The Conservation of Biodiverse and Threatened Dry Rainforest Plant Communities Is Vital in a Changing Climate

Abstract

Dry rainforest communities are globally threatened by anthropogenic pressures and climatic change but are less well researched and more poorly conserved than mesic rainforests. In response to the increasing loss of biodiversity, the Australian Government joined other international signatory parties to adopt the Kunming-Montreal Global Biodiversity Framework (GBF). The GBF emphasises the maintenance of connectivity and genetic diversity of whole ecosystems via landscape-scale conservation initiatives. Rainforest plant diversity, distinctiveness, and the current level of conservation of seasonal rainforest regional ecosystems of the Central Queensland Coast region in Australia were evaluated. Our three-marker DNA barcode dated phylogeny of rainforest plant taxa together with community species lists were used to calculate phylogenetic diversity (PD) estimates and species composition. Levels of rainforest ecosystem protection were assessed using Queensland government data. This study found selection pressures for moisture and geology significantly influence rainforest distribution and species diversity and evidence of a high degree of variability in terms of conservation. While some phylogenetically distinctive rainforest community types were well conserved, restricted or endangered communities were very poorly protected. Additionally, we found smaller dry rainforests in the Central Queensland Coast represent regional plant migration but are inadequately protected, highlighting the need for a revision of conservation objectives within the region.

1. Introduction

Dry rainforest types continue to be poorly understood and conserved when compared to mesic rainforests worldwide, and yet, they have been identified as some of the most biodiverse ecosystems [1,2,3,4,5]. They are considered to be less diverse than mesic rainforest ecosystems and often contain deciduous species [5,6]. Compared to wet tropical rainforest ecosystems, dry rainforests are characterised by less rainfall and greater seasonality and fire sensitivity [7]. Globally, anthropogenic disturbances like agriculture, grazing, and urbanisation have reportedly caused fragmentation and deterioration of dry rainforest types [3,6,8,9,10,11,12]. Climate change predictions of fluctuations in frequency of rainfall, increased severe weather occurrences, and rising global temperatures have been identified as major risks to these ecosystems [2,13,14,15].

It has been widely accepted that rainforests are distributed in a global latitudinal pattern of highly diverse wet tropical forests about the equatorial region and fewer mesic rainforests of decreasing diversity and complexity toward the poles [16,17,18]. In rainforest communities of China, Zu et al. (2023) found that species richness and phylogenetic diversity (PD) increased along a latitudinal gradient toward the equator and also longitudinally toward habitats of higher rainfall [19]. Similar patterns were observed in the dry tropical forests of Central America [9] and central Africa [6].

The discontinuous nature of rainforest distribution in eastern Australia has been attributed to historical climatic variations [20]. The dry tropical and dry subtropical rainforest types are mostly located in central Queensland between the northern Wet Tropics and subtropical region of Southeast Queensland. These dry rainforest communities have been noted to contain sclerophyllous species that are not typically associated with perceptions of “rainforest”, and other rainforest communities have been found to exist within the substrata or understory of more sclerophyllous vegetation [21]. Few studies have investigated these vulnerable communities, and our understanding of Central Queensland Coast rainforest types is lacking.

The distribution of species and the suitability of habitats can be affected by variations in geological characteristics, which can lead to variations in the spatial structure of vegetation [22,23,24]. The Interim Biogeographic Regionalization for Australia (IBRA) is a landscape-based method that categorizes Australia’s landscape into broad-scale bioregions, which are then subdivided into subregions based on landform similarities, finer-scale geology, and vegetation types. In Queensland, subregions are further classified into regional ecosystems (REs), identifying plant communities associated with specific landforms, soil types, or geology. The implementation of this framework, which provides a methodical way to describe biodiversity across a range of habitats, provides a means to identify gaps in the network of protected areas to ultimately enhance the reserve system [25,26]. The RE system mapping has led to a fine-scale mapping of the variation in rainforest ecosystem types within Queensland and permits the calculation of the extent of remnant vegetation [25,27].

Distinctive, highly biodiverse rainforest communities with rare and endemic species have been identified worldwide; several of these areas have been considered “biodiversity hotspots” and, as such, have become a priority for conservation initiatives [18,28,29,30]. However, biodiversity assessments based solely at the species level may be restricted by limited taxonomic knowledge in poorly known areas, such as tropical and subtropical rainforests, and may not adequately capture genetic variation within plant communities [31,32]. To improve estimates of biodiversity, DNA barcoding has been documented as a rapid, standardised tool for the measurement of genetic diversity between organisms and for species identification [32,33,34]. Phylogenetic diversity [35] has been identified as a means to ensure greater biodiversity for the conservation of distinctive rainforest communities [36,37]. DNA barcode-based phylogenetic analysis has been shown to provide rigorous investigations of rainforest communities that have revealed phylogenetically distinctive patterns of vegetation within Australia [38,39,40,41] and elsewhere [42,43,44,45,46].

It was hypothesised that rainforest communities that are found to be phylogenetically even, containing taxa evenly dispersed across a phylogeny, may be examples of stable climatic refugia that contain more species that are dispersal limited, endemic, and longer lived than random [47,48,49]. Phylogenetic clustering was hypothesised to be the result of factors such as speciation, competition, tolerance of abiotic characteristics, or recolonisation of disturbed areas from more stable communities [48,50]. Patterns of both phylogenetically even and clustered rainforest communities were found in southeast Queensland [38,40,51] and the Wet Tropics in northern Queensland [44,50].

A comparative study of RE rainforest communities in Southeast Queensland revealed that both distinctive and diverse RE types are required to enhance the conservation of biodiversity [52]. Das et al. (2019) identified the Central Mackay Coast bioregion as containing refugial rainforest communities and a putative centre for future climatic refugia but was based on a limited number of species [53]. To better conserve maximum biodiversity within the Central Queensland Coast region, it is essential to discover the most distinctive and diverse rainforest RE types in this understudied area.

The Australian Government, together with other international signatory parties, adopted the Kunming-Montreal Global Biodiversity Framework (GBF) [54]. The GBF established 23 new action-focused global targets to protect and conserve 30% of both land and oceans by 2030 [55]. Of particular importance to the conservation of the rainforest ecosystems in the Central Queensland Coast region are Targets 1–8 and 10, which aim to reduce threats to biodiversity by the effective management of ecosystems and taxa [55].

The conservation of biologically diverse ecosystems requires a targeted approach for the movement of taxa to be able to track anticipated climatic changes [56,57,58]. Central Queensland Coast drier rainforest communities have been considered bridging ecosystems between the moist Wet Tropics of north Queensland and subtropical rainforests of Southeast Queensland that contain taxa composed of lineages from both the continental Australian Sahul shelf (Sahul) and southeast Asian Sunda shelf (Sunda) [41,59,60]. The loss of dry rainforest patches was hypothesised to impact the conservation and connectivity of adjacent more mesic rainforest types, to the detriment of dispersal-limited plant taxa, and limit the range and resources for faunal species [61,62]. Imbach et al. (2013) argued that it is critical to improve the connectivity of protected areas in low altitudinal hot, dry regions as most are likely to be impacted by climate change [63]. In addition to the information afforded by the RE mapping data of rainforest habitats, the study of vegetation communities for targeted biodiversity conservation requires estimates of the diversity and distinctiveness of taxa of different rainforest types [40,52,64,65,66].

Shapcott et al. (2017) found that, in southeast Queensland, not all RE types are well conserved and that highly distinctive communities remain vulnerable to biodiversity loss [52]. An analysis of rainforest conservation at a broader subregion scale within the central Queensland Coast demonstrated inconsistent levels of protection [41]. In order to target the most diverse and distinctive rainforest types, an investigation into the existing conservation of finer-scale RE rainforest types of the Central Queensland Coast, within each subregion, is imperative for the effective connectivity of the most diverse and distinctive ecosystems to benefit both flora and fauna.

Dry rainforests plant species are considered to exhibit a robustness due to selection pressures of seasonality, reduced rainfall, and edge effects associated with fragmentation [61]. However, many dry rainforest communities are considered to be less diverse than more mesic rainforest types, such as littoral rainforests found along the Central Queensland Coast [66], and some are listed as an endangered ecological community under the Commonwealth Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act) [67]; yet, diversity and phylogenetic distinctiveness of the dry coastal rainforest habitats of the Central Queensland Coast region have not been comprehensively evaluated.

This study focused on dry and moist tropical and subtropical RE rainforest types of the Central Queensland Coast region of Queensland, Australia, to assess the diversity, distinctiveness, and current level of conservation of this unique region by addressing the following questions:

Are some rainforest RE types more diverse or distinctive than others within the Central Queensland Coast?

Are there some RE types that are potential refugial areas within the Central Queensland Coast region?

What environmental factors influence the diversity and distinctiveness of the Central Queensland Coast rainforest REs?

Are the rainforest RE types of the Central Queensland Coast region adequately protected within Australia’s National Reserve System?

2. Materials and Methods

2.1. Central Queensland Coast Study Area

The study area was comprised of the Central Mackay Coast bioregion and sections of neighbouring bioregions and extended from Rockhampton in the south to Paluma in the north (Figure 1). This region experiences highly variable moisture gradients (https://www.worldclim.org/data/bioclim.html, accessed on 16 March 2021) [68] and is typified by highly heterogeneous landscapes due to complex geological and topographical formations [69]. It has been considered that intensive geologic activity during the early Cretaceous (120–100 Ma) to the early Tertiary (66 Ma) produced a mosaic of geologic patterns, including sections of younger, recently exposed fertile soils as well as older, highly weathered, very poor nutrition soil [70].

