Plant Communities of the Tern Sanctuary on the Matsu Islands as a Breeding Habitat for Seabirds

Abstract

The Matsu Islands Tern Refuge comprises eight reefs located at a relay station on the East Asian bird migration route, and it attracts many transiting, wintering, or breeding birds to inhabit and live on the reefs every year. In order to understand the compositions of plant communities as a breeding habitat for seabirds, we investigated the plant communities of the eight reefs. A total of 130 plots of 10 × 10 square meters were established, from which we found 107 species of plants in 102 genera and 51 families. Among this, we found one critically endangered (CR) species, four vulnerable (VU) species, and three near-threatened (NT) species. The result of two-way indicator species analysis (TWINSPAN) and indicator value (IndVal) showed 130 samples were divided into 11 vegetation types; most of the vegetation types had significant indicator species. We also use the two-way to present the plot of detrended correspondence analysis (DCA) by vegetation types and reefs. Moreover, this result reveals that these samples were more clearly cluster divided by islands. Our results reveal that the compositions and characteristics of plant communities were related clearly to the environmental factors for each reef in the Matsu Islands Tern Refuge. Canonical correspondence analysis (CCA) indicated that species composition of vegetation yielded high correlation with soil property, especially with soil pH. In addition, we found that the traces of bird activity is relevant to the characteristics and structures of plant communities. We found that the plant communities comprising low-grass shrubs would provide relatively soft nesting materials and sheltering effects for eggs or hatchlings for terns. Compared to low-grass shrubs, the traits of high-grass shrubs would not be beneficial to nest for breeding of terns on the ground, and no nested trace was found in these plant communities.

1. Introduction

Reefs in the tropics have been receiving much attention from many ecologists [1,2,3,4], especially under the influence of current climate change at different scales [4]. In addition to being a unique natural landscape, the vegetation on the reef also reflects its adaptability to the reef and directly and indirectly impacts the stability and resilience of the island ecosystem [5,6]. Many studies also pointed out that the vegetation on these reefs is closely related to the environment where they grow, especially soil factors [5,6,7,8,9]. The animal communities that inhabit plant communities on the reefs choose plants suited to their needs for rest, feeding, breeding, and sheltering from enemies [10]. Although terns mainly feed on sea fishes, they must return to land for resting and breeding, and through the elimination effect, the soil properties on the reef are affected [10]. Therefore, the reef ecosystem is relatively close to being severely threatened by the impact of climate change and sea-level rising, which is also an issue of concern in many studies [3,4,10].

The Matsu Islands Tern Refuge comprises eight reefs, including Shuangzih, Baimiao, Tiejian, Jhongdao, Sanlianyu, Jinyu, Lioucyuan, and Sheshan Reefs (Appendix A Figure A1), located at the southwestern end of the Zhoushan Islands (world-renowned fishing grounds) nearby Mainland China. The Matsu Islands Tern Refuge abounds with fishery resources due to the confluence of warm and cold currents, providing seabirds with a sufficient food source. Additionally, located at a relay station of the bird migration route in East Asia, the Matsu Islands attract many birds in transit or that winter or breed every year, making it an important habitat for bird conservation. More than 250 species of birds have been recorded in the Matsu Islands Tern Refuge, and it mainly consists of seven populations, which include the bridled tern (Onychoprion anaethetus Scopoli), roseate tern (Sterna dougallii, Montagu), black-naped tern (Sterna sumatrana Raffles), black-tailed gull (Larus crassirostris Vieillot), Pacific reef heron (Egretta sacra Gmelin), and Pacific swift (Apus pacificus Latham) [11]. The Chinese crested tern (CCT) (Thalasseus bernsteini Schlegel), found on Jhongdao Reef, is listed as an endangered bird in the IUCN Red List of Threatened Species [12], and its breeding behaviors were first identified and recorded in 2000 [13,14]. All the reefs of the Matsu Islands Tern Refuge are covered by bare granite, along with a little grass-covered land, where the Compositae and Aizoaceae plants are dominant [11].

Compared to the studies of tern gathering and breeding [13,14,15,16,17,18], relatively few studies have been conducted on the flora and vegetation in the Matsu Islands Tern Refuge. The composition of plant resources in the Matsu Islands Tern Refuge is mainly influenced by the proportion of soil area to island area, island area, and the distance to neighboring large islands [19]. Most related studies [20,21,22] pertain to the inhabited big islands of the Matsu Islands. Thus, data on the flora and plant communities on the Matsu Islands are scarce. Only Tzeng et al. [19] investigated the plant resources on the Matsu Islands. They obtained detailed flora data, recording a total of 107 species of vascular plants within 102 genera of 53 families, including 3 types of naturalized plants (Conyza bonariensis, Physalis angulate, and Cotula australis), and 3 species of ferns (Lygodium japonicum, Pteris fauriei, and Cyrtomium falcatum). The reef with the highest plant species diversity was Jinyu Reef (70 species), followed by the Tiejian Reef (62 species); the reefs with the lowest plant species diversity were Baimiao Reef (20 species) and Shuangzih Reef (19 species).

This study marked the inaugural survey of plant communities on the eight uninhabited reefs within the Matsu Islands Tern Refuge. It serves as valuable foundational data in validating the vegetation ecology present on these tropical reefs. Our analysis focused on detailing the composition characteristics of the plant communities and delving into the interconnectedness between tern habitats and vegetation. Additionally, we documented seabird activity to inform discussions on the conservation efforts of the Chinese crested tern.

2. Materials and Methods

2.1. Study Area

Located in the northwest of the Taiwan Straits and outside the estuary of the Minjiang River (Fujian Province, Mainland China), the Matsu Islands Tern Refuge comprises eight reefs of the Matsu Islands, including Shuangzih, Baimiao, Tiejian, Jhongdao, Sanlianyu, Jinyu, Lioucyuan, and Sheshan Reefs (Figure 1 and Figure A1), and covers an area of approximately 72 ha (land area: 12 ha; sea area: 60 ha) [11]. The Matsu Islands Tern Refuge is administered by Lianjiang County and is intended to conserve the island ecology, the inhabiting seabirds, and the unique geographic landscape [11].

The Matsu Islands are abundant in pyroclastic rocks of the Continental Shelf, with ages ranging from 90 million to 100 million years; these pyroclastic rocks are classified as pyrogenetic and metamorphic rocks, in which granites and granodiorites are dominant. Metamorphic rocks are only found in stream valleys or beaches [23]. Granites are solid and impervious to weathering, and superficial granites weather slowly into the soil. There is a lack of complete pedogenesis in certain areas [23]. The soil’s structural properties are the main factors affecting the overall vegetation and plant growth on the Matsu Islands [19,21,22].

The Matsu Islands have a subtropical maritime monsoon climate, with northeasterly winds prevailing throughout the year. They have four distinctive seasons, with a warm spring and autumn and a significant temperature difference between winter and summer: The coldest month is January, which has an average temperature of approximately 8.9 °C, and the warmest month is August, with an average temperature of 27.1 °C. The average annual temperature is 18.2 °C, and the average annual rainfall is approximately 1035.2 mm, with uneven rainfall distribution and distinct wet and dry seasons. The wet season is mainly concentrated in the plum rain period from April to June, and the typhoon period occurs in summer [19]. Because of the small size of the islands and their steep terrain, their ability to retain rainwater is poor; thus, the Matsu Islands suffer a year-round water shortage. During the winter months (October to February of the following year), the northeast monsoons are intense and rainfall is low, possibly because of the lack of hills offering shelter. In particular, drought is experienced every October [20,22,23].

