Tenerife, Canary Islands - Mountain Invasion Research Network

3 downloads 3 Views 921KB Size Report
Jul 11, 2013 - to singular island ecology and topography (Ullmann et al. 1998) ..... de Nascimento L, Willis KJ, Fernández-Palacios JM, Criado C, Whittaker RJ ...

Folia Geobot (2014) 49:65–82 DOI 10.1007/s12224-013-9159-z

Road Edge Effect and Elevation Patterns of Native and Alien Plants on an Oceanic Island (Tenerife, Canary Islands) Rüdiger Otto & Manuel A. Arteaga & Juan D. Delgado & José R. Arévalo & Cristina Blandino & José M. Fernández-Palacios

# Institute of Botany, Academy of Sciences of the Czech Republic 2013

Abstract We studied road edge effects on floristic composition and richness of alien and native plants on five zonal ecosystems, following a steep altitudinal gradient from arid coastal and mid-elevation scrublands, through laurel and pine forests, to summit scrub, on Tenerife (Canary Islands). We analyzed vegetation within transects running from the road edge to the core of natural habitats. Alien richness significantly decreased with distance to road edge for most ecosystems. Native richness also decreased with increasing road distance for the coastal scrub and pine forest, but increased for the thermophilous scrub. We found a decrease in both native and alien species richness with elevation. Our results suggest that road edge effects in alien plant invasion are stronger in native shrub communities at low elevations than in forests (laurel and Canary Island pine forest), where aliens were limited to a narrow road edge band. Detrended correspondence analyses showed that road edge plots were floristically very different from interior plots and that each ecosystem harboured a specific alien assemblage with few species present in more than one ecosystem, suggesting a marked species turnover of roadside alien species across altitudinal belts up to the pine forest. However, at high elevations, very few aliens invaded roadsides, probably due to harsh environmental conditions and still relatively low propagule pressure. Keywords Disturbance . Elevation . Island scrublands . Oceanic islands . Roadside assemblages

Introduction Roads favor dispersal of alien plant species and may connect inaccessible conserved natural areas with alien propagule sources such as urban areas or intensively used R. Otto : M. A. Arteaga : J. R. Arévalo : C. Blandino : J. M. Fernández-Palacios Ecology Department, University of La Laguna, 38206 Tenerife, Canary Islands, Spain J. D. Delgado (*) Ecology Area, Department Physical, Chemical and Natural Systems, University Pablo de Olavide, 41013 Seville, Spain e-mail: [email protected]

