Interactions between global change components

1 downloads 0 Views 1MB Size Report
Sep 10, 2018 - AIC, the adjusted determination coefficient R2 (Radj. 2) and model comparison ..... vLATE Extension für ArcGIS – vektor- basiertes Tool zur ...
|

|

Received: 27 April 2018    Revised: 13 August 2018    Accepted: 10 September 2018 DOI: 10.1111/jvs.12683

Journal of Vegetation Science

RESEARCH ARTICLE

Interactions between global change components drive plant species richness patterns within communities in mountain grasslands independently of topography Antonio Rodríguez1,2

 | Xavier de Lamo2 | Maria-Teresa Sebastià1,2

1 Laboratory of Functional Eology and Global Change (ECOFUN), Forest Sciences Centre of Catalonia (CTFC), Solsona, Spain 2

Group GAMES and Department of Horticulture, Botany and Landscaping, School of Agrifood and Forestry Science and Engineering, University of Lleida, Lleida, Spain Correspondence Antonio Rodríguez, Laboratory of Functional Eology and Global Change (ECOFUN), Forest Sciences Centre of Catalonia (CTFC), Solsona, Spain. Email: [email protected] Present address Xavier de Lamo, UN Environment World Conservation Monitoring Centre (UNEPWCMC), Cambridge, UK Funding information University of Lleida, Grant/Award Number: Ph.D. Fellowship; Spanish Science Foundation, Grant/Award Number: BIOGEI: GL2013-49142-C2-1-R Co-ordinating Editor: Beverly Collins

Abstract Questions: How do interactions between global change factors (climate and land use, including livestock management) shape plant species richness patterns in mountain grassland communities? Does topography interact with global change factors to modulate their effect on within-­community plant diversity? Location: Pyrenees, Spain. Methods: We used an initial set of 20 predictors: climatic, biogeographic, livestock management and topographic. Our data set included a wide range of management and climatic conditions from mediterranean to alpine environments. After a variable selection procedure with random forest, we built GLM to explain species richness in plant communities through 100-­m2 plots (SR100), employing backward–forward selection with AIC and other techniques. Results: The main filters of SR100 were the regional factors (climate and biogeography, with 23% and 17%, respectively, of the total contribution to SR100), followed by livestock management (14%) and topography (9%). Interactions between climatic and biogeographic variables were almost as important as the main effects (18%). Fragmentation effects were higher under low mean minimum temperatures and summer precipitation compared with other conditions. Connectivity interacted with most of the climatic variables. Moderately high stocking rates mitigated plant community species losses triggered by decreased connectivity. Sheep-­grazed grassland communities had lower SR100 than differently managed grasslands under low fragmentation scenarios; conversely, sheep grazing enhanced SR100 when fragmentation was high. Topographic predictors accounted for lower variability in SR100 than regional and management factors and were independent from these. Conclusions: Several multi-­scale spatial filters determine SR100 of grassland communities in the Pyrenees, interacting through complex processes. Climate can enhance negative effects of fragmentation and lack of connectivity on SR100 under stressful conditions. Livestock management modified the impact of patch disaggregation and fragmentation on SR100. KEYWORDS

climate change, eastern Pyrenees, grazing ecology, habitat fragmentation, mountain grasslands, plant species richness

J Veg Sci. 2018;1–11.

wileyonlinelibrary.com/journal/jvs   © 2018 International Association |  1 for Vegetation Science

| Journal of Vegetation Science

RODRÍGUEZ et al.

2      

1 |  I NTRO D U C TI O N

In addition to these reported effects of specific global change components, the existence of interacting effects is widely assumed

Global change includes alterations of the atmosphere and oceans,

(Newbold et al., 2015), although such interactions are less frequently

and local changes so common that they can be considered of global

tested explicitly (Mantyka-­pringle, Martin, & Rhodes, 2012). The

importance, including changes in land use or biodiversity (Steffen

study of these potential interactions between global change com-

et al., 2015). Global change is particularly worrying for European

ponents is critical to prevent synergistic effects that lead to more

mountain grasslands because of their potential vulnerability to an-

serious biodiversity loss (Ferger et al., 2017). In addition to global

thropogenic climate change (Dullinger et al., 2012), and also because

change factors, other abiotic landscape factors can structure vege-

of the widespread changes in land use promoted by global socio-

tation. In the eastern Pyrenees, topography affects soils and act as

economic changes (Hülber et al., 2017). These global trends pro-

a strong driver of grassland vegetation at landscape scales (Sebastià,

mote abandonment of the less productive land, commonly located

2004).

