MODELING THE EFFECT OF TEMPERATURE CHANGES ON PLANT LIFE-FORM
In this study, we used an adaptation of Herdberg´s (1964 Acta Phytogeogr Suec) life-form classification system, to develop multiple-regression models for the current distribution of the dominant PLF across a tropical treeline and to evaluate the possible effect of temperature changes on their distribution under climate change sce- narios. The adaptations introduced to the classification system, allowed us to explore the importance of adaptive traits with a direct influence on the leaf thermal/radiation balance (i.e. leaf pubescence), in modulating the spatial distribution of emblematic PLF in the northern páramos. We hypothesized that (i) dominant PLF respond differ- ently to altitude and slope orientation, influencing treeline vegetation structure and the spatial distribution of veg- etation belts; and (ii) PLF respond differently under temperature increase scenarios, modifying vegetation physi- ognomy across the treeline ecotone.
Study area and sampling design.
The study area was located along an altitudinal gradient from 3300 to 3550 m of altitude near La Aguada station of the Mérida cable car system, in the forest-páramo ecotone at the Sierra Nevada mountain range (Mérida, Venezuela), within the limits of the Sierra Nevada National Park (Fig. 1A). Mean temperature is 7.1 °C and annual precipitation is 1811 mm (La Aguada weather station, 3452 m of altitude).
For the PLF selection, we considered pubescence in caulescent rosettes as a morphological characteristic with a clear adaptive value related with contrasting adaptive strategies. Five plant life-forms were chosen for fur- ther analysis based on their relative abundance and structural importance in defining the major vegetation physi- ognomic types across the treeline ecotone: Caulescent pubescent rosettes (CPR), caulescent non-pubescent rosettes (CNR), sclerophyllous shrubs (SS), tussock grasses (TG) and trees (T).
To quantify the cover of each PLF, six samplings transects were selected along the altitudinal gradient, just above the upper limit of continuous forest (approximately 3300 m of altitude). The sampling was carried out during July and August 2007 and transects were placed in the dominant slope orientations in the area -North, Norwest and West- with two transects per orientation. Over each transect, 11 sampling units separated by 25 m in altitude were placed regularly along the gradient. Each sampling unit consisted of two 10 m long lines separated by 2 m.
PLF cover was measured using the point quadrat method (Greig-Smith, 1983 Book), sampling 50 points per line separated by 20 cm each.
Distribution models and climate change scenarios.
Two spatial variables (altitude and slope orientation) were considered for the models according to previous studies (Bader & Ruijten, 2008 J Biogoegr; Arzac et al., 2011). Altitude values were replaced with temperature values using the linear regression model: temperature = 27.98 – 0.006 * altitude (Suárez del Moral & Chacón- Moreno, 2011 Ecotropicos); this allowed us to create distribution models under temperature change scenarios. All models were produced using ITC-ILWIS 3.3 a remote sensing and GIS software. Nonlinear multiple regression models were performed to analyze the response of cover for each plant life-form (dependent variable) as a function of altitude and slope orientation (independent variables)
To determine the possible response of the PLF in new temperature scenarios, projections based on climat e change models used in the Fifth Assessment Report of the Intergovernmental Panel of Climate Change (AR5) (IPCC, 2013 Rprt) were run. We employed the projections for temperature expressed as anomalies with respect to the reference period 1986-2005, for RCP4.5, RCP6.0 and RCP8.5 analyzed in two periods of 30 years (2025 to 2055 and 2045 to 2075), the models were focused on 2040 and 2060 respectively (IPCC, 2013).
Results and discussion.
Our results indicated that within an altitudinal gradient of 250 m, PLF showed distribution patterns highly influenced by altitude and slope orientation, with a pronounced decrease of trees abundance with altitude. Multiple regression models suggested a strong link between both variables and PLF cover values.
TG were the dominant PLF in the study area with cover values of 84.02 % followed by CNR (9.79 %), SS (6.85 %), CPR (5.23 %), and T (4.64 %). Additionally, PLF distributional patterns showed differences in their optimal distribution limits along the gradient, albeit with some range overlap. CPR and SS showed the highest optimum altitudinal distribution (3489 and 3476 m of elevation), while TG and CNR showed lower values for their optimum altitudinal distribution (3455 and 3425 m of elevation respectively). Finally at the bottom of the gradient, tress showed the lowest optimal altitudinal distribution (3390 m of elevation).
Vegetation distribution models and climate change scenarios.
PLF spatial distribution patterns showed clear differences in the surface and altitudinal range they covered within the study site (Fig. 1B-F). CPR and SS were constrained to the upper limit of the gradient and the models predicted predominant cover values between 15-20% for CPR and between 8-12 % in the case of SS. With a wider distributional range (centered on intermediate elevations along the gradient), CNR rosettes showed cover values between 4-16 %. The models predicted higher cover for CPR in West facing slopes opposed to CNR, which showed higher covers in North facing slopes. Trees were restricted to low elevations and their cover was also influenced by slope orientation with higher cover values (4-6 %) in Norwest facing slopes, showing the treeline
limit to lie between 3280-3340 m of elevation. On the other hand, TG showed the highest relative cover values across the gradient (60-90 %, except at very low elevations).