2.2. Development of Regional Ecosystem Species Datasets

To compile species lists for each RE type, a set of RE types within the central Queensland study area, classified as rainforest ecosystems or containing rainforest elements, such as understory or well-developed vine thickets, was first compiled by utilising the Queensland Herbarium databases and RE mapping (Version 11.0) [71]. Because some locations were sampled less extensively than others, two forms of species dataset were collated to obtain a more comprehensive species representation for each RE type. The first was compiled from 201 plots from the Queensland Herbarium database and 297 fixed-area plots (0.1 ha; W.J.F. McDonald, personal data) (CORVEG, currently available from https://www.data.qld.gov.au/dataset/?tags=QBEIS; accessed on 9 March 2019) where RE type was assigned.

Secondly, a supplementary species list of rainforest plants was compiled from the Queensland Herbarium (BRI) species occurrence dataset (Herbrecs, currently available from https://www.gbif.org/; accessed on 22 February 2018) following the methods of Shapcott et al. (2017) [52]. Only the most recent georeferenced records, sampled and registered by Queensland Herbarium (BRI) botanists, were used, following the methods of Howard et al. (2023) and consistent with Shapcott et al. (2017) [41,52]. Second, using the techniques of Shapcott et al. (2017), a supplementary species list of rainforest plants was created from the Queensland Herbarium (BRI) species occurrence dataset (Herbrecs, currently available from https://www.gbif.org/; accessed on 22 February 2018) [52]. Consistent with the methods of Howard et al. (2023) and Shapcott et al. (2017), only the most current georeferenced records that were sampled and registered by Queensland Herbarium (BRI) botanists were used [41,52].

The RE type for these records was assigned using GPS coordinates overlaid with RE mapping data (QSpatial, 2019; https://qldspatial.information.qld.gov.au/catalogue/custom/detail.page?fid={01972496-CD6D-4314-B0C0-DA0E0421FB0A; accessed on 16 April 2020) in ArcGIS v10.7.1 [72]. For instances where Herbrecs collection records did not align with a rainforest RE polygon in the mapping data, the Queensland Herbarium (BRI) botanist’s location description and the suite of species recorded at the site were used to determine the RE type.

To identify endemic, rare, or threatened rainforest species of the central Queensland region under the Nature Conservation Act 1992 (NCA), we used the publicly available “Census of the Queensland Flora” (https://www.data.qld.gov.au/dataset/census-of-the-queensland-flora-2021; accessed on 10 February 2022). In addition, rare or threatened floral species, according to federal legislation of the Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act), were collated from the publicly available dataset for the Central Mackay Coast IBRA bioregion (https://wetlandinfo.des.qld.gov.au/wetlands/facts-maps/ibra-bioregion-central-mackay-coast-cmc/; accessed on 20 February 2018) and the EPBC Act (https://www.environment.gov.au/cgi-bin/sprat/public/publicthreatenedlist.pl?wanted=flora; accessed on 21 October 2022). The conservation status of RE types, as listed by the federal EPBC Act List of Threatened Ecological Communities (https://www.environment.gov.au/cgi-bin/sprat/public/publiclookupcommunities.pl/publicshowcommunity.pl?id=3&status=Endangered; accessed on 8 November 2022), was used for consistency with IUCN threatened ecosystem categories [72].

2.3. Development of Phylogenetic Analysis

We then used the combined Central and Southeast Queensland phylogeny (CSEQ) constructed from the Queensland DNA barcoded reference library developed by Howard et al. (2023) [41] of a standard three-marker barcode of rbcL, matK, and psbA–trnH for each species (data available at: https://doi.org/10.25907/00879 accessed on 13 August 2023). The Interactive Tree of Life online tool (iTOL v5; https://itol.embl.de/, accessed 17 October 2021) was used to produce a phylogenetic tree to discern phylogenetic patterns within and among Regional Ecosystem rainforest communities [73]. To determine if some rainforest RE communities were more or less distinctive or diverse, the dated phylogeny of 1629 rainforest plant species (excluding epiphytic orchids and ferns) and a species file for each RE type were used to calculate PD metrics, as described by Howard et al. (2023) [41].

Additionally, the variation in species composition within and between RE communities across distance was investigated using geographic dissimilarity matrices calculated from species presence/absence data. Phylogenetic diversity among REs was calculated using unweighted Unifrac [74]. Bray–Curtis dissimilarity matrices were calculated to quantify the dissimilarity for variables of NRI significance, the 15 highest or lowest PD scores, sclerophyllous content, rainfall and elevation categories, and rainforest types using the species composition data of ecosystems’ communities in the vegan: Community Ecology Package v2.5-7 (https://cran.r-project.org/src/contrib/Archive/vegan/; accessed on 28 August 2020) with a maximum of 20 random starts, in RStudio [75,76]. Non-metric multidimensional scaling (NMDS) was then used to evaluate potential patterns of relatedness quantified by the Bray–Curtis dissimilarity matrices. The phylogenetic distance between species and the community’s phylogenetic structure were both measured by the net relatedness index (NRI) [47]. Even dispersion was indicated by a positive NRI, whereas clustering, or more related than by chance, was shown by a negative NRI [47]. Multivariate correlations for RE communities were conducted using pairwise dissimilarity matrices with a Mantel test, in the Ape package v5.5 [77], between species composition and phylogenetic distance in RStudio [75,76].

2.4. Mapping of Regional Ecosystem Communities

Patterns of phylogenetic distinctiveness of RE rainforest communities were examined using the distribution data of Queensland rainforests, downloaded from the Qspatial database (2019; https://qldspatial.information.qld.gov.au/catalogue/custom/detail.page?fid={01972496-CD6D-4314-B0C0-DA0E0421FB0A}; accessed on 16 April 2020) in ArcGIS v10.7.1 [72]. The distribution of phylogenetically even and clustered (net relatedness index: NRI) and the ten highest and ten lowest PD scores for RE communities were mapped and coloured. A visual comparison of diversity and distinctiveness in terms of harsh environmental conditions was afforded by mapping the approximate locations of the Burdekin and St. Lawrence dry climatic barriers, of the Queensland Coast in ArcGIS v10.7.1 [72].

2.5. Environmental Data

To test for potential impacts on the diversity of Central Queensland Coast rainforest Res, abiotic data, such as annual rainfall, were collated from a number of sources, as follows. Elevation data were compiled from plot data and georeferenced location data of plots were used in Google Earth Pro (2021) where needed to fill gaps in the data. RE types were allocated to elevation categories (m): Lowland = 1–150, Midland = 151–300, Upland = 301–1160 (Table S1). Climate data were downloaded from the WorldClim v2 database at a spatial resolution of 30 s (~1 km²) (https://www.worldclim.org/data/bioclim.html, accessed on 16 March 2021). Due to the wide variation in rainfall data, REs were categorised based on rainfall to compare mean annual rainfall (mm): Low = 801–1000, Low-Medium = 1001–1200, Medium = 1201–1400, Medium-High = 1401–1600, High = 1601–2000, Very-High = 2001–3060 (Table S2). Spearman’s rank correlations were conducted in RStudio [76,78] to explore relationships between mean annual rainfall and species richness, family richness, PD, number of endemic rainforest species present within a RE, rainforest type, and the total pre-clearing extent (Ha) of rainforest REs within protected areas (P–C total PA).

A one-way ANOVA calculated in RStudio was used to test for significant differences among PD, family, and genus richness of RE types and elevation category. Non-parametric Kruskal–Wallis tests with the Dunn–Bonferroni post hoc method were performed to test for significant differences among elevation category and a range of variables, including the number of central Queensland endemic rainforest species, the number of NCA threatened species, and species richness, and also between mean elevation and the presence/absence of sclerophyllous vegetation for each RE of the study area in RStudio [75,76]. One-way ANOVA was also conducted in RStudio to test for significant differences among mean elevation and rainforest type, rainfall category, and NRI significance (random, even, or clustered).

To determine the geology of RE types, ‘Detailed Surface Geological Data’, obtained from Queensland Spatial Catalogue—Qspatial (2018; https://www.business.qld.gov.au/industries/mining-energy-water/resources/geoscience-information/gsq; accessed on 22 October 2020), was overlaid with RE mapping data in ArcGIS v10.7.1 [72]. Because of the geological complexity and distribution of RE types across multiple formations, communities were grouped by four categories according to the geological process by which they were formed: (Metamorphic (M), Igneous (I), Sedimentary (S), and Volcanic (V);

Table A1. Groupings of REs into dominant rock types used to test for possible geological factors that may influence the diversity and distinctiveness of Central Queensland Coast rainforest REs. Shown are the major dominant rock groupings, the rock types that were categorized within each group, and the characteristics that justify each of the groupings.

