Acute and Chronic Performance Enhancement in Rowing: A Network Meta-analytical Approach on the Effects of Nutrition and Training

Introduction This systematic review and network meta-analysis assessed via direct and indirect comparison the occurrence and magnitude of effects following different nutritional supplementation strategies and exercise interventions on acute and chronic rowing performance and its surrogates. Methods PubMed, Web of Science, PsycNET and SPORTDiscus searches were conducted until March 2022 to identify studies that met the following inclusion criteria: (a) controlled trials, (b) rowing performance and its surrogate parameters as outcomes, and (c) peer-reviewed and published in English. Frequentist network meta-analytical approaches were calculated based on standardized mean differences (SMD) using random effects models. Results 71 studies with 1229 healthy rowers (aged 21.5 ± 3.0 years) were included and two main networks (acute and chronic) with each two subnetworks for nutrition and exercise have been created. Both networks revealed low heterogeneity and non-significant inconsistency (I2 ≤ 35.0% and Q statistics: p ≥ 0.12). Based on P-score rankings, while caffeine (P-score 84%; SMD 0.43) revealed relevantly favorable effects in terms of acute rowing performance enhancement, whilst prior weight reduction (P-score 10%; SMD − 0.48) and extensive preload (P-score 18%; SMD − 0.34) impaired acute rowing performance. Chronic blood flow restriction training (P-score 96%; SMD 1.26) and the combination of β-hydroxy-β-methylbutyrate and creatine (P-score 91%; SMD 1.04) induced remarkably large positive effects, while chronic spirulina (P-score 7%; SMD − 1.05) and black currant (P-score 9%; SMD − 0.88) supplementation revealed impairment effects. Conclusion Homogeneous and consistent findings from numerous studies indicate that the choice of nutritional supplementation strategy and exercise training regimen are vital for acute and chronic performance enhancement in rowing.


Introduction
Rowing is considered a strength [1] endurance sport [2] that has been part of the Olympic program since 1896 [3]. In addition to a high and primarily aerobic endurance capacity [4], strength capabilities are crucial in rowing [1]. Therefore, 2000-m time trials are considered the gold standard for rowing performance testing [5,6].
Furthermore, the enhancement of acute 2000-m timetrial performance was intended via (i) postactivation potentiation [20], (ii) respiratory preconditioning [21], (iii) precooling [22], (iv) weight loss management [23], or (v) nutritional supplementation [19]. In the context of nutritional supplementation, β-alanine [24], spirulina [25], black currant [26], elk velvet antler [27], creatine monohydrate [28], beetroot [29], sodium bicarbonate [30], and sodium citrate [31] were used. Despite the multitude of different acute and chronic interventional approaches, only few rowing-specific meta-analyses on nutritional supplementation strategies [19] and exercise interventions [7,32] are available. Thereby, the effects of resistance training [7], preconditioning [32], and caffeine [19] have been meta-analytically reviewed only via direct pairwise comparisons. Accordingly, the rowing-specific findings on plyometric training [8,9], respiratory training [10,11], sprint interval training [12,13], high-intensity training [14,15], blood flow restriction methods [16], altitude training [17,18], weight loss management [23], β-alanine [24], spirulina [25], black currant [26], elk velvet antler [27], creatine monohydrate [28], beetroot [29], sodium bicarbonate [30], and sodium citrate [31] have not yet been examined via meta-analytical approaches. This is partly explained by the lack of a sufficient number of studies to perform pairwise meta-analyses in each case. Therefore, the evidence resulting from pairwise comparisons does not sufficiently provide compelling evidence and does not allow for well informed decision-making by trainers, athletes, and practitioners in the field of rowing-related training, preconditioning, and nutritional strategies. Hence, a network meta-analysis (NMA) rather than pairwise approaches can address this issue adequately by accounting for direct and indirect comparisons of different interventions [33]. A NMA does not require experimental studies to include similiar comparators, the evidence that can be integrated for the relative comparison of different intervention types is extended and more comprehensive [33]. Since a NMA enables the comparison of numerous different intervention and treatment approaches, all the above-mentioned rowing-specific findings could be examined within one analysis. In addition, a NMA approach enables a treatment ranking based on effectiveness [34].
Against this background, the present systematic review and NMA aimed to examine and compare the effects of different nutritional and exercise-based interventions on acute and chronic rowing performance through indirect and direct network-analytical comparisons. The overall results might enable athletes and coaches to select evidence-based strategies to improve rowing performance acutely and chronically, respectively.

