1. Introduction
Temperate and tropical forests make up much of the world’s biomass, with tropical forests alone accounting for over 60% of terrestrial global carbon [
1]. Forests can mitigate the effects of climate change, such as elevated temperature, through carbon dioxide (CO
2) uptake during photosynthesis. However, global surface-air temperature is expected to increase by 1 to 5 °C by 2100, with an even greater increase expected in northern latitudes, along with an increase in the frequency and intensity of short-term heat waves [
2,
3]. As temperatures increase, there is a theoretical thermal tipping point after which photosynthesis begins to decline while plant respiration may still be increasing. If declines in CO
2 uptake are severe enough that forest-wide respiration exceeds photosynthesis, forests could become net sources of carbon to the atmosphere [
4,
5]. Determining where this thermal threshold exists and whether different forest types are close to shifting from carbon sources to sinks is crucial to understanding and modeling global climate feedbacks [
6].
Many global vegetation models are based on Farquhar’s model of photosynthesis [
7], but our ability to accurately parameterize these models is hampered by a lack of mechanistic data on physiological responses to warming and thermal acclimation potential of canopy photosynthesis and respiration, especially for tropical forests [
8,
9,
10,
11,
12,
13,
14,
15,
16]. Models analyzing the responses of tropical carbon storage to elevated temperature have produced inconsistent predictions on whether these forests will remain sinks or become sources [
14,
17]. In part, this is because there is great variability among tropical forests in annual temperature, precipitation, soil types, geographic range, and species assemblage; likely showing a variety of responses to changing climate across different systems [
18,
19,
20]. Additionally, the greatest uncertainties in model projections of global carbon balance over the next century are for tropical latitudes, due to a combination of large carbon fluxes and a severe lack of relevant tropical-specific data [
14]. Since tropical forests in particular account for such a large proportion of the global carbon cycle, any shifts in tropical photosynthesis could alter global carbon balances and feedbacks to climate change.
Though canopy warming studies will be crucial in determining whether different forest types can physiologically acclimate [
9,
11,
14,
21,
22,
23], instantaneous photosynthesis measurements in response to a range of temperatures are necessary to improve models and set a baseline to determine how close these forests are to the temperature optima for photosynthesis. Even in temperate forests, there is a scarcity of data of the short-term effects of elevated temperatures on photosynthesis in mature forest canopies and for long-term temperature acclimation [
24].
Light saturated photosynthetic response to temperature increases to a maximum (
Aopt) at an optimum leaf temperature (
Topt), then decreases again, following a parabolic curve (
Figure 1) [
7,
25,
26].
Below
Topt, photosynthesis typically increases as the rates of biochemical reactions increase with temperature due to low-temperature limited enzymes [
26,
27]. In most temperate C
3 species, photosynthetic decline occurs at leaf temperatures between 26 and 32 °C, due to direct biochemical limitations and/or to indirect stomatal responses to increases in vapor pressure deficit [
7,
26]. The two components of net photosynthesis commonly used for model parameterization are the rate of ribulose-1,5-bisphosphate (RuBP) carboxylation (
Vcmax) and the rate of RuBP regeneration from the electron transport chain (
Jmax) [
7,
28]. At elevated leaf temperatures, primary limitations to net photosynthesis include: temperature sensitivity to Ribulose-1,5-bisphosphate carboxylase oxygenase activase (Rubisco activase) and thylakoid membrane dysfunction, which affects efficiency of light reactions and electron transport [
29,
30], and increased rates of photorespiration due to the decrease of both specificity of Rubisco for CO
2 and the relative solubility of CO
2 versus O
2 [
31]. Irreversible leaf damage from high temperatures typically begins above 45 °C as plasma membranes in the chloroplast become permeable and proteins begin to denature [
26].
Perhaps due to relatively low diurnal and inter-annual variability, tropical trees have been shown to be more susceptible to warming in comparison to temperate and boreal species, and have shown photosynthetic declines at temperatures of only 3 °C above mean ambient air temperatures [
32,
33]. Recent evidence suggests that lowland tropical forests, which are already consistently warm throughout the year, may already be operating close to their photosynthetic thermal optima, beyond which carbon uptake declines [
5,
22,
34,
35,
36,
37,
38]. Due to greater diurnal, seasonal, and inter-annual variability in temperature in temperate ecosystems, the range of temperature from ambient to
Topt will likely be greater in temperate species and therefore these high latitude species may not be operating as close to their thermal thresholds as tropical species [
25,
32,
39,
40,
41].