Dominant Rock Groupings	Rock Types	Characteristics
Metamorphic (M)	Metamorphic	altered
Igneous (I)	Granitoid	intrusive
	Mafities
	Felsites
	Gabbroid
Sedimentary (S)	Sand	water/wind movement
	Colluvium
	Alluvium
	Sediments
Volcanic (V)	Ultramafic	extrusive
	Basalt
	Volcanic–Sedimentary
	Volcanics

). RE types were also grouped by mean age of the geological period of dominant rock types on which they were located, from the oldest to youngest: late Neoproterozoic Era (mean age 685 Ma) to Quaternary Period (mean age 0.8 Ma), respectively. One-way ANOVA and non-parametric Kruskal–Wallis with the Dunn’s post hoc test were used to determine if there were statistically significant relationships among RE communities with different dominant rock types, mean geological age and PD, PD metrics, species, and family richness in RStudio [76,78].

2.6. Analysis of Central Queensland Coast Rainforest Protected Area Estate

The publicly available Queensland government analysis of REs by subregion dataset (Queensland Government 2022; https://www.publications.qld.gov.au/dataset/; accessed on 7 July 2021) was used to determine the degree of protection of the Central Queensland Coast rainforest estate. For each rainforest RE, data were extracted to calculate the sum of the total pre-clearing area of extent (P–C Total), total pre-clearing area of extent in protected areas (P–C Total PA), the total remnant vegetation area of extent (REM Total), and the total area of the remnant vegetation located within protected areas (REM Total PA) in Excel v.2408. Then, for each RE, the percents of pre-clearing (P–C % PA) and remnant rainforest (REM% PA) within a protected area were calculated. To determine if the degree of protectedness was significantly greater or less than average area protected within the whole region, a z-test was applied to all pre-clearing (P–C Total; P–C Total PA) and remnant rainforest vegetation categories (REM Total; REM Total PA) in Excel v.2408.

3. Results

3.1. Regional Ecosystem Diversity

We identified 65 REs containing rainforest. There were nine near-basal lineages represented in the research area: Eupomatiaceae (one species), Hernandiaceae (one species), Piperaceae (seven species), and Lauraceae (40 species from 7 genera) (data available at: https://doi.org/10.25907/00879 accessed on 13 August 2023). A summary of rainforest RE types was collated from the Regional Ecosystem Description Database (REDD), version 11.1 [21] (Table S2). Some rainforest types, such as “Evergreen notophyll feather palm vine forest” (RE 8.3.1b), could be characterised as “typical wet rainforests” in areas with high mean annual rainfall (mm) and a variety of plant forms. Others, like “Microphyll vine forest (“beach scrub”)” (RE 11.2.3), were drier and had a simpler structure in harsher environments (

Table 1. Rainforest RE types found in the Central Queensland Coast, the study area. Adapted from Neldner et al. (2019) [79]. “Rainforest Type” is a category used to group REs for analyses. Location category: Island, Coastal, or Inland location of RE types. RE ID (e.g., 11.2.3): where 11 is bioregion, 2 is land zone, and 3 is vegetation community.

Rainforest Type	RE ID	Elevation Category	Location Category
Evergreen Rainforest	8.2.5; 8.3.1b	Lowland	Coastal
	8.12.3c	Midland	Coastal
	8.12.3a	Midland	Island; Coastal; Inland
	8.12.1b; 8.12.30	Upland	Coastal
	8.12.17b; 8.12.2	Upland	Coastal; Inland
	8.12.17a; 8.12.1a; 8.8.1a; 8.8.1b	Upland	Inland
	8.12.17c	Upland	Island
Rainforest Patches	11.12.9; 8.12.13; 8.12.13a	Lowland	Inland
	11.12.16; 8.12.29; 8.12.29b	Lowland	Island; Coastal
	8.12.29a	Midland	Island
Rainforest Understory	8.2.6a	Lowland	Coastal
	7.3.16b; 7.3.26a	Lowland	Inland
	8.12.14a; 8.12.14b	Lowland	Island
	8.12.26	Lowland	Island; Coastal
	8.2.6b	Midland	Coastal; Inland
	7.12.21b; 7.12.22b; 7.12.26b	Upland	Inland
Semi-Deciduous Rainforest	11.3.40; 8.3.9	Lowland	Coastal
	11.3.11; 8.3.1a	Lowland	Coastal; Inland
	8.12.19	Midland	Island; Coastal; Inland
Semi-Evergreen Rainforest	11.3.11x1; 8.11.2x1a; 8.12.11c	Lowland	Coastal
	11.11.21; 11.11.5a; 11.12.4a; 8.11.2	Lowland	Coastal; Inland
	8.10.1	Lowland	Island
	11.2.3; 8.12.11; 8.12.11a; 8.12.28; 8.2.2; 8.3.10	Lowland	Island; Coastal
	8.12.18	Lowland	Island; Coastal; Inland
	11.5.15	Midland	Coastal
	11.12.4	Midland	Island; Coastal; Inland
	8.12.16	Upland	Coastal; Inland
	11.11.5; 11.4.1; 11.8.3; 8.12.3b; 9.12.34	Upland	Inland
Wet Tropics Vine Forest	7.12.11b; 7.12.1a; 7.3.50b;	Lowland	Inland
	7.12.48	Midland	Inland
	7.12.10a; 7.12.11a; 7.12.16b	Upland	Inland

).

Four RE types were significantly (p < 0.05) higher in diversity (species richness, genus richness, family richness, PD) compared to all other RE types (

Table 2. Summary of PD measures for species composition data by RE within the Central Queensland Coast region. Where PD = phylogenetic diversity; SR = species richness; GR = genus richness; FR = family richness; MPD = mean pairwise distance; MNTD = mean nearest taxon index; NRI = net relatedness index; NTI = nearest taxon index; CR = critically endangered; EN = endangered; EPBCAct = the Federal Environment Protection and Biodiversity Conservation Act (1999); NCAct = the number of rainforest species listed under the Queensland Nature Conservation Act 1992; CQC Endemics = the number of Central Queensland Coast endemic rainforest species present within an RE type; Gymno = the number of gymnosperm species recorded from RE types. ^H is higher; ^L is lower; ^E indicates even; ^C indicates clustered; (*) indicates significant values (p < 0.05). RE ID (e.g., 11.2.3) where 11 is bioregion, 2 is land zone, and 3 is vegetation community.