Search and Screening Procedures
This network-analytical review was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Network Meta-Analyses (PRISMA-NMA) (Hutton et al. 2015). The literature search and screening processes were independently conducted by two researchers (LR and SH). Four health-related, biomedical, and psychological databases (PubMed, Web of Science, PsycNET, and SPORTDiscus) were screened from inception until March 7, 2022. Relevant search terms (operators) were combined with Boolean conjunctions (OR/AND) and applied to three search levels ( Table 1). In addition, tracking of cited articles and manual searching of relevant primary articles and reviews were also carried out. Duplicates were removed and the remaining studies underwent manual screening. The remaining studies were gradually screened using (i) titles, (ii) abstracts, and (iii) full texts for potentially eligible articles. Two researchers (LR and SH)made the final decision regarding inclusion or exclusion. The following inclusion criteria were applied based on the PICOS approach [population (P), intervention (I), comparators (C), main outcome (O), and study design (S)] : Full-text article published in English in a peer-reviewed journal; participants were healthy rowers (P), without any cognitive, neurological, orthopedic, and/or cardiac conditions that could affect physical testing and training; acute (≤ 7 days) or chronic (> 7 days) treatments or interventions (I); active and/or passive inactive control group(s) that received a placebo Search #1 "rowing" OR "rower" OR "row" OR "oarsmen" Search #2 #1 AND ("VO2peak" OR "VO 2max " OR "maximal oxygen uptake" OR "maximal oxygen consumption" OR "aerobic capacity" OR "threshold" OR "time trial" OR "time to exhaustion" OR "one repetition maximum" OR "1RM" OR "1 repetition maximum" OR "MVC" OR "maximal voluntary contraction" OR "rowing performance") Search #3 #2 NOT ("patient" OR "patients") treatment or did not receive any intervention served as a comparator (C); at least one rowing-related outcome such as time trial (≥ 500 m), time to exhaustion, maximal oxygen consumption (VO 2max ), power at VO 2max , or power at given lactate concentration (O); and prospective two-or multi-armed controlled intervention study with pre-and post-testing (S). The exclusion criterion was an inadequate control condition, which made integration into the network impossible.

Assessment of Methodological Quality of the Studies
The methodological quality (including risk of bias) of the included studies was independently rated by two researchers (LR and SH) using the PEDro (Physiotherapy Evidence Database) scale [35]. The PEDro scale consists of 11 dichotomous (yes or no) items, in which criteria 2-9 rate randomization and internal validity, and criteria 10-11 rate the presence of statistically replicable results. Criterion 1 merely relates to external validity and was not includeed PEDro score sum. Studies with a PEDro score ≥ 6 on a scale of 0 to 10 [35] we considered high-quality study.

Data Extraction
Relevant data (required for calculating effect sizes) were extracted independently by two researchers ( LR and SH) using a standardized extraction Excel spreadsheet adapted from the Cochrane Collaboration [36]. Means and standard deviations of pre-and post-test scores on rowing-related performance outcomes were extracted along with the number of participants assessed in each group. If these point and variability measures were not reported in the full-text article, either the means and pooled within-group standard deviations of change scores were entered in an electronic spreadsheet or the authors were contacted and missing values were requested up to three times. If studies only presented means and standard deviations in figures, WebPlotDigitizer Version 4 (Free Software Foundation, Boston, MA, USA) was used to extract means with standard deviations [37]. WebPlotDigitizer was used in 10 studies. Data from three author requests are included. For acute effects, only time-trial performance was extracted. The following ranking was used to select respective outcome parameters for chronic effects: time trial > time to exhaustion > power at VO 2max > power at a blood lactate concentration of 4 mmol/L > VO 2max . This ranking is based on the high correlations between 2000-m timetrial performance and power at VO 2max (r = 0.95, p < 0.001), power at 4 mmol/L (r = 0.92, p < 0.001), or VO 2max (r = 0.88, p < 0.001), respectively [38]. All outcomes were categorized as acute or chronic effects. In addition to these outcomes, relevant study information regarding author, year, number of participants, interventional data (weeks, frequency, duration per session, type of intervention), control condition, and PEDro scale scores were also recorded. Similar treatments are summarized in Table 2 for simplification of both networks. The corresponding interventions were classified as acute (≤ 7 days) or chronic (> 7 days).

