ArticlePDF Available

Complexity Analysis of the Viking Labeled Release Experiments

Authors:

Abstract and Figures

The only extraterrestrial life detection experiments ever conducted were the three which were components of the 1976 Viking Mission to Mars. Of these, only the Labeled Release experiment obtained a clearly positive response. In this experiment 14C radiolabeled nutrient was added to the Mars soil samples. Active soils exhibited rapid, substantial gas release. The gas was probably CO2 and, possibly, other radiocarbon-containing gases. We have applied complexity analysis to the Viking LR data. Measures of mathematical complexity permit deep analysis of data structure along continua including signal vs. noise, entropy vs.negentropy, periodicity vs. aperiodicity, order vs. disorder etc. We have employed seven complexity variables, all derived from LR data, to show that Viking LR active responses can be distinguished from controls via cluster analysis and other multivariate techniques. Furthermore, Martian LR active response data cluster with known biological time series while the control data cluster with purely physical measures. We conclude that the complexity pattern seen in active experiments strongly suggests biology while the different pattern in the control responses is more likely to be non-biological. Control responses that exhibit relatively low initial order rapidly devolve into near-random noise, while the active experiments exhibit higher initial order which decays only slowly. This suggests a robust biological response. These analyses support the interpretation that the Viking LR experiment did detect extant microbial life on Mars.
Content may be subject to copyright.
Copyright The Korean Society for Aeronautical & Space Sciences
Received: November 19, 2011 Accepted: February 21, 2012
14 http://ijass.org pISSN: 2093-274x eISSN: 2093-2480
Technical Paper
Int’l J. of Aeronautical & Space Sci. 13(1), 14–26 (2012)
DOI:10.5139/IJASS.2012.13.1.14
Complexity Analysis of the Viking Labeled Release Experiments
Giorgio Bianciardi*
Department of Patologia Umana e Oncologia, Università degli Studi di Siena, Via delle Scotte 6, 53100 Siena, Italy
Joseph D. Miller**
Department of Cell and Neurobiology, Keck School of Medicine at USC, 1333 San Pablo St./BMT401, Los Angeles, CA 90033,
USA jdm@usc.edu 323-442-1629
Patricia Ann Straat***
830 Windy Knoll, Sykesville, Maryland 21784
Gilbert V. Levin****
Beyond Center, College of Liberal Arts and Sciences, Arizona State University, Tempe, AZ 85287
Abstract
e only extraterrestrial life detection experiments ever conducted were the three which were components of the 1976 Viking
Mission to Mars. Of these, only the Labeled Release experiment obtained a clearly positive response. In this experiment 14C
radiolabeled nutrient was added to the Mars soil samples. Active soils exhibited rapid, substantial gas release. e gas was
probably CO2 and, possibly, other radiocarbon-containing gases. We have applied complexity analysis to the Viking LR data.
Measures of mathematical complexity permit deep analysis of data structure along continua including signal vs. noise, entropy
vs.negentropy, periodicity vs. aperiodicity, order vs. disorder etc. We have employed seven complexity variables, all derived from
LR data, to show that Viking LR active responses can be distinguished from controls via cluster analysis and other multivariate
techniques. Furthermore, Martian LR active response data cluster with known biological time series while the control data
cluster with purely physical measures. We conclude that the complexity pattern seen in active experiments strongly suggests
biology while the dierent pattern in the control responses is more likely to be non-biological. Control responses that exhibit
relatively low initial order rapidly devolve into near-random noise, while the active experiments exhibit higher initial order
which decays only slowly. is suggests a robust biological response. ese analyses support the interpretation that the Viking
LR experiment did detect extant microbial life on Mars.
Key words: Astrobiology, extraterrestrial microbiology, Mars, Viking lander labeled release
1. Introduction
e possibility of extraterrestrial life has excited the human
imagination for hundreds of years. However, the rst (and
only) dedicated life detection experiments on another planet
were not performed until the Viking Landers of 1976. One
experiment in particular, the Labeled Release (LR) experiment
of Levin and Straat [1-4] satised a stringent set of prior agreed-
upon criteria for the detection of microbial life on Mars (i.e. a
signicant increase in evolved radioactive carbon-containing
gas over baseline after 14C radiolabeled nutrient administration
to a Mars soil sample, and abolition of that response by pre-
This is an Open Access article distributed under the terms of the Creative Com-
mons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-
nc/3.0/) which permits unrestricted non-commercial use, distribution and reproduc-
tion in any medium, provided the original work is properly cited.
***
Corresponding author : E-mail : gbianciardi@unisi.it
***
Retired (NIH)
15
Giorgio Bianciardi Complexity Analysis of the Viking Labeled Release Experiments
http://ijass.org
heating the soil to 160° C). However, controversy has reigned
ever since over these ndings. Until recently, chemical
interpretation of the LR results has dominated but discoveries
of Martian atmospheric methane [5, 6], sub-surface water ice
on Mars [7], drops of liquid water at the Phoenix landing site
[8], and the incredible hardiness of terrestrial extremophiles
[9] have all led to the re-examination of the possibility of
extant Martian microbial life.
In past work [10], we have shown that the “active” (gas-
evolving) Viking LR experiments exhibited strong circadian
rhythms in radiolabeled gas release. ese oscillations rapidly
grew in amplitude and regularity in the rst two sols (one
sol=24.66 hr, a Martian solar day) of the active experiments to
reach a near steady state of constant amplitude and period.
Perhaps, this reects the synchronization of a population of
microbes to the temperature cycle imposed by the Viking
landers. When tested, heat-treated (control) samples of the
same soil showed a greatly attenuated rhythm, or no rhythm
whatsoever. In the two experiments in which the active soil
samples were stored for several months before administering
the nutrient solution, rhythmicity was almost completely
absent.
2. New Approach
We now report a new methodological approach to these
data, complexity analysis. Due to the high order present in
biological systems [11] , time series of biological variables, with
their short- and long-range correlations, scale-invariance,
complex periodic cycles, quasi-periodicities, positive and
inverse “memory” and the like, exhibit behaviours that
are dierent from the complete unpredictability of pure
random physical processes (white noise). Moreover, they
are also distinguishable from the trivially smooth landscape
of a completely predictable deterministic process, often
manifesting themselves with icker (pink) noise (temporal
scale statistical invariance) [12, 13]. We have now found that
a set of complexity measures (appendix#1 for denition)
unambiguously distinguishes the active LR experiments, or
portions thereof, from various abiotic controls (p<0.001).
ese measures very strongly suggest, in agreement with
terrestrial analyses, that the active LR experiments in all
likelihood detected microbial life on Mars.
3. LR Results on Mars
Summary of initial analyses
In the thousands of tests that were conducted on a wide
variety of terrestrial microorganism-laden soils in 20 years of
testing before and after the Viking mission, radiolabeled gas,
presumably CO2 (or possibly CO2 plus some other carbon-
containing gas such as CH4) was produced by cellular
metabolism, always evolving immediately after the injection
of the radiolabeled LR nutrient (e.g. Biol 5, see Methods).
Heat-treated control soils produced insignicant responses
(e.g. Biol 6). ere was never a false positive or ambiguous
result in the terrestrial experiments. In the current study,
terrestrial LR pilot experiments using bacteria-laden active
(Biol 5) and sterilized (Biol 6) soil samples were analyzed,
using the same nonlinear approaches that were employed
for analysis of the Martian data.
On Mars, injected soil samples evolved radioactive gas [3,
14] rapidly, subsequently approaching plateaus of 10,000 –
15,000 cpm after several sols (Fig 3, top panel). ese “actives”
(VL1c1, VL1c3, VL2c1, VL2c3), were run at Viking Lander
sites 1 and 2, with similar results. In contrast, the LR response
in VL1c2, the 160° C control, was very low, essentially nil,
thereby, in conjunction with the active experiment results,
satisfying the pre-mission criteria for life (see appendix #2
for a brief description of the Viking LR results).
Martian soil heated for three hours at 51° C produced an
erratic succession of declining low-amplitude oscillations,
each rising for about a sol, then precipitously falling to
baseline (VL2c2). Soil treated for three hours at 46° C
responded with typical “active” kinetics, but 70% reduced
in amplitude (VL2c4). Further, formerly “active” soils stored
at 10° C for three and ve months, at Lander 2 (VL2c5), and
Lander 1 (VL1c4), respectively, failed to respond to the
nutrient [15].
A second nutrient injection was made to each “active” soil
after seven sols (VL1c1, VL2c1, VL2c3) or 16 sols (VL1c3).