66

R. Otto et al.

agricultural land, accelerating the invasion process (Tyser and Worley 1992; Parendes and Jones 2000; Pauchard and Alaback 2004). Invasibility of natural habitats, and hence richness and compositional patterns within ecosystems, depends on accessibility and suitability for aliens (Davis et al. 2000). Road edges are chronically disturbed by human activities, which frequently result in native plants being displaced, or locally extinguished, and the area invaded by certain alien species (Pauchard and Alaback 2004; Hansen and Clevenger 2005). Furthermore, this disturbed habitat is relatively homogeneous with respect to abiotic and biotic conditions, with exception of meso-climate (Wilson et al. 1992), which favors the spread of alien species. The creation of abrupt ecological boundaries along roads has received remarkable attention on continents, and, to a lesser extent, on oceanic islands. Roads on islands might affect adjacent natural habitats in different ways compared to continental road systems due to singular island ecology and topography (Ullmann et al. 1998). Some alien species can be aggressive invaders on oceanic islands and may cause deleterious damage to native communities (Reaser et al. 2007), but only relatively few aliens are consistently invasive across oceanic archipelagos (Kueffer et al. 2009; Seipel et al. 2012). In general, the number of invaders per island group is well predicted by the degree of human development, habitat diversity, island age, and geographic region. Roads contribute to dispersal facilitation by enabling alien plants to expand their altitudinal ranges and reach equilibrium with their climatic limits on islands, if residence time is long enough (Ullmann et al. 1998; Alexander et al. 2009; Haider et al. 2010). Hence, roads on oceanic islands can serve, at the same time, to investigate the spread of aliens from the road edge into adjacent natural habitats, i.e., ecosystem invasibility, but also climate adaptations or climate matching of alien species along elevation gradients (Alexander et al. 2009; Haider et al. 2010; Jakobs et al. 2010). Urbanization pressures and transportation (e.g., Whittaker and Fernández-Palacios 2007) are heavily altering oceanic islands worldwide. The Canary Islands have the highest road density of all the European islands (6 km/km2) (Martín and FernándezPalacios 2001), and in Tenerife, the road network (primary and secondary roads) occupies 3 % of the area (a conservative figure based only on paved roads mainly on non-protected territory, ca. 8,600 km of paved roads on 2,034 km2). The road net traverses all the five major altitudinal vegetation belts: coastal scrub, thermophilous scrub, laurel forest, Canarian pine forest and summit scrub. Vegetation has changed on the Canary Islands since the arrival of the aborigines about 3,000 years ago (de Nascimento et al. 2009), but human activities since the Spanish conquest have especially facilitated alien plant invasions (Arévalo et al. 2005; Arteaga et al. 2009; Kueffer et al. 2009). Few areas remain free of alien plants. In other oceanic islands (e.g., Hawaii: Vitousek 1992) ubiquity of alien plants prompts alien invasion of even naturally disturbed habitats. On other mountain ecosystems (i.e., South Africa: Kalwij et al. 2008a,b; New Zealand: Wilson et al. 1992) urban settlements act as propagule sources and roads enable the alien plants to reach otherwise unconnected land, thus promoting the dispersal of these alien plants. Road edge effects have been studied in the Canaries regarding environmental variables, vegetation and floristic changes (Delgado et al. 2007; Arévalo et al. 2005, 2008; Arteaga et al. 2009; Haider et al. 2010; Alexander et al. 2011). However, these works did not exclusively focus on roads in conserved natural vegetation, but were carried out within highly disturbed landscapes in the lower and intermediate part and more natural habitats in the upper part of the existing elevation gradient. They showed that both the climatic pre-adaptation of aliens (lowland climatic filtering)

Road Edge Effect and Elevation Patterns of Natives and Aliens

67

and the habitat context are important factors shaping the distribution of these species along the elevation gradients on the Canary Islands. Furthermore, invasive processes were more frequent and intensive near urban nuclei (Arévalo et al. 2005). In general, intensive land uses such as urbanizations and cultivated or deforested areas strongly affect road-side communities (Wester and Juvik 1983; Celesti Grapow and Blasi 1998; Gelbard and Belnap 2003), because for alien plant populations these areas serve as seed sources that facilitate their expansion (Milton et al. 1998; Qian and Ricklefs 2006). Recent work on islands and mountainous systems worldwide have shown that plant species composition, and richness of aliens follow interacting gradients from global to local scales (Kueffer et al. 2009; Alexander et al. 2011). The main forces are global-scale dispersal or/and propagation caused by human activities, elevation at the regional scale, and disturbance at the local scale (Seipel et al. 2012). Our aim was to study distribution of alien plants in the context of two major interacting factors shaping structure and composition of island floras at several scales, the altitudinal organization of native ecosystems and the road edge effect. Our concrete objectives were i) to assess road edge effects on composition and richness of natives and aliens within natural habitats along the major elevation gradient on Tenerife; ii) to compare, for each ecosystem, the magnitude and pattern of changes in alien and native richness from roads to core habitat; and iii) to evaluate the effect of road side characteristics on the distribution of alien species.

Material and Methods Study Area The study was conducted between June–September 2006 within the five main ecosystems on Tenerife (28.28° N, 16.15° W, 2,034 km2, 3,718 m maximum elevation, Teide peak). The prevailing trade winds create a contrast between humid NE (windward) slopes and the more arid SE–SW (leeward) slopes (Fernández-Palacios 1992). The main climatic feature is the orographic cloud layer caused by an inversion in temperature and humidity on the windward slopes at ca. 1,200 m a.s.l. (Fernández-Palacios 1992). The natural vegetation is structured altitudinally from coast to summit in five main vegetation belts, namely coastal scrub, thermophilous scrub, laurel forest (chiefly at the windward slope), Pinus canariensis forest and summit or high mountain scrub. Vegetation types are not strictly discontinuous along the elevation gradient (Fernández-Palacios and de Nicolás 1995). On Tenerife, discontinuity in the vegetation is more sharply defined on the windward slope than on the leeward slope (FernándezPalacios 1992), where it mostly resembles a vegetation continuum or coenocline. The coastal scrub on the Anaga massif is dominated by Euphorbia canariensis and E. lamarkii, Lavandula canariensis and Aeonium spp., with secondarily grown Artemisia thuscula, Hyparrhenia hirta and Cenchrus ciliaris. This belt is located between 0 and 300 m a.s.l., although it may reach 450 m on drier slopes. Mean annual temperature is 21°C; mean annual precipitation ranges between 250–350 mm with a strong hydric stress and water deficit between May–October. The thermophilous scrub follows the coastal scrub between 300–500 m a.s.l. on the leeward slope of Anaga. Mostly shrubs (Globularia salicina, Hypericum