Figure 1. Digital elevation model (DEM) of the study site and spatial distribution models showing the percentage of surface cover for the selected PLF in the sampling site, under current climate conditions. (A) DEM of study site, (B) Caulescent pubescent rosettes, (C) sclerophyllous shrubs, (D) caulescent non-pubescent rosettes, (E) tussock grasses and (F) trees. Colors represent the percent cover.
The distribution of trees is especially relevant in the context of this study, since sampling was carried out immediately above the continuous forest-line, just in the contact zone with the grassland páramos. The highest tree cover value was found at the lower limit of the gradient in Northwest facing slopes, associated with lower incidence of solar radiation in the study area. Bader & Ruijten (2008) found a fluctuation of Andean treeline position as a function of slope orientation in the Andes (Central Ecuador). They reported lower position of the treeline in slopes with higher solar radiation incidence, and proposed this could be linked with low temperature photoinhibition of tree seedlings. Seedling establishment beyond the tropical treeline is mainly limited by temperature and solar ra- diation (Bader et al., 2007b Plant Ecol) but colonization of trees species that seem to be well adapted to the con- ditions outside closed forests (e.g. Diplostephium) may occur above the treeline (Llambí et al., 2013).
Adaptations such as leaf trichromes may be responsible for the contrasting distribution pattern observed for both caulescent rosettes types. Although CPR and CNR belong to the same plant life-form in Hedberg´s (1964) classification scheme, CPR showed higher cover values in the upper limit of the gradient and were mainly associ- ated to West facing slopes, characterized by lower mean average temperatures in our study area; in contrast, CNR showed higher cover at lower elevations in North facing slope. In fact, leaf pubescence has been shown to influence the leaf energy balance, reducing the absorption of solar radiation (Rada et al., 1985 Plant Cell Environ; Vareschi, 1992; Meinzer et al., 1994 Book Chap; Azócar & Rada, 2006 Book). Additionally, pubescent rosettes show a freezing avoidance strategy, through mechanisms such as the protection of organs through insulating structures (e.g. marcescent leaves) and supercooling (Squeo et al., 1996 Oecologia; García-Vare la & Rada, 2003 Acta Oecol).
Small leaf size in sclerophyllous shrubs could also be interpreted as an adaptive strategy favored at higher eleva- tions in the páramo, decreasing the leaf area exposed and increasing drought resistance (Vareschi, 1992; Ely &
Torres, 2003 Plantula).
The TG (dominant PLF in the area) reached maximu m cover values close to 90 %, showing a fairly uniform spatial pattern across different elevations and slope orientation. This distributional pattern suggests that the studied altitudinal range is within the optimal environmental envelope for this life-form, at least in these relatively humid páramos.
Despite several factors (e.g. precipitation, dispersion and migration rates) could be responsible of plant distribution under climate change scenarios, in this work simple spatial distribution models were developed and only changes in temperature were analyzed. Overall, models predicted an upslope migration of PLF as a response of temperature increments; these results are supported by vegetation shift described on high Andean areas (Feeley et al., 2011 J Biogeogr; Feeley, 2012 Glob Chang Biol; Morueta-Holme et al., 2015 Proc Natl Acad Sci USA).
Models suggested a displacement of woody species across the grassland, increasing its density and altitudinal range (Fig. 2A-F). This would be associated with a change in grassland páramo structure where woody species begin to be dominant (Fig. 2G-L).
(B) (C) (D) (E) (F)
3600 3520 3400
3200 3360 3280 Altitude (m) (A)
Figure 2. Cover distribu- tion models for tress (A-F) and tussock grasses (G-L) under climate change sce- narios for the study site.
The Representative Con- centration Pathway (RCP) and the future time periods are indicated. Colors repre- sent the percent cover.
Temperature increments:
0.7 ºC (A & G), 0.9 ºC (B
& H), 1.3 ºC (C & I), 1.9 ºC (I & J), 1.8 (E & K) and 2.6 (F & L).
Several authors have predicted upslope shifts (Suárez del Moral & Chacón-Moreno, 2011; Tovar et al., 2013 PLoS ONE) and population size expansion (Feeley & Silman, 2010) of high mountain forest species as a consequence of climate change on Andean ecosystems. The expansion of forest usually is linked with the reduction of population size of other páramo species (Feeley & Silman, 2010; Tovar et al., 2013). However, the expansion of trees above the treeline will depend on the ability of individuals to settle outside forest boundaries, where climate conditions are different (Llambí et al. 2013).