Each time, the gas briey spiked, followed immediately by a
24% mean decrease in the accumulated 14C gas. Laboratory
simulations [16] showed absorption of CO2 by wetted Mars
analog soils (pH 7.2) indicating that the Viking LR gas was,
at least in part, CO2. In a terrestrial experiment, upon second
injection [17] to an Antarctic soil with known bacterial
content (pH 8.1) a brief spike also occurred, followed by a
decrease in the accumulated gas. CH4, now known to be a
component of the Martian atmosphere (6) and a possible
biological metabolite, is virtually insoluble in aqueous
media at temperatures and pressures recorded in the Viking
Landers. If produced in such experiments, then it must have
remained in the non-reabsorbed 14C-labeled gas fraction.
ese results indicate that a signicant fraction of the
14C-labeled gas evolved on Mars was CO2, at least a part of
which (~24%) was reabsorbed on wetting of what was likely
an alkaline soil [18], while the unabsorbed fraction could
DOI:10.5139/IJASS.2012.13.1.14 16
Int’l J. of Aeronautical & Space Sci. 13(1), 14–26 (2012)
have contained CH4.
As mentioned in the Introduction, circadian oscillations in
the evolved LR gas developed gradually after the rst nutrient
administration in the active experiments. e oscillations
were superimposed on the initial rise in cpm, and also on the
subsequent linear rise following the mean 24% reduction in
cpm after the second nutrient administration. e oscillations
were relatively stable in amplitude, but phase-delayed by
about two hrs compared to the daily oscillation in the lander
temperature. ese oscillations were not slavishly driven
by the diurnal temperature cycle. All these eects are more
characteristic of a biological rhythm than a purely physical
temperature-driven process [10]. Our detailed consideration
of the possible eects of soil pH, thermal variation in CO2
solubility, and a review of the ground-based and LR controls
found that CO2 absorption and release could account for
at most about 50% of the oscillatory response. us, some
fraction of the circadian oscillations as well as the evolved
gas remaining in the headspace of the instrument following
second or third nutrient injection could have been any
water-insoluble carbon-containing gas, such as CH4, which
gas, since Viking, has become of possible biological interest
in studies of Mars [19].
4. Materials and Methods
Nine LR experiments (VL1c1, VL1c2, VL1c3, VL1c4, VL2c1,
VL2c2, VL2c3, VL2c4, VL2c5) were performed on Mars soil
samples collected with a robotic arm from the surface to a
depth of about 4 cm. Each sample (0.5 cc) (Levin and Straat,
1976a)[2] was injected with 0.115 ml of a solution of formate,
glycine, glycolate, D-lactate, L-lactate, D-alanine and
L-alanine, each at 2.5 x 10−4 M, with each ingredient uniformly
labeled with 14C. e soil samples were monitored for the
evolution of 14C gas as preliminary evidence of microbial
life. L-lactate and D-alanine were included to detect alien
metabolism that might require amino acids and sugars with
a chirality dierent from ours [20](Levin et al., 1964) (using
opposite chirality enantiomers in separate experiments was
later proposed as a follow-on life detection experiment [21]
(Levin, 1987) ). To conrm a positive response, a second soil
sample was heated to sterilize it without destroying possible
inorganic chemical agents, these agents presumably being
far more heat resistant than the biological mechanism that
might plausibly have produced the positive response. us,
a negative LR response from a heat-treated soil conrmed
that the initial response in the active experiments was likely
biological, rather than inorganic.
Four experiments (VL1c1, VL1c3, VL2c1, VL2c3) were
performed on untreated soil samples. Another soil sample
was heat-treated (“sterilized”) for three hours at 160° C
(VL1c2). Two experiments utilized samples that were heat-
treated (46°C and 51°C) for three hrs (VL2c4, VL2c2). Two
soil samples (VL1c4, VL2c5), after sub-samples showed
active responses, were stored at 10°C in the dark sample
distribution box for 3 and 5 months, respectively, before
nutrient solution was administered. In all experiments
except VL2c5, a second nutrient injection was given at least
four sols after rst injection, and gas measured as above.
Biol 5 and Biol 6 data were obtained from pre-ight tests
conducted in a test instrument that was essentially identical
to the ight instrument. In these tests, LR nutrient was added
to “active” terrestrial soil with a known microbial population
(Biol 5) or to soil that had been heated for three hours at
160°C (Biol 6). e results for Biol 5 showed immediate and
rapid 14C-labeled gas evolution, typical of terrestrial soils
with modest microbial populations, whereas Biol 6 results
were essentially nil.
We employed both positive and negative controls (known
presence or absence of life) to further characterize the LR
experiments. Pre-nutrient administration background
radioactivity, a series of internal Viking Lander 1 temperature
measurements (1980 data points each taken sequentially every
960 sec), a series of external Mars atmosphere temperature
readings (1000 data points each taken sequentially every
hr) and a terrestrial heat-sterilized sample test (Biol 6)
constituted negative controls. A terrestrial bacteria-laden
active test (Biol 5) and a 23- day series of core temperature
readings taken every minute from a rat in constant darkness
constituted positive controls.
Complexity analysis
Nonlinear indices (Relative LZ Complexity or LZ; Hurst
Exponent H; Largest Lyapunov Exponent λ; Correlation
Dimension CD; Entropy K; BDS Statistic; and Correlation
Time τ; see appendix #1 for denitions) were calculated
(Chaos Data Analyzer Pro, J.C: Sprott & G. Rowlands,
American Institute of Physics, 1995) as an operational
numerical method to measure quantitatively the complexity
of the LR signal during active and control experiments on the
Viking landers and in terrestrial pilot experiments on sterile
and bacteria-laden soil samples. Moreover, negative controls
included complexity analyses of variations in pre-injection
background radioactivity, Mars atmosphere temperature and
lander temperature. A positive terrestrial control consisted
of a twenty-three day series of rat core temperature measures
that were taken every minute. Data were analyzed in three
dierent ways: 1) all usable data from a given experiment
17
Giorgio Bianciardi Complexity Analysis of the Viking Labeled Release Experiments
http://ijass.org
were pooled 2) data from each experiment were analyzed in
sequential 92 bin samples (approximately one sol of data)
throughout the experiment in which each bin constituted one
960- sec data point 3) all data points were detrended of linear
and circadian periodic components. e residuals (noise)
were then analyzed as in 1) above. For the Viking data, only
960- sec data samples were analyzed, to standardize time
and the areas of signal dropout were ignored. e number of
data points/experiments varied from 790 (~8 sols; VL1c2) to
6879 (~69 sols; VL2c3).
For the purpose of this work we dene “order” as relatively
high H, BDS and τ, and relatively low LZ, λ, CD and K.
Similarly, “disorder” may be dened as relatively low H, BDS
and τ, and relatively high LZ, λ, CD and K.
Statistical analyses
K-means cluster analysis (Systat 12) was employed to
determine whether the Viking LR experiments, averaged
over all sols, would automatically sort, on the basis of
the complexity variables, with known physical measures
(terrestrial LR pilot experiment (Biol 6) on sterile desert
soil, pre-injection random background radioactivity, Mars
atmospheric temperature, Viking lander temperature),
or with known biological measures (terrestrial LR pilot
experiment (Biol 5) on a known microbe-positive soil
sample, rat core temperature data series). e cluster
analysis was repeated only on the Viking LR active and
control experiments. ese analyses were applied both to the
raw data series and to the same series following linear and
circadian detrending (SigmaPlot 11). e derived clusters
27
Fig.1
Fig. 1.
28
Fig.2
Fig. 2.
DOI:10.5139/IJASS.2012.13.1.14 18
Int’l J. of Aeronautical & Space Sci. 13(1), 14–26 (2012)
were then validated via discriminant analysis. is allowed
the determination of jack-knifed assignment accuracy of
the individual experiments to the proposed clusters, as well
as a measure of the relative discriminative power of each
complexity measure (sequential F to remove procedure).
Repeated measures multivariate and univariate analysis of
variance was also performed to determine whether the seven
complexity variables could discriminate between active and
control Viking LR experiments over the rst six sols of the
experiments on a sol-by-sol basis. Finally, a stability analysis
compared the complexity scores between the clusters for the
rst and the last sol of each experiment.
Table 1. K-means Cluster Analysis of Averages across Sols of All Detrended Data Sets Summary Statistics for All Cases
24
Table 1 K-means Cluster Analysis of Averages across Sols of All Detrended Data Sets
Summary Statistics for All Cases
Variable Between SS df Within
SS
df F-ratio p<
LZ 10.689 1 3.311 13 41.975 .001
H 8.534 1 5.466 13 20.299 .001
λ 8.988 1 5.012 13 23.312 .001
K 4.889 1 9.111 13 6.976 .05
BDS 8.908 1 5.092 13 22.745 .001
τ 4.427 1 9.573 13 6.011 .05
** TOTAL ** 46.436 6 37.564 78
Cluster 1 (controls/physical) of 2 Contains 8 Cases
Members Statistics
Case Distance Variable Minimum Mean Maximum Standard
Error
VL2C4 0.464 LZ 0.294 0.790 1.379 0.115
Vl1C2 0.593 H -0.984 -0.706 0.111 0.144
VL1C4 0.479 λ 0.000 0.724 2.404 0.276
VL2C5 0.966 K -1.434 0.534 1.449 0.325
BIOL 6 0.311 BDS -2.010 -0.721 0.190 0.293
DT VL2C3 0.790 τ -0.645 -0.508 -0.156 0.076
VL1 Atmo. temp 0.413
Pre-inj radioactivity 0.494
Cluster 2 (actives/biological) of 2 Contains 7 Cases
Members Statistics
Case Distance Variable Minimum Mean Maximum Standard
Error
BIOL5 0.534 LZ -2.136 -0.902 -0.080 0.248
VL1C1 0.285 H -0.218 0.806 2.190 0.320
VL1C3 0.409 λ -1.202 -0.828 -0.219 0.133
VL2C1 0.544 K -1.656 -0.610 0.673 0.277
VL2C3 0.622 BDS 0.664 0.824 1.291 0.084
VL2C2 0.587 τ -0.174 0.581 3.304 0.470
Rat temp 1.354
This table lists F-ratios and p values for the complexity variables discriminating the
two clusters in the K-means cluster analysis, top panel). Cluster members
(individual experiments and data series) of the two clusters are shown (second and
third panels) with distances from the centroid for each experiment, as well as means,
SEs, and ranges for each discriminating complexity variable, as plotted in Fig 1. It
may be seen that the various experiments sort into what can be labeled as control or
physical data (Cluster 1) or active biological data (Cluster 2). Discriminant analysis
indicated that the two clusters differed significantly on the complexity variables
(p<.001).
is table lists F-ratios and p values for the complexity variables discriminating the two clusters in the K-means cluster
analysis, top panel). Cluster members (individual experiments and data series) of the two clusters are shown (second and
third panels) with distances from the centroid for each experiment, as well as means, SEs, and ranges for each discriminating
complexity variable, as plotted in Fig 1. It may be seen that the various experiments sort into what can be labeled as control
or physical data (Cluster 1) or active biological data (Cluster 2). Discriminant analysis indicated that the two clusters diered
signicantly on the complexity variables (p<.001).
19
Giorgio Bianciardi Complexity Analysis of the Viking Labeled Release Experiments
http://ijass.org
5. Results. Complexity Analysis
K-means cluster analysis (Fig 1; Table 1 automatically
sorted the active Viking LR experiment data, averaged across
all sols (VL1c1, VL1c3, VL2c1, VL2c3) with known biological
measures averaged in the same way (terrestrial LR study Biol 5
on a soil sample with known microbial content, terrestrial rat
core temperature data series). ese experiments exhibited
icker (pink) noise in the detrended active samples (LZ
active mean ± SEM =.565 ±.044). In contrast, the Viking LR
160°C control (VL1c2), the sterile soil terrestrial LR control
(Biol 6), the Viking LR sample heated to 46°C (VL2c4) and the
two long-term stored soil samples maintained in the dark at
approximately 10°C (VL1c4, VL2c5) sorted with the purely
physical measures (Viking LR pre-injection background
radioactivity, Mars atmospheric temperature series and the
VL2 lander temperature measured at the beta detector), all
approximating white noise (e.g., LZ control mean ± SEM
=.958+.022).
Another sample (VL2c2), heated to 51°C, sorted with the
actives. e acute LR response in this experiment was much
less attenuated than in the other modestly (46°C) heated
sample (VL2c4). A series of circadian oscillations with
periods identical to those seen in the active experiments,
but with lower amplitudes, was observed in the 51° C heated
sample VL2c2. e concomitant complexity measure H
remained high until the tenth sol, at which time the values
declined rapidly (Fig 2, top panel). In contrast, the 46 °C
heated sample VL2c4 exhibited high H values for a few sols,
but then rapidly declined to the level of random noise (e.g.
pre-injection background radioactivity) for the rest of the
experiment, causing it to sort with the controls in the sol-
averaged cluster analysis (Fig 2, bottom panel; Table 1).
e average cluster proles are nearly mirror images of
each other (Fig 1), with all complexity variables (F ratios
ranging from 6 to 42, df=1,13, p<.001, Table I, top panel)
except Correlation Dimension (CD), well discriminating the
clusters in this analysis. e cluster membership structure
persisted whether the raw LR data were used or data
detrended for linear and circadian components. Table 1 (top
panel) illustrates the relative strengths (F values, p values)
of the complexity variables in sorting the detrended means
of the experiments into Cluster 1 (middle panel) which we
named Actives or Cluster 2 (bottom panel) which we named
Controls on the basis of the cluster membership. Moreover,
the same sorting of the Viking experiments into “active” and
“control” clusters was seen if the analysis was limited only
to the LR experiments (Bottom panel; Fig 1). Euclidean
distances to the cluster centroids, and the mean, range and
SE of each complexity variable are also presented (middle
panel, bottom panel).
e cluster structure was robust across either Euclidean,
Minkowski, or Pearson distance measures. If three clusters
were requested, rather than two, this caused the control
cluster to fragment into sub-clusters while the active cluster
persisted. e cluster membership (all data series) was
further validated by a two-group discriminant analysis
which showed that the two clusters were easily discriminated
(Lambda =0.03, df=5,1,13; approximate overall F=59.1,
df=5,9, p<.001). Assignment of individual experiments to the
two clusters was 100% correct by the jack-knife classication
procedure. e variables best able to discriminate the two
clusters in the discriminant analysis were LZ, H, λ, BDS
and entropy (K) with F to remove ranging from 5.3 (K)
to 36.9 (LZ), approximately p<.02 to p<.001 (Systat12). If
the analysis was limited to the LR data series, results were
essentially the same (discriminant analysis, Lambda=.028,
df=4, 1, 9; approximate overall F=51.4, df=4,6, p<.001, 100%
correct classication of experiments by jackknife). For the
restricted analysis LZ, H, BDS and K were the best cluster
discriminators, with F to removes ranging from 5.7 to 36,
approximate p=.05 to p<.001.
Multivariate and univariate analysis of variance indicated
that most of the complexity variables with the exceptions of
K and CD, discriminated the two clusters averaged across
all sols (Hotelling-Lawley Trace, Wilk’s Lambda, Pillai
29
Fig.3
Fig. 3.
DOI:10.5139/IJASS.2012.13.1.14 20
Int’l J. of Aeronautical & Space Sci. 13(1), 14–26 (2012)
Trace, df=7,1, p<.02 for all three statistics). For the other ve
variables F values ranged from 5.8 to 46.3, with p values from
.007 to .05. e LZ mean value for the detrended active LR
experiments was consistent with pink or icker noise (mean
± SEM =.565 ±.044) while the LZ mean for the controls was
characteristic of white noise (mean ± SEM =.958±.022), and
was signicantly higher than the active LZ mean (two tailed
independent t test, LZ actives vs. controls, t=8, df=7, p<.001).
Changes in complexity restricted to the rst six sols of
each experiment were also examined. Fig 3, Top panel, plots
the raw LR data from an active experiment (VL2c3) and the
160° C control (VL1c2), while the bottom panel plots the
complexity variable, H, over the same time course for the
two experiments.
A two-tailed independent t test showed that this complexity
variable easily discriminated the two experiments (n=16 sol
by sol data points, df=14, t=3.76, p<.005). In order to show
sol to sol variability, additional complexity variables are
plotted over the rst several sols for selected individual
LR experiments in Fig 4. In general, H values are higher
for active experiments than for controls (Note that H=0 for
pre-injection measures (sol 0) of radioactivity for VL2c3,
and VL1c2) while the reverse is true for LZ and λ. In active
experiments (i.e.,VL2c3, Biol 5), λ and LZ climb and H
declines over sols. In contrast, the rat temperature series and
the sterile control series (VL1c2, Biol 6) maintain relatively
stable values.
e cluster mean data for active and control experiments
are plotted across sols for the three best discriminating
complexity variables (LZ, H, λ) in Fig 5. H is signicantly
higher when it is averaged across all the active experiments
than when it is averaged across the controls. e reverse is
true for LZ and λ.
While the condence intervals overlap to some extent, in
18 separate comparisons on these three complexity measures
the active experiments diered consistently in complexity
from the controls. e probability of this occurring by chance
for independent events is 1/218, much less than p<.001. For
the Between-Clusters eects on each complexity variable
(df=1,9), F=9.26, p=.01 for LZ; F=17.46, p<.002 for H; F=6.44,
p=.03 for λ; F=3.95, p<.08 marginal for BDS, F=3.28, p=0.10
marginal for τ. CD and K failed to discriminate the clusters
over the rst six sols. A signicant Sol eect (df=5, p<.05) was
seen for all complexity variables except CD, K and τ. Cluster
x Sol interactions were non-signicant for all the complexity
variables.
30
Fig.4
Fig. 4.
31
Fig.5
Fig. 5.
21
Giorgio Bianciardi Complexity Analysis of the Viking Labeled Release Experiments
http://ijass.org
A stability analysis compared the rst and the last sols
of the experiments. It demonstrated that the complexity
measures evolve over time in the direction of disorder,
dened as relatively high K, LZ, λ and CD, but relatively low
τ, H and BDS in all the experiments, in agreement with the
signicant Sol eect seen in the six sol analyses (Fig 3, 4, 5).
However, the starting values of the complexity variables LZ
(p<.05), H (p<.004), λ (p<.02), BDS (p<.07, marginal) and τ
(p<.01) were signicantly dierent, favoring greater order,
for the active experiments compared to the controls. K and
CD indices were numerically smaller on Sol 1 for the actives
vs. the controls, but these dierences were not signicant.
By the last sol of the experiments, all complexity scores had
evolved in the direction of disorder and there was no longer
any dierence in complexity between the actives and the
controls. Furthermore, active experiments were longer in
duration (mean=34 sols) compared to the controls (mean=20
sols). us, it took over 40 sols for complexity to change in
the active experiment VL2c3, whereas the moderately heat-
treated experiments VL2c2 and VL2c4 exhibited substantial
order for several sols, typical of an active response. However,
the response then rapidly decayed to a near-random state
of disorder for the remainder of the experiments, perhaps
indicating a cessation of biological activity soon after dosing
(Fig 2).
6. Discussion
For almost 35 years a controversy has raged over whether
or not the Viking LR experiment detected life on Mars.
Although the results of the LR experiment met the pre-launch
criteria for the detection of life, the dominant explanation of
the results was that a superoxide in the soil was responsible
for oxidizing the organic molecules in the LR nutrient. Levin
and Straat [22] spent three years seeking a chemical or
physical method of duplicating the Mars LR test and control
data, to no avail. Moreover, Levin [23] reviewed more than
two dozen abiotic explanations that had been proposed
over the years, and found all of them wanting. None of the
strong oxidants proposed over the years exhibit the thermal
prole of the active Martian agent as established by the LR
experiments. Superoxides synthesized in the laboratory [24]
as candidates for the LR response turned out to be unstable
in aqueous media, breaking down in seconds. In contrast,
stable LR signals were detected for many weeks after the
administration of the aqueous nutrient. Furthermore, it is
unclear why sample storage for 3-5 months at 10°C in the
dark would virtually destroy the response from a strong
oxidant or superoxide. e perchlorate discovered in Mars
soil [18] similarly fails as a candidate for the LR response.
On the other hand, in recent years, biological interpreta-
tions of the LR experiment have become more acceptable
with the discovery of equatorial methane-generating regions
on Mars overlapping areas with extensive sub-soil water ice
deposits and atmospheric water vapor (http://www.esa.int/
SPECIALS/Mars_Express/SEML131XDYD_0.html). Further-
more, study of terrestrial extremophiles, including methane-
generating microbes in desert sub-soil [25] indicate that such
organisms can thrive in arid sub-soil environments compa-
rably harsh to the Martian environment. It has been pro-
posed by Levin and Straat [26] and later by Miller et. al, [19]
that both the persistence of methane in the Martian atmo-
sphere and its required sink can be explained by the possible
presence of methane-producing and methane-consuming
microorganisms similar to those on Earth.
In past work [10] we have shown that the active LR signal
is periodic, exhibiting a circadian (more appropriately
circasolar) rhythm with a period of 24.66 hr, approximating
the rotational period of Mars. e periodicity in the LR
experiments rapidly evolves over time, and can be almost
entirely extinguished by heat treatment or long-term soil
sample storage. Circadian rhythms are robust biosignatures,
and the presence of such rhythms in the LR signal is at least
consistent with a biological interpretation.
In the current work we have demonstrated that the LR
signal (and the associated noise) in active LR experiments
is very dierent from the LR signal for heated or long-term-
stored soil samples. Furthermore, the active LR experiment
data cluster with known biological signals (rat temperature
series, the active terrestrial LR pilot study Biol 5), exhibiting
icker (pink) noise in the detrended active samples , whereas
the LR control studies cluster with non-biological signals
such as random background radiation, Mars atmospheric
temperature, Mars lander temperature, and a terrestrial LR
sterile control (Biol 6), approximating white noise . ese
clusters are robust under dierent measures of inter-cluster
distance and persist even when the individual cluster
membership is restricted to the Viking LR experiments, with
active experiments and control experiments sorting into two
distinct clusters. An attempt to form a third cluster simply
fragments the controls. Discriminant analysis conrmed the
cluster structure and classied each experiment correctly
into either the active or control clusters. Repeated measures
analysis of variance indicated that the complexity dierences
between actives and controls were generally stable over
the rst sols of the study, but decreased in magnitude
gradually in that period and very strongly when evaluated
over the entire experiment in the stability analysis. Most
importantly, the active experiments exhibited higher order,
DOI:10.5139/IJASS.2012.13.1.14 22
Int’l J. of Aeronautical & Space Sci. 13(1), 14–26 (2012)
dened on the basis of the complexity measures, early in
the experiments compared to the controls. Order declined
more slowly over time in the active experiments than did
the already low order in the controls. Overall, the complexity
variables that best discriminated actives from controls were
LZ, H, λ, BDS and τ, while K and CD failed to do so. In terms
of the denition of the complexity variables (Methods), the
active experimental data were more predictable and pinker
(LZ), persistent (H), periodic (λ), and less random (BDS, τ)
compared to the control data. Entropy (K) was numerically
higher in the controls, consistent with the other results, but
the dierence from the K values for the active experiments
was not statistically signicant.
7. Conclusion
e multivariate analyses, especially the cluster analyses,
clearly distinguished between active and control Martian LR
experiments. When a number of terrestrial time series, known
to be biological or non-biological, were added to the set of LR
experiments, the biological time series automatically sorted
with the LR active experiments, and the non-biological
time series sorted with the LR controls, forming two distinct
clusters on the basis of the complexity variables. In the multi-
dimensional space dened by those variables, the cluster
analysis indicated that the active LR experiments were more
similar to the terrestrial biological time series and the control
LR experiments were more similar to the non-biological
terrestrial time series. In mathematical terms, the Euclidean
distance between the centroids of the two clusters was
signicantly larger than the intra-cluster distances between
any members of either cluster. It is reasonable to infer from
this analysis that the Martian active LR experiments were
more likely detecting a biological process, whereas the
Martian control LR experiments were more likely detecting
a non-biological process.
us, we have shown that complexity variables distinguish
active LR experiments (Martian and terrestrial samples)
from control LR experiments. e active experiment Viking
LR signals have a complexity that is similar to the biological
signals, such as the terrestrial microbe-positive LR pilot study,
Biol 5, while the controls (heated, Martian and terrestrial)
LR signals are more similar in complexity to non-biological
signals. Moreover, stability analysis indicated that the kinetics
of complexity decay are very dierent for the actives vs. the
controls. It is dicult to see why order declined so strongly
in the active experiments if a purely non-biological process
were responsible. On the other hand, an increase in the
disorder of a biological measure can simply reect the decline
of a microbial population under deprivation or heat stress
(a much smaller extant population surviving in the control
experiments could nevertheless exhibit a small reduction in
residual order over time for the same reason). In contrast the
complexity measures on the rat temperature rhythm are very
stable (e.g. Fig.4), an expected outcome from a continuously
viable preparation. It is also possible that 46°C-51°C is near
threshold for the deleterious eects of heat stress on the
microbial population, since one of the moderately heated
soil samples produced a response that consistently sorted
with the active experiments (VL2c2), while the other (VL2c4)
always sorted with control experiments (In spite of an early,
transient, possibly biological response, the dominant pattern
over the rest of this experiment caused it to sort with the
controls, see Fig.2).
It is important to say that, the nature of the LR gas (es)
and the degree to which apparent circadian oscillations
reect CO2 solubility in moist Mars soil is not completely
resolved. Nevertheless, if the LR gas evolution in the active
experiments were entirely non-biological, it would sort with
the other purely physical, rather than biological processes. In
actuality, LR gas evolution in the active experiments sorted
with the biological measures, while gas evolution controls
(e.g. heat-sterilized) sorted with non-biological measures.
We believe that these results provide considerable support
for the conclusion that the Viking LR experiments did,
indeed, detect extant microbial life on Mars.
Appendix #1. Brief description of complexity vari-
ables
Relative LZ complexity, LZ: Relative LZ complexity is
a measure of the algorithmic complexity of a time series
[27]. According to the Kaspar and Schuster algorithm, each
data point is converted to a single binary digit according to
whether the value is less than, or greater than, the median
value of a set of data points [28]. Applying the software used
in the present paper to known series, LZ results are:
LZ value
White noise 1.04
Pink noise 0.70
Sine+noise 0.16
Heart rate 0.74 (median, range=0.51-0.89)
(Adult healthy subjects)
White noise (a pure random signal, common in physical
systems, that exhibits equal power across all the component
frequencies of the signal), has an LZ value that is close to 1.0.
Pink noise (icker noise or 1/f noise), exhibits decreasing
23
Giorgio Bianciardi Complexity Analysis of the Viking Labeled Release Experiments
http://ijass.org
power as frequency increases, and is associated with a
relatively low LZ value; it is common in biological systems
(see Heart rate). A sine function with 10% superimposed
Gaussian white noise yields an LZ value that is close to zero.
e algorithm for calculating LZ, applied in the present
paper, converts it to a single binary digit which indicates
whether the value was less than, or greater than, the median
value of a set of at least 92 such data points.
Hurst exponent, H: e Hurst exponent is the slope of the
root-mean-square displacement of each data point versus
time. Applying the software used in the present paper to
known series, H values are:
H value
White noise 0.00
Pink noise 0.16
Brownian noise 0.53
Sine+noise 0.51
Heart rate 0.19 (median, range=0.12-0.36)
(Adult healthy subjects)
e H value for white noise is equal to 0. If H ≠ 0.5, then
correlation exists, the noise is “colored” and the process
exhibits a “memory”: if the exponent is greater than 0.5,
persistence occurs (past trends will statistically persist in the
future, see sine function), and, vice versa, if H is less than 0.5,
anti-persistence occurs (past trends tend to reverse in the
future, see Pink noise and biological signals such as Heart
rate). For Brownian motion, a random process in which, on
average, each point moves away from its initial condition by
an amount that is proportional to the square root of time,
the Hurst exponent exhibits a value which is close to 0.5 (no
memory) [29].
Largest Lyapunov Exponent, λ: Lyapunov exponents
measure the rate at which the nearby trajectories in phase
space diverge. Here, the most positive exponent is calculated
according to a published algorithm [30]. e embedding
dimension and the number of sample intervals were always
both xed in the present paper at D=3 and n= 3.Applying
the software used in the present paper to known series, the
results for λ are:
λ value
White noise 0.89
Pink noise 0.73
Sine + noise 0.51
Sine 0.00
Heart rate 0.35 (median, range=0.21-0.60)
(Adult healthy subjects)
e exponent is numerically high for pure randomness
(white noise). Pink noise and biological signals e.g. Heart
rate, exhibit relatively low values. λ =0 (or a negative value)
for purely periodic data, such as the sine function.
Correlation Dimension, CD: e fractional correlation
dimension was obtained by counting the data points that
are inside hyperspheres of various radii centered on each
data point in a phase space of some embedding dimension,
according to a published algorithm [31]. e correlation
dimension in these data sets was calculated with embedding
dimensions between 1 and 10. A plot of correlation dimension
vs embedding dimension was performed and the value of
the Correlation Dimension (CD) at plateau was chosen.
A simple deterministic function, such as a sine function,
exhibits a CD value that is close to 1, while a purely random
distribution exhibits a CD value of 6 or more. Biological
signals, such as Heart rate exhibit CD values that are less
than 6, but are larger than 1.
Entropy, K: e entropy index chosen here [32] is a
measure of the disorder in a data set and was calculated as
the sum of the positive Lyapunov exponents.
Randomness is indicated by numerically high values of
entropy. Ordered series like the sine function exhibit values
that are close to 0.
BDS statistic, BDS: e Brock-Dechert-Scheinkman
statistic detects serial dependence in time series and can
thereby quantitate the deviation of the data from pure
randomness. Applying the software used in the present
paper to known series, the results are as follows:
BDS value
White noise -16.8
Pink noise -0.6
Sine + Noise +2.5
Heart rate +0.2 (median, range= -3.3 to +1.6)
(Adult healthy subjects)
In short discrete time series, pure randomness (white
noise) exhibits BDS values <<0, while more ordered series
exhibit greater values of BDS [33] .
Correlation time, τ: A measure of how dependent data
points are on their temporal neighbours. It is taken as the
time at which the correlation function rst falls to 1/e. By
DOI:10.5139/IJASS.2012.13.1.14 24
Int’l J. of Aeronautical & Space Sci. 13(1), 14–26 (2012)
Acknowledgments
e authors wish to thank Dr. Ralph Mistlberger for kindly
providing the rat temperature data series.
References
[1] Levin, G.V., “Detection of metabolically produced
labeled gas: e Viking Mars Lander”, Icarus, Vol. 16, 1972,
pp. 153-166.
[2] Levin, G.V., and Straat, P.A., “Labeled Release – an
experiment in radiorespirometry”, Origins of Life, Vol. 7,
1976, pp. 293-311.
[3] Levin, G.V., and Straat, P.A., “Viking Labeled Release
biology experiment: interim results”, Science, Vol. 194, 1976,
pp. 1322-1329.
[4] Klein, H.P., Horowitz, N.H.,,Levin, G.V.,Oyama,
V.I.,Lederberg, J., Rich, A.,, Hubbard, J.S., Hobby, G.L., Straat,
P.A., Berdahl, B.J., Carle, G.C., Brown, F.S. and Johnson, R.S.,
“e Viking biological investigations: preliminary results”,
Science, Vol.194, 1976, pp. 99-105.
[5] Formisano, V., Atreya, S., Encrenaz, T., Ignatiev, N., and
Giuranna, M., “Detection of methane in the atmosphere of
Mars”, Science, Vol. 306, 2004, pp.1758–1761.
[6] Mumma, M.J., Villanueva, G.L., Novak, R.E., Hewagama,
T.H., Bonev, B.P., DiSanti, M.A., Mandell, A.M., and Smith,
J.D., “ Strong release of methane on Mars in northern
summer”, Science, Vol. 323, 2009, pp. 1041-1045.
[7] Smith, P.H., Tamppari, L.K., Arvidson, R.E., Bass,
D., Blaney, D., Boynton, W.V., Carswell, A., Catling, D.C.,
Clark, B.C., Duck, T., DeJong, E., Fisher, D., Goetz, W.,
Gunnlaugsson, H.P., Hecht, M.H., Hipkin, V., Homan, J.,
Hvlid, S.F., Keller, H.U., Kounaves, S.P., Lange, C.F., Lemmon,
applying the software used in the present paper to known
series, the results are as follows:
τ value
White noise 0.64
Pink noise 21.2
Sine + noise 11.9
Heart rate 4.9 (median, range=2.7 – 11.4)
(Adult healthy subjects)
Pure random data (white noise) will have no
intercorrelation and the τ value will be close to zero [34].
Larger correlations with time result in larger τ values.
Appendix #2 Viking Mission Labeled Release Ex-
periment Results
e LR controls established that the “active agent”
detected in the Martian soil was destroyed at 160°C, was
greatly impaired at 46°C, essentially destroyed at 51°C, and
fully depleted after storage in the dark inside the sample
distribution box at approximately10°C for three and four
months at the respective sites. All the results were supportive
or consistent with the detection of a biological agent.