68

R. Otto et al.

canariense, Teline canariensis, Periploca laevigata, Jasminum odoratissimum, Olea cerasiformis, and Juniperus turbinata subsp. canariensis, and others) characterize it. Mean annual temperature reaches 17.6°C and annual precipitation 350–450 mm. Insolation is moderate and the water deficit occurs between March-October. The laurel forest (laurisilva) lies between 500–950 m a.s.l. in Anaga. Dominant trees are Laurus novocanariensis, Myrica faya, Erica arborea, Ilex canariensis, Prunus lusitanica, Apollonias barbujana, Picconia excelsa and Persea indica. Mean annual temperature is 16.2°C, mean annual precipitation attains 800 mm but exceeds 1,500 mm when including horizontal fog drip. There is almost no hydric stress and insolation is lower than in the former ecosystems (Aschan et al. 1994). The pine forest at the study area ranged between 1,300–1,600 m a.s.l. on the road from La Laguna to Las Cañadas del Teide. It is a tall (up to 30 m height) formation dominated by Canarian pine (Pinus canariensis), with a sparse understory of Adenacorpus foliolosus, Myrica faya and Erica arborea. There is a moderate thermic stress with mean annual temperature between 10–15°C, and precipitation between 500–600 mm. Insolation is variable depending on elevation, inclination and aspect, and there is a water deficit between August–October. The summit scrub is located above 2,000 m a.s.l. We selected two road segments within the subalpine zone of the Cañadas del Teide National Park, one on the ridge near the Izaña Observatory (E of the Park), and the other near the Roques de García and Parador (Park center). Vegetation within the park is formed mostly by local endemics, with Adenocarpus viscosus, Spartocytisus supranubius, Descurainia bourgeauana, and Pterocephalus lasiospermus dominating. Mean annual temperature is 9.8°C; mean annual precipitation is 440 mm. Insolation is very intense, with a high thermic stress and moderate hydric stress. The soil is water-deficient between June–October, and frost is common above 1,800 m a.s.l. in winter. The Road Scheme and Traffic Data Tenerife is highly populated (ca. 900,000 inhabitants). Because the larger villages are located mainly near the coast and at mid-elevations, human pressure decreases with elevation, with the lowest human population density above 1,000 m. Road density on Tenerife varies between 0.19–2.18 km/km 2 in the most conservative estimate, being lower in mountainous areas and protected natural areas. The studied road segments were regional two-lane roads, on average 7 m wide and traffic density in from 1999–2004 varied between 700–2,000 vehicles/day (Martín and Fernández-Palacios 2001; Cabildo de Tenerife 2004). Elevation in our study roads ranged from 143 (coastal scrub) to 2,114 m a.s.l. (summit scrub). Elevation and aspect cause micro and mesoclimate variation among altitudinal belts and between slopes. A single road encompassing three ecosystems from coastal scrub up to laurel forest through thermophilous scrub was surveyed on the leeward slope of Anaga (NE Tenerife) from San Andrés to El Bailadero (964 m a.s.l.). Pine forest and summit scrub road transects were respectively set 25 and 40 km west of Anaga to the center of the island (see Fig. 1). In all, transects were separated at least by 500 m within a road segment (distance measured on the road) to minimize spatial dependence.