2.2. Sampling and Vegetation Survey

In the Matsu Islands Tern Refuge, the terns’ annual breeding season occurs from May to August; thus, field surveys can only be conducted during the nonbreeding season to avoid disturbing their breeding. Eight uninhabited islands are small and have no docks; thus, survey personnel must lean fishing boats against the protruding reefs and quickly disembark. However, the large waves caused by the northeast monsoon from October to March complicate disembarking to conduct field surveys [19]. Therefore, the personnel could only survey eight uninhabited islands in April and September 2010, during which the winds and waves were relatively calm.

In this study, for selecting sample plots, we mainly considered the morphology, composition, and height of plants, as well as the homogeneity of the environments. Most of the Matsu Islands Tern Refuge reefs are small and have shallow soil; thus, they are mainly covered by herbs and shrubs. In addition, the composition and height of plants vary from reef to reef; thus, the size of the sample plots differed from the growth pattern of the plants [9]. Tall shrubs and Miscanthus floridulus mainly cover Jinyu Reef; thus, each sample area was 10 m × 10 m and was divided into four 5 m × 5 m subareas. Those low herbs and shrubs mostly covered the seven other reefs; each sample area was 4 m × 4 m and was divided into four 2 m × 2 m subareas. We recorded the species and coverage of species in each sample plot and study area. Plant species records are based on the Catalogue of Life in Taiwan (https://taicol.tw (accessed on 5 July 2024)) [24], and we also marked the rare species according to The Red List of Vascular Plants of Taiwan (https://www.tbri.gov.tw/A6_2/open/29633 (accessed on 5 July 2024)) [25].

2.3. Environmental Factor Survey

Plant growth is influenced by environmental factors, which are the main cause of variations in vegetation. To assess plant fertility and distribution, we investigated the following environmental factors: altitude (Alt), slope (Slo), whole-light sky (WLS), aspect (Asp), and moisture gradient (MG). The altitude of each sample plot was determined using a GPS device (GPSMAP 60CSx, Garmin, Olathe, KS, USA), and we recorded the TWD 97 coordinate of each sample plot. A compass measured each sample plot’s slope, WLS, and aspect. The WLS refers to the scope of sky accessible to solar radiation in a sample plot and is an estimated value that considers factors such as direction, slope, terrain masking, and solar radiation energy [26,27]. Aspect refers to the direction in which the sample plot slope surface faces. Furthermore, we considered aspects such as the plants corresponding to moisture gradients and assigned the relative values of 1 (the driest) to 16 (the wettest) [28].

We selected one soil sample at each of the four sample subplots and sampled the topsoil after the dead leaves were removed, and then we mixed the topsoil and brought it back to the laboratory. Subsequently, the soil samples were air-dried at room temperature and were sifted through a 2 mm filter mesh to determine the moisture content, pH value, organic matter, and total nitrogen [29]. After the soil samples were air-dried, they were sifted using a 2 mm aperture filter mesh; we collected 5 g of air-dried soil to be baked continuously at 105 °C for 24 h, and we weighed the absolute-dry soil and calculated its moisture content using the following equation: [(air-dried soil weight) − (absolute-dry soil weight)] ÷ (absolute-dry soil weight) × 100% [29]. We mixed the soil with distilled water in a proportion of 1:2.5 (w/v), placed the mixture in a test tube, stirred it well, and left it to rest for one day [30]. We measured its pH value using the pH meter (model 6173 pH, Jenco). Soil organic matter and total nitrogen were analyzed using the Walkley–Black wet combustion method and the Semimacro Kjeldhal method [31].

2.4. Data Analysis

The plant community mediation in a sample area is expressed as the importance value (IV): plant species’ occurrence frequency and coverage in a sample plot. The IV can be converted into the sum of relative frequency and relative dominance, indicating the importance of a plant species in the plant community of a sample plot.

Vegetation analysis can be conducted using the two-way indicator species analysis (TWINSPAN) method [32]. In this study, we converted the IV index of a sample plot into decile values (0 to 9), and we identified its social unit and characteristic species to establish a vegetation structure. Each vegetation type was named based on the principle of characteristic species first and then dominant species; if a characteristic species was also the dominant species of a plant community, the vegetation type was named after the plant species. We also counted the indicator value (IndVal) to verify the result of TWINSPAN [33,34]. The equation is as follows:

For each species i in group j, define the specificity as

Aij = Nindividualsij/Nindividualsi

(1)

Bij = Nsitesij/Nsitesj

(2)

INDVALij = Aij Bij

(3)

where Nindividualsij is the mean number of individuals of species i across sites in group j, and Nindividualsi is the sum of the mean numbers of individuals of species i over all groups. Nsitesij is the number of sites in group j where species i is present, and Nsitesj is the total number of sites in group j. The indicator value of species i in group j is then a value between 0 and 1. We conducted INDVAL with R version 4.3.3 [35] and the “labdsv” package. We conducted a detrended correspondence analysis (DCA) based on the IV index of the species in each sample area to investigate the relationship between the species composition and environmental factors of each reef. If the axial length of the DCA was greater than four standard deviations (SDs), canonical correspondence analysis (CCA) was conducted [36]. Ordination analysis was conducted with PC-ORD 7.0 software [37].

3. Results

3.1. Vegetation Classification

We investigated 130 sampling plots distributed on eight Matsu Islands Tern Refuge reefs.

Table 1. The environments of each reef in the Matsu Islands Tern Refuge of the West Pacific Ocean (modified by [31]).

REEFS	Lioucyuan	Jhongdao	Baimiao a	Tiejian	Jinyu	Sheshan	Sanlianyu	Shuangzih
Area (m2) b	1420	820	750 b	980	3110	1500	1500	500 b
Code	1	2	3	4	5	6	7	8
Longitude (°)	119°58′32″	119°59′60″	120°00′10″	119°58′60″	119°56′42″	120°00′10″	119°03′32″	120°29′34″
Latitude (°)	26°13′32″	26°15′50″	26°17′00″	26°16′50″	26°11′42″	25°17′00″	26°14′17″	26°21′43″
Distance (m) d	6550	1050	820	130	1720	550	4650	50
Direction e	East South	East South	East	West	West South	West	East	South
Altitude (m)	63	32	32	29	42	29	39	33
Soil ratio c	3	4	1	4	5	3	3	2
Species number	18	33	20	62	70	33	26	19
Species density (spp./ha)	12.7	40.2	26.7	63.3	22.5	22.0	17.3	38.0
Species richness (spp./log(area))	5.71	11.33	6.96	20.73	20.05	10.39	8.19	7.04

^a Data not for the main island; ^b estimated area of surveyed island; ^c soil ratio: soil area ratio of the island, 1: <5%; 2: 5–10%; 3: 10–25%; 4: 25–50%; 5: >50%; ^d distance: distance to the nearest larger island; ^e direction: direction to the nearest larger island.

and Appendix A provide data about the number of vascular plant species and environmental characteristics of different reefs and sample plots. A total of 52 families, 102 genera, and 107 species were surveyed in the sampling plots, including three ferns (

Table 2. The plant list and indicator value in the Matsu Islands Tern Refuge.