RE ID	SR	GR	FR	PD	MPD	NRI	MNTD	NTI	Threatened			Endemic	Gymno
									(EPBCAct)		(NCAct)	CQC
									Ecosys	Spp.	Spp.	Spp.
11.11.21	173	140	61	6968	195.1	*C 2.95	53.7	−0.65	(Rare)		2	4	0
11.11.5	134	115	54	5988	200.2	1.51	59.9	−0.49				3	1
11.11.5a	48	46	29	2902	191.9	*C 1.75	86.6	0.09				1	0
11.12.16	149	125	59	6019	191.8	*C 3.38	49.6	1.22	CR		1	7	0
11.12.4	*H 394	270	97	11,667	204	*C 1.73	36.9	−1.01		1	3	10	1
11.12.4a	188	144	72	7506	212.7	−0.95	48.5	0.31			1	5	2
11.12.9	229	179	71	8550	205.8	0.69	45.9	−0.21		1	1	8	1
11.2.3	219	167	65	7594	195	*C 3.22	42	1.31	EN		1	5	0
11.3.11	140	114	48	5553	194.6	*C 2.67	48.4	*C 1.80	EN			4	0
11.3.11x1	38	36	31	2769	203.2	0.48	107.6	−1.02				1	0
11.3.40	28	27	20	2082	209.4	−0.05	103.1	0.2				0	0
11.4.1	4	4	*L 3	433	142.7	*C 1.82	88	*C 2.23	EN			0	0
11.5.15	104	91	44	4576	184.8	*C 3.69	52	*C 2.09	EN		1	3	0
11.8.3	35	33	25	2264	197.5	1.02	90.5	0.62	EN			1	0
7.12.10a	82	75	42	4271	210.7	−0.35	67.2	0.53				2	2
7.12.11a	120	101	52	5565	212.3	−0.68	52.2	1.62	EN	1	2	4	2
7.12.11b	152	130	64	6464	218.4	*E −2.12	53.6	0.22	EN		1	4	2
7.12.16b	259	163	79	8101	215.1	*E −1.88	34	*C 3.03		2	2	5	2
7.12.1a	7	8	8	832	221.2	−0.43	169.9	−0.33	EN			0	0
7.12.21b	61	53	40	3347	197.6	1.23	70.8	1.07		2	1	0	0
7.12.22b	51	44	29	2921	207.4	0.08	75	1.1				1	0
7.12.26b	64	49	33	3222	224.5	*E −1.99	57.7	*C 2.37				1	0
7.12.48	9	9	9	1033	216.4	−0.35	182	−1.17				0	0
7.3.16b	50	43	29	3108	231.3	*E −2.48	82.4	0.41			1	1	1
7.3.26a	37	35	24	2520	220.9	−1.13	89.2	0.6				1	0
7.3.50b	48	43	31	3101	233.9	*E −2.70	83.4	0.4	EN		1	1	1
8.10.1	34	31	25	2363	191.4	*C 1.47	105	−0.45			1	1	0
8.11.2	202	163	71	7656	210.1	−0.35	48	−0.03		2	2	12	2
8.11.2x1a	136	120	65	6237	200.4	*C 1.64	62.7	−1.14			1	1	0
8.12.11	190	150	68	7362	205.7	0.64	47.3	0.54	CR	1	3	7	1
8.12.11a	373	*H 269	92	11,538	203.3	*C 1.81	38.2	−1.14	CR	4	4	21	2
8.12.11c	159	128	63	6414	215	−1.36	44.8	*C 2.06	CR	1	2	5	1
8.12.13	136	112	55	5880	200.5	1.53	57.8	−0.08			1	2	1
8.12.13a	119	109	56	5428	193.9	*C 2.47	56.9	0.73		1	1	2	0
8.12.14a	101	90	45	4796	199.9	1.38	59.4	0.95				4	1
8.12.14b	79	76	43	4077	197.8	1.58	66.1	0.8			1	5	0
8.12.16	73	66	41	4087	186.6	*C 2.80	79.8	−0.64	CR			2	0
8.12.17a	80	66	43	3970	220.9	−1.7	61.7	1.27			3	3	0
8.12.17b	95	65	40	3922	219.5	*E −1.72	46.3	*C 3.20			3	5	0
8.12.17c	34	32	25	2315	226.4	−1.54	89.5	0.79			1	3	0
8.12.18	*H 483	*H 303	98	*H 13,009	205.7	1.3	32.8	−0.65		7	7	25	2
8.12.19	365	238	85	10,087	206.4	0.67	31.1	*C 2.65		4	4	19	0
8.12.1a	253	179	79	8429	221.1	*E −3.43	38.3	*C 1.68			4	8	1
8.12.1b	191	143	67	6711	209.7	−0.3	42.7	*C 1.83			1	9	0
8.12.2	*H 404	271	*H 104	11,442	211.8	−1.26	31.6	*C 1.69		2		11	1
8.12.26	75	71	42	4076	214.5	−0.82	69.6	0.48				2	1
8.12.28	161	130	67	6777	201.9	1.3	53.6	−0.08			2	12	1
8.12.29	127	109	51	5766	208.1	0.13	59.2	−0.07			2	5	1
8.12.29a	55	52	36	3266	198.8	1.09	81.1	0.22				2	1
8.12.29b	117	99	56	5451	193.9	*C2.45	61.9	−0.1			1	2	0
8.12.30	95	78	45	4678	217.4	−1.34	60.9	0.9		2	3	9	0
8.12.3a	*H 479	*H 301	*H 102	*H 12,469	205.8	1.24	30	1.28		6	7	18	2
8.12.3b	35	32	26	2391	205.8	0.23	93	0.41				0	0
8.12.3c	245	187	76	8997	219.2	*E −2.88	43.2	0.29			4	8	1
8.2.2	354	256	89	10,933	208.2	0.15	38.2	−0.66	CR		2	13	2
8.2.5	84	72	47	4610	220.5	*E −1.75	71.4	−0.14			2	3	0
8.2.6a	68	65	36	3673	197.8	1.38	72.5	0.53				1	0
8.2.6b	48	42	29	2857	211.9	−0.34	79.5	0.77				0	0
8.3.10	338	237	87	10,497	204	1.58	37.8	−0.04		4	6	20	1
8.3.1a	272	207	82	9253	204.2	1.25	40.6	0.41		1	6	19	1
8.3.1b	255	187	84	9206	218.5	*E −2.74	42.6	0.17		1	4	4	1
8.3.9	221	166	73	7830	206.1	0.55	43	0.89			1	11	0
8.8.1a	120	90	52	4994	222.1	*E −2.27	48.7	*C 2.23			2	4	1
8.8.1b	156	125	61	6695	214	−1.11	55.7	−0.43			3	3	0
9.12.34	150	130	63	6573	208	0.24	55.8	−0.19			6	6	0
Mean (Stdv)				5786 (2996)	207 (13)	0.23 (1.71)	63.2 (27.9)	0.57 (1.03)
CQC Total	996	525	141	19,916							27	42	3

). The RE with the highest diversity measures was “Semi-evergreen notophyll/microphyll to complex notophyll vine forest” (RE 8.12.18) with significantly (p < 0.05) higher scores than expected in terms of PD (13,009) and species richness (483: Table 2). RE type “Evergreen to semi-evergreen, notophyll to microphyll, vine forest” (RE 8.12.3a) was significantly (p < 0.05) higher in all diversity measures: species richness (479), genus richness (301), family richness (102), and PD (12,504; Table 2). These RE types were mostly located at low to mid elevations in protected areas with small sections found throughout the Central Mackay Coast bioregion (Figure 1 and Figure 2).

The RE with the highest significant (p < 0.05) family richness (104) was upland “Evergreen notophyll to complex notophyll vine forest” (RE 8.12.2), which also had a high PD (11,442) and significantly (p < 0.05) higher than expected species richness (404; Table 2). The RE that scored the lowest PD (433) with only 123 Ha of remnant vegetation remaining was “Semi-evergreen vine thicket ± Casuarina cristata on Cainozoic clay plains” (RE 11.4.1; Table 2; Figure 3). Interestingly, some of the least diverse RE types with lower PD and fewer families were located in the more tropical north including “Complex notophyll vine forests” (RE 7.12.11a; PD = 5565; SR = 120; FR = 52) (Table 2; Figure 3). These results were inconsistent with a north-to-south diversity gradient between tropical wet and subtropical rainforest types (Figure 3). As expected, species richness of the RE types on continental islands were some of the lowest recorded within the study area and had low PD scores, such as “Evergreen microphyll fern forest” (RE 8.12.17c; PD = 2315) found on Whitsunday Island (Table 2; Figure 2). As expected, PD was strongly significantly correlated with species richness (rho = 0.9949186, p < 0.001) and genus richness (rho = 0.996503, p < 0.001) among RE types.

Sixty-nine percent of distinctive REs provided habitat for threatened species listed under the NCA (Table 2;

Table A2. Summary table of threatened species, as listed under the federal Environment Protection and Biodiversity Conservation Act (1999; EPBC) and the Nature Conservation Act (1992; NC Act) of Queensland, and endemic species found within the Central Queensland Coast study area. CR = critically endangered; EN = endangered; NT = near threatened; V = vulnerable; No. = number; RE ID = the regional ecosystem where the species was located; CEQ = central eastern Queensland; QLD = Queensland; AU = Australia.