Statistical Analysis
The standardized mean difference (SMD) and 95% confidence intervals were calculated for all interventional treatments as a measure of treatment effectiveness. SMDs were calculated as differences between means divided by the pooled standard deviations (trivial: SMD < 0.2, small: 0.2 ≤ SMD < 0.5, moderate: 0.5 ≤ SMD < 0.8, large SMD ≥ 0.8) [40]. Subsequently, two separate network models were computed for acute and chronic effects. Therefore, a frequentist approach was chosen. To visualize the networks, a network graph was created for each network. The estimations of treatment effects were calculated based on a random-effects model [41].The control group was defined as usual preparation for the acute effects and usual training for chronic effects and served as the reference treatment. A ranking was created based on the P-score of the individual treatments. The P-score represents the means of one-sided P-values under the normality assumption in a frequentist NMA [33]. This is interpreted as the mean extent of certainty that one intervention is superior to any other and is analogous to the surface under the cumulative ranking curve (SUCRA) [34] values of Bayesian NMA [33]. P-scores range from 0 to 100% with 0 and 1 being the theoretically worst and best treatment, respectively. Additionally, a forest plot was created to further visualize the ranking and effects of the treatments. Decomposed Q-statistics (within and between designs) were used to interpret potential heterogeneity and inconsistency. Heterogeneity and inconsistency were quantified using I 2 [42]. Funnel plots were created to check for potential publication bias and Egger's test for asymmetry of the funnel plot was used [43]. All calculations and presentational figures were made using the R software (version 4.1.1; The R Foundation for Statistical Computing) and the package 'netmeta' [44].

Study Characteristics and Quality
After screening and study selection (Fig. 1), 71 studies were included in the NMA. The full list of selected studies, with the corresponding study details is displayed in Table 3.
1RM one-repetition maximum, 4minP mean power of 4 min TT, ET endurance training, HiT high-intensity endurance training, HR heart rate, HR max maximal heart rate, ISO isometric, LiT low-intensity endurance training, NA not available, P4 power at 4 mmol/L lactate, PAP postactivation potentiation, PPO peak power output, PVO 2max power at VO 2max , RT resistance training, SiT sprint interval training, ThT threshold-intensity endurance training, TT time trial, TTE time to exhaustion, VO 2max maximal oxygen consumption
Both the acute-and chronic-effects networks revealed low heterogeneity and non-significant heterogeneity (see I 2 and Details of each included study are given in Table 3  Usual rowing-specific endurance and resistance training, with about > 75% low-intensity training (below first lactate threshold), < 20% threshold training (between first and second lactate threshold), and < 10% high-intensity training (above second lactate threshold), including placebo (if nutritional supplementation was used for intervention) Fig. 1 Flow chart of study screening and selection Q statistics; Fig. 2), which applied to the total, nutritionrelated, and exercise-related networks. In addition, funnel plot evaluations and non-significant Egger's tests revealed no risk of bias for all networks (Fig. 3). Only the funnel plot of the chronic exercise network (Fig. 5D) revealed a significant Egger's test result (p < 0.01). However, visual inspection indicated that this asymmetry was contrary to the corresponding publication bias. Correspondingly, the P-score-based rankings of treatments are shown in Fig. 4. In addition, pairwise comparisons of both acute and chronic effects are presented as forest plots (Fig. 5). Thereby, nutrition-related and exercise-related data for both acute-and chronic-effects networks are displayed separately.

Discussion
This is the first network meta-analytical review that reviewed acute and chronic effects of different nutritional supplementation strategies and exercise-based interventions on rowing performance. To the best of our knowledge, there has not been a summary or ranking of rowing-related interventions of this scale available to [March 7, 2022]. Our key findings indicated (i) favorable effects of caffeine supplementation on acute rowing time-trial performance, and (ii) large positive effects of blood flow restriction training and the combination of β-hydroxy-β-methylbutyrate and creatine supplementation on chronic adaptation of rowing performance indices. In contrast, our network analytical approach suggested small negative effects on acute rowing-related time-trial performance through prior weight reduction or extensive preload. Furthermore, chronic spirulina and black currant supplementations may hamper rowing performance improvements. Despite different acute and chronic interventional approaches on rowing performance, only a few rowing-specific meta-analyses on caffeine supplementation [19], preconditioning [32], and resistance training [7] are available. In contrast to the pairwise meta-analyses, our network analytical approach allowed not only two treatments to be compared, but many different treatments to be integrated into the network. Accordingly, the current study is the first to analyze acute and chronic effects of different nutritional supplementation strategies and exercise-based interventions on rowing performance.