Family	Scientific Name	Code	Veg- Type a	IndVal	p-Value	Ree- Type a	IndVal	p-Value	Spread Type b	IUCN c
Dryopteridaceae	Cyrtomium falcatum (L. f.) C. Presl	Cyrfal	II	0.150	0.050	8	0.097	0.103	W, A
Lygodiaceae	Lygodium japonicum (Thunb.) Sw.	Lygjap	I	0.500	0.001	8	0.385	0.001	W, A
Pteridaceae	Pteris fauriei Hieron.	Ptefau	II	0.080	0.382	8	0.048	0.546	W, A
Acanthaceae	Justicia procumbens L.	Juspro	V	0.060	0.710	5	0.059	0.472	A
Aizoaceae	Sesuvium portulacastrum (L.) L.	Sespor	IX	0.250	0.011	3	0.136	0.056	A
Aizoaceae	Tetragonia tetragonoides (Pall.) Kuntze	Tettet	IX	0.262	0.005	1	0.359	0.001	A
Amaranthaceae	Achyranthes aspera var. indica L.	Achasp	II	0.250	0.007	8	0.154	0.023	A
Amaranthaceae	Chenopodium acuminatum subsp. virgatum (Thunb.) Kitam.	Cheacu	VIII	0.277	0.005	5	0.393	0.001	A
Apiaceae	Peucedanum japonicum Thunb.	Peujap	II	0.400	0.005	8	0.225	0.027	W
Apocynaceae	Gymnema sylvestre (Retz.) R. Br. ex Schult.	Gymsyl	III	0.360	0.002	7	0.535	0.001	W
Apocynaceae	Trachelospermum jasminoides (Lindl.) Lem.	Trajas	I	0.379	0.007	8	0.231	0.006	W
Asteraceae	Artemisia capillaris Thunb.	Artcap	V	0.594	0.001	5	0.593	0.001	W
Asteraceae	Aster hispidus Thunb.	Astcil	VI	0.943	0.001	4	0.647	0.001	W, A
Asteraceae	Cirsium japonicum var. australe Kitam.	Cirjap	II	0.250	0.007	8	0.154	0.033	W, A
Asteraceae	Conyza bonariensis (L.) Cronquist	Conbon	-	-	-	-	-	-	W
Asteraceae	Cotula australis (Sieber ex Spreng.) Hook. f.	Cotaus	-	-	-	-	-	-	W, A
Asteraceae	Crepidiastrum lanceolatum (Houtt.) Nakai	Crelan	IX	0.441	0.001	3	0.320	0.004	W, A
Asteraceae	Crossostephium chinense (L.) Makino	Crochi	VIII	0.397	0.001	3	0.419	0.001	W, A	VU
Asteraceae	Dendranthema indicum (L.) Des Moul.	Denind	II	0.385	0.002	6	0.351	0.001	W
Asteraceae	Emilia sonchifolia var. javanica (Burm. f.) Mattf.	Emison	II	0.110	0.209	8	0.070	0.293	W
Asteraceae	Farfugium japonicum var. formosanum (Hayata) Kitam.	Petfor	II	0.375	0.002	8	0.231	0.004	W
Asteraceae	Sonchus oleraceus L.	Sonole	I	0.110	0.148	8	0.029	0.872	W
Brassicaceae	Coronopus didymus (L.) Sm.	Cordid	-	-	-	-	-	-	A
Cannabaceae	Celtis biondii Pamp.	Celbio	I	0.200	0.037	8	0.077	0.326	A
Cannabaceae	Celtis sinensis Pers.	Celsin	-	-	-	-	-	-	A
Caprifoliaceae	Lonicera japonica Thunb.	Lonjap	III	0.724	0.001	7	0.618	0.001	A
Caryophyllaceae	Dianthus superbus var. longicalycinus (Maxim.) F.N. Williams	Diasup	V	0.120	0.122	5	0.118	0.027	A
Caryophyllaceae	Sagina japonica (Sw.) Ohwi	Sagjap	-	-	-	-	-	-	A
Caryophyllaceae	Silene aprica Turcz.	Silapr	-	-	-	-	-	-	A
Caryophyllaceae	Stellaria media (L.) Vill.	Stemed	-	-	-	-	-	-	A
Celastraceae	Celastrus hindsii Benth.	Celhin	II	0.271	0.003	8	0.538	0.001	A
Celastraceae	Euonymus japonicus Thunb.	Euojap	II	0.781	0.001	8	0.615	0.001	A	CR
Celastraceae	Maytenus diversifolia (Maxim.) Ding Hou	Maydiv	II	0.803	0.001	1	0.803	0.001	A
Convolvulaceae	Dichondra micrantha Urb.	Dicmic	V	0.060	0.700	5	0.059	0.496	A
Crassulaceae	Sedum formosanum N.E. Br.	Sedfor	I	0.070	0.573	8	0.131	0.100	A
Elaeagnaceae	Elaeagnus glabra Thunb.	Elagla	II	0.454	0.001	8	0.629	0.001	A
Fabaceae	Callerya reticulata (Benth.) Schot	Calret	IV	0.312	0.003	6	0.233	0.003	A
Fabaceae	Canavalia lineata (Thunb.) DC.	Canlin	II	0.199	0.024	8	0.142	0.065	A
Fabaceae	Rhynchosia volubilis Lour.	Rhyvol	I	0.600	0.001	8	0.231	0.003	A
Lamiaceae	Leucas chinensis (Retz.) R. Br.	Leuchi	II	0.232	0.008	8	0.136	0.052	A
Malvaceae	Grewia rhombifolia Kaneh. & Sasaki	Grerho	IV	0.397	0.001	6	0.302	0.002	A
Malvaceae	Sida rhombifolia subsp. insularis (Hatus.) Hatus.	Sidrho	VIII	0.100	0.372	3	0.045	0.769	A
Menispermaceae	Cocculus orbiculatus (L.) DC.	Cocorb	II	0.185	0.048	6	0.230	0.018	A
Menispermaceae	Stephania japonica (Thunb.) Miers	Stejap	I	0.160	0.080	8	0.231	0.005	A
Moraceae	Ficus erecta var. beecheyana (Hook. & Arn.) King	Ficere	I	0.310	0.004	8	0.538	0.001	A
Moraceae	Ficus pumila L.	Ficpum	I	0.433	0.001	8	0.538	0.001	A
Moraceae	Maclura cochinchinensis (Lour.) Corner	Maccoc	III	0.581	0.001	7	0.560	0.001	A
Nyctaginaceae	Boerhavia diffusa L.	Boedif	XI	0.466	0.002	2	0.503	0.001	A
Oxalidaceae	Oxalis corniculata L.	Oxacor	V	0.130	0.191	5	0.183	0.049	A
Pentaphylacaceae	Eurya emarginata (Thunb.) Makino	Eurema	II	0.942	0.001	8	0.692	0.001	A
Phyllanthaceae	Breynia officinalis Hemsl.	Breoff	IV	0.266	0.009	6	0.329	0.001	A
Phyllanthaceae	Glochidion rubrum Blume	Glorub	III	0.894	0.001	7	0.816	0.001	A
Piperaceae	Piper kadsura (Choisy) Ohwi	Pipkad	II	0.120	0.152	8	0.077	0.331	A