RUN TYPE1AGE2TEMP.3PLATEAU42NDINJ.53RDINJ.6,
(A/C) (SOLS) (OC) (SOLS/CPM) (SOLS/CPM) (SOLS/CPM),
V1c1 A 2 ~10 8/10,000 8/ ~7,500 NA
V1c2 C 28 160 NA/~1,000 27/~800 NA
V1c3 A 3 ~10 16/~16000 16/~11,500 41/~11,000
V1c4 A 140 ~10-26 18/~2,000 NA, NA
V2c1 A 3 ~10 7/~14,000 7/~11,000 NA
V2c2 C 6 51 NA/~1,000 10/~200, NA
V2c3 A 2 ~9 7/~10,000 7/~7,500 NA
V2c4 C 2 46 14/~6,000 7/~4,200 NA
V2c5 A 84 ~7 NA NA/~250 NA
1Active or Control sample
2Sols after sample rst taken
3Treatment prior to run
4Maximum approached
5Minimum immediately after 2nd injection
6Minimum immediately after 3rd injection
25
Giorgio Bianciardi Complexity Analysis of the Viking Labeled Release Experiments
http://ijass.org
M.T., Madsen, M.B., Markiewicz, W.J., Marshall, J., McKay,
C.P., Mellon, M.T., Ming, D.W., Morris, R.V., Pike, W.T., Renno,
N., Staufer, U., Stoker, C., Taylor, P., Whiteway, J.A., and Zent,
A.P., “H2O at the Phoenix landing site”, Science, Vol. 325,
2009, pp. 58-61.
[8] Renno, N.O., Bos, B.J., Catling, D., Clark, B.C.,
Drube, L., Fisher, D., Goetz, W., Hviid, S.F., Keller, H.U.,
Kok, J.F., Kounaves, S.P., Leer, K., Lemmon, M., Madsen,
M.B., Markiewicz, W.J., Marshall, J., McKay, C., Mehta,
M., Smith, M., Zorzano, M.P., Smith, P.H., Stoker, C., and
Yound, S.M.M., “Possible physical and thermodynamical
evidence for liquid water at the Phoenix landing site”,
Journal Of Geophyical. Research, Vol. 114, 2009, E00E03,
doi:10.1029/2009JE003362.
[9] Cavicchioli, R., “Extremophiles and the search for
extraterrestrial life”, Astrobiology, Vol. 2, 2002, pp. 281-292.
[10] Miller, J.D., Straat, P.A., and Levin, G.V., “Periodic
analysis of the Viking lander Labeled Release experiment”,
Instruments, Methods, and Missions for Astrobiology IV,
SPIE Proceedings, Vol. 4495, 2002, pp. 96-107.
[11] Ross, J., and Arkin, A.P., “ Complex systems: from
chemistry to system biology”, Proceedings National Academy
Of Science, Vol. 106, 2009, pp. 6433-6434.
[12] Mosconi, F., Julou, T., Desprat, N., Sinha, D.K.,
Allemand, J.F., Croquette, V., and Bensimon, D., “Some
nonlinear challenges in biology”, Nonlinearity, Vol. 21, 2008,
T131-T147.
[13] Gisiger, T., “Scale invariance in biology: coincidence
or footprint of a universal mechanism?”, Biological Reviews,
Vol. 76, 2001, pp. 161-209.
[14] Levin, G. V., and Straat, P.A., “Life on Mars? e Viking
Labeled Release experiment”, BioSystems, Vol. 9, 1977, pp.
165-174.
[15] Levin, G.V., and Straat, P.A., “Completion of the Viking
Labeled Release experiment on Mars”, Journal of. Molecular
Evolution, Vol. 14, 1979, pp.167-183.
[16] Levin, G.V., and Straat, P.A., “ Laboratory simulations
of the Viking Labeled Release experiment: kinetics following
second nutrient injection and the nature of the gaseous end
product”, Journal of. Molecular Evolution, Vol. 14, 1979, pp.
185-197.
[17] Levin, G.V., and Straat, P.A., (1986)., “A reappraisal of
life on Mars, Proceedings of the NASA/Mars Conference at
the National Academy of Sciences, Science and Technology
Series, Univelt, Inc., San Diego, Vol. 71, 1986, pp. 187- 207.
[18] Hecht , M.H., Kounaves, S.P., Quinn, R.C., West,
S.J., Young, S.M.M., Ming, D.W., Catling, D.C., Clark, B.C.,
Boynton, W.V., Homan, J., DeFlores, L.P., Gospodinova, K.,
Kapit, J., and Smith, P.H., “Detection of perchlorate and the
soluble chemistry of Martian soil at the Phoenix lander site”,
Science, Vol. 325, 2009, pp. 64-67.
[19] Miller, J.D., Case, M.J., Straat, P.A., and Levin, G.V.,
“Likelihood of methane-producing microbes on Mars”,
Instruments, Methods, and Missions for Astrobiology
XIII, SPIE Proceedings, Vol. 7819, 2010, 78190I;
doi:10.1117/12.862230.
[20] Levin, G.V., Heim, A.H., ompson, M.F., Beem,
D.R., and Horowitz, N.H., “Gulliver- An experiment for
extraterrestrial life detection and analysis”, Life Sciences and
Space Research, Vol. 2, 1964, pp. 124-132.
[21] Levin, G.V.,“e life on Mars dilemma and the sample
return mission”, Proceedings of the Mars Sample Return
Science Workshop, Houston, 1987, pp.109-110.
[22] Levin, G.V., and Straat, P.A., A Search for a
Nonbiological Explanation of the Viking Labeled Release Life
Detection Experiment”, Icarus, Vol. 45, 1981, pp. 494-516.
[23] Levin, G.V., “e oxides of Mars”, Instruments,
Methods, and Missions for Astrobiology, SPIE Proceedings,
Vol. 4495, 2002, pp. 131-135.
[24] Yen, A.A., Kim, S.S., Hecht, M.H., Frant, M.S., and
Murray, B., “Evidence that the reactivity of the Martian soil
is due to superoxide ions”, Science, Vol. 289, 2000, pp. 1909-
1912.
[25] Moran, M., Miller, J.D., Kral, T., and Scott, D., “Desert
methane: implications for life detection on Mars”, Icarus, Vol.
178, 2005, pp. 277-280.
[26] Levin, G.V., and Straat, P.A., “Methane and life on
Mars”, Instruments, Methods and Missions for Astrobiology
and Planetary Missions XII, SPIE Proceedings,. Vol. 7441,
2009, 74410D1 – 74410D16.
[27] Lempel, A., and Ziv, J., “On the complexity of nite
sequence”, IEEE Transactions on Information eory, IT-22,
1976, pp. 75-81.
[28] Kaspar, F., and Schuster, H.G., “Easily calculable
measure for the complexity of spatiotemporal patterns”,
Physical Review A, Vol. 36,1987, pp. 842-848.
[29] Feder, J., Fractals, Plenum, New York and London,
1988, pp 149-153.
[30] Wolf, A., Swift, J.B., Swinney, H.L., and Vastano, J.A.,
“Determining Lyapunov exponents from a time series”,
Physica D, Vol. 16, 1985, pp. 285-317.
[31] Grassberger, P., and Procaccia, I., “Characterization of
strange attractors”, Physical Review Letters, Vol. 50, 1983, pp.
346-349.
[32] Grassberger, P., and Procaccia, I., “Estimation of the
Kolmogorov entropy from a chaotic signal”, Physical Review
A, Vol. 28, 1983, pp. 2591-2593
[33] Brock W.A.,“Distinguishing random and deterministic
systems: abridged version”, Journal of Economic eory, Vol.
40, 1986, pp. 168-195.
DOI:10.5139/IJASS.2012.13.1.14 26
Int’l J. of Aeronautical & Space Sci. 13(1), 14–26 (2012)
[34] Wei, W.W., Time series analysis: univariate and
multivariate methods, Addison-Wesley, New York, 1989.
Figure Legends
Figure 1. Cluster analysis discriminates active and control
LR experiments. e top panel gives the two cluster result for
K-means clustering of all Viking LR studies, VL1c1-VL1c4,
VL2c1-VL2c5, the two ground-based studies, microbe-
positive Biol 5 and sterilized Biol 6, VL1 atmospheric
temperatures, VL2c3 detector temperatures, plus two anchor
cases, a terrestrial rat circadian rhythm, and random pre-
injection radioactivity. e means ±S.E. for the active (red
bars) and control (blue bars) clusters are given for each
of the six complexity variables which discriminated the
clusters. e bottom panel shows only the VLR study means
across the same complexity variables for the active (red bars)
and control (black bars) clusters. e cluster membership
(all data series) was further validated by a two-group
discriminant analysis. is showed that the two clusters were
easily discriminated (Lambda =0.03, df=5,1,13; approximate
overall F=59.1, df=5,9, p<.001). For the LR studies alone
(Lambda=.028, df=4, 1, 9; approximate overall F=51.4, df=4,6,
p<.001. N for each experiment varied from 790 (VL1c2) to
6879 (VL2c3) data points.
Figure 2. Hurst analysis of VL2c2 and VL2c4. is graph
compares the raw LR scores with the complexity variable
H on every sol throughout experiments VL2c2 and VL2c4
in which the soil samples were heated to intermediate
temperatures (51°C and 46°C, respectively) before the
nutrient administration. e top panel shows the complexity
score H which is maintained at a high level for most of VL2c2,
even though the amplitude of the circadian oscillations is
markedly reduced following partial sterilization. When the
LR values drop to baseline noise (~140 cpm; e.g. sols 10-11, or
after sol 13 in VL2c2), H values drop to near zero. is pattern
resulted in this experiment automatically sorting with active
experiments in the cluster analysis. In contrast, the bottom
panel shows relatively high H values for VL2c4 for the rst
few sols which decline rapidly over the rest of the experiment.
Since the cluster analysis is performed on the average
complexity response, indices like H cause this experiment to
automatically sort with the control experiments, in spite of
an early transient response. is possibly indicates a rapidly
expiring biosignature due to the thermal exposure. N=1203
data points and 14 H measures for VL2c2; 2128 data points
and 23 H measures for VL2c4.
Figure 3. Complexity measures dier radically between the
active and control LR experiments. Top panel: Radioactivity
counts (RC) in cpm plotted against time in sols from rst
nutrient injection to second nutrient injection for an active
experiment (VL2c3) and the 160 °C sterilization control
(VL1c2). Note oscillations beginning at about Sol 1 on the x
axis. N=555 data points for VL2c3 and VL1c2.
Bottom panel: e Hurst exponent H is plotted as in the
top panel but Sol 0 is pre-injection background radioactivity
for which the expected value of H is zero. An independent
two-tailed t test was calculated for eight daily H values in the
active sample (VL2c3) vs. the eight daily H values for the full
sterilization control (VL1c2), N=16 data points, df=14, t=3.76,
p<.005.
Figure 4. Complexity scores for rst six sols discriminate
individual active and control experiments. is graph plots
individual complexity scores starting with pre-nutrient
injection background radioactivity (a good example of white
noise) and extending across the rst six sols of an active
VLR experiment (VL2c3), the sterilization control (VL1c2), a
microbe-positive terrestrial control (Biol 5) a sterile terrestrial
control (Biol 6) and a rat circadian temperature rhythm that
functions as another terrestrial positive control. e top
panel plots LZ vs Sols, the middle panel plots H vs. Sols and
the bottom panel plots λ vs. Sols for the various time series.
N=92 data points, each taken every 16 min, constituting one
sol of data per complexity measure. Note that H=0 for pre-
injection radioactivity and that Biol 5 and Bio6 do not have
pre-injection radioactivity scores or scores on Sol 6.
Figure 5. Average complexity measures vs. time in sols
discriminate active and control LR experiments.
is gure plots the mean values of the active (N=5;
VL1c1, VL1c3, VL2c1, VL2c2, VL2c2) vs. the control Viking
LR experiments (N=4; VL1c2, VL1c4, VL2c4, VL2c5) for each
of the three complexity indices (LZ, H, λ) against time in sols.
Top panel: LZ vs Sols. Middle panel: H vs. Sols. Bottom panel:
λ vs. Sols. Error bars are SEM. For the Between Clusters
eects on each complexity variable (df=1,9), F=9.26, p=.01
for LZ; F=17.46, p<.002 for H; F=6.44, p=.03 for λ
... Controversies are no doubt present in astrobiology. To give a few examples, controversies can be about (i) facts, for instance: about the putative traces of life on Mars discovered by the Viking landers in 1976 (Klein et al. 1976;Biemann et al. 1977;Bianciardi et al. 2012;Levin and Straat 2016); about the biogenicity of the oldest fossilized candidate micro-traces of life (Schopf 2006;Brasier et al. 2015;Javaux 2019); or about the existence of "arsenic life" or microbes capable of incorporating arsenic instead of phosphorus into their DNA (Wolfe-Simon et al. 2011;Erb et al. 2012;Reaves et al. 2012). There are also controversies about (ii) theories or theoretical constructs, for instance about the heterotrophic or autotrophic nature of the first living entities, the temporal priority of genetic information over metabolism, and the importance of cellularization in the early stages of life (Peretó 2005). ...
Article
Full-text available
Disagreement in science exists in a variety of strengths, from doubt-raising articles and issues of non-reproducibility up to raging disputes and major controversies. An often-latent form of disagreement consists of misconceptions whereby false ideas are held that run contrary to what is commonly accepted as knowledge. Misconceptions have been the focus of much research in education science and psychology. Here we draw attention to misconceptions that may arise in the very practice of science. We highlight formal features that can be used to characterize misconceptions and distinguish them from controversies, in addition to how they relate to knowledge creation.
... Acqua allo stato liquido non vuol dire vita, anche se questa è un prerequisito indipensabile per la vita. Possiamo comunque ricordare come negli ultimi 20 anni evidenze per una possibile vita marziana nell'antico Marte non sono mancate (il metorite ALH84001, [17]) così come nel presente Marte (analisi frattale dei gas rilasciati nell'esperimento biologico Labeled Release dei Viking, [18]). ...
... Also, the reports of methane in the Martian atmosphere suggested to reconsidering the question of organics on Mars (Ten Kate 2010). A biological interpretation of the Viking experiment was suggested by applying mathematical and fractal analysis to the record of monitored data, which have proved an occurrence of a circadian sequence related to a bacterial metabolism that should prove the existence of an extant microbial life on Mars (Bianciardi et al. 2012). At the same time, fossil evidence of microorganisms was, also, suggested about the Martian meteorite ALH84001 (McKay et al. 1996). ...
Article
Full-text available
The study analyses possible parallels of the microbialite-known structures with a set of similar settings selected by a systematic investigation from the wide record and data set of images shot by NASA rovers. Terrestrial cases involve structures both due to bio-mineralization processes and those induced by bacterial metabolism, that occur in a dimensional field longer than 0.1 mm, at micro, meso and macro scales. The study highlights occurrence on Martian sediments of widespread structures like microspherules, often organized into some higher-order settings. Such structures also occur on terrestrial stromatolites in a great variety of 'Microscopic Induced Sedimentary Structures', such as voids, gas domes and layer deformations of microbial mats. We present a suite of analogies so compelling (i.e. different scales of morphological, structural and conceptual relevance), to make the case that similarities between Martian sediment structures and terrestrial microbialites are not all cases of 'Pareidolia'.
... Will never be able to definitively prove the existence of life on the Red Planet? The search for definitively proving the presence of life on Mars is one of the outstanding scientific challenges of our time [69,83,84]. We have described in this paper how the Curiosity landed region was clearly demonstrated as a fluvialdeltaic-lacustrine environment [2]. ...
... The Viking lifedetection experiments were inconclusive and demonstrated difficulty in confirming or excluding the occurrence of extant life (Schuerger & Clark, 2008;Schulze-Makuch et al., 2015). Consequently, the question about life on Mars remains unanswered (Bianciardi et al., 2012;Levin & Straat, 2016;Guaita, 2017). It is worth mentioning that during all three biological experiments (the Labeled Release, the Pyrolytic Release, and the Gas Exchange or the Carbon Assimilation), the temperature was considerably elevated (to 8-26°C, depending on experiment; Horowitz et al., 1977;Klein, 1977;Levin & Straat, 1977;Oyama & Berdahl, 1977;Levin & Straat, 2016) as compared to that outside both Viking landers (-90°C to -5°C;Kieffer, 1976). ...
Article
Full-text available
Life-detection experiments carried out in extraterrestrial locations provided inconclusive results whether processes observed were biological or chemical. In this study, the typical effect of temperature on metabolic rates is described and a life-detection method that is easy to perform is proposed. The method comprises observing changes in microbial metabolic rates after temperature shift. The method was demonstrated by experiments on aquatic microorganisms in the Gulf of Gdansk (Baltic Sea). First experiment, in which temperature was shifted within the temperature range encountered at the sampling site, demonstrated a typical Q10 coefficient (2.84). The experiment in which temperature was shifted beyond the environmental temperature range provided an unexpectedly low Q10 coefficient (1.44), which indicated that excessive temperature exerted an inhibitory effect on metabolism. This response is not expected for chemical reactions, but it is typical for biological processes. In summary, a pair of properly-tailored experiments permitted separating biological and chemical reactions.