For accurate model representation, it is important to measure photosynthesis across the entire vertical transect of the canopy, as upper and lower branches have different traits, are exposed to different environmental gradients, and can respond differently to temperature [
42]. For example, light conditions are highly variable across a vertical gradient within canopies due to density of foliage limiting light access in the lower canopy [
40,
43], and light can be especially limiting in ecosystems with high annual cloud cover and mean annual precipitation (MAP), such as tropical rainforests [
44]. As light availability typically increases with canopy height, leaf temperature does as well. The direct radiation hitting upper canopy leaves warms sun leaves more than diffuse scattered light in lower shaded leaves [
45]. In many forests, leaf temperatures can be 1–7 °C above ambient air temperature, especially in the upper canopy [
46]. Therefore, upper canopy leaves of both temperate and tropical species have been shown to have greater photosynthetic capacity, but they also may be more susceptible to heat stress due to elevated temperatures and vapor pressure deficit [
47].
Various leaf functional traits can also be strong predictors of photosynthetic capacity and, in turn, photosynthetic response to elevated temperature. Leaf nitrogen (N) content in canopies has been found to correlate with greater photosynthetic capacity in various forest types [
48,
49,
50], and leaf-area based phosphorus (P) has also shown similar patterns across upper canopy gradients as light-saturated rates of photosynthesis [
51,
52,
53]. Additionally, leaf mass per area (LMA) is an extremely useful leaf trait that correlates with numerous within-canopy traits that are more challenging to measure (e.g., gas exchange, leaf nutrients) [
54]. LMA typically increases with height in temperate and tropical forest canopies, either due to light or hydrostatic gradients [
55,
56,
57] and since LMA has been found to have significant relationships with light saturated rates of photosynthesis (
Amax), it may also be a strong predictor of maximum rates of photosynthesis under varying temperatures [
12,
56,
58]. Thus, leaf P, N, and/or LMA may be able to predict photosynthetic temperature response parameters (
Aopt and
Topt). The highly weathered clay soils of tropical forests are often considered to be more limited by P while the younger temperate forests are generally considered to be more limited by N [
59,
60,
61]. We may therefore expect P to be strongly correlated to photosynthetic parameters in the tropical forest with N more important in the temperate forest canopies [
62,
63,
64].
Overall, this study sought to investigate how ambient leaf temperatures and instantaneous photosynthetic responses to temperature varied within mature tree canopies and across three forest types: temperate deciduous, tropical moist, and tropical wet, in order to see how close to their physiological temperature optima the dominant tree species of these forests may be operating. Our specific hypotheses were:
- (A)
Compared to lower leaves, upper canopy leaves at all sites will have higher temperature optima for photosynthesis (Topt), higher maximum rates of photosynthesis (Aopt), and both parameters will increase with increasing light availability.
- (B)
Topt and Aopt will increase with increasing mean annual temperatures across forest types, but tropical tree species will be functioning closer to photosynthetic thresholds than temperate species.
- (C)
Topt and Aopt will increase with increasing LMA at all sites but will be better predicted by leaf N in the temperate site, and leaf P at the tropical sites.
2. Materials and Methods
Sampling took place from four canopy access towers at three sites: a temperate deciduous forest in Upper Peninsula Michigan and two sites in Puerto Rico. The Puerto Rico sites were defined as subtropical moist and subtropical wet forests, respectively by the Holdridge life zone system and both tropical forests were dominated by broadleaf evergreen tree species [
19].
2.1. Temperate Deciduous Forest
The temperate deciduous study site was located at the Michigan Tech Ford Center and Forest near L’Anse, MI (46°64′ N, 88°48′ W). Mean annual temperature (MAT) was 5 °C, though mean annual growing temperature (MAGT) was 17.4 °C. Mean annual precipitation (MAP) was 879 mm from 2009 to 2011, with 401 mm occurring during the growing season, typically May through September [
65]. The soil at the site was an Allouez gravelly coarse sandy loam, extending to a depth of 40 cm (
http://websoilsurvey.sc.egov.usda.gov/).
