Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Phenology and Productivity of C3 and C4 Grasslands in Hawaii

Phenology and Productivity of C3 and C4 Grasslands in Hawaii

  • Stephanie Pau, 
  • Christopher J. Still
PLOS
x

Abstract

Grasslands account for a large proportion of global terrestrial productivity and play a critical role in carbon and water cycling. Within grasslands, photosynthetic pathway is an important functional trait yielding different rates of productivity along environmental gradients. Recently, C3-C4 sorting along spatial environmental gradients has been reassessed by controlling for confounding traits in phylogenetically structured comparisons. C3 and C4 grasses should sort along temporal environmental gradients as well, resulting in differing phenologies and growing season lengths. Here we use 10 years of satellite data (NDVI) to examine the phenology and greenness (as a proxy for productivity) of C3 and C4 grass habitats, which reflect differences in both environment and plant physiology. We perform phylogenetically structured comparisons based on 3,595 digitized herbarium collections of 152 grass species across the Hawaiian Islands. Our results show that the clade identity of grasses captures differences in their habitats better than photosynthetic pathway. Growing season length (GSL) and associated productivity (GSP) were not significantly different when considering photosynthetic type alone, but were indeed different when considering photosynthetic type nested within clade. The relationship between GSL and GSP differed most strongly between C3 clade habitats, and not between C3-C4 habitats. Our results suggest that accounting for the interaction between phylogeny and photosynthetic pathway can help improve predictions of productivity, as commonly used C3-C4 classifications are very broad and appear to mask important diversity in grassland ecosystem functions.

Introduction

A critically important problem in understanding ecosystem responses to global change is the relationship between growing season length and productivity. This is because growing season length is commonly used to predict net ecosystem exchange and to inform terrestrial biosphere models of vegetation dynamics and carbon exchange [1], [2]. The length of the growing season should be a primary factor controlling the terrestrial carbon cycle by setting the length of time available for photosynthesis and plant growth [3][5]. For example, an extension of the growing season either through earlier spring onset or later autumn senescence has been shown to be positively related to primary productivity and increased carbon assimilation across several different vegetation types [6] and refs. above but also see [7][8] for examples of no or negative relationships. However most studies examine temperate deciduous or evergreen forests, while examples from grasslands and tropical sites are limited.

Grasslands and tropical savannas comprise approximately 29% of the global extent of terrestrial biomes and this corresponds to some 33% of global terrestrial NPP [9]. Photosynthetic pathway is a dominant plant functional trait that has consequences for the global carbon cycle because of the distribution of C3 and C4 plants and their different rates of productivity [10]. In general, the efficiency of C4 photosynthesis should be greater than C3 photosynthesis under conditions of high temperatures, high light, and greater aridity [11], [12]. The classic turnover of C3-C4 grasses along altitudinal gradients has been well documented [13][19]. This turnover illustrates the physiological advantage of C4 grasses in warmer and drier environments, which confers dominance in species composition and relative cover. The occurrence of C4 grasses in more marginal environments, however, should have consequences for their productivity; thus they may not be more productive than C3 grasses even with an advantage in resource-use efficiencies under resource-poor conditions [15].

C3 and C4 grasses should also partition their activity along temporal environmental gradients resulting in different phenologies and growing season lengths. Differences in phenology have been demonstrated in both mixed C3-C4 grasslands [20], [21] as well as across large environmental gradients [15]. Studies of C3 and C4 grasslands in the Great Plains of North America have demonstrated that, compared to C3 grasses, C4 grasses are active later in the growing season when temperatures warm and water becomes limiting in their respective environment [21][25]. The Great Plains region has been the focus of many subsequent studies using remote sensing to discriminate C3-C4 regions based on their distinct seasonality and responses to climate variability [26][29].

Yet there is a growing body of work that has re-examined C3-C4 comparisons, benefiting from the recent development of well-resolved grass phylogenies [30]. The vast majority of grasses (Poaceae) belong to either the ‘BEP’ or ‘PACMAD’ clade, which last shared a common ancestor an estimated 50–80 Mya [31]. ‘BEP’ is an acronym for the Bambusoideae, Ehrhartoideae, and Pooideae lineages, which contain only C3 species. ‘PACMAD’ refers to Panicoideae, Aristidoideae, Chloridoideae, Micrairioideae, Arundinoidae, and Danthonioideae lineages, which contain all of the C4 species and some C3 species [30]. Thus, there is considerable diversity within C3 and C4 groups associated with different lineages. Comparing C3 and C4 grasses restricted to the PACMAD clade enables our understanding of ecological differences previously confounded by different evolutionary histories [32][37]. Furthermore C3 BEP and PACMAD taxa may exhibit ecological differences relevant to community structure and ecosystem functions.