Acute Effects
Acute caffeine supplementation scored the highest in the P-score ranking, with small-to moderate-positive effects. In line with these findings, previous multisports-based metaanalytical reviews revealed relevant improvements in time  [93,94]. Similarly, a rowing-related meta-analysis [19] revealed acute timetrial performance enhancement effects via caffeine supplementation, which is in line with our findings. Although a systematic review [95] and a meta-analysis [96] showed acute multisports-based timetrial performance enhancements via beetroot supplementation, our networkanalytical approach revealed only trivial effects on rowing time-trial performance. Similarly, several meta-analytical reviews have revealed improved muscular endurance [97], 200-400 m swimming performance [98], and (running or cycling) time to exhaustion performance [99] via acute sodium bicarbonate supplementation, whereas our findings revealed only trivial effects. These contrasting findings may be due to the small number of rowing-related studies on beetroot (n = 2) and sodium bicarbonate (n = 1) supplementation in our network model. Furthermore, our network revealed only trivial effects of acute creatine supplementation on the 2000-m time-trial performance. These findings are in line with those of previous multisports-based meta-analyses, since acute creatine supplementation increased only time-trial performance ≤ 3 min [100] and has even shown negative effects on VO 2max [101]. Based on the P-score ranking, our network showed that the effect of sodium bicarbonate on performance was enhanced by its combination with caffeine. In contrast, the effects of caffeine appeared to be impaired when combined with sodium bicarbonate. However, because of considerable overlap in the effect sizes (95% confidence intervals of standard mean differences), these differences are difficult to interpret. Future studies should investigate the effects of combining various supplementation strategies. Apart from nutritional supplementation strategies, our acute network revealed merely trivial effects of precooling on the 2000-m time-trial performance (under usual temperature conditions ≤ 23 °C). These findings were in line with previous multisports-based metaanalyses, which revealed enhancement effects of precooling on time-trial performance only in hot environments [102,103]. Likewise, a multisports-based meta-analytical review revealed small performance-enhancing effects on jumping,   [104]. In contrast, our data revealed that these PAP effects are only trivial for rowing-related 2000-m time-trial performance improvements. In addition, only one meta-analytical review indicated that an adequate warm-up procedure could improve performance [105]. Nevertheless, our network analytical approach indicated small but relevant negative effects of prior weight reduction and preload (heavy resistance training, high-intensity training or longer low-intensity training prior to testing) on subsequent rowing-specific time-trial performances. Therefore, weight reduction, heavy resistance training, high-intensity training, and longer low-intensity training should be strictly avoided within the 48 h prior to a crucial time-trial testing.