Road Edge Effect and Elevation Patterns of Natives and Aliens

69

Fig. 1 Study area and location of road-interior transects (four transects per ecosystem; symbols are cluttered due to scale)

Sampling Design Transects were placed where habitat was accessible from the road avoiding cliffs and other vertical obstacles. We selected comparable areas to establish transect replicates minimizing within-ecosystem variation (within-ecosystem beta diversity), a source of variation potentially interacting with edge effects. In each ecosystem, we placed four road edge-interior transects perpendicular to the road. Four plots sized 5 × 5 m were installed on each transect at 0 (= road edge), 25 m, 50 m and 100 m distance from the asphalt margin towards the interior, giving 16 plots/ecosystem and 80 plots overall. We performed 100-m long transects to limit topographical variation and avoid overlapping roads. On each plot we recorded all vascular plants, annuals included, by species and status (alien or native), and species richness. We classified plants as natives (including endemics) and aliens. We used recent checklists for classification (Acebes et al. 2004; Stierstorfer 2005; Stierstorfer and von Gaisberg 2005; von Gaisberg 2005). Species cover at the ground level was estimated on a scale of 1–10 (percent cover classes): 1) traces; 2) 90 %) (Arévalo et al. 2005). At each plot we noted coordinates (UTM), elevation (m a.s.l.), aspect of the slope and inclination (º), substrate type cover (rock, bare soil, leaf litter, and total vegetation were expressed as percentage). We also estimated disturbance intensity within plots and classified it using a semi-quantitative (categorical) scale with the following classes: very high (5), high (4), moderate (3), little (2), and absent (1). We used these

70

R. Otto et al.

disturbance proxies: i) mechanical impact by vehicles and pedestrians (called “trampling”, inspecting road edges for foot and tire prints); ii) roadside management (including one or more of the following: presence of concrete/asphalt residues, furrows, soil substrate removed, cuts of road shoulder vegetation); iii) trash dumping (visual categorical estimate of ordinary rubbish, including plastics, envelopes, etc.). Statistical Analysis We applied generalized linear mixed-effects models to predict richness of alien and native species (GLMMs, using the R package lme4; Bates and Sarkar 2007). GLMMs are extensions of generalized linear models (GLMs, McCullagh and Nelder 1989), including random effects to deal with correlated data structures, in particular, with clustered, spatial autocorrelated and/or longitudinal data (Hedeker 2005; Bolker et al. 2009). We selected a Poisson error distribution using a log-link function, as recommended for count data when dependent variables and residuals of the models were not normally distributed (Zuur et al. 2007). We introduced total richness, richness of alien and native species as dependent variables, and distance to road, transect elevation, ecosystem, trash, trampling and road management (disturbance proxies) as fixed factor in the model, while road and transect identity were included as random factors. We also checked for interaction between elevation and distance to road. AIC (Akaike information criteria) was used to obtain the optimal set of predictor variables, choosing the lowest AIC value for every possible combination of explanatory variables. We first constructed GLMMs for each species group (total, alien, native) including all fixed and random factors and all ecosystems, with exception of the very species poor summit scrub. We then separately run GLMs for each ecosystem using distance to road as the only explanatory factor, not considering random factors because there is no road effect within each habitat and spatial dependence of plots is considerably reduced. GLMMs and GLMs were performed in R (R Development Core Team 2008). Ordination techniques help to describe and interpret variation in plant community structure and composition along environmental gradients (Gauch 1982; ter Braak and Šmilauer 1998). We used detrended correspondence analysis (DCA) within the CANOCO Package to explain variation in species composition from road edge to interior (ter Braak and Šmilauer 1998). DCA ordination diagrams were separately constructed for the four ecosystems studied, including all vascular plant species. Additionally, we analyzed composition of alien species at the road edges for all ecosystems, except the summit scrub. Detrending by segments, with no data transformation and down weighting of rare species were selected as options in the CANOCO package. Due to very low species richness, summit scrub was not included in the statistical analysis.