Previous phylogenetic modeling of Hawaiian grasses has demonstrated that BEP and PACMAD clades capture habitat distinctions not strictly associated with photosynthetic pathway [33]. This work suggested that there are larger differences in aridity between the habitats of C3 and C4 grasses compared to temperature differences [33], [35], [37], a finding largely driven by the much higher precipitation and woody cover of C3 PACMAD grass environments compared to other grass groupings [38]. Subsequent ecological niche modeling described the changes from C4 grass-dominated habitats in lower elevation, warm and dry regions, to C3 PACMADs in mid-elevation, cooler and wetter regions, and finally to C3 BEPs at the highest elevations in cooler and drier conditions [37]. It was furthermore shown that C3 PACMADs tend to occur in habitats that receive early season (winter) precipitation, whereas the habitat sorting of C3 BEPs and C4 grasses was associated with temperature differences. This work and others add to the growing evidence that there are important ecological differences among lineages that were not captured by previous C3-C4 comparisons. Likewise, phylogenetically structured comparisons should highlight when and where photosynthetic pathway may in fact be responsible for habitat sorting and differences in resource strategies.

In this study we examine the phenology and inferred productivity of C3-C4 grass habitats using herbarium collection localities across the Hawaiian Islands and a timeseries of satellite images. C3 and C4 grasses dominate distinct regions along spatial gradients, thus our examination of the phenology of these habitats reflects both the dominant plant functional type and the environment in which they occur. We compare the habitats of C3 and C4 grass species within the PACMAD clade to try to isolate the effects of photosynthetic pathway on the timing, magnitude, and estimated productivity (integrated NDVI) of these grasslands. We also examine habitats of C3 grass species from both PACMAD and BEP clades to highlight the functional diversity that is commonly grouped. Specifically, we ask the following questions: Do the habitats of C3 and C4 grass species within and between clades differ in their phenology (start-of-season, end-of-season, and growing season length) and productivity? Is the relationship between growing season length and productivity of grass species' habitats within and between clades different? The Hawaiian Islands provide an ideal setting for this work because of the considerable variation in both species diversity and climatic gradients within a small geographic region. The majority of grasses in Hawaii are introduced, and thus their distribution should reflect recent ecological sorting and not insular evolutionary history or patterns of extinction.

Methods

Species and environmental data

We used a dataset of 3,595 digitized geo-referenced herbarium records for Poaceae created by [33]. Records from across the main seven Hawaiian Islands (Kauai, Oahu, Molokai, Lanai, Maui, Kahoolawe, and Hawaii) were obtained from the Smithsonian Flora of the Hawaiian Islands Website and The Bishop Museum Herbarium. The dataset represents 152 grass species (there are multiple collections for most species), each assigned to either the ‘BEP’ or ‘PACMAD’ clade that together represent most species of Poaceae. Each species' life history was further categorized as annual or perennial based on [39]. Because individual occurrence records may not indicate a large enough expanse of grasses for remote sensing analysis, herbarium point localities were excluded if they did not fall within a grassland landcover class according to the Hawaii GAP Analysis. The Hawaii GAP Analysis is a land cover dataset derived from 30-m resolution Landsat imagery from 1999–2003 and edited using ancillary data and expert knowledge (US Geological Survey, Gap Analysis Program (GAP) August 2011. National Land Cover, Version 2). As a consequence, our C3 PACMAD samples represent species that occur open grassland environments, yet C3 PACMADs tend to occur in regions with the greater tree cover [37], [38].

Additionally, C3 PACMADs present a challenge in using remotely sensed data to capture grass phenology because they are globally uncommon. Because of this, we verified that at least ten C3 PACMAD species in our herbarium records were collected recently based on collection dates. Five species have been collected at recently as 2000 (Cortaderia jubata, Dichanthelium hillebrandianum, Oplimensus compositus, Rytidosperma pilosum, and Schizostachyum glaucifolium); in addition, another five have been collected since 1990 (Cortaderia selloana, Dichanthelium cynodon, Isachne distichophylla, Oplimenus hirtella, Sacciolepsis indica). These species are furthermore all considered common and/or occurring densely in open areas in Hawaii, with the exception of Schizostachum [39].