in high-­altitude areas, and intensification of the most productive and

In this paper, we focus on how different elements of global

accessible areas, mainly those located at the valley bottoms (Lasanta

change and their interactions could shape within-­community plant

et al., 2017).

diversity of mountain grasslands in the eastern Pyrenees, and assess

One of the main implications of land abandonment in mountain

the role of local factors, including topography, as regulators of global

grasslands is the re-­wilding by natural succession, with an expansion

change effects. We focus on plant species richness in 100-­m 2 plots

of shrub and forest cover, which can reduce open spaces and grass-

(SR100) as a proxy for within-­community plant diversity, and consid-

land biodiversity (Lasanta, Nadal-­Romero, & Arnáez, 2015). Most of

ered four categories of predictors: climatic, biogeographic, grazing

those grasslands are distributed below the tree line and have been

management and topographic variables.

used as feed for livestock since Neolithic times (Poschlod & Wallis

We chose species richness as a measure of biodiversity that

DeVries 2002). In addition, this unleashes a habitat fragmentation

tends to be positively correlated with several ecosystem func-

process that can have negative consequences for grassland plant

tions (Ma & Chen, 2016). In fact, studying the multivariate drivers

biodiversity and other services provided by mountain grasslands

of species richness is a momentous aim for ecologists (Fraser et al.,

(Cojoc, Postolache, Olariu, & Beierkuhnlein, 2016; Lasanta et al.,

2015). As species richness measurements are sensitive to sample

2015).

size and sampled area (Brown & Williams, 2016), we use plant spe-

Grasslands from the eastern Pyrenees are a good example of this

cies richness within communities, measured through 100-­m2 plots,

process, because in the last decades the tendency is strong towards

and investigate how global change components affect the number

abandonment (Lasanta-­Martínez et al. 2005); this implies that the

of species within those grassland plant communities. Therefore, we

more accessible grasslands are maintained at relatively high stock-

use SR100 as an indicator of plant community structuring patterns.

ing rates, while the most remote are scarcely used. Therefore, the

In particular, the specific questions in this study are: (a) how do inter-

forest advances and the grassland patches become smaller, more

actions between global change factors (climate and land use, includ-

fragmented and disconnected than in former times (Ameztegui et al.

ing livestock management) shape plant species richness patterns in

2016). The biogeographic pattern here is thus tightly dependent on

mountain grassland communities; and (b) does topography interact

land use and management.

with global change factors modulating their effect on plant diversity?

In this region, land abandonment has been accompanied by changes in livestock management, with a severe decrease in stocking rates and the substitution of sheep by cattle as the main

2 | M E TH O DS

livestock species (Lasanta et al., 2017). High grazing intensity can increase plant diversity by promoting landscape heterogeneity (de Bello, Lepš, & Sebastià, 2007) and, at the community scale,

2.1 | Location and sampling design

by facilitating the entrance of ruderal species (Canals & Sebastià,

The set of data used in this study was extracted from the PASTUS

2000). Furthermore, different grazer species can affect plant di-

Database (http://ecofun.ctfc.cat/?p=3538), which was compiled by

versity differently, depending on how they use the pasture. In the

the Laboratory of Functional Ecology and Global Change (ECOFUN)

Pyrenees, cattle are reported to increase plant diversity at the plot

of the Forest Sciences Centre of Catalonia (CTFC) and the University

scale by reducing populations of the most competitive species due

of Lleida (UdL), Spain. We sourced a wealth of data for 118 grass-

to their generalist feeding habit, while the preference of sheep

land patches distributed across the eastern Pyrenees, including

for palatable plants can produce a strong vegetation homogeniza-

geographic, topographic, climatological, management and plant di-

tion effect (Sebastià, de Bello, Puig, & Taull, 2008). On the other

versity variables, as explained below.

hand, sheep are described as outstanding seed dispersers by

The sampled area (Figure 1; Table 1) encompasses a wide vari-

means of epi-­ and endozoochory (Liu et al., 2016; Will, Maussner,

ety of temperate and cold-­temperate climates, depending on alti-

& Tackenberg, 2007), and have been reported to increase plant

tude and geographic location, from mediterranean to continental

diversity at landscape scales in the central Pyrenees (Aldezabal,

and boreo-­alpine (Lamo & Sebastià, 2006). Sampling was planned

2001).

according to a stratified random sampling design, and samples were

|

      3

RODRÍGUEZ et al.