Pittosporaceae	Pittosporum tobira (Thunb.) W.T. Aiton	Pittob	I	0.140	0.090	8	0.103	0.092	A
Plumbaginaceae	Limonium sinense (Girard) Kuntze	Limsin	II	0.130	0.131	8	0.118	0.120	A
Polygonaceae	Polygonum chinense L.	Polchi	II	0.262	0.010	8	0.277	0.003	A
Polygonaceae	Polygonum multiflorum var. hypoleucum (Nakai ex Ohwi) T.S. Liu, S.S. Ying & M.J. Lai	Polmul	I	0.400	0.004	8	0.154	0.037	W
Polygonaceae	Rumex crispus var. japonicus (Houtt.) Makino	Rumcri	IX	0.170	0.089	3	0.080	0.485	W, A
Portulacaceae	Portulaca pilosa L.	Porpil	-	-	-	-	-	-	A
Primulaceae	Anagallis arvensis L.	Anaarv	-	-	-	-	-	-	A
Primulaceae	Ardisia sieboldii Miq.	Ardsie	I	0.255	0.009	8	0.308	0.001	A
Primulaceae	Lysimachia mauritiana Lam.	Lysmau	X	0.394	0.002	2	0.449	0.001	A
Rhamnaceae	Berchemia lineata (L.) DC.	Berlin	IX	0.080	0.451	3	0.045	0.791	A
Rhamnaceae	Rhamnus brachypoda C.Y. Wu ex Y.L. Chen & P.K. Chou	Rhabra	I	0.200	0.045	8	0.077	0.325	A
Rhamnaceae	Sageretia thea (Osbeck) M.C. Johnst.	Sagthe	I	0.508	0.002	8	0.219	0.010	A
Rosaceae	Prunus japonica Thunb.	Prujap	-	-	-	5	0.044	0.599	A
Rosaceae	Rhaphiolepis indica var. umbellata (Thunb.) H. Ohashi	Rhaind	IV	0.110	0.161	6	0.145	0.068	A	NT
Rosaceae	Rosa bracteata J.C. Wendl.	Rosbra	II	0.120	0.151	8	0.077	0.329	A	VU
Rosaceae	Rosa cymosa Tratt.	Roscym	I	0.200	0.036	8	0.077	0.324	A	VU
Rosaceae	Rubus parvifolius L.	Rubpar	IV	0.173	0.031	6	0.174	0.029	A
Rubiaceae	Galium gracilens (A. Gray) Makino	Galgra	-	-	-	-	-	-	A
Rubiaceae	Hedyotis strigulosa var. parvifolia (Hook. & Arn.) T. Yamaz.	Hedstr	X	0.295	0.011	2	0.302	0.003	A
Rubiaceae	Paederia foetida L.	Paefoe	I	0.368	0.004	8	0.486	0.001	A
Rutaceae	Zanthoxylum nitidum (Roxb.) DC.	Zannit	I	0.798	0.001	8	0.769	0.001	A
Salicaceae	Scolopia oldhamii Hance	Scoold	XI	0.100	0.354	1	0.100	0.137	A
Solanaceae	Physalis peruviana L.	Phyang	II	0.250	0.007	8	0.154	0.028	A
Solanaceae	Solanum nigrum L.	Solnig	XI	0.180	0.052	1	0.211	0.037	A
Urticaceae	Boehmeria nivea var. tenacissima (Gaudich.) Miq.	Boeniv	II	0.417	0.001	8	0.462	0.001	A
Vitaceae	Ampelopsis brevipedunculata var. hancei (Planch.) Rehder	Ampbre	I	0.372	0.001	8	0.580	0.001	A
Vitaceae	Vitis thunbergii Siebold & Zucc.	Vitthu	III	0.321	0.001	7	0.353	0.001	A	NT
Amaryllidaceae	Allium grayi Regel	Allgra	VIII	0.331	0.005	2	0.229	0.006	A	DD
Amaryllidaceae	Lycoris radiata (L’Hér.) Herb.	Lycrad	IV	0.214	0.012	6	0.103	0.092	A
Araceae	Arisaema heterophyllum Blume	Arihet	III	0.140	0.103	7	0.125	0.061	A
Araceae	Pinellia ternata (Thunb.) Ten. ex Breitenb.	Pinter	II	0.120	0.140	8	0.077	0.311	A
Asparagaceae	Asparagus cochinchinensis (Lour.) Merr.	Aspcoc	IX	0.633	0.001	3	0.297	0.004	A
Asparagaceae	Scilla sinensis (Lour.) Merr.	Barjap	XI	0.110	0.135	1	0.112	0.061	A	VU
Asphodelaceae	Dianella ensifolia (L.) DC.	Diaens	III	0.385	0.001	7	0.359	0.001	A
Commelinaceae	Commelina communis L.	Comcom	III	0.220	0.023	6	0.185	0.018	A
Cyperaceae	Carex wahuensis C.A. Mey.	Carwah	-	-	-	-	-	-	A
Cyperaceae	Cyperus rotundus L.	Cyprot	VIII	0.193	0.025	3	0.099	0.117	A
Cyperaceae	Fimbristylis ovata (Burm. f.) J. Kern	Fimova	X	0.167	0.049	2	0.167	0.027	A
Cyperaceae	Mariscus cyperinus Vahl	Marcyp	IX	0.329	0.002	3	0.172	0.041	A
Cyperaceae	Pycreus polystachyos (Rottb.) P. Beauv.	Pycpol	XI	0.192	0.018	1	0.194	0.012	A
Dioscoreaceae	Dioscorea futschauensis Uline ex R. Knuth	Diofut	I	0.424	0.002	8	0.385	0.001	W
Liliaceae	Liriope minor var. angustissima (Ohwi) S.S. Ying	Lirmin	III	0.160	0.062	7	0.169	0.031	A
Poaceae	Calamagrostis epigeios (L.) Roth	Calepi	IV	0.188	0.035	6	0.244	0.004	W, A
Poaceae	Dactyloctenium aegyptium (L.) Willd.	Dacaeg	XI	0.500	0.001	1	0.500	0.001	W, A
Poaceae	Digitaria setigera Roth	Digset	IV	0.070	0.599	6	0.077	0.350	W, A
Poaceae	Ischaemum crassipes (Steud.) Thell.	Isccra	V	0.060	0.710	5	0.059	0.469	W, A	DD
Poaceae	Miscanthus floridulus (Labill.) Warb. ex K. Schum. & Lauterb.	Misflo	II	0.322	0.003	8	0.500	0.001	W, A
Poaceae	Paspalum scrobiculatum L.	Passcr	IX	0.080	0.478	3	0.045	0.787	W, A
Poaceae	Polypogon fugax Nees ex Steud.	Polfug	-	-	-	-	-	-	W, A
Poaceae	Setaria pallide-fusca (Schumach.) Stapf & C.E. Hubb.	Setpal	VIII	0.638	0.001	3	0.310	0.004	W, A
Poaceae	Setaria pumila (Poir.) Roem. & Schult.	Setgla	V	0.389	0.001	5	0.412	0.001	W, A
Poaceae	Zoysia tenuifolia Thiele	Zoyten	VIII	0.277	0.008	3	0.077	0.028	W, A	NT
Smilacaceae	Smilax china L.	Smichi	III	0.403	0.001	7	0.458	0.001	A