Results Species Richness We recorded 201 vascular plant species, of which 78 (38.8 %) were endemic to the Canary Islands, 69 (34.3 %) were native non-endemic species and 54 (26.87 %) were aliens. Species pools differed considerably between ecosystems (Table 1), with the lowland

Road Edge Effect and Elevation Patterns of Natives and Aliens

71

Table 1 Species pools of endemic, native and alien species for the five main altitudinal ecosystems on Tenerife Ecosystem

Coastal scrub

Thermophilous scrub

Laurel forest

Pine forest

Summit scrub

Endemics

30

33

Native non endemics

41

42

26

7

6

13

19

Total natives

71

75

0

39

26

6

Aliens

21

34

13

8

1

Total

92

109

52

34

7

scrubs being very rich and the summit scrub very poor in species for all species groups. For all ecosystems alien species richness was highest in the road edge plots (Table 1, Fig. 2), the same holds true for the native species richness with the exception of the thermophilous scrub, where richness peaked at the interior plots. This trend was especially pronounced for the pine forest, where understory vegetation was very scarce in the interior of the forest. We only found one alien species (Bromus tectorum) in the summit scrub road edge plots, whereas 34 non-natives were recorded in the thermophilous scrub plots. Alien species formed between 14 % and 31 % of the overall ecosystem species pool. There were only few alien plant species reaching more than 5 % of cover in at least one ecosystem plot: Opuntia maxima, Rubus ulmifolius, Nicotiana glauca, Crassula lycopodioides, Scorpiurus muricatus, Echium plantagineum, Crocosmia ×crocosmiiflora and Poa annua. Only the first two species were recorded with more than 5 % of cover further from the road edge (>50 m distance) at the two low-altitude scrubs. There was Increasing elevation Total richness

COASTAL SCRUB

THERMOPHIL. SCRUB

LAUREL FOREST

PINE FOREST

SUMMIT SCRUB

40 30 20 10

Alien richness

0 0

25 50 100

0

25

50 100

0

25

50 100

0

25

50 100

0

25 50 100

0

25 50 100

0

25

50 100

0

25

50 100

0

25

50 100

0

25

50 100

0

25 50 100

0

25

50 100 0 25 50 100 0 Distance to road edge (m)

25

50 100

0

25

50 100

12 8 4

Native richness

0

30 20 10 0

Fig. 2 Richness of all vascular plants, native and alien species with respect to distance to road and along the elevation gradient represented by the five main altitudinal ecosystems. Box-and-whisker plots are shown: boxes − upper and lower quartile, whiskers − range minimum-maximum, horizontal line − median

72

R. Otto et al.

only one alien species (1.9 %) shared by all four ecosystems (Sonchus oleraceus, typical cosmopolitan ruderal herb of Mediterranean origin with a wide ecological niche), and one shared by three ecosystems (Bidens pilosa). Problematic invaders (species for which negative effects on the island native biota and/or landscape have been reported; see SanzElorza et al. 2005) included Conyza bonariensis, Nicotiana glauca, and Opuntia maxima (mainly but not limited to coastal and thermophilous scrub), and Crocosmia ×crocosmifolia, Oxalis pes-caprae and Rubus ulmifolius (mainly in laurel forest). Results of GLMMs including all ecosystems revealed that elevation was the best predictor for overall species richness followed by distance to road and road management (Table 2). Regarding only alien species, distance to road had the strongest influence on richness, followed by elevation and road management. Native richness was strongly affected by elevation and much less by distance to road, whereas road management was not included as predictor in the final model after AIC selection (Table 2, Fig. 2). The largest variation for alien and overall richness occurred within the first 25 m from the road margin to the habitat interior (Fig. 2). No significant interactions between elevation and distance to road were found. Richness patterns were different for each of the analyzed ecosystems. Although alien richness significantly decreased with distance to road in all ecosystems (Table 3), laurel and pine forests were almost free of non-natives 50 m from the road, whereas alien plants penetrated much further into lowland scrubs (Fig. 2). Hence, the effect of road distance on alien richness seems to be stronger for forests than for lowland scrubs. Native richness decreases with distance to road in the coastal scrub and pine forest, but increased in the thermophilous scrub.

Table 2 Results of generalized linear mixed-effect models (GLMM) with total richness, richness of alien and native species as dependent variables and distance to road, elevation, trash, trampling and road management as fixed factors. Road and transect identity were introduced as random factors. Only the optimal sets of significant predictors selected by the AIC (Akaike information criteria) are shown Model

Estimate

z value

Pr(>|z|)

Intercept

3.197

18.380