Family	Botanical Name	NC Act (1992)	EPBC Act (1999)	Endemicity	No. of REs	RE ID
Acanthaceae	Graptophyllum excelsum	NT		QLD	4	8.12.3a; 9.12.34; 11.11.5; 11.5.15
Acanthaceae	Graptophyllum ilicifolium	V	V	QLD	3	8.11.2; 8.12.19; 8.12.3a
Annonaceae	Meiogyne heteropetala			NEQ; CEQ	17	11.12.4; 11.12.9; 11.2.3; 7.12.11a; 8.12.11; 8.12.11a; 8.12.11c; 8.12.18; 8.12.19; 8.12.2; 8.12.28; 8.12.3a; 8.12.3c; 8.2.2; 8.3.10; 8.3.1a; 9.12.34
Apocynaceae	Cerbera dumicola	NT		QLD	2	8.12.13; 11.5.15
Apocynaceae	Neisosperma kilneri *	V	V	CEQ	4	8.3.10; 8.12.18; 8.3.1a; 8.12.3a
Apocynaceae	Parsonsia larcomensis	V	V	QLD	3	8.11.2; 8.12.17b; 8.12.3c
Brownlowiaceae	Berrya rotundifolia	V		QLD	3	8.12.11a; 8.12.14b; 8.12.29
Capparaceae	Capparis ornans			NEQ; CEQ	7	11.2.3; 11.12.4; 8.12.16; 8.12.18; 11.11.21; 11.8.3; 11.12.16
Combretaceae	Macropteranthes fitzalanii			CEQ	10	8.12.18; 8.2.2; 8.12.19; 8.3.9; 8.3.10; 8.3.1a; 8.12.3a; 8.12.28; 8.12.11a; 8.12.11
Cycadaceae	Cycas media			CYP; NEQ; CEQ	12	11.12.9; 8.11.2; 8.12.11a; 8.12.14a; 8.12.18; 8.12.26; 8.12.29; 8.12.29a; 8.12.3a; 8.2.2; 8.8.1b; 9.12.34
Cyperaceae	Cyperus sp. Eungella NP (P.R.Sharpe 5052)			CEQ	1	8.12.1a
Elaeocarpaceae	Elaeocarpus largiflorens			NEQ; CEQ	10	8.12.1b; 8.12.19; 8.12.1a; 8.12.17b; 8.8.1a; 7.12.16b; 8.12.2; 8.12.18; 8.12.30; 8.12.3a
Euphorbiaceae	Croton magneticus	V		QLD	4	8.12.29b; 11.12.16; 11.12.4; 11.12.4a
Euphorbiaceae	Trigonostemon inopinatus	V		QLD	2	8.12.2; 8.12.3a
Hernandiaceae	Hernandia bivalvis	NT		QLD	2	8.12.18; 8.3.1a
Laxmanniaceae	Cordyline petiolaris *			CEQ	1	8.11.2
Leguminosae (Caesalpiniaceae)	Cassia sp. Paluma Range (G.Sankowsky+ 450)			NEQ	10	8.12.18; 8.3.10; 11.12.4a; 8.12.28; 7.12.11b; 9.12.34; 11.12.16; 11.12.4; 8.12.11a; 8.12.17c
Leguminosae (Fabaceae)	Hovea clavate *			CEQ	1	8.3.1b
Malvaceae	Brachychiton compactus	NT		QLD	7	8.12.11; 8.12.11a; 8.12.18; 8.12.28; 8.2.2; 8.3.10; 8.3.1a
Malvaceae	Corchorus hygrophilus	V		AU	2	11.12.4; 11.12.9
Myrsinaceae	Myrsine crassifolia			CEQ	8	11.2.3; 8.12.11c; 8.12.29; 8.12.17b; 8.12.1a; 8.12.2; 8.2.2; 8.2.5
Myrsinaceae	Myrsine ireneae subsp. ireneae *			NEQ; CEQ	1	8.12.2
Myrtaceae	Gossia pubiflora			CEQ	8	8.12.18; 8.12.19; 8.3.10; 8.2.2; 8.3.9; 8.3.1a; 8.12.11a; 8.12.28
Myrtaceae	Rhodamnia glabrescens	NT		QLD	4	8.3.10; 8.3.1a; 8.12.18; 8.12.19
Myrtaceae	Rhodamnia rubescens	CR	CR	AU	1	8.12.2
Myrtaceae	Ristantia gouldii	V	V	QLD	5	8.12.18; 8.12.19; 8.12.30; 8.3.10; 8.3.1a
Myrtaceae	Ristantia waterhousei	V		CEQ	2	8.12.30; 8.3.10
Phyllanthaceae	Actephila bella	V		QLD	1	8.12.3a
Phyllanthaceae	Actephila plicata			CEQ	2	8.12.18; 8.3.10
Phyllanthaceae	Cleistanthus dallachyanus			NEQ; CEQ	19	8.12.3a; 8.12.18; 8.12.19; 8.3.10; 8.3.1a; 11.12.4; 11.12.4a; 8.11.2; 8.12.11a; 8.2.2; 8.12.28; 7.12.11b; 11.12.9; 7.12.10a; 8.12.29b; 8.12.13; 8.12.14a; 8.12.14b; 11.12.16
Picrodendraceae	Dissiliaria indistincta			CEQ	9	8.3.9; 8.3.10; 8.12.18; 8.3.1a; 8.12.11a; 11.12.4; 8.12.14b; 8.12.30; 9.12.33
Polygalaceae	Comesperma oblongatum *	V		CEQ	1	8.12.11c
Rhamnaceae	Pomaderris clivicola	EN	V	AU	4	7.12.16b; 7.12.21b; 8.12.13a; 8.3.1b
Rubiaceae	Antirhea putaminosa			NEQ; CEQ	5	11.11.21; 11.3.11; 11.12.4; 11.11.5; 11.5.15
Rubiaceae	Atractocarpus fitzalanii			NEQ; CEQ	20	8.12.18; 8.12.3a; 8.12.1a; 8.3.10; 8.12.19; 8.3.1a; 8.11.2; 8.12.11; 8.12.1b; 9.12.34; 7.12.11b; 7.12.16b; 8.3.9; 8.12.2; 8.2.2; 7.12.10a; 7.3.26a; 11.12.9; 8.12.11a; 8.12.29a
Rubiaceae	Larsenaikia jardinei			CEQ	14	8.11.2; 8.12.11a; 8.12.18; 8.12.19; 8.12.1b; 8.12.28; 8.12.29; 8.12.3a; 8.2.2; 8.3.10; 8.3.1a; 8.3.9; 11.2.3; 11.3.11x1
Rubiaceae	Randia sp. Shute Harbour (D.A.Halford Q811)			CYP; NEQ; CEQ	8	7.12.11a; 8.12.18; 8.12.19; 8.12.2; 8.12.3a; 8.3.1a; 8.3.9; 8.11.2
Rutaceae	Acronychia eungellensis	NT		QLD	4	8.12.17a; 8.12.18; 8.12.19; 8.12.1a
Rutaceae	Medicosma obovata	V	V	QLD	6	8.12.11a; 8.12.18; 8.12.30; 8.3.10; 8.3.1a; 8.3.9
Rutaceae	Phebalium distans	EN	EN	QLD	3	7.12.11a; 7.12.16b; 7.12.21b
Salicaceae	Homalium sp. South Molle Island (J.A.GrestyAQ208995) *			CEQ	4	8.12.18; 8.3.10; 8.12.11a; 8.12.28
Salicaceae	Xylosma ovata	NT		AU	9	8.10.1; 8.11.2x1a; 8.12.11a; 8.12.11c; 8.12.13a; 8.12.26; 8.12.3c; 8.2.2; 11.2.3
Sapindaceae	Arytera dictyoneura	NT		QLD	5	7.12.11a; 7.12.11b; 8.12.18; 8.12.3a; 8.3.1a
Sapindaceae	Arytera sp. Dryander Creek (P.R.Sharpe 4184)			CEQ		8.12.18; 8.3.1a; 8.12.3a
Sapindaceae	Diploglottis obovata			CEQ	12	8.3.1a; 8.3.9; 8.12.19; 8.12.3a; 8.12.18; 8.3.10; 8.12.2; 8.11.2; 8.12.1b; 8.12.11c; 11.3.11; 8.8.1a
Sapindaceae	Lepiderema punctulata			NEQ; CEQ	1	8.12.18
Sapindaceae	Lepiderema sp. Impulse Creek (A.B.Pollock 73) *			CEQ	1	8.3.9
Sapindaceae	Sarcotoechia heterophylla	NT		CEQ	5	8.12.2; 8.12.1a; 8.12.19; 8.8.1a; 8.12.17a
Simaroubaceae	Samadera bidwillii *	V		CEQ	2	8.2.5; 11.11.21
Solanaceae	Solanum sporadotrichum *	NT		CEQ	2	8.12.28; 8.12.11
Sterculiaceae	Argyrodendron sp. Whitsundays (W.J.McDonald+ 5831)			CEQ	4	8.12.18; 8.12.19; 8.3.10; 8.12.17c
Zamiaceae	Bowenia serrulata			CEQ	3	8.12.3c; 8.3.1b; 8.11.2
Zamiaceae	Macrozamia serpentina *	EN		CEQ	1	11.11.21

(*) indicates georeferenced endemic species that were substituted with DNA barcoded congeneric species for the purposes of inclusion in the phylogenetic analysis when a specimen was not available for DNA extraction.

). The highest numbers (n = 7) of NCA-listed threatened species were found in mainland, coastal, and island environments of “Semi-evergreen notophyll/microphyll to complex notophyll vine forest” (RE 8.12.18) and “Evergreen to semi-evergreen, notophyll to microphyll, vine forest” (RE 8.12.3a; Table 1 and Table 2). Even though both REs were significantly (p < 0.05) higher in species and genus richness and PD than expected by chance, taxa were randomly distributed across the phylogeny (Table 2).

3.2. Regional Ecosystem Distinctiveness

Most RE types overlapped considerably in terms of species composition, but some were significantly (p < 0.05) different to others (Figure 4). Of the 65 RE communities, 62 ecosystem types tended to group together in terms of species composition, but three were outliers (Figure 4; Table 2). Two were listed as endangered (EPBC Act), “Mesophyll to notophyll vine forest” (RE 7.12.1a) and “Semi-evergreen vine thicket ± Casuarina cristata on Cainozoic clay plains” (RE 11.4.1). The third was “Wind-sheared notophyll vine forest on exposed granite/rhyolite, steep slopes” (RE 7.12.48). All three had low diversity measures (PD, SR, GR, FR; Table 2) but were not significantly correlated (p > 0.05) with mean annual rainfall (mm) (Table S3).

The majority of RE types contained taxa that were randomly distributed across the phylogeny. Of the 11 RE types found to be phylogenetically even (NRI, p < 0.05), 7 were moist rainforest types located in three main regions conspicuously separated by the Burdekin and St. Lawrence dry gaps (Figure 2; Table 2), for example, “Evergreen notophyll to complex notophyll vine forest” (RE 8.8.1a) located in Eungella National Park and Crediton State Forest; “Simple notophyll vine forest” in lowland (RE 7.12.11b), located north of Townsville, and “Evergreen notophyll vine forest” (RE 8.2.5) in lowland coastal areas in the southern part of the study area (Figure 2; Table 2).

The remaining five contained sclerophyllous species, such as Melaleuca fluviatilis in lowland “Low notophyll vine forest” (RE 7.3.50b) or “Tall open forest and woodland well-developed rainforest understory” (RE 7.12.26b) with dominant Syncarpia and Allocasuarina sclerophyllous species (Figure 4a,c; Table 1). Some RE types were comprised of significantly (p < 0.05) evenly dispersed (NRI) taxa at a higher taxonomic rank (order or family) but contained co-occurring species that were more closely related than expected due to chance (clustered; NTI, p < 0.05) (Table 2). These included the upland communities of “Simple notophyll vine forest” (RE 7.12.16b) in Paluma National Park and “Evergreen notophyll mossy closed forest” (RE 8.12.17b) in Eungella National Park, both of which contained Central Queensland Coast threatened and endemic rainforest plant species (Figure 2; Table 1, Table 2 and Table A2). These results may be indicative of sympatric speciation and resource partitioning.