Remote sensing data

We used the NASA Terra Moderate Resolution Imaging Spectroradiometer (MODIS) NDVI (MOD13Q1) Collection 5, which provides 16-day composite data at 250-m spatial resolution. Data were screened using the ‘QC_Day’ scientific data set for only ‘good quality’ pixels (i.e., not contaminated by clouds or aerosols) or ‘check other QA’ in the VI Quality dataset. Pixels falling into the latter category were discarded if flagged for clouds (adjacent cloud, mixed clouds and possible shadow), aerosols (high and climatology aerosols), or possible shadow. NDVI has been shown to be less sensitive to view-angle differences compared to the other commonly used index of plant greenness, the Enhanced Vegetation Index (EVI) [40].

An NDVI time series was created from one pixel centered over each collection locality for a 10-year period using 16-day MODIS data (February 2000–February 2010). Each time series was first smoothed using a centered moving-window average of five data points. For each smoothed time series (for each collection locality), piece-wise logistic functions were fit to each ascending and descending phase of NDVI, following [41]. The points of local maxima or minima in the rate of change in the curvature of the fitted logistic models were used to identify the ‘Start-of-Season’ (SOS) and ‘End-of-Season’ (EOS) for each year [41]. ‘Growing Season Length’ (GSL) is the length of time between SOS and EOS, and ‘Growing Season Productivity’ (GSP) is the time-integrated NDVI between the SOS and EOS (integrated across days - NDVI is a ratio and is unitless). An attempt was made to use each of the 3,595 herbaria point localities for logistic fits. However, in a given year, collection localities were excluded if the fitting procedure failed to converge, for example, because too few data points remained after the QC_Day screening.

On average 107 logistic fits were made each year for C3 BEPs, 19 for C3 PACMADs, and 284 for C4 PACMADs, which in part reflects the number of collections for each group (there were fewer C3 PACMADs in the herbarium records) as well as potential biases in tree-covered and cloudy regions that were screened more heavily. Of these successful logistic fits, SOS and EOS dates were then averaged across all species falling into each photosynthetic type-clade combination (3 groups: C3 PACMAD, C3 BEP, and C4 PACMAD), for each year (10 years). Annuals and perennials within each photosynthetic type-clade combination were also determined (except there were no annual C3 PACMADS), which together with the 3 photosynthetic type-clade combinations produced a total of 50 samples for statistical analyses.

Statistical analyses

First, linear models were used to examine the effect of photosynthetic pathway on SOS, EOS, and GSL irrespective of clade (all C3 vs. C4 grasses). Then, photosynthetic pathway nested within clades (PACMAD:C3, PACMAD:C4, BEP:C3) were used to assess differences in SOS, EOS, and GSL. Using photosynthetic type nested within clade tests for differences in means among photosynthetic type and differences in means among photosynthetic type-clade combinations. Tukey post-hoc tests were used to examine independent contrasts between each group and to test for differences more likely due to photosynthetic type by controlling for clade identity (C3 PACMAD vs. C4 PACMAD). Linear models were also used to examine the relationship between GSL and GSP. Four models were compared using a corrected Akaike's Information Criterion (AICc) for small sample sizes and with greater penalties for extra model parameters [42]. The simplest model considered only GSL as a predictor of GSP (Model A; see Table 1). Then the interactions between GSP and clade (Model B), between GSP and photosynthetic type (Model C), and between GSP and photosynthetic pathway nested within clade (Model D) were included. These models were further compared using Akaike weights (wi), which indicate a measure of model selection uncertainty, i.e., the probability that a model is ‘best’ among a set of candidate models.

thumbnail
Table 1. Comparison of models predicting growing season productivity (GSP; estimated with integrated NDVI) using growing season length (GSL) and its interaction with clade and photosynthetic type.