Note: “-” means not in survey plots; retrieved 2024-07-09 from https://taicol.tw (accessed on 5 July 2024) [24]. Type ^a was according to the result of TWINSPAN and Table 1. Spread type ^b was based on the fruit of the species; W: wind, A: animal. IUCN ^c was according to the Taiwanese IUCN species list [25].

). Among this, we found eight threatened species and two species that were data deficient (DD). The critically endangered (CR) species was Euonymus japonicus. Four vulnerable (VU) species were Rosa bracteate, R. cymose, Scilla sinensis and Crossostephium chinense. Three near-threatened (NT) species were Rhaphiolepis indica var. umbellate, Vitis thunbergii, and Zoysia tenuifolia.

We conducted TWINSPAN on the sample area data to re-rank the sample plots and tree species, where Level 3 showed the results of the differentiation of sample plots and species. According to the differential species, 11 vegetation types were identified (Figure 2). At Level 1, we classified Jinyu and Tiejian Reefs and the plant communities on five other reefs into two vegetation physiognomies. A high-grass shrub with a height of more than 1 m dominated the first vegetation physiognomy, with four species, including Breynia officinalis, Mis. floridulus, Dendranthema indicum, and Rubus parvifolius. The second vegetation physiognomy was dominated by low-grass shrubs with a height of less than 60 cm, which were distributed on all reefs except for Jinyu Reef. The vegetation type had four species: Crossostephium chinense, Chenopodium acuminatum subsp. virgatum, Setaria glauca, and Tetragonia tetragonoides.

The following section describes the composition characteristics of each vegetation type and summarizes its environmental characteristics and plant composition (presented in Table 2 and

Table 3. The characteristics of vegetation type and plant composition in the Matsu Islands Tern Refuge.

Vegetation Types	Distributed Reefs	Vegetation Physiognomy	Characteristic Species	Dominated Compositions	Environmental Condition a
I. Sageretia thea-Rhynchosia volubilis-Miscanthus floridulus type	Jinyu Reef	High grass with vegetation height of more than 2 m	Sageretia thea and Rhynchosia volubilis	Miscanthus floridulus	Altitude (m): 28.00 ± 10.27 Slope (°): 23.50 ± 14.15	pH: 5.50 ± 0.33 SWP: 2.46 ± 0.33 SOM: 20.01 ± 6.38 STN: 0.91 ± 0.37
II. Eurya emarginata-Euonymus japonicus type	Jinyu Reef	Shrubs with vegetation height of more than 2 m	Eurya emarginata and Euonymus japonicas	Eur. emarginata and Euo. japonicus	Altitude (m): 16.13 ± 5.33 Slope (°): 32.13 ± 4.74	pH: 5.22 ± 0.63 SWP: 3.24 ± 1.3 SOM: 46.16 ± 44.56 STN: 2.12 ± 1.65
III. Glochidion rubrum-Maclura cochinchinensis-Smilax china type	Tiejian Reef	Shrub with vegetation height of less than 1.5 m	Glochidion rubrum and Maclura cochinchinensis	Gl. rubrum, Ma. cochinchinensis, and Br. officinalis	Altitude (m): 23.86 ± 5.01 Slope (°): 9.64 ± 7.01	pH: 4.79 ± 0.47 SWP: 2.84 ± 0.61 SOM: 31.19 ± 13.91 STN: 0.18 ± 0.29
IV. Millettia reticulate-Grewia rhombifolia-Dendranthema indicum type	Tiejian Reef	Shrub mix herb with vegetation height of less than 1.5 m	Millettia reticulate and Grewia rhombifolia	Br. officinalis, Grewia rhombifolia, Rhaphiolepis indica var. umbellate, Lonicera japonica, Coc. Orbiculatus, Rub. Parvifolius, Millettia reticulate, and Dendranthema indicum	Altitude (m): 21.14 ± 4.28 Slope (°): 20.39 ± 11.15	pH: 4.97 ± 0.61 SWP: 3.50 ± 0.60 SOM: 52.46 ± 17.88 STN: 0.70 ± 1.05
V. Artemisia capillaris-Chenopodium acuminatum subsp. virgatum-Setaria glauca type	Tiejian Reef, Lioucyuan Reef (3 plots), and Shuangzih Reef (1 plot)	Herb vegetation height of less than 60 cm	Artemisia capillaris and Chenopodium acuminatum subsp. virgatum	Chenopodium acuminatum subsp. virgatum, Artemisia capillaris, Setaria glauca, Solanum nigrum, and Crossostephium chinense	Altitude (m): 20.82 ± 5.77 Slope (°): 24.59 ± 11.41	pH: 5.35 ± 0.69 SWP: 3.66 ± 2.11 SOM: 47.36 ± 30.28 STN: 2.88 ± 5.1
VI. Aster asagrayi-Crossostephium chinense type	Jhongdao Reef and Lioucyuan Reef (1 plot)	Herb vegetation height of less than 60 cm	Aster asagrayi	Aster asagrayi, Crossostephium chinense, Chenopodium acuminatum subsp. virgatum, and Coc. orbiculatus	Altitude (m): 29.45 ± 8.86 Slope (°): 23.23 ± 9.15	pH: 4.87 ± 0.59 SWP: 2.74 ± 0.81 SOM: 23.50 ± 11.83 STN: 0.15 ± 0.07
VII. Chenopodium acuminatum subsp. virgatum-Tetragonia tetragonoides type	Sanlianyu Reef and Sheshan Reef (2 plots)	Herb vegetation height of less than 60 cm	Chenopodium acuminatum	Chenopodium acuminatum subsp. virgatum, Tetragonia tetragonoides, Coc. orbiculatus, and Oxalis corniculata	Altitude (m): 32.67 ± 4.44 Slope (°): 24.28 ± 14.67	pH: 4.38 ± 0.87 SWP: 3.61 ± 1.35 SOM: 69.31 ± 38.56 STN: 2.67 ± 1.82
VIII. Setaria pallide-fusca-Zoysia tenuifolia type	Sanlianyu Reef	Herb vegetation height of less than 60 cm	Setaria pallide-fusca and Zoysia tenuifolia	Crossostephium chinense and Chenopodium acuminatum subsp. virgatum	Altitude (m): 27.00 ± 7.90 Slope (°): 25.40 ± 14.61	pH: 5.23 ± 0.62 SWP: 3.42 ± 1.65 SOM: 67.32 ± 55.43 STN: 2.22 ± 2.33
IX. Asparagus cochinchinensis-Crepidiastrum lanceolatum type	Sheshan Reef	Herb vegetation height of less than 60 cm	Asp. Cochinchinensis and Crepidiastrum lanceolatum	Asp. cochinchinensis and Cre. lanceolatum	Altitude (m): 25.08 ± 2.57 Slope (°): 20.89 ± 15.90	pH: 5.72 ± 0.43 SWP: 2.87 ± 0.93 SOM: 86.8 ± 162.26 STN: 3.78 ± 7.07
X. Boerhavia diffusa-Allium macrostemon type	Baimiao Reef, Lioucyuan Reef (4 plots), and Tiejian Reef (1 plot)	Herb vegetation height of less than 60 cm	Boerhavia diffusa and Cre. lanceolatum	Lysimachia mauritiana and Setaria glauca	Altitude (m): 29.92 ± 18.69 Slope (°): 12.97 ± 11.53	pH: 5.10 ± 1.02 SWP: 12.62 ± 20.22 SOM: 114.76 ± 138.39 STN: 0.95 ± 0.75
XI. Maytenus diversifolia-Dactyloctenium aegyptium-Tetragonia tetragonoides type	Shuangzih Reef	Herb mix shrub vegetation with height of less than 60 cm	Maytenus diversifolia and Dactyloctenium aegyptium	Tetragonia tetragonoides	Altitude (m): 26.10 ± 3.90 Slope (°): 14.2 ± 12.64	pH: 4.71 ± 0.67 SWP: 4.05 ± 2.19 SOM: 80.31 ± 83.45 STN: 5.16 ± 5.03

Note: environmental condition ^a: SWP is soil water potential (MPa), SOM is soil organic matter (%), STN is soil total nitrogen (g/kg).