Acer saccharum was the dominant species on the site, making up 97% of total tree density, with only five species per hectare at the site. Other non-dominant species included
Ulmus americana,
Tilia americana,
Betula alleghaniensis, and
Ostrya virginiana. The average height of the canopy was 23 m, while the average height to live crown was 14 m [
56]. For additional information about forest composition, structure, and site history, see Campione et al. [
66].
Gas exchange and leaf trait data were acquired at the temperate site with a 19-m tall mobile aluminum walk-up tower (Upright, Inc., Selma, CA, USA), constructed in the summer of 2012.
2.2. Tropical Moist Forest
The tropical moist forest study site was located in El Tallonal Forest Reserve, a 114 ha privately owned natural area in the municipality of Arecibo, Puerto Rico (18°40′ N, 66°73′ W). The MAT was 25.5 °C and the MAP was 1295 mm (
Table 1). The wet season at this site lasts from July to September, while the dry season is generally from January to March. This north central region of Puerto Rico is dominated by karst topography and has soils derived from limestone parent material with an overlay of Inceptisols in the low lying valleys [
67]. There are an estimated 24 tree species per hectare at the site, and the dominant tree species is
Castilla elastica [
68]. The forest was estimated to be over 70 years old at the time of the study [
68,
69]. Historically, the area was used for grazing cattle and agriculture until 1950, when these lands were abandoned and naturally regenerated to forest [
69].
Two identical 25 m steel towers, constructed 12 m apart in 2008, were used at the tropical moist forest site (BilJax, Archbold, OH, USA).
C. elastica, Guarea guidonia, and
Ocotea leucoxylon were measured from these towers.
C. elastica is a fast growing, early successional broadleaved evergreen tree with relatively high LMA and light saturated photosynthetic rates (
Amax) [
70].
G. guidonia is a mid-successional deciduous tree native to Central and South America [
71,
72].
O. leucoxylon, which is also a Caribbean native, is an early successional broad-leaved evergreen tree [
73].
Leaves of these three species were sampled from seven heights (7, 12, 14, 16, 20, 22, and 25 m) across both towers. Other species were accessible, however, they were not sampled if they were not present at a minimum of two heights.
2.3. Tropical Wet Forest
The tropical wet forest was located in the Luquillo Experimental Forest in northeastern Puerto Rico (18°31′ N, 65°74′ W) [
19,
74,
75,
76]. MAP of the site was 3936 mm and MAT is 25 °C, at an elevation of 361 m above sea level [
74]. Temperature varies just 4 °C from month to month throughout the year with no days with temperatures below freezing. A drier season typically occurs from January through April, while the wettest period ranges from May to November [
75]. The soils at the tropical wet forest are Ultisols, clayey poorly drained soil with red mottles [
75]. This forest as a whole averages 10 tree species per hectare, with an average of 516 stems per ha [
77,
78]. The average height of the canopy was 20 m [
79].
Dacryodes excelsa (common name: tabonuco), was the dominant tree species of the forest [
80].
Dacryodes excelsa is one of the dominant native trees in Puerto Rico, growing at elevations between 200 to 800 m [
81]. Other species located near the tower include:
Prestoea montana, Casearia arborea, Inga laurina, Manilkara bidentata, and
Sloanea berteriana [
82]. For additional site history, see [
75,
83].
The tropical wet forest tower was a 24.7-m tall mobile aluminum walk-up built in 1991, with a footprint of 2.5 m
2 (Upright, Inc., Selma, CA, USA) (
Table 1).
Dacryodes excelsa was the only species accessible across the entire vertical canopy gradient.
Dacryodes excelsa is a long-lived broad-leaved evergreen species, and it is estimated that mature trees live up to 400 years with average heights of 30 to 35 m [
81].
Dacryodes excelsa has historically shown relatively low light saturated photosynthesis, about 2.7 µmol CO
2 m
−2 s
−1 [
84]. Leaves were sampled from seven heights (8, 11, 15, 18, 20, 23, and 25 m).
2.4. Environmental Measurements
Diffuse non-interceptance (DIFN%), defined as diffuse light transmitted through the canopy as a proportion of incident (above canopy) radiation, was measured at each study site with an LAI-2200 (LI-COR Biosciences, Lincoln, NE, USA). DIFN% in forests is closely related to seasonally integrated photosynthetic photon flux density [
85] and we use it here as a measurement of canopy openness and light availability integrated over time. ‘Above canopy’ measurements were taken at the top of the towers, and measurements were taken at every sampling height for each tower at each forest site. DIFN% was only sampled under uniformly cloudy conditions whenever possible at mid-day for all three sites, and a white diffuser cap was used to correct for direct sunlight when uniformly cloudy conditions were not present. DIFN% was measured on 1 June 2014 in Puerto Rico, and on 20 June 2014 in the temperate forest.
Leaf temperature was measured using a Fluke 572 infrared thermometer (Fluke Corporation, Everett, WA, USA). Five leaves of all accessible species were measured at every sampling height once an hour, for three days at all sites. Leaf temperature was measured from the 1–5 June and the 4–22 August at the tropical moist and tropical wet sites, and 8–16 August at the temperate deciduous site.
Air temperature was measured in thirty-minute increments and automatically logged from 26 May–27 August 2014 at the tropical moist and wet forests (HOBO Pro V2 temp/RH, Onset Computer Corporation, Cape Cod, MA, USA). Air temperature sensors in solar shields were tethered to the towers with zip ties at 5, 15, and 25 m at the tropical moist forest, and at 7, 13, 18, and 23 m at the tropical wet forest. Air temperature data were collected at the temperate deciduous forest from May to August of 2013 at heights 3, 8, and 11 m.
2.5. Sampling Design for Photosynthetic Measurements
In the temperate forest, photosynthetic temperature response curves were conducted on two leaves of Acer saccharum at every accessible height of the tower. Leaves were collected from four individual mature trees between 7 to 20 m, and from five individual understory saplings at less than 1 m in height. Photosynthetic light response curves were conducted on two healthy mature leaves at the top (20 m), middle (11 m), and understory (<1 m) to extract the light saturation point for use in temperature response curves. All physiological measurements were performed from the towers on intact leaves. All sampling from the temperate forest site took place during the growing season, from 7–21 July 2014 between 8 am and 4 pm.
In the tropical wet forest site, two mature leaves of Dacryodes excelsa were measured at every height for the temperature response curves from three accessible trees. Light response curves were taken on two healthy mature leaves at 25, 20, and 11 m to extract the light saturation point (the light concentration at which photosynthesis levels off). Sampling took place in two field campaigns: one from 6–9 June 2014 and another from 6–28 August 2014 between 8 am and 4 pm.
In the tropical moist site, two leaves of each species (
C. elastica from four trees,
G. guidonia two trees, and
O. leucoxylon two trees) were sampled at each accessible height for the photosynthetic temperature response curves.
C. elastica was sampled at 12, 14, 16, 20 and 22 m,
G. guidonia at 8 and 12 m, and
O. leucoxylon at 20, 22, and 25 m. The light saturation point used for these species was from a previous study [
86]. Sampling took place between 9 am and 5 pm in two campaigns: one from 11–13 June 2014 and another from 13–22 August 2014.
2.6. Photosynthesis Measurements and Parameter Extractions
Photosynthesis measurements were taken with a LI-6400XT with a 6400-02B Red/Blue light source (LI-COR Biosciences, Lincoln, NE, USA). Light response curves were conducted to obtain the light saturation point (
Isat) for each species, since it is a necessary parameter for the temperature response curves. Temperature response curves were conducted using a water jacket (6400-88 expanded temperature kit, Licor Biosciences, Lincoln, NE, USA) to increase and decrease leaf temperature. Leaf temperatures were achieved by introducing heated or cooled water through a Bev-A-Line tube through the water jacket and awaiting equilibrium at target temperatures. Water was continuously introduced through the system using gravity, with receptacles both above and below the measurement chamber. Using this method, we measured photosynthetic rates at 22, 24, 27, 29, 30, 31, 32, 33, and 35 °C. Relative humidity was kept between 50% and 60%, and was regulated by flow (200 umols
−1 for temperate samples and 300 umols
−1 for tropical samples) and desiccant. CO
2 was kept at ambient concentrations of 400 ppm, and Photosynthetically active radiation (PAR)was kept at the average
Isat for each species (800, 700, 1200, 1300, and 1200 µmol m
−2 s
−1 respectively for Dac, Cas, Acer, Oco, and Gua) as extracted from light response curves.