https://doi.org/10.1371/journal.pone.0107396.t001

Results and Discussion

Ignoring phylogeny (i.e., comparing only C3 and C4 grass groupings), SOS was later in C4 habitats compared to C3 habitats (F = 9.42, p<0.01, df = 48; Figure 1a). The SOS for C4 habitats occurred during the end of September (mean day of year or DOY = 269 based on logistic models) compared to the beginning of September for C3 habitats (mean DOY = 245). In Hawaii, climate seasonality is governed primarily by precipitation, with November-April generally considered the wet season and May-October the dry season [39], [43]. Therefore, the SOS for both C3 and C4 grasslands appear to fall within the dry season, but C4 habitats green-up later in the dry season when plants should be more drought-stressed. C4 grasses in temperate climates have been shown to be active and more productive later in the growing season as temperatures increase and water becomes limiting in their respective environments [20], [25], [44]. Indeed, this temporal offset forms the basis for predicting C3-C4 mixtures in temperate grasslands [26][29], [45], [46]. The physiological advantage of greater water-use efficiency [47] and dominance of arid, warm environments (reviewed in [12]) by C4 grasses is well recognized and furthermore shown to be robust in phylogenetically structured screening experiments [36].

thumbnail
Figure 1. Differences in C3 and C4 start-of-season (SOS) (a), end-of-season (EOS) (b), growing season length (c), and growing season productivity (d).

Y-axis for panels (a)–(c) is the Julian day-of-year or DOY; y-axis for panel (d) is integrated NDVI, which is unitless, based on logistic models using 10-year timeseries of MODIS NDVI (see Methods). The growing season in Hawaii crosses the calendar year so that SOS begins at a later DOY than EOS. Most Poaceae species fall within the ‘BEP’ or ‘PACMAD’ clade. All C4 grasses are in the ‘PACMAD’ clade.

https://doi.org/10.1371/journal.pone.0107396.g001

When we considered phylogenetic structure by accounting for photosynthetic pathway nested within clade, results revealed that the temporal offset in SOS between photosynthetic types reflected differences between C4 PACMAD habitats and C3 BEP habitats (F = 4.66, p<0.02, df = 47; posthoc Tukey's p<0.05; Figure 1a). C4 habitats had a significantly later SOS compared to C3 BEP habitats by about 23 days on average (p = 0.03). There was no statistically significant SOS difference between C3 and C4 PACMADs (posthoc Tukey's p = 0.17). Thus SOS differences were due to BEP-PACMAD identity rather than strictly photosynthetic pathway, otherwise this difference should be apparent in the C3 and C4 PACMAD comparison. Our data show large variability in SOS for C3 PACMAD habitats indicating a diversity of strategies and growing environments for this group. Statistical significance notwithstanding, C4 habitats started their growing season an estimated 20 days later into the dry season on average compared to C3 PACMAD habitats. C3 PACMAD SOS was only about 3 days later than C3 BEPs on average, and this difference was not statistically significant (p = 0.99).

EOS showed the greatest variability among phenological metrics. Although there was a significant difference when considering photosynthetic type alone (F = 4.85, p<0.05, df = 48), there was no significant difference when photosynthetic type was nested within clade (F = 2.81, p = 0.07, df = 47; post-hoc Tukey's p>0.05 for all comparisons; Figure 1b). C3 and C4 habitats brown-down during the transition into the dry season in mid-April/beginning of May (mean EOS DOY for C3 = 124; mean EOS DOY for C4 = 107). C4 habitats had an earlier EOS compared to C3 habitats when clade identity was not considered, and this difference appears to be due to the later EOS for BEP taxa (Figure 1b). Environmental controls on EOS or leaf senescence are not well understood (Menzel et al. 2006; Taylor et al. 2007). In other analyses of phenological shifts associated with climate change, EOS dates have exhibited more variability in comparison to SOS dates, [48][52].

GSL was not significantly different when considering photosynthetic type alone (F = 3.18, p = 0.08, df = 48), but was indeed different when considering photosynthetic type nested within clade (F = 4.44, p<0.02, df = 47; Figure 1c). C3 PACMAD habitats had a longer GSL by about 30 days on average compared to C4 habitats in the same clade (posthoc Tukey's p<0.05); no other comparisons were significantly different. Thus the data show that physiological and/or habitat differences governing GSL are likely masked when C3 taxa are grouped across clades. But when clades are accounted for, the difference in GSL due to photosynthetic pathway becomes more apparent (i.e., when comparing C3 and C4 PACMAD GSL – Figure 1c). Similarly, GSP was not significantly different with respect to just photosynthetic type (F = 0.10, p = 0.32, df = 48), but was significantly different when considering photosynthetic type nested within clade (F = 9.59, P<0.001, df = 47). C3 PACMAD habitats were significantly more productive than C3 BEP and C4 habitats (post-hoc Tukey's p<0.001 for both comparisons; Figure 1d). The longer GSL and greater GSP of C3 PACMAD habitats reflect the difference in their growing environment—they prefer or are possibly restricted to more favorable habitats for plant growth—wetter than C4 habitats and warmer than C3 BEP habitats [37].

GSL explained a large proportion of the variability in GSP—as GSL increased, there was a corresponding increase in GSP (Figure 1c,d). There were two equivalent best-fit models predicting yearly differences in GSP and these were the two models that included clade as a model parameter (Models B and D). Akaike weights showed more support for the model with photosynthetic type nested within clade (Model D) over the model with clade alone (Model B), and there was strong support for both models over other candidate models that excluded clade. Using photosynthetic type alone to predict GSP resulted in the worst-fit model.

The slope of the relationship between GSL and GSP differed for each photosynthetic type-clade combination (Model D; F = 108.4, p<0.01, df = 45; Figure 2 and Table 2). Productivity increased more per unit of increase in GSL in C3 PACMAD habitats than other grass habitats. Results show that controlling for GSL, C3 PACMAD habitats were still more productive than C4 or C3 BEP habitats. Interestingly, there was a larger difference within the C3 functional group than between photosynthetic pathways, i.e., there was a larger clade effect (BEP vs. PACMAD) than photosynthetic pathway effect (C3 vs. C4). There was not a significant difference in the GSL of habitats where C3 BEPs and C3 PACMADs occur (Figure 1c), but these habitats had the largest difference in GSP. In comparison to C4 regions, BEP environments had a longer GSL but were less productive (Figure 2), although these differences were not statistically significant. This pattern is consistent with a resource strategy of longer growing seasons at a cost of lower productivity [53], [54]. Particularly in resource-poor environments, such as the cold and dry habitats of BEP taxa, species with longer growing seasons are often associated with lower rates of productivity as a strategy to balance energy requirements for tissue growth. BEPs may have traits that have adapted to these marginal environments by adopting a slow and conservative approach to energy-use. Conversely, plants in high-resource environments tend to produce short-lived leaves with high photosynthetic rates at a cost of being less resistant to environmental stresses and having to produce new leaf tissue [54]. This tradeoff between leaf lifespan and rates of photosynthesis has also been demonstrated in tropical species [55], where the growing season is potentially year-round. Although these strategies may explain C3 BEP and C4 differences, C3 PACMADs did not appear to fit this tradeoff, yet they exhibited greater rates of productivity even when controlling for growing season length.

thumbnail
Figure 2. The slope of the relationship between GSL and GSP differed for each photosynthetic type-clade combination (Model D, Table 2).

C3 PACMAD habitats exhibited higher rates of greenness than C4 PACMAD or C3 BEP habitats for a given growing season length (slope coefficients = 0.87, 0.83, and 0.59 respectively). Differences within the C3 functional group were larger than between photosynthetic pathways, i.e., there was a larger clade effect than photosynthetic pathway.

https://doi.org/10.1371/journal.pone.0107396.g002

thumbnail
Table 2. The relationship between growing season length (GSL) and productivity (GSP) (Model D in Table 1) was marginally different between C3 PACMAD and C4 grasses (GSL*clade:photosynthetic type), and significantly different between clades irrespective of photosynthetic type (GSL*clade). See Figure 2.

https://doi.org/10.1371/journal.pone.0107396.t002

Notably, life history differences (annual vs. perennial) were not significant in explaining any phenological metric except for GSP (F = 4.63, p<0.05, df = 48), with perennial species being more productive based on integrated NDVI compared to annual species (posthoc Tukey's p<0.05). However the slope of the relationship between GSL and GSP was not significantly different when considering life history (p = 0.07).

Our proxy for productivity (integrated NDVI) only considers aboveground greenness. It is possible that C4 and C3 BEP grasses may be less productive aboveground because they are allocating resources to roots, a strategy that would make sense in water- or nutrient-limited environments [56]. A further limitation of the study is the influence of tree cover on the phenology of understory grasses. In our case, although we only used pixels occurring in grassland vegetation classes, C3 PACMADs preferentially occur in regions with the greater tree cover, which represents a microclimate that is cooler and shadier [37], [38]—consequently these samples were excluded and our C3 PACMADs are represented by species that occur only in open environments.

Conclusions

Our results provide a working hypothesis for understanding C3 and C4 grass habitats and growing environments. We show that the clade identity of grasses captures differences in their habitats better than photosynthetic pathway. SOS differences that appear to be associated with the habitats of C3 and C4 grasses are in fact associated with habitats of different clades. EOS was not significantly different among any comparisons. GSL and GSP are indeed constrained by photosynthetic pathway; however, the relationships between GSL and GSP were not as expected given the well-recognized habitat sorting of C3-C4 photosynthetic pathways. Importantly, the relationship between growing season length and associated productivity differed most strongly between C3 clade habitats and not between C3-C4 habitats. C3 PACMAD habitats, which in Hawaii represent highly favorable growing environments, had the longest growing season and exhibited the largest variability in all phenological metrics. Taken together, the characteristics of C3 PACMAD habitats shown here may possibly indicate greater trait diversity within this lineage of globally rare grasses and an advantage in adapting to novel climates [57]. Our results highlight the large functional diversity within Poaceae, one of the largest flowering plant families, and how grass species may respond to future global change. We furthermore identify a need for ecosystem and vegetation models of plant productivity to refine relationships between GSL and GSP. Although our work is limited to grasses and their habitats, in theory, different species or plant functional groups should possess distinct growth strategies resulting in complex relationships between phenology and productivity [58].

Acknowledgments

We thank E.J. Edwards for developing the herbarium dataset and J. Regetz for assistance with the timeseries analysis. We additionally thank the National Center for Ecological Analysis & Synthesis (NCEAS) and Florida State University.

Author Contributions

Conceived and designed the experiments: SP CJS. Analyzed the data: SP. Wrote the paper: SP CJS.