Journal of Vegetation Science

TA B L E   1   Location and climatic ranges of the sampled sites in the eastern Pyrenees

LatR (°)

LongR (°)

42–43 N

0–3 E

EW (km)

AltR (m)

TempR (°C)

PrecR (mm)

120

965

2.2–8.1

790–1325

Notes. LatR: Latitudinal range; LongR: Longitudinal range; EW: East–west extension; AtltR: Altitudinal range; TempR: Range of annual temperatures; PrecR: Range of annual precipitation. Climatic source: Ninyerola, Pons, and Roure (2000a).

selected at random within strata. This process was done in the soft-

index (Gustafson & Parker, 1994) calculated with the corresponding

ware ArcMap 10 (ESRI, Redlands, CA, USA). The basis for random-

tool in V-­L ATE.

ization was the map of habitats of Catalonia 1:50,000 (Carreras &

Regional variables are climate and bedrock variables. Bedrock

Diego, 2006). Four variables were initially considered for sampling

type was determined in the field and confirmed from the Geological

stratification: altitude (2,300 m), slope

Map (ICGC, 2007). Bedrock was categorized into three categories:

(0–20°, 20–30°; >30°), macrotopography (mountain top/south-­

basic (marls and calcareous rocks), acidic (mostly sandstones and

facing slope, valley bottom/north-­facing slope) and management

slates) and mixed. Climatic variables were determined using the

(sheep grazing, cattle grazing, mixed grazing). In this way, we deter-

digital database atlas, Climatic Digital Atlas of Catalonia (Ninyerola,

mined a set of homogeneous grassland patches by crossing the strat-

Pons, Roure 2000b). We determined mean annual precipitation

ification variable layers. Grassland patches were then listed by type

(MAP), mean summer precipitation (MSP), mean maximum tem-

and ordered within each list randomly to determine sampling prior-

perature (MTmax) and mean minimum temperature (MTmin) for each

ity. At least one to two replicates of each patch type were sampled.

grassland plot in the study. In addition, we included in the model

In each sampled grassland patch, a 10 m × 10 m plot was sys-

the intra-­annual difference of temperature, calculated as the dif-

tematically placed in the middle of each homogeneous patch, includ-

ference between mean summer temperature and mean annual

ing a specific plant community. All the plant species found in each

temperature.

plot after detailed inspection were listed and counted. The variable

Topographic variables included Slope, Aspect, Macrotopography

SR100 was thus defined as the number of plant species in 100 m2,

and Microtopography. Slope and aspect were determined in the field

which is an indicator of plant species richness in the grassland com-

by clinometer and compass, respectively. Macrotopography and mi-

munity at the 100 m2 scale. This plot size (100 m2) is bigger than the

crotopography were determined visually. Four macrotopographic

minimum sampling area for these grassland communities (de Bello

positions were initially considered: valley bottom, north-­facing

et al., 2007). Almost all species in the studied grasslands are peren-

slopes, south-­facing slopes and mountain tops. Three microtopo-

nial (Sebastià, 2004). Plots were surveyed at peak biomass, when ba-

graphic positions were initially considered: flat areas, convexities or

sically all species can be observed, if present, including leaves of the

mounds, and concavities, hollows or depressions.

few early growth geophytes and the few annuals (Sebastià, 2007).

Regarding management variables, detailed surveys were carried out among farmers, shepherds and land managers. Two management variables were considered: Grazing Intensity and Grazer Type.

2.2 | Environmental variables

Grazing intensity was determined by estimating livestock stocking

In order to assess the relationship between SR100 and possible po-

rates measured as livestock units/ha (LU/ha), and treated as a semi-­

tential environmental drivers, 20 independent variables were ini-

quantitative variable with three categories (Sebastià et al., 2008):

tially considered (Supporting information Table S1). These variables

low (1: