- Email: [email protected]
Temperature preference across life stages and acclimation temperatures investigated in four species of Drosophila
Journal Pre-proof Temperature preference across life stages and acclimation temperatures investigated in four species of Drosophila Heidi J. MacLean, Johannes Overgaard, Torsten N. Kristensen, Catrine Lyster, Leander Hessner, Esajas Olsvig, Jesper G. Sørensen PII:
S0306-4565(19)30345-6
DOI:
https://doi.org/10.1016/j.jtherbio.2019.102428
Reference:
TB 102428
To appear in:
Journal of Thermal Biology
Received Date: 28 June 2019 Revised Date:
24 September 2019
Accepted Date: 4 October 2019
Please cite this article as: MacLean, H.J., Overgaard, J., Kristensen, T.N., Lyster, C., Hessner, L., Olsvig, E., Sørensen, J.G., Temperature preference across life stages and acclimation temperatures investigated in four species of Drosophila, Journal of Thermal Biology (2019), doi: https:// doi.org/10.1016/j.jtherbio.2019.102428. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.
1
Temperature preference across life stages and acclimation temperatures investigated in
2
four species of Drosophila
3
4 5
Heidi J MacLean1*, Johannes Overgaard1, Torsten N Kristensen2, Catrine Lyster1, Leander
6
Hessner1, Esajas Olsvig1, Jesper G Sørensen1
7
8
1
Department for Bioscience, Aarhus University, Ny Munkegade 116, 8000 Aarhus C, Denmark
9
2
Department of Chemistry and Bioscience, Aalborg University, Fredrik Bajers Vej 7H, 9220
10
Aalborg E, Denmark
11
12
*Corresponding author: [email protected]
13
1
14
Abstract: Ectotherms can use microclimatic variation and behavioral thermoregulation to cope
15
with unfavorable environmental temperatures. However, relatively little is known about how and
16
if thermoregulatory behavior is used across life-stages in small ectothermic insects. Here we
17
investigate differences between three specialized Drosophila species from temperate, tropical or
18
desert habitats and one cosmopolitan species by estimating the preferred temperature (Tpref) and
19
the breadth (Tbreadth) of the distribution of adults, adult egg-laying, and larvae in thermal
20
gradients. We also assess the plasticity of thermal preference following developmental
21
acclimation to three constant temperatures. For egg-laying and larvae we observe significant
22
species differences in preferred temperature but this is not predicted by thermal ecology of the
23
species. We corroborated this with previous studies of other Drosophila species and found that
24
Tpref for egg laying and larvae have no relationship with annual mean temperature of the species
25
natural habitat. While adults have the greatest mobility, they show the greater variation in
26
preference compared to juveniles contradicting common assumptions. We found evidence of
27
developmental thermal acclimation for adult egg-laying preferred temperature, Tpref increasing
28
with acclimation temperature, and in the breadth of the temperature preference distributions,
29
Tbreadth decreasing with increasing acclimation temperature. Together, these data provide a high
30
resolution and comprehensive look at temperature preferences across life stages and in response
31
to acclimation. Results suggest, that thermal preference, particularly in the early life stages, is
32
relatively conserved among species and unrelated to temperature at species origin. Measuring
33
thermal preference, in addition to thermal performance, is essential for understanding how
34
species have adapted/will adapt to their thermal environment.
35 36
Key words: thermoregulatory behavior, climate change, thermal preference, thermal optima 2
37 38
1. Introduction:
39
The thermal biology of ectotherms is often investigated by assessing extreme thermal tolerance
40
measures or by measuring thermal performance of various traits across the permissive
41
temperature range (Angilletta 2009; Sørensen et al. 2018). Using thermal tolerance as a proxy for
42
thermal adaptation enable detection of an association between species and the thermal
43
characteristics of their habitats (Magnuson et al., 1979; Kellermann et al., 2009, 2012; Sunday et
44
al., 2011; Overgaard et al., 2014; Andersen et al., 2015). Further, characterization of thermal
45
performance curves can be used to inform spatial distribution models (SDMs) and parameterize
46
modelling approaches of population growth rates (Kearney & Porter, 2009; Richardson et al.,
47
2011; Buckley & Kingsolver, 2012). However, the assessments of tolerance and performance are
48
typically studied under conditions where behavioral responses are constrained and therefore fail
49
to capture how insects navigate temperature and temperature variation through behavioral choice
50
(but see Woods et al., 2015).
51
Behavior can allow organisms to exploit microclimatic variation at a small spatial scale and
52
potentially avoid unfavorable environmental temperatures (Huey, Hertz & Sinervo 2003; Potter,
53
Woods & Pincebourde 2013; Sunday et al. 2014; Buckley, Ehrenberger & Angilletta 2015).
54
Thus, behavioral thermoregulation may relieve some organisms from selective pressure on
55
thermal physiological traits and explain cases where no clear adaptive variation exists among
56
species or populations despite substantial climate/seasonality differences (see Hoffmann,
57
Sørensen & Loeschcke 2003). In a recent study, we found strikingly similar thermal optima for a
58
component of fitness based on three life-history traits (fecundity, developmental time and egg-to-
59
adult viability) among 22 species of Drosophila despite representing markedly different ecotypes
3
60
(MacLean et al. 2019). Here we assess temperature preference across life-stages of four
61
Drosophila species to determine if behavior can partly explain the apparent lack of variation in
62
thermal optima across Drosophila species.
63
The knowledge of behavioral thermoregulation is well-established in large ectotherms such as
64
lizards that can use behavioral thermoregulation to achieve body temperatures that deviate
65
considerably from ambient air temperatures (Hertz, Huey & Stevenson 1993). Small ectotherms
66
also rely on behavioral thermoregulation (Hoffmann, Sørensen & Loeschcke 2003; Dillon et al.
67
2009). For example, Drosophila subobscura and D. melanogaster overwintering in compost
68
heaps have the potential to avoid extreme temperatures in their natural habitat (Schou, Loeschcke
69
& Kristensen 2015; Sørensen et al. 2015). Insects with complex life-cycles may face different
70
selection pressures because of e.g. life-stage-specific thermal sensitivity, difference in
71
microclimates available, or different mobility of different life stages (Kingsolver et al. 2011;
72
Potter, Woods & Pincebourde 2013; Woods, Dillon & Pincebourde 2015). For example, local
73
adaptation in thermal performance have been reported in the larvae (Higgins et al. 2014) but not
74
adults of a highly-thermoregulating butterfly species (MacLean et al. 2016). Importantly, the
75
variation in thermoregulatory behavior (thermal preferences) across life stage and the plasticity
76
in preferences resulting from developmental acclimation are poorly investigated (but see
77
Rajpurohit & Schmidt 2016).
78
Existing knowledge on thermal preference and thermal preference plasticity in Drosophila
79
typically stem from studies using one or two species, and comparisons among studies are
80
challenged by different treatments (e.g. acclimation) and methodology (see Table S1). To
81
address this lack of knowledge, we performed a comprehensive study looking at four species
82
with distinct thermal ecology and assessed among life stages using standardized, comparable 4
83
methods. We compared the thermoregulatory behavior of adult and larvae Drosophila; D.
84
melanogaster, D. birchii, D. subobscura, and D. mojavensis representing cosmopolitan, tropical,
85
temperate, and desert habitats, respectively. We hypothesize that thermal preference should be
86
influenced by both habitat type (evolutionary history) and short-term thermal experience
87
(ecological history). Further, we expect that adult behavior (both egg-placement and adult
88
choice) should be more narrowly distributed across favorable temperatures due to their increased
89
mobility. By measuring the distributions of thermal preference (Figure 1) in different life stages
90
we examined if different Drosophila species from different environments show adaptive
91
differences in thermal preference. Specifically, that species from warmer environments prefer
92
warmer temperatures. We also investigated the effect of developmental acclimation at three
93
temperatures in adult egg-laying and adult thermal preference to investigate the potential
94
adaptive role of phenotypic plasticity in thermal preference. We expect that increased
95
developmental acclimation temperature would translate into increased thermal preference. We
96
then analyzed the patterns of preference among life stages within and among species to test the
97
prediction that mobile adults might rely more on behavior than less mobile juveniles.
98
99
2. Materials and Methods
100
2.1 Experimental animals
101
To quantify species specific differences in thermal preference (Tpref) we used populations of four
102
Drosophila species reared under common conditions (19 °C, 12/12 L:D). The four species
103
represent disparate habitat specializations: Drosophila melanogaster (cosmopolitan), D. birchii
104
(tropical), D, subobscura (cold-hardy, temperate), and D. mojavensis (xeric, desert) (summarized 5
105
in Table S2). All four species were bred on standard Drosophila medium. Each species was
106
reared in two bottles with 50 mL of fly food at 19°C and controlled larval density (unless
107
otherwise noted).
108
109
2.2 Adult and egg-laying thermal preference
110
Prior to experimentation, newly emerged adults of both sexes were transferred to fresh fly food
111
and fed yeast until they were sexually mature (1-4 days depending on the species). During the
112
thermal preference experiments the flies were placed in a linear thermal gradient with
113
temperature spanning from 9 to 34°C. The thermal gradient was created by a large aluminum
114
block (156 cm L x 60 cm W, Figure S1A) which was cooled/heated using a circulating water
115
bath at either end. Eight horizontal lanes were established in parallel on the thermal gradient (6
116
cm W x 3cm H) and in each lane, we placed 13 petri dishes (diameter 5.5 cm) containing 14 mL
117
of standard fly food. Petri dishes with food were placed every 10 cm and petri dishes filled with
118
silica gel were placed in between to mitigate condensation (Figure S1A). The lanes were covered
119
with a Plexiglas lid and weighed down to prevent individuals from migrating between lanes or
120
escaping the experimental setup. Approximately 60 flies of mixed sex (sd = ± 22) were loaded
121
through holes across the gradient (at 6, 43, 77, 117 and 150 cm corresponding to temperatures of
122
approximately 10, 14, 21, 27 and 33°C of the temperature gradient). While the number of adults
123
was variable, previous work on thermal preference in Drosophila found no effect of number of
124
flies tested on the preferred temperature (Krstevska & Hoffmann, 1994). Flies were given 20
125
hours to explore and lay eggs. Hereafter, adults were anesthetized on the gradient by flushing the
126
air with CO2 to anaesthetize the flies and the distribution of adults was then registered by
6
127
counting number of adults in each 10-cm bin (representing approximately a span of 1.6°C within
128
the gradient). We observed that the application of CO2 simply slowed the activity until the flies
129
entered into a comatose state and did not result in any extraordinary activity. Subsequent to the
130
experiment, we counted the number of eggs in each petri dish to establish the distribution of egg-
131
laying preference. To ensure that the obtained data reflected the thermal “choice” on the gradient
132
rather than an effect of loading position or edge, we ran 10 replicate controls for D. melanogaster
133
(and two replicate controls for all other species) using the apparatus without applying a thermal
134
gradient. Under these conditions, we found adult presence and egg laying to be evenly
135
distributed across the entire gradient (data not shown). When the gradient was turned on, we did
136
encounter a “loading effect” for D. birchii at the cold end of the gradient (10°C) so this species
137
was subsequently loaded at 14, 21, 27°C. See discussion for further details.