References

  1. 1. White M, Running S, Thornton P (1999) The impact of growing-season length variability on carbon assimilation and evapotranspiration over 88 years in the eastern US deciduous forest. Int J Biometeorol 42: 139–145.
  2. 2. Richardson AD, Anderson RS, Arain MA, Barr AG, Bohrer G, et al. (2012) Terrestrial biosphere models need better representation of vegetation phenology: results from the North American Carbon Program Site Synthesis. Glob Chang Biol 18: 566–584.
  3. 3. Myneni RB, Keeling CD, Tucker CJ, Asrar G, Nemani RR (1997) Increased plant growth in the northern latitudes from 1981 to 1991. Nature 386: 698–701.
  4. 4. Baldocchi D, Falge E, Gu L, Olson R, Hollinger D, et al. (2001) FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull Amer Meteor Soc 82: 2415–2434.
  5. 5. Churkina G, Schimel D, Braswell BH, Xiao A (2005) Spatial analysis of growing season length control over net ecosystem exchange. Glob Chang Biol 1777–1787.
  6. 6. Piao S, Friedlingstein P, Ciais P, Viovy N, Demarty J (2007) Growing season extension and its impact on terrestrial carbon cycle in the Northern Hemisphere over the past 2 decades. Global Biogeochem Cycles 21: GB3018.
  7. 7. Dunn AL, Barford CC, Wofsy SC, Goulden ML, Daube B (2007) A long-term record of carbon exchange in aboreal black spruce forest: means, responses to interannual variability, and decadal trends. Glob Chang Biol 13: 577–590.
  8. 8. Hu J, Moore DJP, Monson RK (2010) Longer growing seasons lead to less carbon sequestration by a subalpine forest. Glob Chang Biol 16: 771–783.
  9. 9. Saugier B, Roy J, Mooney HA (2001) Estimations of global terrestrial productivity: Converging on a single number? In: Roy J, Saugier B, Mooney HA, editors. Terrestrial Global Productivity. San Diego: Academic Press. pp. 543–557.
  10. 10. Still CJ, Berry JA, Collatz GJ, DeFries RS (2003) Global distribution of C 3 and C 4 vegetation: Carbon cycle implications. Global Biogeochem Cycles 17.
  11. 11. Ehleringer JR, Cerling TE, Helliker BR (1997) C4 photosynthesis, atmospheric CO2, and climate. Oecologia 112: 285–299.
  12. 12. Sage RF, Monson RK (1999) C4 Plant Biology. San Diego: Academic Press.
  13. 13. Newell C (1968) A phytosociological study of the major vegetation types in Hawai'i Volcanoes National Park, Hawai'i University of Hawai'i.
  14. 14. Chazdon RL (1978) Ecological Aspects of the Distribution of C4 Grasses in Selected Habitats of Costa Rica. Biotropica 10: 265–269.
  15. 15. Tieszen LL, Senyimba MM, Imbamba SK, Troughton JH (1979) The distribution of C3 and C4 grasses and carbon isotope discrimination along an altitudinal and moisture gradient in Kenya. Oecologia 37: 337–350.
  16. 16. Rundel PW (1980) The ecological distribution of C4 and C3 grasses in the Hawaiian Islands. Oecologia 45: 354–359.
  17. 17. Cabido M, Pons E, Cantero JJ, Lewis JP, Anton A (2007) Photosynthetic pathway variation among C 4 grasses along a precipitation gradient in Argentina. Journal of Biogeography 24: 197–204.
  18. 18. Cavagnaro JB (1988) Distribution of C3 and C4 grasses at different altitudes in a temperate arid region of Argentina. Oecologia 76: 273–277.
  19. 19. Angelo CL, Daehler CC (2013) Upward expansion of fire-adapted grasses along a warming tropical elevation gradient. Ecography 36: 551–559.
  20. 20. Fargione J, Tilman D (2005) Niche differences in phenology and rooting depth promote coexistence with a dominant C4 bunchgrass. Oecologia 143: 598–606.
  21. 21. Ode DJ, Tieszen LL, Lerman JC (1980) The seasonal contribution of C3 and C4 plant species to primary production in a mized prairie. Ecology 61: 1304–1311.
  22. 22. Kemp PR, Williams GJ III (1980) A physiological basis for niche separation between agropyron Smithii (C3) and Bouteloua gracilis (C4). Ecology 61: 846–858.
  23. 23. Still CJ, Berry Ja, Ribas-Carbo M, Helliker BR (2003) The contribution of C3 and C4 plants to the carbon cycle of a tallgrass prairie: an isotopic approach. Oecologia 136: 347–359.
  24. 24. Epstein HE, Lauenroth WK, Burke IC, Coffin DP (1997) Productivity patterns of C3 and C4 functional types in the U.S. Great Plains. Ecology 78: 722–731.
  25. 25. Tieszen LL, Reed B, Bliss NB, Wylie B, DeJong D (1997) NDVI, C3 and C4 production and distribution in Great Plains grassland land cover classes. Ecol Appl 7: 59–78.
  26. 26. Goodin DG, Henebry GM (1997) Monitoring ecological disturbance in tallgrass prairie using seasonal NDVI trajectories and a discriminant function mixture model. Remote Sens Environ 61: 270–278.
  27. 27. Davidson A, Csillag F (2003) A comparison of three approaches for predicting C4 species cover of northern mixed grass prairie. Remote Sens Environ 86: 70–82.
  28. 28. Foody GM, Dash J (2010) Estimating the relative abundance of C3 and C4 grasses in the Great Plains from multi-temporal MTCI data: issues of compositing period and spatial generalizability. Int J Remote Sens 31: 351–362.
