. 58: 433-453. doi:10.1146/annurev-ento-120811-153618.

Tripet, F., Christe, P., and Møller, A.P. 2002. The importance of host spatial distribution for

parasite specialization and speciation: a comparative study of bird fleas (Siphonaptera:

Ceratophyllidae). J. Anim. Ecol. 71(5): 735-748.

Walther, B., and Morand, S. 1998. Comparative performance of species richness estimation

methods. Parasitol. 116(04): 395-405. doi:10.1017/s0031182097002230.

Walther, B., Cotgreave, P., Price, R., Gregory, R., and Clayton, D. 1995. Sampling effort and

parasite species richness. Parasitology Today, 11(8): 306-310. doi:10.1016/0169-

4758(95)80047-6.

Ward, M.J. 2005. Patterns of box mistletoe Amyema miquelii infection and pink gum Eucalyptus

fasciculosa condition in the , South Australia. Forest Ecol. Manage.

213(1): 1-14.

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 24 of 32

24

Watson, D.M. 2008. Determinants of parasitic plant distribution: the role of host quality. Botany,

87(1): 16-21.

Watson, D.M. 2019. Mistletoes of southern Australia. CSIRO Publishing, Collingwood, Victoria.

Watson, D.M., Milner, K.V., and Leigh, A. 2017. Novel application of species richness

estimators to predict the host range of parasites. Int. J. Parasitol. 47(1): 31-39.

doi:10.1016/j.ijpara.2016.10.001.

Yan, Z. 1990. Host specificity of Lysiana exocarpi subsp exocarpi and other mistletoes in

southern south-Australia. Aust. J. Bot. 38(5): 475-486.

Yan, Z. 1993. Seed dispersal of Amyema preissii and Lysiana exocarpi by mistletoe-birds and

spiny-cheeked honeyeaters. Emu, 93: 214-219. Draft

https://mc06.manuscriptcentral.com/botany-pubs Page 25 of 32 Botany

25

Table 1. Similarity matrix of β-specificity values of host use for Lysiana exocarpi. Pairwise

comparisons of bioregions based on Chao-Sørensen similarity index.

Broken Hill Mulga Riverina Cobar Sydney Complex Lands Peneplain Basin Broken Hill - Complex Mulga Lands 0.85 - Riverina 0.43 0.56 - Cobar Peneplain 0.31 0.15 0.42 - Sydney Basin 0.20 0.01 0.01 0 -

Draft

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 26 of 32

26

Table 2. Similarity matrix of β-specificity values for Amyema quandang. Pairwise comparisons of bioregions based on Chao-Sørensen similarity index.

Darling Riverine Broken Hill Cobar Peneplain Riverina Plains Complex Cobar Peneplain - Darling Riverine 0.90 - Plains Broken Hill 0.19 0 - Complex Riverina 0.17 0.93 0 -

Draft

https://mc06.manuscriptcentral.com/botany-pubs Page 27 of 32 Botany

27

Figure 1. Mean mistletoe specialisation/generalisation scores (G) in response to scale. (a) Lysiana

exocarpi (sample plot, n =25; habitat, n =12; bioregion, n =5; state-level, n =1); (b) Amyema

quandang (sample plot, n =13; habitat, n =6; bioregion, n =4; state-level, n =1). (c) Amyema

lucasii (plot, n =5; bioregion, n =2, habitat was nested in bioregion so only bioregion displayed;

state-level, n =1). Error bars show the standard error.

Figure 2. Preference scores (ω) for Lysiana exocarpi hosts pooled into habitat by bioregion.

Dashed line shows ω = 1, lower confidence interval >1 indicates host is significantly preferred.

Species denoted with * are significantly preferred hosts of L. exocarpi. (a) Broken Hill Complex

(b) Mulga Lands (c) Riverina (d) Cobar Peneplain (e) Sydney Basin. Note y axes have different

scales. Draft Figure 3. Preferred host species of Lysiana exocarpi over the state-level range of the study.

Dashed line shows ω = 1, lower confidence interval >1 indicates host is significantly preferred.

Species denoted with * are significantly preferred hosts of L. exocarpi.

Figure 4. Preference scores (ω) for Amyema quandang hosts pooled into habitat by bioregion.

Dashed line shows ω = 1, lower confidence interval >1 indicates host is significantly preferred.

Species denoted with * are significantly preferred hosts of A. quandang. Bioregions sampled are:

(a) Cobar Peneplain; (b) Riverina; (c) Broken Hill Complex; and (d) Darling Riverine Plains. Note

Y axes show different scales.

Figure 5. Preferred hosts of Amyema quandang over the state-level range of the study. Dashed

line shows ω = 1, lower confidence interval >1 indicates host is significantly preferred. Species

denoted with * are significantly preferred hosts of A. quandang.

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 28 of 32

Draft

Figure 1. Mean mistletoe specialisation/generalisation scores (G) in response to scale. (a) Lysiana exocarpi (sample plot, n = 25; habitat, n = 12; bioregion, n = 5; state-level, n = 1); (b) Amyema quandang (sample plot, n = 13; habitat, n = 6; bioregion, n = 4; state-level, n = 1). (c) Amyema lucasii (plot, n = 5; bioregion, n = 2, habitat was nested in bioregion so only bioregion displayed; state-level, n = 1). Error bars show the standard error.

185x159mm (300 x 300 DPI)

https://mc06.manuscriptcentral.com/botany-pubs Page 29 of 32 Botany

Draft

Figure 2. Preference scores (ω) for Lysiana exocarpi hosts pooled into habitat by bioregion. Dashed line shows ω = 1, lower confidence interval >1 indicates host is significantly preferred. Species denoted with * are significantly preferred hosts of L. exocarpi. (a) Broken Hill Complex (b) Mulga Lands (c) Riverina (d) Cobar Peneplain (e) Sydney Basin. Note y axes have different scales.

177x288mm (300 x 300 DPI)

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 30 of 32

Figure 3. Preferred host species of Lysiana exocarpi over the state-level range of the study. Dashed line shows ω = 1, lower confidence interval >1 Draftindicates host is significantly preferred. Species denoted with * are significantly preferred hosts of L. exocarpi.

177x101mm (300 x 300 DPI)

https://mc06.manuscriptcentral.com/botany-pubs Page 31 of 32 Botany

Draft

Figure 4. Preference scores (ω) for Amyema quandang hosts pooled into habitat by bioregion. Dashed line shows ω = 1, lower confidence interval >1 indicates host is significantly preferred. Species denoted with * are significantly preferred hosts of A. quandang. Bioregions sampled are: (a) Cobar Peneplain; (b) Riverina; (c) Broken Hill Complex; and (d) Darling Riverine Plains. Note Y axes show different scales.

177x204mm (300 x 300 DPI)

https://mc06.manuscriptcentral.com/botany-pubs Botany Page 32 of 32

Figure 5. Preferred hosts of Amyema quandang over the state-level range of the study. Dashed line shows ω = 1, lower confidence interval >1 indicatesDraft host is significantly preferred. Species denoted with * are significantly preferred hosts of A. quandang.

177x101mm (300 x 300 DPI)

https://mc06.manuscriptcentral.com/botany-pubs