138
To investigate the effect of thermal acclimation on both adult thermal preference and egg-laying
139
thermal preference we compared flies that had developed and been reared at three different
140
acclimation temperatures (15, 19, 23°C for D. subobscura; 19, 23, 27°C for the other three
141
species). In these experiments, we tested the acclimation treatments for a given species in
142
different lanes of the same round of experiments to ensure that any difference was not associated
143
with block effects.
144
145
2.3 Larval thermal preference
146
Larval thermal preference was assayed on a smaller thermal gradient established on an aluminum
147
plate (800 mm L x 200 mm W x10 mm H) with temperature controlled at both ends by
148
individual Peltier devices. The Peltier devices established a largely linear temperature gradient 7
149
over the length of the apparatus running from 17-31°C on the surface (Figure S1B). Flat
150
bottomed U-shaped aluminum lanes (800 mm L x 20 mm W x 10 mm H) placed in parallel and
151
spanning the length of the apparatus, were filled with 100 mL of standard fly food (Figure S1B).
152
The apparatus was placed in an insulated box and the experimental arena was covered by a
153
Plexiglas lid weighted down to prevent individuals from migrating across lanes. Up to eight
154
lanes were run simultaneously surrounded by empty lanes (to avoid edge effects). For each lane,
155
we transferred 200 eggs evenly spaced on the gradient (in groups of 10 eggs). Egg-laying was
156
timed so that we could collect eggs close to hatching. The eggs were allowed to hatch and larvae
157
to randomly distribute for 48 h at 19°C. The whole lane was then placed on the thermal gradient
158
for 24 h to allow larvae relocate to their preferred temperature. After this treatment, the lanes
159
were removed from the temperature gradient and the food from each lane was divided into 4 cm
160
bins (20 bins total – each representing approximately 0.7°C of gradient). The food was then
161
transferred to a vial with 7 mL additional fly food and all vials were returned to 19°C to allow
162
the larvae to develop. After 18-24 days when all larvae had developed we counted the number of
163
emerging adults in each vial. We could also use this test to verify that there was no migration
164
between lanes as only one species emerged from each lane. Recovery (survival) rates were
165
similar among species and replicates (mean % ± sem, D. melanogaster (N = 946): 76.2 ± 3.1%;
166
D. birchii (N = 374): 71.1 ± 1.5%; D. subobscura (N = 674): 67.9 ± 5.6%; D. mojavensis (N =
167
1503): 77.1 ± 4.1%, respectively). Similar to the experiments with adult flies, controls for larval
168
distribution were run without the Peltier devices turned on (i.e. no thermal gradient) to verify that
169
the observed patterns of thermal preference were a function of temperature. Under these
170
conditions, we found larval preference to be evenly distributed across the entire width of the
171
gradient (data not shown).
8
172
2.4 Analysis
173
All analyses were conducted in R version 3.3.3. To account for variation in total number of
174
adults used, eggs laid or larvae recovered among different replicates, we standardized the data by
175
calculating the proportional distribution per replicate lane. Two different approaches were
176
employed to estimate preferred temperature (Tpref). First, we selected the temperature
177
corresponding to the highest proportional value (Tpref_max). Second, for each lane we multiplied
178
the proportion of individuals in each bin by the corresponding temperature and summed these
179
values across all bins to determine a weighted average (Tpref_ave). However, as the two measures
180
were strongly correlated across all data (Pearson's product-moment correlation, t(26) = 18.2, p <
181
0.05, r = 0.96, Figure S2) we continued analyses with only the weighted average (Tpref_ave) as our
182
estimate for Tpref (see Figure 1).
183
We fitted a generalized additive model (GAM) using the mcgv package (Wood 2001) with a 0.01
184
smoothing parameter to the proportional distribution data for each species, life stage, and
185
acclimation combination. Based on this model we estimated the temperature at which 10% (P10)
186
and 90% (P90) of the animals occur. The range between these two (P90 – P10) yielded breadth
187
(Tbreadth) of the distribution (all estimates are summarized in Table 1, Figure 2).
188
Further, we assessed general relationships between the thermal preferences (at different life
189
stages) and environmental temperatures of species origin, across data from this study and from
190
the literature. To do this, we fitted linear models for Tpref as a function of annual mean
191
temperature for each species (as given in Kellermann et al. 2012a) across species and life stages,
192
respectively (Figure 3).
9
193
Finally, we correlated the thermal preferences with the thermal optima estimated in for egg-
194
laying and overall fitness in MacLean (2019). Specifically, we correlated egg-laying thermal
195
optima with egg-laying preference and overall fitness thermal optima with adult thermal
196
preference for all developmental acclimation regimes.
197
3. Results
198
We examined if Drosophila exhibit different preferred temperatures (Tpref) as a function of
199
species, acclimation or life stage life-stage. We found small, but significant differences between
200
species in Tpref for adult egg-laying and for larval preference (Table 2) but these differences are
201
not ordered by thermal habitat/ecotype of each species. For adults, we find larger differences in
202
Tpref between the four species (Figure 2, Table 2). We found a large difference in Tbreadth between
203
the four species in the adult and the adult egg-laying choice, but not in larvae choice (Figure 2,
204
Table 2). Analysis of both P10 and P90 show species effects for larvae, adult and adult egg-laying
205
choice (Table 2, Figure S3).
206
To determine if developmental thermal acclimation (plasticity) alters the preferred temperature
207
of adults or adult egg-laying we investigated the distribution of adult female flies and their egg-
208
laying when reared at 19, 23, and 27°C (or at 15, 19 and 23°C for D. subobscura).
209
Developmental acclimation did significantly increase Tpref of adult egg-laying preference, while
210
no significance of developmental temperature was found for adult preference (Figure S3, Table
211
2). The strongest acclimation response in egg-laying Tpref (shown as increased Tpref °C/°C
212
acclimation) was found for D. birchii (where Tpref increased by 0.46°C/°C acclimation) followed
213
by D. subobscura (0.39°C/°C acclimation), D. mojavensis (0.25°C/°C acclimation) and the
214
weakest response was observed in D. melanogaster (0.15°C/°C acclimation). Further, there was
10
215
a significant decrease in egg-laying Tbreadth as a function of increasing acclimation temperature
216
(Table 2, Figure S3). When we consider P10 and P90 individually we see evidence of the curve
217
shifting towards higher temperatures in egg-laying distribution with increasing acclimation
218
temperature. The response to acclimation for adult egg-laying P10 and P90 were on average ~0.36
219
and ~0.18°C/°C acclimation, respectively, however it should be noted that variation among
220
species was marked. This result suggest that the upper marginal bound of the distribution is
221
potentially less labile than the lower bound and the body of the distribution is centered around
222
Tpref. All species except D. melanogaster showed a stronger acclimation response for P10 as
223
compared to P90 (D. birchii: 0.48 & 0.31°C/°C acclimation, D. melanogaster: 0.11 & 0.15°C/°C
224
acclimation, D. mojavensis: 0.30 & 0.07°C/°C acclimation, D. subobscura: 0.58 & 0.20°C/°C
225
acclimation).
226
As our dataset does not include enough species to perform a robust analysis of patterns across
227
acclimation temperatures or the climatic variables representing the origin of the species, we
228
collated our data with estimates of thermal preference from published literature (Table S1). We
229
divided this analyses in adult thermal preference or juvenile thermal preference. Here, juvenile
230
thermal preference was defined as egg-laying/placement (admittedly determined by adult
231
females) and larval thermal preferences. Further subdivision was deemed redundant due to low
232
numbers of records. Analyses showed that both adult and juvenile thermal preference was not
233
significantly related to annual mean temperature of the habitat of origin of the species (t(36) =
234
0.91, p = 0.57 & t(21) = 1.48, p = 0.15, respectively, Figure 3). Excluding adult D. birchii from the
235
dataset led to a significant relationship (t(33) = 2.74, p = 0.01, r = 0.43).
236
Finally, in order to determine if behavior can partly explain the apparent lack of variation in
237
thermal optima across Drosophila species, we correlated our estimates of thermal preference to 11
238
existing thermal optima estimates from the same populations and acclimation regimes (MacLean
239
et al. 2019). Analyses showed a relatively strong correlation between the two temperatures (r =
240
0.61, p < 0.01, Figure 4) but with optimal temperature always higher than preferred temperature.
241
242
4. Discussion
243
A deeper understanding of the role of behavioral thermoregulation among species and between
244
life stages may complement information from traditional measures of thermal physiology and
245
shed light on the associations between the climate of a species distribution and its thermal
246
physiology (e.g. thermal tolerance). Behavioral responses may alleviate some of the direct
247
environmental influence on physiology by allowing organisms to avoid variable and periodically
248
stressful body temperatures (Huey et al., 2003). Organisms (or life-stages) that are good at
249
thermoregulation may distribute across a narrow range of favorable temperatures potentially
250
explaining cases where clear adaptive patterns between ambient environmental conditions and
251
organismal thermal biology, specifically thermal optima, might not exist (Stratman & Markow
252
1998; Huey, Hertz & Sinervo 2003; MacLean et al. 2019). Despite this, little is known about
253
how animals distribute across a thermally variable microhabitats (Fogleman 1979; Dillon et al.
254
2009). Here we investigated thermal preferences of adults and larvae, as well as preference for
255
adult egg-laying across a thermal gradient in four species of Drosophila representing different
256
habitat types (Kellermann et al. 2012b) to provide a better understanding of the role of
257
behavioral thermoregulation in Drosophilids. Our data produced high resolution thermal
258
distributions allowing us to estimate mean preferred temperatures and aspects of the curvature of
259
these thermal preference distributions. The use of temperature gradients to estimate thermal
12
260
preference in insects is not a new idea (Deal 1941) but this method is not widely used (but see
261
Dillon et al., 2009; Coggan et al., 2011), in part, because measurement of preferred temperature
262
is complicated and time consuming.
263
Differences in temperature preference across life-stages may be due to evolved differences
264
driven by variation in mobility or microhabitat (Kingsolver et al., 2011; Potter et al., 2011). For
265
adult egg-laying and larvae we found the least amount of variation, while adults, the most mobile
266
life-stage, exhibited the greatest amount of variation across species (Figure 2). The adult life-
267
stage is exposed to the largest variation in environmental (i.e. air) temperature and also have the
268
greatest potential for behavioral thermoregulation due to their high mobility (Dillon et al. 2009).
269
Laid eggs and larvae are restricted to food sources and have less opportunity to behaviorally
270
avoid the likely exposure to stressful thermal environments (Feder & Krebs 1998; Marais &
271
Chown 2008). We found adults produced the broadest distributions and largest species
272
differences in temperature preference, while larvae and adult egg-laying showed preference for a
273
narrower range of temperatures (Figure 2, Figure S3). This similarity in thermal preference
274
across species could be evidence of common physiological temperature dependency. Measures
275
of fitness, including egg-to-adult survival, suggest that the thermal optima for fitness is highly
276
conserved across Drosophila species. Measures of fitness, including egg-to-adult survival,
277
suggest that the thermal optima for fitness is highly conserved across Drosophila species
278
(MacLean et al., 2019) which may explain why adult flies lay their eggs within a narrow span of
279
temperatures (Figure 2) regardless of evolutionary history.
280
Notably, adults of the tropical D. birchii yielded a surprisingly low Tpref estimate. However, we
281
do not believe this is not reflective of a true preference but rather the unidirectional nature of
282
thermal gradients resulting in D. birchii adults wandering into the cold end of the gradient where 13
283
they were cold immobilized and thus could not get out (Dillon et al., 2009). Drosophila birchii
284
reared at 19°C has a CTmin of approximately 7.6°C whereas D. melanogaster, D. subobscura, and
285
D. mojavensis have CTmin of 5.0°C, -1.6°C, and 6.0°C respectively (MacLean et al., 2019). This
286
could suggest that D. birchii show increased thermal sensitivity or reduced sensing ability to cold
287
temperature compared to the other three species, which would make sense from an evolutionary
288
standpoint given that D. birchii is adapted to warm and humid tropical environments and rarely
289
experience low temperatures in its natural habitat. In contrast, e.g. D. subobscura, which is
290
adapted to temperate environments and has a much lower CTmin (MacLean et al., 2019), was
291
observed to maintain mobility across the whole gradient.
292
No clear pattern in thermal preference across species of different ecotypes were found in our
293
study. Four species is insufficient to generate robust correlations, so we assembled information
294
from the literature and analyzed all data together (Table S1, Figure 3). The juvenile life stages
295
(egg-laying and larvae) show no relationship with annual mean temperature of origin was found
296
for thermal preference. Thus, including these published data in our analysis supports our result
297
that a conserved preferred temperature exists for these stages across species. We note that our
298
data accounts for the vast majority of studies investigating egg-laying and larvae preference.
299
Excluding our potentially biased results for adult D. birchii, a weak relationship between thermal
300
preference and annual mean temperature of origin of each species was detected. However, there
301
was no strong evidence for thermal adaptation in preferred temperatures and, compared to the
302
observed difference in mean temperatures of origin across species investigated, preference
303
temperature seems rather evolutionary constrained. This is also supported by recent estimates of
304
narrow sense heritability of thermal preference in D. subobscura, are as low as h2 = 0.066,
305
suggesting limited evolutionary potential for thermal adaptation in preferred temperature 14
306
(Castañeda et al. 2019). Further, the relatively broad distribution of adult preferred temperatures
307
does not point to adults relying more on behavioral thermoregulation compared to the other life
308
stages. While, we did not find that preferred temperature strongly indicated evolutionary history.
309
Even if we provide evidence of differences among species and life stages, our data does not
310
support the prediction that adults (as the most mobile life-stage) show the greatest potential for
311
behavioral thermoregulation. Instead, we suggest that less mobile stages may benefit
312
physiologically from similar optimal temperatures achieved through narrow selection of
313
preferred temperatures.
314
We also evaluated the ability to shift preference through developmental acclimation within a
315
species. Acclimation to cold environments has been shown to lead to a compensatory increase in
316
preferred temperature (Fogleman, 1979) but more recent work on developmental acclimation and
317
preference has yielded somewhat mixed results (Dillon et al., 2009). We find little evidence of
318
such compensation in the thermal preference among the species we investigated. Rather, we see
319
a relatively strong increase in preferred egg-laying temperature and no response in preferred
320
adult temperature to thermal acclimation (Table 2). When we estimate the effect of acclimation
321
in egg laying preference we found the effect was approximately 0.3 °C/°C acclimation. This is
322
notable as it is comparable to the acclimation effects of critical lower thermal limits, which is a
323
highly labile trait (Sørensen et al., 2016; Schou et al., 2017). The direction of the acclimation
324
response in this study corresponds to an adaptive interpretation of the role of thermal preference
325
in adult egg-laying preference. However, when we consider the results from the literature there is
326
no general support for increased thermal preference with increasing acclimation temperature
327
(Table S2). Scarcity of comparable data across species presently prevent general conclusions
328
regarding the acclimation response in preferred temperatures in response to e.g. developmental 15
329
or adult acclimation. However, this comprehensive study across life stages and species suggest
330
that developmental acclimation has a large effect on adult egg placement but not on the adult
331
preference. In contrast, evolved differences between species seem largely to be affecting adult
332
preference.
333
Theoretically, organism should prefer temperatures at which they optimally perform. Our
334
analysis (Figure 4) suggests that thermal optima is always higher than thermal preference. This
335
could be due to one of two reasons; either the used estimates of thermal optima we not of
336
adequate resolution or we lacking crucial traits (see MacLean et al. 2019), or the flies are in fact
337
selecting lower than optimal temperatures as part of a bet-hedging strategy (Martin & Huey
338
2008). Due to the typical (left-skewed) shape of a thermal performance curve, the fitness
339
consequence of higher than optimal temperature strongly outweighs the consequences of lower
340
than optimal temperatures. Following this, preferring lower than optimal temperatures might
341
constitute an evolutionary stable strategy.
342
343
5. Conclusion:
344
Traditional measures of thermal tolerance and performance and the acclimation capacity thereof
345
are important for the assessment of animal fitness across thermal environments. While measures
346
of physiological capacity can be interpreted to reflect thermal adaptation when correlated with
347
environmental temperatures (Sunday, Bates & Dulvy 2011; Kellermann et al. 2012a; Overgaard,
348
Kearney & Hoffmann 2014), we should take care when applying these measures to ecological
349
and evolutionary predictions, as this thermal adaptation is not reflected in behaviorally preferred
350
temperatures. Behavioral changes are very fast and likely important in spatially and temporally 16
351
heterogeneous thermal environments. Further, if individuals are able to exploit microclimates it
352
is possible that we overestimate the negative consequences of climate change and extreme
353
climate events because ectotherms typically never experience temperatures reported. Including
354
thermal physiological capacity when assessing components of fitness may have little ecological
355
relevance because individuals may rarely experience these conditions in nature (Sunday, Bates &
356
Dulvy 2011; Kellermann et al. 2012a; Overgaard, Kearney & Hoffmann 2014). The optimal
357
temperature for an organism is complex, comprised of the optimization of many physiological
358
systems and may be evolutionary or ecologically constrained (Angilletta 2009; Schulte, Healy &
359
Fangue 2011; MacLean et al. 2019) and behavior may enable organisms to achieve these body
360
temperatures regardless of ambient air temperature in a given environment (Huey, Hertz &
361
Sinervo 2003). Measuring behavioral thermoregulation and integrating these measures with
362
measures of thermal tolerance and optimal performance from the laboratory offers a context for a
363
broader adaptive understanding of the impact of temperature on shaping past, current and future
364
distribution of species.
365
366
Acknowledgements
367
We thank Mads Fristrup Schou and Sarah E. Diamond for helpful conversations regarding data
368
analysis and Kirsten Kromand, Trine Bech Søgaard and Annemarie Højmark for excellent help
369
in the fly labs. This work was supported by grants from the Villum Foundation (JO), Aarhus
370
University Research Foundation (JGS), and the Danish Council for Independent Research (TNK
371
and JO, respectively).
372
17
373
Supplementary material: Table S1: Table of previous studies on thermal preference in
374
Drosophila species, Table S2: Description of fly stocks and origins, Figure S1: Graphical
375
depiction of experimental set-up, Figure S2: Correlation between estimated preferred
376
temperatures, Figure S3: Distribution of adult and egg-laying thermal preference following
377
developmental acclimation, Figure S4: Correlation of trait values across all temperatures.
378
379
Author Contributions: Conceptualization, HJM, JO, JGS. and TNK.; Methodology, HJM, JO,
380
JGS; Formal Analysis, HJM and JGS.; Investigation, HJM, CL, LH, EO, JO, and JGS.;
381
Resources, JO, TNK, and JGS.; Data Curation, CL, LH, EO, HJM and JGS.; Writing – Original
382
Draft Preparation, HJM and JGS.; Writing – Review & Editing, TNK, JO, CL, LH, EO;
383
Visualization, JGS and HJM.; Funding Acquisition, JO, JGS, and TNK.
384 385
Conflicts of Interest: The authors declare no conflict of interest.
386
387
References
388 389 390 391 392 393 394 395 396 397 398
Angilletta, M.J. (2009) Thermal adaptation: a theoretical and empirical synthesis. Oxford University Press. Buckley, L.B., Ehrenberger, J.C. & Angilletta, M.J. (2015) Thermoregulatory behavior limits local adaptation of thermal niches and confers sensitivity to climate change. Functional Ecology, n/an/a. Castañeda, L.E., Romero-Soriano, V., Mesas, A., Roff, D.A. & Santos, M. (2019) Evolutionary potential of thermal preference and heat tolerance in Drosophila subobscura. Journal of Evolutionary Biology, 0. Deal, J. (1941) The temperature preferendum of certain insects. The Journal of Animal Ecology, 323-356. Dillon, M.E., Wang, G., Garrity, P.A. & Huey, R.B. (2009) Thermal preference in Drosophila. Journal of Thermal Biology, 34, 109-119.
18
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
Feder, M.E. & Krebs, R.A. (1998) Natural and genetic engineering of the heat-shock protein Hsp70 in Drosophila melanogaster: Consequences for thermotolerance. American Zoologist, 38, 503-517. Fogleman, J. (1979) Oviposition site preference for substrate temperature inDrosophila melanogaster. Behavior Genetics, 9, 407-412. Hertz, P.E., Huey, R.B. & Stevenson, R. (1993) Evaluating temperature regulation by field-active ectotherms: the fallacy of the inappropriate question. The American naturalist, 142, 796-818. Higgins, J.K., MacLean, H.J., Buckley, L.B. & Kingsolver, J.G. (2014) Geographic differences and microevolutionary changes in thermal sensitivity of butterfly larvae in response to climate. Functional Ecology, 28, 982-989. Hoffmann, A.A., Sørensen, J.G. & Loeschcke, V. (2003) Adaptation of Drosophila to temperature extremes: bringing together quantitative and molecular approaches. Journal of Thermal Biology, 28, 175-216. Huey, R.B., Hertz, P.E. & Sinervo, B. (2003) Behavioral drive versus behavioral inertia in evolution: a null model approach. The American naturalist, 161, 357-366. Kellermann, V., Loeschcke, V., Hoffmann, A.A., Kristensen, T.N., Fløjgaard, C., David, J.R., Svenning, J.C. & Overgaard, J. (2012a) Phylogenetic Constraints in Key Functional Traits Behind Species' Climate Niches: Patterns of Desiccation and Cold Resistance Across 95 Drisiohila Species. Evolution, 66, 3377-3389. Kellermann, V., Overgaard, J., Hoffmann, A.A., Fløjgaard, C., Svenning, J.-C. & Loeschcke, V. (2012b) Upper thermal limits of Drosophila are linked to species distributions and strongly constrained phylogenetically. Proceedings of the National Academy of Sciences, 109, 16228-16233. Kingsolver, J.G., Woods, H.A., Buckley, L.B., Potter, K.A., MacLean, H.J. & Higgins, J.K. (2011) Complex life cycles and the responses of insects to climate change. Integrative and Comparative Biology, icr015. MacLean, H.J., Higgins, J.K., Buckley, L.B. & Kingsolver, J.G. (2016) Geographic divergence in upper thermal limits across insect life stages: does behavior matter? Oecologia, 181, 107-114. MacLean, H.J., Sørensen, J.G., Kristensen, T.N., Loeschcke, V., Beedholm, K., Kellermann, V. & Overgaard, J. (2019) Evolution and plasticity of thermal performance: An analysis of variation in thermal tolerance and fitness in 22 Drosophila species. Philosophical Transactions of the Royal Society B: Biological Sciences. Marais, E. & Chown, S.L. (2008) Beneficial acclimation and the Bogert effect. Ecology Letters, 11, 10271036. Martin, T.L. & Huey, R.B. (2008) Why “suboptimal” is optimal: Jensen’s inequality and ectotherm thermal preferences. The American naturalist, 171, E102-E118. Overgaard, J., Kearney, M.R. & Hoffmann, A.A. (2014) Sensitivity to thermal extremes in A ustralian D rosophila implies similar impacts of climate change on the distribution of widespread and tropical species. Global change biology, 20, 1738-1750. Potter, K.A., Woods, H.A. & Pincebourde, S. (2013) Microclimatic challenges in global change biology. Global change biology, 19, 2932-2939. Rajpurohit, S. & Schmidt, P.S. (2016) Measuring thermal behavior in smaller insects: A case study in Drosophila melanogaster demonstrates effects of sex, geographic origin, and rearing temperature on adult behavior. Fly, 10, 149-161. Schou, M.F., Loeschcke, V. & Kristensen, T.N. (2015) Strong costs and benefits of winter acclimatization in Drosophila melanogaster. PloS one, 10, e0130307. Schulte, P.M., Healy, T.M. & Fangue, N.A. (2011) Thermal performance curves, phenotypic plasticity, and the time scales of temperature exposure. Integrative and Comparative Biology, 51, 691-702.
19
445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460
Sørensen, J.G., Kristensen, T.N., Loeschcke, V. & Schou, M.F. (2015) No trade-off between high and low temperature tolerance in a winter acclimatized Danish Drosophila subobscura population. Journal of Insect Physiology, 77, 9-14. Sørensen, J.G., White, C.R., Duffy, G.A. & Chown, S.L. (2018) A widespread thermodynamic effect, but maintenance of biological rates through space across life's major domains. Proceedings of the Royal Society B: Biological Sciences, 285, 20181775. Stratman, R. & Markow, T. (1998) Resistance to thermal stress in desert Drosophila. Functional Ecology, 12, 965-970. Sunday, J.M., Bates, A.E. & Dulvy, N.K. (2011) Global analysis of thermal tolerance and latitude in ectotherms. Proceedings of the Royal Society B: Biological Sciences, 278, 1823-1830. Sunday, J.M., Bates, A.E., Kearney, M.R., Colwell, R.K., Dulvy, N.K., Longino, J.T. & Huey, R.B. (2014) Thermal-safety margins and the necessity of thermoregulatory behavior across latitude and elevation. Proceedings of the National Academy of Sciences, 111, 5610-5615. Wood, S.N. (2001) mgcv: GAMs and generalized ridge regression for R. R news, 1, 20-25. Woods, H.A., Dillon, M.E. & Pincebourde, S. (2015) The roles of microclimatic diversity and of behavior in mediating the responses of ectotherms to climate change. Journal of Thermal Biology, 54, 86-97.
461 462
20
463
Table 1. Characteristics of thermal preference in four species of Drosophila and in three life stages. Acclimation temperatures were
464
applied for two life stages (adult preference and egg-laying preference). The traits estimated are mean (± sem). Tpref was estimated as
465
the average temperature of animals. For the remaining traits, we fit a gam model to the proportional distribution data for each species,
466
life stage, and acclimation combination to predict frequency across the range of temperatures used. We then estimated a cumulative
467
density to estimate the temperature at which 10% (P10) and 90% (P90) of the animals occur. Tbreadth was calculated as the difference
468
between P90 and P10.
469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
Species
Stage
Accl. Temp (°C)
n
Tpref
Tbreadth
P10
P90
D. birchii -
Adult Adult Adult Egg-laying Egg-laying Egg-laying Larvae
19 23 27 19 23 27 19
12 12 12 9 12 5 8
17.4 (0.4) 16.8 (0.8) 16.7 (0.4) 23.2 (0.6) 25.1 (0.3) 26.9 (0.5) 22.9 (0.3)
16.0 (0.5) 15.7 (0.8) 15.3 (0.7) 11.6 (0.8) 11.0 (0.6) 10.2 (0.2) 8.5 (0.2)
10.9 (0.4) 10.6 (0.4) 10.7 (0.3) 17.1 (0.6) 19.1 (0.6) 20.9 (0.4) 19.5 (0.1)
27.0 (0.3) 26.3 (0.9) 26.0 (0.7) 28.7 (0.6) 30.1 (0.3) 31.2 (0.3) 28.0 (0.3)
D. melanogaster -
Adult Adult Adult Egg-laying Egg-laying Egg-laying Larvae
19 23 27 19 23 27 19
12 12 12 12 12 12 7
24.3 (0.5) 25.0 (0.4) 24.7 (0.4) 24.9 (0.3) 25.7 (0.3) 26.1 (0.4) 25.0 (0.5)
12.7 (0.6) 11.8 (0.4) 12.5 (0.7) 8.9 (0.4) 9.2 (0.3) 9.3 (0.3) 8.2 (0.6)
17.3 (0.8) 18.6 (0.5) 18.0 (0.6) 20.4 (0.3) 21.0 (0.3) 21.3 (0.4) 20.9 (0.6)
30.0 (0.3) 30.4 (0.3) 30.5 (0.3) 29.3 (0.3) 30.1 (0.3) 30.5 (0.3) 29.1 (0.1)
21
488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503
D. mojavensis -
Adult Adult Adult Egg-laying Egg-laying Egg-laying Larvae
19 23 27 19 23 27 19
14 12 12 12 12 12 9
20.8 (0.4) 22.0 (0.5) 23.0 (0.5) 24.4 (0.4) 25.5 (0.5) 26.4 (0.5) 22.9 (0.1)
15.9 (0.8) 14.8 (0.5) 15.2 (0.6) 11.5 (0.7) 9.2 (0.5) 9.7 (0.5) 7.6 (0.3)
12.8 (0.5) 14.4 (0.4) 15.2 (0.6) 18.6 (0.7) 20.7 (0.6) 21.0 (0.4) 19.6 (0.1)
28.7 (0.5) 29.2 (0.5) 30.4 (0.3) 30.1 (0.2) 29.9 (0.4) 30.7 (0.4) 27.2 (0.2)
D. subobscura -
Adult Adult Adult Egg-laying Egg-laying Egg-laying Larvae
15 19 23 15 19 23 19
12 12 12 8 11 6 8
19.0 (0.7) 18.3 (0.4) 18.8 (0.5) 20.9 (0.6) 23.8 (0.5) 24.0 (0.4) 23.3 (0.2)
16.3 (0.8) 16.0 (0.4) 17.1 (0.5) 12.9 (0.8) 10.5 (1.0) 10.0 (0.4) 8.0 (0.4)
11.4 (0.4) 11.2 (0.2) 11.0 (0.3) 14.4 (0.7) 18.3 (0.9) 18.8 (0.5) 19.7 (0.1)
27.7 (0.8) 27.1 (0.5) 28.1 (0.6) 27.2 (0.6) 28.8 (0.5) 28.8 (0.2) 27.7 (0.3)
504 505 506 507
22
508
Table 2. Summary of the F-values from linear models for Tpref, Tbreadth, P10, and P90 as a function of species as a fixed categorical
509
variable and acclimation temperature (Accl. (°C)) as a continuous variable. * indicate statistical significance (* p < 0.05, ** p < 0.01.
510
*** p < 0.001).
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
Stage
Source (df)
Tpref
Tbreadth
P10
P90
Adult
Species (3,138) Accl. (°C) (1,138)
134.7*** 1.6
24.6*** 0.3
141.0*** 3.6
32.8*** 1.2
Adult Egg-laying
Larvae
Species*Accl (3,138)
3.6*
0.6
Species (3,115) Accl. (°C) (1,115)
20.2*** 46.7***
7.5*** 8.1**
Species* Accl (3,115)
2.8*
2.5
3.6*
1.7
Species (3,28)
11.1***
1.1
5.7**
9.7***
528 529 530 531
23
3.6*
22.4*** 36.8***
2.1
14.2*** 22.2***
532 533 534 535 536 537 538
Figure 1: Example of a proportional distribution plot used to assess thermal preference. Points reflect the mean (± sem) proportion of individuals in a specific temperature bin. The temperature at which the highest number occurred is Tpref max. The curve is predicted using a generalized linear model and used to estimate the temperature of inflection (Tpref ave), the temperature at which 10% of the eggs occur (P10), the temperature at which 90% of the eggs occur (P90), and the difference between these two points (Tbreadth).
24
539 540 541 542 543 544 545
Figure 2: Proportionally distribution across a thermal gradient of adult females (left), deposited eggs of same females (middle), and of larvae (right) for four different species all acclimated at 19 °C. A clear thermal preference is observable in most cases. Note that the thermal gradient used for larval preference only covers a part of the thermal range examined for adult and egg-laying preference (as indicated by vertical dashed grey lines in the plots for adult and egg-laying preference).
546 547 548
25
549 550 551 552 553 554 555 556 557 558 559 560
Figure 3. Correlation of thermal preference versus annual mean temperature (AMT) as a measure of environmental temperature. Plots represent adult (Panel A) or juvenile (position of egg-laying or larvae) thermal preference (Panel B). AMT was extracted from for each species. Circles represent previous studies (see Table S2) and triangles represent the data from the current study. Dashed lines represent correlation across all data. Solid line (Panel A) represent correlations on all data except D. birchii (see discussion on the rationale for this). Neither juvenile nor adult preference were significantly related to AMT (for adults the exclusion of D. birchii changed this to a significant positive relationship). We do note that the data are limited and compares (for juveniles) different metrics. However, including published data both the number of investigated species and the environmental range covered is strongly increased – especially for the juvenile stages.
561
26
28 26 24 18
Topt (°C)
22 20
D. birchii D. melanogaster D. mojavensis D. subobscura
Adult Egg-laying
18
20
22
24
26
28
Tpref (°C)
562 563 564 565 566 567 568
Figure 4. Correlation of thermal preference versus temperature of optimal performance (Topt) as estimated in MacLean 2019. Plots represent adult thermal preference correlated to composite fitness optima and position of egg-laying thermal preference correlated to optimal temperature for egg-laying. We observe a correlation of r = 0.61. These data show that the preferred temperature is always lower than the optimal temperature and suggest a potential bet-hedging strategy may be employed
569 570
27
J thermal biology HIGHLIGHTS •
• •
• •
We take on the role of thermoregulatory behavior in Drosophila in a new and comprehensive way. Specifically, we apply the same methodology to four species (twice as many as has ever been published in one study before) across life-stages and following acclimation regimes. We did this to quantify evolved difference and potential plasticity in preferred temperature. We then took this two steps further. First, we extracted all available data on thermal preference in Drosophila and regressed it against annual mean temperature in species origin. Then, we related preference temperature to estimates of optimal temperature for the same populations. Because we find surprisingly little interspecific variation in temperature preference suggests behavior is an important factor in thermal adaptation or lack thereof of thermal optima. Many conceptualize thermal responses and expectations of thermal adaptation in the context of an insect’s mean thermal environment (ie air temperature) but we show organisms
S0306-4565(19)30345-6
DOI:
https://doi.org/10.1016/j.jtherbio.2019.102428
Reference:
TB 102428
To appear in:
Journal of Thermal Biology
Received Date: 28 June 2019 Revised Date:
24 September 2019
Accepted Date: 4 October 2019
Please cite this article as: MacLean, H.J., Overgaard, J., Kristensen, T.N., Lyster, C., Hessner, L., Olsvig, E., Sørensen, J.G., Temperature preference across life stages and acclimation temperatures investigated in four species of Drosophila, Journal of Thermal Biology (2019), doi: https:// doi.org/10.1016/j.jtherbio.2019.102428. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.
1
Temperature preference across life stages and acclimation temperatures investigated in
2
four species of Drosophila
3
4 5
Heidi J MacLean1*, Johannes Overgaard1, Torsten N Kristensen2, Catrine Lyster1, Leander
6
Hessner1, Esajas Olsvig1, Jesper G Sørensen1
7
8
1
Department for Bioscience, Aarhus University, Ny Munkegade 116, 8000 Aarhus C, Denmark
9
2
Department of Chemistry and Bioscience, Aalborg University, Fredrik Bajers Vej 7H, 9220
10
Aalborg E, Denmark
11
12
*Corresponding author: [email protected]
13
1
14
Abstract: Ectotherms can use microclimatic variation and behavioral thermoregulation to cope
15
with unfavorable environmental temperatures. However, relatively little is known about how and
16
if thermoregulatory behavior is used across life-stages in small ectothermic insects. Here we
17
investigate differences between three specialized Drosophila species from temperate, tropical or
18
desert habitats and one cosmopolitan species by estimating the preferred temperature (Tpref) and
19
the breadth (Tbreadth) of the distribution of adults, adult egg-laying, and larvae in thermal
20
gradients. We also assess the plasticity of thermal preference following developmental
21
acclimation to three constant temperatures. For egg-laying and larvae we observe significant
22
species differences in preferred temperature but this is not predicted by thermal ecology of the
23
species. We corroborated this with previous studies of other Drosophila species and found that
24
Tpref for egg laying and larvae have no relationship with annual mean temperature of the species
25
natural habitat. While adults have the greatest mobility, they show the greater variation in
26
preference compared to juveniles contradicting common assumptions. We found evidence of
27
developmental thermal acclimation for adult egg-laying preferred temperature, Tpref increasing
28
with acclimation temperature, and in the breadth of the temperature preference distributions,
29
Tbreadth decreasing with increasing acclimation temperature. Together, these data provide a high
30
resolution and comprehensive look at temperature preferences across life stages and in response
31
to acclimation. Results suggest, that thermal preference, particularly in the early life stages, is
32
relatively conserved among species and unrelated to temperature at species origin. Measuring
33
thermal preference, in addition to thermal performance, is essential for understanding how
34
species have adapted/will adapt to their thermal environment.
35 36
Key words: thermoregulatory behavior, climate change, thermal preference, thermal optima 2
37 38
1. Introduction:
39
The thermal biology of ectotherms is often investigated by assessing extreme thermal tolerance
40
measures or by measuring thermal performance of various traits across the permissive
41
temperature range (Angilletta 2009; Sørensen et al. 2018). Using thermal tolerance as a proxy for
42
thermal adaptation enable detection of an association between species and the thermal
43
characteristics of their habitats (Magnuson et al., 1979; Kellermann et al., 2009, 2012; Sunday et
44
al., 2011; Overgaard et al., 2014; Andersen et al., 2015). Further, characterization of thermal
45
performance curves can be used to inform spatial distribution models (SDMs) and parameterize
46
modelling approaches of population growth rates (Kearney & Porter, 2009; Richardson et al.,
47
2011; Buckley & Kingsolver, 2012). However, the assessments of tolerance and performance are
48
typically studied under conditions where behavioral responses are constrained and therefore fail
49
to capture how insects navigate temperature and temperature variation through behavioral choice
50
(but see Woods et al., 2015).
51
Behavior can allow organisms to exploit microclimatic variation at a small spatial scale and
52
potentially avoid unfavorable environmental temperatures (Huey, Hertz & Sinervo 2003; Potter,
53
Woods & Pincebourde 2013; Sunday et al. 2014; Buckley, Ehrenberger & Angilletta 2015).
54
Thus, behavioral thermoregulation may relieve some organisms from selective pressure on
55
thermal physiological traits and explain cases where no clear adaptive variation exists among
56
species or populations despite substantial climate/seasonality differences (see Hoffmann,
57
Sørensen & Loeschcke 2003). In a recent study, we found strikingly similar thermal optima for a
58
component of fitness based on three life-history traits (fecundity, developmental time and egg-to-
59
adult viability) among 22 species of Drosophila despite representing markedly different ecotypes
3
60
(MacLean et al. 2019). Here we assess temperature preference across life-stages of four
61
Drosophila species to determine if behavior can partly explain the apparent lack of variation in
62
thermal optima across Drosophila species.
63
The knowledge of behavioral thermoregulation is well-established in large ectotherms such as
64
lizards that can use behavioral thermoregulation to achieve body temperatures that deviate
65
considerably from ambient air temperatures (Hertz, Huey & Stevenson 1993). Small ectotherms
66
also rely on behavioral thermoregulation (Hoffmann, Sørensen & Loeschcke 2003; Dillon et al.
67
2009). For example, Drosophila subobscura and D. melanogaster overwintering in compost
68
heaps have the potential to avoid extreme temperatures in their natural habitat (Schou, Loeschcke
69
& Kristensen 2015; Sørensen et al. 2015). Insects with complex life-cycles may face different
70
selection pressures because of e.g. life-stage-specific thermal sensitivity, difference in
71
microclimates available, or different mobility of different life stages (Kingsolver et al. 2011;
72
Potter, Woods & Pincebourde 2013; Woods, Dillon & Pincebourde 2015). For example, local
73
adaptation in thermal performance have been reported in the larvae (Higgins et al. 2014) but not
74
adults of a highly-thermoregulating butterfly species (MacLean et al. 2016). Importantly, the
75
variation in thermoregulatory behavior (thermal preferences) across life stage and the plasticity
76
in preferences resulting from developmental acclimation are poorly investigated (but see
77
Rajpurohit & Schmidt 2016).
78
Existing knowledge on thermal preference and thermal preference plasticity in Drosophila
79
typically stem from studies using one or two species, and comparisons among studies are
80
challenged by different treatments (e.g. acclimation) and methodology (see Table S1). To
81
address this lack of knowledge, we performed a comprehensive study looking at four species
82
with distinct thermal ecology and assessed among life stages using standardized, comparable 4
83
methods. We compared the thermoregulatory behavior of adult and larvae Drosophila; D.
84
melanogaster, D. birchii, D. subobscura, and D. mojavensis representing cosmopolitan, tropical,
85
temperate, and desert habitats, respectively. We hypothesize that thermal preference should be
86
influenced by both habitat type (evolutionary history) and short-term thermal experience
87
(ecological history). Further, we expect that adult behavior (both egg-placement and adult
88
choice) should be more narrowly distributed across favorable temperatures due to their increased
89
mobility. By measuring the distributions of thermal preference (Figure 1) in different life stages
90
we examined if different Drosophila species from different environments show adaptive
91
differences in thermal preference. Specifically, that species from warmer environments prefer
92
warmer temperatures. We also investigated the effect of developmental acclimation at three
93
temperatures in adult egg-laying and adult thermal preference to investigate the potential
94
adaptive role of phenotypic plasticity in thermal preference. We expect that increased
95
developmental acclimation temperature would translate into increased thermal preference. We
96
then analyzed the patterns of preference among life stages within and among species to test the
97
prediction that mobile adults might rely more on behavior than less mobile juveniles.
98
99
2. Materials and Methods
100
2.1 Experimental animals
101
To quantify species specific differences in thermal preference (Tpref) we used populations of four
102
Drosophila species reared under common conditions (19 °C, 12/12 L:D). The four species
103
represent disparate habitat specializations: Drosophila melanogaster (cosmopolitan), D. birchii
104
(tropical), D, subobscura (cold-hardy, temperate), and D. mojavensis (xeric, desert) (summarized 5
105
in Table S2). All four species were bred on standard Drosophila medium. Each species was
106
reared in two bottles with 50 mL of fly food at 19°C and controlled larval density (unless
107
otherwise noted).
108
109
2.2 Adult and egg-laying thermal preference
110
Prior to experimentation, newly emerged adults of both sexes were transferred to fresh fly food
111
and fed yeast until they were sexually mature (1-4 days depending on the species). During the
112
thermal preference experiments the flies were placed in a linear thermal gradient with
113
temperature spanning from 9 to 34°C. The thermal gradient was created by a large aluminum
114
block (156 cm L x 60 cm W, Figure S1A) which was cooled/heated using a circulating water
115
bath at either end. Eight horizontal lanes were established in parallel on the thermal gradient (6
116
cm W x 3cm H) and in each lane, we placed 13 petri dishes (diameter 5.5 cm) containing 14 mL
117
of standard fly food. Petri dishes with food were placed every 10 cm and petri dishes filled with
118
silica gel were placed in between to mitigate condensation (Figure S1A). The lanes were covered
119
with a Plexiglas lid and weighed down to prevent individuals from migrating between lanes or
120
escaping the experimental setup. Approximately 60 flies of mixed sex (sd = ± 22) were loaded
121
through holes across the gradient (at 6, 43, 77, 117 and 150 cm corresponding to temperatures of
122
approximately 10, 14, 21, 27 and 33°C of the temperature gradient). While the number of adults
123
was variable, previous work on thermal preference in Drosophila found no effect of number of
124
flies tested on the preferred temperature (Krstevska & Hoffmann, 1994). Flies were given 20
125
hours to explore and lay eggs. Hereafter, adults were anesthetized on the gradient by flushing the
126
air with CO2 to anaesthetize the flies and the distribution of adults was then registered by
6
127
counting number of adults in each 10-cm bin (representing approximately a span of 1.6°C within
128
the gradient). We observed that the application of CO2 simply slowed the activity until the flies
129
entered into a comatose state and did not result in any extraordinary activity. Subsequent to the
130
experiment, we counted the number of eggs in each petri dish to establish the distribution of egg-
131
laying preference. To ensure that the obtained data reflected the thermal “choice” on the gradient
132
rather than an effect of loading position or edge, we ran 10 replicate controls for D. melanogaster
133
(and two replicate controls for all other species) using the apparatus without applying a thermal
134
gradient. Under these conditions, we found adult presence and egg laying to be evenly
135
distributed across the entire gradient (data not shown). When the gradient was turned on, we did
136
encounter a “loading effect” for D. birchii at the cold end of the gradient (10°C) so this species
137
was subsequently loaded at 14, 21, 27°C. See discussion for further details.
138
To investigate the effect of thermal acclimation on both adult thermal preference and egg-laying
139
thermal preference we compared flies that had developed and been reared at three different
140
acclimation temperatures (15, 19, 23°C for D. subobscura; 19, 23, 27°C for the other three
141
species). In these experiments, we tested the acclimation treatments for a given species in
142
different lanes of the same round of experiments to ensure that any difference was not associated
143
with block effects.
144
145
2.3 Larval thermal preference
146
Larval thermal preference was assayed on a smaller thermal gradient established on an aluminum
147
plate (800 mm L x 200 mm W x10 mm H) with temperature controlled at both ends by
148
individual Peltier devices. The Peltier devices established a largely linear temperature gradient 7
149
over the length of the apparatus running from 17-31°C on the surface (Figure S1B). Flat
150
bottomed U-shaped aluminum lanes (800 mm L x 20 mm W x 10 mm H) placed in parallel and
151
spanning the length of the apparatus, were filled with 100 mL of standard fly food (Figure S1B).
152
The apparatus was placed in an insulated box and the experimental arena was covered by a
153
Plexiglas lid weighted down to prevent individuals from migrating across lanes. Up to eight
154
lanes were run simultaneously surrounded by empty lanes (to avoid edge effects). For each lane,
155
we transferred 200 eggs evenly spaced on the gradient (in groups of 10 eggs). Egg-laying was
156
timed so that we could collect eggs close to hatching. The eggs were allowed to hatch and larvae
157
to randomly distribute for 48 h at 19°C. The whole lane was then placed on the thermal gradient
158
for 24 h to allow larvae relocate to their preferred temperature. After this treatment, the lanes
159
were removed from the temperature gradient and the food from each lane was divided into 4 cm
160
bins (20 bins total – each representing approximately 0.7°C of gradient). The food was then
161
transferred to a vial with 7 mL additional fly food and all vials were returned to 19°C to allow
162
the larvae to develop. After 18-24 days when all larvae had developed we counted the number of
163
emerging adults in each vial. We could also use this test to verify that there was no migration
164
between lanes as only one species emerged from each lane. Recovery (survival) rates were
165
similar among species and replicates (mean % ± sem, D. melanogaster (N = 946): 76.2 ± 3.1%;
166
D. birchii (N = 374): 71.1 ± 1.5%; D. subobscura (N = 674): 67.9 ± 5.6%; D. mojavensis (N =
167
1503): 77.1 ± 4.1%, respectively). Similar to the experiments with adult flies, controls for larval
168
distribution were run without the Peltier devices turned on (i.e. no thermal gradient) to verify that
169
the observed patterns of thermal preference were a function of temperature. Under these
170
conditions, we found larval preference to be evenly distributed across the entire width of the
171
gradient (data not shown).
8
172
2.4 Analysis
173
All analyses were conducted in R version 3.3.3. To account for variation in total number of
174
adults used, eggs laid or larvae recovered among different replicates, we standardized the data by
175
calculating the proportional distribution per replicate lane. Two different approaches were
176
employed to estimate preferred temperature (Tpref). First, we selected the temperature
177
corresponding to the highest proportional value (Tpref_max). Second, for each lane we multiplied
178
the proportion of individuals in each bin by the corresponding temperature and summed these
179
values across all bins to determine a weighted average (Tpref_ave). However, as the two measures
180
were strongly correlated across all data (Pearson's product-moment correlation, t(26) = 18.2, p <
181
0.05, r = 0.96, Figure S2) we continued analyses with only the weighted average (Tpref_ave) as our
182
estimate for Tpref (see Figure 1).
183
We fitted a generalized additive model (GAM) using the mcgv package (Wood 2001) with a 0.01
184
smoothing parameter to the proportional distribution data for each species, life stage, and
185
acclimation combination. Based on this model we estimated the temperature at which 10% (P10)
186
and 90% (P90) of the animals occur. The range between these two (P90 – P10) yielded breadth
187
(Tbreadth) of the distribution (all estimates are summarized in Table 1, Figure 2).
188
Further, we assessed general relationships between the thermal preferences (at different life
189
stages) and environmental temperatures of species origin, across data from this study and from
190
the literature. To do this, we fitted linear models for Tpref as a function of annual mean
191
temperature for each species (as given in Kellermann et al. 2012a) across species and life stages,
192
respectively (Figure 3).
9
193
Finally, we correlated the thermal preferences with the thermal optima estimated in for egg-
194
laying and overall fitness in MacLean (2019). Specifically, we correlated egg-laying thermal
195
optima with egg-laying preference and overall fitness thermal optima with adult thermal
196
preference for all developmental acclimation regimes.
197
3. Results
198
We examined if Drosophila exhibit different preferred temperatures (Tpref) as a function of
199
species, acclimation or life stage life-stage. We found small, but significant differences between
200
species in Tpref for adult egg-laying and for larval preference (Table 2) but these differences are
201
not ordered by thermal habitat/ecotype of each species. For adults, we find larger differences in
202
Tpref between the four species (Figure 2, Table 2). We found a large difference in Tbreadth between
203
the four species in the adult and the adult egg-laying choice, but not in larvae choice (Figure 2,
204
Table 2). Analysis of both P10 and P90 show species effects for larvae, adult and adult egg-laying
205
choice (Table 2, Figure S3).
206
To determine if developmental thermal acclimation (plasticity) alters the preferred temperature
207
of adults or adult egg-laying we investigated the distribution of adult female flies and their egg-
208
laying when reared at 19, 23, and 27°C (or at 15, 19 and 23°C for D. subobscura).
209
Developmental acclimation did significantly increase Tpref of adult egg-laying preference, while
210
no significance of developmental temperature was found for adult preference (Figure S3, Table
211
2). The strongest acclimation response in egg-laying Tpref (shown as increased Tpref °C/°C
212
acclimation) was found for D. birchii (where Tpref increased by 0.46°C/°C acclimation) followed
213
by D. subobscura (0.39°C/°C acclimation), D. mojavensis (0.25°C/°C acclimation) and the
214
weakest response was observed in D. melanogaster (0.15°C/°C acclimation). Further, there was
10
215
a significant decrease in egg-laying Tbreadth as a function of increasing acclimation temperature
216
(Table 2, Figure S3). When we consider P10 and P90 individually we see evidence of the curve
217
shifting towards higher temperatures in egg-laying distribution with increasing acclimation
218
temperature. The response to acclimation for adult egg-laying P10 and P90 were on average ~0.36
219
and ~0.18°C/°C acclimation, respectively, however it should be noted that variation among
220
species was marked. This result suggest that the upper marginal bound of the distribution is
221
potentially less labile than the lower bound and the body of the distribution is centered around
222
Tpref. All species except D. melanogaster showed a stronger acclimation response for P10 as
223
compared to P90 (D. birchii: 0.48 & 0.31°C/°C acclimation, D. melanogaster: 0.11 & 0.15°C/°C
224
acclimation, D. mojavensis: 0.30 & 0.07°C/°C acclimation, D. subobscura: 0.58 & 0.20°C/°C
225
acclimation).
226
As our dataset does not include enough species to perform a robust analysis of patterns across
227
acclimation temperatures or the climatic variables representing the origin of the species, we
228
collated our data with estimates of thermal preference from published literature (Table S1). We
229
divided this analyses in adult thermal preference or juvenile thermal preference. Here, juvenile
230
thermal preference was defined as egg-laying/placement (admittedly determined by adult
231
females) and larval thermal preferences. Further subdivision was deemed redundant due to low
232
numbers of records. Analyses showed that both adult and juvenile thermal preference was not
233
significantly related to annual mean temperature of the habitat of origin of the species (t(36) =
234
0.91, p = 0.57 & t(21) = 1.48, p = 0.15, respectively, Figure 3). Excluding adult D. birchii from the
235
dataset led to a significant relationship (t(33) = 2.74, p = 0.01, r = 0.43).
236
Finally, in order to determine if behavior can partly explain the apparent lack of variation in
237
thermal optima across Drosophila species, we correlated our estimates of thermal preference to 11
238
existing thermal optima estimates from the same populations and acclimation regimes (MacLean
239
et al. 2019). Analyses showed a relatively strong correlation between the two temperatures (r =
240
0.61, p < 0.01, Figure 4) but with optimal temperature always higher than preferred temperature.
241
242
4. Discussion
243
A deeper understanding of the role of behavioral thermoregulation among species and between
244
life stages may complement information from traditional measures of thermal physiology and
245
shed light on the associations between the climate of a species distribution and its thermal
246
physiology (e.g. thermal tolerance). Behavioral responses may alleviate some of the direct
247
environmental influence on physiology by allowing organisms to avoid variable and periodically
248
stressful body temperatures (Huey et al., 2003). Organisms (or life-stages) that are good at
249
thermoregulation may distribute across a narrow range of favorable temperatures potentially
250
explaining cases where clear adaptive patterns between ambient environmental conditions and
251
organismal thermal biology, specifically thermal optima, might not exist (Stratman & Markow
252
1998; Huey, Hertz & Sinervo 2003; MacLean et al. 2019). Despite this, little is known about
253
how animals distribute across a thermally variable microhabitats (Fogleman 1979; Dillon et al.
254
2009). Here we investigated thermal preferences of adults and larvae, as well as preference for
255
adult egg-laying across a thermal gradient in four species of Drosophila representing different
256
habitat types (Kellermann et al. 2012b) to provide a better understanding of the role of
257
behavioral thermoregulation in Drosophilids. Our data produced high resolution thermal
258
distributions allowing us to estimate mean preferred temperatures and aspects of the curvature of
259
these thermal preference distributions. The use of temperature gradients to estimate thermal
12
260
preference in insects is not a new idea (Deal 1941) but this method is not widely used (but see
261
Dillon et al., 2009; Coggan et al., 2011), in part, because measurement of preferred temperature
262
is complicated and time consuming.
263
Differences in temperature preference across life-stages may be due to evolved differences
264
driven by variation in mobility or microhabitat (Kingsolver et al., 2011; Potter et al., 2011). For
265
adult egg-laying and larvae we found the least amount of variation, while adults, the most mobile
266
life-stage, exhibited the greatest amount of variation across species (Figure 2). The adult life-
267
stage is exposed to the largest variation in environmental (i.e. air) temperature and also have the
268
greatest potential for behavioral thermoregulation due to their high mobility (Dillon et al. 2009).
269
Laid eggs and larvae are restricted to food sources and have less opportunity to behaviorally
270
avoid the likely exposure to stressful thermal environments (Feder & Krebs 1998; Marais &
271
Chown 2008). We found adults produced the broadest distributions and largest species
272
differences in temperature preference, while larvae and adult egg-laying showed preference for a
273
narrower range of temperatures (Figure 2, Figure S3). This similarity in thermal preference
274
across species could be evidence of common physiological temperature dependency. Measures
275
of fitness, including egg-to-adult survival, suggest that the thermal optima for fitness is highly
276
conserved across Drosophila species. Measures of fitness, including egg-to-adult survival,
277
suggest that the thermal optima for fitness is highly conserved across Drosophila species
278
(MacLean et al., 2019) which may explain why adult flies lay their eggs within a narrow span of
279
temperatures (Figure 2) regardless of evolutionary history.
280
Notably, adults of the tropical D. birchii yielded a surprisingly low Tpref estimate. However, we
281
do not believe this is not reflective of a true preference but rather the unidirectional nature of
282
thermal gradients resulting in D. birchii adults wandering into the cold end of the gradient where 13
283
they were cold immobilized and thus could not get out (Dillon et al., 2009). Drosophila birchii
284
reared at 19°C has a CTmin of approximately 7.6°C whereas D. melanogaster, D. subobscura, and
285
D. mojavensis have CTmin of 5.0°C, -1.6°C, and 6.0°C respectively (MacLean et al., 2019). This
286
could suggest that D. birchii show increased thermal sensitivity or reduced sensing ability to cold
287
temperature compared to the other three species, which would make sense from an evolutionary
288
standpoint given that D. birchii is adapted to warm and humid tropical environments and rarely
289
experience low temperatures in its natural habitat. In contrast, e.g. D. subobscura, which is
290
adapted to temperate environments and has a much lower CTmin (MacLean et al., 2019), was
291
observed to maintain mobility across the whole gradient.
292
No clear pattern in thermal preference across species of different ecotypes were found in our
293
study. Four species is insufficient to generate robust correlations, so we assembled information
294
from the literature and analyzed all data together (Table S1, Figure 3). The juvenile life stages
295
(egg-laying and larvae) show no relationship with annual mean temperature of origin was found
296
for thermal preference. Thus, including these published data in our analysis supports our result
297
that a conserved preferred temperature exists for these stages across species. We note that our
298
data accounts for the vast majority of studies investigating egg-laying and larvae preference.
299
Excluding our potentially biased results for adult D. birchii, a weak relationship between thermal
300
preference and annual mean temperature of origin of each species was detected. However, there
301
was no strong evidence for thermal adaptation in preferred temperatures and, compared to the
302
observed difference in mean temperatures of origin across species investigated, preference
303
temperature seems rather evolutionary constrained. This is also supported by recent estimates of
304
narrow sense heritability of thermal preference in D. subobscura, are as low as h2 = 0.066,
305
suggesting limited evolutionary potential for thermal adaptation in preferred temperature 14
306
(Castañeda et al. 2019). Further, the relatively broad distribution of adult preferred temperatures
307
does not point to adults relying more on behavioral thermoregulation compared to the other life
308
stages. While, we did not find that preferred temperature strongly indicated evolutionary history.
309
Even if we provide evidence of differences among species and life stages, our data does not
310
support the prediction that adults (as the most mobile life-stage) show the greatest potential for
311
behavioral thermoregulation. Instead, we suggest that less mobile stages may benefit
312
physiologically from similar optimal temperatures achieved through narrow selection of
313
preferred temperatures.
314
We also evaluated the ability to shift preference through developmental acclimation within a
315
species. Acclimation to cold environments has been shown to lead to a compensatory increase in
316
preferred temperature (Fogleman, 1979) but more recent work on developmental acclimation and
317
preference has yielded somewhat mixed results (Dillon et al., 2009). We find little evidence of
318
such compensation in the thermal preference among the species we investigated. Rather, we see
319
a relatively strong increase in preferred egg-laying temperature and no response in preferred
320
adult temperature to thermal acclimation (Table 2). When we estimate the effect of acclimation
321
in egg laying preference we found the effect was approximately 0.3 °C/°C acclimation. This is
322
notable as it is comparable to the acclimation effects of critical lower thermal limits, which is a
323
highly labile trait (Sørensen et al., 2016; Schou et al., 2017). The direction of the acclimation
324
response in this study corresponds to an adaptive interpretation of the role of thermal preference
325
in adult egg-laying preference. However, when we consider the results from the literature there is
326
no general support for increased thermal preference with increasing acclimation temperature
327
(Table S2). Scarcity of comparable data across species presently prevent general conclusions
328
regarding the acclimation response in preferred temperatures in response to e.g. developmental 15
329
or adult acclimation. However, this comprehensive study across life stages and species suggest
330
that developmental acclimation has a large effect on adult egg placement but not on the adult
331
preference. In contrast, evolved differences between species seem largely to be affecting adult
332
preference.
333
Theoretically, organism should prefer temperatures at which they optimally perform. Our
334
analysis (Figure 4) suggests that thermal optima is always higher than thermal preference. This
335
could be due to one of two reasons; either the used estimates of thermal optima we not of
336
adequate resolution or we lacking crucial traits (see MacLean et al. 2019), or the flies are in fact
337
selecting lower than optimal temperatures as part of a bet-hedging strategy (Martin & Huey
338
2008). Due to the typical (left-skewed) shape of a thermal performance curve, the fitness
339
consequence of higher than optimal temperature strongly outweighs the consequences of lower
340
than optimal temperatures. Following this, preferring lower than optimal temperatures might
341
constitute an evolutionary stable strategy.
342
343
5. Conclusion:
344
Traditional measures of thermal tolerance and performance and the acclimation capacity thereof
345
are important for the assessment of animal fitness across thermal environments. While measures
346
of physiological capacity can be interpreted to reflect thermal adaptation when correlated with
347
environmental temperatures (Sunday, Bates & Dulvy 2011; Kellermann et al. 2012a; Overgaard,
348
Kearney & Hoffmann 2014), we should take care when applying these measures to ecological
349
and evolutionary predictions, as this thermal adaptation is not reflected in behaviorally preferred
350
temperatures. Behavioral changes are very fast and likely important in spatially and temporally 16
351
heterogeneous thermal environments. Further, if individuals are able to exploit microclimates it
352
is possible that we overestimate the negative consequences of climate change and extreme
353
climate events because ectotherms typically never experience temperatures reported. Including
354
thermal physiological capacity when assessing components of fitness may have little ecological
355
relevance because individuals may rarely experience these conditions in nature (Sunday, Bates &
356
Dulvy 2011; Kellermann et al. 2012a; Overgaard, Kearney & Hoffmann 2014). The optimal
357
temperature for an organism is complex, comprised of the optimization of many physiological
358
systems and may be evolutionary or ecologically constrained (Angilletta 2009; Schulte, Healy &
359
Fangue 2011; MacLean et al. 2019) and behavior may enable organisms to achieve these body
360
temperatures regardless of ambient air temperature in a given environment (Huey, Hertz &
361
Sinervo 2003). Measuring behavioral thermoregulation and integrating these measures with
362
measures of thermal tolerance and optimal performance from the laboratory offers a context for a
363
broader adaptive understanding of the impact of temperature on shaping past, current and future
364
distribution of species.
365
366
Acknowledgements
367
We thank Mads Fristrup Schou and Sarah E. Diamond for helpful conversations regarding data
368
analysis and Kirsten Kromand, Trine Bech Søgaard and Annemarie Højmark for excellent help
369
in the fly labs. This work was supported by grants from the Villum Foundation (JO), Aarhus
370
University Research Foundation (JGS), and the Danish Council for Independent Research (TNK
371
and JO, respectively).
372
17
373
Supplementary material: Table S1: Table of previous studies on thermal preference in
374
Drosophila species, Table S2: Description of fly stocks and origins, Figure S1: Graphical
375
depiction of experimental set-up, Figure S2: Correlation between estimated preferred
376
temperatures, Figure S3: Distribution of adult and egg-laying thermal preference following
377
developmental acclimation, Figure S4: Correlation of trait values across all temperatures.
378
379
Author Contributions: Conceptualization, HJM, JO, JGS. and TNK.; Methodology, HJM, JO,
380
JGS; Formal Analysis, HJM and JGS.; Investigation, HJM, CL, LH, EO, JO, and JGS.;
381
Resources, JO, TNK, and JGS.; Data Curation, CL, LH, EO, HJM and JGS.; Writing – Original
382
Draft Preparation, HJM and JGS.; Writing – Review & Editing, TNK, JO, CL, LH, EO;
383
Visualization, JGS and HJM.; Funding Acquisition, JO, JGS, and TNK.
384 385
Conflicts of Interest: The authors declare no conflict of interest.
386
387
References
388 389 390 391 392 393 394 395 396 397 398
Angilletta, M.J. (2009) Thermal adaptation: a theoretical and empirical synthesis. Oxford University Press. Buckley, L.B., Ehrenberger, J.C. & Angilletta, M.J. (2015) Thermoregulatory behavior limits local adaptation of thermal niches and confers sensitivity to climate change. Functional Ecology, n/an/a. Castañeda, L.E., Romero-Soriano, V., Mesas, A., Roff, D.A. & Santos, M. (2019) Evolutionary potential of thermal preference and heat tolerance in Drosophila subobscura. Journal of Evolutionary Biology, 0. Deal, J. (1941) The temperature preferendum of certain insects. The Journal of Animal Ecology, 323-356. Dillon, M.E., Wang, G., Garrity, P.A. & Huey, R.B. (2009) Thermal preference in Drosophila. Journal of Thermal Biology, 34, 109-119.
18
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
Feder, M.E. & Krebs, R.A. (1998) Natural and genetic engineering of the heat-shock protein Hsp70 in Drosophila melanogaster: Consequences for thermotolerance. American Zoologist, 38, 503-517. Fogleman, J. (1979) Oviposition site preference for substrate temperature inDrosophila melanogaster. Behavior Genetics, 9, 407-412. Hertz, P.E., Huey, R.B. & Stevenson, R. (1993) Evaluating temperature regulation by field-active ectotherms: the fallacy of the inappropriate question. The American naturalist, 142, 796-818. Higgins, J.K., MacLean, H.J., Buckley, L.B. & Kingsolver, J.G. (2014) Geographic differences and microevolutionary changes in thermal sensitivity of butterfly larvae in response to climate. Functional Ecology, 28, 982-989. Hoffmann, A.A., Sørensen, J.G. & Loeschcke, V. (2003) Adaptation of Drosophila to temperature extremes: bringing together quantitative and molecular approaches. Journal of Thermal Biology, 28, 175-216. Huey, R.B., Hertz, P.E. & Sinervo, B. (2003) Behavioral drive versus behavioral inertia in evolution: a null model approach. The American naturalist, 161, 357-366. Kellermann, V., Loeschcke, V., Hoffmann, A.A., Kristensen, T.N., Fløjgaard, C., David, J.R., Svenning, J.C. & Overgaard, J. (2012a) Phylogenetic Constraints in Key Functional Traits Behind Species' Climate Niches: Patterns of Desiccation and Cold Resistance Across 95 Drisiohila Species. Evolution, 66, 3377-3389. Kellermann, V., Overgaard, J., Hoffmann, A.A., Fløjgaard, C., Svenning, J.-C. & Loeschcke, V. (2012b) Upper thermal limits of Drosophila are linked to species distributions and strongly constrained phylogenetically. Proceedings of the National Academy of Sciences, 109, 16228-16233. Kingsolver, J.G., Woods, H.A., Buckley, L.B., Potter, K.A., MacLean, H.J. & Higgins, J.K. (2011) Complex life cycles and the responses of insects to climate change. Integrative and Comparative Biology, icr015. MacLean, H.J., Higgins, J.K., Buckley, L.B. & Kingsolver, J.G. (2016) Geographic divergence in upper thermal limits across insect life stages: does behavior matter? Oecologia, 181, 107-114. MacLean, H.J., Sørensen, J.G., Kristensen, T.N., Loeschcke, V., Beedholm, K., Kellermann, V. & Overgaard, J. (2019) Evolution and plasticity of thermal performance: An analysis of variation in thermal tolerance and fitness in 22 Drosophila species. Philosophical Transactions of the Royal Society B: Biological Sciences. Marais, E. & Chown, S.L. (2008) Beneficial acclimation and the Bogert effect. Ecology Letters, 11, 10271036. Martin, T.L. & Huey, R.B. (2008) Why “suboptimal” is optimal: Jensen’s inequality and ectotherm thermal preferences. The American naturalist, 171, E102-E118. Overgaard, J., Kearney, M.R. & Hoffmann, A.A. (2014) Sensitivity to thermal extremes in A ustralian D rosophila implies similar impacts of climate change on the distribution of widespread and tropical species. Global change biology, 20, 1738-1750. Potter, K.A., Woods, H.A. & Pincebourde, S. (2013) Microclimatic challenges in global change biology. Global change biology, 19, 2932-2939. Rajpurohit, S. & Schmidt, P.S. (2016) Measuring thermal behavior in smaller insects: A case study in Drosophila melanogaster demonstrates effects of sex, geographic origin, and rearing temperature on adult behavior. Fly, 10, 149-161. Schou, M.F., Loeschcke, V. & Kristensen, T.N. (2015) Strong costs and benefits of winter acclimatization in Drosophila melanogaster. PloS one, 10, e0130307. Schulte, P.M., Healy, T.M. & Fangue, N.A. (2011) Thermal performance curves, phenotypic plasticity, and the time scales of temperature exposure. Integrative and Comparative Biology, 51, 691-702.
19
445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460
Sørensen, J.G., Kristensen, T.N., Loeschcke, V. & Schou, M.F. (2015) No trade-off between high and low temperature tolerance in a winter acclimatized Danish Drosophila subobscura population. Journal of Insect Physiology, 77, 9-14. Sørensen, J.G., White, C.R., Duffy, G.A. & Chown, S.L. (2018) A widespread thermodynamic effect, but maintenance of biological rates through space across life's major domains. Proceedings of the Royal Society B: Biological Sciences, 285, 20181775. Stratman, R. & Markow, T. (1998) Resistance to thermal stress in desert Drosophila. Functional Ecology, 12, 965-970. Sunday, J.M., Bates, A.E. & Dulvy, N.K. (2011) Global analysis of thermal tolerance and latitude in ectotherms. Proceedings of the Royal Society B: Biological Sciences, 278, 1823-1830. Sunday, J.M., Bates, A.E., Kearney, M.R., Colwell, R.K., Dulvy, N.K., Longino, J.T. & Huey, R.B. (2014) Thermal-safety margins and the necessity of thermoregulatory behavior across latitude and elevation. Proceedings of the National Academy of Sciences, 111, 5610-5615. Wood, S.N. (2001) mgcv: GAMs and generalized ridge regression for R. R news, 1, 20-25. Woods, H.A., Dillon, M.E. & Pincebourde, S. (2015) The roles of microclimatic diversity and of behavior in mediating the responses of ectotherms to climate change. Journal of Thermal Biology, 54, 86-97.
461 462
20
463
Table 1. Characteristics of thermal preference in four species of Drosophila and in three life stages. Acclimation temperatures were
464
applied for two life stages (adult preference and egg-laying preference). The traits estimated are mean (± sem). Tpref was estimated as
465
the average temperature of animals. For the remaining traits, we fit a gam model to the proportional distribution data for each species,
466
life stage, and acclimation combination to predict frequency across the range of temperatures used. We then estimated a cumulative
467
density to estimate the temperature at which 10% (P10) and 90% (P90) of the animals occur. Tbreadth was calculated as the difference
468
between P90 and P10.
469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
Species
Stage
Accl. Temp (°C)
n
Tpref
Tbreadth
P10
P90
D. birchii -
Adult Adult Adult Egg-laying Egg-laying Egg-laying Larvae
19 23 27 19 23 27 19
12 12 12 9 12 5 8
17.4 (0.4) 16.8 (0.8) 16.7 (0.4) 23.2 (0.6) 25.1 (0.3) 26.9 (0.5) 22.9 (0.3)
16.0 (0.5) 15.7 (0.8) 15.3 (0.7) 11.6 (0.8) 11.0 (0.6) 10.2 (0.2) 8.5 (0.2)
10.9 (0.4) 10.6 (0.4) 10.7 (0.3) 17.1 (0.6) 19.1 (0.6) 20.9 (0.4) 19.5 (0.1)
27.0 (0.3) 26.3 (0.9) 26.0 (0.7) 28.7 (0.6) 30.1 (0.3) 31.2 (0.3) 28.0 (0.3)
D. melanogaster -
Adult Adult Adult Egg-laying Egg-laying Egg-laying Larvae
19 23 27 19 23 27 19
12 12 12 12 12 12 7
24.3 (0.5) 25.0 (0.4) 24.7 (0.4) 24.9 (0.3) 25.7 (0.3) 26.1 (0.4) 25.0 (0.5)
12.7 (0.6) 11.8 (0.4) 12.5 (0.7) 8.9 (0.4) 9.2 (0.3) 9.3 (0.3) 8.2 (0.6)
17.3 (0.8) 18.6 (0.5) 18.0 (0.6) 20.4 (0.3) 21.0 (0.3) 21.3 (0.4) 20.9 (0.6)
30.0 (0.3) 30.4 (0.3) 30.5 (0.3) 29.3 (0.3) 30.1 (0.3) 30.5 (0.3) 29.1 (0.1)
21
488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503
D. mojavensis -
Adult Adult Adult Egg-laying Egg-laying Egg-laying Larvae
19 23 27 19 23 27 19
14 12 12 12 12 12 9
20.8 (0.4) 22.0 (0.5) 23.0 (0.5) 24.4 (0.4) 25.5 (0.5) 26.4 (0.5) 22.9 (0.1)
15.9 (0.8) 14.8 (0.5) 15.2 (0.6) 11.5 (0.7) 9.2 (0.5) 9.7 (0.5) 7.6 (0.3)
12.8 (0.5) 14.4 (0.4) 15.2 (0.6) 18.6 (0.7) 20.7 (0.6) 21.0 (0.4) 19.6 (0.1)
28.7 (0.5) 29.2 (0.5) 30.4 (0.3) 30.1 (0.2) 29.9 (0.4) 30.7 (0.4) 27.2 (0.2)
D. subobscura -
Adult Adult Adult Egg-laying Egg-laying Egg-laying Larvae
15 19 23 15 19 23 19
12 12 12 8 11 6 8
19.0 (0.7) 18.3 (0.4) 18.8 (0.5) 20.9 (0.6) 23.8 (0.5) 24.0 (0.4) 23.3 (0.2)
16.3 (0.8) 16.0 (0.4) 17.1 (0.5) 12.9 (0.8) 10.5 (1.0) 10.0 (0.4) 8.0 (0.4)
11.4 (0.4) 11.2 (0.2) 11.0 (0.3) 14.4 (0.7) 18.3 (0.9) 18.8 (0.5) 19.7 (0.1)
27.7 (0.8) 27.1 (0.5) 28.1 (0.6) 27.2 (0.6) 28.8 (0.5) 28.8 (0.2) 27.7 (0.3)
504 505 506 507
22
508
Table 2. Summary of the F-values from linear models for Tpref, Tbreadth, P10, and P90 as a function of species as a fixed categorical
509
variable and acclimation temperature (Accl. (°C)) as a continuous variable. * indicate statistical significance (* p < 0.05, ** p < 0.01.
510
*** p < 0.001).
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
Stage
Source (df)
Tpref
Tbreadth
P10
P90
Adult
Species (3,138) Accl. (°C) (1,138)
134.7*** 1.6
24.6*** 0.3
141.0*** 3.6
32.8*** 1.2
Adult Egg-laying
Larvae
Species*Accl (3,138)
3.6*
0.6
Species (3,115) Accl. (°C) (1,115)
20.2*** 46.7***
7.5*** 8.1**
Species* Accl (3,115)
2.8*
2.5
3.6*
1.7
Species (3,28)
11.1***
1.1
5.7**
9.7***
528 529 530 531
23
3.6*
22.4*** 36.8***
2.1
14.2*** 22.2***
532 533 534 535 536 537 538
Figure 1: Example of a proportional distribution plot used to assess thermal preference. Points reflect the mean (± sem) proportion of individuals in a specific temperature bin. The temperature at which the highest number occurred is Tpref max. The curve is predicted using a generalized linear model and used to estimate the temperature of inflection (Tpref ave), the temperature at which 10% of the eggs occur (P10), the temperature at which 90% of the eggs occur (P90), and the difference between these two points (Tbreadth).
24
539 540 541 542 543 544 545
Figure 2: Proportionally distribution across a thermal gradient of adult females (left), deposited eggs of same females (middle), and of larvae (right) for four different species all acclimated at 19 °C. A clear thermal preference is observable in most cases. Note that the thermal gradient used for larval preference only covers a part of the thermal range examined for adult and egg-laying preference (as indicated by vertical dashed grey lines in the plots for adult and egg-laying preference).
546 547 548
25
549 550 551 552 553 554 555 556 557 558 559 560
Figure 3. Correlation of thermal preference versus annual mean temperature (AMT) as a measure of environmental temperature. Plots represent adult (Panel A) or juvenile (position of egg-laying or larvae) thermal preference (Panel B). AMT was extracted from for each species. Circles represent previous studies (see Table S2) and triangles represent the data from the current study. Dashed lines represent correlation across all data. Solid line (Panel A) represent correlations on all data except D. birchii (see discussion on the rationale for this). Neither juvenile nor adult preference were significantly related to AMT (for adults the exclusion of D. birchii changed this to a significant positive relationship). We do note that the data are limited and compares (for juveniles) different metrics. However, including published data both the number of investigated species and the environmental range covered is strongly increased – especially for the juvenile stages.
561
26
28 26 24 18
Topt (°C)
22 20
D. birchii D. melanogaster D. mojavensis D. subobscura
Adult Egg-laying
18
20
22
24
26
28
Tpref (°C)
562 563 564 565 566 567 568
Figure 4. Correlation of thermal preference versus temperature of optimal performance (Topt) as estimated in MacLean 2019. Plots represent adult thermal preference correlated to composite fitness optima and position of egg-laying thermal preference correlated to optimal temperature for egg-laying. We observe a correlation of r = 0.61. These data show that the preferred temperature is always lower than the optimal temperature and suggest a potential bet-hedging strategy may be employed
569 570
27
J thermal biology HIGHLIGHTS •
• •
• •
We take on the role of thermoregulatory behavior in Drosophila in a new and comprehensive way. Specifically, we apply the same methodology to four species (twice as many as has ever been published in one study before) across life-stages and following acclimation regimes. We did this to quantify evolved difference and potential plasticity in preferred temperature. We then took this two steps further. First, we extracted all available data on thermal preference in Drosophila and regressed it against annual mean temperature in species origin. Then, we related preference temperature to estimates of optimal temperature for the same populations. Because we find surprisingly little interspecific variation in temperature preference suggests behavior is an important factor in thermal adaptation or lack thereof of thermal optima. Many conceptualize thermal responses and expectations of thermal adaptation in the context of an insect’s mean thermal environment (ie air temperature) but we show organisms