). Based on the calculation of the indicator value, we found that most plant groups had obvious indicator species, except for VI. Aster asagrayi-Crossostephium chinense type (only 1 indicator species), and VII. Chenopodium acuminatum subsp. virgatum − Tetragonia tetragonoides type (no indicator species). A total of 94 species were used to calculate indicator values for the 130 samples. The indicator values for 65 species were significant (p < 0.05) and the indicator values for 51 species were above 0.25 and significant (p < 0.05). We also counted the indicator value by reefs, and all reefs had corresponding indicator species. The indicator values for 61 species were significant (p < 0.05), and the indicator values for 39 species were above 0.25 and significant (p < 0.05).

3.2. Ordination

The vegetation data of the Matsu Islands Tern Refuge were ordinated using detrended correspondence analysis (DCA). The lengths of Axes_1, _2, and _3 were 5.430, 3.940, and 3.860, respectively, and the axes had eigenvalues of 0.815, 0.517, and 0.391, respectively. Moreover, the total rate of variation explanation was 16.71%. In this study, Axes_1 and _2 were selected to explain the variation of vegetation because of the high explanatory power of the vegetation distribution. We described these vegetation types (Figure 3a) and different reefs (Figure 3b) as follows:

Figure 3a shows the ranking results of the vegetation types distributed among the 130 sample plots. On Axis_1, four vegetation types (Sageretia thea-Rhynchosia volubilis-Miscanthus floridulus type, Eurya emarginata-Euonymus japonicus type, Glochidion rubrum-Maclura cochinchinensis-Smilax china type, and Millettia reticulata-Grewia rhombifolia-Dendranthema indicum type) were distinguished from the other low herbaceous plant communities (as described in Table 2). On Axis_2, two vegetation types (including Aster asagrayi-Crossostephium chinense type and Setaria pallide-fusca-Zoysia tenuifolia-Crossostephium chinense-Chenopodium acuminatum subsp. virgatum type) were distinguished from two other vegetation types (Boerhavia diffusa-Crepidiastrum lanceolatum-Lysimachia mauritiana-setaria glauca type and Maytenus diversifolia-Dactyloctenium aegyptium-Tetragonia tetragonoides type), mainly due to the differences in the plant composition of the different reefs. The former two vegetation types were mainly distributed on Jhongdao Reef, and the latter two were mainly distributed on Shuangzih Reef. Three vegetation types (Artemisia capillaris-Chenopodium acuminatum subsp. Virgatum-Setaria glauca type, Chenopodium acuminatum subsp. Virgatum-Tetragonia tetragonoides type, and Asparagus cochinchinensis-Crepidiastrum lanceolatum type) were widely distributed on and shared by six reefs (Jhongdao, Sheshan, Lioucyuan, Baimiao, Sanlianyu, and Shuangzih Reefs). Hence, they were mostly located in the middle of Axis_2.

According to the distribution of the 130 sample plots on the different reefs of the Matsu Islands Tern Refuge (Figure 3b), the species composition and quantity in the sample plots varied with sole and dominant species on different reefs. On Axis_1, Jinyu and Tiejian Reefs were distinguished from six other reefs, indicating that these two reefs had the most diverse plant species among the eight reefs. In addition, the ranking diagram showed that Tiejian Reef had similar vegetation composition to that of Jinyu Reef and the six other reefs. Axis_2 roughly reflected the difference between Jhongdao Reef, Shuangzih Reef, and Baimiao Reef. The vegetation compositions of the Sanlianyu and Sheshan Reefs were distributed between them, showing that the composition of Jhongdao Reef was significantly different from that of Shuangzih and Baimiao Reefs.

As the length of the DCA Axis_1 was more than four SDs, the vegetation data and environment data were subjected to CCA. We computed 116 samples to continue the CCA section for the part of the samples that lacked soil at these reefs. The first three axis lengths of the CCA were 5.295, 3.870, and 3.883, respectively, and the eigenvalues were 0.810, 0.530, and 0.423, respectively (total explained variance was 6.23%). The variables with high correlation on the first axes and environmental factors were soil pH (r = 0.699, p < 0.05) and altitude (r = −0.623, p < 0.05), the variables with high correlation on the second axis were soil pH (r = 0.536, p < 0.05) and altitude (r = −0.484, p < 0.05), and the variables with high correlation on the third axis were soil total nitrogen (r = 0.725, p < 0.05) and soil organic matter (r = −0.697, p < 0.05). Although the first and second axes of the CCA explained the most variance (4.6%), the third axis was related to more environmental factors, so the bi-plot is presented with the first and third axes of the CCA (Figure 4). The distribution of the samples on the CCA bi-plot is consistent with the DCA results, but the first and third axes of the CCA are mirror symmetrical with the ordination diagram of the DCA.

3.3. Description of Plant Communities on Different Reefs

3.3.1. Shuangzih Reef

Shuangzih Reef is located at the northernmost end of the Matsu Islands Tern Refuge and was the nearest to the neighboring large island, Dongyin (Figure 1). On Shuangzih Reef, soil covered a small area and was only distributed on the flat recessions on sloping fields. In addition to plants growing on flat recessions with a thick soil layer, shrubs (e.g., Maytenus Diversifolia and Eur. emarginata) were found on hill ridges (Figure A2a). Crossostephium chinense was mainly distributed in steep rock crevices on the reefs (Append 2b). Tetragonia tetragonoides, Lysimachia mauritiana, Setaria glauca, Dactyloctenium aegyptium, and Zoysia tenuifolia were mostly distributed in places with rich soil, where several unhatched eggs were distributed (Figure A3a).

3.3.2. Baimiao Reef

Among the eight reefs, Baimiao Reef had a relatively small number of plant species. As the main body of the reef was too steep for boat docking, we only investigated its second-largest reef. On Baimiao Reef, as the area covered by soil was very small, plants could only grow in environments with rock debris and soil (e.g., rocky recessions and crevices), and the proportion of vegetation cover was tiny (Figure A2c). The main dominant plant species included Crepidiastrum lanceolatum, Allium grayi, Crossostephium chinense, and Zoysia tenuifolia (Figure A2d).

3.3.3. Tiejian Reef

On Tiejian Reef, the plant species were quite diverse, possibly due to the relatively large area of soil on the reef. On the windward side of northeast monsoon (e.g., northern and northeastern slopes) and the dry and low-soil environment on the southwest side, the herbaceous plant community (e.g., Chenopodium acuminatum subsp. virgatum and Artemisia capillaries) dominated and Calamagrostis epigeios became gradually dominant along the ridge line (Figure A2e,f). On the upslope of the south-to-southeast slope, we observed a zonally distributed shrub and vine plant community, which mainly comprised Smilax china, Rhaphiolepis indica var. umbellata, Br. officinalis, Gl. rubrum, Vitis thunbergii, and Am. brevipedunculata var. hancei intermingled with herbaceous plants such as Arisaema heterophyllum and Lycoris radiate (Figure A2f,g). On the mesoslope of the south-to-southeast slope, the dominant plant community comprised Dianella ensifolia, Lycoris radiate, and Dendranthema indicum (Figure A2g,h).