Topt was calculated by taking the second order polynomial function on the temperature response curves and solving for the highest point on the x-axis, and
Aopt was extracted by solving for the photosynthetic value from the regression equation at
Topt (
Figure 1) [
87].
2.7. Leaf Traits
Leaf area for all leaves measured for gas exchange at the temperate site was measured using a leaf area meter (LI-3100 LI-COR Biosciences, Lincoln, NE, USA). The leaf area of leaves measured for gas exchanged at the tropical sites was measured by scanning pictures of leaves alongside a ruler (HP deskjet), then tracing the outline of the scanned image in ImageJ (Rasband, W.J 1997–2014,
http://imagej.nih.gov/ij). Leaves were then dried for 24 h at 65 °C and weighed. Leaf mass per area (LMA; g m
−2) was determined by taking the ratio of dry weight to leaf area. Samples were ground with a ball bearing grinder (8000 M Mixer/Mill, Spex Sample Prep, Metuchen, NJ, USA) for three minutes. Prior to elemental analysis, ground samples were dried for an additional 24 h at 65 °C, weighed with a Sartorius Cubis microbalance (Data Weighing Systems, Elk Grove, IL, USA) for 4–6 mg of leaf material, and folded in to 5 by 9 mm tin capsules. Leaf nitrogen (N) analysis was performed using an Elementar vario Microcube elemental analyzer (Elementar Inc., Hanau, Germany). Leaf phosphorus (P) analysis was analyzed with inductively coupled plasma optical emission spectrometry using a Thermo Jarrell Ash IRIS Advantage Inductively Coupled Plasma Optimal Emission Spectrometer (Precision Dynamics Corporation, San Fernando, CA, USA).
2.8. Data Analysis
An F-protected Least Significant Difference Test was performed to determine if means were significantly different among species and forest types. For all subsequent regression analyses, species data within tropical moist forest were pooled. Maximum daily leaf temperature (TLEAFmax), maximum daily air temperature (TAIRmax), and photosynthetic temperature optima (Topt) of leaves were each analyzed with height using simple linear regression. The points where TLEAFmax and Topt intersected were found by algebraically comparing where the coordinates of Topt = a1(height) + b1 and TLEAFmax = a2(height) + b2 intersected. Simple linear regression was also used to compare Topt and Aopt with foliar nutrients (Nmass, Narea, Pmass, Parea), and LMA. DIFN% was expressed as a percent of light reaching each canopy level graphically and was transformed with a natural log function for statistical analyses. All statistical analyses were performed in R (R Core Team, Vienna, Austria, 2013).
5. Conclusions
In the upper canopy of all forest types, TLEAFmax was found to exceed the temperature optima for photosynthesis, indicating that leaves in the upper canopies of these forests are already showing declines as a result of high temperatures. The tropical moist forest, which had the highest mean annual temperature of all sites, had the greatest amount of canopy operating above the thermal optima. Contrary to expectations, the thermal optima for photosynthesis exceeded maximum leaf temperature in the temperate forest, more so than the tropical wet forest. Unless forest canopies can acclimate to increasing temperature and shift thermal optima for photosynthesis higher, a majority of the canopy of each site could be operating above the thermal optima by 2100, leading to potential global decreases in carbon storage.
As seen in previous studies, LMA and N (on an area and mass basis) were significant predictors of
Aopt for the temperate forest and showed a strong correlation with
Topt as well. The relationships in both tropical forests, however, were not as pronounced. Contrary to predictions, P was not significantly related to
Aopt or
Topt in either tropical forest. This could indicate that N and P were not limiting in the respective forests, though additional sampling could further elaborate on this. For future studies, more data on the mechanistic responses of photosynthesis will be needed, such as rates of electron transport and RuBP carboxylation, to determine what is limiting. In situ warming experiments could prove to be very useful in determining if acclimation to temperature does occur and to what extent. While additional data for model parameterization are needed, atmosphere-biosphere models are only as accurate as their underlying species-specific or plant functional type-specific physiological data, and different species show different thermal tolerances [
8]. This could be critical for atmosphere-biosphere models and improving parameterization and for understanding changes in carbon flux due to climate change.