  29. 29. Wang C, Jamison B, Spicci A (2010) Trajectory-based warm season grassland mapping in Missouri prairies with multi-temporal ASTER imagery. Remote Sens Environ 531–539.
  30. 30. Grass Phylogeny Working Group II (2012) New grass phylogeny resolves deep evolutionary relationships and discovers C4 origins. New Phytol 193: 304–312.
  31. 31. Vicentini A, Barber JC, Aliscioni SS, Giussani LM, Kellogg Ea (2008) The age of the grasses and clusters of origins of C 4 photosynthesis. Glob Chang Biol 14: 2963–2977.
  32. 32. Edwards EJ, Still CJ, Donoghue MJ (2007) The relevance of phylogeny to studies of global change. Trends Ecol Evol 22: 243–249.
  33. 33. Edwards EJ, Still CJ (2008) Climate, phylogeny and the ecological distribution of C4 grasses. Ecol Lett 11: 266–276.
  34. 34. Osborne CP, Freckleton RP (2009) Ecological selection pressures for C4 photosynthesis in the grasses. Proc Biol Sci 276: 1753–1760.
  35. 35. Edwards EJ, Smith Sa (2010) Phylogenetic analyses reveal the shady history of C4 grasses. Proc Natl Acad Sci U S A 107: 2532–2537.
  36. 36. Taylor SH, Hulme SP, Rees M, Ripley BS, Woodward FI, et al. (2010) Ecophysiological traits in C3 and C4 grasses: a phylogenetically controlled screening experiment. New Phytol 185: 780–791.
  37. 37. Pau S, Edwards EJ, Still CJ (2013) Improving our understanding of environmental controls on the distribution of C3 and C4 grasses. Glob Chang Biol 19: 184–196.
  38. 38. Still CJ, Pau S, Edwards EJ (2014) Land surface skin temperature captures thermal environments of C3 and C4 grasses. Glob Ecol Biogeogr 23: 286–296.
  39. 39. Wagner WL, Herbst DR, Sohmer S (1999) Manual of the Flowering Plants of Hawai'i. Revised. Honolulu: University of Hawai'i Press.
  40. 40. Sims DA, Rahman AF, Cordova VD, El-Masri BZ, Baldocchi DD, et al. (2006) On the use of MODIS EVI to assess gross primary productivity of North American ecosystems. J Geophys Res 111: 1–16.
  41. 41. Zhang X, Friedl MA, Schaaf CB, Strahler AH, Hodges JCF, et al. (2003) Monitoring vegetation phenology using MODIS. Remote Sens Environ 84: 471–475.
  42. 42. Burnham KP, Anderson DR (2010) Model selection and multimodel inference: a practical information-theoretic approach. Second Edi. New York: Springer.
  43. 43. Giambelluca TW, Chen Q, Frazier AG, Price JP, Chen Y-L, et al. (2013) Online Rainfall Atlas of Hawai'i. Bull Am Meteorol Soc 94: 313–316.
  44. 44. Reed BC, Brown JF, Vanderzee D, Loveland TR, Merchant W, et al. (2009) Measuring phenological variability from satellite imagery. Inf Syst 5: 703–714.
  45. 45. Foody G, Dash J (2007) Discriminating and mapping the C3 and C4 composition of grasslands in the northern Great Plains, USA. Ecol Inform 2: 89–93.
  46. 46. Powell RL, Yoo E-H, Still CJ (2012) Vegetation and soil carbon-13 isoscapes for South America: integrating remote sensing and ecosystem isotope measurements. Ecosphere 3: art109.
  47. 47. Pearcy RW, Ehleringer JR (1984) Comparative ecophysiology of C3 and C4. plants. Plant Cell Environ 7: 1–13.
  48. 48. Menzel A, Sparks TH, Estrella N, Koch E, Aasa A, et al. (2006) European phenological response to climate change matches the warming pattern. Glob Chang Biol 12: 1969–1976.
  49. 49. Menzel A, Fabian P (1999) Growing season extended in Europe. Nature 397: 659.
  50. 50. Zhou L, Tucker CJ, Kaufmann RK, Slayback D, Shabanov NV, et al. (2001) Variations in northern vegetation activity inferred from satellite data of vegetation index during 1981 to 1999. J Geophys Res 106: 20069–290083.
  51. 51. Taylor G, Tallis MJ, Giardina CP, Percy KE, Miglietta F, et al. (2007) Future atmospheric CO2 leads to delayed autumnal senescence. Glob Chang Biol 14: 264–275.
  52. 52. Jeong SJ, Ho CH, Gim HJ, Brown ME (2011) Phenology shifts at start vs. end of growing season in temperate vegetation over the Northern Hemisphere for the period 1982–2008. Glob Chang Biol 17: 2385–2399.
  53. 53. Mooney HA (1972) The carbon balance of plants. Annu Rev Ecol Syst 3: 315–346.
  54. 54. Chapin FS (1980) The mineral nutrition of wild plants. Annu Rev Ecol Syst 11: 233–260.
  55. 55. Reich PB, Uhl C, Waiters MB, Ellsworth DS (1991) Leaf lifespan as a determinant of leaf structure and function among 23 amazonian tree species. Oecologia 16–24.
  56. 56. Nippert JB, Wieme Ra, Ocheltree TW, Craine JM (2012) Root characteristics of C4 grasses limit reliance on deep soil water in tallgrass prairie. Plant Soil 355: 385–394.
  57. 57. Holt RD (2009) Bringing the Hutchinsonian niche into the 21st century: Ecological and evolutionary perspectives. Proceedings of the National Academy of Sciences 106: 19659–19665.
  58. 58. Reich PB (2014) The world-wide “fast-slow” plant economics spectrum: a traits manifesto. Journal of Ecology 102: 275–301.