As expected, rainforest RE types containing species more significantly related than random (NRI; p < 0.05) were found predominantly in harsher, drier environments. Examples included “Microphyll vine forest (“beach scrub”)” (RE 11.2.3) on coastal sand within the Burdekin and St. Lawrence dry gaps or on Ultramafic (serpentine) rocks such as “Semi-evergreen vine thicket on serpentinite” (RE 11.11.21) located inland, near Princhester Conservation Park (CP) (Figure 1 and Figure 3; Table 2). These communities provide habitat for threatened and endemic species (Table A2). Two REs that were significantly clustered (NRI; p < 0.05), “Mixed low woodland to shrubland with rainforest patches” (RE 11.12.16) and “Semi-evergreen microphyll vine thicket” (RE 8.12.11a), were located on islands or lowland coastal areas and were listed as critically endangered (CR) under the EPBC Act 1999. These contained threated (NC Act, 1992) and endemic species, such as Croton magneticus (RE 11.12.16) and Berrya rotundifolia (RE 8.12.11a) (Table 2 and Table A2).

3.3. Environmental Factors

3.3.1. Rainfall

Rainforest RE diversity, distinctiveness, and distribution were influenced by abiotic environmental factors. RE types that were significantly evenly dispersed (NRI; p < 0.05) were found in areas of medium–high mean annual rainfall (mm) while those that were significantly clustered (NRI, p < 0.05) were more frequently located in regions of low–medium mean rainfall (mm; Figure 4a,d). Phylogenetic diversity dissimilarity among RE types showed that groupings of “Semi-evergreen/deciduous Rainforest” ecosystems were found in mostly lowland areas, with low–medium rainfall (Figure 4d–f) and contained species more significantly (p < 0.05) closely related than expected by chance (NRI; Figure 4a). As expected, evergreen rainforest types were shown to be found in mostly upland, moist areas (Figure 4d–f). Most REs with the highest PD (Figure 4b) were mixed in terms of rainforest type, elevation, and mesic qualities (Figure 4b–f).

3.3.2. Geology

Dominant rock type strongly influenced patterns of RE diversity and distinctiveness. RE types grouped by igneous/sedimentary/volcanic (I/S/V) rock were significantly higher in terms of mean PD, species genus, and family richness than those grouped by sedimentary (S; p < 0.001) and volcanic (V; p < 0.05) rock types (

Table 3. Summary of REs grouped by dominant rock type based on species composition data. The mean values of Central Queensland Coast rainforest REs that differed significantly (p < 0.05) in terms of dominant rock groups are shown, where DomRock = dominant rock and Ma = million years before present. Diversity measures are SR = species richness, GR = family richness, FR = family richness, PD = phylogenetic diversity. Distinctive measures are NCAct = the Nature Conservation Act 1992 and No. Endemic spp. = number of Central Queensland Coast endemic rainforest species. Dominant Rock groups are I = igneous, I/M/S/V = igneous/metamorphic/sedimentary/volcanic, I/S = igneous/sedimentary, I/S/V = igneous/sedimentary/volcanic, I/V = igneous/volcanic, S = sedimentary, V = volcanic; (*) stipulates statistically significant difference in mean values in Kruskal–Wallis test between dominant rock type and SR, GR, FR, PD, “No. NCA listed”, and “No. Endemic spp.”. The results of Dunn’s post hoc tests between “Dominant Rock group” and diversity measures are indicated by superscript codes that are equivalent to dominant rock grouping codes.

RE ID	No. REs	DomRock Group	GeoAge (Ma) Mean (stdv)	SR	GR	FR	PD Mean (stdv)	REs NCAct	Endemic spp./Group	Rainfall Mean (stdv)
11.12.16; 7.12.11a; 7.12.48; 8.12.14a; 8.12.17a; 8.12.1a	6	I	291 (81)	501	321	110	4969 (2234)	4	18	1546 (500)
11.12.4; 8.12.3a; 8.2.2	3	* I/M/S/V	197 (13)	682 S	406	120 S	11,690 (627) S	8	28 S	1225 (193)
11.11.5; 11.12.4a; 11.3.11; 7.12.10a; 7.12.11b; 7.12.16b; 8.12.11c; 8.12.17b; 8.12.3c; 9.12.34	10	I/S	174 (43)	671	388	124	6379 (1498)	4	20	1293 (204)
8.11.2; 8.12.11a; 8.12.18; 8.12.19; 8.12.28; 8.3.10; 8.3.1a;	7	* I/S/V	151 (26)	673 S;V	395 S;V	116 S;V	9831 (1997) S;V	12 S;V	33 S;V	1641 (103)
8.12.2; 8.3.9; 8.8.1a	3	I/V	194 (13)	470	299	108	8089 (2639)	3	16	1594 (307)
11.11.5a; 11.12.9; 11.2.3; 11.3.11x1; 11.3.40; 11.5.15; 7.12.1a; 7.12.21b; 7.12.22b; 7.12.26b; 7.3.16b; 7.3.26a; 7.3.50b; 8.10.1; 8.12.16; 8.12.3b; 8.2.5; 8.2.6a; 8.2.6b; 8.3.1b	20	* S	29 (80)	661	397	123	3836 (2121)	2	16	1407 (481)
11.11.21; 11.4.1; 11.8.3; 8.11.2x1a; 8.12.11; 8.12.13; 8.12.13a; 8.12.14b; 8.12.17c; 8.12.1b; 8.12.26; 8.12.29; 8.12.29a; 8.12.29b; 8.12.30; 8.8.1b	16	V	162 (170)	630	387	120	4850 (1913)	9	28	1555 (549)

). Regional ecosystem types grouped by igneous/metamorphic/sedimentary/volcanic (I/M/S/V) were also significantly higher than sedimentary (S; p < 0.05) dominant rock types in terms of species and family richness and mean PD (Table 3). Moderately significant correlations were found between mean age (Ma) and PD (rho = 0.510958, p < 0.001) and species richness (rho = 0.5208758, p < 0.001). There was a significant difference between dominant rock type and rainforest group (χ²(36) = 52.774, p = 0.035). The highest number of rainforest groups were located on sedimentary dominant rock types (n = 20; Table 3), and, of these, semi-evergreen rainforest types were the most common (70%).

Dominant rock type was significantly correlated with the number of Central Queensland Coast endemic rainforest species present within RE types (p < 0.001; Table 3). The number of endemic rainforest species within RE types on sedimentary rocks (S) were significantly different to those grouped by igneous/sedimentary/volcanic/metamorphic rocks (I/S/V/M; p < 0.01), igneous/sedimentary/volcanic (I/S/V; p < 0.001), and igneous/sedimentary rock types (I/S; p < 0.05) (Table 3). Regional ecosystem types on igneous/sedimentary/volcanic (I/S/V) rock types included significantly more threatened species listed under the NCA compared with RE types on sedimentary rocks (S; p < 0.001) and volcanic rock types (V; p < 0.05) (Table 3). RE type “Evergreen complex notophyll feather palm vine forest, of uplands and highlands, on basalt” (RE 8.8.1a) contained species that were significantly phylogenetically even (NRI; p < 0.05; Table 2; Figure 3).

The number of Central Queensland Coast endemic rainforest species recorded in each RE type was weakly significantly correlated with mean geological age (rho = 0.4836044, p < 0.001). The RE with the most Central Queensland Coast endemic rainforest species (n = 25) was “Semi-evergreen notophyll/microphyll to complex notophyll vine forest” (RE 8.12.18), found on substrates with a mean geological age from the mid-Jurassic period (M = 157.0, SD = 125.7 Ma; Table 2). The RE type “Semi-evergreen vine thicket on serpentinite” (RE 11.11.21), located on the oldest substrates of the Neoproterozoic (685 Ma) age had only four Central Queensland Coast endemic species (Table 2). As was expected, those on the youngest, sedimentary dominant rock types of the Quaternary (0.8 Ma) period had the fewest number of Central Queensland Coast endemic rainforest species (Table 2). There was a significant difference in the number of threatened species within RE types among the different dominant rock types (H(6) = 22.40, p < 0.01).

3.4. Central Queensland Coast Rainforest Protected Area Estate

Overall, we found that the protection of RE types was highly variable across the Central Queensland Coast rainforest estate (

Table 4. Summary of dry rainforest REs and areas of extent and protection within the Central Queensland Coast study area, based on species composition data. RE ID (e.g., 11.2.3) where 11 is bioregion, 2 is land zone, and 3 is vegetation community; where PA is protected area; P-C is pre-clearing; REM is remnant; (Ha) is hectares.