3.3.4. Jhongdao Reef

On Jhongdao Reef, the plant community comprised almost only a single dominant species (i.e., Crossostephium chinense), which was mainly distributed on hill ridges, along ridge lines, and on cliffs (Figure A2i). In areas with a thick soil layer near hill ridges or in corries, the plant community mainly comprised Chenopodium acuminatum subsp. virgatum and Aster asagrayi, which were interspersed with Crossostephium chinense, Tetragonia tetragonoides, Rhaphiolepis indica var. umbellata, Polygonum chinense, and Oxalis Corniculata (Figure A2j,k). Unhatched eggs, oviposited eggshells, or birdling carcasses were often found in Aster asagrayi-Crossostephium chinense type (Figure A3e).

3.3.5. Sanlianyu Reef

On Sanlianyu Reef, there were almost no woody plants, and several dominant species composed a mosaic plant community. Near the hilltops and on leeward slopes, Tetragonia tetragonoides (Figure A2l) and Chenopodium acuminatum subsp. virgatum were the dominant species of the plant community; in particular, Tetragonia tetragonoides was the dominant species. Crossostephium chinense revealed dominant coverage on the mesoslopes with shallow soil and in the rocky crevices of steep hill walls (Figure A2m). On the mesoslopes with thick soil, Oxalis corniculata, Chenopodium acuminatum subsp. virgatum, Polygonum chinense, and Allium grayi were dominant (Figure A2n,o). Unhatched eggs, oviposited eggshells, or birdling carcasses were often found in such plant communities (Figure A3b,c).

3.3.6. Jinyu Reef

Jinyu Reef is the largest reef in the Matsu Islands Tern Refuge. Compared to the seven other reefs, Jinyu Reef is composed of many large rocks with wide geological variation, which has also manifested in the relatively complex plant composition and structure (Figure A1f). On the hilltops, the high-grass plant community was dominated by Mis. floridulus, intermingled with a few vine plants (e.g., Sageretia thea, Rhy. volubilis, Za. nitidum, Rub. parvifolius, and Vitis thunbergii) and low shrubs (e.g., Br. officinalis and El. glabra) (Figure A2p). On the slopes, the dominant species included Eur. emarginata and Euo. japonicas, accompanied by patchy shrubs (e.g., Ficus erecta var. beecheyana, Pittosporum tobira, and Ardisia sieboldii) in Mis. floridulus, interspersed with large herbaceous plants (e.g., Mis. floridulus and Alpinia zerumbet), and accompanied by vine plants (e.g., Polygonum chinense, Ficus pumila, Lonicera japonica, and Coc. orbiculatus) (Figure A2q). In offshore areas, a narrow-zone plant community was dominated by Calamagrostis epigeios, Dianella ensifolia, Polygonum chinense, Peucedanum japonicum, and Dendranthema indicum, or vine plants (e.g., Smilax china, Ficus pumila, and Am. brevipedunculata var. hancei) (Figure A2r).

3.3.7. Lioucyuan Reef

Among the eight reefs, Lioucyuan Reef has the highest altitude and is the farthest from the neighboring large island (Figure A1g). Lioucyuan Reef is located amongst Nanga Island, Dongju Island, and Xiju Island (Figure 1), with low soil coverage except for a thick soil layer on the flat areas of hilltops. Across Lioucyuan Reef, Crossostephium chinense was dominant (Figure A2s) but was sparsely distributed only in the flat areas of hilltops. In these areas, Tetragonia tetragonoides, Setaria glauca, Solanum nigrum, Chenopodium acuminatum subsp. virgatum, and Boerhavia diffusa were dominant, and Zoysia tenuifolia, Lysimachia mauritiana and Oxalis corniculata were sporadically distributed, as well as accompanied by shrubs such as Rhaphiolepis indica var. umbellata (Figure A2t).

3.3.8. Sheshan Reef

Located in the southernmost part of the Matsu Islands Tern Refuge (Figure 1), Sheshan Reef is approximately 500 m away from the nearest large island (i.e., Xiju Island). In flat hilltops and areas with a thick soil layer, Tetragonia tetragonoides, Boerhavia diffusa, and Oxalis corniculata were dominant and interspersed with Setaria glauca, Coc. orbiculatus, Asp. cochinchinensis, and Liriope minor var. angustissima (Figure A2u). On gentle slopes, Mariscus cyperinus was dominant (Figure A2v). Crossostephium chinense was dominant in rocky crevices of steep hill walls or on slopes with little soil (Figure A2w); it was intermingled with Chenopodium acuminatum subsp. virgatum, Crepidiastrum lanceolatum, Setaria glauca, Zoysia tenuifolia, and Paspalum scrobiculatum and interspersed with Rhamnus lineate and Rhaphiolepis indica var. umbellate. In some areas, Zoysia tenuifolia and Paspalum scrobiculatum were dominant (Figure A2x).

4. Discussion

4.1. Vegetation Characteristics of Reefs

The reef environment and species composition significantly affected the Matsu Islands Tern Refuge plant communities. Vitousek et al. [38] indicated that the presence or absence of specific critical taxa can result in essential differences in ecosystems, particularly on islands. In our previous study, the dispersal syndromes of fruits/seeds for most of the species’ compositions were anemochory and zoochory, which disperse across long distances [19]. These species would have a higher probability of reaching the habitats of reefs via dispersal from nearby islands than dispersal by autochory. Furthermore, Tzeng et al. [19] found that the plant life-form spectrums of eight reefs were different, of which phanerophytes and hemicryptophytes plants dominated Jinyu Reef and Tiejian Reef, and the ratio of therophytes plant was the highest on Jhongdao Reef and Sanlianyu Reef. The characteristics of life form and growth dominate floristic compositions and determine the physiognomy, structure, and function of the essential characteristics of vegetation, which forms the major biological component of terrestrial island ecosystems. We found that vegetation physiognomy showed shrub and high-grass shrub plant communities with vegetation higher than 1.5 m on Jinyu Reef and Tiejian Reef. Moreover, the vegetation on Tiejian Reef tended to change gradually between the two low- and high-grass shrub vegetation types, while the other six reefs were covered with low-grass shrub vegetation.

Island biodiversity and levels of endemism probably depend on island size, geology, and isolation, among other variables [19,39]. Tzeng et al. [19] indicated that the floristic compositions of the Matsu Islands Tern Refuge are affected by the environmental characteristics of reefs, such as the position relative to the nearest largest island, size of the reef, topography, property and coverage of soil, etc. According to the CCA results, altitude, soil organic matter, soil total nitrogen, and soil pH were the critical abiotic factors of eight environmental factors affecting vegetation types. Although the overall altitude range of the vegetation distribution was small (10.0–60.5 m), the plant species distribution differentiated along the altitude of the reefs and were distributed at higher altitudes, such as up slope hills and on hilltops. Those species comprised most vegetation types, such as Maytenus diversifolia, Dactyloctenium aegyptium, Tetragonia tetragonoides type, and Aster asagrayi-Crossostephium chinense type.