Chronic Effects
Our network analytical approach revealed large (combination of β-hydroxy-β-methylbutyrate and creatine), small (β-hydroxy-β-methylbutyrate or β-alanine), and trivial (creatine, colostrum, or elk velvet antler) beneficial effects of chronic nutritional supplementation strategies on the 2000-m timetrial performance. Interestingly, the combination of β-hydroxy-β-methylbutyrate and creatine induced more pronounced beneficial effects on rowing timetrial performance than the separate supplementation of β-hydroxy-β-methylbutyrate or creatine. The positive effects of β-hydroxy-β-methylbutyrate and creatine were partly surprising, since previous multisports-based meta-analyses and systematic reviews revealed (i) only performance-enhancing effects via creatine supplementation when timetrial duration was ≤ 3 min [100]; (ii) negative effect of creatine supplementation on maximal oxygen uptake [101]; and (iii) no effects on hypertrophy or strength if β-hydroxy-β-methylbutyrate was combined with resistance training [106]. Apart from this, other multisports-based meta-analyses [107,108] revealed only small beneficial effects of chronic β-alanine supplementation on endurance performance indices, which was confirmed by our findings. Another recent meta-analysis revealed positive effects of spirulina supplementation on oxidative stress and pro-inflammatory biomarkers [109], systolic and diastolic blood pressure [110], and body weight reduction in obese individuals [111]. However, our network analytical approach revealed that these positive effects of spirulina supplementation are not transferable to improved timetrial rowing performance. In fact, based on the P-score ranking and calculated effect sizes, negative effects on rowing-specific performance might be expected. Similarly, our results show trivial to large negative effects of black currant supplementation, although a previous multisports-based meta-analysis showed only a small, but relevant, positive effect on sport performance, with no known detrimental side effects [112]. These contrasting findings may be explained by different intervention durations. While black currant is usually supplemented for only about seven days [112], the rowing study, which is integrated in the current network analytical approach, lasted six weeks [26]. Therefore, future research on black currant should target different intervention durations. Furthermore, these partly contrasting findings may be due to the fact that only one spirulina and one black currant supplementation study was included in our network analytical approach.
Apart from these supplementation strategies, numerous previously published meta-analyses have demonstrated the beneficial effects of low-intensity and threshold-intensity training [113], high-intensity training [113,114], and sprint-interval training [114,115] on relevant endurance performance surrogate parameters such as VO 2max , lactate threshold power, or timetrial performance. Thereby,    our network also corroborated positive but trivial effects of threshold-training, high-intensity training, and sprintinterval training on rowing-specific timetrial performance. These varying magnitudes of the effect sizes could be due to the comparative conditions used in each case: Whereas pairwise meta-analyses selected a comparison condition that was substantially contrasting (e.g., low-vs high-intensity training) [113][114][115], we chose the usual rowing training as the reference intervention for our network. Since usual rowing training also contains a certain amount of threshold-, high-intensity, and sprint-interval training, the effects could partially overlap, which might account for the lower effect sizes. This usual training comparator was chosen because it best represented the actual training of successful rowers. Based on the P-score ranking and the calculated effect sizes, our network indicated that respiratory training via breathing against resistance has similar to higher effects on rowing-specific performance than threshold-, high-intensity, and sprint-interval training. Similarly, the positive effects of respiratory training on sports performance were concluded in a multisports-based meta-analysis [116]. These authors assumed that a more inclined progression of respiratory training intensity may induce even greater performance improvement [116]. Regarding resistance training, several multisports-based meta-analytical [117] and systematic reviews [118] concluded that the implementation of resistance training in addition to traditional sport-specific training improves endurance performance, mainly through improvements in the energy cost of locomotion, maximal power, and maximal strength. A recent rowing-related systematic review and meta-analysis [7] indicated that resistance training is effective in improving lower limb maximal strength and sport-specific performance in rowers. While this rowerspecific meta-analysis was based on nine studies, our network analytical approach was able to consider a total of 41 chronic intervention studies which were linked by combining direct and indirect evidence. Overall, the positive effects of resistance training are also reflected by our results. However, the different resistance training approaches show only trivial effects on the rowing-specific timetrial performance which are similar to non-resistance training approaches. These multifaceted approaches are also reflected in high-performance rowing, as successful rowing carriers can be achieved both with and without resistance training [119]. Interestingly, non-failure-based resistance training approaches such as velocity-based training [64] or repetition in reserve-based training [48] scored similar or even slightly better than the other resistance training approaches in our network. These findings are supported by a recent multisports-based metaanalysis [120] that concluded that resistance training to muscle failure does not seem to be required for gains in strength and muscle size [120]. Overall, several researchers concluded that moderate strength training volume and training not to repetition failure may be more favorable for achieving greater strength gains, muscle power, and rowing performance than with higher training volumes to repetition failure [48,121,122]. Previous non-rowing-related met a-analyses have revealed, besides improved vertical jump [123] and repeated sprint [124] abilities, endurance running performance improvements [125]. These improvements have been mainly attributed to plyometric exercises. Likewise, our network analytical approach revealed positive effects on rowing performance via plyometric training. However, the two included studies showed partly contradictory results. While one intervention study (n = 18, 4 weeks) revealed rowingspecific performance improvements through plyometric training [8], another intervention study (n=24, 9 weeks) observed no rowing-specific performance improvements [9]. These contradictory findings may be partly explained by methodological issues. For example, the sequence of stretching and contraction of a muscle tendon unit is described as a stretch-shortening cycle (SSC) [126]. In addition, an SSC enables up to 50% higher muscle force, work, and power output during the shortening phase of the SSC compared to isolated muscle shortening [127][128][129]. Considering that usual rowing results in a notable performance enhancement of ~ 10% compared to purely concentric rowing [130], it has been speculated that this is due to SSC-based mechanisms at the muscle level [130][131][132]. A differentiation between slow (> 250 ms) and fast SSC (< 250 ms) must be considered in discipline-specific movement analyses and training [133,134]. Furthermore, training adaptations in the fast SSC are not necessarily transferable to performance increases in the slow SSC (and vice versa) [126,[134][135][136]. For rowing, it has been recently shown that examinations of surface electromyographic activity of selected leg muscles (m. vastus medialis and m. gastrocnemius medialis) showed no preactivation or reflex activity, which implies that any form of muscle action in the fast SSC domain does not reflect discipline-specific muscle actions and could hamper rowing performance enhancement during training and competitions [132]. These SSC mechanism are rescently confirmed on the fascicle level in rowing. Since both rowing-related plyometric intervention studies [8,9] used slow and fast SSC exercises to different extents, a comparison of the results is difficult. Accordingly, further research on the effects of plyometric training in rowers with application of exclusively slow SSC exercises is needed.
Although previous multisports-based meta-analyses [137,138] have revealed improved endurance adaptations via altitude or hypoxic training, our network suggested that even performance declines via altitude training compared to usual rowing training. As the effect of altitude training is highly dependent on the protocol employed (e.g., sleep high, train low vs train high, sleep low) [139] and the limited number of included studies (n = 3), future rowing-related research should challenge or confirm this finding.
In contrast, our P-score ranking and calculated effect sizes showed superior adaptation via blood flow restriction training. Although these results were based on only one included study [16], they were confirmed by numerous multisportsbased meta-analyses [140][141][142]. Thereby, numerous positive effects of blood flow restriction training such as increased strength, hypertrophy, and endurance adaptations have been reported [140][141][142].
The strengths of this study outweigh potential limitations of this network meta-analysis. These strengths include (i) the large number of included studies and overall comparisons, (ii) the robust homogeneity and consistency of the formed networks, and (iii) the methodological quality of the included studies (PEDro scores > 6). Additionally, most of the findings in this analysis are a solid condensation of many trials and are largely consistent with previous literature, further supporting the plausibility of these findings. With all this in mind, it is reasonable to assume that this network metaanalysis provides valuable and important evidence despite its limitations. In addition, the current study enabled the first meta-analytical investigation of rowing-specific findings on plyometric training [8,9], respiratory training [10,11], sprint-interval training [12,13], high-intensity training [14,15], blood flow restriction methods [16], altitude training [17,18], weight loss management [23], β-alanine [24], spirulina [25], black currant [26], elk velvet antler [27], creatine monohydrate [28], beetroot [29], sodium bicarbonate [30], and sodium citrate [31].

Conclusion
This network meta-analytical review revealed (i) moderate positive effects of caffeine supplementation on acute rowing timetrial performance; (ii) small to moderate negative effects on acute rowing-related time-trial performance via prior weight reduction or extensive preload; (iii) large positive effects of blood flow restriction training and the combination of β-hydroxy-β-methylbutyrate and creatine supplementation on (chronic) improvement of rowing performance indices, and (iv) large impairment effects of rowing performance adaptations via chronic spirulina and black currant supplementation. Overall, these findings indicate that the choice of the nutritional supplementation strategy and the exercise training approach has a meaningful impact on the magnitude of the effects and should therefore be carefully considered. Future research should focus on the optimal combination of nutritional and exercise modalities.

Conflict of interest
This study has no conflicts of interest to declare.
Funding Open Access funding enabled and organized by Projekt DEAL. This study was financially supported by the German Federal Institute of Sports Science (BISp) under Grant (ZMVI4-070513/ [19][20].
Author contributions Conceptualization: SH, LR and LD; formal analysis and investigation: SH and LR; writing-original draft preparation: SH; writing-review and editing: SH, LR and LD; funding acquisition: SH and LD; supervision: LD.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.