RE ID	P–C Total (Ha)	P–C Total PA (Ha)	P–C % PA	REM Total (Ha)	REM Total PA (Ha)	REM% PA
11.11.21	1894	249	13	1518	248	16
11.11.5	37,342	2901	8	14,221	2839	20
11.11.5a	2456	1748	71	2348	1747	74
11.12.16	1911	1220	64	1893	1220	64
11.12.4	37,972	11,448	30	30,691	11,328	37
11.12.4a	413	76	18	385	76	20
11.12.9	*H 111,709	8259	7	*H 95,722	8220	9
11.2.3	2915	343	12	2418	337	14
11.3.11	1810	46	3	313	46	15
11.3.11x1	52	37	71	51	37	72
11.3.40	503	10	2	197	10	5
11.4.1	3035	0	0	123	0	0
11.5.15	5119	1	0	4972	1	0
11.8.3	510	8	2	394	8	2
7.12.10a	1719	1203	70	1717	1203	70
7.12.11a	12,882	11,874	92	12,876	11,873	92
7.12.11b	2437	1829	75	2432	1829	75
7.12.16b	33,790	*H 31,755	94	33,353	*H 31,556	95
7.12.1a	5515	4936	90	5477	4924	90
7.12.21b	5137	4335	84	5112	4329	85
7.12.22b	3448	3258	94	3389	3248	96
7.12.26b	295	294	100	295	294	100
7.12.48	506	475	94	506	475	94
7.3.16b	11,839	2271	19	4408	1496	34
7.3.26a	2407	506	21	2176	503	23
7.3.50b	60	50	82	60	50	82
8.10.1	310	163	53	302	163	54
8.11.2	2875	521	18	2585	498	19
8.11.2x1a	206	5	3	201	5	3
8.12.11	116	114	98	116	114	98
8.12.11a	14,737	13,373	91	14,661	13,416	92
8.12.11c	1553	283	18	1542	284	18
8.12.13	254	220	86	248	213	86
8.12.13a	4456	2908	65	3932	2650	67
8.12.14a	5150	4571	89	5104	4579	90
8.12.14b	9167	8507	93	9122	8506	93
8.12.16	4155	1157	28	4152	1157	28
8.12.17a	3638	3317	91	3505	3299	94
8.12.17b	577	408	71	571	403	71
8.12.17c	707	0	0	707	0	0
8.12.18	26,725	*H 21,570	81	25,977	*H 21,562	83
8.12.19	13,623	10,537	77	12,911	10,480	81
8.12.1a	21,771	15,882	73	18,351	15,676	85
8.12.1b	1392	1391	100	1392	1391	100
8.12.2	33,516	*H 20,951	63	30,596	*H 20,756	68
8.12.26	4523	1096	24	3106	1176	38
8.12.28	1321	772	58	1300	771	59
8.12.29	217	189	87	216	188	87
8.12.29a	1798	1771	99	1830	1802	99
8.12.29b	2728	1989	73	2726	1989	73
8.12.30	486	486	100	486	486	100
8.12.3a	*H 62,910	*H 28,029	45	*H 58,241	*H 27,900	48
8.12.3b	1930	1689	88	1914	1683	88
8.12.3c	2127	236	11	1934	234	12
8.2.2	2501	212	8	2183	213	10
8.2.5	2726	212	8	2411	213	9
8.2.6a	5689	997	18	3771	992	26
8.2.6b	1259	29	2	927	29	3
8.3.10	2043	1034	51	1664	1030	62
8.3.1a	11,636	142	1	5543	154	3
8.3.1b	2314	571	25	1773	295	17
8.3.9	1275	1275	100	1275	1275	100
8.8.1a	2340	700	30	1238	690	56
8.8.1b	943	943	100	943	943	100
9.12.34	10,363	973	9	10,354	973	9
Total	545,230	233,461	43	461,306	231,208	50

(*^H) indicates significant values (p < 0.05) that were higher than expected (z-test).

). Of 65 RE types, only 5 restricted dry REs were very well protected, with 100% of remnant vegetation conserved within protected areas (Table 4). One of these, “Tall open forest and woodland with a well-developed rainforest understory” (RE 7.12.26b; 100%), was significantly phylogenetically even (NRI, p < 0.05) but did not contain any threatened or endemic species (Table 2). Three were phylogenetically random (NRI), for example, “Semi-deciduous complex notophyll vine forest on perched alluvials in valleys of undulating mountain ranges” (RE 8.3.9; 100%), which contained 1 threatened and 11 endemic species, which were protected in Conway National Park of the Whitsunday Region (Table 2). Also, well conserved was “Semi-evergreen notophyll/microphyll to complex notophyll vine forest” (RE 8.12.18; 83%) that contained 21 rainforest species endemic to Central Queensland Coast and 7 species listed under the NCA, such as Medicosma obovata (Table 2, Table 4 and Table A2). Phylogenetically distinct RE types such as “Low notophyll vine thicket” (7.3.50b; NRI, p < 0.05), listed as endangered (EPBC Act), and “Simple notophyll vine forest” (7.12.16b; NRI, p < 0.05) had 82% and 95%, respectively, of remnant rainforest conserved within protected areas (Table 2 and Table 4).

Alarmingly, four dry RE types, with an average mean rainfall of 811–2084 mm, had 0–2% of their remnant extent conserved within a protected area (Table 4). For example, “Semi-evergreen vine thicket on Cainozoic sand plains and/or remnant surfaces” (RE 11.5.15) and the heavily cleared and endangered (EPBC Act) “Semi-evergreen vine thicket ± Casuarina cristata on Cainozoic clay plains” (RE 11.4.1), located in the dry Burdekin and St. Lawrence Gaps, had no remaining remnant vegetation protected (Figure 2; Table 4).

The RE type that underwent one of the greatest rates of clearing of a rainforest ecosystem with only 9% within a protected area was the critically endangered (EPBC Act) “Evergreen notophyll vine forest” (RE 8.2.5), situated on the high parabolic dunes of the Byfield area and found to be significantly phylogenetically even (NRI; p < 0.05) (Figure 2; Table 2 and Table 4). Other dry RE types listed as critically endangered or endangered (EPBC Act) contained less than 28% of remnant rainforest vegetation conserved within protected areas, for example, “Microphyll vine forest (“beach scrub”)” (RE 11.2.3; 14%) and “Semi-evergreen microphyll vine thicket to vine forest” (RE 8.2.2; 10%; Table 4).

Two lowland RE types listed by the Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act) as critically endangered ecological communities, “Evergreen notophyll feather palm vine forest” (8.3.1b; 17%) and “Evergreen notophyll vine forest” (8.2.5; 9%), were determined to be phylogenetically evenly dispersed (NRI, p < 0.05) but were poorly conserved (Figure 2; Table 2 and Table 4). Some critically endangered dry rainforest RE types had less than 16% of remnant rainforest within a protected area and, yet, contained rainforest plant species listed as threatened or endemic, such as the highly distinctive, rare “Semi-evergreen vine thicket on serpentinite” (11.11.21; 16%) and “Evergreen to semi-evergreen, notophyll to microphyll, vine forest to vine thicket” (8.12.3c; 12%) (Table 2 and Table 4).

3.5. Dry Regional Ecosystem Ttypes

Weak but significant positive correlations were found between mean annual rainfall and percent of pre-clear (r = 0.2983215, p < 0.05) and percent remnant rainforest vegetation (r = 0.298528, p < 0.05) in protected areas and were consistent with more mesic rainforest types afforded higher rates of protection (Table 4). Phylogenetic diversity and the number of Central Queensland Coast endemic rainforest species were also significantly but weakly correlated with percent remnant rainforest vegetation (r = 0.3399443, p < 0.01) within protected areas. However, some RE types with a high number of endemic rainforest and threatened species listed under the NCA were poorly conserved. For example, “Semi-deciduous to evergreen notophyll to mesophyll vine forest” (RE 8.3.1a) in lowland areas, with 17 endemic rainforest species, had only 3% remnant rainforest within a protected area (Table 2 and Table 4).

4. Discussion

Typically, the diversity of rainforests has been found to vary along elevation and latitudinal gradients, with the highest diversity in wet tropical biomes at lower latitudes and lower diversity found in drier, more seasonal higher latitudes [19,40,80]. However, contrary to expectations, this study found some of the least diverse RE types (“Mesophyll to notophyll vine Forest”; RE 7.12.1a) were located in the uplands of Paluma National Park of the Wet Tropics. Hilbert et al. (2007) reported that the “Mesophyll to notophyll vine forest” (RE 7.12.1a) was reduced to only 608 km² during the Last Glacial Maximum (LGM), which may explain this lower diversity [81]. Indeed, this study found that the most diverse RE rainforest types were drier and found in more seasonal locations within the Central Mackay Coast bioregion, particularly in “Semi-evergreen notophyll/microphyll to complex notophyll vine forest” (RE 8.12.18). This RE was located in lowland areas across the Central Mackay Coast bioregion on fertile dominant rock types aged between 0.76 and 282.7 Ma. and may have remained stable throughout previous climatic cycles. Similar patterns of the highest phylogenetic and species diversity were found in seasonal semi-deciduous Atlantic forests of South America, while lower diversity was reported for wet forests [82]. Rezende et al. (2020) attributed these results to historical processes amplified by current environmental conditions such as seasonality and water availability [82].

High species richness in Australian rainforests is considered to be influenced by higher proportions of migratory plant taxa from the Sunda region than relictual Sahul elements within tropical northern Queensland, while subtropical rainforest communities have been found to have lower species diversity and fewer Sunda species [50,83,84,85]. Similarly, we found higher numbers of families of known Sunda rather than Sahul ancestry in the tropical areas north of the Burdekin Gap but equal proportions within Central Mackay Coast rainforest RE types. These patterns indicate that tropical and subtropical taxa form a region of overlap in the coastal drier rainforest communities of central Queensland, and this is similar to Yap et al. (2018), who reported almost equal proportions of Sunda and Sahul rainforest tree species within the central Queensland region [84].