Seabirds influence terrestrial vegetation capably by altering edaphic conditions with guano, imposing physical disturbance, and affecting seed dispersal [40,41]. In the Matsu Islands Tern Refuge, we found that vegetation with heights of less than 60 cm, such as Maytenus diversifolia-Dactyloctenium aegyptium-Tetragonia tetragonoides type, Boerhavia diffusa-Allium macrostemon type, Asparagus cochinchinensis-Crepidiastrum lanceolatum type, Setaria pallide-fusca-Zoysia tenuifolia type, Chenopodium acuminatum subsp. Virgatum-Tetragonia tetragonoides type, and Aster asagrayi-Crossostephium chinense type, could be destroyed by terns for nesting, and leftover carcasses of birds or unhatched eggs were found in the habitat (see discussions below). Furthermore, we analyzed the soil properties and found that the soil pH of vegetation inhabited by terns was lower and that the organic matter and total nitrogen of the soil were higher than those for vegetation types with a height of more than 1.5 m, such as Sageretia thea-Rhynchosia volubilis-Miscanthus floridulus type, Eurya emarginata-Euonymus japonicus type, Glochidion rubrum-Maclura cochinchinensis-Smilax china type, and Millettia reticulate-Grewia rhombifolia-Dendranthema indicum type. These vegetation types were rarely found to have traces of tern inhabitation (e.g., carcasses of birds or unhatched eggs; see discussion below).

Although the guano of seabirds is generally alkaline [42,43], guano decomposition in the soil often results in increased soil acidity [43,44]. Waits et al. [45] compared the soil conditions between the “bird islands” and “non-bird islands” in the Gulf of California and found that six of the “bird islands” receive seabird guano deposition. Species richness was significantly lower on islands with guano (bird islands) than islands without guano (non-bird islands), with soil properties of significantly high nutrients, soil moisture, and respiration, but lower soil pH on bird islands than on non-bird islands. Zwolicki et al. [46] found that with increasing distance from bird colonies, the concentration of nutrients and soil conductivity decreased, while pH increased. The vegetation zones were related to this gradient of seabird colony influence.

We inferred that the soil properties would be affected by the activities of terns in the Matsu Islands Tern Refuge. Then, the influences further affect the compositions of vegetation [45,46,47]. Aerts et al. [47] pointed out that the soil in their study area of seagull activity is rich in nitrogen and phosphorus, so the suspended particles in the air and seagull feces are transferred to the volcanic ash soil and then absorbed by plants; this phenomenon is also reflected in the CCA bi-plot above the third axis. As mentioned in this article, Sanlianyu Reef is an area where terns are more active, so the vegetation type distributed on the island (mainly VII. Chenopodium acuminatum subsp. virgatum-Tetragonia tetragonoides type and VIII. Setaria pallide-fusca-Zoysia tenuifolia type) is rich in nitrogen, which is consistent with the current situation of the plot described by the section of vegetation classification, which proves that the area with frequent seabird activity has a higher nitrogen content. Furthermore, the increase in soil organic matter is also highly correlated with the activity of seabirds [47], and the relationship between the N content of organic matter and soil is reciprocal. The increased fertility of the soil significantly improves the soil conditions of the local islands, which in turn increases opportunities for colonization and growth of plant communities [48]. Although many studies indicate that plant dispersal mechanisms are often closely related to the activity of birds [47,48,49,50], terns primarily feed on fish [51]. Therefore, plant dispersal might occur through the transport of plant seeds by tern feathers to other reef islands. However, this process is highly random, and terns’ movements, which often involve wetting their bodies from hunting fish at sea, may reduce seed germination rates [48]. Overall, this study does not provide evidence to reveal the dispersal mechanisms of plants on reef islands.

4.2. Relationship between Plant Communities and Tern Habitat Environments

During the annual breeding season from May to August, the Matsu Islands, in addition to being world-renowned fishing grounds, become an important place where tens of thousands of terns gather to feed and breed. The Matsu Islands Tern Refuge aims to conserve the terns that inhabit the different reefs of the refuge. Although terns mainly feed on sea fishes, they must return to land for resting and breeding. However, do terns choose special plant communities for resting, feeding, breeding, and shelter from enemies? Because no plant community survey was conducted during the tern breeding season for this study, we only conducted a preliminary discussion using the results of plant community grouping combined with the photos of unhatched tern eggs and birdling carcasses taken in the field in 2010 (Figure A3). The results showed that the largest number of unhatched eggs and carcasses were found in the Aster asagrayi-Crossostephium chinense type community on Jhongdao Reef (Figure A3e); this may be related to the fact that Jhongdao Reef had the highest density of tern gathering in 2010 (data source: Lianjiang County’s government). Aster asagrayi and Tetragonia tetragonoides in the Aster asagrayi-Crossostephium chinense type often wither due to trampling by nesting terns, whereas Crossostephium chinense survives due to inhabiting woody shrubs.

Chenopodium acuminatum subsp. Virgatum-Tetragonia tetragonoides type on Sanlianyu Reef was the plant community where the second largest number of unhatched eggs or young carcasses was found (Figure A3b,c), and Tetragonia tetragonoides also withered due to trampling. A similar situation occurred for the Maytenus diversifolia-Dactyloctenium aegyptium-Tetragonia tetragonoides type on Shuangzih Reef; Tetragonia tetragonoides, a dominant ground cover species, withered on a large scale (Figure A3a). In other low-shrub plant communities of the Matsu Islands Tern Refuge, unhatched eggs or birdling carcasses were scarcely found. In addition, unhatched eggs or birdling carcasses were occasionally found below the Crossostephium chinense plants growing on the steep cliffs of reefs. However, no excreta, unhatched eggs, or birdling carcasses of terns were found in the high-grass plant communities on Jinyu or Tiejian Reefs, whereas we found abundant tern excreta on the scarcely vegetated reef rocks offshore on Jinyu Reef.

The Wild Bird Society of Taipei [52] found that Tiejian Reef was the main reef where CCTs gathered to breed in 2016. Specifically, CCTs bred and nested at the top of Tiejian Reef in early June, and the hatchlings moved to the shrubs around the top of Tiejian Reef to hide after they hatched out on the north side of the reef in early July. Yuan [53] argued that slope and vegetation cover are the main factors affecting terns’ breeding. Our analysis of the slope and vegetation cover of Tiejian Reef showed that at the northern end of the reef, the habitat is quite flat, the area of short-grass sloping field is large, and abandoned eggs were relatively few, indicating that the habitat at the northern end is relatively good for nesting. According to the results of a vegetation survey in 2010 and aerial photos in a survey report published by the Wild Bird Society of Taipei in 2016, terns predominantly chose to breed and nest in the herbaceous vegetation types of Artemisia capillaris-Chenopodium acuminatum subsp. Virgatum-Setaria glauca and Millettia reticulata-Grewia rhombifolia-Dendranthema indicum on Tiejian Reef. The shrubs of the Glochidion rubrum-Maclura cochinchinensis-Smilax china type were the places for sheltering and feeding tern hatchlings.

5. Conclusions

Because the nesting environment of terns in the breeding season was not observed in this study, we assumed—based on the 2010 vegetation survey and the unhatched eggs or dead hatchlings under different vegetation types—that terns seemingly prefer to breed in low-grass shrub and herbaceous plant communities with both soft-structured herbs and some shrubs. Specifically, low-grass shrubs and herbaceous plants provide relatively soft nesting materials, and shrubs provide a specific sheltering effect for eggs or hatchlings. Compared to low-grass shrubs, the traits of high-grass shrubs would not be beneficial to nests for the breeding of terns on the ground. However, terns rarely nested in specific low-grass shrub plant communities on the reefs. This would be influenced by the number of terns that gather, breed, and nest annually on appropriate reefs or affected by environmental factors, characteristics, and structures of plant communities. In order to understand the impact of the composition of island reef vegetation on tern colony and breeding habitat selection, further long-term investigation and monitoring are necessary.