... Subsequently it was determined that Martian perchlorates are not reactive given the parameters of the LR experiments: massive amounts of hydrogen peroxide would have been required to simulate the LR results (Levin & Straat,1981). Subsequently, Bianciardi, Miller, Straat, and Levin (2012) performed a mathematical complexity deep analysis of the Viking LR data, employing seven complexity variables. It was determined that the LR responses from the Raw Soil exhibited highly organized responses typical of biological activity whereas responses from the 160°C samples resembled random noise. ...
Article
Full-text available
Vesicular rocks and thick clumps of green-colored matter photographed in Utopia Planitia and Chryse Planitia by NASA's Viking landers were subject to morphological and computerized quantitative pattern analysis. These vesicular rocks are not homogenous and include those similar to vesicular basalts, marine trace fossils, and "tafoni" which on Earth are fashioned via the interactional influences of moisture, powerful winds, the leaching of salts and lichen-chemical weathering. Upon magnification the green-colored vesicular substances closely resemble "vegetative matter" similar to green algae, lichens, mosses and vesicular mats. The green colors (based on false colors derived from spectra) may be indicative of chlorophyll and the capacity to produce oxygen via photosynthesis. These observations, when coupled with the continual replenishment of atmospheric oxygen and evidence of surface frost, subsurface water-ice, and past cycles of flooding and ponding of water, are supported by the positive results from the Viking Labeled Release and Gas Exchange experiments and should be viewed as confirming that beginning in 1976 the USA and NASA's Viking Landers 1 and 2 detected, photographed and discovered life and evidence of past life on Mars.
Article
Organisms act stochastically. A not uncommon view in the ecological literature is that this is mainly due to the observer having insufficient information or a stochastic environment-and not partly because organisms themselves respond with inherent unpredictability. In this study, I compile the evidence that contradicts that view. Organisms generate uncertainty internally, which results in irreducible stochastic responses. I consider why: for instance, stochastic responses are associated with greater adaptability to changing environments and resource availability. Over longer timescales, biologically generated uncertainty influences behavior, evolution, and macroecological processes. Indeed, it could be stated that organisms are systems defined by the internal generation, magnification, and record-keeping of uncertainty as inputs to responses. Important practical implications arise if organisms can indeed be defined by an association with specific classes of inherent uncertainty: not least that isolating those signatures then provides a potential means for detecting life, for considering the forms that life could theoretically take, and for exploring the wider limits to how life might become distributed. These are all fundamental goals in astrobiology.
Article
Full-text available
Possibilities of intelligent extraterrestrial life.
Article
Full-text available
Ten ovoid life-like specimens, less than a cm in size, that appear to possess forward facing eyes and pleopod-pareiopod-like appendages have been photographed by the rover Curiosity attached to the sides of sediment in Gale Crater. Dozens of ovoid specimens with multiple appendages were also photographed within and on the outside of honeycombed sediment. If these appendages are pleopods or pareiopods they would enable movement and feeding behavior. One specimen with forward-facing eye-like orifices was photographed occluding a hole in a sedimentary side surface and the following day the specimen was photographed several centimeters outside the now empty hole upon a ledge. Two specimens with forward-facing eye-like orifice were photographed engaged in what appears to be feeding behavior. Over three dozen investigators have published evidence indicating that Mars was and is still inhabited by stromatolite-constructing organisms, and that acritarchs, green algae, fungi, lichens, corals, sponges, mollusks, tube worms, and other metazoan invertebrates evolved on the Red Planet. Hence, although it cannot be precisely determined if these ovoid specimens are in fact Martian “animalia” the evidence presented here supports the hypothesis that complex living organisms evolved on and still inhabit Mars.
Article
A new measure of strange attractors is introduced which offers a practical algorithm to determine their character from the time series of a single observable. The relation of this new measure to fractal dimension and information-theoretic entropy is discussed.
Article
The Viking Landers were unable to detect evidence of life on Mars but, instead, found a chemically reactive soil capable of decomposing organic molecules. This reactivity was attributed to the presence of one or more as-yet-unidentified inorganic superoxides or peroxides in the martian soil. Using electron paramagnetic resonance spectroscopy, we show that superoxide radical ions (O2 –) form directly on Mars-analog mineral surfaces exposed to ultraviolet radiation under a simulated martian atmosphere. These oxygen radicals can explain the reactive nature of the soil and the apparent absence of organic material at the martian surface.
Article
The Labeled Release extraterrestrial life detection experiment onboard the Viking spacecraft is described as it will be implemented on the surface of Mars in 1976. This experiment is designed to detect heterotrophic life by supplying a dilute solution of radioactive organic substrates to a sample of Martian soil and monitoring for evolution of radioactive gas. A significantly attenuated response by a heat-sterilized control sample of the same soil would confirm a positive metabolic response. Experimental assumptions as well as criteria for the selection of organic substrates are presented. The Labeled Release nutrient has been widely tested, is versatile in eliciting terrestrial metabolic responses, and is stable to heat sterilization and to the long-term storage required before its use on Mars. A testing program has been conducted with flight-like instruments to acquire science data relevant to the interpretation of the Mars experiment. Factors involved in the delineation of a positive result are presented and the significance of the possible results discussed.
Article
Did Viking Lander biology experiments detect life on Mars? The strongest evidence for biology resulted from the Labeled Release (LR) experiment1. A radiolabeled (14C) nutrient solution was added to a Martian soil sample and the subsequent evolution of radioactive gas was observed. Flight data showed an initial release of labeled gas followed by strong periodic fluctuations in amount of gas in the headspace above the soil, superimposed on a slow rise in release. Current analyses show, at steady state, these fluctuations exhibit a periodicity of 24.66+/- 0.27 hr, statistically indistinguishable from the Martian solar period. The gas fluctuation appears synchronized to a mean 2 degree(s)C periodic fluctuation in internal temperature in the experimental chamber, which in turn is synchronous with almost 50 degree(s)C daily fluctuations in ambient Mars surface temperature. Calculations based on LR data indicate that the daily gas fluctuation amplitude could be in part accounted for by change in temperature-dependent soil solubility of CO2, but total amount of gas accumulated cannot be accounted for in this way. Recent observations of circadian rhythmicity in microorganisms and entrainment of terrestrial circadian rhythms by low amplitude temperature cycles argue that a Martian circadian rhythm in the LR experiment may constitute a biosignature.
Article
Mumma et al. 1 have confirmed earlier detections of methane in the Martian atmosphere, finding it localized and correlated with atmospheric water vapor. They determined that, because of the short half-life of methane, a continual replenishment is required to account for its presence. They also conclude that the dynamics of methane on Mars require a methane sink in the soil. It is suggested here that both phenomenon could be accounted for by an ecology of methane-producing and methane-consuming microorganisms. Such ecologies exist on Earth, where, generally, anaerobic methanogens live at depth and aerobic methanotrophs live at or near the surface. On Mars, with its essentially anaerobic atmosphere, both types of microorganisms could co-exist at or near the surface. It is possible that the Viking Labeled Release (LR) experiment detected methanogens in addition to other microorganisms evolving carbon dioxide since the LR instrumentation would detect methane, carbon dioxide, or any other carbon gas derived from one of the LR substrates. A simple modification of the LR experiment that could resolve the life on Mars issue is discussed.
Article
Ever since the Viking Mission landed on Mars, a hypothetical film of highly oxidizing material has been applied to the Red Planet by a host of articles in the scientific literature. This putative chemical is credited with destroying all organic matter and preventing extant life. The only 'evidence' cited for the oxidant is a re-interpretation of the Viking biology experiments. On the other hand, direct experimental evidence from Mariner 9, Viking, Pathfinder, and Kitts Peak clearly demonstrate that Mars does not have a highly oxidative surface. This should remove the primary reason commonly cited against the Viking LR experiment having detected microorganisms in the Martian soil. For those requiring further evidence, an unambiguous test is proposed for the next Mars lander.
Article
We present a likelihood estimate that methane was a significant component of the gas detected by the Labeled Release (LR) experiment in the Viking Mission to Mars of 1976. In comparison with terrestrial methanogen production of methane we estimate the size of the putative microbe population necessary to produce the LR gas, had it been primarily methane. We extrapolate that figure to estimate the number of methanogens necessary to produce the methane content of the Martian atmosphere. Next, we estimate the amount of Martian soil and the amount of water needed for that global population of microbes. Finally, assuming a globally distributed population of such microbes, we estimate the likely sub-surface depth at which such methanogens could be detected.