The theory of island biogeography predicts lower diversity on smaller islands proportional to isolation and size [86]. Communities on islands were found to be less diverse as isolation increased and area decreased, and it was hypothesised to be the result of the environmental filtering of conserved traits [87,88,89,90]. Rainforest species richness on the continental island of K’gari (Fraser Island) in Southeast Queensland was found to be lower than that of mainland communities [40]. As predicted, we also found a decreased number of rainforest community types on islands. However, since the continental islands of the Whitsunday group were once part of the mainland during the Last Glacial Maximum, there was evidence of shared taxa with mainland rainforest groups but overall less diversity.

Islands were found to have a varied, distinctive community structure [91,92]. Phylogenetically clustered communities and lower diversity were reported on smaller islands favouring stronger competitors [91]. While phylogenetically even communities were found on large islands such as K’gari (Fraser Island) [40], the rainforest communities on the continental islands of the Central Queensland Coast were found to be phylogenetically clustered, consistent with others found on smaller islands.

Rainforest ecosystems with distinctive patterns of overdispersion and clustering have been found worldwide [18,19,43,93,94,95]. Theories predict that phylogenetically clustered rainforest communities are the result of factors such as the recolonisation of disturbed areas from more stable communities, but REs that are phylogenetically even were considered to be examples of stable climatic refugial communities in mesic habitats containing long-lived, dispersal-limited, or endemic taxa [48,49,50,53]. We found RE types of significantly phylogenetically even and clustered communities, but some did not correspond with current hypotheses. For example, one RE type that exhibited a distinctive, significant phylogenetic clustering also contained high numbers of central Queensland endemic rainforest species. These results might be explained by an extended history of climatic variability during the Quaternary period; these have been supported by studies in the Wet Tropics and eastern central and Southeast Queensland [81,96,97,98]. These climatic fluctuations may have resulted in isolation and speciation and may be evidence of strong selective pressures and environmental filtering, suggesting plant species specialisation. However, these hypotheses would require future paleontological research within the Central Queensland Coast region.

Contrary to expectations, the REs with the highest numbers of endemic rainforest species, and PD such as “Semi-evergreen notophyll/microphyll to complex notophyll vine forest” (RE 8.12.18) and “Evergreen to semi-evergreen, notophyll to microphyll, vine forest” (RE 8.12.3a), were found to have a random phylogenetic distribution. These results suggest that these rainforest ecosystems do not fit the accepted definition of a potential refugial area and may be overlooked for conservation. However, factors such as habitat filtering or dispersal syndromes may be active in these areas, and further assessment is required.

Rainfall has been considered to impact the distribution, diversity, and distinctiveness of rainforest communities from global to landscape scales [20,99,100,101,102]. Widely accepted hypotheses suggest the most biodiverse ecosystems are located in more mesic habitats of high rainfall [49,103]. This study found results contrary to this. RE types in areas of very high rainfall did not contain the most diversity or distinctiveness and were phylogenetically random. These results may be influenced by low levels of soil nutrient, such as those located on rhyolite dominant rock, which weathers into clayey soils [104], for example, “Evergreen notophyll Ristantia waterhousei mossy forest of uplands on rhyolite” (RE 8.12.30).

Research has found that, in dry conditions, taxa share similar phylogenetic traits that enable them to survive in harsh environments, otherwise known as niche conservatism or habitat filtering [105]. In Australia, Fensham (1995) found that a great number of the floral taxa of the dry rainforest of inland Central Queensland are specialised and form a continuum of floristic variation between more mesic rainforest types [13]. In this study, we found that phylogenetically clustered rainforest REs existed in dry, harsher environments, consistent with rainforest patterns elsewhere.

Because the Central Queensland Coast region has exceedingly complex and varied geology and highly weathered substrates, we anticipated differences between the rainforest RE types based on geology. Our results showed significant differences between communities based primarily on a geological variation in dominant rock types in terms of diversity. Dominant rock types with unique chemical compositions such as ultramafic (serpentinite) rock are known to sustain distinctive taxa with high levels of endemism [23,87,106], and those communities found in Central Queensland are the most diverse of their kind in Australia [107]. This study found “Semi-evergreen vine thicket on serpentinite” (Re 11.11.21) was distinctive in its community composition, consistent with other studies, and was also significantly phylogenetically clustered, suggesting environmental filtering.

The distribution of climatic refugia has also been considered to be influenced by a region’s geology, as substrates, like igneous and metamorphic rocks, provide greater topographic variation, particularly in drier climates [22]. This research found the RE rainforest community types with the highest PD occurred predominantly on dominant rock groups containing higher-nutrient igneous rock. RE rainforest communities grouped by sedimentary (S) dominant rock types were distinctive in terms of the lowest mean PD but had high species diversity, which suggests selective pressures for certain attributes.

The age of geological substrates has been considered to influence the distribution and diversity of rainforest plant species and may contribute to the existence of habitats that exhibit refugial characteristics, such as endemism, due to geographic and edaphic discontinuity, isolation, and nutrient content [23,104]. Therefore, it may be hypothesised that geological age may correlate significantly with PD. According to this study, rainforest communities on the youngest and oldest substrates included the fewest endemic rainforest species, consistent with previous findings [41]. These communities also exhibited lower PD and an overrepresentation of closely related taxa that is consistent with severely eroded or more recently formed low-nutrient soils that inevitably result in habitat specialisation and conserved traits.

One of four long-term goals of the Kunming-Montreal Global Biodiversity Framework (GBF) is to maintain the integrity, connectivity, genetic diversity, and resilience of all ecosystems and to increase their area by enhancement or restoration by 2050 [55]. We expected to find variation in the extent and percentage of rainforest REs protected within the National Reserve System of Australia. Indeed, this study found that 63% of rainforest RE types within the Central Queensland Coast region had greater than 30% of pre-clearing and remnant extent conserved within protected areas, surpassing the GBF targets, but some were not well protected.

Too often, research has found inconsistencies or no protection of distinctive and diverse ecosystems, especially for dry forest types [8,38,108,109,110]. We found that some rainforest RE types were very well conserved but more restricted or endangered communities were very poorly protected. One example, located along the coastal dunes of the Central Mackay Coast, is the critically endangered and highly diverse “Semi-evergreen Rainforest” type (RE 8.2.2). This RE type contains 13 endemic species and 2 near-threatened species, Brachychiton compactus and Xylosma ovata, listed (NC Act 1992); these represent potential corridors for the movement of rainforest species along the coast.

In this study, significantly phylogenetically even (NRI) evergreen vine forest communities of low to midland coastal areas that contain threatened and Central Queensland Coast endemic rainforest species were found to have less than 17% of remnant vegetation conserved in protected areas. In stark contrast, “Complex notophyll vine forests” (RE 7.12.11a) and “Mesophyll to notophyll vine forest” (RE 7.12.1a) in upland areas with more than 90% of remnant vegetation remaining that were not found to be phylogenetically distinct or highly diverse were listed as endangered ecosystems (EPBC Act). These results are highly incongruent with conservation ideals and highlight the importance of phylogenetic analyses of rainforest community plant communities to identify rainforest ecosystems that require increased protection.

Protected areas alone, however, are considered insufficient for the long-term survival of many rainforest ecosystems [63,111,112]. Rossetto and Kooyman (2021) argued that refugia should not simply be considered as static conservation elements but incorporated into a dynamic conservation model that considers both the persistence and emergence of refugial areas [112]. A departure from the goal of achieving a historic reference state in restoration ecology has been advocated to adjust conservation initiatives to facilitate suitable habitat in order to minimise the loss of biodiversity under climate change [113,114]. Corridors interconnecting existing reserves and remnant forest patches to enable the movement of species throughout the landscape are deemed the most effective way to mitigate the effects of climate change [63,115].

The rainforest estate of the Central Queensland Coast is not only exceedingly threatened by anthropogenic activity, including urban expansion, grazing, mining, and regional development, but also expected increases in the occurrences of wildfires, severe weather events, rainfall variation, and temperature fluctuations due to climate change [81,106,116,117,118,119]. Because of the highly fragmented nature of the Central Queensland Coast rainforest and increased threats, this region would benefit substantially from an increased network of connectivity as some smaller fragments may be of higher importance to meet the current GBF conservation of biodiverse ecosystem targets than previously recognised.

5. Conclusions

This study found that the Central Queensland Coast region contains floristically diverse rainforest ecosystem types but there is considerable overlap in terms of species composition, consistent with a region of overlap. However, some types are significantly different taxonomically and phylogenetically. This research showed that diversity in the Central Queensland Coast region was not consistent with the expected latitudinal and elevational diversity gradients; rather, strong selection pressures for moisture and harsh abiotic conditions, similar to those seen in dry rainforest types around the world, are the primary drivers of species diversity and community distinctiveness. While this study found some RE types consistent with phylogenetic climatic refugia that are well protected within the National Reserve System of Australia, some phylogenetically clustered rainforest communities that are not consistent with recent expansion showed evidence of strong selective pressures and environmental filtering, suggesting a high degree of specialisation to substrates, such as serpentine geologies, with many endemic and congeneric plant species.

We identified the most important diverse and distinct Central Queensland Coast rainforests’ communities for the conservation of genetic and ecological biodiversity. We located communities that require increased protection, particularly littoral rainforest and dry, coastal vine thickets of eastern Australia classified as critically endangered (EPBC Act) that have little to no conservation in the current National Reserve System. Regional Ecosystem communities found in dry areas, on depleted soils, or exposed to harsh environmental conditions need to be protected in order to maintain the high diversity of this region lest we lose unique lineages.